Maxwell Kadel v. Dale Folwell

100 F.4th 122

Opinion

PUBLISHED

UNITED STATES COURT OF APPEALS FOR THE FOURTH CIRCUIT

No. 22-1721

MAXWELL KADEL; JASON FLECK; CONNOR THONEN-FLECK; JULIA MCKEOWN; MICHAEL D. BUNTING, JR.; C.B., by his next friends and parents; SAM SILVAINE; DANA CARAWAY,

Plaintiffs - Appellees,

v.

DALE FOLWELL, in his official capacity as State Treasurer of North Carolina; EXECUTIVE ADMINISTRATOR OF THE NORTH CAROLINA STATE HEALTH PLAN FOR TEACHERS AND STATE EMPLOYEES,

Defendants – Appellants,

and

NORTH CAROLINA STATE HEALTH PLAN FOR TEACHERS AND STATE EMPLOYEES; STATE OF NORTH CAROLINA DEPARTMENT OF PUBLIC SAFETY,

Defendants.

--------------------------------------- COMMONWEALTH OF KENTUCKY; COMMONWEALTH OF VIRGINIA; STATE OF ALABAMA; STATE OF ALASKA; STATE OF ARKANSAS; STATE OF FLORIDA; STATE OF GEORGIA; STATE OF INDIANA; STATE OF IOWA; STATE OF KANSAS; STATE OF LOUISIANA; STATE OF MISSISSIPPI; STATE OF MISSOURI; STATE OF MONTANA; STATE OF NEBRASKA; STATE OF NORTH DAKOTA; STATE OF OHIO; STATE OF OKLAHOMA; STATE OF SOUTH CAROLINA; STATE OF TEXAS; STATE OF UTAH,

Amicus Supporting Appellants.

NEW YORK; CALIFORNIA; COLORADO; DELAWARE; HAWAII; ILLINOIS; MAINE; MARYLAND; MASSACHUSETTS; MINNESOTA; NEVADA; NEW JERSEY; NEW MEXICO; OREGON; RHODE ISLAND; VERMONT;

2

WASHINGTON; DISTRICT OF COLUMBIA; CONSTITUTIONAL LAW PROFESSORS; AMERICAN MEDICAL ASSOCIATION; AMERICAN COLLEGE OF OBSTETRICIANS AND GYNECOLOGISTS; AMERICAN PSYCHIATRIC ASSOCIATION; AMERICAN ACADEMY OF PEDIATRICS; ENDOCRINE SOCIETY; NORTH AMERICAN SOCIETY FOR PEDIATRIC AND ADOLESCENT GYNECOLOGY; NATIONAL ASSOCIATION OF NURSE PRACTITIONERS IN WOMEN’S HEALTH; SOCIETY OF OB/GYN HOSPITALISTS,

Amici Supporting Appellees.

Appeal from the United States District Court for the Middle District of North Carolina, at Greensboro. Loretta C. Biggs, District Judge. (1:19-cv-00272-LCB-LPA)

No. 22-1927

SHAUNTAE ANDERSON, individually and on behalf of all others similarly situated, Plaintiffs - Appellees,

v.

WILLIAM CROUCH, in his official capacity as Cabinet Secretary of the West Virginia Department of Health and Human Resources; CYNTHIA BEANE, in her official capacity as Commissioner for the West Virginia Bureau for Medical Services; WEST VIRGINIA DEPARTMENT OF HEALTH AND HUMAN RESOURCES, Bureau for Medical Services,

Defendants – Appellants.

--------------------------------------- STATE OF WEST VIRGINIA; COMMONWEALTH OF KENTUCKY; COMMONWEALTH OF VIRGINIA; STATE OF ALABAMA; STATE OF ALASKA; STATE OF ARKANSAS; STATE OF FLORIDA; STATE OF GEORGIA; STATE OF INDIANA; STATE OF IOWA; STATE OF KANSAS; STATE OF LOUISIANA; STATE OF MISSISSIPPI; STATE OF MISSOURI; STATE OF MONTANA; STATE OF NEBRASKA; STATE OF NORTH DAKOTA; STATE OF OHIO; STATE OF OKLAHOMA; STATE OF SOUTH CAROLINA; STATE OF TEXAS; STATE OF UTAH,

Amicus Supporting Appellants,

3

COLORADO; DELAWARE; DISTRICT OF COLUMBIA; FAIRNESS WEST VIRGINIA; MOUNTAIN STATE JUSTICE, INC.; NATIONAL HEALTH LAW PROGRAM; CENTER FOR MEDICARE ADVOCACY; CONSTITUTIONAL LAW PROFESSORS; AMERICAN MEDICAL ASSOCIATION; ENDOCRINE SOCIETY; NATIONAL ASSOCIATION OF NURSE PRACTITIONERS IN WOMEN'S HEALTH; AMERICAN PSYCHIATRIC ASSOCIATION; SOCIETY OF OB/GYN HOSPITALISTS; ILLINOIS; MAINE; MARYLAND; MASSACHUSETTS; MINNESOTA; NEVADA; NEW JERSEY; NEW MEXICO; NEW YORK; OREGON; RHODE ISLAND; VERMONT; WASHINGTON,

Amici Supporting Appellees.

Appeal from the United States District Court for the Southern District of West Virginia, at Huntington. Robert C. Chambers, District Judge. (3:20-cv-00740)

Argued: September 21, 2023 Decided: April 29, 2024

Before DIAZ, Chief Judge, WILKINSON, NIEMEYER, KING, GREGORY, AGEE, WYNN, THACKER, HARRIS, RICHARDSON, QUATTLEBAUM, RUSHING, HEYTENS, and BENJAMIN, Circuit Judges.

Affirmed by published opinion. Judge Gregory wrote the opinion, in which Chief Judge Diaz, Judge King, Judge Wynn, Judge Thacker, Judge Harris, Judge Heytens, and Judge Benjamin joined. Judge Richardson wrote a dissenting opinion, in which Judge Wilkinson, Judge Niemeyer, Judge Quattlebaum joined, and in which Judge Agee and Judge Rushing joined except for part II.A.3. Judge Wilkinson wrote a dissenting opinion. Judge Quattlebaum wrote a dissenting opinion, in which Judge Agee, Judge Richardson, and Judge Rushing joined.

No. 22-1721. ARGUED: John Guyton Knepper, LAW OFFICE OF JOHN G. KNEPPER, LLC, Cheyenne, Wyoming, for Appellants. Tara Lynn Borelli, LAMBDA LEGAL DEFENSE & EDUCATION FUND, INC., Decatur, Georgia, for Appellees. ON BRIEF: Kevin G. Williams, Mark A. Jones, BELL, DAVIS & PITT, P.A., Winston-Salem, North Carolina, for Appellants. Amy E. Richardson, Lauren E. Snyder, HWG LLP, Washington, D.C.; Michael W. Weaver, Chicago, Illinois, Dmitriy G. Tishyevich, Warren Haskel, MCDERMOTT WILL & EMERY, New York, New York; Carl S. Charles, Decatur, Georgia, Omar Gonzalez- Pagan, LAMBDA LEGAL DEFENSE AND EDUCATION FUND, INC., New York, New York; David P. Brown, Ezra Cukor, TRANSGENDER LEGAL DEFENSE & EDUCATION FUND, INC., New York, New York, for Appellees. Howard S. Suskin, Chicago, Illinois, Matthew D. Cipolla, New York, New York, Illyana A. Green,

4

Christina M. Isnardi, JENNER & BLOCK LLP, Washington, D.C.; Shana L. Fulton, Sarah M. Saint, BROOKS PIERCE MCLENDON HUMPHREY & LEONARD, LLP, Greensboro, North Carolina, for Amici American Medical Association and Seven Additional Health Care Organizations. Katie R. Eyer, RUTGERS LAW SCHOOL, Camden, New Jersey; Andrew Barr, Denver, Colorado, Kathleen Hartnett, COOLEY LLP, San Francisco, California, for Amici Constitutional Law Professors. Letitia James, Attorney General, Barbara D. Underwood, Solicitor General, Ester Murdukhayeva, Deputy Solicitor General, Daniel S. Magy, Assistant Solicitor General, Andrea W. Trento, Assistant Solicitor General, OFFICE OF THE ATTORNEY GENERAL OF NEW YORK, New York, New York, for Amicus State of New York. Rob Bonta, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF CALIFORNIA, Sacramento, California, for Amicus State of California. Philip J. Weiser, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF COLORADO, Denver, Colorado, for Amicus State of Colorado. Kathleen Jennings, Attorney General, DELAWARE DEPARTMENT OF JUSTICE, Wilmington, Delaware, for Amicus State of Delaware. Holly T. Shikada, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF HAWAI’I, Honolulu, Hawai’i, for Amicus State of Hawai’i. Kwame Raoul, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF ILLINOIS, Chicago, Illinois, for Amicus State of Illinois. Aaron M. Frey, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF MAINE, Augusta, Maine, for Amicus State of Maine. Brian E. Frosh, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF MARYLAND, Baltimore, Maryland, for Amicus State of Maryland. Maura Healey, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF MASSACHUSETTS, Boston, Massachusetts, for Amicus Commonwealth of Massachusetts. Keith Ellison, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF MINNESOTA, St. Paul, Minnesota, for Amicus State of Minnesota. Aaron D. Ford, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF NEVADA, Carson City, Nevada, for Amicus State of Nevada. Matthew J. Platkin, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF NEW JERSEY, Trenton, New Jersey, for Amicus State of New Jersey. Hector Balderas, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF NEW MEXICO, Santa Fe, New Mexico, for Amicus State of New Mexico. Ellen F. Rosenblum, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF OREGON, Salem, Oregon, for Amicus State of Oregon. Peter F. Neronha, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF RHODE ISLAND, Providence, Rhode Island, for Amicus State of Rhode Island. Susanne R. Young, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF VERMONT, Montpelier, Vermont, for Amicus State of Vermont. Robert W. Ferguson, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF WASHINGTON, Olympia, Washington, for Amicus State of Washington. Karl A. Racine, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF THE DISTRICT OF COLUMBIA, Washington, D.C., for Amicus District of Columbia. Andrew Bailey, Attorney General, Joshua M. Divine, Solicitor General, Kenneth C. Capps, Assistant Attorney General, OFFICE OF THE ATTORNEY GENERAL OF MISSOURI, Jefferson City, Missouri, for Amicus State of Missouri. Steve Marshall, Attorney General, OFFICE

5

OF THE ATTORNEY GENERAL OF ALABAMA, Montgomery, Alabama, for Amicus State of Alabama. Treg Taylor, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF ALASKA, Anchorage, Alaska, for Amicus State of Alaska. Tim Griffin, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF ARKANSAS, Little Rock, Arkansas, for Amicus State of Arkansas. Ashley Moody, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF FLORIDA, Tallahassee, Florida, for Amicus State of Florida. Christopher M. Carr, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF GEORGIA, Atlanta, Georgia, for Amicus State of Georgia. Theodore E. Rokita, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF INDIANA, Indianapolis, Indiana, for Amicus State of Indiana. Brenna Bird, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF IOWA, Des Moines, Iowa, for Amicus State of Iowa. Kris W. Kobach, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF KANSAS, Topeka, Kansas, for Amicus State of Kansas. Daniel Cameron, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF KENTUCKY, Frankfort, Kentucky, for Amicus Commonwealth of Kentucky. Jeff Landry, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF LOUISIANA, Baton Rouge, Louisiana, for Amicus State of Louisiana. Lynn Fitch, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF MISSISSIPPI, Jackson, Mississippi, for Amicus State of Mississippi. Austin Knudsen, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF MONTANA, Helena, Montana, for Amicus State of Montana. Michael T. Hilgers, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF NEBRASKA, Lincoln, Nebraska, for Amicus State of Nebraska. Drew H. Wrigley, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF NORTH DAKOTA, Bismarck, North Dakota, for Amicus State of North Dakota. Dave Yost, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF OHIO, Columbus, Ohio, for Amicus State of Ohio. Gentner Drummond, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF OKLAHOMA, Oklahoma City, Oklahoma, for Amicus State of Oklahoma. Alan Wilson, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF SOUTH CAROLINA, Columbia, South Carolina, for Amicus State of South Carolina. Ken Paxton, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF TEXAS, Austin, Texas, for Amicus State of Texas. Sean D. Reyes, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF UTAH, Salt Lake City, Utah, for Amicus State of Utah. Jason S. Miyares, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF VIRGINIA, Richmond, Virginia, for Amicus Commonwealth of Virginia.

No. 22-1927. ARGUED: Michael Ray Williams, OFFICE OF THE ATTORNEY GENERAL OF WEST VIRGINIA, Charleston, West Virginia, for Amicus Curiae. Caleb David, SHUMAN MCCUSKEY SLICER PLLC, Charleston, West Virginia, for Appellants. Tara Lynn Borelli, LAMBDA LEGAL DEFENSE & EDUCATION FUND, INC., Decatur, Georgia; Anna Purna Prakash, NICHOLAS KASTER, LLP, Minneapolis, Minnesota, for Appellees. ON BRIEF: Kimberly M. Bandy, Lou Ann S. Cyrus, Roberta F. Green, SHUMAN MCCUSKEY SLICER PLLC, Charleston, West Virginia, for Appellants. Avatara Smith-Carrington, Washington, D.C., Carl Charles, Decatur, Georgia, Nora Huppert,

6

LAMBDA LEGAL DEFENSE AND EDUCATION FUND, INC., Chicago, Illinois; Nichole J. Schladt, NICHOLS KASTER, PLLP, Minneapolis, Minnesota; Walt Auvil, THE EMPLOYMENT LAW CENTER, PLLC, Parkersburg, West Virginia, for Appellees. Patrick Morrisey, Attorney General, Lindsay S. See, Solicitor General, OFFICE OF THE ATTORNEY GENERAL OF WEST VIRGINIA, Charleston, West Virginia, for Amicus State of West Virginia. Howard S. Suskin, Lillian M. McGuire, Chicago, Illinois, Matthew D. Cipolla, New York, New York, Christina M. Isnardi, JENNER & BLOCK LLP, Washington, D.C.; Shana L. Fulton, Sarah M. Saint, BROOKS HUMPHREY MCLENDON HUMPHREY & LEONARD, LLP, Greensboro, North Carolina, for Amici American Medical Association and Four Additional Health Care Organizations. Jah Akande, Alicia M. Penn, Evan X. Tucker, MCGUIREWOODS LLP, Richmond, Virginia, for Amici Fairness West Virginia and Mountain State Justice, Inc. Letitia James, Attorney General, Barbara D. Underwood, Solicitor General, Ester Murdukhayeva, Deputy Solicitor General, Daniel S. Magy, Assistant Solicitor General, OFFICE OF THE ATTORNEY GENERAL OF NEW YORK, New York, New York, for Amicus State of New York. Philip J. Weiser, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF COLORADO, Denver, Colorado, for Amicus State of Colorado. Kathleen Jennings, Attorney General, DELAWARE DEPARTMENT OF JUSTICE, Wilmington, Delaware, for Amicus State of Delaware. Kwame Raoul, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF ILLINOIS, Chicago, Illinois, for Amicus State of Illinois. Aaron M. Frey, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF MAINE, Augusta, Maine, for Amicus State of Maine. Brian E. Frosh, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF MARYLAND, Baltimore, Maryland, for Amicus State of Maryland. Maura Healey, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF MASSACHUSETTS, Boston, Massachusetts, for Amicus Commonwealth of Massachusetts. Keith Ellison, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF MINNESOTA, St. Paul, Minnesota, for Amicus State of Minnesota. Aaron D. Ford, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF NEVADA, Carson City, Nevada, for Amicus State of Nevada. Matthew J. Platkin, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF NEW JERSEY, Trenton, New Jersey, for Amicus State of New Jersey. Hector Balderas, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF NEW MEXICO, Santa Fe, New Mexico, for Amicus State of New Mexico. Ellen F. Rosenblum, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF OREGON, Salem, Oregon, for Amicus State of Oregon. Peter F. Neronha, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF RHODE ISLAND, Providence, Rhode Island, for Amicus State of Rhode Island. Susanne R. Young, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF VERMONT, Montpelier, Vermont, for Amicus State of Vermont. Robert W. Ferguson, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF WASHINGTON, Olympia, Washington, for Amicus State of Washington. Karl A. Racine, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF THE DISTRICT OF COLUMBIA, Washington, D.C., for Amicus District of Columbia. Katie R. Eyer, RUTGERS LAW SCHOOL, Camden, New Jersey; Andrew Barr, Denver, Colorado, Kathleen Hartnett,

7

COOLEY LLP, San Francisco, California, for Amici Constitutional Law Professors. Martha Jane Perkins, NATIONAL HEALTH LAW PROGRAM, Chapel Hill, North Carolina; Alice Bers, Wey-Wey Kwok, CENTER FOR MEDICARE ADVOCACY, Willimantic, Connecticut, for Amici National Health Law Program and Center for Medicare Advocacy. Andrew Bailey, Attorney General, Joshua M. Divine, Solicitor General, Kenneth C. Capps, Assistant Attorney General, OFFICE OF THE ATTORNEY GENERAL OF MISSOURI, Jefferson City, Missouri, for Amicus State of Missouri. Steve Marshall, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF ALABAMA, Montgomery, Alabama, for Amicus State of Alabama. Treg Taylor, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF ALASKA, Anchorage, Alaska, for Amicus State of Alaska. Tim Griffin, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF ARKANSAS, Little Rock, Arkansas, for Amicus State of Arkansas. Ashley Moody, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF FLORIDA, Tallahassee, Florida, for Amicus State of Florida. Christopher M. Carr, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF GEORGIA, Atlanta, Georgia; Todd Rokita, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF INDIANA, Indianapolis, Indiana, for Amicus State of Indiana. Brenna Bird, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF IOWA, Des Moines, Iowa, for Amicus State of Iowa. Kris W. Kobach, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF KANSAS, Topeka, Kansas, for Amicus State of Kansas. Daniel Cameron, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF KENTUCKY, Frankfort, Kentucky, for Amicus State of Kentucky. Jeff Landry, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF LOUISIANA, Baton Rouge, Louisiana, for Amicus State of Louisiana. Lynn Fitch, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF MISSISSIPPI, Jackson, Mississippi, for Amicus State of Mississippi. Austin Knudsen, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF MONTANA, Helena, Montana, for Amicus State of Montana. Michael T. Hilgers, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF NEBRASKA, Lincoln, Nebraska, for Amicus State of Nebraska. Drew H. Wrigley, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF NORTH DAKOTA, Bismarck, North Dakota, for Amicus State of North Dakota. Dave Yost, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF OHIO, Columbus, Ohio, for Amicus State of Ohio. Gentner Drummond, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF OKLAHOMA, Oklahoma City, Oklahoma, for Amicus State of Oklahoma. Alan Wilson, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF SOUTH CAROLINA, Columbia, South Carolina, for Amicus State of South Carolina. Ken Paxton, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF TEXAS, Austin, Texas, for Amicus State of Texas. Sean D. Reyes, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF UTAH, Salt Lake City, Utah, for Amicus State of Utah. Jason S. Miyares, Attorney General, OFFICE OF THE ATTORNEY GENERAL OF VIRGINIA, Richmond, Virginia, for Amicus Commonwealth of Virginia.

8

GREGORY, Circuit Judge:

These two cases present the same question: Do healthcare plans that cover medically necessary treatments for certain diagnoses but bar coverage of those same medically necessary treatments for a diagnosis unique to transgender patients violate either the Equal Protection Clause or other provisions of federal law? We hold that they do, and therefore affirm the judgments of the district courts.

***

North Carolina provides healthcare coverage to state employees and their dependents through its state-operated insurance plan, the North Carolina State Health Plan for Teachers and State Employees (“the Plan”). Though all healthcare covered by the Plan is medically necessary, the Plan does not cover all medically necessary healthcare. At issue here is the Plan’s coverage exclusion of “[t]reatment or studies leading to or in connection with sex changes or modifications and related care.” Kadel, J.A. 181.

West Virginia’s Medicaid Program (“the Program”) covers some gender-affirming care, but not gender-affirming surgery, or, as the Program calls it, “[t]ranssexual surgery.” Anderson, J.A. 934–35. The Program does, however, cover the same surgical procedures when conducted to treat non-gender dysphoria diagnoses. For example, the Program covers mastectomies to treat cancer, but not to treat gender dysphoria; breast-reduction surgery to treat excess breast tissue in cisgender men, but not to treat gender dysphoria in transgender men; and chest-reconstruction surgery for cisgender women post-mastectomy, but not for gender dysphoria in transgender women. Anderson, J.A. 304, 2385–96, 2403–08, 2412–15.

9

Appellees in both cases are transgender individuals who were denied coverage for healthcare prescribed for their gender-dysphoria diagnoses. In North Carolina, Appellees are Plan members and dependents of Plan members. In West Virginia, Appellees are Medicaid beneficiaries. Both sets of Appellees say that the coverage exclusions violate their right to equal protection under the Fourteenth Amendment. So they sued the State Health Plan and Medicaid Program, respectively, as well as the state administrators in charge of those entities, to restore their rights, arguing that the coverage exclusions discriminate against them based on their sex and gender identity. The West Virginia Appellees also alleged violations of the Medicaid Act and Affordable Care Act.

The district courts in both cases agreed with Appellees. They granted summary judgment in Appellees’ favor and enjoined Appellants from enforcing the coverage exclusions. Both sets of Appellants appealed those decisions. The North Carolina Appellants also appealed certain evidentiary rulings underlying the district court’s judgment, and the West Virginia Appellants appealed the district court’s denial of their motion for summary judgment, as well as the district court’s certification of Appellees’ proposed class pursuant to Federal Rule of Civil Procedure 23. The North Carolina and West Virginia Appellants’ central argument is that the coverage exclusions do not discriminate against a suspect or quasi-suspect class and are rationally related to legitimate government interests. Because we hold that the coverage exclusions facially discriminate on the basis of sex and gender identity, and are not substantially related to an important government interest, we affirm the district courts. We further hold that the West Virginia exclusion violates the Medicaid Act and the Affordable Care Act.

10

I.

A. North Carolina’s Health Plan The North Carolina State Health Plan is part of the compensation package provided to state employees and the largest purchaser of healthcare and pharmaceuticals in North Carolina. J.A. 154.1 It funds healthcare for more than 740,000 teachers, legislators, state and local government employees, retirees, and their dependents. J.A. 160, 167. The Plan is administered by two third parties: BlueCross BlueShield North Carolina and CVS/Caremark. J.A. 156, 183. As third-party administrators, BlueCross and CVS process reimbursement claims from medical providers on behalf of Plan members. J.A. 184. The administrators do not, however, decide what benefits to cover. The State Health Plan alone does that. Id.

Each year, the State Health Plan publishes Plan Benefit Booklets that list the covered healthcare, as well as the coverage exclusions for healthcare it will not reimburse. J.A. 186. No procedural or diagnostic codes are assigned to the treatments listed in the booklet. 2 But to receive reimbursement, a healthcare provider must submit a claim with

1

Citations to the Joint Appendix and party briefs in this section and the next, I.A.

and B. (“North Carolina’s Health Plan” and “Kadel Procedural History”), refer to the Joint Appendix and briefs in the North Carolina case, Kadel. Citations in sections I.C. and D. (“West Virginia’s Medicaid Program” and “Anderson Procedural History”) refer to the Joint Appendix and briefs in the West Virginia case, Anderson. Each citation in II.A. (“Equal Protection”) specifies which Joint Appendix or brief it refers to. Citations in II.B. (“Evidentiary and Injunctive Challenges”) refer to the Kadel Joint Appendix and briefs. Citations in II.C, D., and E. (“Class Certification,” “Medicaid Act,” and “Affordable Care Act”) refer to the Anderson Joint Appendix and briefs.

2

The healthcare industry uses these alphanumeric codes to identify every possible diagnosis and medical service a patient might receive. See J.A. 185–87. Diagnostic codes classify diseases as provided by the ICD (“International Classification of Diseases”). J.A. (Continued)

11

both of these codes, J.A. 185, so BlueCross, in consultation with Plan staff, assigns codes to each of the benefits covered by the Plan, J.A. 186. When BlueCross receives a claim, its “automated claims systems review[] the claim to determine whether it is for a benefit covered by the Plan.” Id. If the medical treatment is a covered treatment, BlueCross authorizes reimbursement. Id. If the treatment is not covered, BlueCross does not authorize reimbursement. J.A. 188. 3 At issue is a coverage exclusion for “[t]reatment or studies leading to or in connection with sex changes or modifications and related care.” J.A. 181. Except for the 2017 calendar year, this exclusion has been in effect and administered by BlueCross each year since the 1990s. 4 Based on the exclusion, four procedures are not covered by the State Health Plan

185. Procedural codes, or CPT codes (“Current Procedural Terminology”), identify services and procedures. Id.

3

There are two exceptions to this. First, although the Plan theoretically excludes behavioral health services for treating gender dysphoria—either because they are treatments “leading to or in connection with sex changes or modifications” or are “related” to such treatments—BlueCross does not exclude these services. J.A. 191. That is because its automated system does not “distinguish between an individual diagnosed with gender dysphoria or another psychiatric diagnosis.” Id. Second, BlueCross has never implemented the benefit booklet’s exclusion of “surgery for psychological or emotion [sic] reasons” because it has no diagnostic or procedural codes for that (broad) category of surgery. Id.

4

The one-year change was in response to a 2016 final rule by the U.S. Department of Health and Human Services prohibiting “categorical coverage exclusion[s] or limitation[s] for all health services related to gender transition.” Nondiscrimination in Health Programs and Activities, 81 Fed. Reg. 31375, 31471–72 (May 18, 2016). To comply with the rule, the State Health Plan Board of Trustees voted to remove “the blanket exclusions that relate to treatment or studies leading to or in connection with sex changes or modifications and related care[,] and psychological assessment and psychotherapy treatment in conjunction with proposed gender transformation[,] resulting in the provision of medically necessary services for the treatment of gender dysphoria.” S.J.A. 4685, 4689 (Continued)

12

“regardless of the diagnostic code”: “Intersex Surgery, Male to Female,” “Intersex Surgery, Female to Male,” “Vaginoplasty for Intersex State,” and “Clitoroplasty for Intersex State.” J.A. 188–89. Roughly two dozen other procedures are not covered when the diagnostic code is for “Transsexualism” or “Personal history of sex reassignment.” J.A. 189–90.

North Carolina Appellees are members of the State Health Care Plan or their dependents. With the exception of next-friend Appellees, all Appellees are among the approximately 1.4 million people in the United States who identify as transgender. See generally J.A. 324–28, 342–46, 376–79, 389–93, 403–06; see also Brief of Amici Curiae the American Medical Association, et al., (Br. of Medical Amici) at 6. This means that their gender identity—that is, their deeply felt, inherent sense of their gender—is not aligned with their sex assigned at birth. This is in contrast to cisgender people’s gender identity, which does align with their sex assigned at birth. Id. at 9. Each Appellee (with the exception of next friends) has also been diagnosed with gender dysphoria, J.A. 324– 448, a condition characterized by clinically significant distress and anxiety resulting from the incongruence between an individual’s gender identity and birth-assigned sex, see Am. Psychiatric Ass’n, Diagnostic and Statistical Manual of Mental Disorders 451 (5th ed. 2013) (DSM-5). 5 “If untreated, gender dysphoria can cause debilitating distress,

(State Health Plan Board of Trustees Meeting Minutes, Dec. 2, 2016). The Board removed the exclusion only for 2017, S.J.A. 4690, and it went back into effect in 2018.

5

Federal Rule of Evidence 201(b) permits the Court to take judicial notice of a fact “not subject to reasonable dispute” either because it is (1) generally known or (2) capable of accurate and ready determination by resort to sources whose “accuracy cannot reasonably be questioned.” Both parties have cited to the DSM-5 for the definition of (Continued)

13

depression, impairment of function, self-mutilation to alter one’s genitals or secondary sex characteristics, other self-injurious behaviors, and suicide.” Br. of Medical Amici at 14 (citing DSM-5 at 455, 458). Although every patient with gender dysphoria requires care specific to their individual medical needs, id. at 17, the medical community uses generally accepted protocols from the Standards of Care for the Health of Transgender and Gender Diverse People (Version 8), https://perma.cc/8DMN-DN33 (last visited Nov. 29, 2023), developed by the World Professional Association for Transgender Health. Br. of Medical

gender dysphoria. See Opening Br. at 6–7; Resp. Br. at 12. The DSM-5 offers standardized criteria for the classification of mental disorders. It was published by the American Psychiatric Association after a twelve-year revision process in coordination with the National Institute of Mental Health (NIMH) and World Health Organization and a two- month public- and professional-review period. See Introduction, DSM-5. We therefore take judicial notice of the DSM-5. See Jacobs v. N.C. Admin. Office of the Courts, 780 F.3d 562, 565 n.2 (4th Cir. 2015) (taking judicial notice of the DSM–4 because the expert witnesses in that case applied the diagnostic criteria of the DSM–4); see also Williams v. Kincaid, 45 F.4th 759, 767–69 (4th Cir. 2022) (relying on the DSM-5 in determining that gender dysphoria is not a “gender identity disorder” under the ADA, which “reflected a significant shift in medical understanding”); United States v. Charboneau, 914 F.3d 906, 908 n.2 (4th Cir. 2019) (citing the DSM-5 for “paraphilic disorder” diagnosis criteria).

The North Carolina Appellants dispute the DSM-5’s reliability as a scientific authority given their expert testimony that the NIMH stopped funding projects that use the DSM-5 and that the DSM-5 is generally controversial. See J.A. 742, 764. But the director of NIMH issued a press release clarifying that “NIMH has not changed its position on DSM-5,” and that the DSM-5 still “represents the best information currently available for clinical diagnosis of mental disorders.” Sharon Jayson, NIH official clarifies criticism of diagnostic manual, USA Today, https://perma.cc/VU2L-MWZ8 (last visited Nov. 29, 2023). The NIMH’s research focus, he said, will be on a new system called Research Domain Criteria (RDoc), which will aim to find causes of disorders rather than focusing on symptoms. Id. Findings from RDoc may then be incorporated into future DSM revisions, he said. Id.

14

Amici at 15–16. These are known as the WPATH Standards. 6 To treat gender dysphoria, the WPATH Standards recommend “assessment, counseling, and, as appropriate, social transition, hormone therapy, and surgical interventions to bring the body into alignment with one’s gender identity.” Id. at 16. Appellees sought many of these treatments but were denied coverage based on the Plan’s exclusion.

6

The North Carolina Appellants dispute the scientific validity of these standards and the district court’s reliance on amici (which, in turn, heavily rely on the Standards) for incorporation of facts outside the record. See J.A. 788, 863 (defense expert declarations that WPATH’s recommendations are not scientifically based).

But nothing about Appellants’ experts’ criticisms undermines the consensus around WPATH’s recommendations that gender dysphoria treatments may include surgery and hormone therapy. Appellants’ experts question the methodology, but not the consensus it has garnered. Compare, e.g., J.A. 863 (Defense expert Dr. McHugh criticizing WPATH for using “consensus-seeking methodologies, including voting”), with S.J.A. 4298 (Plaintiff expert Dr. Schechter explaining that voting in medical societies is a means for experts to voice their scientific opinions rather than one figurehead making a top-down decision). As Dr. George Brown put it, “WPATH Standards of Care . . . have been recognized as the authoritative treatment protocols by the major medical and mental health associations in the United States.” J.A. 3567. “The Veterans Health Administration []—the largest integrated health care system in the United States—treats transgender veterans largely based on the guidelines set forth in the current version of the WPATH [Standards].” Id. In fact, BlueCross’s default policy (the policy BlueCross uses when contracting with organizations that do not make their own coverage decisions) requires patients seeking medically necessary treatments for gender dysphoria to, among other things, provide a letter from the patient’s established healthcare provider indicating whether the provider follows the WPATH Standards and/or is part of a gender identity dysphoria treatment team. S.J.A. 4706–14. Given the record and the fact that “amici often make useful contributions to litigation,” Stuart v. Huff, 706 F.3d 345, 355 (4th Cir. 2013), we reject Appellants’ contentions. See Grimm v. Gloucester County Sch. Bd., 972 F.3d 586, 594–96 (4th Cir. 2020) (citing to substantially same amici for the proposition that WPATH promulgates “modern accepted treatment protocols for gender dysphoria”); Peters v. Aetna, Inc., 2 F.4th 199, 234 (4th Cir. 2021) (concluding that “the brief of amici, the American Medical Association,” bolstered the Court’s interpretation of “Network Provider” under a health plan).

15

B. Kadel Procedural History The North Carolina Appellees sued the Executive Administrator of the State Health Plan and State Treasurer Dale Folwell for their roles in the administration of the Plan. Appellees also sued the Plan itself. 7 J.A. 47, 51, 53. They alleged violations of the Equal Protection Clause and Affordable Care Act. The Plan moved to dismiss, asserting that it was entitled to sovereign immunity under the Eleventh Amendment. The district court denied that motion, holding that the Plan waived its immunity by accepting federal financial assistance. We affirmed on appeal. See Kadel v. N.C. State Health Plan, 12 F.4th 422, 426 (4th Cir. 2021).

The parties then cross-moved for summary judgment. Appellees also moved to exclude Appellants’ expert testimony. As relevant to this appeal, the district court granted Appellees’ motion for summary judgment on their Equal Protection Claim and granted partial relief to Appellees on their motions to exclude evidence. J.A. 3701–13, 3674–99. 8 The district court reserved judgment on the Affordable Care Act claims. J.A. 3726–27. 9 On the Equal Protection claim, the district court concluded that the Plan’s coverage exclusion facially discriminates based on sex and transgender status and, therefore, must

7

Appellees also sued three public universities in North Carolina and the North Carolina Department of Public Safety. J.A. 52–53. For reasons not relevant here, none of these defendants is a party to the current appeal.

8

The district court’s opinion can be found at 620 F. Supp. 339 (M.D.N.C. 2022).

9

The Court reserved judgment pending resolution of Administrative Procedure Act challenges to a revised rule from the Department of Health and Human Services and expected changes to that rule by the Biden administration.

16

withstand intermediate scrutiny. It found no real dispute that the Plan’s exclusion is not substantially related to important government interests. Appellants raised two justifications: cost and efficacy. The district court readily dismissed the first reason because fiscal justifications cannot withstand intermediate scrutiny. J.A. 3710 (citing Mem’l Hosp. v. Maricopa County, 415 U.S. 250, 263 (1974)). And, although the court agreed that “[t]he state has an obvious interest in protecting its employees and their families from ineffective medical treatments,” it said the record did not support the notion that the treatments were actually ineffective. J.A. 3710–11. The court enjoined Appellants from enforcing the coverage exclusion and ordered them to reinstate coverage for “medically necessary services for the treatment of gender dysphoria.” J.A. 3734.

The district court also granted in part and denied in part Appellees’ motions to exclude Appellants’ expert testimony. As a general matter, the district court excluded all of Appellants’ expert evidence that appeared to be based on unreliable methodology. J.A. 3685–86. The district court also rejected theories about the “Transgender Treatment Industry” as “speculation designed to distract or inflame the jury.” J.A. 3694. It also excluded testimony from Appellants’ experts opining on areas of medicine and science in which they had no specific experience or expertise. J.A. 3690–92.

At the same time, the district court held that testimony about issues directly within the experts’ professional purviews was admissible. For instance, the court found that Dr. Patrick Lappert, a surgeon, was qualified to opine on the risks associated with surgery used to treat gender dysphoria. J.A. 3692–93. It also found that Dr. Stephen Levine, a physician and professor of psychiatry, was qualified as a mental health provider and

17

researcher to testify to “the treatment of gender dysphoria and the efficacy and findings of research studies evaluating gender dysphoria treatments.” J.A. 3697.

Of Appellants’ five proposed experts, four were allowed to testify. Proposed testimony from the fifth expert, Dr. Peter Robie, was excluded because half of it was not expert testimony and the other half was irrelevant, the district court said. J.A. 3678–79.

C. West Virginia’s Medicaid Program Medicaid is a federal-state program that provides health insurance for low-income people. J.A. 2562–63; 42 U.S.C. § 1396-1. Though states are not required to participate in Medicaid, “once a state elects to join the program, it must administer a state plan that meets federal requirements.” Frew ex rel. Frew v. Hawkins, 540 U.S. 431, 433 (2004). To ensure state compliance, each state must submit a written state plan for approval by the Secretary of Health and Human Services. 42 U.S.C. §§ 1396a, 1396c. The plan must describe the nature and scope of the state’s program and affirm the state’s commitment to adhere to the requirements of the Medicaid Act and its associated regulations. 42 C.F.R. § 430.10. It “consists of a standardized template, issued and updated by CMS [the Centers for Medicaid and Medicare Services], that includes both basic requirements” common to every state and “individualized content that reflects the characteristics of the State’s program.” 42 C.F.R. § 430.12(a).

For “categorically needy” populations, 10 states must cover certain basic categories of services and may cover other optional categories of services. 42 C.F.R. § 440.210

10

Medicaid distinguishes between “categorically needy” and “medically needy”

populations. States must cover the categorically needy and may cover the medically needy. See Medicaid Eligibility, https://perma.cc/C4LC-64MY (last visited Nov. 29, 2023). The (Continued)

18

(listing mandatory services). Some mandatory categories of services are inpatient hospital services; outpatient hospital services; laboratory and X-ray services; nursing facility services; early and periodic screening, diagnostic, and treatment services for people under 21; family-planning services for people of child-bearing age; and physicians’ services. 42 U.S.C. § 1396d(a)(1)–(5). Optional service categories for adults include physical therapy and prescription drugs. 42 U.S.C. §§ 1396a(a)(10)(A), 1396d(a)(11), (12).

The West Virginia Department of Health and Human Resources, Bureau of Medical Services, administers the state’s Medicaid Program and receives funding from the U.S. Department of Health and Human Services. J.A. 2564. The Department’s Cabinet Secretary and the Bureau’s Commissioner, both of whom are named defendants, are responsible for ensuring that the Program complies with federal law. J.A. 2564–65. The Bureau Commissioner is also responsible for administering the Program. J.A. 2565.

Like the North Carolina State Health Plan, West Virginia’s Medicaid Program does not cover every medically necessary procedure, but every procedure it covers is medically necessary. J.A. 458. To determine what is medically necessary, the state contracts with a company called Kepro. J.A. 2567. Kepro, in turn, relies on InterQual, which establishes nationally accredited criteria that insurers use to make coverage decisions. J.A. 2567.

categorically needy are those who are eligible for certain federal welfare programs, those who are not eligible for those programs but whose income falls below a certain level, and other distinct groups (for example, qualifying pregnant women). See 42 U.S.C. § 1396a(10)(A). The medically needy are people with significant health needs whose incomes are too high to otherwise qualify for Medicaid but who will spend enough money on medical care that their income after medical costs falls below a certain threshold. See Medicaid Eligibility, https://perma.cc/C4LC-64MY. The distinction between the two groups is not relevant to this case.

19

InterQual criteria are derived from the “systematic continuous review and critical appraisal of the most current evidence based literature” and include input from an independent panel of experts. J.A. 573. To develop its guidance on treatments for gender dysphoria, InterQual relies on guidelines from the World Professional Association of Transgender Health (WPATH) and the Endocrine Society. J.A. 2567.

The state’s Medicaid Program covers some gender-affirming care, including counseling, office visits, hormones, and lab work. J.A. 1136–37; Opening Br. at 3; Resp. Br. at 8. 11 It does not, however, cover gender-affirming surgery. Specifically, Medicaid’s Policy Manual lists twenty-one services it does not cover, including “transsexual surgery.” J.A. 934–35. Such surgery is excluded “regardless of medical necessity,” J.A. 459—a relevant caveat because, under InterQual’s criteria, surgery to treat gender dysphoria is medically necessary for certain individuals. J.A. 2143–58; see also J.A. 459 (Deposition of BMS Commissioner Cynthia Beane) (testifying that the Program does not cover gender-affirming surgery “regardless of whether or not there’s a physician or a review team saying it’s medically necessary”). The coverage exclusion was adopted around 2004 and has been maintained since without review. J.A. 473–74, 2564. Appellants admit they do not know why it was adopted, nor are they aware of what information, if any, the Program relied on in adopting the exclusion. See J.A. 1127, 2212.

11

The district court did not make a finding about whether Medicaid covers these types of gender-affirming care, but both parties agree that it does and the Joint Appendix supports the same conclusion.

20

Medicaid contracts with three managed care organizations to provide coverage. J.A.

2564. Its contract with each says the organization is “not permitted to provide” certain services, including “[s]ex transformation procedures and hormone therapy associated with sex transformation procedures.” J.A. 1040–41. As a result, each organization’s own member handbook explicitly states that those services are not covered. J.A. 947, 953, 958. While the Medicaid Program does not follow InterQual’s coverage criteria for what Medicaid refers to as “transsexual surgery,” it does follow the criteria for the same surgeries when they are not performed to treat gender dysphoria. Specifically, the Program partially or fully covers the following procedures for non-gender dysphoria diagnoses: mastectomy (removal of breast tissue), breast-reduction surgery, post-mastectomy chest-reconstruction surgery, hysterectomy (removal of uterus), oophorectomy (removal of ovaries), vaginoplasty (creation or repair of vagina), orchiectomy (removal of testicles), penectomy (removal of penis), and phalloplasty (creation or reconstruction of penis). J.A. 304. 12

12

InterQual deems mastectomy/reduction mammoplasty medically necessary in certain cases of macromastia/gigantomastia (a medical condition where the breasts of patients assigned female at birth become excessively large) and gynecomastia (enlarged breasts in patients assigned male at birth). J.A. 2397–2406. It deems hysterectomy or salpingo-oophorectomy (removal of fallopian tubes and ovaries) or salpingectomy (removal of fallopian tubes) medically necessary in certain cases of endometriosis (uterine-like tissue growing outside the uterus), endometrial cancer, presence of the BRCA gene, cervical adenocarcinoma in situ (a premalignant precursor to cervical cancer), postmenopausal endometrial hyperplasia (thickening of the uterine lining that can lead to uterine cancer), Lynch syndrome (a genetic condition that increases the risks of certain types of cancers, including endometrial cancer), suspected ovarian or tubal cancer, abnormal uterine bleeding or postmenopausal bleeding, adenomyosis (endometrial tissue growing into the muscular walls of the uterus), fibroids, chronic abdominal or pelvic pain, and cervical dysplasia (abnormal cell growth in the cervix). Id. 2351–2415.

21

D. Anderson Procedural History Plaintiff Shauntae Anderson is a transgender Medicaid patient who has been diagnosed with gender dysphoria. J.A. 289–90, 294–96. She is seeking gender-affirming surgery, specifically breast augmentation and vaginoplasty. J.A. 296. Before she was on Medicaid, she began medically transitioning through self-treatment: taking birth control pills for estrogen. J.A. 294. Once she was on Medicaid, her doctors recommended hormone replacement therapy. J.A. 295. She began the therapy in 2019. Id. Still, she struggles with her body. Id. at 295–96. She also worries about her safety in public, where strangers have mocked her for being transgender. J.A. 296. She is concerned that future interactions will escalate to violence. J.A. 296–97.

Doctors have not yet recommended her for surgery; to the extent they have discussed it with her, they have simply said that Medicaid does not cover the surgeries, so “there is nothing that they can do about it.” Id.

Anderson sued in the Southern District of West Virginia on behalf of herself and others similarly situated. 13 She argued that the coverage exclusion discriminates against transgender people in violation of the Fourteenth Amendment, the Medicaid Act, and the Affordable Care Act, and sought class certification. Both parties moved for summary judgment. Appellants also argued that Appellees lacked standing and opposed class certification. The district court found in favor of Appellees on all claims, J.A. 2562–91,

13

Christopher Fain also sued as a named plaintiff. Between oral argument and publication of this opinion, his income made him ineligible to participate in West Virginia’s Medicaid Program. His individual claims in this case are therefore now moot, but Anderson and the class members still have standing.

22

and certified a class of “all transgender people who are or will be enrolled in West Virginia Medicaid and who are seeking or will seek gender-confirming care barred by the Exclusion,” J.A. 2552. 14 The court enjoined Appellants from enforcing or applying the exclusion. J.A. 2592.

II.

The central dispute in this case is about the fate of the coverage exclusions. 15 Appellants in both cases ask us to reverse the district courts’ summary judgment rulings that the exclusions violate the Fourteenth Amendment. We review that decision de novo. See Bostic v. Schaefer, 760 F.3d 352, 370 (4th Cir. 2014). And we will affirm a summary judgment ruling only if we find “no genuine dispute as to any material fact” after considering the evidence in the light most favorable to the nonmovant. Ret. Comm. of DAK Ams. LLC v. Brewer, 867 F.3d 471, 479 (4th Cir. 2017) (quoting Fed. R. Civ. P. 56(a)); Libertarian Party of Va. v. Judd, 718 F.3d 308, 312–13 (4th Cir. 2013).

14

The district court’s opinion can be found at 618 F. Supp. 3d 313 (S.D.W. Va. 2022).

15

We can quickly dispose of the Anderson Appellants’ argument that Anderson lacks standing. Appellants say she cannot demonstrate an actual, concrete injury because she did not submit claims to Medicaid for gender-affirming surgery, and therefore has not shown that her claims would be denied. Opening Br. at 49. Submitting a claim for a procedure that the policy manual explicitly excludes from coverage would be futile and is therefore not required to show standing. See Townes v. Jarvis, 577 F.3d 543, n.1 (4th Cir. 2009). Appellants’ argument that Anderson’s injury is speculative because she is not yet in a position to undergo surgery is similarly unconvincing. Though Anderson has not yet sought formal approval for surgery from a physician, Supp. J.A. 1–2, which would be required if the exclusion were lifted, doing so would be futile. See, e.g., Pinchback v. Armistead Homes Corp., 907 F.2d 1447, 1451 (4th Cir. 1990) (plaintiff not required to apply for a job that company’s racially discriminatory policy would bar him from getting).

23

A. Equal Protection

The Equal Protection Clause of the Fourteenth Amendment forbids a state from denying “to any person within its jurisdiction the equal protection of the laws.” U.S. Const. amend. XIV, § 1, cl. 4. It prohibits states from placing people into different classes and treating them unequally for reasons “wholly unrelated” to permissible government objectives. Reed v. Reed, 404 U.S. 71, 75–76 (1971). If the state does seek to treat different groups of people differently, it must do so “upon some ground of difference having a fair and substantial relation to the object of the [policy].” Id. at 76 (quoting Royster Guano Co. v. Virginia, 253 U.S. 412, 415 (1920)).

Appellants argue that the district courts’ equal-protection analyses were flawed because, they say, the exclusions distinguish on the basis of diagnosis. The exclusions therefore only have to withstand rational-basis review. We disagree. In this case, discriminating on the basis of diagnosis is discriminating on the basis of gender identity and sex. The coverage exclusions are therefore subject to intermediate scrutiny. They cannot meet that heightened standard.

1.

We start by determining the proper level of scrutiny with which to review the coverage exclusions. When a state law regulates on the basis of something other than a protected characteristic, we apply rational-basis review and will uphold the law if it rationally relates to a legitimate government objective. City of Cleburne v. Cleburne Living Ctr., 473 U.S. 432, 440 (1985). When the state draws distinctions based on a protected classification, however, a more searching review is required. Classifications along racial

24

lines, for example, are inherently suspect and subject to strict scrutiny. Id. Classifications based on sex are also suspect but are subject to intermediate, or “quasi-suspect,” scrutiny. Grimm v. Gloucester County Sch. Bd., 972 F.3d 586, 607–08 (4th Cir. 2020).

The distinction between rational basis and intermediate scrutiny is significant. We have described rational-basis review as a “deferential” standard under which “the plaintiff bears the burden to negate every conceivable basis which might support” the differential treatment. Giarratano v. Johnson, 521 F.3d 298, 303 (4th Cir. 2008) (citation and quotations omitted). By contrast, an intermediate-scrutiny analysis requires the proponent of the policy to produce an “exceedingly persuasive justification” for treating individuals differently based on quasi-suspect characteristics. Miss. Univ. for Women v. Hogan, 458 U.S. 718, 724 (1982) (citation and quotations omitted). The district courts considered the coverage exclusions under intermediate scrutiny because they viewed the exclusions as facially discriminating on the basis of sex and gender identity. Because Appellants dispute that conclusion, we consider the question anew.

The central disagreement between the parties is whether the exclusion discriminates on the basis of diagnosis and procedure (Appellants’ view) or on the basis of sex and transgender identity (Appellees’ view). As a reminder, the exclusions respectively bar coverage of “[t]reatment or studies leading to or in connection with sex changes or modifications and related care,” Kadel, J.A. 181 (North Carolina), and “transsexual surgery,” Anderson, J.A. 2566 (West Virginia). Appellants argue that this language is facially neutral because it simply excludes treatments for gender dysphoria; it does not bar transgender patients from receiving the same treatments as cisgender patients. In fact,

25

Appellants argue, treatments for gender dysphoria—all treatments in North Carolina and surgical treatments in West Virginia—are excluded from coverage for everyone, regardless of their gender identity. Appellees argue that the language of the exclusions is facially discriminatory because it makes coverage for certain procedures hinge on the sex of the patient and bars coverage of treatments for a condition that is bound up in transgender identity (gender dysphoria).

The parties spend much of their briefs arguing over the meanings of “surgery,”

“procedure,” and “treatment.” Is a procedure defined by the diagnosis it treats or simply by what happens in the operating room? Is removing a patient’s breasts to treat cancer the same procedure as removing a patient’s breasts to treat gender dysphoria? Is testosterone therapy to address “hypogonadotropic hypogonadism” (“a lack of sex hormones, . . . prevent[ing] normal sexual maturity in children and normal function of the testicles or ovaries in adults” 16) the same treatment as testosterone therapy to address gender dysphoria? There is no caselaw to ground this discussion nor obvious first principles to work from.

Instead, we answer the following questions, starting from the premise that gender identity is a protected characteristic, see Grimm, 972 F.3d at 610: (1) Is gender dysphoria a proxy for transgender identity?, (2) Can proxy discrimination be facial discrimination?,

16

Hypogonadotropic Hypogonadism, MedlinePlus, https://perma.cc/A4V2-6WLU (last visited Dec. 17, 2023); see also familial hypogonadotropic hypogonadism, Stedman’s Medical Dictionary, 428990 (Westlaw 2014) (defining the term as characterized by failure of sexual development, owing to inadequate secretion of pituitary gonadotropins). “MedlinePlus is a service of the National Library of Medicine (NLM), the world’s largest medical library, which is part of the National Institutes of Health (NIH).” About MedlinePlus, MedlinePlus, https://perma.cc/S75C-J939 (last visited Dec. 17, 2023).

26

and (3) In this case, is discrimination on the basis of gender dysphoria discrimination on the basis of gender identity? We also address whether the coverage exclusions discriminate on the basis of sex. We answer each of these questions in the affirmative.

a.

We begin by reiterating our holding in Grimm v. Gloucester County School Board that gender identity is a protected characteristic under the Equal Protection Clause. The school board in that case passed a policy limiting the use of boys’ and girls’ restrooms to students with “the corresponding biological genders.” 972 F.3d at 599. Because of that policy, Gavin Grimm, a transgender boy, was barred from using the boys’ restroom. Grimm sued on Equal Protection grounds, as well as Title IX grounds, claiming that the policy discriminated on the basis of sex and gender identity. Id. at 593. In addressing Grimm’s gender-identity argument, this Court had to decide whether gender identity is a protected characteristic subject to heightened scrutiny. The Court applied the Supreme Court’s four factors for determining whether a group of people constitutes a suspect or quasi-suspect class. Id. at 611. It found that transgender people have historically been subjected to discrimination, transgender status “bears [no] relation to ability to perform or contribute to society,” transgender people are a discrete group with immutable characteristics, and transgender people are a minority lacking political power. Id. at 611– 13 (quotations and citation omitted). Because transgender people constitute a quasi- suspect class, the Grimm Court held, discrimination on the basis of gender identity is subject to heightened scrutiny. Id. at 613. If the coverage exclusions here discriminate

27

against transgender people, they must withstand that scrutiny to stay in place. We next address whether the exclusions discriminate against transgender individuals.

b.

The coverage exclusions do not explicitly mention transgender people. Instead, they mention the types of treatments that are not covered: “[t]reatment or studies leading to or in connection with sex changes or modifications and related care,” Kadel, J.A. 181 (North Carolina), and “transsexual surgery,” Anderson, J.A. 2566 (West Virginia). In other words, treatments for gender dysphoria. Appellees argue that targeting gender dysphoria is targeting the people with gender dysphoria, all of whom are, by definition, transgender. Appellants argue that gender dysphoria is not a proxy for transgender identity. They make two arguments: (1) not all transgender people have gender dysphoria, and (2) the policies apply to everyone, not just transgender people. We address these arguments in turn.

i.

Not all transgender people are diagnosed with gender dysphoria. 17 And not all people with gender dysphoria seek gender-affirming surgery, as the West Virginia Appellants note. Anderson, Opening Br. at 7–8. But “a law is not immune to an equal protection challenge if it discriminates only against some members of a protected class but not others.” Hecox v. Little, 79 F.4th 1009, 1025 (9th Cir. 2023). In Rice v. Cayetano, for instance, the Supreme Court struck down a Hawaiian constitutional provision that allowed

17

As North Carolina Appellees’ counsel noted at oral argument, transgender people without gender dysphoria may not suffer from gender dysphoria because they were treated for it. Oral Arg. at 1:15:57–1:16:24.

28

people to vote in certain elections only if they were descendants of aboriginal people who inhabited Hawaii in or before 1778, the year the British made landfall in Hawaii. 528 U.S. 495, 498–500 (2000). Hawaii was settled by Polynesians who voyaged from Tahiti. Id. at 500. The island was isolated from migration, so the aboriginal people living in Hawaii in 1778 were all Polynesian. See id. at 514. Their descendants were therefore at least part Polynesian. The state argued that the law was not racially discriminatory against non-Polynesians because not all Polynesians were allowed to vote; those who had come after 1778 could not. Id. at 516. The Court rejected that argument: “Simply because a class defined by ancestry does not include all members of the race does not suffice to make the classification race neutral.” Id. at 516–17 (emphasis added). 18 Indeed, the Court has consistently taken the view that discrimination within a certain class does not mean there is no discrimination between classes. See Mathews v. Lucas, 427 U.S. 495, 504 n.11 (1976) (“That the statutory classifications challenged here discriminate among illegitimate children does not mean, of course, that they are not also properly described as discriminating between legitimate and illegitimate children.”) (emphasis added); Weber v. Aetna Cas. & Sur. Co., 406 U.S. 164, 167–68 (1972) (striking down workers’ compensation law that allowed recovery for, among others, children born to married parents, as well as children born to unmarried parents and acknowledged by their biological fathers, but did not allow recovery for children born to unmarried parents and unacknowledged by their biological fathers); Frontiero v. Richardson, 411 U.S. 677, 678–

18

Rice was decided under the Fifteenth Amendment because it barred a certain group from voting. But its analysis is just as relevant to the Equal Protection context.

29

79, 690–91 (1973) (invalidating rule that servicemen could claim their wives as dependents, and servicewomen providing at least half of their husbands’ financial support could claim their husbands as dependents, but servicewomen providing less than half of their husbands’ financial support could not claim their husbands as dependents); Nyquist v. Mauclet, 432 U.S. 1, 3–4, 12 (1977) (striking down state law that offered tuition assistance to citizens and to non-citizens who had applied to become citizens or submitted statement affirming intent to apply for citizenship or were refugees, but did not offer tuition assistance to other non- citizens); Graham v. Richardson, 403 U.S. 365, 366–67, 376 (1971) (invalidating law that provided state welfare benefits to U.S. citizens and non-citizens who had been in the state for at least 15 years, but not to non-citizens who had been in the state for less than 15 years). 19 Geduldig v. Aiello, on which Appellants heavily rely, does not alter the meaning of these cases. 417 U.S. 484 (1974). Nor could it—both Mathews and Nyquist were decided after Geduldig. The Court in Geduldig dealt with a California disability insurance system that compensated workers for “disability stemming from a substantial number of mental or physical illness(es) and mental or physical injur(ies).” Id. at 488. Certain disabilities were not covered:

19

At least one other circuit has applied these holdings to recognize that a law need not affect every transgender person to discriminate against transgender people as a class. In Hecox v. Little, the Ninth Circuit preliminarily enjoined an Idaho law that barred student athletes assigned male at birth from competing on girls’ and women’s sports teams. 79 F.4th 1009. Defendant-appellants argued that the Act did not discriminate on the basis of transgender identity because it did “not prohibit biologically female athletes who identify as male from competing on male sports teams consistent with their gender identity.” Id. at 1025. In other words, although the Act prohibited transgender girls and women from participating on girls’ and women’s sports teams, it did not prohibit transgender boys and men from participating on boys’ and men’s sports teams. The Court found this argument unconvincing for the same reason we find Appellants’ argument unconvincing here. Id. at 1025–26.

30

disability lasting less than eight days, unless the employee was hospitalized; disability that resulted from someone’s court commitment as a “dipsomaniac” (someone who struggles with alcohol addiction), “drug addict,” or “sexual psychopath”; and disability resulting from “normal” pregnancy. Id. at 486, 488, 490. Plaintiffs sued, arguing that the plan’s pregnancy exclusion discriminated against women. The Court rejected this argument in a footnote:

The California insurance program does not exclude anyone from benefit eligibility because of gender but merely removes one physical condition—

pregnancy—from the list of compensable disabilities. While it is true that only women can become pregnant[,] it does not follow that every legislative classification concerning pregnancy is a sex-based classification . . . . Normal pregnancy is an objectively identifiable physical condition with unique characteristics. Absent a showing that distinctions involving pregnancy are mere pretexts designed to effect an invidious discrimination against the members of one sex or the other, lawmakers are constitutionally free to include or exclude pregnancy from the coverage of legislation such as this on any reasonable basis, just as with respect to any other physical condition.

The lack of identity between the excluded disability and gender as such under this insurance program becomes clear upon the most cursory analysis. The program divides potential recipients into two groups—pregnant women and nonpregnant persons. While the first group is exclusively female, the second includes members of both sexes. The fiscal and actuarial benefits of the program thus accrue to members of both sexes.

Geduldig, 417 U.S. at n.20.

The West Virginia Appellants say that, like in Geduldig, Medicaid’s policy “does not create a sex-based classification, because it divides members into two groups—those who seek gender-confirming surgery, and all other persons. While the first group may be exclusively comprised of transgender individuals, the second group includes all other

31

persons, whether cisgender, transgender, or other identity, who do not seek gender-confirming surgery.” Anderson, Opening Br. at 28. 20 Appellants’ argument—that Geduldig compels us to find in their favor because not all transgender individuals seek treatment for gender dysphoria—might be correct if we read Geduldig as broadly as possible. But Geduldig must be read in light of Mathews, Weber, Frontiero, Nyquist, Graham, and Rice, all of which say that a state cannot immunize itself from violating the Equal Protection Clause by discriminating against only a subset of a protected group. Appellants’ reading of Geduldig cannot be squared with these cases. Read in conjunction with these cases, Geduldig is best understood as standing for the simple proposition that pregnancy is an insufficiently close proxy for sex. The same cannot be said for the inextricable categories of gender dysphoria and transgender status.

Three facts support this conclusion. First, the Supreme Court has only relied on Geduldig to reject proxy-based arguments in cases where pregnancy was at issue. See Gen. Elec. Co. v. Gilbert, 429 U.S. 125, 136 (1976) (holding that a similar benefits exclusion for pregnancy-related disability did not violate Title VII and reiterating that “exclusion of pregnancy from a disability-benefits plan providing general coverage is not a gender-based discrimination at all”), superseded by statute, Pregnancy Discrimination Act of 1978, Pub.

20

As explained below, II.A.1.b.ii.B., II.A.1.d., the policy does cover certain kinds of gender-affirming surgery for cisgender people, so it is inaccurate to say that the people seeking gender-affirming surgery are exclusively transgender. We assume for purposes of analysis that Appellants intend to compare gender-affirming surgery sought by transgender people with surgery unrelated to gender and sought by people of all gender identities—for instance, a mastectomy sought by a transgender man for gender-affirming purposes versus mastectomies sought by cisgender women and transgender men to treat cancer.

32

L. No. 95-555, 92 Stat. 2077, as recognized in Newport News Shipbuilding & Dry Dock Co. v. EEOC, 462 U.S. 669, 685 (1983); Nashville Gas Co. v. Satty, 434 U.S. 136, 142 (1977) (holding, in part, that a policy denying sick-leave pay to pregnant employees was permissible under Title VII, so long as the policy was not a pretext for invidious discrimination); Bray v. Alexandria Women’s Health Clinic, 506 U.S. 263, 271 (1993) (“‘While it is true,’ we said [in Geduldig], ‘that only women can become pregnant, it does not follow that every legislative classification concerning pregnancy is a sex-based classification.’”). Thus, while Geduldig held that pregnancy is not a proxy for sex, it did not hold that a characteristic of a subset of a protected group cannot be a proxy for that group.

Second, gender dysphoria is so intimately related to transgender status as to be virtually indistinguishable from it. The excluded treatments aim at addressing incongruity between sex assigned at birth and gender identity, the very heart of transgender status. In contrast to pregnancy—which is a condition that can be described entirely separately from a person’s sex—gender dysphoria is simply the medical term relied on to refer to the clinical distress that can result from transgender status.

Finally, the exclusions cannot function without relying on direct—not just proxy-

based—discrimination. Determining whether someone requires pregnancy-related treatment—the issue in Geduldig—does not turn on or require inquiry into a protected characteristic. True, when a doctor determines a person is pregnant, they will generally, as a consequence, also have reached a conclusion about the person’s sex assigned at birth. But that is true only because, as Geduldig recognized, pregnancy is often a reliable indicator of a person’s sex. In contrast, determining whether a treatment like reduction mammoplasty

33

constitutes “transsexual surgery” or whether a testosterone supplement is prescribed in connection with a “sex change[] or modification[]” is impossible—literally cannot be done— without inquiring into a patient’s sex assigned at birth and comparing it to their gender identity. Indeed, those procedures are routinely covered by the Plan and Program in situations where the only material difference is the patient’s sex.

For those reasons, Appellants’ arguments that Geduldig requires us to find in their favor is unpersuasive.

ii.

A.

Appellants next argue that gender dysphoria is not being used as a proxy for transgender identity here because treatment for that diagnosis is not covered for anyone, transgender or cisgender. This argument elides common sense and is inconsistent with Supreme Court precedent about how to approach equal-protection analyses. “The proper focus of constitutional inquiry is the group for whom the law is a restriction, not the group for whom the law is irrelevant.” City of Los Angeles v. Patel, 576 U.S. 409, 418 (2015) (quotations omitted).

The argument is also tautological, akin to saying that the law “applies equally to all to whom it applies.” See Joseph Tussman & Jacobus tenBroek, The Equal Protection of the Laws, 37 Calif. L. Rev. 341, 345 (1949); see also Giovanna Shay, Similarly Situated, 18 Geo. Mason L. Rev. 581, 587 (2011). The Supreme Court explicitly rejected this line of reasoning in McLaughlin v. Florida, where it struck down a ban on interracial couples living together. 379 U.S. 184 (1964). In doing so, it overturned a prior case, Pace v. Alabama. Id. at 188–91. In Pace, it held that people convicted of violating two separate

34

laws—one prohibiting sex outside of marriage and the other prohibiting sex outside of marriage specifically for interracial couples—were not similarly situated because the same- race couple had committed a different offense than the interracial couple. 106 U.S. 583, 585 (1883). Rejecting Pace’s cramped approach, the McLaughlin Court wrote:

The [Pace v. Alabama] opinion acknowledged that the purpose of the Equal Protection Clause “was to prevent hostile and discriminating state legislation against any person or class of persons” and that equality of protection under the laws implies that any person, “whatever his race . . . shall not be subjected, for the same offense, to any greater or different punishment.” But taking quite literally its own words, “for the same offense,” the Court pointed out that Alabama had designated as a separate offense the commission by a white person and a Negro of the identical acts forbidden by the general provisions. There was, therefore, no impermissible discrimination because the difference in punishment was “directed against the offence designated” and because in the case of each offense all who committed it, white and Negro, were treated alike . . . . Because each of the Alabama laws applied equally to those to whom it was applicable, the different treatment accorded interracial and intraracial couples was irrelevant.

Id. at 188–90 (emphasis added) (quoting Pace, 106 U.S. at 584–85).

This “narrow view” of the Equal Protection Clause—that a law does not discriminate if it applies equally to all—made no sense, the Court said. Id.; see also Loving v. Virginia, 388 U.S. 1, 10 (1967) (noting rejection of Pace). Indeed, the analysis collapses in on itself. Take other examples. A tax on wearing kippot would apply to non-Jews and Jews alike, but would affect only Jews. See Bray, 506 U.S. at 270. A ban on same-sex marriage would apply to straight, gay, lesbian, and bisexual people equally, but would affect only gay, lesbian, and bisexual people—straight people would not choose to marry someone of the same sex. Finally, a literacy test only required of people whose ancestors were not allowed to vote before 1866 would apply to everyone, but would affect only Black

35

people. See Guinn v. United States, 238 U.S. 347, 364–65 (1915). Put differently, all these barriers or bans, although they do not use the words “Jews,” “gays, lesbians, or bisexuals,” or “Black people,” targeted these groups by proxy, which is just as impermissible under the Equal Protection Clause. See Christian Legal Soc’y v. Martinez, 561 U.S. 661, 689 (2010) (“Our decisions have declined to distinguish between status and conduct in th[e] context [of discrimination].”); see also Lawrence v. Texas, 539 U.S. 558, 575 (2003) (“When homosexual conduct is made criminal by the law of the State, that declaration in and of itself is an invitation to subject homosexual persons to discrimination.”); id. at 583 (O’Connor, J., concurring in judgment) (“While it is true that the law applies only to conduct, the conduct targeted by this law is conduct that is closely correlated with being homosexual. Under such circumstances, [the] law is targeted at more than conduct. It is instead directed toward gay persons as a class.”); Bray, 506 U.S. at 270 (“Some activities may be such an irrational object of disfavor that, if they are targeted, and if they also happen to be engaged in exclusively or predominantly by a particular class of people, an intent to disfavor that class can readily be presumed.”).

B.

Just as both Appellants claim that they do not provide certain types of gender-dysphoria treatment to anyone, the West Virginia Appellants claim that they do not provide gender-affirming treatment to anyone. Anderson, Opening Br. at 21–24. This argument fails for two reasons: (1) for many procedures, it is not true, and (2) for those procedures of which it is true, the coverage ban only affects transgender people.

36

First, the record shows that cisgender people do receive coverage for certain gender-affirming surgeries, specifically vaginoplasty (for congenital absence of a vagina), breast reconstruction (post-mastectomy), and breast reduction (for gynecomastia). Anderson, J.A. 304, 332, 2385–87, 2418–27.

Second, the gender-affirming surgeries that are not covered for anyone are surgeries that only transgender people would get; they are either not physically possible for other groups or would not be gender-affirming for them. Specifically, any surgeries involving removing genitals or internal parts of the body are not covered when performed for gender- affirming purposes. So neither a cisgender woman nor cisgender man would be entitled to a hysterectomy, oophorectomy, vaginectomy, orchiectomy, or penectomy for gender- affirming purposes. Appellants argue that this fact shows that the Program does not discriminate against transgender people.

This is just another version of Appellants’ “applies equally to all to whom it applies”

argument. Anderson, Opening Br. at 6, 21; Reply Br. at 6–7; Kadel, Opening Br. at 23; Reply Br. at 16. Just as cisgender people would not seek any treatment for gender dysphoria, they would not seek certain surgeries for gender-affirming purposes. For instance, a cisgender woman would never seek a hysterectomy, oophorectomy, or vaginectomy for gender-affirming reasons because, for her, those surgeries are not gender-affirming. Nor would a cisgender man ever seek an orchiectomy or penectomy for gender-affirming reasons because, for him, those surgeries are not gender-affirming. Again, while the exclusion may apply to everyone, for many treatments, it is only relevant to transgender individuals.

37

In sum, targeting a subset of a protected group does not preclude a finding of proxy discrimination. Nor does the fact that a law applies equally to all, when it only affects a protected group. We hold that gender dysphoria, a diagnosis inextricable from transgender status, is a proxy for transgender identity. And coverage exclusions that bar treatments for gender dysphoria bar treatments on the basis of transgender identity by proxy.

c.

We next address whether proxy discrimination can be a form of facial discrimination. At oral argument, the North Carolina Appellants argued that we only ask whether a trait is being used as a proxy once we have found that a law does not facially discriminate. Oral Arg. at 1:31:30–1:31:40, 1:32:48–1:36:48. In other words, they say, we would have to look beyond the face of the exclusions to find that gender dysphoria was being used as a proxy for gender identity. Id. at 1:35:50–1:36:08. Because Appellees only advance a facial-discrimination theory, and not an invidious-intent theory, proxy discrimination does not enter our analysis, Appellants say.

This argument about how to approach proxy discrimination has significant practical implications. If a plaintiff needs discovery about extratextual factors—say, a legislator’s intent—to argue that a statute is using a proxy to discriminate, that plaintiff will rarely make it past a motion to dismiss. 21 Government officials who pass discriminatory policies

21

Appellees made it past the motion-to-dismiss stage here because the district courts recognized that discrimination on the basis of gender dysphoria is discrimination on the basis of gender identity. Kadel, J.A. 3706 (“Discrimination against individuals suffering from gender dysphoria is also discrimination based on sex and transgender status.); Anderson, J.A. 2573 (“[I]nherent in a gender dysphoria diagnosis is a person’s identity as (Continued)

38

(generally) do not say the quiet part out loud. So, under Appellants’ view, a plaintiff bringing an Equal Protection claim would be left with only two avenues to get to summary judgment: the statute itself must explicitly name the protected group in its text or a government official must let slip the real purpose of the policy. Both virtually never happen.

This approach also cannot be squared with Supreme Court precedent. The Court has consistently found the text of statutes and constitutions, coupled with basic facts, enough to find facial discrimination, even when the text does not explicitly name a protected group. In Guinn v. United States, for example, the Court invalidated an Oklahoma law that required all voters to pass a literacy test, except those whose ancestors were eligible to vote in or before 1866 (i.e., before the Fifteenth Amendment was passed). 238 U.S. at 364–65. “It is true [the law] contains no express words of an exclusion . . . on account of race, color, or previous condition of servitude,” the Court said. Id. at 364. “[B]ut the [1866] standard itself inherently brings that result into existence since it is based purely upon a period of time before the enactment of the 15th Amendment, and makes that period the controlling and dominant test of the right of suffrage.” Id. at 364–65. The Court reaffirmed that proxy discrimination can be facial discrimination in Bray (a tax on kippot is a tax on Jews), Christian Legal Society (exclusion based on same-sex conduct is exclusion based on sexual orientation), and Lawrence (criminalization of same-sex conduct is discrimination based on sexual orientation). 506 U.S. at 270; 561 U.S. at 689; 539 U.S. at 575. And while Rice v. Cayetano, explained above, canvassed the legislative history of

transgender. In other words, a person cannot suffer from gender dysphoria without identifying as transgender.”).

39

the constitutional provision, 528 U.S. at 509–10, 515–16, the Court’s ruling did not hinge on this evidence of invidious discrimination. It was enough to know the history of the island—including the importance of 1778—to conclude that the provision used ancestry as a proxy for race. 22 Similarly, it is enough to know that gender dysphoria, and therefore treatment for gender dysphoria, is unique to transgender individuals in order to conclude that the exclusions use gender dysphoria as a proxy for transgender identity. 23

22

The principal dissent says that, because we hold that proxy discrimination can be established solely through the text of a law, coupled with basic facts, we also hold that evidence of invidious discrimination isn’t necessary. Dissent Op. at n.3. Not so. Evidence of discriminatory intent is always necessary. But just as text alone can be enough to show that intent (e.g., women cannot receive heart transplants), text coupled with basic facts can also be enough (people with XX chromosomes cannot receive heart transplants).

23

At least one other circuit has addressed proxy discrimination as a form of facial discrimination in other contexts. In Hecox v. Little, the Ninth Circuit preliminarily enjoined an Idaho law that barred student athletes assigned male at birth from competing on women’s sports teams, see supra at n.199. The Court noted that the Act’s definition of “biological sex” was written with “seemingly neutral criteria that are so closely associated with the disfavored group that discrimination on the basis of such criteria is, constructively, facial discrimination against the disfavored group [transgender athletes].” 79 F.4th at 1024 (quoting Pac. Shores Props., LLC v. City of Newport Beach, 730 F.3d 1142, 1160 n.23 (9th Cir. 2013)). The Ninth Circuit reached the same conclusion in Davis v. Guam, 932 F.3d 822, 839 (9th Cir. 2019). Guam law restricted voting rights to “those persons who became U.S. Citizens by virtue of the authority and enactment of the 1950 Organic Act of Guam and descendants of those persons”—a seemingly innocuous definition on its own. Id. The Organic Act granted citizenship to three categories of people, all of whom had to have been born in Guam before April 11, 1899—(not) coincidentally, the date that Spain ceded Guam to U.S. control. Id. The Ninth Circuit held that the law facially discriminated by using qualification under the Organic Act as a proxy for race. Id.; see also id. at 837–38 (noting that discriminating against individuals with gray hair would be facial discrimination on the basis of age because “the fit between age and gray hair is sufficiently close”) (citation and quotations omitted). The Seventh Circuit, in another context, has noted that policies excluding service dogs and wheelchairs would “no doubt” discriminate on the basis of disability. See McWright v. Alexander, 982 F.2d 222, 228 (7th Cir. 1992).

40

The principal dissent sees the discrimination analysis differently. To begin, we agree with the principal dissent that, as a default rule, we do not presume discriminatory intent from a facially neutral statute. Because of that, plaintiffs claiming that a facially neutral statute violates the Equal Protection Clause must conduct a more searching evidentiary inquiry. See Vill. of Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252, 266–68 (1977). We agree, however, that when a statute is very clearly using a proxy to target a protected characteristic, we need not conduct a full-blown Arlington Heights inquiry. See Dissent Op. at 75 (“Sometimes a law uses a classification that is so obviously a proxy for a suspect class that ‘an intent to disfavor that class can be readily presumed.’” (quoting Bray, 506 U.S. at 270)). From there, we diverge.

The first difference between our approaches is where in the analysis we ask about proxy discrimination. 24 As stated above, usually, if we have determined that something is facially neutral, we cannot find discriminatory intent without first conducting a more searching evidentiary inquiry. In the principal dissent’s view, there is an exception to that rule. Even if something is facially neutral, we can find discriminatory intent without an evidentiary inquiry if there is incredibly clear proxy discrimination. But, the principal dissent says, the proxy inquiry is never part of the facial classification inquiry. Id. at n.4. That is where we disagree.

24

The question about proxy discrimination—whether procedures are being used as such an obvious proxy for a protected characteristic that the policies cannot be facially neutral—is only relevant to whether the text of the policies discriminate on the basis of gender identity. As we explain below, the text of the policies discriminates on the basis of sex, even without using a procedure as a proxy.

41

In the principal dissent’s view:

That we must ask whether a law uses a classification that is merely a substitute for a protected trait means that the law does not explicitly—i.e., facially—

classify based on that protected trait. Cf. Hunt v. Cromartie, 526 U.S. 541, 546 (1999) (“When . . . classifications are explicit, no inquiry into legislative purpose is necessary.”). Indeed, the Supreme Court has expressly rejected this argument, noting that courts only inquire into “covert” classifications—i.e., ostensibly neutral classifications that “could not be plausibly explained on a neutral ground”—after concluding that a statute is “gender-neutral on its face.”

Feeney, 442 U.S. at 274–75. Instead, the proxy inquiry is better understood as a species of intentional discrimination: Is the government targeting something because of its close connection to another thing?

Id.

We agree that the covert-classifications inquiry only happens after concluding that a statute is facially neutral. But it doesn’t follow that the proxy inquiry only happens after concluding that a statute is facially neutral. That’s because not every proxy is covert. Indeed, some are glaringly—facially—obvious. And when that’s the case, when there is incredibly clear proxy discrimination, the law is not facially neutral. In Califano v. Westcott, for example, the Court held that a welfare law that differentiated between unemployed fathers and unemployed mothers facially discriminated on the basis of sex. 443 U.S. 76, 83–89 (1979); see also Sessions v. Morales-Santana, 582 U.S. 47, 58 (2017) (holding that law that gave children born abroad to unwed U.S. citizen mother and non-citizen father an easier path to U.S. citizenship than children born abroad to unwed non-citizen mother and U.S. citizen father facially discriminated on the basis of gender); Reed v. Reed, 404 U.S. 71, 74–76 (1971) (holding facially discriminatory a state law that gave preference to fathers to act as administrators of their deceased child’s estate). The Court has similarly treated laws differentiating between wives and husbands as facial discrimination on the basis of sex. See,

42

e.g., Orr v. Orr, 440 U.S. 268, 278 (1979) (law providing alimony for wives but not husbands facially discriminatory). The principal dissent acknowledges that the laws in these cases did not use the words “men” or “women.” Dissent Op. at n.1. Still, it says, the laws were nevertheless facial classifications because “discriminating between mothers and fathers is just another way of discriminating ‘on the basis of the sex of the qualifying parent.’” Id. (quoting Morales-Santana, 582 U.S. at 58).

That is exactly our point. A law is not facially neutral simply because, in place of explicit references to protected identities, the law uses different words that mean the same thing. This case is a good example. Had the West Virginia and North Carolina policies barred “surgical procedures or treatments related to a patient’s transgender status,” the policies would no doubt discriminate on the basis of gender identity. Rewording the policies to use a proxy, by barring “[t]reatment or studies leading to or in connection with sex changes or modifications and related care” and “transsexual surgery,” does not make the classification covert. The policies remain just as obviously discriminatory as before. Our definition of facial discrimination is thus broader than the principal dissent’s.

This brings us to the second difference between our approaches: what makes a proxy obvious. The principal dissent says a proxy is obvious when plaintiffs can show “both discriminatory effects and that no rational, nondiscriminatory explanation exists for the law’s classification.” Dissent Op. at 77.

There are two problems with these criteria. First, they assume that the presence of a nondiscriminatory reason means the absence of a discriminatory reason. But “[r]arely can it be said that a legislature or administrative body operating under a broad mandate

43

made a decision motivated solely by a single concern.” Vill. of Arlington Heights, 429 U.S. at 265. The question therefore is not whether there is a non-discriminatory reason for a policy, but instead whether there is a discriminatory reason for it. When there is “proof that a discriminatory purpose has been a motivating factor in the decision, [] judicial deference is no longer justified.” Id. at 265–66 (emphasis added).

Especially where government budgets are involved, there will frequently be a “rational” basis for discrimination. A law that pays state employees with XX chromosomes 75 percent of what state employees with XY chromosomes are paid has a rational, nondiscriminatory reason: it saves the state large sums of money. But under the principal dissent’s framework, not only would that law be facially neutral; it would also be supported by a “rational, nondiscriminatory reason.” A court therefore could not find that the law discriminated on the basis of gender until it conducted a full-blown Arlington Heights evidentiary inquiry. This would require us to ignore the obvious.

Second, the principal dissent’s “no rational, nondiscriminatory explanation” criteria would muddle the traditional equal-protection analysis. The second step of that analysis asks whether a discriminatory law can be justified by the state’s nondiscriminatory interest in the law. The principal dissent’s analysis would require asking the state-interest question twice: first to determine whether a facially neutral law is nevertheless discriminatory 25 and second to determine whether a discriminatory law can nevertheless be justified.

25

Of course, and as the principal dissent notes, even if the state had a rational, nondiscriminatory interest in the law, plaintiffs might be able to show discrimination through the Arlington Heights factors.

44

d.

In addition to discriminating on the basis of gender identity, the exclusions discriminate on the basis of sex. Certain gender-affirming surgeries that could be provided to people assigned male at birth and people assigned female at birth are provided to only one group under the policy. Those surgeries include vaginoplasty (for congenital absence of a vagina), breast reconstruction (post-mastectomy), and breast reduction (for gynecomastia). Anderson, J.A. 304, 332, 2385–87, 2418–27. Those assigned female at birth can receive vaginoplasty and breast reconstruction for gender-affirming purposes, but those assigned male at birth cannot. And those assigned male at birth can receive a mastectomy for gender-affirming purposes, 26 but those assigned female at birth cannot. In other words, when the purpose of the surgery is to align a patient’s gender presentation with their sex assigned at birth, the surgery is covered. When the purpose is to align a patient’s gender presentation with a gender identity that does not match their sex assigned at birth, the surgery is not covered.

This is textbook sex discrimination, for two reasons. For one, we can determine whether some patients will be eliminated from candidacy for these surgeries solely from knowing their sex assigned at birth. And two, conditioning access to these surgeries based on a patient’s sex

26

Appellants note that cisgender men with excess breast tissue (gynecomastia) can only have a covered mastectomy if they also experience breast pain or tenderness. Anderson, Reply Br. at 21–22 (citing J.A. 2405). But it is not clear why this is relevant. It seems this is an argument that the two are not similarly situated because one surgery (mastectomy for cisgender men with symptomatic gynecomastia) is medically necessary, while the other (mastectomy for transgender men with gender dysphoria) is not. As explained below, though, there is no threshold similarly situated inquiry in the equal- protection analysis.

45

assigned at birth stems from gender stereotypes about how men or women should present. See Bostock v. Clayton County, 590 U.S. ----, ----, 140 S. Ct. 1731, 1742–49 (2020).

First, as the North Carolina district court noted, the policy cannot be applied “without referencing sex.” Kadel, J.A. 3704 (quoting Grimm, 972 F.3d at 608). Try figuring out whether the State Health Plan or Medicaid Program will cover a certain patient’s vaginoplasty. By virtue of the fact that they are seeking a vaginoplasty, we know that they were born without a vagina. But we do not know what sex they were assigned at birth. Without that information, we cannot say whether the Plan or Program will cover the surgery.

The Supreme Court used this type of thought experiment in Bostock v. Clayton County.

There, it imagined a job applicant asked to disclose the applicant’s sexual orientation. 140 S. Ct. at 1746. “There is no way for an applicant to decide whether to check the homosexual or transgender box without considering sex,” it wrote. Id. “To see why, imagine an applicant doesn’t know what the words homosexual or transgender mean. Then try writing out instructions for who should check the box without using the words man, woman, or sex (or some synonym). It can’t be done.” Id. The same is true here. A third-party administrator cannot make the coverage decision without knowing whether the vaginoplasty is to treat gender dysphoria—in other words, whether the patient was assigned male at birth.

Second, a policy that conditions access to gender-affirming surgery on whether the surgery will better align the patient’s gender presentation with their sex assigned at birth is a policy based on gender stereotypes. For instance, while mastectomies are available for both people assigned male at birth and those assigned female at birth, when they are conducted for gender-affirming purposes, they are only available to those assigned male at birth. This

46

difference in coverage is rooted in a gender stereotype: the assumption that people who have been assigned female at birth are supposed to have breasts, and that people assigned male at birth are not. No doubt, the majority of those assigned female at birth have breasts, and the majority of those assigned male at birth do not. But we cannot mistake what is for what must be. And because gender stereotypes can be so ingrained, we must be particularly careful in order to keep them out of our Equal Protection jurisprudence. “[T]he test for determining the validity of a gender-based classification is straightforward, [but] it must be applied free of fixed notions concerning the roles and abilities of males and female.” Miss. Univ. for Women, 458 U.S. at 724–25. Policies based on gender stereotypes impermissibly discriminate on the basis of sex. See Price Waterhouse v. Hopkins, 490 U.S. 228, 255–58 (1989), superseded by statute on other grounds, 42 U.S.C. §§ 2000e-2(m), 2000e-5(g)(2)(B), as recognized in Comcast Corp. v. Nat’l Ass’n of Afr. Am.-Owned Media, 589 U.S. ----, ----, 14 S. Ct. 1009, 1017 (2020); Grimm, 972 F.3d at 608–09 (noting that sex stereotyping is sex discrimination under the Equal Protection Clause); Peltier v. Charter Day Sch., Inc., 37 F.4th 104, 124–26 (4th Cir. 2022) (en banc) (“[W]e will reject sex-based classifications that appear to rest on nothing more than conventional notions about . . . males and females.”) (citation and quotations omitted). Because the exclusions here condition access to certain surgeries on whether those surgeries will better align the patient’s sex assigned at birth with their gender, they discriminate on the basis of sex.

e.

Having addressed Appellants’ two primary arguments—that the exclusions discriminate on the basis of diagnosis and not gender identity or sex, and that Geduldig mandates this finding—we move on to Appellants’ other equal-protection arguments.

47

Both Appellants argue that the district courts incorrectly determined that the two groups at issue are similarly situated. Anderson, Opening Br. at 20–21; Kadel, Opening Br. at 32. Appellants define the groups as those seeking surgery for non-gender dysphoria diagnoses and those seeking surgery for gender-dysphoria diagnoses. Appellants call this similarly situated analysis a “foundational requirement.” Kadel, Opening Br. at 32. If the court finds that two groups are not similarly situated, the equal-protection analysis goes no further, they say.

Appellants misunderstand the similarly situated directive. Far from a threshold step, the similarly situated inquiry is “one and the same as the equal protection merits inquiry.” Shay, supra, at 598. As the Court in City of Cleburne v. Cleburne Living Center said, “The Equal Protection Clause,” not the first step of an Equal Protection Clause analysis, “is essentially a direction that all persons similarly situated should be treated alike.” 473 U.S. at 439. The “similarly situated” language preceded the modern tiers of scrutiny, and the Court has continued to use the phrase. Shay, supra, at 598. But it has never used it as a threshold hurdle. Id.; see, e.g., Brown v. Bd. of Educ., 347 U.S. 483, n.10 (1954) (mentioning the phrase in a footnote near the end of the opinion); Loving v. Virginia, 388 U.S. 1 (1967) (never mentioned); United States v. Virginia, 518 U.S. 515 (1996) (never mentioned); Grutter v. Bolinger, 539 U.S. 306, 375 (2003) (mentioned in concurrence and not as a threshold inquiry). It has instead used the similarly situated inquiry to decide whether the governmental interest for discrimination is justified. See, e.g., Tuan Anh Nguyen v. INS, 533 U.S. 53, 62–68 (2001) (asking whether the fact that biological mother and father are not similarly situated with regard to proof of biological parenthood justifies state’s different citizenship rules for children born abroad and to unmarried parents, depending on whether the citizen parent is the mother or father).

48

This makes sense. The similarly situated inquiry does not just ask whether two groups are similarly situated; it asks whether they are similarly situated with respect to the statute’s objective. See, e.g., Reed, 404 U.S. at 77 (“The objective of [the statute] clearly is to establish degrees of entitlement of various classes of persons in accordance with their varying degrees and kinds of relationship to the intestate. Regardless of their sex, persons within any one of the enumerated classes of that section are similarly situated with respect to that objective.”) (emphasis added); Stanton v. Stanton, 421 U.S. 7, 14 (1975) (“A classification . . . must rest upon some ground of difference having a fair and substantial relation to the object of legislation, so that all persons similarly circumstanced shall be treated alike.” (emphasis added) (quotations and citation omitted)). But the modern equal- protection analysis does not reach a statute’s objectives until after determining whether it discriminates on the basis of a protected characteristic. Adding a threshold similarly situated inquiry confuses the proper sequence of the analysis.

Next, the North Carolina Appellants argue that there is a genuine dispute about whether the healthcare sought by Appellees is medically necessary. Kadel, Opening Br. at 32–33. They frame this as a similarly situated argument: those seeking gender-dysphoria treatment may not be similarly situated to those not seeking it. Putting aside that there is no similarly situated threshold inquiry, this argument ignores the coverage exclusion’s language. The North Carolina exclusion prohibits treatment “leading to or in connection with sex changes or modifications and related care,” irrespective of medical necessity. To the extent Appellants are arguing that treatments for gender dysphoria are never medically necessary, that argument is better understood as a back-end justification for the facial discrimination rather than an ex ante

49

argument that it is not subject to heightened scrutiny.27 Because the exclusions discriminate on the basis of transgender identity and sex, they are subject to intermediate scrutiny.

2.

Having determined that the challenged coverage exclusions receive intermediate scrutiny, we now turn to whether the coverage exclusions can withstand that scrutiny. To survive intermediate-scrutiny review, the government must provide an “exceedingly persuasive justification” for the classification. Miss. Univ. for Women, 458 U.S. at 724. At a minimum, the government must show that “the classification serves important governmental objectives and that the discriminatory means employed are substantially related to the achievement of those objectives.” Id. (citation and quotations omitted). A law that discriminates against a quasi-suspect class “must be genuine, not hypothesized or invented post hoc in response to litigation.” Virginia, 518 U.S. at 533. “And it must not rely on overbroad generalizations about the different talents, capacities, or preferences of males and females.” Id.

“Under intermediate scrutiny, the government bears the burden of establishing a reasonable fit between the challenged statute and a substantial governmental objective.” United States v. Chapman, 666 F.3d 220, 226 (4th Cir. 2012). The party defending the statute must “present[] sufficient probative evidence in support of its stated rationale for enacting a

27

Appellants’ remaining arguments—that the coverage exclusion does not constitute facial discrimination simply because it contains the word “sex” and that the district court improperly relied on Title VII precedents—also fail. The phrase “sex change” is not merely descriptive; it forecloses medical coverage based on a patient’s choice to diverge from sex stereotypes, which, as explained above, constitutes sex discrimination. Moreover, the district court properly rooted its analysis in Grimm, an Equal Protection Clause case.

50

gender preference, i.e., . . . the evidence [must be] sufficient to show that the preference rests on evidence-informed analysis rather than on stereotypical generalizations.” H.B. Rowe Co. v. Tippett, 615 F.3d 233, 242 (4th Cir. 2010) (citation and quotations omitted). The classification must be based on “reasoned analysis rather than [on] the mechanical application of traditional, often inaccurate, assumptions.” Miss. Univ. for Women, 458 U.S. at 726.

The North Carolina Appellants say that they excluded gender-dysphoria treatments because the treatments cost too much and were not effective. The first justification is a non- starter. “[A] state may not protect the public fisc by drawing an invidious distinction between classes of its citizens.” Mem’l Hosp., 415 U.S. at 263.

Protecting public health from ineffective medicine is an important government interest, as the North Carolina district court noted. Kadel, J.A. 3710–11. But the district court properly rejected the contention that the coverage exclusion is substantially related to that end. Some of the expert testimony that Appellants rely on to argue that gender- dysphoria treatments are ineffective does not actually support their argument. Kadel, J.A. 3712. And the anecdotal evidence that does call into question medical efficacy challenges only some of Appellees’ evidence. In any event, those criticisms do not support the notion that gender-dysphoria treatments are ineffective so much as still developing. That alone does not create a genuine dispute that is material to the heightened-scrutiny analysis. Without evidence to show that gender-dysphoria treatments are ineffective, the North Carolina Appellants cannot show that the coverage exclusion is narrowly tailored to serve the state’s substantial interest in not covering medically ineffective treatment.

51

The West Virginia Appellants also argue that saving costs and not covering medically ineffective treatments justify the exclusion. Anderson, Opening Br. at 33–35. Their arguments are even weaker than the North Carolina Appellants’ arguments. CMS Commissioner Cynthia Beane testified that she did not know why the exclusion was adopted; in fact, she was not even sure when it was adopted. Anderson, J.A. 436–37. What’s more, West Virginia Department of Health and Human Resources Secretary Bill Crouch said he did not know if Medicaid had conducted any research or analysis about the cost of providing access to gender-affirming care. Anderson, J.A. 393. That testimony shows that Appellants’ proffered rationales were created for the purposes of litigation. They therefore cannot justify the policy under a heightened-scrutiny analysis. See Virginia, 518 U.S. at 533.

B. Evidentiary and Injunctive Challenges We now address the Kadel Appellants’ remaining complaints about the district court’s order. First, they argue that the district court impermissibly relied on facts in an amicus brief filed by medical organizations. Second, they challenge the district court’s exclusion of certain expert testimony. And third, they say the district court’s injunction order was too vague for them to comply with absent risking a contempt sanction.

1.

Appellants challenge the district court’s reliance on an amicus brief filed by eight medical organizations. 28 J.A. 3539–59. Appellants’ central contention is that the district

These are the same organizations who have filed an amicus brief with this Court

28

on appeal: the American Medical Association, the American Academy of Pediatrics, American College of Obstetricians and Gynecologists, the American Psychiatric (Continued)

52

court used the amicus brief to establish evidence contrary to Appellants’ expert testimony that “no reliable medical studies show that plaintiffs’ desired treatments . . . improve the health and wellbeing of patients with gender dysphoria over time.” Opening Br. at 47 (citing J.A. 3698, n.3, Declaration of defense expert Stephen B. Levine).

This argument is unpersuasive. The district court relied on the amicus brief to anchor its discussion in well-accepted facts about what it means to be transgender, how transgender people may be affected by gender dysphoria, and what treatments exist to mitigate the symptoms of gender dysphoria. See J.A. 3669–71 (e.g., “[w]hile being transgender is not itself a psychiatric condition, many transgender individuals experience severe anxiety and distress as a result of having physiology or an assigned sex that does not match their deeply felt, inherent sense of their gender” and “[t]he current Standards of Care (WPATH-7) recommended treatments include[] assessment, counseling, and, as appropriate, social transition, hormone therapy, and surgical interventions” (citations and quotations omitted)). None of this information contradicts any party’s proffered testimony. To the contrary, the district court clearly laid out the dispute between the parties in the final paragraphs of the section. Compare J.A. 3671 (“Plaintiffs’ experts testify that . . . these are ‘safe and effective treatment[s] for gender dysphoria’ that are governed by ‘well- established community standards.’”), with id. (“Defendants’ experts . . . testify that medical and surgical treatments have significant medical risks and consequences, and the research

Association, the Endocrine Society, the North American Society for Pediatric and Adolescent Gynecology, National Association of Nurse Practitioners in Women’s Health, and the Society of OB/GYN Hospitalists.

53

supporting such treatments is of ‘low quality.’”). The district court did not improperly rely on amicus briefing.

2.

The North Carolina Appellants also claim that the district court misapplied Federal Rule of Evidence 702 when it rejected portions of their expert witnesses’ proffered evidence. We review the district court’s rulings on this matter for abuse of discretion. McKiver v. Murphy-Brown, LLC, 980 F.3d 937, 958 (4th Cir. 2020). Rule 702 sets forth the requirements a witness must satisfy to qualify as an expert. When determining the reliability of experiential expert testimony for purposes of Rule 702, a court must require the witness to “explain how [the witness’s] experience leads to the conclusion reached, why that experience is a sufficient basis for the opinion, and how that experience is reliably applied to the facts.” Fed. R. Evid. 702 advisory committee’s note to 2000 amendments.

“A witness who is qualified as an expert by knowledge, skill, experience, training, or education may testify in the form of an opinion” if (1) the expert’s specialized knowledge will help the jury understand the evidence or determine a fact in issue; (2) the testimony is based on sufficient facts or data; (3) the testimony comes from reliable principles and methods; and (4) the expert reliably applied those principles and methods to the facts of the case. Fed. R. Evid. 702. A witness’s qualifications are “liberally judged by Rule 702,” and “a person may qualify to render expert testimony in any one of the five ways listed” by the Rule. Kopf v. Skyrm, 993 F.2d 374, 377 (4th Cir. 1993); see Cooper v. Lab’y Corp. of Am. Holdings, Inc., 150 F.3d 376, 380 (4th Cir. 1998).

54

Appellants see three problems in the district court’s analysis: the district court (1)“artificially constrained the ‘technical area’ of the substantive issues at hand,” Opening Br. at 52–53; (2) diminished Dr. Hruz’s credentials “with untrue statements” and erroneously excluded “his testimony about the treatment of gender dysphoria,” id. at 55; and (3) misunderstood the relevance of Dr. Robie’s testimony that “there is no such thing as a gender-neutral diagnosis or gender-neutral medicine,” id. at 56. None of these arguments points to an abuse of discretion.

First, the district court constrained Appellants’ experts to the specific technical areas in which they had expertise because that is what Fourth Circuit precedent requires. In undertaking its gatekeeping role to ensure that evidence is reliable under Rule 702, a district court “must decide whether the expert has ‘sufficient specialized knowledge to assist the jurors in deciding the particular issues in the case.’” Belk, Inc. v. Meyer Corp., U.S., 679 F.3d 146, 162 (4th Cir. 2012) (quoting Kumho Tire Co. v. Carmichael, 526 U.S. 137, 156 (1999) (emphasis added)). The district court did not abuse its discretion by rejecting expert testimony about the treatment of gender dysphoria from witnesses who, although medical professionals, did not demonstrate an expertise in treating gender dysphoria. See, e.g., Zellers v. NexTech Ne., LLC, 533 F. App’x 192, 199 (4th Cir. 2013) (finding that a Ph.D.-holding neuropsychologist and neurotoxicologist was not a medical doctor and was thus “not qualified to diagnose the cause of [plaintiff’s] alleged symptoms”); see also Shreve v. Sears, Roebuck & Co., 166 F. Supp. 2d 378, 391 (D. Md. 2001) (stating “[t]he fact that a proposed witness is an expert in one area, does not ipso facto qualify him to testify as an expert in all related areas” and collecting cases supporting that proposition).

55

Second, Dr. Hruz’s testimony that he has overseen two medical fellows who performed research on gender dysphoria does nothing to contradict the district court’s conclusion that Dr. Hruz himself has not conducted research on gender dysphoria. J.A. 3681–82. The district court’s point was that “[m]erely reading literature” about gender dysphoria does not qualify Dr. Hruz as an expert on the subject. J.A. 3682. And, notwithstanding its finding, the district court still allowed Dr. Hruz “to testify to the risks associated with puberty blocking medication and hormone therapy,” based on his “long career treating patients and conducting academic research on the effects of hormone treatments.” Id. The district court’s decision to exclude a portion of Dr. Hruz’s testimony did not “rest[] upon a clearly erroneous factual finding.” Bryte ex rel. Bryte v. Am. Household, Inc., 429 F.3d 469, 475 (4th Cir. 2005).

Third, Appellants’ argument that the district court misunderstood the relevance of Dr. Robie’s opinion that “physicians must know the chromosomal sex of patients” to provide competent medical care, J.A. 3500, fails to address the district court’s other, independent basis for rejecting it. The district court held that, in addition to the testimony being irrelevant, “Robie’s failure to submit an expert report or provide any basis for his opinion other than a vague reference to his years of practice precludes this Court from finding that his expert opinion is based on a reliable methodology under Rule 702.” J.A. 3679; see Sardis v. Overhead Door Corp., 10 F.4th 268, 281 (4th Cir. 2021) (requiring district courts “to ensur[e] that an expert’s testimony both rests on a reliable foundation and is relevant to the task at hand”). Because Appellants did not present competent support for the reliability of Dr. Robie’s testimony at the district court or on appeal, we are

56

unpersuaded by their argument. See Cooper v. Smith & Nephew, Inc., 259 F.3d 194, 199 (4th Cir. 2001) (“The proponent of [expert] testimony must establish its admissibility by a preponderance of proof.”).

Finding no abuse of discretion in the district court’s evidentiary rulings, we reject Appellants’ claims on this issue.

3.

Finally, the North Carolina Appellants argue that the injunctive order’s language enjoining them from “enforcing the Plan’s exclusion,” J.A. 3734, was too vague for them to comply with, without risking a contempt sanction. Opening Br. at 39.

Federal Rule of Civil Procedure 65(d)(1) requires courts granting an injunction to “(A) state the reasons why it issued; (B) state its terms specifically; and (C) describe in reasonable detail . . . the act or acts restrained or required.” The rule “was designed to prevent uncertainty and confusion on the part of those faced with injunctive orders, and to avoid the possible founding of a contempt citation on a decree too vague to be understood.” Schmidt v. Lessard, 414 U.S. 473, 476 (1974). Considering these goals, “the specificity provisions of Rule 65(d) are no mere technical requirements,” and “basic fairness requires that those enjoined receive explicit notice of precisely what conduct is outlawed.” Id. To comply with Rule 65(d), the district court’s order must be clear enough to inform Appellants of what they may and may not do. See also CPC Int’l, Inc. v. Skippy Inc., 214 F.3d 456, 459 (4th Cir. 2000).

The injunction order contains two components. First, it permanently enjoins Appellants “from enforcing the Plan’s exclusion.” J.A. 3734. Appellants’ position that

57

this language does not indicate the specific coverage exclusion they are enjoined from enforcing is baseless. The injunction order represents the culmination of a detailed, 73-page opinion discussing why the coverage exclusion prohibiting treatments leading to or in connection with sex changes violates the Equal Protection Clause. Plainly speaking, the district court’s injunction refers to the only coverage exclusion at issue in the case. See, e.g., Ciena Corp. v. Jarrard, 203 F.3d 312, 315–16, 322 (4th Cir. 2000) (relying on the plain meaning of the language in an injunction in affirming its requirements).

Second, the injunction order requires Appellants “to reinstate coverage for ‘medically necessary services for the treatment of gender dysphoria.’” J.A. 3734. Appellants again inject vagueness into this command by stripping it of its context. In its decision, the district court specified that it was “reimposing the 2017 rule” that covered “medically necessary services for the treatment of gender dysphoria.” J.A. 3729. We find nothing vague about this. Appellants understood the meaning of “medically necessary services for the treatment of gender dysphoria” well enough in 2017 to implement it without incident that year. They can do it again now.

We reject Appellants’ claims that the district court’s injunction falls below the standards required by the Federal Rules of Civil Procedure.

C. Class Certification

We now move on to the Anderson Appellants’ other arguments, beginning with their challenge to the district court’s class certification. The district court certified a class of “all transgender people who are or will be enrolled in West Virginia Medicaid and who are seeking or will seek gender-confirming care barred by the Exclusion.” J.A. 2552.

58

Appellants say this class definition does not satisfy Rule 23(b)’s numerosity requirement, but what they are actually arguing is that the definition does not meet a threshold ascertainability requirement. See Opening Br. at 50–54. Specifically, Appellants claim that of the definition’s three criteria—1) transgender, 2) is or will be enrolled in Medicaid, and 3) is seeking or will seek gender-affirming care—only the Medicaid-enrollment criterion is objective and therefore ascertainable. Opening Br. at 51–52.

This Circuit, and many others, have recognized an implicit requirement in 23(b)(1)

and 23(b)(3) cases that members of a proposed class be “readily identifiable.” EQT Prod. Co. v. Adair, 764 F.3d 347, 358 (4th Cir. 2014); see also Marcus v. BMW of N. Am., LLC, 687 F.3d 583, 593–94 (3d Cir. 2012); John v. Nat’l Sec. Fire & Cas. Co., 501 F.3d 443, 445 (5th Cir. 2007); In re Initial Pub. Offerings Sec. Litig., 471 F.3d 24, 44–45 (2d Cir. 2006). This makes sense when issues of notice and damages are at play, i.e., for 23(b)(1) and (23)(b)(3) classes. But courts of appeals have consistently declined to impose an ascertainability requirement in 23(b)(2) cases requesting that a party be enjoined from certain actions. See Yaffe v. Powers, 454 F.2d 1362, 1366 (1st Cir. 1972), abrogated on other grounds by Gardner v. Westinghouse Broad. Co., 437 U.S. 478 (1978); Shelton v. Bledsoe, 775 F.3d 554, 559–63 (3d Cir. 2015) (vacating order denying class certification in 23(b)(2) case); Cole v. City of Memphis, 839 F.3d 530, 542 (6th Cir. 2016) (affirming district court’s certification of 23(b)(2) class in a challenge to city’s street-sweep policy); Shook v. El Paso County, 386 F.3d 963, 972–73 (10th Cir. 2004) (stating that “identifiability” is not a concern with 23(b)(2) classes).

59

These holdings are supported by the Advisory Committee Notes to Rule 23(b)(2), which state that “illustrative” examples of a Rule 23(b)(2) class “are various actions in the civil-rights field where a party is charged with discriminating unlawfully against a class, usually one whose members are incapable of specific enumeration.” Fed. R. Civ. P. 23 advisory committee’s note (1966) (emphasis added). There is no threshold ascertainability requirement in this Rule 23(b)(2) case, which seeks only declaratory and injunctive relief from a discriminatory policy. The district court did not abuse its discretion in certifying the class.

D. Medicaid Act

The Anderson Appellants next challenge the district court’s finding that the exclusion violates the availability and comparability requirements of the Medicaid Act.

1.

The Act’s availability provision requires states to cover both mandatory and optional services in sufficient “amount, duration, and scope to reasonably achieve its purpose.” 42 C.F.R. § 440.230(b). States can “place appropriate limits on a service based on such criteria as medical necessity or on utilization control procedures.” 42 C.F.R. § 440.230(d). 29 But they cannot “arbitrarily deny or reduce the amount, duration, or scope of a required service . . . to an otherwise eligible beneficiary solely because of the diagnosis, type of illness, or condition.” 42 C.F.R. § 440.230(c).

29

Neither the statute nor regulations define medical necessity or utilization control procedures. As explained above, West Virginia relies on a third party, Kepro (which, in turn, relies on InterQual criteria), to decide which services are medically necessary. J.A. 571.

60

The Supreme Court has said that “serious statutory questions might be presented if a state Medicaid plan excluded necessary medical treatment from its coverage.” See Beal v. Doe, 432 U.S. 438, 444 (1977). Other circuits have held that medically necessary procedures that fall within mandatory categories of coverage must be covered. See Bontrager v. Ind. Fam. Soc. Servs. Admin., 697 F.3d 604, 608 (7th Cir. 2012); Ellis ex rel. Ellis v. Patterson, 859 F.2d 52, 54 (8th Cir. 1988); Meusberger v. Palmer, 900 F.2d 1280, 1282 (8th Cir. 1990); Dexter v. Kirschner, 984 F.2d 979, 983 (9th Cir. 1992); Moore ex rel. Moore v. Reese, 637 F.3d 1220, 1232–33 (11th Cir. 2011). 30 At least one court has noted that medical necessity alone cannot compel coverage because the regulations say that the state Medicaid agency “may place appropriate limits on a service based on such criteria as medical necessity or on utilization control procedures.” 42 C.F.R. § 440.230(d) (emphasis added). Cruz v. Zucker, 195 F. Supp. 3d 554, 570–71 (S.D.N.Y. 2016), order reversed on other grounds on motion for reconsideration, 218 F. Supp. 3d 246 (S.D.N.Y. 2016). Presumably, “[p]roper utilization control procedures, as distinct from medical necessity, may limit the provision of services.” Id. at 571 (citing Pharm. Rsrch. & Mfrs. of Am. v. Walsh, 538 U.S. 644 (2003) (plurality opinion) (upholding prior authorization procedures)). Another court has suggested that, in

30

The corollary of this is that states do not have to cover treatments that are not medically necessary—even if those treatments fall within the “mandatory” categories of coverage—so long as the coverage decision is not based solely on the patient’s “diagnosis, type of illness, or condition.” See Moore ex rel. Moore, 637 F.3d at 1232–33.

61

particular cases, but not as a general rule, a state could deny coverage for a service deemed medically necessary. See Hern v. Beye, 57 F.3d 906, 911 (10th Cir. 1995). 31 Under any of those readings, though, West Virginia’s categorical exclusion violates the availability requirement. It does so in two ways. First, it is inconsistent with the objectives of the Act: to provide medical assistance to people too poor to afford it. See Md. Dep’t of Health & Mental Hygiene v. Ctrs. for Medicare & Medicaid Servs., 542 F.3d 424, 429 (4th Cir. 2008). Though state plans have discretion to determine what to cover and the scope of that coverage, all state plans must “include reasonable standards . . . for determining eligibility for and the extent of medical assistance under the plan which . . . are consistent with the objectives” of the Act. 42 U.S.C. § 1396a(a)(17). “But a state law that categorically denies coverage for a specific, medically necessary procedure . . . is not a ‘reasonable standard [] . . . consistent with the objectives of [the Act].’” Hern, 57 F.3d at 911.

Second, the exclusion violates the availability requirement by “arbitrarily den[ying] or reduc[ing] the amount, duration, or scope of a required service . . . to an otherwise eligible beneficiary solely because of the diagnosis, type of illness, or condition.” 42 C.F.R.

31

As examples, the Hern Court cited Miller ex rel. Miller v. Whitburn, 10 F.3d 1315, 1321 (7th Cir. 1993) (“stating that a participating state may deny coverage for experimental treatments so long as its definition of ‘experimental’ and its application of the restriction are reasonable”); Charleston Mem’l Hosp. v. Conrad, 693 F.2d 324, 330 (4th Cir. 1982) (“holding that a state’s annual limits on Medicaid coverage to twelve inpatient hospital days—which met the needs of 88 percent of Medicaid recipients—and eighteen outpatient hospital visits—which met the needs of 99 percent of Medicaid recipients—was consistent with Title XIX and applicable regulations”); and Curtis v. Taylor, 625 F.2d 645, 651–53 (5th Cir. 1980) (“upholding a state’s limit on Medicaid coverage to three physicians’ visits per month where only 3.9 percent of the state’s Medicaid population had required more than three physicians’ visits in any one month in the year before the regulation was adopted”). 57 F.3d at 911.

62

§ 440.230(c). Determinations about proper medical treatment will always be based on the patient’s diagnosis. But they cannot be arbitrarily denied to a patient for whom the treatment is medically necessary based on that diagnosis alone. West Virginia’s exclusion, which bars coverage of all surgeries to treat gender dysphoria, regardless of medical necessity, does just that.

The exclusion violates the availability requirement of the Medicaid Act.

2.

Under Medicaid’s comparability requirement, states must ensure that services available to any categorically needy individual are “equal in amount, duration, and scope for all beneficiaries within the group.” 42 C.F.R. § 440.240(b)(1). The same applies to individuals in a covered medically needy group. 42 C.F.R. § 440.240(b)(2).

West Virginia Appellants argue that the Medicaid policy does not offer services different in amount, duration, or scope to individuals within each group because surgical treatment for gender dysphoria is not covered for any Medicaid participant. Opening Br. at 45–46. This is the same argument they made in the Equal Protection context.

In support, they cite to a Second Circuit case in which the court held that New York City did not violate the Medicaid Act when it covered certain in-home personal care services but did not cover safety monitoring for individuals who suffered from mental disabilities. Rodriguez v. City of New York, 197 F.3d 611, 613–14 (2d Cir. 1999). The court rejected plaintiff’s argument, noting that “Section 1396(a)(10)(B) does not require a state to fund a benefit that it currently provides to no one. Its only proper application is in situations where the same benefit is funded for some recipients but not others.” Id. at 616. This is no doubt true. But as the Second Circuit itself pointed out in 2016, that does not mean that the court

63

defers to a state’s definition of what the relevant service is. Davis v. Shah, 821 F.3d 231, 257 (2d Cir. 2016). It did not question that states may, “within reason, define the scope and purpose of the services it provides.” Id. But “allowing a state to deny medical benefits to some categorically needy individuals that it provides to others with the exact same medical needs simply by defining such services—however arbitrarily—as aimed at treating only some medical conditions would risk swallowing the comparability provision whole.” Id.

The Court continued:

If, for example, New York defined the purpose of an arm cast as supporting regrowth of broken bones in the right arm only, or defined the purpose of a prosthetic leg as enhancing mobility in disabled individuals born without limbs, surely it would violate the comparability requirement to deny equivalent services to categorically needy individuals who break their left arms, or who lose limbs through amputation, but who have the same indisputable medical needs for a cast or prosthetic. Such a scenario would seem an archetypal instance of denying some categorically needy individuals the same “scope” of medical assistance available to others under a state plan.

Id. at 257–58.

The same is true here. West Virginia cannot get around the comparability requirement by defining the relevant services as services aimed at treating only some medical conditions (i.e., non-gender dysphoria conditions) any more than it can get around the Equal Protection Clause by doing so. See also White v. Beal, 555 F.2d 1146, 1151 (3d Cir. 1977) (“We find nothing in the federal statute that permits discrimination based upon etiology rather than need.”). The policy violates Medicaid’s comparability requirement.

3.

Appellants argue that the district court erred by “not afford[ing] any deference to the fact that CMS has approved Medicaid’s State plan and thereby has made an implicit

64

judgment that the plan complies with federal law.” Opening Br. at 38. This misstates the record. While CMS approved the plan, it did not approve the exclusion—the exclusion was not included in the plan that was submitted to CMS. See West Virginia State Medicaid Plan, https://perma.cc/BJ86-GNLX (last visited Dec. 18, 2023); West Virginia State Medicaid Plan, Attachment 3 (last updated March 2022), https://perma.cc/Y7FF-74LN (last visited Nov. 29, 2023). The scope-of-coverage template that CMS provides to the states to complete and return asks whether the state plans cover certain services for the categorically needy, including eyeglasses, inpatient psychiatric services, hospice care, and others. The template does not ask whether the states cover access to “transsexual surgery” or gender-affirming surgery for transgender individuals, nor did West Virginia volunteer this information. Id. CMS therefore made no judgment about whether West Virginia’s plan complies with the Medicaid Act, and there is nothing for this Court to defer to.

E. Affordable Care Act

The Affordable Care Act’s anti-discrimination mandate provides that, “[e]xcept as otherwise provided . . . an individual shall not, on the ground prohibited under Title VI of the Civil Rights Act . . . [and] Title IX . . . be excluded from participation in, be denied the benefits of, or be subjected to discrimination under, any health program or activity, any part of which is receiving Federal financial assistance.” 42 U.S.C. § 18116(a). “[F]or guidance” in evaluating a Title IX claim, this Circuit relies on caselaw interpreting Title VII. Jennings v. Univ. of N.C., 482 F.3d 686, 695 (4th Cir. 2007); see Grimm, 972 F.3d at 616. The district court therefore applied Bostock v. Clayton County, which held that “an

65

employer who intentionally treats a person worse because of sex . . . discriminates against that person in violation of Title VII.” 140 S. Ct. at 1740.

Appellants argue that Bostock is the wrong standard because it was “limited to Title VII claims involving employers who fired employees because they were gay or transgender.” Opening Br. at 36. But there is nothing in Bostock to suggest the holding was that narrow. Appellants also argue that “[h]istorically in terms of Title IX jurisprudence, the term ‘sex’ referred to the binary sex of male and female, and ‘gender identity’ was understood as a distinct concept.” Opening Br. at 36. But Bostock was based on that assumption. 140 S. Ct. at 1739 (“[B]ecause nothing in our approach to these cases turns on the outcome of the parties’ [historical] debate . . . we proceed on the assumption that ‘sex’ . . . referr[ed] only to biological distinctions between male and female.”). So even if the definition of sex under Title IX encompasses only binary sex, West Virginia’s policy still violates the ACA.

III.

The North Carolina State Health Plan and the West Virginia Medicaid Program discriminate on the basis of gender identity and sex in violation of the Equal Protection Clause. The West Virginia Medicaid Program violates the Medicaid Act’s availability and comparability provisions and violates the Affordable Care Act’s anti-discrimination provision. The North Carolina district court did not abuse its discretion by using an amicus brief to provide context about transgender healthcare, striking certain portions of expert testimony, and enjoining the state from “enforcing the Plan’s exclusion.” Nor did the West

66

Virginia district court abuse its discretion in certifying the Appellee class. The decisions of the district courts in both cases are therefore

AFFIRMED.

67

RICHARDSON, Circuit Judge, with whom Judges WILKINSON, NIEMEYER, and QUATTLEBAUM join, and with whom Judges AGEE and RUSHING join except for Part II.A.3, dissenting:

In Grimm v. Gloucester County School Board, we heralded the victory of “the burgeoning values of our bright youth” over “the prejudices of the past.” 972 F.3d 586, 620 (4th Cir. 2020). Our en banc Court treats these cases as new fronts upon which this conflict must be waged. But not every battle is part of a larger war. In the majority’s haste to champion plaintiffs’ cause, today’s result oversteps the bounds of the law. The majority asserts that the challenged exclusions use medical diagnosis as a proxy for transgender persons, despite the complete lack of evidence for this claim. It then blatantly sidesteps controlling Supreme Court precedent by conjuring up an imagined conflict with another, unrelated line of cases. Finally, it misrepresents how the challenged exclusions actually work in order to malign them as sex-based and grounded in stereotypes. The result is a holding that speaks the language of Equal Protection yet departs wholly from its established principles.

I respectfully dissent. The Equal Protection Clause does not license judges to strike down any policy we disagree with. It instead grants the states leeway to tailor policies to local circumstances, while providing a carefully calibrated remedy for truly illicit discrimination. No such discrimination appears in these cases. North Carolina and West Virginia do not target members of either sex or transgender individuals by excluding coverage for certain services from their policies. They instead condition coverage on whether a patient has a qualifying diagnosis. Anyone—regardless of their sex, gender identity, or combination thereof—can obtain coverage for these services if they have a

68

qualifying diagnosis. And no one—regardless of their sex, gender identity, or combination thereof—can obtain coverage if they lack one. There is therefore nothing about these policies that discriminates on the basis of sex or transgender status. I. Background North Carolina and West Virginia (together, the “states”) operate respective health-

insurance and Medicaid plans that reimburse individuals for a variety of healthcare needs. Yet neither plan covers every attainable medical service. Today’s cases concern the choice of both states to exclude coverage for certain sex-change services. The North Carolina State Health Plan for Teachers and State Employees excludes coverage for “[t]reatment or studies leading to or in connection with sex changes or modifications and related care.” Kadel, J.A. 3836. The West Virginia State Medicaid Program similarly excludes coverage for “transsexual” or “[s]ex change” surgeries. Anderson, J.A. 935, 941–43. Both exclusions operate to deny coverage for certain treatments of gender dysphoria, a mental disorder defined as “the distress that may accompany the incongruence between one’s experienced or expressed gender and one’s assigned gender.” Am. Psychiatric Ass’n, Diagnostic and Statistical Manual of Mental Disorders 451 (5th ed. 2013). And both exclusions were challenged in court on various statutory and constitutional grounds.

Plaintiffs in Kadel are members of the North Carolina plan who identify as transgender and have been diagnosed with gender dysphoria. Each of them sought certain treatments for gender dysphoria but was denied coverage because of the exclusion. This prompted them to sue North Carolina, alleging, among other things, an Equal Protection

69

Clause violation. The district court granted plaintiffs summary judgment on the Equal Protection claim and permanently enjoined North Carolina from enforcing the exclusion.

Plaintiffs in Anderson are transgender participants in West Virginia’s Medicaid program who also have been diagnosed with gender dysphoria. They too desire certain treatments for gender dysphoria but would be denied coverage for them. So they sued West Virginia, alleging that the program violates the Equal Protection Clause, § 1557 of the Affordable Care Act, and certain provisions of the Medicaid Act. The district court granted summary judgment to plaintiffs on all three grounds, denied summary judgment to West Virginia, entered a declaratory judgment, and enjoined West Virginia from enforcing its exclusion. II. Discussion These appeals involve two issues. First, plaintiffs argue that the challenged exclusions violate the Equal Protection Clause and, in Anderson, § 1557 of the Affordable Care Act. Second, plaintiffs in Anderson argue that the exclusions violate the Medicaid Act. I consider each issue in turn.

A. Discrimination Claims 1. Equal Protection Doctrine The Equal Protection Clause of the Fourteenth Amendment commands that “no State shall . . . deny to any person within its jurisdiction the equal protection of the laws.” U.S. Const. amend. XIV, § 1. At its core, the Clause prevents states from “treating differently persons who are in all relevant respects alike.” Nordlinger v. Hahn, 505 U.S. 1, 10 (1992); City of Cleburne v. Cleburne Living Ctr., 473 U.S. 432, 439 (1985). Yet the

70

scope of this prohibition should not be exaggerated. Laws often deal in classifications to solve particular problems or to achieve targeted outcomes. “The Equal Protection Clause does not forbid classifications” categorically. Nordlinger, 505 U.S. at 10. Rather, classifications ordinarily are valid so long as they have a rational basis. City of Cleburne, 473 U.S. at 440; Pers. Adm’r of Mass. v. Feeney, 442 U.S. 256, 271 (1979) (“When the basic classification is rationally based, uneven effects upon particular groups within a class are ordinarily of no constitutional concern.”). “[T]he Constitution presumes that even improvident decisions will eventually be rectified by the democratic processes.” City of Cleburne, 473 U.S. at 440.

Yet this presumption gives way when a law treats people differently because of their membership in a protected class. Sex is one such protected class. When a law discriminates based on sex, we fear that it is rooted in “outmoded notions of the relative capabilities of men and women,” id. at 441, or “traditional, often inaccurate, assumptions about the proper roles of men and women,” Miss. Univ. for Women v. Hogan, 458 U.S. 718, 726 (1982). At the same time, however, we know that “[p]hysical differences between men and women . . . are enduring.” United States v. Virginia, 518 U.S. 515, 533 (1996). Recognizing biological reality is “not a stereotype.” Nguyen v. INS, 533 U.S. 53, 68 (2001). To split the difference, we subject sex discrimination to intermediate scrutiny, which requires a law to be substantially related to a sufficiently important government interest. Virginia, 518 U.S. at 533.

The easiest way for a plaintiff to prove sex discrimination is to show that a law facially classifies based on sex. A facial classification triggering heightened scrutiny is

71

one that explicitly “distributes burdens or benefits on the basis of” membership in a protected class. Parents Involved in Cmty. Schs. v. Seattle Sch. Dist. No. 1, 551 U.S. 701, 720 (2007); see also Shaw v. Reno, 509 U.S. 630, 642 (1993) (describing a suspect facial classification as one that “explicitly distinguish[es] between individuals on [protected] grounds”); Adarand Constructors, Inc. v. Pena, 515 U.S. 200, 213 (1995) (same). In other words, a law facially classifies when, by its own terms, it identifies sex as a ground for discriminatory treatment. Consider an obvious example. Suppose a health-insurance policy said: “Women may not receive reimbursement for heart transplants.” This policy would be a facial classification based on sex: Whether a person would be denied reimbursement would turn (at least in part) on whether they were a man or a woman.

But not every law that references or relates to sex necessarily classifies on that basis.

For instance, imagine a Medicaid policy that said: “Neither men nor women may receive reimbursement for heart transplants.” This policy might be unartfully worded, but it would not be a sex-based classification. Both sexes would be treated the same, as neither could receive reimbursement for heart transplants. See Vacco v. Quill, 521 U.S. 793, 800 (1997) (“Generally speaking, laws that apply evenhandedly to all ‘unquestionably comply’ with the Equal Protection Clause.” (quoting N.Y.C. Transit Auth. v. Beazer, 440 U.S. 568, 587 (1979))). So the fact that a policy uses terms like “sex,” “men,” or “women” does not automatically mean that it facially classifies on these grounds. See, e.g., Atkins v. Rumsfeld, 464 F.3d 456, 468 (4th Cir. 2006) (holding that a law providing retirement benefits to divorced military spouses and defining spouse as “the husband or wife . . . of a member” was not a facial classification). Determining whether a law facially classifies based on sex

72

thus involves more than a mere word search for particular terms. Rather, we must examine whether the policy uses those terms to draw distinctions between the sexes.

I break no new ground by saying this. Over and over, the Supreme Court has said that sex-based facial classifications explicitly identify sex as the basis for favorable or unfavorable treatment. See Reed v. Reed, 404 U.S. 71, 73 (1971) (providing that “males must be preferred to females” when appointing the administrator of a decedent’s estate); Frontiero v. Richardson, 411 U.S. 677, 691 (1973) (requiring female, but not male, service members to prove that their spouses are financially dependent in order to receive benefits); Stanton v. Stanton, 421 U.S. 7, 13–14 (1975) (setting a lower age of majority for women); Weinberger v. Weisenfeld, 420 U.S. 636, 637–38 (1975) (denying widowers certain Social Security benefits); Craig v. Boren, 429 U.S. 190, 192 (1976) (allowing women under the age of twenty-one, but not men under that age, to buy beer); Orr v. Orr, 440 U.S. 268, 271 (1979) (requiring only men to pay alimony); Michael M. v. Superior Ct., 450 U.S. 464, 466 (1981) (plurality opinion) (holding only men liable for statutory rape); Hogan, 458 U.S. at 720 (denying admission to men); J.E.B. v. Alabama ex rel. T.B., 511 U.S. 127, 129 (1994) (excluding potential jurors based on sex); Virginia, 518 U.S. at 519–20 (denying admission to women); Sessions v. Morales-Santana, 582 U.S. 47, 51 (2017) (establishing different immigration rules for fathers versus mothers). 1 These cases demonstrate that our task is

1

Admittedly, the laws in some of these cases didn’t use the words “men” or “women” but rather used sex-identifying language such as “father,” “mother,” “husband,” or “wife.” Yet these were still facial classifications. A law that discriminates between mothers and fathers, for example, identifies a trait—being a parent—and expressly distinguishes between people who have that trait based on whether they are male or female. (Continued)

73

not simply to note the words used in a law, but to determine what function those words serve in that law’s operation. 2 One way for a plaintiff to prove an Equal Protection violation is to show that a law facially classifies based on sex. But there are other ways. Even a facially neutral classification may warrant heightened scrutiny if it uses a proxy to camouflage intentional discrimination based on a protected trait. Of course, to trigger the Equal Protection Clause in the first place, the challenged law must first make some classification of persons. See Palmer v. Thompson, 403 U.S. 217, 219–20 (1971) (finding that a city policy closing public pools, even if motivated by a desire to avoid integration, did not deny anyone “the equal protection of the laws” where the city closed the pools “to all its citizens”). But once a classification has been made, the law offends Equal Protection principles if “a gender- based discriminatory purpose has, at least in some measure, shaped the [challenged] legislation.” Feeney, 442 U.S. at 276, 279 (explaining that a legislature acts with a discriminatory purpose when it “select[s] or reaffirm[s] a particular course of action at least

It therefore facially classifies based on sex, even though it also classifies based on a second characteristic (parenthood). Morales-Santana, 582 U.S. at 58 (explaining that discriminating between mothers and fathers is just another way of discriminating “on the basis of the sex of the qualifying parent” (quoting Califano v. Westcott, 443 U.S. 76, 84 (1979))).

2

To be clear, once a facial classification based on a protected trait has been shown, the government cannot evade heightened scrutiny by claiming that the law applies equally to everyone. For instance, even though the anti-miscegenation law in Loving v. Virginia technically applied to all citizens, it still facially classified based on race by prohibiting marriages for persons of one race that it permitted for persons of the other race. 388 U.S. 1, 8–9 (1967); see also McLaughlin v. Florida, 379 U.S. 184, 184–86, 191–92 (1964) (striking down a law that prohibited cohabitation between interracial couples).

74

in part ‘because of,’ not merely ‘in spite of,’ its adverse effects upon an identifiable group”). In making this determination, an analysis of the law’s disparate “impact provides an ‘important starting point,’ but purposeful discrimination is ‘the condition that offends the Constitution.’” Id. at 274 (first quoting Vill. of Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252, 266 (1977); and then quoting Swann v. Charlotte-Mecklenburg Bd. of Educ., 402 U.S. 1, 16 (1971)).

Discriminatory purpose or intent is usually proved through a fact-intensive inquiry, requiring investigation into things like the law’s impact, its historical background, and its legislative history. See Arlington Heights, 429 U.S. at 264–68. Thus, in Rice v. Cayetano, the Supreme Court struck down a provision of the Hawaii Constitution that limited the right to vote for certain public officials to persons descended from the aboriginal peoples inhabiting the Hawaiian Islands in 1778. 528 U.S. 495, 499 (2000). Hawaii argued that the provision was race-neutral because it turned on a person’s ancestry, not their race. Id. at 514. But the Court disagreed and found that the state was really using ancestry as a proxy for race. Id. The inhabitants of Hawaii in 1778, the Court explained, shared common physical and cultural characteristics. Id. at 514–15. And an examination of the legislative history, including prior versions of the provision and statements from its enactors, revealed that the provision was intended to “preserve that commonality of people to the present day.” Id. at 515–16 (“The very object of the statutory definition in question . . . is to treat the early Hawaiians as a distinct people, commanding their own recognition and respect.”).

75

The Court therefore held that the provision discriminated on the basis of race because of “its express racial purpose and by its actual effects.” Id. at 517. 3 Yet a full-blown evidentiary inquiry is not always necessary to prove discriminatory intent. Sometimes a law uses a classification that is so obviously a proxy for a suspect class that “an intent to disfavor that class can be readily presumed.” Bray v. Alexandria Women’s Health Clinic, 506 U.S. 263, 270 (1993). 4 But that presumption doesn’t come

3

The majority cites Rice for the proposition that proxy discrimination can be established by only looking at the text of the challenged law “coupled with basic facts,” and thus that “evidence of invidious discrimination” isn’t necessary. Majority Op. at 38– 39. But Rice explicitly looked at both the statute’s effects and its discriminatory purpose to hold that it discriminated based on race. 528 U.S. at 517 (“[T]he State’s argument is undermined by its express racial purpose and its actual effects.”). And it identified that purpose not by simply looking at the statute’s text and “basic facts” but also by examining one type of evidence that Arlington Heights specifically recognized: legislative history. Rice, 528 U.S. at 515–17; Arlington Heights, 429 U.S. at 268. It was only with that legislative history that the Court was able to determine that “Hawaiian” was used as a proxy for race. Rice, 528 U.S. at 515–17. So Rice confirms that an evidentiary inquiry into a law’s purpose is typically required.

4

The majority thinks that the proxy inquiry is a subset of the facial classification inquiry. Majority Op. at 37–39. We disagree. That we must ask whether a law uses a classification that is merely a substitute for a protected trait means that the law does not explicitly—i.e., facially—classify based on that protected trait. Cf. Hunt v. Cromartie, 526 U.S. 541, 546 (1999) (“When . . . classifications are explicit, no inquiry into legislative purpose is necessary.”). Indeed, the Supreme Court has expressly rejected this argument, noting that courts only inquire into “covert” classifications—i.e., ostensibly neutral classifications that “could not be plausibly explained on a neutral ground”—after concluding that a statute is “gender-neutral on its face.” Feeney, 442 U.S. at 274–75. Instead, the proxy inquiry is better understood as a species of intentional discrimination: Is the government targeting something because of its close connection to another thing? See Rice, 528 U.S. at 517. And (as we will explain) we can conclusively presume such intent, without conducting a full-blown evidentiary inquiry, in the rare set of cases where the classification is so irrational that nothing could explain it but an intent to discriminate. See Bray, 506 U.S. at 270 (explaining that when a statute blatantly uses a proxy for a protected class, “an intent to disfavor that class can readily be presumed”); Arlington Heights, 429 U.S. at 266.

76

easy. We can only presume discriminatory intent when the law’s explicit target is “an irrational object of disfavor” and the law “happen[s] to [affect] exclusively or predominantly . . . a particular class of people.” Id.; see also Arlington Heights, 429 U.S. at 266 (“Sometimes a clear pattern, unexplainable on grounds other than race, emerges . . . even when the governing legislation appears neutral on its face. The evidentiary inquiry is then relatively easy. But such cases are rare.” (emphasis added) (internal citations omitted)). In other words, we must find that the law overwhelmingly affects a suspect class and that there’s no logical reason for the distinction the law makes other than targeting that suspect class. See Feeney, 442 U.S. at 275 (“If the impact of [the] statute could not be plausibly explained on a neutral ground, impact itself would signal that the real classification made by the law was in fact not neutral.”).

Consider the cases in which the Supreme Court has deployed this presumption. In Yick Wo v. Hopkins, the Court found that a city’s facially neutral permitting requirement, which was applied almost exclusively to the detriment of Chinese workers, violated the Equal Protection Clause. 118 U.S. 356, 374 (1886). “No reason for [the unequal treatment] is shown,” the Court found, “and the conclusion cannot be resisted that no reason for it exists except hostility to the race and nationality to which the petitioners belong, and which, in the eye of the law, is not justified.” Id. 5 Similarly, in Guinn v. United States, the Court

5

Yick Wo technically involved the discriminatory enforcement of a facially neutral law, not the reasons behind that law’s enactment. Yick Wo, 118 U.S. at 373–74. But the principles are the same: Whether in enacting or enforcing a law that draws classifications between people, official action taken for a discriminatory purpose triggers Equal Protection scrutiny. Arlington Heights, 429 U.S. at 266.

77

invalidated an Oklahoma law that required voters to take a literacy test unless their ancestors were eligible to vote in or before 1866 (conveniently, right before the enactment of the Fifteenth Amendment). 238 U.S. 347, 364–65 (1915). 6 The Court found that this law discriminated on the basis of race, despite its facial neutrality, because the Court could not identify “any basis of reason for the standard thus fixed other than” to contravene the Fifteenth Amendment. Id. at 365. And in Gomillion v. Lightfoot, the Court refused to dismiss a Fifteenth Amendment suit challenging a twenty-eight-sided electoral district that allegedly excluded all black voters, since the government could not identify “any countervailing municipal function” for the bizarre shape. 364 U.S. 339, 342 (1960). If these allegations were true, the Court reasoned, then “the conclusion would be irresistible, tantamount for all practical purposes to a mathematical demonstration,” that the district was designed to discriminate against black voters. Id. at 341; see also Shaw, 509 U.S. at 649 (“[A] plaintiff challenging a reapportionment statute under the Equal Protection Clause may state a claim by alleging that the legislation, though race-neutral on its face, rationally cannot be understood as anything other than an effort to separate voters into different districts on the basis of race, and that the separation lacks sufficient justification.”).

These cases reveal that what is critical to obtaining a presumption of discriminatory intent is showing both discriminatory effects and that no rational, nondiscriminatory explanation exists for the law’s classification. Indeed, the Supreme Court has been clear

6

We, like the Supreme Court, use Fifteenth Amendment cases to inform our Fourteenth Amendment cases, see, e.g., Arlington Heights, 429 U.S. at 266, as they expound general principles of antidiscrimination law.

78

that the mere fact that a law primarily—or even exclusively—affects a protected class cannot alone establish an Equal Protection claim. Washington v. Davis, 426 U.S. 229, 242 (1976) (“Disproportionate impact is not irrelevant, but it is not the sole touchstone of an invidious racial discrimination forbidden by the Constitution.”). So if we can identify rational, nondiscriminatory reasons for why the law targets who or what it does, then we cannot presume an intent to discriminate. 7 This is illustrated by failed attempts to establish the presumption. In Myers v.

Anderson (decided on the same day as Guinn), the Supreme Court held that a Maryland provision conferring the right to vote on all taxpayers assessed for at least $500 did not itself violate the Fifteenth Amendment. 238 U.S. 368, 379 (1915). 8 “[A]s there is a reason other than discrimination on account of race or color discernible upon which the standard may rest,” the Court explained, “there is no room for the conclusion that it must be assumed, because of the impossibility of finding any other reason for its enactment, to rest alone upon a purpose to violate the 15th Amendment.” Id. Likewise, in Personnel Administrator of Massachusetts v. Feeney, the Court refused to hold that a hiring preference for veterans was mere pretext for sex discrimination, even though it overwhelmingly favored men. 442 U.S. at 274–75. The Court found that the preference could not “plausibly be explained only as a gender-based classification,” since it was

7

Of course, such an intent still can be proven though the fuller evidentiary inquiry of Arlington Heights.

8

The Court nonetheless invalidated the provision because, although it was itself constitutional, it was inseverable from a different, unconstitutional provision. Myers, 238 U.S. at 380–83.

79

gender-neutral by definition, it placed a significant number of non-veteran males at a disadvantage, and it served “legitimate and worthy purposes.” Id. Finally, in Bray v. Alexandria Women’s Health Clinic, the Court rejected the argument that opposition to abortion is necessarily sex discrimination. 506 U.S at 270. “[O]pposition to voluntary abortion cannot possibly be considered such an irrational surrogate for opposition to (or paternalism towards) women,” the Court concluded, because “[w]hatever one thinks of abortion, it cannot be denied that there are common and respectable reasons for opposing it, other than hatred of, or condescension towards (or indeed any view at all concerning), women as a class.” Id. 9 This brings us to Geduldig v. Aiello, 417 U.S. 484 (1974). Geduldig involved an Equal Protection challenge to California’s disability-insurance system, which excluded coverage for “any injury or illness caused by or arising in connection with pregnancy.” Id. at 489. The dissenting Justices argued that the exclusion discriminated on the basis of sex by “singling out for less favorable treatment a gender-linked disability peculiar to women.” Id. at 501 (Brennan, J., dissenting). But the Court disagreed. California had not, the Court found, denied insurance eligibility to any group of persons; it had simply chosen to underinsure a particular risk (i.e., pregnancy). Id. at 494 (majority opinion). Its reasons for doing so—maintaining a self-supporting, cost-effective, and affordable insurance program—were legitimate, given the substantial cost of insuring pregnancy, and provided

9

At the same time, the Court gave an example of a law that would support the presumption: “A tax on wearing yarmulkes is a tax on Jews.” Bray, 506 U.S. at 270. This is because yarmulkes are “such an irrational object of disfavor” that only an intent to discriminate against Jews could explain such a tax. See id.

80

“an objective and wholly noninvidious basis” for the exclusion. Id. at 496. And what risk coverage California did afford, it afforded equally to both men and women. Id. at 496–97. As the Court explained:

The California insurance program does not exclude anyone from benefit eligibility because of gender but merely removes one physical condition—

pregnancy—from the list of compensable disabilities. While it is true that only women can become pregnant it does not follow that every legislative classification concerning pregnancy is a sex-based classification . . . .

Normal pregnancy is an objectively identifiable physical condition with unique characteristics. Absent a showing that distinctions involving pregnancy are mere pretexts designed to effect an invidious discrimination against the members of one sex or the other, lawmakers are constitutionally free to include or exclude pregnancy from the coverage of legislation such as this on any reasonable basis, just as with respect to any other physical condition.

The lack of identity between the excluded disability and gender as such under this insurance program becomes clear upon the most cursory analysis. The program divides potential recipients into two groups—pregnant women and nonpregnant persons. While the first group is exclusively female, the second includes members of both sexes. The fiscal and actuarial benefits of the program thus accrue to members of both sexes.

Id. at 496 n.20. The Court therefore held that the plan did not violate the Equal Protection Clause. Id. at 497.

Geduldig was no outlier. For one, the Court has repeatedly reaffirmed its holding.

See Bray, 506 U.S. at 271; Gen. Elec. Co. v. Gilbert, 429 U.S. 125, 136 (1976); 10 Nashville Gas Co. v. Satty, 434 U.S. 136, 142 (1977); Dobbs v. Jackson Women’s Health Org., 597 U.S. 215, 236 (2023). Moreover, Geduldig’s principles accord with the broader Equal

10

Gilbert was superseded by statute. See Pregnancy Discrimination Act of 1978, Pub. L. No. 95-555, 92 Stat. 2076; see also Newport News Shipbuilding & Dry Dock Co. v. EEOC, 462 U.S. 669, 685 (1983).

81

Protection doctrine. That a state plan doesn’t cover a medical condition that only members of one sex experience does not itself mean that it facially classifies based on sex. Geduldig, 417 U.S. at 496 n.20. Nor is this fact alone sufficient to establish a presumption of discriminatory intent, given a state’s legitimate interests in maintaining a self-supporting, cost-effective, and affordable healthcare program. Id. Some additional evidence of discriminatory intent beyond underinclusive risk coverage is required to trigger heightened scrutiny. See also Dobbs, 597 U.S. at 236 (“The regulation of a medical procedure that only one sex can undergo does not trigger heightened constitutional scrutiny unless the regulation is a ‘mere pretext designed to effect an invidious discrimination against members of one sex or the other.’” (alteration omitted) (quoting Geduldig, 417 U.S. at 496 n.20)).

2. The challenged exclusions do not violate the Equal Protection Clause.

I now turn to the cases before us. To prevail on their Equal Protection claims, plaintiffs must show that the challenged exclusions discriminate against them because of their sex or transgender status. 11 But they fail to make this showing. The challenged exclusions do not facially classify based on either. Instead, they turn on medical diagnosis and apply evenhandedly to everyone. And their use of medical diagnosis as the

I need not decide in this section whether discrimination based on transgender

11

status merits intermediate scrutiny, either because transgender individuals make up a quasi- suspect class, see Majority Op. at 26–27, or because discrimination based on transgender status is necessarily sex discrimination, see infra Section II.A.3. For even if discrimination against transgender persons does trigger intermediate scrutiny, neither plan discriminates on this basis.

82

discriminating factor is not so irrational that we can presume that they discriminate by proxy. Put simply, whether an individual receives coverage for medical services does not turn on their sex or transgender status. As a result, neither exclusion violates the Equal Protection Clause. 12 I begin with the facial-classification inquiry. At first blush, one might think that the exclusions at issue here are sex- or transgender-based classifications. After all, they collectively deny coverage for certain “sex change” or “transsexual” treatments. 13 And in the past, transgender people were sometimes called “transsexuals.” See, e.g., Transsexual, Merriam-Webster’s Collegiate Dictionary (10th ed. 1993) (“[A] person with a psychological urge to belong to the opposite sex that may be carried to the point of undergoing surgery to modify the sex organs to mimic the opposite sex.”). So surely, as plaintiffs argue, policies that use the words “sex change” and “transsexual” facially discriminate based on sex or transgender status, right?

12

Plaintiffs in Anderson offer no evidence of discriminatory intent, since, as the district court explained, “there is no known reason as to why this Exclusion was ever adopted in the first place.” See Anderson, J.A. 2569 n.1. And because the district court in Kadel determined that North Carolina facially classifies based on sex, it did not address plaintiffs’ evidence of discriminatory intent. So neither appeal presents us with evidence of intent-based discrimination.

13

To be clear, the challenged exclusions are not perfectly identical: North Carolina excludes all sex change “treatments or studies,” while West Virginia only excludes sex change “surgeries.” Kadel, J.A. 3836; Anderson, J.A. 935, 941–43. But this difference does not matter for our purposes. Regardless of the extent of the exclusions, both states exclude from coverage certain treatments for gender dysphoria. Accordingly, I will elide this nuance for the remainder of my analysis.

83

Not so. The exclusions do not use “sex change” or “transsexual” as nouns to identify certain persons who cannot receive coverage. The exclusions use the terms as adjectives. And these adjectives are not used to describe “people,” but “treatment” or “surgery.” On their face, therefore, the exclusions do not deny someone coverage for medical services based on the person’s sex or transgender status. Rather, they deny everyone coverage for certain services based on the medical diagnosis for which the person is seeking those services.

An example shows the difference. Suppose an individual sought a hysterectomy because they had uterine cancer. Both programs would cover the surgery. And they would do so whether the person was male or female, transgender or not. Indeed, Christopher Fain—one of the plaintiffs below—received coverage for a hysterectomy based on a diagnosis unrelated to Fain’s transgender status. 14 But if that same person did not have uterine cancer and instead sought the hysterectomy on the basis of a non-covered diagnosis, like gender dysphoria, then they would not get coverage.

Thus, a person is not covered for certain medical services if they are seeking that service as treatment for gender dysphoria. But if they are seeking the same service for a different, qualifying diagnosis, then North Carolina and West Virginia would cover it— regardless of that person’s sex or transgender status. In other words, there is a list of acceptable diagnoses that would entitle a person to coverage for each service. Every

14

For privacy reasons, Fain has not disclosed the reason for the hysterectomy. But, in a deposition, Fain testified that Medicaid paid for it, and that it was “not related to . . . being transgender.” Anderson, J.A. 1327.

84

person—regardless of their sex, gender identity, or combination thereof—will be covered if they seek that service for one of those diagnoses. And no person—regardless of their sex, gender identity, or combination thereof—will be covered if they seek that service for a diagnosis that’s not on the list, such as gender dysphoria. Neither policy, therefore, facially classifies based on sex or transgender status.

Plaintiffs insist otherwise. They argue that gender dysphoria is a diagnosis exclusively tied to transgender identity. Accordingly, by excluding gender dysphoria, the plans really classify based on transgender identity itself.

But this argument is foreclosed by Geduldig. As in Geduldig, the challenged exclusions do not deny coverage to anyone because of their sex or transgender status. See 417 U.S. at 494–95. Instead, they merely decline coverage for a particular risk: gender dysphoria. See id. And Geduldig held that a health plan that declines to cover a risk that only members of a protected class face does not facially classify people based on their membership in that class. Id. at 496 n.20; see also Dobbs, 597 U.S. at 236. So the fact that only transgender individuals experience gender dysphoria does not mean the exclusions discriminate based on transgender status, any more than the fact that “only women can become pregnant” made the exclusion in Geduldig facially discriminatory. 417 U.S. at 496 n.20. Rather, the dispositive question is whether the plans provide equal risk coverage for all persons. Id. at 496–97. And that is the case here—there is “no risk from which [non- transgender persons] are protected and [transgender persons] are not. Likewise, there is no risk from which [transgender persons] are protected and [non-transgender persons] are not.” Id.

85

Still, plaintiffs contend that these plans really do provide unequal risk coverage, because the plans allegedly deny coverage to transgender individuals for treatments that they provide to others. For example, in North Carolina, men can obtain testosterone if “their bodies do not produce enough,” but transgender men cannot obtain it to treat gender dysphoria. Kadel, Response Br. at 34. In West Virginia, meanwhile, women can receive coverage for a vaginoplasty to treat the congenital absence of a vagina, but transgender women cannot receive a vaginoplasty to treat gender dysphoria. West Virginia similarly covers chest surgery for men who experience gynecomastia, but not for transgender men who experience gender dysphoria. See Gynecomastia, Dorland’s Illustrated Medical Dictionary (28th ed. 1993) (defining gynecomastia as “excessive growth of the male mammary glands”). And both states cover surgery to reconstruct a feminine chest contour following cancer treatment, but not if needed to treat gender dysphoria.

Yet these examples actually demonstrate that the plans do not provide unequal risk coverage based on sex or transgender status. They instead show that, for every medical service, the states have established a list of diagnoses that qualify someone for that service. Which diagnoses qualify is determined by the kinds of risks the state is willing to cover. Here, the states have chosen to cover alterations of a person’s breasts or genitalia only if the person experiences physical injury, disease, or (in West Virginia) congenital absence of genitalia. 15 Anyone who has a diagnosis of this kind can receive coverage for such

15

Each example plaintiffs identify to support their claims involves treatment for physical injury, disease, or congenital absence of genitalia. First, North Carolina covers testosterone treatment for diagnoses like “primary hypogonadism” (testicular failure) and (Continued)

86

medical services, regardless of their sex or transgender status. That the plans do not also cover additional risks, like conditions that only manifest psychological or emotional symptoms (including gender dysphoria), does not change the fact that what coverage they do provide is provided equally to all. See Geduldig, 417 U.S. at 496–97.

Next, plaintiffs contend that the policies unlawfully discriminate because they are based on sex stereotypes. As the Kadel district court explained, the plans supposedly “limit[] members to coverage for treatments that align their physiology with their biological sex and prohibit[] coverage for treatments that ‘change or modify’ physiology to conflict with their assigned sex.” Kadel, J.A. 3704. The challenged exclusions therefore punish transgender persons for gender nonconformity, according to plaintiffs.

Before addressing this argument, it’s important to explain what a stereotype is. A sex stereotype is a generalization about the relative capabilities of, or socially acceptable behavior for, members of each sex. See Glenn v. Bumbry, 663 F.3d 1312, 1316 (11th Cir. 2011) (defining a stereotype as “failing to act and appear according to expectations defined

“hypogonadotropic hypogonadism” (failure of the testes to create testosterone), but not for diminished or lower-than-desired testosterone levels generally. Second, West Virginia covers vaginoplasty for the congenital absence of a vagina. Third, West Virginia provides chest surgery to men with excessive chest tissue (i.e., gynecomastia), but only “if the patient has actual physical pain.” Anderson, J.A. 2527. By contrast, “psychological symptoms”—without physical ones—“are not sufficient to meet the coverage criteria for surgical treatment of gynecomastia.” Anderson, J.A. 1819. And fourth, both plans cover chest reconstruction surgery as part of the treatment for those who have undergone cancer treatment, but not for unrelated cosmetic purposes. Thus, far from showing that the states provide unequal risk coverage, plaintiffs’ examples show a consistent trend: The states provide equal coverage to everyone for certain treatments to redress physical injury, disease, or congenital absence of genitalia, but they do not cover such treatments for anyone experiencing a condition with only psychological or emotional symptoms, like gender dysphoria.

87

by gender”); Hogan, 458 U.S. at 724–25 (explaining that the Equal Protection Clause “must be applied free of fixed notions concerning the roles and abilities of males and females”); accord Price Waterhouse v. Hopkins, 490 U.S. 228, 250–51 (1989) (plurality opinion). Examples abound: Women are unfit for military service. Virginia, 518 U.S. at 549–50. Men should not become nurses. Hogan, 458 U.S. at 729. Women are not “macho.” Price Waterhouse, 490 U.S. at 235. And so on. Whatever their form, the Supreme Court has made clear that the government may not discriminate between men and women based on stereotypes. See Virginia, 518 U.S. at 541; Hogan, 458 U.S. at 724–25.

But that’s not what’s happening here. The plans do not condition coverage based on whether a treatment aligns with or departs from a patient’s sex. Nor do they bar certain persons from treatment if they don’t identify with their sex. Instead, the plans grant or withhold coverage based on a patient’s diagnosis, i.e., a certain physical condition with unique causes, risks, and susceptibility to treatment. See Geduldig, 417 U.S. at 496 n.20 (“Normal pregnancy is an objectively identifiable physical condition with unique characteristics.”). The different coverage accorded to treatments for different diagnoses is therefore based on medical judgment of biological reality, which is “not a stereotype.” Nguyen, 533 U.S. at 68; see also Virginia, 518 U.S. at 533 (“Physical differences between men and women . . . are enduring . . . .”). So plaintiffs fail on this basis, too. 16

16

Of course, it is possible that the selection of certain risks for coverage was pretextual and was really based on gender stereotypes or some other discriminatory purpose. But we need some evidence for this claim beyond the mere selection of risks itself. No such evidence is before us in these appeals.

88

Finally, plaintiffs argue that the plans use gender dysphoria as a proxy for transgender persons. Rather than point to evidence of discriminatory intent, they argue that we can presume such intent because gender dysphoria “happen[s] to [occur] in exclusively” transgender persons. See Bray, 506 U.S. at 270. In other words, they assert that gender dysphoria is so closely tied to transgender identity that the choice to exclude the former can only be explained as intending to exclude the latter.

Yet plaintiffs conveniently fail to mention the other half of the inquiry. That a law targets something closely or exclusively associated with a protected class cannot alone support a presumption of discriminatory intent. See Feeney, 442 U.S. at 274–75; Geduldig, 417 U.S. at 496 n.20; Bray, 506 U.S. at 270; Dobbs, 597 U.S. at 236. The classification a law uses must also be inexplicable on grounds other than an intent to discriminate against a suspect class. See Yick Wo, 118 U.S. at 374; Guinn, 238 U.S. at 364–65; Gomillion, 364 U.S. at 342. So to establish a presumption that the exclusions discriminate by proxy, plaintiffs must show that the choice to exclude gender dysphoria from coverage is so irrational that nothing could explain it other than an intent to discriminate against transgender persons. Bray, 506 U.S. at 270; Arlington Heights, 429 U.S. at 266.

No matter one’s view of the challenged exclusions, one cannot deny that the states have put forth legitimate, nondiscriminatory reasons for denying coverage for certain gender-dysphoria treatments. The main reason is cost. The states have finite and diminishing resources to spend on healthcare. If they must spend money to cover medical services for gender dysphoria, then they either must cut spending (e.g., take away coverage

89

for other diagnoses) or raise taxes. 17 Here, states can reasonably decide that certain gender- dysphoria services are not cost-justified, in part because they question the services’ medical efficacy and necessity. And the evidence on record shows that there is an ongoing debate over this issue. See Anderson, J.A. 1860–1935 (Expert Disclosure Report of Dr. Stephen B. Levine, M.D.); Kadel J.A. 3327–3441 (Expert Witness Declaration of Paul W. Hruz, M.D., Ph.D.). 18 Accordingly, given these legitimate, nondiscriminatory explanations for the exclusions, it cannot be said that the plans obviously use gender dysphoria to discriminate by proxy against transgender persons.

I therefore conclude that the challenged exclusions do not discriminate because of sex or transgender status. Plaintiffs advance a host of arguments for reaching the opposite result. Yet none of their arguments are persuasive. At bottom, the exclusions turn on the basis of medical diagnosis, not on sex or transgender status. The Constitution doesn’t subject such coverage decisions to heightened scrutiny. Hence, Kadel should be remanded to the district court. The § 1557 challenge in Anderson fails. 19 And as to the Equal

17

The majority brushes aside West Virginia’s cost-based arguments because the state failed to provide specific evidence of the cost of surgical treatment for gender dysphoria. Majority Op. at 51. But it didn’t need to provide such evidence. And even if it did, it is undisputed that West Virginia Medicaid anticipates budget deficits within two years, that it cannot add services without sacrificing coverage for existing services, and that it will likely have to cut even existing services soon. It therefore stands to reason that adding surgical treatment for gender dysphoria will be unworkable without compromising current coverage for other conditions.

18

As Judge Quattlebaum’s separate dissent explains, the district court erroneously excluded one of North Carolina’s key witnesses on this point.

19

Section 1557 of the Affordable Care Act provides that “an individual shall not, on the ground prohibited under . . . title IX of the Education Amendments of 1972 . . . be (Continued)

90

Protection challenge in Anderson, the exclusion need only survive rational-basis review. It clearly does so. 20 3. Bostock v. Clayton County Before responding to the majority, I pause to consider a question that is lurking in the background: Does the Supreme Court’s decision in Bostock v. Clayton County have any implications for Equal Protection doctrine? 140 S. Ct. 1731 (2020). I conclude that it does, though not in the way that plaintiffs expect.

Bostock involved a suit brought under Title VII of the Civil Rights Act of 1964, which provides that no employer shall “discriminate . . . because of . . . sex.” 42 U.S.C. § 2000e-2(a)(1). The plaintiffs alleged that their former employers violated Title VII by firing them because of their respective homosexual and transgender statuses. Bostock, 140 S. Ct. at 1737–38. The Supreme Court agreed, holding that an employer who intentionally

subjected to discrimination under[] any health program or activity, any part of which is receiving Federal financial assistance.” 42 U.S.C. § 18116. Medicaid is obviously a federally funded health program. And the “ground” on which Title IX prohibits discrimination is “sex.” 20 U.S.C. § 1681 (“No person in the United States shall, on the basis of sex, be excluded from participation in, be denied the benefits of, or be subjected to discrimination under any education program or activity receiving Federal financial assistance . . . .”). But plaintiffs have not suffered discrimination because of their sex. So West Virginia does not violate § 1557.

20

Under rational-basis review, we presume that a challenged law is valid unless the challenger shows that the law is not “rationally related to a legitimate state interest.” City of Cleburne, 473 U.S. at 440. That means that the challenger must “negative every conceivable basis which might support” the law. Lehnhausen v. Lake Shore Auto Parts Co., 410 U.S. 356, 364 (1973) (emphasis added) (quoting Madden v. Kentucky, 309 U.S. 83, 88 (1940)). Plaintiffs have failed to make this showing. As already explained, the states put forth at least two legitimate reasons for their policy: cost and concerns over medical efficacy and necessity. These reasons are certainly sufficient to establish a rational basis for this policy.

91

discriminates because of homosexual or transgender status necessarily discriminates because of sex, since that employer chooses to tolerate a characteristic in members of one sex that it penalizes in members of the other. Id. at 1741.

Bostock’s holding was based on the plain meaning of Title VII’s text. Id. at 1739.

And the Court declined to explain how its reasoning would affect other antidiscrimination laws. Id. at 1753–54. Hence, several of our colleagues on other Circuits argue that Bostock does not apply outside of Title VII. See Williams ex rel. L.W. v. Skrmetti, 83 F.4th 460, 484–86 (6th Cir. 2023); Eknes-Tucker v. Governor of Ala., 80 F.4th 1205, 1228–29 (11th Cir. 2023); Brandt ex rel. Brandt v. Rutledge, No. 21-2875, 2022 WL 16957734, at *1 n.1 (8th Cir. Nov. 16, 2022) (Stras, J., dissenting from denial of rehearing en banc).

With respect for their thoughtful opinions, I believe they are wrong. I recognize that Bostock left many questions unanswered. Yet Bostock’s principles reverberate in other areas of the law. One such area is Equal Protection.

Though a Title VII case, Bostock addressed generally applicable principles of but-

for causation. The Court concluded that the ordinary, legal meaning of the words “because of” incorporates principles of but-for causation. Bostock, 140 S. Ct. at 1739. This result was unsurprising—in several prior cases, the Court had found this to be true of similar phrases in other statutes, like “by reason of” and “based on.” See Univ. of Tex. Sw. Med. Ctr. v. Nassar, 570 U.S. 338, 352 (2013) (“because of”); Gross v. FBL Fin. Servs., Inc., 557 U.S. 167, 176 (2009) (“because of”); Bridge v. Phx. Bond & Indem. Co., 553 U.S. 639, 652–55 (2008) (“by reason of”); Safeco Ins. Co. of Am. v. Burr, 551 U.S. 47, 63 (2007) (“based on”). The Court in Bostock then clarified that the causation standard incorporated

92

by these statutes is not an idiosyncratic one but rather the “simple” and “traditional” approach to causation in fact used throughout the law. 140 S. Ct. at 1739 (quoting Nassar, 570 U.S. at 346, 360). Finally, the Court explained how to conduct a general but-for causation test: “[C]hange one thing at a time and see if the outcome changes. If it does, we have found a but-for cause.” Id. Bostock, then, did more than simply define the meaning of words in Title VII. It recognized that Title VII incorporates a widely used standard of but-for causation and articulated one way to establish it.

These principles formed the backbone of Bostock’s holding that discrimination based on homosexual or transgender status is necessarily sex discrimination. When two employees of opposite sexes are both attracted to men, for instance, they are “materially identical in all respects,” except for their sex. Id. at 1741. If their employer subsequently fires one of them, but not the other, for their attraction, then the employer has chosen to tolerate traits in members of one sex that it penalizes in members of the other. Id. In other words, “the employer intentionally singles out an employee to fire based in part on the employee’s sex, and the affected employee’s sex is a but-for cause of his discharge.” Id. Bostock held that this type of discrimination is necessarily sex discrimination since it is “impossible” to discriminate on the basis of homosexual or transgender status without discriminating on the basis of sex. Id.

The employers in Bostock tried to evade this result by arguing that they were discriminating based on distinct criteria—sexual orientation and gender identity—rather than sex itself. Id. at 1746–48. But the Court was not moved. It explained that when an employer adopts a “sex-based rule[]” that “makes hiring turn on [sex], the employer

93

violates the law, whatever it might know or not know about individual applicants.” Id. at 1745–46. In other words, if the employer’s very policy holds a man and a woman in identical factual circumstances to different standards, then that employer discriminates based on sex. See id. at 1746. And how do we know when this has occurred? We know when the policy cannot be explained without reference to sex. See id. (“To see why, imagine an applicant doesn’t know what the words homosexual or transgender mean. Then try writing out instructions for who should check the box without using the words man, woman, or sex (or some synonym). It can’t be done.”). A rule against hiring homosexual or transgender people is a rule that tolerates behavior in members of one sex that it penalizes in members of another. Id. When an employer uses such a rule, it necessarily makes sex a but-for cause of its hiring decisions and thereby discriminates based on sex. Id.

Whether and how this translates into the Equal Protection context is not immediately obvious. Unlike Title VII, the Fourteenth Amendment does not use the language of but- for causation. See U.S. Const. amend. XIV, § 1 (“No state shall make or enforce any law which shall . . . deny to any person within its jurisdiction the equal protection of the laws.”). And both laws look different in operation. Once but-for discrimination has been shown, a Title VII claim is open and shut, absent the applicability of a statutory defense. Bostock, 140 S. Ct. at 1741 (“Title VII’s message is ‘simple but momentous’: An individual employee’s sex is ‘not relevant to the selection, evaluation, or compensation of employees.’” (quoting Price Waterhouse, 490 U.S. at 239)); see 42 U.S.C. §§ 2000e-1, -2. The Equal Protection Clause handles things differently, subjecting discriminatory laws to

94

tiers of scrutiny. See Shaw, 509 U.S. at 642 (“This Court never has held that race-conscious state decisionmaking is impermissible in all circumstances.”); Morrison v. Garraghty, 239 F.3d 648, 654 (4th Cir. 2001) (explaining that, once intentional discrimination is proven, “the court proceeds to determine whether the disparity in treatment can be justified under the requisite level of scrutiny”).

Despite these salient differences, there is nonetheless a crucial similarity between the two laws. At their cores, Title VII and the Equal Protection Clause both target the same conduct: treating people who are otherwise similarly situated differently because of their membership in a protected class. Compare Bostock, 140 S. Ct. at 1740 (“To ‘discriminate against’ a person, then, would seem to mean treating that individual worse than others who are similarly situated.” (quoting Burlington N. & S.F. Ry. Co. v. White, 548 U.S. 53, 59 (2006))), and id. (“[A]n employer who intentionally treats a person worse because of sex— such as by firing the person for actions or attributes it would tolerate in an individual of another sex—discriminates against that person in violation of Title VII.”), with City of Cleburne, 473 U.S. at 439 (explaining that the Equal Protection Clause is “essentially a direction that person similarly situated should be treated alike”), Yick Wo, 118 U.S. at 373– 74 (holding that laws effect “the denial of equal justice . . . within the prohibition of the constitution” when they “make unjust and illegal discriminations between persons in similar circumstances”), and Students for Fair Admissions, Inc. v. President & Fellows of Harvard Coll., 600 U.S. 181, 205 (2023) (“[T]he Constitution . . . forbids . . . discrimination by the General Government, or by the States, against any citizen because of

95

his [protected trait].” (quoting Bolling v. Sharpe, 347 U.S. 497, 499 (1954))). 21 And both are triggered if a person’s membership in the protected class was one reason—not necessarily the only or the primary reason—for their dissimilar treatment. Compare Bostock, 140 S. Ct. at 1741 (“If the employer intentionally relies in part on an individual employee’s sex when deciding to discharge the employee . . . a statutory violation has occurred.”), with Feeney, 442 U.S. at 276 (“The dispositive question, then, is whether the appellee has shown that a gender-based discriminatory purpose has, at least in some measure, shaped the [challenged] legislation.” (emphasis added)).

Asking whether a protected trait was a reason for discriminatory treatment is precisely what Bostock described as a but-for causation inquiry. 140 S. Ct. at 1739. If two people are otherwise similarly situated except for their sex, and they are treated differently

21

While the Equal Protection Clause and Title VII contain different words—most notably, that the latter includes the words “because of”—that distinction doesn’t give me pause, for two reasons. First, as explained, the Clause’s text has been interpreted to subject disparate treatment “on the basis of” a protected characteristic to heightened scrutiny. See, e.g., Washington, 426 U.S. at 239. Second, the Supreme Court has instructed that a provision doesn’t need the Equal Protection Clause’s precise wording to inform how and when the Clause prohibits discrimination. See, e.g., Arlington Heights, 429 U.S. at 266 (applying precedents interpreting the Fifteenth Amendment’s prohibition on denying the right to vote “on account of race” to understand the meaning of the Fourteenth Amendment’s Equal Protection Clause’s prohibition on “deny[ing] . . . the equal protection of the laws”).

Nor is it helpful to point to the author of Bostock’s concurring opinion in Students for Fair Admissions, 600 U.S. at 308 (Gorsuch, J., concurring), as some of my colleagues on other Circuits have done. See Williams, 83 F.4th at 484–85; Eknes-Tucker, 80 F.4th at 1229. While Bostock’s author did note several differences between Title VII and the Equal Protection Clause—including which actors they govern, which classes of people they protect, and whether discrimination can be justified under judicial scrutiny—he did not say that they demand different inquiries into whether intentional discrimination has occurred in the first place. See Students for Fair Admissions, 600 U.S. at 308–10 (Gorsuch, J., concurring).

96

because of their sex, then sex is a but-for cause of the result. Title VII and the Equal Protection Clause both prohibit this from occurring. They thus share a common inquiry into but-for causation.

Now to the punchline. The Equal Protection Clause requires a showing of but-for causation. Bostock gave us a test for identifying “traditional” and “simple” but-for causation. Id. It therefore follows that Bostock’s test can identify but-for causation under the Equal Protection Clause. A plaintiff can establish the first step of an Equal Protection claim by showing that they suffered intentional discrimination because of their protected trait. Washington, 426 U.S. at 239. And they can prove any such discrimination was because of that trait (i.e., but-for causation) by “chang[ing] one thing at a time and see[ing] if the outcome changes.” Bostock, 140 S. Ct. at 1739. If the outcome changes based on their protected trait, then that trait was a but-for cause of their mistreatment, and the burden ought to shift to the government to justify the law under heightened scrutiny. 22

22

Some might hesitate to make this connection absent a clearer mandate from the Supreme Court. But the Court instructs us to identify shared principles between laws with common elements, especially in antidiscrimination cases. In Bray, for instance, the Court used Equal Protection precedents to clarify the elements of discriminatory purpose in a federal cause of action. 506 U.S. at 271–73. It did so not because the statute automatically incorporated the constitutional standard but because common principles underlay both legal rules. See id. at 272 n.4. The Court did the same in General Electric Co. v. Gilbert, noting the “similarities” between Title VII and the Equal Protection Clause and using the latter as a “useful starting point in interpreting the former.” 429 U.S. at 133. Sometimes this works in the opposite direction, too. In Dobbs v. Jackson Women’s Health Organization, the Court used Bray—a statutory holding—to inform its analysis under the Equal Protection Clause. 597 U.S. at 236–37. So finding that Bostock’s but-for causation principles apply in Equal Protection cases is consistent with the method the Supreme Court has prescribed in similar cases.

97

This leads me to conclude that discrimination on the basis of homosexual or transgender status triggers heightened scrutiny under the Equal Protection Clause. Not because these groups constitute “quasi-suspect classes,” as the majority believes transgender persons do, 23 but rather because Bostock tells us that to discriminate on the basis of these traits is necessarily to discriminate “because of” sex. When sex is a but-for cause of official mistreatment, the Equal Protection Clause proscribes that action unless it can be justified under intermediate scrutiny. 24 Plaintiffs think that Bostock provides them another avenue for relief in these cases.

Yet by now, it should be clear why plaintiffs cannot show that their sex or transgender status was a but-for cause of any injury they suffered. Under Bostock, “if changing the [patient’s] sex would have yielded a different choice by the [states],” then the patient’s sex would be a but-for cause of their discrimination. Id. at 1741. But here, changing plaintiffs’ sex (or even their transgender status) would not change either state’s choice to decline coverage for the requested services. Even if we changed the biological sex of Maxwell Kadel—one of the plaintiffs below—from female to male, North Carolina would still deny

23

I am highly skeptical of our Circuit’s holding in Grimm that transgender individuals make up a quasi-suspect class. 972 F.3d at 610–13. For some of my reasons, see Williams, 83 F.4th at 486–88. Even so, this disagreement is immaterial, because I ultimately agree that laws targeting transgender individuals trigger intermediate scrutiny (since they are necessarily sex-based under Bostock).

24

Bostock also shows why discrimination by stereotype violates the Equal Protection Clause. A stereotype, as I have explained, is a generalization about the capabilities of and socially acceptable behavior for members of each sex. To discriminate based on such a generalization is therefore to tolerate behavior or attributes in members of one sex that one penalizes in members of the other sex, which Bostock said is sex discrimination. Bostock, 140 S. Ct. at 1741; cf. Price Waterhouse, 490 U.S. at 250–51.

98

Kadel coverage for a testosterone prescription. And even if we changed Christopher Fain’s biological sex from female to male, West Virginia would still deny Fain coverage for a mastectomy. So too if we changed their transgender identities. Both would still lack a qualifying diagnosis for the treatments. The only way that Kadel or Fain could get these treatments is if they had some other diagnosis (e.g., hypogonadotropic hypogonadism or cancer) that was covered. But if they had that other diagnosis, then they could obtain coverage for these treatments regardless of their sex or transgender status. Thus, a patient’s diagnosis, and not their sex or transgender status, is the but-for cause of their ability or inability to obtain coverage under both plans.

Nor do the plans discriminate via “sex-based rules” that necessarily make coverage “turn on” sex or transgender status. See id. at 1745–46. To see why, let’s return to the example Bostock used where an employer asks applicants to check a box if they are homosexual or transgender and then refuses to hire anyone who checks the box. Id. at 1746. Bostock held that this is sex discrimination, even if the employer never learns an individual’s sex, because the rule the employer uses holds men and women in the same factual circumstances to different standards (thus making sex a but-for cause of the discriminatory treatment). Id.

But now let’s modify the hypothetical. Imagine an employer announces that it will only hire a candidate with certain qualifications—a college degree, one year of experience, and two references—and that it will not consider other things in an application—such as race, religion, sex, sexual orientation, or gender identity—when making its hiring decision. To be clear, the employer will not deny employment based on these latter traits, either;

99

these traits just will not themselves qualify someone for the job. So, in this example, a transgender person who applies for the job and doesn’t have a college degree will not be hired. But neither will anyone else who lacks that or any other of the employer’s required qualifications. And a transgender person who has all the qualifications will be hired—just not because of their transgender identity.

The policy I’ve described does not discriminate on the basis of sex. It does not use a sex-based rule that holds men and women to different standards. Rather, it holds everyone to the same standard: Anyone who has the relevant qualifications will be hired, but no one will be hired simply because of their race, religion, sex, sexual orientation, or gender identity. Someone who has these latter traits can still be hired. But they will not be hired because they have these traits—they will be hired because they have the relevant qualifications, just like everyone else. The qualifications, and not the protected traits, are therefore the but-for cause of the hiring decision.

The challenged plans work a lot like this hiring policy. The states have decided that they will only pay for procedures that alter a patient’s breasts or genitalia if the patient suffers from physical injury, disease, or congenital absence of genitalia. Based on these criteria, the states identify a set of diagnoses that qualify someone for every treatment. They then grant coverage for those treatments only to people with qualifying diagnoses. As it turns out, gender dysphoria does not meet these criteria, so the states do not treat it as a qualifying diagnosis. Anyone who seeks to qualify for coverage on the basis of gender dysphoria alone thus will not receive treatment. Yet neither will anyone else who lacks a qualifying diagnosis, whether or not they have gender dysphoria. The only way anyone

100

receives coverage for a treatment is if they have a qualifying diagnosis. And if they have one, then they will receive coverage, regardless of their sex or transgender status, and even if they also happen to have gender dysphoria.

The plans, therefore, have not adopted a sex-based rule that makes coverage turn on a person’s sex or their transgender status. Indeed, they both can be described without reference to sex: No patient who seeks to alter their breasts or genitalia will receive coverage unless they experience physical injury, disease, or congenital absence of genitalia. Rather, the plans merely condition coverage for certain treatment on medical diagnosis. Anyone of either sex or who is transgender can obtain those treatments if they have a qualifying diagnosis. The exclusions therefore do not discriminate because of sex or transgender status.

4. The majority’s arguments are unpersuasive.

As I’ve explained, the challenged exclusions do not discriminate on the basis of sex or transgender status. So why does the majority conclude differently? Frankly, it’s hard to tell. Rather than beginning with an affirmative case for why these plans are discriminatory, the majority instead begins by refuting the states’ counterarguments. It then strings together a line of unrelated Supreme Court precedents to distinguish away Geduldig. Along the way, it announces various holdings with very little substantive analysis. I find none of these arguments remotely persuasive.

The main thrust of the majority opinion is that the plans use gender dysphoria as a proxy for transgender persons. The majority gives us several formulations of this conclusion, but each is essentially the same: Gender dysphoria is “virtually

101

indistinguishable” from “transgender status.” Majority Op. at 32. It is “inextricable” from transgender identity. Majority Op. at 37. And it is “unique” to transgender persons. Majority Op. at 39. To the majority, then, it is enough to know that gender dysphoria is closely related to transgender identity for us to conclude that “discriminating on the basis of diagnosis is discriminating on the basis of gender identity and sex.” Majority Op. at 23.

Conspicuously absent from the majority’s analysis, however, is any discussion of the actual legal standard for presuming intentional discrimination by proxy. As cases like Bray make clear, we cannot presume that a law intentionally discriminates just because the targeted activity is “engaged in exclusively or predominantly by a particular class of people”; it also must be “such an irrational object of disfavor that,” if targeted, “an intent to disfavor that class can be readily presumed.” Bray, 506 U.S. at 270; see also Arlington Heights, 429 U.S. at 266 (explaining that discrimination may be presumed if a classification is “unexplainable on grounds other than” a protected trait). This means we can presume intentional discrimination by proxy only if the distinction drawn is so obviously discriminatory that we can find an illicit purpose without requiring further evidence. See, e.g., Yick Wo, 118 U.S. at 374 (“[T]he conclusion cannot be resisted that no reason for [the unequal treatment] exists except hostility to the race and nationality to which the petitioners belong . . . .”); Guinn, 238 U.S. at 365 (finding no “basis of reason for the standard thus fixed other than” an intent to discriminate); Gomillion, 364 U.S. at 341 (explaining that “the conclusion would be irresistible, tantamount for all practical purposes to a mathematical demonstration,” that the electoral district was drawn to discriminate against black voters).

102

The majority does not engage with, let alone mention, this part of the standard. To be sure, it cites the outcomes that the Supreme Court reached in various proxy- discrimination cases. 25 See Majority Op. at 37–39. Yet it never mentions why the Court reached these outcomes—namely, that the challenged classifications were so obviously irrational that no other reason but a discriminatory purpose could explain them. Instead, without asking whether there might be rational, nondiscriminatory reasons to exclude coverage for gender-dysphoria treatments, the majority simply asserts that “it is enough to know that gender dysphoria, and therefore treatment for gender dysphoria, is unique to transgender individuals” in order to presume proxy discrimination. Majority Op. at 39.

But it is not, and has never been, “enough to know” that something targeted is “unique” to a protected class to presume that it is being used as a proxy for that class. It was not enough to know that only women can get pregnant for the Court to find that the refusal to cover pregnancy-related disabilities targeted women. Geduldig, 417 U.S. at 496

25

Here and elsewhere, the majority relies on Lawrence v. Texas, 539 U.S. 558 (2003), and Christian Legal Society v. Martinez, 561 U.S. 661 (2010). But Lawrence was not an Equal Protection case; indeed, the Court explicitly declined to rest its decision on this basis. See 539 U.S. at 574–75. And the majority’s heavily edited quotation of Christian Legal Society elides the limited scope of that holding. The Court did not say that it has “declined to distinguish between status and conduct in th[e] context [of discrimination]” generally. Majority Op. at 35. Rather, the Court said it has “declined to distinguish between status and conduct in this context,” Christian Legal Soc’y, 561 U.S. at 689 (emphasis added), i.e., in the context of policies that discriminate based on homosexual conduct. In other words, the Court determined that discrimination against persons who engage in homosexual conduct is discrimination against homosexual persons themselves. But that’s not what is happening here. The challenged exclusions do not prohibit anyone who cross dresses, for instance, from obtaining coverage. Instead, they simply decline to recognize a particular diagnosis as one that qualifies for certain treatments. So Christian Legal Society simply has no relevance here.

103

n.20 (“While it is true that only women can become pregnant[,] it does not follow that every legislative classification concerning pregnancy is a sex-based classification . . . . Absent a showing that distinctions involving pregnancy are mere pretexts designed to effect an invidious discrimination against the members of one sex or the other, lawmakers are constitutionally free to include or exclude pregnancy from the coverage of legislation such as this on any reasonable basis . . . .”). It was not enough to know that veterans were overwhelmingly men in the 1970s to find that a veteran hiring preference discriminated against women. Feeney, 442 U.S. at 274–75 (noting that “[a]lthough few women benefit from the preference,” the “legitimate and worthy purposes” for preferring veterans over non-veterans precluded a finding that the law intended to disadvantage women). And it was not enough to know that only women can get abortions to find that opposition to abortion targeted women. 26 Bray, 506 U.S. at 270 (holding that, though abortion is “engaged in exclusively” by women, discriminatory intent could not be presumed since there are “common and respectable reasons for opposing” abortion besides discriminatory intent towards women); see also Dobbs, 597 U.S. at 236. So why, exactly, is it enough to know that gender dysphoria is unique to transgender individuals for us to conclude that the

26

The majority frequently cites Bray’s statement that “[a] tax on wearing yarmulkes is a tax on Jews.” Bray, 506 U.S. at 270. But the reason such a tax is obviously discriminatory is because, unlike opposition to abortion, there is no rational basis to single out yarmulkes other than to discriminate against Jews. See id. The same is not true here.

104

plans use the former as a proxy for the latter? The majority does not say—it just asserts this to be so. Well, forgive me for remaining unpersuaded by mere assertion. 27 Equally shocking is the majority’s treatment of Geduldig. The majority acknowledges Geduldig’s holding that the choice to underinsure a particular medical condition is not sex discrimination, absent further evidence of pretext. But rather than wrestling with that holding, the majority states that Geduldig is inconsistent with another line of Supreme Court precedents, which hold that “a state cannot immunize itself from violating the Equal Protection Clause by discriminating against only a subset of a protected group.” Majority Op. at 31. So the majority simply limits Geduldig to its facts (i.e., pregnancy discrimination) and finds it inapplicable here.

Yet what these other cases have to do with Geduldig is an utter mystery. Each of them concerned whether policies already found to discriminate against members of a protected class were immune from heightened scrutiny because they targeted only a subset

27

The majority responds by citing several cases where the use of a proxy was “glaringly—facially—obvious.” Majority Op. at 41. Yet these were not proxy cases. Rather, each involved a law that facially classified based on sex in addition to other characteristics. See Califano, 442 U.S. at 83–89 (classifying based on employment status, parenthood, and sex); Morales-Santana, 582 U.S. at 58 (classifying based on citizenship, parenthood, and sex); Reed, 404 U.S. at 74–76 (classifying based on parenthood and sex); Orr, 440 U.S. at 278 (classifying based on marital status and sex). The challenged exclusions, by contrast, do not facially classify based on sex or transgender status. Nor do they even “use[] different words that mean the same thing,” like a person’s chromosomal makeup. Majority Op. at 42–43. Instead, they classify based on a single, facially neutral criteria—a class of medical treatments—and deny coverage to everyone who seeks those treatments—regardless of their sex or transgender status. That those treatments are predominantly or exclusively sought by transgender persons may serve as evidence of discriminatory impact, but it cannot by itself prove the existence of discriminatory intent.

105

of that protected class. See Graham v. Richardson, 403 U.S. 365, 366–67, 371–72 (1971) (facially distinguishing between the requirements for citizens and the suspect class of noncitizens); Weber v. Aetna Cas. & Sur. Co., 406 U.S. 164, 167–68 (1972) (facially imposing more requirements on illegitimate children than on legitimate children); Frontiero, 411 U.S. at 688 (“The sole basis of the classification established in the challenged statutes is the sex of the individuals involved.”); Mathews v. Lucas, 427 U.S. 495, 504 n.11 (1976) (facially discriminating against illegitimate children); Nyquist v. Mauclet, 432 US. 1, 3–4, 12 (1977) (facially discriminating between citizens and noncitizens); Rice, 528 U.S. at 498–99 (intentionally using ancestry as a proxy for race). Geduldig, by contrast, concerned whether discrimination on the basis of a suspect class occurs at all when a policy excludes coverage for something closely associated with members of a protected class. This is the question we must answer today. The states do not admit that they discriminate against transgender persons and then ask for lenience because they only target a subset of that community. They rather deny that any such discrimination has occurred in the first place. So the cases cited by the majority provide no reason to limit Geduldig to its facts and have no relevance to the question presented in these appeals. 28

28

The majority comes up with three additional reasons that distinguish Geduldig:

(1) the Supreme Court has only applied Geduldig in cases involving pregnancy discrimination; (2) unlike pregnancy, gender dysphoria is a proxy for transgender status; and (3) the plans engage in direct sex discrimination. See Majority Op. at 31–33. The first point is true, and yet proves nothing—that Geduldig has only been applied in cases involving pregnancy discrimination does not mean its reasoning is limited to such cases. The second point likewise fails. It is true that Geduldig “did not hold that a characteristic (Continued)

106

Why, then, does the majority imagine up this conflict of precedents? Probably because Geduldig, read fairly, obviously applies to the cases before us. See Majority Op. at 31 (“Appellants’ arguments . . . might be correct if we read Geduldig as broadly as possible.”). 29 As the Supreme Court recently explained in Dobbs, Geduldig established that “[t]he regulation of a medical procedure that only one sex can undergo does not trigger heightened constitutional scrutiny unless the regulation is a ‘mere pretext designed to effect an invidious discrimination against members of one sex or the other.’” 597 U.S. at 236 (alteration omitted) (quoting Geduldig, 417 U.S. at 496 n.20). As in Geduldig, the challenged plans here do not exclude a class of persons from coverage, but rather exclude coverage of treatments for a particular diagnosis. That only transgender persons happen to experience this diagnosis cannot alone support a finding of discriminatory intent, any more than the fact that only women can become pregnant could do so in Geduldig. 417 U.S. at

of a subset of a protected group cannot be a proxy for that group.” Majority Op. at 32. But this only proves that Geduldig might not control all cases—it does not prove why this case is distinguishable. As I have already explained, the mere fact that gender dysphoria relates to transgender status does not itself prove that the plans use it as a proxy for transgender status. In order to presume discriminatory intent, we must find that nothing else could explain the exclusions other than discriminatory intent—a finding the majority has not made and cannot make. And the third point will be addressed later. Suffice it to say this argument is equally unconvincing and an inadequate basis upon which to distinguish Geduldig.

29

After all, the majority’s argument is precisely the argument that the dissenting Justices made in Geduldig—and therefore precisely the argument that the majority in Geduldig rejected. Geduldig, 417 U.S. at 501 (Brennan, J., dissenting) (arguing that the exclusion discriminated based on sex by “singling out for less favorable treatment a gender- linked disability peculiar to women”).

107

496 n.20. Some evidence of pretext is needed—evidence that the majority lacks and does not even discuss.

The majority next contends that the fact that the exclusions apply equally to everyone doesn’t matter because they only affect transgender persons. Majority Op. at 33– 35. This is largely a repetition of the failed attempt at distinguishing Geduldig: Those that are impacted by the law all fall within the greater group of transgender people. But the fact that a law affects only a certain group of people does not itself mean that it discriminates based on membership in that group. At risk of sounding like a broken record, the Supreme Court has made this crystal clear: Disparate impact alone cannot alone sustain an Equal Protection claim. A plaintiff must offer some evidence of discriminatory intent or purpose to prevail. See Arlington Heights, 429 U.S. at 264–65 (“[O]fficial action will not be held unconstitutional solely because it results in a . . . disproportionate impact. . . . Proof of . . . discriminatory intent or purpose is required to show a violation of the Equal Protection Clause.”); Feeney, 442 U.S. at 272. Rather than identifying such evidence, however, the majority itself becomes a broken record, repeating over and over that gender dysphoria and transgender status are closely linked. 30 Yet no amount of repetition can turn nondiscrimination into discrimination.

Finally, the majority claims that the exclusions directly discriminate based on sex.

According to the majority, the plans cover certain “gender-affirming” surgeries “when the

30

The majority also cites McLaughlin v. Florida, 379 U.S. 184. But McLaughlin involved a facial classification that explicitly varied punishment based on whether couples were of the same or different race. Id. at 184–86. It therefore has no relevance to a case involving the disparate impact of a facially neutral policy, like ours.

108

purpose of the surgery is to align a patient’s gender presentation with their sex assigned at birth,” but not when “the purpose is to align a patient’s gender presentation with a gender identity that does not match their sex.” Majority Op. at 44. This, the majority claims, is “textbook sex discrimination, for two reasons.” Id. First, the exclusions cannot be applied without referencing sex. Id. Second, the exclusions are based on “gender stereotypes about how men or women should present.” Id. at 44–45.

Before responding to these arguments, I must again clarify how these plans actually work. Neither state covers surgery to alter breasts or genitalia for “gender-affirming purposes,” i.e., solely because a person wishes to align their outward appearance with their biological sex. Majority Op. at 44. The majority’s statement to the contrary simply is not supported by the record. To obtain coverage, a person must be afflicted with physical injury, disease, or congenital absence of genitalia. In other words, they must have a particular kind of qualifying diagnosis. Anyone can seek coverage for a vaginoplasty to correct the congenital absence of a vagina. Anyone can seek coverage for a breast reconstruction to restore what was destroyed by cancer treatment. And anyone can seek coverage for breast reduction to alleviate symptomatic gynecomastia. 31 But no one—man

31

The majority feigns ignorance as to why it is relevant that West Virginia only covers symptomatic gynecomastia. See Majority Op. at 44 n.26. But the relevance of this fact should be obvious. Like a person with gender dysphoria, a person with gynecomastia cannot obtain coverage for surgery because they wish to bring their body into line with how they believe it should appear. Rather, they can only receive it if they have physical symptoms, like breast pain—the very symptoms that gender dysphoria does not cause.

109

or woman, transgender or not—can seek coverage for these surgeries simply out of a desire to “affirm” their gender. 32 With that out of the way, it is easy to see why the majority’s first argument holds no water. The policies can be applied without reference to sex. Indeed, they are applied without reference to sex. The states do not use a person’s sex or transgender status to make coverage decisions. Instead, for each kind of surgery, the states keep a list of diagnoses that qualify someone for that surgery. When someone submits a coverage request, the states grant or deny coverage based on whether that person has a qualifying diagnosis. So for instance, if a person requests coverage from West Virginia for a vaginoplasty, whether they receive coverage or not depends on whether their diagnosis does or does not qualify. Nothing about this turns on a patient’s sex; the plans need only know whether the patient has a qualifying diagnosis. 33

32

The majority’s argument only works because it draws a line between “gender-

affirming” surgery and other kinds of surgery. So, for instance, it labels a mastectomy sought to treat diagnoses like gynecomastia or gender dysphoria as gender-affirming, but a mastectomy sought to treat cancer as something else. Majority Op. at 31 n.20. But this distinction is arbitrary and divorced from reality. A mastectomy for symptomatic gynecomastia is not performed to affirm a patient’s biological sex; it is aimed to treat the pain caused by a particular medical condition, just like a mastectomy to treat cancer.

To make this point even clearer, consider a female who naturally has little-to-no breast tissue. The lack of breast tissue is not a result of a diagnosed illness—it’s just genetics. She may want to obtain breast augmentation surgery in order for her body to align with what she views as a “female” body. But she wouldn’t get coverage for this “gender-affirming” care in either North Carolina or West Virginia. That lack of coverage is not because she is a female or because her gender-identity aligns with her sex; it’s because the reason she is seeking the surgery is not one covered by either plan.

33

The majority seems to think that because a third-party administrator must know a person’s diagnosis in order to make a coverage decision, and because they can infer a (Continued)

110

The majority can only label these policies as sex-based by reading medical diagnosis completely out of the picture. On the majority’s telling, the only difference between two people who request a vaginoplasty is sex—they are otherwise identical because both “were born without a vagina.” Majority Op. at 45. But phrasing it in these terms omits the medical reason they have this condition. One of them was “born without a vagina” in the sense that they have a congenital defect. The other was “born without a vagina” not because they have any such congenital defect, but because they have a diagnosed psychological disorder. These are not the same! Only by treating them as such can the majority sidestep the determinative role diagnosis plays and characterize these coverage decisions as necessarily sex-based.

The majority then contends that the exclusions discriminate based on gender stereotypes because they “condition[] access to gender-affirming surgery on whether the surgery will better align the patient’s gender presentation with their sex assigned at birth.” Majority Op. at 45. Yet this is the same error as before, just repackaged under a different label. Neither plan makes coverage available to anyone simply to “better align [their] gender presentation with their sex assigned at birth.” Id. Instead, they condition coverage based on whether a patient has a certain identifiable medical condition. And there is simply no evidence in these appeals that the states chose which conditions to cover with a view towards punishing gender-nonconformity.

patient’s sex from their diagnosis, the coverage decision itself is necessarily sex-based. See Majority Op. at 45. But a diagnosis and a person’s sex are not the same thing. That an administrator can infer a person’s sex from sex-neutral facts does not thereby mean they must know a person’s sex in order to make a coverage decision.

111

The majority rebukes the states for mistaking “what is for what must be.” Id. at 46.

It is the majority, however, and not the states, that has committed this error. States have finite resources to spend on healthcare, so they must prioritize those treatments that they deem cost-effective and medically necessary. As a result, they have chosen to cover treatment for some, but not all, diagnoses, while making treatment for those covered diagnoses available to all on an equal basis. The majority may disagree with this choice. But by castigating it as illicit discrimination, the majority imposes its own vision of what “must be” upon the states. This is not law—it is policy, plain and simple.

B. Medicaid Act Claims On top of their antidiscrimination challenges, the Anderson plaintiffs assert two claims under the Medicaid Act. 34 First, they allege that West Virginia’s program violates the Act’s “availability requirement,” which—in broad terms—requires states to cover certain categories of care under their Medicaid programs. See 42 U.S.C. § 1396a(a)(10)(A). Second, they contend that it violates the Act’s “comparability requirement,” which prevents states from discriminating between certain groups of Medicaid beneficiaries when covering care. See 42 U.S.C. § 1396a(a)(10)(B). Both arguments fail.

The Medicaid Act’s “availability requirement” is found in § 1396a(a)(10)(A). It says that participating states must “provide . . . for making medical assistance available” to

When I say the “Medicaid Act” I am referring to Title XIX of the Social Security

34

Amendments of 1965, Pub. L. No. 89-97, §§ 1901–05, 79 Stat. 286, 343–53, codified as amended at 42 U.S.C. §§ 1396 et seq.

112

eligible individuals, “including at least” an enumerated list of “care and services.” Id. That list is described in a different part of the statute and includes broad categories of care, like “inpatient hospital services,” “outpatient hospital services,” “rural health clinic services,” “laboratory and X-ray services,” and others. See § 1396d(a)(1)–(5), (13)(B), (17), (21), (28), (29)–(30). The Act’s “comparability requirement,” meanwhile, is found in the next subparagraph, § 1396a(a)(10)(B). That provision says that states must provide “that the medical assistance made available to any individual” covered by the availability requirement “shall not be less in amount, duration, or scope than the medical assistance made available to any other such individual.” Id.

Read alone, these provisions look sweeping. For instance, what does it mean that a state’s Medicaid program must “provide for making ‘inpatient’ and ‘outpatient hospital services’ available”? Does it mean that any time a categorically needy participant goes to the hospital asking for a procedure, the state must provide coverage, no matter what the procedure was or why the person wanted it?

No. The Supreme Court has made clear that these provisions of the Medicaid Act must be read alongside another provision—§ 1396a(a)(17)—which allows states to “include reasonable standards . . . for determining eligibility for and the extent of medical assistance under the plan which . . . are consistent with the objectives of this subchapter.” See Beal v. Doe, 432 U.S. 438, 444 (1977). This provision, the Court held, means that states have “broad discretion . . . to adopt standards for determining the extent of medical assistance” under their Medicaid programs, so long as those standards are “‘reasonable’ and ‘consistent with the objectives’ of the Act.” Id. And the Act’s “broadly stated primary

113

objective,” said the Court, is “to enable each State, as far as practicable, to furnish medical assistance to individuals whose income and resources are insufficient to meet the costs of necessary medical services.” Id. (emphasis added). 35 Notice what the Court did and did not say. Dicta notwithstanding, the Court did not hold that the purpose of the Act is to provide all medically necessary services to everyone who requests them. But see id. at 445 (suggesting that “serious statutory questions might be presented if a state Medicaid plan excluded necessary medical treatment from its coverage”). Rather, the Court held that the Act’s objective is for states, “as far as practicable,” to provide “medical assistance” to a certain category of people, i.e., those who cannot afford medically necessary services. See Preterm, Inc. v. Dukakis, 591 F.2d 121, 124 (1st Cir. 1979). The objective is thus to serve a certain population, not to provide a certain level of services for each and every person. See § 1396a(a)(19) (providing that coverage decisions must be in the “best interests of the recipients,” plural); accord Alexander v. Choate, 469 U.S. 287, 303 (1985) (“Medicaid programs do not guarantee that each recipient will receive that level of healthcare precisely tailored to his or her particular needs.”). And Medicaid—like every government-funded program—has limited resources. So the decision to spend money covering one procedure comes with a tradeoff: The program must forgo funding a different procedure, either now or later.

35

This purpose is reflected in the opening section of the Act. See § 1396-1 (“For the purpose of enabling each State, as far as practicable under the conditions in such State, to furnish (1) medical assistance on behalf of families with dependent children and of aged, blind, or disabled individuals, whose income and resources are insufficient to meet the costs of necessary medical services . . . .” (emphasis added)).

114

The resulting system is one where states have broad discretion to structure fiscally workable Medicaid programs to serve the interests of the Medicaid population as a whole. When making these decisions, a state will have to evaluate a procedure’s cost given the actual benefit that it provides to the recipient; and it will then have to compare that cost- to-benefit ratio to the same ratio for each of the alternative procedures that it could have provided other Medicaid recipients with the same money. Along the way, it will have to make tough judgment calls. Suppose that a patient has terminal cancer and that a procedure exists with a 10% chance of extending their life by a year at the cost of $5,000,000. Does the state cover the procedure? What if it could use that money to cover 500 cataract surgeries instead? How, after all, does one quantify the “benefit” of a procedure? The Act does not supply a one-size-fits-all answer. Instead, it simply requires that—wherever each state ultimately decides to draw the line—the decision be “reasonable.”

Thus, read in light of § 1396a(a)(17), the Act’s “availability” and “comparability”

requirements each impose a “reasonableness” test much like a rational-basis test. Whenever a state’s Medicaid program limits coverage for a procedure that would otherwise fall within § 1396a(a)(10)(A)’s enumerated list, its decision to do so must be “reasonable.” Similarly, when a state decides that some Medicaid participants get coverage for a given procedure, but that other participants do not, that decision must likewise be “reasonable.” And like under rational-basis review, when I say “reasonable,” I mean objectively reasonable. In other words, the state must merely provide a justification for its decision— which may be after-the-fact—that could lead a reasonable person to believe that the decision was made in the “best interests” of the state’s Medicaid recipients as a whole.

115

The Act’s implementing regulations support this reading. Title 42 C.F.R.

§ 440.230(b) states that a service provided by a state plan must be “sufficient in amount, duration, and scope to reasonably achieve its purpose.” Neither the regulations nor the statute, however, define what the purpose of any individual service is, so its purpose can be understood only in relation to the broader purpose of the Act—“furnish[ing] medical assistance to individuals whose income and resources are insufficient to meet the costs of necessary medical services.” Beal, 432 U.S. at 444. And by including the word “reasonably,” the regulation does no more than restate what the statute and Beal already told us: Decisions about the extent of coverage must be reasonable and in line with the statute’s purpose. § 440.230(b).

Similarly, § 440.230(d) establishes that the state Medicaid agency may “place appropriate limits on a service based on such criteria as medical necessity or on utilization control procedures.” This simply offers a non-exhaustive list of factors that states may consider when determining how to limit the amount, duration, or scope of a provided service. Paragraph 440.240(b)(1) then provides that once a state chooses to make certain services available, it must “provide that the services available to any individual in the [categorically needy group] are equal in amount, duration, and scope for all beneficiaries within the [categorically needy group].” Again, this just restates the comparability requirement, which we already know gives states broad discretion to make reasonable coverage decisions.

Subsection 440.230(c), which provides that a state “may not arbitrarily deny or reduce the amount, duration, or scope of a required service . . . to an otherwise eligible

116

beneficiary solely because of the diagnosis, type of illness, or condition,” does not compel a contrary reading. This regulation constrains a state’s ability to make initial coverage decisions reducing available services based on arbitrary limits turning solely on a beneficiary’s diagnosis. For instance, if a state has a general rule that it covers outpatient hospital services for all dental surgeries but decides not to cover outpatient hospital services for surgeries to treat gingivitis, a patient seeking gingivitis care would be “otherwise eligible” for coverage under the plan but for their gingivitis diagnosis. And the state would have “den[ied] or reduce[d] the amount, duration, or scope” of available services to that patient “solely because of the diagnosis.” But § 440.230(c) permits a state Medicaid agency to make such limitations, even those based “solely” on a particular “diagnosis, type of illness, or condition” as long as the decision is “not arbitrarily” (i.e., reasonably) made. 36 In other words, nothing in this regulation prohibits states from considering diagnosis when making their initial eligibility determinations, so long as those determinations are—you guessed it—reasonable.

West Virginia easily fulfills its obligations under the Medicaid Act. Although specific, empirical data is not required, the state presented undisputed evidence that its Medicaid program had limited funds and that covering plaintiffs’ surgeries would require

36

Admittedly, it is not altogether clear whether this regulation actually does limit what a state can consider when initially determining what makes someone eligible for a particular service. It could rather be read as requiring that once a state has deemed someone eligible for that service—i.e., once they are “otherwise eligible”—it cannot arbitrarily limit their access to that service solely because of another diagnosis. That is, if a state decides that a person must have five diagnostic markers to be eligible for heart surgery, it cannot deny surgery to someone with those markers just because they also happen to have, say, depression.

117

it to either “cut existing services or receive additional appropriations from the legislature.” Anderson, J.A. 1203–04. Given that fact, the state may make its own judgment about the relative value of the surgeries plaintiffs request and the other procedures that it could use the same money to cover, provided that its judgment is not arbitrary. And a state might reasonably conclude that the value these procedures provide in treating some diagnoses is higher than any value that surgery has in treating gender dysphoria. See Anderson, J.A. 1860–1935 (Expert Disclosure Report of Dr. Stephen B. Levine, M.D) (questioning the benefit of such surgery). For instance, a mastectomy might be used to treat both breast cancer and gender dysphoria. But the state might reasonably conclude that covering the former will benefit the Medicaid population as a whole more than covering the latter.

The majority, in concluding otherwise, does not even try to parse the text of the statute or its implementing regulations. Instead, the majority just declares, relying on Supreme Court dicta and out-of-circuit precedent, that a state may only exclude services based on comparability of medical need and not based on the underlying diagnosis. See Majority Op. at 60. But this is an absurd reading of the statute. The purpose of the Medicaid Act is not to provide all attainable medically necessary services but to provide medical services for the Medicaid population as a whole, so far as feasible. And neither the statute nor its regulations prohibit states from limiting coverage, so long as those limits are reasonable and consistent with the objectives of the Act. There is therefore no requirement that states provide equal services to everyone with the same level of medical need. Coverage distinctions need only be reasonable. And they certainly are here.

118

* * *

Today’s result is a victory for plaintiffs but a defeat for the rule of law. To reach its holding, the majority misconstrues the challenged policies and steamrolls over the careful distinctions embedded in Equal Protection doctrine. It finds unlawful discrimination where there is none, stripping the states of their prerogative to create health-insurance and Medicaid systems that serve the best interests of their overall populations.

More troubling, however, are the implications of today’s result for future cases involving state classifications in the healthcare context. Running a healthcare system is no easy task. Because the states have limited resources, they must make hard judgment calls about which services they will and will not cover. Ordinarily, such line-drawing is of no concern to the Equal Protection Clause. It is only in a narrow set of cases—when lines are drawn based on membership in a protected class—that heightened scrutiny is triggered. It is therefore incumbent on those in robes to exercise caution before jumping to conclusions about the reasons for particular judgements and distinctions drawn in the medical field. In failing to heed this warning, the majority sets a dangerous precedent and threatens the feasibility of state regulation in this area.

I thus respectfully dissent.

119

WILKINSON, Circuit Judge, dissenting:

Why the rush to constitutionalize? Why the dash to create a substantive Fourteenth Amendment right to transgender surgery and treatment underwritten by the State?

Of course the controversies surrounding transgender status will reach the courts. But how they reach us is the all-important thing. There is a big difference between, say, reading a statute and discovering a novel unenumerated constitutional right.

I see no need to revisit the debates swirling over Roe v. Wade, 410 U.S. 113 (1973).

I should note only that the infirmity of that decision lay not in the shortcomings of a perspective protecting the rights of the unborn or of one safeguarding reproductive freedoms. No, the infirmity lay in the courts reserving the weighing and balancing of those heartfelt perspectives for themselves.

There will, of course, always be those who applaud and those who decry the decision of the day. But that is transient, much as a fleeting goldfinch wings before our eyes. And in the long tomorrow, the recurrent creation of rights so unmoored from constitutional text or history will deplete the store of public respect on which a branch devoid of sword or purse must ultimately rely. See The Federalist No. 78 (Alexander Hamilton). Courts have been thrust into an unprecedented and transparently political thicket from which extrication has proven uncommonly hard.

And yet here we go again. We now confront a lengthy majority opinion without limits on what other statutory dominos will fall. In the era of Roe, it was substantive due process. Now it is substantive equal protection. Make no mistake. The fundamental rights prong of equal protection is what is at play here, and while constitutionally mandating

120

state-funded transgender rights will please some, it will politicize the courts in the eyes of all as assuredly as its substantive due process predecessor did.

Had the majority’s result been reached through the democratic process, it would have been perceived as the product of a process in which many good people of many varied views had had their voices heard. But even those who most passionately approve of the outcome here must recognize that those who do not approve have been ever so wrongly denied their rightful say. Even more so than in Roe, because that decision was never thought to require public funding of reproductive freedoms, Maher v. Roe, 432 U.S. 464, 480 (1977), whereas this decision presumes to dictate how public officials should prioritize the competing requests of deserving claimants for insurance coverage and financial support.

This is all transparently a creative, not an interpretive, judicial exercise, one which is most aptly termed constitutional common law. But even the great common law judges could always be overturned by a legislature, whereas we, their descendants, hold ourselves above amendment by the States, the Congress, or indeed any agency which dares murmur a dissent.

This is imperial judging at its least defensible. It is the law, we say. Why? Because we proclaim it so. I suppose that one day we shall exchange our robes of black for a purple more befitting our new regal state. But until that time, a basic respect for the legitimate and diverse views of our fellow Americans should prevail. Because I believe that ours should not be the first, last, and only word on this volatile set of issues, I respectfully dissent.

I.

121

Plaintiffs put forth claims on the medical necessity of hormonal and surgical treatments for gender dysphoria, a condition they say is ineluctably intertwined with transgender identity. The North Carolina State Health Plan, for example, excludes these very treatments from coverage by prohibiting reimbursement of “treatment . . . leading to or in connection with sex changes or modifications.” J.A. 181. Plaintiffs insist this exclusion is a facial classification based on sex. They further contend that the exclusion constitutes sex-based discrimination because it punishes transgender individuals for failing to conform to sex stereotypes. And they assert that the exclusion evinces an invidious intent to discriminate against transgender people by targeting individuals with gender dysphoria. Under the Equal Protection Clause, then, plaintiffs claim that the exclusion must survive heightened scrutiny. This, they tell us, it cannot do.

These arguments, whether alone or in combination, fail to show that the coverage exclusion constitutes an equal protection violation. What plaintiffs propose is nothing less than to use the Constitution to establish a nationwide mandate that States pay for emerging gender dysphoria treatments. Plaintiffs envision an Equal Protection Clause that is dogmatic and inflexible, one that leaves little room for a national dialogue about relatively novel treatments with substantial medical and moral implications. Plaintiffs’ clause would encroach on a State’s prerogative under its basic police power to safeguard the health and welfare of its citizens. I would resist allowing the Equal Protection Clause to expand to such proportions, bloating the judicial power commensurately. The gender dysphoria treatments at issue—including puberty blocking drugs, cross-sex hormones, and gender reassignment surgery—are matters of significant scientific debate and uncertainty. As

122

such, the arguments made before this court are advanced in the wrong forum. The right forum is a legislative hearing.

It is true, of course, that the Equal Protection Clause applies to the States and supplants offending state enactments. The Supreme Court’s ruling striking down the patently dehumanizing practice of state-enforced segregation is only one of many such examples. The moral tone struck by Brown v. Board of Education, 347 U.S. 483 (1954), rang clear as a bell. Many subsequent cases expanded Brown beyond education to other facets of life, and beyond race to other suspect and quasi-suspect classifications. See Frontiero v. Richardson, 411 U.S. 677 (1973) (heightened scrutiny for sex classifications); Graham v. Richardson, 403 U.S. 365 (1971) (alienage); Oyama v. California, 332 U.S. 633 (1948) (nationality).

Few, if any, of those steps involved this litigation’s mix of medicine and morality at such an incipient and experimental stage. To say the Equal Protection Clause supplies only one answer to issues where parties advance legitimate but deeply conflicting views is to ascribe to the Fourteenth Amendment a power over subjects on which its Framers had very little to say. We cannot ask our Constitution for answers which it does not have and which it cannot give. The Framers expected the people of a great nation to figure out many great issues for themselves.

II.

The Supreme Court has repeatedly acknowledged the broad discretion given to the States in the allocation of public benefits. As the Court has emphasized, “the Fourteenth Amendment gives the federal courts no power to impose upon the States their views of

123

what constitutes wise economic or social policy.” Dandridge v. Williams, 397 U.S. 471, 486 (1970). Indeed, “the intractable economic, social, and even philosophical problems presented by public welfare assistance programs are not the business of th[e] Court,” as “the Constitution does not empower th[e] Court to second-guess State officials charged with the difficult responsibility of allocating limited public welfare funds among the myriad of potential recipients.” Id. at 487.

Thus “[i]n the area of economics and social welfare, a State does not violate the Equal Protection Clause merely because the classifications made by its laws are imperfect.” Id. at 485; see also Lindsley v. Nat. Carbonic Gas Co., 220 U.S. 61, 78 (1911) (“A classification having some reasonable basis does not offend against [the Equal Protection Clause] merely because it is not made with mathematical nicety, or because in practice it results in some inequality.”). The State’s allocation of benefits must simply be rational, a judgment to which we owe great deference.

As my colleague Judge Richardson cogently demonstrates, Geduldig v. Aiello provides the proper framework for this case. 417 U.S. 484 (1974). The Court held that the Equal Protection Clause does not require state insurance programs to protect against specific health risks, even risks that are only experienced by one sex. Id. at 494–96. Because the Plan includes “no risk from which men are protected and women are not,” and vice versa, it does not constitute sex-based discrimination under the Equal Protection Clause. Id. at 496–97. And plaintiffs have failed to put forth evidence that the Plan’s exclusion for gender dysphoria care was motivated by invidious intent.

124

By sidestepping Geduldig, the majority negates the ability of the State to select which procedures, operations, and health risks it insures. The majority insists Geduldig is inapplicable because the exclusion is facially discriminatory. But the neutrality of the provision is readily apparent. The exclusion treats males and females, cisgender individuals and transgender individuals, precisely the same. It merely removes one medical condition, gender dysphoria, from coverage. As Geduldig made clear, “[t]here is nothing in the Constitution . . . that requires the State to subordinate or compromise its legitimate interests solely to create a more comprehensive social insurance program than it already has.” Id. at 496.

The majority, however, sees things differently. It arrogates to itself the authority to tell States how to draft insurance policies covering state employees on state healthcare plans. This is a breach of our federal system. It is an intrusion upon the residual powers that the Constitution guarantees to the States. It is a usurpation of the prerogatives of fifty sovereigns, supplanting difficult judgments on issues in their very infancy with an ill- advised, self-assured ukase of our own.

III.

While the amicus briefs before us are thoughtful and edifying, they also underscore the impropriety of constitutionalizing this complex issue. The brief by the American Medical Association is particularly revealing. See Br. for Am. Med. Ass’n et al. as Amici Curiae Supporting Plaintiffs-Appellees, Kadel v. Folwell (No. 22-1721). It elucidates the healthcare profession’s understanding of advances in treating gender dysphoria. Id. at 10. And it stresses the detrimental consequences that a lack of treatment could have on the

125

wellbeing of individuals struggling with this condition. Id. at 14–15. The brief of States supporting Plaintiffs is equally enlightening. See Br. for New York et al. as Amici Curiae Supporting Plaintiffs-Appellees, Kadel v. Folwell (No. 22-1721). It traces the steps amici are taking to increase access to gender dysphoria care and the benefits their citizens have reaped from these state policies. Id. at 6–16.

I do not disparage the importance of this information. Amici make clear that gender dysphoria is a serious condition which, left untreated, can result in real harm to affected individuals. But the briefs fail to answer the question of why this court ought to find the Plan’s exclusion contrary to the Constitution. Rather, the information methodically presented by our good amici is a classic legislative argument. It presents but one view of a highly disputed matter, and that view must compete for funding with other poignant and deserving claims for state insurance coverage.

Other States present other views. See Br. for Missouri et al. as Amici Curiae Supporting Defendants-Appellants, Kadel v. Folwell (No. 22-1721). There we are reminded that states have significant discretion in areas affecting the health and welfare of their citizens, especially those areas where the science is unsettled. Id. at 3. Healthcare costs stress state budgets mightily. See Appellants’ Opening Br. 2. Whether States should pay for emerging hormonal and surgical interventions to treat gender dysphoria is unclear when so many diseases visit such tragic consequences upon their victims.

As the Missouri brief also makes clear, the science behind gender dysphoria care is far from settled. See Br. for Missouri et al. at 6–11. A recent systematic review of cross- sex hormone treatments for minors revealed that “long-term studies are lacking” and “long-

126

term effects of hormone therapy on psychosocial and somatic health are unknown.” Jonas F. Ludvigsson et al., A Systematic Review of Hormone Treatment for Children with Gender Dysphoria and Recommendations for Research, Acta Paediatrica, Apr. 2023, at 12. Many European nations have questioned the wisdom of hormonal and surgical interventions, particularly when used to treat children. For instance, Finnish medical authorities stress that, when it comes to youth struggling with gender dysphoria, “there is no medical treatment that can be considered evidence-based,” and that “gender reassignment of minors is an experimental practice.” Council for Choices in Health Care, Medical Treatment Methods for Dysphoria Related to Gender Variance in Minors 6, 8 (2020). Likewise, the French Academy of Medicine urges doctors to prioritize psychological support for adolescents identifying as transgender, as the alternative therapies can come with “many undesirable effects, and even serious complications.” Press Release, French Nat’l Acad. Med., Medicine and Gender Transidentity in Children and Adolescents (Feb. 25, 2022); see also Jennifer Block, Gender Dysphoria in Young People is Rising—And So Is Professional Disagreement, BMJ, Feb. 2023, at 1–4.

These different sets of briefs offering their different perspectives illustrate perfectly why the whole issue should be left to percolate in what Justice Brandeis famously called the laboratories of democracy. New State Ice Co. v. Liebmann, 285 U.S. 262, 311 (1932) (Brandeis, J., dissenting). Providing the best possible care to adults and youth struggling with gender dysphoria is a challenging task for our States. But it is one that they are entitled to perform without premature judicial interference. It will require them to engage in rigorous cost-benefit analyses, community outreach, and expert consultation. It is almost

127

certain no two approaches will look the same—a testament to the rich variety in policy our federalist system encourages. Indeed, even the amici States supporting Plaintiffs have not taken a uniform approach to gender dysphoria care. For instance, Nevada’s state employee insurance plan contains certain limitations on gender dysphoria care, while California’s plan provides full coverage. Compare Nevada Public Employees’ Benefits Program, Consumer Driven Healthcare Master Plan Document: Plan Year 2023, 57 (2022), with Blue Shield of California, Trio HMO Basic Plan: Plan Year 2023, 24 (2023).

States have “wide discretion to pass legislation in areas where there is medical and scientific uncertainty.” Gonzales v. Carhart, 550 U.S. 124, 163 (2007) (citing Kansas v. Hendricks, 521 U.S. 346, 360 n.3 (1997)). Yet the majority wrests this discretion out of the hands of North Carolina, West Virginia, and untold other states besides. Self-governance is notably absent when the many voices seeking to provide answers are silenced by federal judges shrouded in an authority of their own design.

IV.

The parties and amici lay bare a dilemma with implications that could not be more profound. On the one hand, we have the powerful arguments of transgender men and women for dignity and open access to desired medical care. This side of the argument is not merely about diagnostic codes and treatment plans. At base, we encounter individuals on a quest for wholeness, for a sense of self which is not fractured, for a quelling of deep tumult and conflict within. Courts must respect those who wish only to become more fully themselves.

128

There is, however, another side. Some States are reluctant to fund emerging treatments until the science can tell us more. Not only is the medical data conflicting, but there is a moral caution in this case as well. Self-righteous folly has long run through us all. The Tower of Babel toppled of its own hubristic weight. Yet still we moderns strive to bend nature to desire. The quest is too important to be left to science and technology alone. “If humanity wants to survive technology, [J. Robert Oppenheimer] believed, it needs to pay attention not only to technology but also to ethics, religion, values, forms of political and social organization, and even feelings and emotions.” See David Nirenberg, J. Robert Oppenheimer’s Defense of Humanity, WSJ, July 15–16, 2023 at C5. That is democracy in action. The untutored and the lettered alike must have their say. Those who wear no robe must not be shunted to the sidelines.

Where to draw the line? How to refashion our beings tomorrow? When is the Rubicon between healing and remaking ourselves irrevocably crossed? What improvements to the handiwork of nature shall we next seek? What ever-receding horizons of happiness shall greet the elusive search for the more perfect self?

The majority and the dissents have no answer to these questions, at least none if we are honest with ourselves. Science is a discipline of many wonders, but also of many limits. We have seen medical breakthroughs and medical overreach, and human history is rife with the triumphs and failures of judgment and morality. The Framers gave us no sure answers to transgender treatments or indeed to many questions confronting succeeding generations. Their gift to us is one of process, and a priceless gift it is. Our Constitution directs that controversies such as these must be hashed out over time by the people and

129

their chosen representatives. The glories of our federalist system are laid before us in these dueling briefs, and we must heed their implicit, collective call. What substance the Constitution does not resolve, the democratic process, along its halting and imperfect paths, yet may.

130

QUATTLEBAUM, Circuit Judge, dissenting:

We do not—or, at least, we should not—bend the Federal Rules of Evidence just because a case involves important constitutional issues. But that is what the majority seems to be doing here. In order to conclude that no legitimate, non-discriminatory reasons support denying coverage for certain treatments of gender dysphoria, the majority abandons settled evidentiary principles. Properly accounting for the record, questions about the medical necessity and efficacy of such treatments linger. And those lingering questions support the states’ coverage decisions.

In its first improper evidentiary move, the majority misapplies Federal Rule of Evidence 702 by affirming the exclusion of Dr. Paul W. Hruz’s gender dysphoria testimony. That exclusion kept evidence of the debate concerning the medical necessity and efficacy about the treatment the plaintiffs seek out of the record.

In its second evidentiary misstep, the majority improperly declares as fact the plaintiffs’ position on this debate. It first states as a fact that “[i]f untreated, gender dysphoria can cause debilitating distress, depression, impairment of function, self- mutilation to alter one’s genitals or secondary sex characteristics, other self-injurious behaviors, and suicide.” Maj. Op. at 12–13. In making this declaration, the majority cites to the fifth edition of the American Psychiatric Association’s Diagnostic and Statistical Manual of Mental Disorders (DSM-5). The majority then states as a fact that “the medical community uses generally accepted protocols from the World Professional Association for Transgender Health’s Standards of Care for the Health of Transgender and Gender Diverse People (WPATH Standards), which it explains recommend “assessment,

131

counseling, and, as appropriate, social transition, hormone therapy, and surgical interventions to bring the body into alignment with one’s gender identity.” Maj. Op 13–14 (quoting Br. of Medical Amici, which cites the WPATH Standards). Despite these declarations of fact, the record reveals that there is a dispute within the medical community on these two points.

Why the evidentiary shortcuts? When the dust settles from our court’s equal protection debate, lawyers and district courts will see that we have applied the Federal Rules of Evidence in ways at odds with their textual requirements and our precedent interpreting them. So, questions naturally follow. Do we cut evidentiary corners when the constitutional stakes are high? Or have we altered evidentiary norms? The answer is not clear to me. But what is clear is that these evidentiary decisions improperly stack the deck against West Virginia and North Carolina. So, in addition to the reasons articulated in Judge Richardson’s dissenting opinion, I respectfully dissent. 1

I.

Before the district court, North Carolina sought to introduce Dr. Hruz as an expert to testify about the treatment of gender dysphoria. As a pediatric endocrinologist, Dr. Hruz has “participated in the care of hundreds of infants and children, including adolescents,

1

My only deviation from Judge Richardson’s dissent is that I would assume, without deciding, that Bostock v. Clayton County, 590 U.S. 644 (2020) applies to the Equal Protection Clause. Assuming it does, I join in Judge Richardson’s analysis and conclusion that the plaintiffs have not established but-for causation.

132

with disorders of sexual development.” Kadel, J.A. 737. In this role, he has treated hormone-related conditions in patients with gender dysphoria, including obesity, diabetes and dyslipidemia associated with gender dysphoria treatment. He has “participated in local and national meetings where the endocrine care of children with gender dysphoria has been discussed in detail and debated in depth.” Kadel, J.A. 737. He has also “consulted with, met with, and had detailed discussions with dozens of parents of children with gender dysphoria to understand the unique difficulties experienced by [that] patient population.” Kadel, J.A. 737. Additionally, Dr. Hruz has given grand round presentations 2 regarding gender dysphoria at major universities’ medical centers. And he has previously testified as an expert witness in litigation concerning issues of sex and gender.

Despite this the district court determined that Dr. Hruz was not qualified to testify about “the diagnosis of gender dysphoria, the DSM, gender dysphoria’s potential causes, the likelihood that a patient will ‘desist,’ or the efficacy of mental health treatments.” Kadel, J.A. 3587. The district court reasoned that Dr. “Hruz is not a psychiatrist, psychologist, or mental healthcare professional.” Kadel, J.A. 3587. The district court also reasoned that Dr. Hruz “has never diagnosed a patient with gender dysphoria, treated gender dysphoria, treated a transgender patient, conducted any original

2

As explained by Dr. Hruz, “Grand rounds are usually a recurring series of talks given by experts in various fields to the relevant scientific community about topics of interests to those physicians. And it generally involves the presentation of high quality scientific evidence for the conditions that those physicians in the audience would encounter.” Kadel, J.A. 1257.

133

research about gender dysphoria diagnosis or its causes, or published any scientific, peer- reviewed literature on gender dysphoria.” Kadel, J.A. 3587.

On appeal, the majority concludes that the district court did not abuse its discretion in limiting Dr. Hruz’s expert testimony. I disagree. Appropriately, we give district courts discretion in exercising their gatekeeping function under Federal Rule of Evidence 702, which governs the admissibility of expert testimony. Belk, Inc. v. Meyer Corp., 679 F.3d 146, 162 (4th Cir. 2012) (quoting Kumho Tire Co. v. Carmichael, 526 U.S. 137, 156 (1999)). But that discretion does not permit ignoring the plain language of Rule 702.

Rule 702 provides:

A witness who is qualified as an expert by knowledge, skill, experience, training, or education may testify in the form of an opinion or otherwise if:

(a) the expert’s scientific, technical, or other specialized knowledge will help the trier of fact to understand the evidence or to determine a fact in issue;

(b) the testimony is based on sufficient facts or data;

(c) the testimony is the product of reliable principles and methods; and (d) the expert has reliably applied the principles and methods to the facts of the case.

Fed. R. Evid. 702 (2011). 3 Relevant to qualifications, our Court has held that because Rule 702 “uses the disjunctive, a person may qualify to render expert testimony in any one of

3

In December 2023, the Advisory Committee amended Rule 702 to read, in relevant part, “A witness who is qualified as an expert by knowledge, skill, experience, training, or education may testify in the form of an opinion or otherwise if the proponent demonstrates to the court that it is more likely than not that” the Rule 702’s other conditions are satisfied. Fed. R. Evid. 702 (2023) (emphasis added). Given the district court necessarily applied the prior version of Rule 702, so must our Court. However, my analysis does not change even if this newly amended version of Rule 702 applies. My focus, much like the district court’s and the majority’s, is whether Dr. Hruz was qualified to offer expert testimony in the first (Continued)

134

the five ways listed: knowledge, skill, experience, training, or education.” Kopf v. Skyrm, 993 F.2d 374, 377 (4th Cir. 1993) (emphasis added). This means that an expert should be able to testify on the basis of knowledge alone, independent of experience or education.

Although he does not treat patients for “the purpose of alleviating gender dysphoria,” Dr. Hruz is an endocrinologist who has treated hundreds of juveniles diagnosed with sexual development disorders and many transgender “patients that have experienced side effects related to . . . hormone treatment.” Kadel, J.A.737, 1256. He has also “extensively studied” scientific literature on gender dysphoria treatments while his hospital developed a transgender clinic, consulted with professionals specializing in this area, presented on gender dysphoria at medical universities and met with “dozens of parents of children with gender dysphoria to understand the unique difficulties experienced by this patient population.” Kadel, J.A. at 737. Dr. Hruz, therefore, has the necessary knowledge to qualify him to testify on the subject of gender dysphoria.

Knowledge is supposed to be an independent basis that qualifies an expert to testify.

See Kopf, 993 F.2d at 377. And given Dr. Hruz’s knowledge qualified him to testify about gender dysphoria, concerns of his lack of experience in diagnosing, treating or researching gender dysphoria went to the weight of his proffered testimony, not its admissibility. See Fed. R. Evid. 702 (2011); United States v. Fuertes, 805 F.3d 485, 496 (4th Cir. 2015). The

place—a threshold question preceding the inquiry into whether Rule 702’s other conditions are also met.

135

district court abused its discretion in excluding Dr. Hruz’s testimony about gender dysphoria.

In addition, the district court’s exclusion of Dr. Hruz’s testimony on the basis of his qualifications conflicts with the way our circuit has traditionally reviewed decisions about the admissibility of expert witness testimony. For decades, we have recognized that “qualifications to render an expert opinion are . . . liberally judged by Rule 702.” Kopf, 993 F.2d at 377; see also Fuertes, 805 F.3d at 496. And where an expert’s qualifications are challenged, we have stated that “the test for exclusion is a strict one, and the purported expert must have neither satisfactory knowledge, skill, experience, training nor education on the issue for which the opinion is proffered.” Kopf, 993 F.2d at 377 (quoting Thomas J. Kline, Inc. v. Lorillard, Inc., 878 F.2d 791, 799 (4th Cir. 1989)). Importantly, “[o]ne knowledgeable about a particular subject need not be precisely informed about all details of the issues raised in order to offer an opinion.” Id. (quoting Thomas J. Kline, 878 F.2d at 799).

Our precedent demonstrates a more relaxed construction of Rule 702. In Garrett v.

Desa Industries, Inc., 705 F.2d 721, 724 (4th Cir. 1983), we determined that the district court abused its discretion in prohibiting a Navy gunnery officer from testifying about the “design and manufacture” of stud drivers “simply because he lacked one of the five qualifications, namely, prior experience” with the tool. In recognizing that the officer had two engineering degrees, worked as a professional engineer, and worked with handguns that operated similarly to stud drivers, we held that he was “qualified by his education, knowledge, training, and skill.” Id. at 724–25.

136

And in Gober v. Revlon, Inc., 317 F.2d 47, 52 (4th Cir. 1963), we found that the district court did not err in admitting the testimony of the plaintiff’s expert witness, a dermatologist whom she called to testify about whether Revlon’s nail polish caused her painful skin condition. The dermatologist “testified as to certain matters concerning chemicals.” Id. Though Revlon argued that the dermatologist’s testimony about chemical matters should have been stricken because he “was not qualified as a chemical expert,” we disagreed. Id. We reasoned that the dermatologist “was testifying as a dermatological expert to the reaction of humans to certain chemicals. Certainly this is within the scope of his medical qualifications. His lack of qualifications as a chemist went to the weight of his testimony, not its admissibility.” Id. (emphasis added).

More recently, in Fuertes, we determined that the district court did not abuse its discretion in permitting a physician who served as the director of a child abuse center to testify as an expert in a trial concerning the alleged sex trafficking of adults. 805 F.3d at 496–97. Though the criminal defendants argued on appeal that the physician was not qualified to testify as an expert because her experience was limited to working with juveniles and her “training and experience were not in the formation and treatment of adult scars,” we rejected this argument. Id. at 496. In addition to noting that the physician testified that few distinctions exist between the scarring of juvenile and adult skin, we stated that the defendants’ “objection to [the physician’s] training and experience [went] to the weight, not the admissibility, of her testimony, and counsel had the opportunity to cross-examine her on these issues.” Id. (emphasis added). Admitting the physician’s expert

137

testimony was not an abuse of discretion, given she “had ample knowledge, skill, experience, training, and education with regard to cutaneous findings of abuse.” Id.

Other circuits have also applied this traditional, relaxed approach specifically in the context of medical expert testimony. In Holbrook v. Lykes Brothers Steamship Co., 80 F.3d 777, 782–83 (3d Cir. 1996), the Third Circuit found that the trial court erred in prohibiting the decedent’s treating physician from testifying about the decedent’s diagnosis of mesothelioma. The trial court had reasoned that the physician was not a “pathologist, oncologist or expert in ‘definitive cancer diagnosis.’” Id. at 782. But the Third Circuit explained, “Because of our liberal approach to admitting expert testimony, most arguments about an expert’s qualifications relate more to the weight to be given the expert’s testimony than to its admissibility.” Id. The Third Circuit concluded that “the court’s mistaken approach restricted [the physician]’s testimony based on a requirement that the witness practice a particular specialty concerning certain matters.” Id.

Similarly, the First Circuit has recognized that “[t]he proffered expert physician need not be a specialist in a particular medical discipline to render expert testimony relating to that discipline.” Gaydar v. Sociedad Instituto Gineco-Quirurgico y Planificacion Familiar, 345 F.3d 15, 24 (1st Cir. 2003); see also Pages-Ramirez v. Ramirez-Gonzalez, 605 F.3d 109, 114 (1st Cir. 2010). As the First Circuit explained, “it would be an abuse of discretion to exclude testimony that would otherwise ‘assist the trier better to understand a fact in issue,’ simply because the expert does not have the specialization that the court considers most appropriate.” Pages-Ramirez, 605 F.3d at 114 (quoting Gaydar, 345 F.3d at 24–25).

138

Finally, the implications of affirming the exclusion of Dr. Hruz’s testimony about gender dysphoria should not be overlooked. Reviewing these cases, there is really no question that the majority applies a much more restrictive approach to expert qualifications than we and other courts of appeal have applied in the past. So, unless we tighten the reins on expert qualifications only in constitutional cases that we deem too important to be bothered by the Federal Rules of Evidence—which, of course, we cannot do—the majority’s evidentiary decisions will reverberate in cases beyond those involving equal protection claims. For example, I suspect lawyers representing defendants in medical malpractice, products liability and other personal injury cases will use the majority’s decision to seek to exclude experts who have been permitted to testify for years despite not having backgrounds perfectly aligned with the subject matter of their opinions. And if district courts grant such motions following the majority’s reasoning, consistency will require us to affirm those exclusions.

To sum up the Rule 702 issue, the district court strayed from the text of the rule. It also departed from the manner we and other courts have interpreted Rule 702 for years. Thus, the district court abused its discretion in determining that Dr. Hruz is not qualified to offer expert testimony on gender dysphoria.

II.

On top of the exclusion of Dr. Hruz’s testimony, the majority improperly declares statements from the WPATH Standards and the DSM-5 about the treatment of gender dysphoria to be facts. The majority describes gender dysphoria as “a condition

139

characterized by clinically significant distress and anxiety resulting from the incongruence between an individual’s gender identity and birth-assigned sex.” See Maj. Op. at 12 (citing the DSM-5). It then states as a fact that “[i]f untreated, gender dysphoria can cause debilitating distress, depression, impairment of function, self-mutilation to alter one’s genitals or secondary sex characteristics, other self-injurious behaviors, and suicide.” Maj. Op. at 12–13 (quoting Br. of Medical Amici, which cites the DSM-5). The majority also states, again as a fact, that “the medical community uses generally accepted protocols from the [WPATH Standards],” which it explains recommend “assessment, counseling, and, as appropriate, social transition, hormone therapy, and surgical interventions to bring the body into alignment with one’s gender identity.” Maj. Op 13–14 (again quoting Br. of Medical Amici, which cites the WPATH Standards).

I disagree with these statements of fact by the majority for two reasons. First, the majority improperly determines the statements qualify as indisputable adjudicative facts under Federal Rule of Evidence Rule 201. Second, even if the statements are legislative facts and thus not subject to Rule 201, the majority declares that there is a consensus of the medical community on the treatment of gender dysphoria when the record indicates otherwise.

A.

I begin with Rule 201. With respect to its statements that quote Medical Amici’s citations to the DSM-5, the majority uses Rule 201 to “take judicial notice of the DSM-5.” Maj. Op. 13 n.5. There are a number of problems with this analysis.

140

First, Rule 201 permits courts, if its requirements are satisfied, to take judicial notice of facts. And if a fact is judicially noticed under Rule 201, it is deemed conclusive in a non- criminal case. Fed. R. Evid. 201(f). But the DSM-5 is not a fact. It is a publication. The Federal Rules of Evidence address evidentiary issues related to publications elsewhere. For example, Rule 803(13) provides an exception to the prohibition on hearsay when a statement in a learned treatise, periodical or pamphlet is (1) “called to the attention of an expert witness on cross-examination or relied on by the expert on direct examination”; (2) the reliability of that statement is established “by the expert’s admission or testimony, by another expert’s testimony, or by judicial notice”; and (3) the statement is read into evidence rather than being received as an exhibit. Fed. R. Evid. 803(13). But the majority does not address this Rule or any other basis for admitting an entire publication into evidence.

Second, even if, rather than the entire publication, the majority is referring to the excerpts from the DSM-5 it cites, Rule 201 does not work. Rule 201 applies to adjudicative facts. “Adjudicative facts are simply the facts of the particular case.” Goldfarb v. Mayor & City Council of Baltimore, 791 F.3d 500, 508 n.6 (4th Cir. 2015) (quoting Fed. R. Evid. 201 advisory committee’s note to 1972 proposed rule); see also 2 McCormick on Evidence § 328 (8th ed. 2022) (explaining that adjudicative facts are “facts about the particular event which gave rise to the lawsuit and, like all adjudicative facts, they help[] explain who did

141

what, when, where, how, and with what motive and intent”). Whatever one’s view of the DSM-5 excerpts, they are not adjudicative facts. 4 Third, judicial notice under Rule 201 is reserved for adjudicative facts that are not “subject to reasonable dispute” because the facts are “generally known” within the court’s jurisdiction or “can be accurately and readily determined from sources whose accuracy cannot reasonably be questioned.” Fed. R. Evid. 201(b). The definition of gender dysphoria might satisfy this requirement. After all, as the majority notes, “both parties have cited to the DSM-5 for the definition of gender dysphoria.” Ma. Op. 13 n.5. But as to the excerpt about the consequences of not treating gender dysphoria, the majority glosses over these requirements, reasoning that “[t]he DSM-5 offers standardized criteria for the classification of mental disorders” and “was published by the American Psychiatric Association after a twelve-year revision process in coordination with the National Institute of Mental Health (NIMH) and World Health Organization and a two-month public- and professional-review period.” Maj. Op at 12–13 n.5.

But North Carolina challenged the DSM-5’s reliability as a scientific authority, arguing, among other things, that “the NIMH stopped funding projects that use the DSM-5 and that the DSM-5 is generally controversial.” Maj. Op. at 12–13 n.5 (citing J.A. 742,

4

I realize that in Jacobs v. North Carolina Administrative Office of the Courts, 780 F.3d 562, 565–66 & n.2 (4th Cir. 2015), we applied Rule 201 to judicially noticed excerpts of the DSM-4 describing “social anxiety disorder” because “experts witnesses in [the] case applied the diagnostic criteria of the DSM-IV.” Id. at 566 n.2. While I think the excerpt from that case is a legislative fact more than an adjudicative fact, Jacobs at most supports the finding that the DSM-5’s definition of gender dysphoria may be judicially noticed under Rule 201.

142

764). To the majority, however, this does did not matter. It brushes this objection aside, quoting a news article reporting that “the director of NIMH issued a press release clarifying that ‘NIMH has not changed its position on DSM-5,’ and that the DSM-5 still ‘represents the best information currently available for clinical diagnosis of mental disorders.’” Maj. Op. at 12–13 n.5 (quoting USA Today article). Still citing the news article, the majority added that the director of NIMH has also stated that NIMH was committed to working on a new system called Research Domain Criteria that will aim to focus on causes of disorders, not symptoms.

I disagree with the majority’s analysis on this point. Under Rule 201(b), the question is not who has the better argument about the authoritativeness of a document. It is whether there is any reasonable basis for disputing it. And whether we ultimately agree with North Carolina or not, its argument, at minimum, frames a reasonable dispute about the reliability of the DSM-5 as a scientific authority. 5

5

There are other reasonable disputes as to the DSM-5’s reliability as a scientific authority. For instance, consider DSM-5’s replacement of the diagnosis of “gender identity disorder” with the diagnosis of “gender dysphoria.” DSM-5 at 451. The DSM-5 states that “[g]ender dysphoria is a new diagnostic class in DSM-5 and reflects a change in conceptualization of the disorder’s defining features by emphasizing the phenomenon of ‘gender incongruence’ rather than cross-gender identification per se, as was the case in [] gender identity disorder.” Id. at 814. What’s more, when previewing this change to the DSM, the American Psychiatric Association (APA) stated, “In the upcoming fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5), people whose gender at birth is contrary to one they identify with will be diagnosed with gender dysphoria. This diagnosis is a revision of DSM-IV’s criteria for gender identity disorder and is intended to better characterize the experiences of affected children, adolescents, and adults.” Gender Dysphoria, Am. Psych. Ass’n (2013), https://perma.cc/TQ4V-R4A6 (last visited Feb. 21, 2024). But the APA not only previewed this change; it gave the reasons for it. The APA stated that the DSM-5 “replaces the diagnostic name ‘gender identity (Continued)

143

As for the WPATH Standards, the majority does not expressly state that it is using Rule 201 to take judicial notice of either the entire publication or the specific excerpt from the Medical Amici brief it cites. But its analysis of that excerpt is essentially the same as its analysis of the DSM-5 excerpts. So, the majority appears to consider the excerpt from the WPATH Standards as a fact of which it can take judicial notice under Rule 201.

The majority again noted North Carolina’s objections. North Carolina argues that the district court relied on facts from the WPATH Standards, which contain facts outside the record. North Carolina also contests the reliability of the WPATH Standards. In support of its objections, it cited its experts’ opinions. Id. Dismissing those arguments as concerning “methodology” and whether the WPATH Standards represent a consensus view, the majority rejects North Carolina’s arguments. In so doing, it cites the plaintiffs’ experts’ opinions that the WPATH Standards do, in fact, represent a consensus of the medical community on the treatment for gender dysphoria.

Once again, even if the majority is right about the ultimate resolution of the North Carolina’s position, which I do not concede, that is not the point. The point is whether it has a reasonable argument on reliability. And it does. The majority ignores that requirement

disorder’ with ‘gender dysphoria’” with the “aim[] to avoid stigma” from characterizing the condition as a disorder. Id. It reasoned that “individuals need a diagnosis” to get insurance coverage, but “diagnostic terms . . . can also have a stigmatizing effect.” Id. Reducing stigma and preserving insurance coverage may be good reasons to change the name of the diagnosis from gender identity disorder to gender dysphoria. But they support North Carolina’s challenge to the DSM-5’s scientific authoritativeness on the issues we face today. To be clear, none of this is to say that North Carolina is ultimately right. But it is to say that there is a reasoned debate about the authoritativeness of the DSM-5 statements the majority declares to be facts.

144

of Rule 201, seemingly taking on the role of a factfinder and declaring that because it finds the plaintiffs’ argument more persuasive, North Carolina’s argument is unreasonable. 6 For these reasons, the majority’s declarations of fact are improper under Rule 201.

B.

While I do not believe the excerpts from the DSM-5 and the WPATH standards described above are adjudicative facts, that does not necessarily mean the majority cannot rely on them in its analysis. True, the typical way this type of information would come in as evidence is through witnesses, most likely expert witnesses. That tried-and-true method would allow the adversary process to identify their relevance and reliability and expose any weaknesses in those areas. I see no reason the plaintiffs could not have followed that traditional course here. But apparently, they did not.

Even so, the majority could cite to the excerpts from the DSM-5 and WPATH Standards as legislative facts. “Legislative facts . . . are those which have relevance to legal reasoning and the lawmaking process, whether in the formulation of a legal principle or ruling by a judge or court or in the enactment of a legislative body.” Fed. R. Evid. 201

6

Most instances in which we have taken judicial notice of facts under Rule 201 involve referencing indisputable facts or statistics from government websites. See, e.g., Mays v. Smith, 70 F.4th 198, 206 (4th Cir. 2023) (“The Court takes judicial notice of these uncontested facts from Defendants’ Response Brief, which are publicly available on the [Bureau of Prison’s] website.”); Murphy v. Capella Educ. Co., 589 F. App’x 646, 654 (4th Cir. 2014) (“We can take judicial notice of the statistics available on this [National Center for Education] website.”); Hall v. Virginia, 385 F.3d 421, 424 & n.3 (4th Cir. 2004) (taking judicial notice of information publicly available on official government website); Sierra Club v. U.S. Dep’t of the Interior, 899 F.3d 260, 276 (4th Cir. 2018) (same); United States v. Garcia, 855 F.3d 615, 621 (4th Cir. 2017) (“This court and numerous others routinely take judicial notice of information contained on state and federal government websites.”).

145

advisory committee’s note to 1972 proposed rule. That is, “[l]egislative facts are established truths, facts or pronouncements that do not change from case to case but apply universally.” United States v. Gould, 536 F.2d 216, 220 (8th Cir. 1976); Robinson v. Liberty Mut. Ins. Co., 958 F.3d 1137, 1142 (11th Cir. 2020). For example, “[d]ictionary definitions establish legislative facts when used to answer a question of law, such as how to interpret contractual terms.” Robinson, 958 F.3d at 1142.

Courts can, and increasingly do, take judicial notice, ungoverned by Rule 201, of legislative facts—even disputed ones. See Kenneth Culp Davis, An Approach to Problems of Evidence in the Administrative Process, 55 Harv. L. Rev. 364, 403–07 (1942); Wilson R. Huhn, Teaching Legal Analysis Using A Pluralistic Model of Law, 36 Gonz. L. Rev. 433, 452 & n.86 (2001); Allison Orr Larsen, Factual Precedents, 162 U. Pa. L. Rev. 59, 71–72 (2013) (“Legislative facts come to judges’ attention by way of a procedural hodgepodge: sometimes on the record and sometimes not, sometimes briefed by the parties and sometimes not. In fact, legislative facts are specifically exempted from the Federal Rule of Evidence on Judicial Notice--the rule most on point--and the advisory notes actually encourage their ‘unfettered use.’”). And although the majority does not address this issue, without deciding the issue, I concede it is possible the excerpts from the DSM- 5 and WPATH Standards might be used as legislative facts. See Williams v. Kincaid, 45 F.4th 759, 767–68 & n.3 (4th Cir. 2022).

But even if the majority could rely on legislative facts, in my view, it oversteps here.

Take the majority’s declaration that “the medical community uses generally accepted protocols from the [WPATH Standards],” which it explains recommend “assessment,

146

counseling, and, as appropriate, social transition, hormone therapy, and surgical interventions to bring the body into alignment with one’s gender identity.” Maj. Op 13–14 (quoting Br. of Medical Amici, which cites the WPATH Standards). That declaration ignores the ongoing dispute over the medical necessity and efficacy of the gender dysphoria treatment the states exclude from coverage. See, e.g., Anderson, J.A. 1860–935 (Expert Disclosure Report of Dr. Stephen B. Levine, M.D.); Kadel, J.A. 3327–441 (Expert Witness Declaration of Paul W. Hruz, M.D., Ph.D.). The majority may feel that the plaintiffs have the better argument on that dispute. But it’s one thing to cite competing facts and decide which is more compelling. It’s quite another to declare there is a consensus when there is an ongoing debate. 7 See Gibson v. Collier, 920 F.3d 212, 221 (5th Cir. 2019) (explaining that “the WPATH Standards of Care reflect not consensus, but merely one side in a sharply contested medical debate”); see also Kosilek v. Spencer, 774 F.3d 63, 90 (1st Cir. 2014) (“The law is clear that where two alternative course of medical treatment exist, and both

7

Or take the DSM-5’s statements about the potential for suicide if gender dysphoria goes untreated. Some recent literature suggests that gender dysphoria is not predictive of youth suicide when psychiatric treatment history is accounted for. See Sami-Matti Ruuska et al., All-Cause and Suicide Mortalities Among Adolescents and Young Adults Who Contacted Specialised Gender Identity Services in Finland in 1996–2019: A Register Study, 27 BMJ Mental Health 1 (2024). Other literature identifies the need for more comprehensive research into the long-term effects of gender dysphoria treatment among the pediatric population due to the shortcomings of existing studies, including “insufficient details on drug administration and dosages, treatment durations, and the type of surgery performed” and the failure to conduct randomized controlled trials to account for biases. Jonas F. Ludvigsson et al., A Systematic Review of Hormone Treatment for Children With Gender Dysphoria and Recommendations for Research, 112 Acta Paediatrica (No. 11) 2279 (2023). Again, my point in citing this literature is not to resolve the debate. It is to point out that a reasoned debate exists.

147

alleviate negative effects within the boundaries of modern medicine, it is not the place of our court to ‘second guess medical judgments’ or require that the [Department of Corrections] adopt the more compassionate of the two adequate options.”). And if, as one well-known treatise on evidence puts it, “the intellectual legitimacy of [using legislative facts] turns upon the actual truth-content of the legislative facts taken into account by the judges who propound the decision,” 2 McCormick on Evidence § 331 (8th ed. 2022), the majority’s factual declaration that there is a consensus when the record reveals there is not jeopardizes “intellectual legitimacy.”

III.

To conclude, the majority makes two evidentiary missteps. It improperly affirms the exclusion of Dr. Hruz’s expert testimony about gender dysphoria. And it improperly declares statements from the DSM-5 and the WPATH Standards to be facts. Individually and combined, these missteps improperly stack the deck, effectively ignoring the fair- minded debate about the medical necessity and efficacy of the treatments the plaintiffs seek. For these additional reasons, I respectfully dissent.