Reiyn Keohane v. Florida Department of Corrections Secretary

952 F.3d 1257

Opinion

[PUBLISH]

IN THE UNITED STATES COURT OF APPEALS

FOR THE ELEVENTH CIRCUIT

________________________

No. 18-14096

________________________

D.C. Docket No. 4:16-cv-00511-MW-CAS

REIYN KEOHANE, Plaintiff - Appellee,

versus

FLORIDA DEPARTMENT OF CORRECTIONS SECRETARY, Defendant - Appellant.

________________________

Appeal from the United States District Court for the Northern District of Florida

________________________

(March 11, 2020)

Before WILSON and NEWSOM, Circuit Judges, and COOGLER, * District Judge.

*

Honorable L. Scott Coogler, United States District Judge for the Northern District of Alabama, sitting by designation.

NEWSOM, Circuit Judge:

This appeal requires us to decide whether the Florida Department of Corrections violated the Eighth Amendment’s prohibition on cruel and unusual punishment in its treatment of a transgender inmate’s gender dysphoria. Specifically, we must determine whether the FDC acted with deliberate indifference to Reiyn Keohane’s serious medical need when it (1) enforced a since- repealed policy that strictly limited transgender inmates to the particular medical treatments they were receiving when taken into custody, (2) delayed providing hormone therapy to Keohane for two years pursuant to that policy, and (3) refused Keohane’s “social transitioning” requests—in particular, to wear long hair, makeup, and female undergarments. We must also determine whether the FDC’s post-suit decisions to rescind what the parties have called its “freeze-frame” policy and to prescribe Keohane hormone therapy moot this appeal with respect to the first two issues.

Keohane brought this action under 42 U.S.C. § 1983 alleging violations of her Eighth Amendment rights and seeking (as relevant here) declaratory and injunctive relief. The district court entered a three-part order (1) declaring the FDC’s former freeze-frame policy unconstitutional and permanently enjoining the FDC from “reenacting and enforcing” it, (2) requiring the FDC to continue to provide Keohane with hormone therapy “so long as it is not medically

3

contraindicated,” and (3) directing the FDC to permit Keohane “to socially transition by allowing her access to female clothing and grooming standards.” Keohane v. Jones, 328 F. Supp. 3d 1288, 1319 (N.D. Fla. 2018).

We hold that Keohane’s challenges to the prior freeze-frame policy and the FDC’s initial denial of hormone therapy are moot in light of the FDC’s subsequent repeal and replacement of the policy and its provision of hormone treatment. We reject on the merits Keohane’s claim that the FDC violated the Eighth Amendment by refusing to accommodate her social-transitioning requests.

I

A

Reiyn Keohane is an FDC inmate currently serving a 15-year sentence for attempted murder. Keohane was born male, but she began to identify as female sometime during her preadolescent years. Beginning at age 14—and up until the time she was incarcerated at 19—Keohane wore women’s clothing, makeup, and hairstyles. At 16, she was formally diagnosed with gender dysphoria—which, in general terms, “refers to the distress that may accompany the incongruence between one’s experienced or expressed gender and one’s assigned gender.” American Psychiatric Ass’n, Diagnostic and Statistical Manual of Mental Disorders 451 (5th ed. 2013). About six weeks before her arrest, Keohane began hormone therapy under the care of a pediatric endocrinologist.

4

Following her arrest, Keohane was initially housed at the Lee County Jail, where she says her request to continue hormone therapy was immediately denied. When, several months later, in July 2014, Keohane was transferred to an FDC prison in south Florida, she asked to resume her hormone-therapy treatment because, as she explained to prison officials in a written grievance, “[w]ithout it [she] consider[ed] self-harm and suicide every single day.” She made similar requests (accompanied by similar threats of self-harm) during the ensuing two years, all of which were either disregarded or rejected.1 Keohane alleges—and the FDC doesn’t dispute—that her hormone-therapy requests were denied pursuant to a policy specifying that “[i]nmates who have undergone treatment for [gender dysphoria] will be maintained only at the level of change that existed at the time they were received by the Department.” Under this “freeze-frame” policy, the care of inmates suffering from gender dysphoria was determined not by their current, individualized medical needs, but rather by the treatment they were (or weren’t) receiving at the time of their incarceration.

In December 2014, Keohane’s grievances began to include requests relating to “social transitioning”—that is, the ability to live consistently with one’s gender identity, including by dressing and grooming accordingly. In particular, Keohane

1 It is undisputed that, throughout the course of her incarceration, Keohane has consistently been provided mental-health counseling for her gender dysphoria.

5

expressed a desire to wear female undergarments and makeup, and to grow out her hair in a long, feminine style—as the district court described it, “to possess and wear the same bras, panties, hairstyles, and makeup items permitted in [the FDC’s] female facilities.” The FDC refused Keohane’s social-transitioning requests on the grounds that they violated prison policy—which required male inmates to wear “[u]nder shorts” and to “have their hair cut short to medium uniform length at all times with no part of the ear or collar covered,” Fla. Admin. Code r. 33- 602.101(2), (4)—and that they posed a security risk. Specifically, the FDC was concerned that an inmate wearing makeup and female undergarments would inevitably become a target in an all-male prison, thereby endangering not only the inmate but also the prison employees who would have to step in to protect her. Additionally, the FDC concluded that there are clear advantages to maintaining uniformity in a prison setting, including the ability to more readily detect contraband.

During this protracted request-denial cycle, Keohane made multiple attempts to self-harm. In October 2014, Keohane tried to hang herself. In January 2015, she tried to castrate herself. And in April 2017, she tried to kill herself twice more.

B

Having exhausted her efforts to obtain relief within the prison system, Keohane filed a single-count complaint in the United States District Court for the

6

Northern District of Florida alleging that the FDC’s denial of her hormone-therapy and social-transitioning requests violated the Eighth Amendment. As relevant here, Keohane sought three forms of relief: (1) a declaration that the FDC was acting with deliberate indifference to her gender dysphoria, a serious medical need; (2) a permanent injunction ordering the FDC to provide her with hormone therapy and social-transitioning accommodations, including “access to female clothing and grooming standards”; and (3) a permanent injunction prohibiting the FDC from enforcing its freeze-frame policy.

Not long after Keohane filed suit, the FDC altered its behavior in two material respects. First, just two weeks after the complaint was filed, the FDC referred Keohane to an outside endocrinologist who immediately prescribed her hormone therapy. Second, about six weeks after that, the FDC formally repealed its freeze-frame policy and replaced it with a policy that calls for individualized assessment and treatment of inmates who claim to be suffering from gender dysphoria and related conditions. With the lone exception of a sports bra to help with her hormone-related breast enlargement, however, the FDC has continued to refuse Keohane’s social-transitioning requests.

Keohane’s case proceeded to a bench trial. Helpfully, the parties agreed—

and still do—both that Keohane’s gender dysphoria constitutes a “serious medical need” for deliberate-indifference purposes and that hormone therapy is medically

7

necessary to treat that need. Most notably, Keohane’s FDC treatment team— which comprised her psychologist, her mental-health counselor, and a psychiatric physician assistant—supported the determination that hormone therapy is medically necessary. And since initially acceding to Keohane’s request for hormone therapy in September 2016, the FDC has consistently provided it and has repeatedly represented (both at trial and on appeal) that it will continue to do so “as long as [her] treatment team believes the hormones are medically necessary to treat her gender dysphoria.” Br. of Appellant at 7–8 (citing testimony).

The parties and medical professionals disagreed, however—and still do—

about the medical necessity of Keohane’s social-transitioning-related requests to dress and groom herself as a woman. For his part, Keohane’s retained medical expert testified (1) that allowing an individual to present consistently with her gender identity is one “of the medically necessary components for the treatment of gender dysphoria,” (2) that it would be “medically and logically inconsistent” and “potentially harmful” to provide Keohane hormone therapy while denying her the ability to socially transition, and (3) that forcing one to live in conformity with a gender with which she doesn’t identify “would likely” cause her to engage in self- harm.

By contrast, the members of Keohane’s treatment team, who had supported the provision of hormone therapy, denied that social transitioning is medically

8

necessary to treat Keohane’s gender dysphoria—as did a staff psychiatrist with the FDC’s medical vendor Wexford, the FDC’s chief clinical officer, and the FDC’s retained expert. According to the treatment team, Keohane’s current regimen— hormone therapy and mental-health counseling, together with other accommodations, including the use of female pronouns (“she,” “her,” etc.), safer housing accommodations, and private shower facilities—is sufficient to treat her gender dysphoria. The treatment team also explained that requiring Keohane to comply with the FDC’s clothing and grooming policies does not place her at a substantial risk of self-harm or severe psychological pain. The FDC’s retained expert acknowledged that the sorts of social-transitioning-related accommodations that Keohane sought may be “psychologically pleasing” to her, but he too rejected the suggestion that they are medically necessary. Finally, FDC witnesses testified—as FDC personnel had explained from the beginning—that granting Keohane’s social-transitioning requests would pose unacceptable security risks. Notably, though, despite the FDC’s steadfast refusal to accommodate Keohane’s social-transitioning requests, it has repeatedly stated—since this suit was filed, anyway—that “if [those] requests are deemed medically necessary, they will be fulfilled,” and that it will take additional security measures as needed. Br. of Appellant at 9.

9

Following trial, the district court issued an opinion in Keohane’s favor. The court rejected the FDC’s contention that Keohane’s claims relating to the former freeze-frame policy and its initial refusal to provide hormone therapy were moot— concluding, in particular, that the FDC’s “voluntary cessation” of the challenged conduct was insufficient to render those claims nonjusticiable. On the merits, the district court held (1) that the FDC’s former freeze-frame policy was an unconstitutional “blanket ban on medically necessary care,” (2) that the FDC’s earlier denial of hormone therapy—which the district court thought resulted from “bigotry and ignorance”—evinced “deliberate indifference to [Keohane’s] serious medical need in violation of the Eighth Amendment,” and (3) that allowing Keohane to clothe and groom herself as a woman is medically necessary to treat her gender dysphoria and that the FDC’s ongoing denial of her social-transitioning requests likewise violates the Eighth Amendment. To effectuate its judgment, the court entered a three-part order (1) declaring the FDC’s former freeze-frame policy unconstitutional and “permanently enjoin[ing]” the FDC from “reenacting and enforcing” it, (2) requiring the FDC to continue to “provide Ms. Keohane with hormone therapy so long as it is not medically contraindicated,” and (3) directing

10

the FDC to “permit Ms. Keohane to socially transition by allowing her access to female clothing and grooming standards.”2

* * *

The FDC’s appeal presents the following issues for our consideration: (1)

Did the FDC’s former freeze-frame policy manifest deliberate indifference to Keohane’s serious medical need and thereby violate the Eighth Amendment’s prohibition against cruel and unusual punishment, and relatedly, is Keohane’s challenge to that policy—and requested injunction against its enforcement—now moot in light of its repeal and replacement? (2) Did the FDC’s refusal to provide Keohane with hormone therapy during the first two years of her incarceration violate the Eighth Amendment, and again, is Keohane’s challenge to that refusal— and requested injunction—now moot in light of the FDC’s decision to allow the treatment? And (3) does the FDC’s ongoing refusal to provide Keohane with

2 We review the district court’s mootness determination, including its voluntary-cessation analysis, de novo, and any related findings of fact for clear error. Troiano v. Supervisor of Elections, 382 F.3d 1276, 1282 (11th Cir. 2004). On the merits, “[a]lthough we review the district court’s entry of a permanent injunction for an abuse of discretion, the district court’s underlying legal conclusion—that there was an Eighth Amendment violation warranting equitable relief—is reviewed de novo.” Thomas v. Bryant, 614 F.3d 1288, 1303 (11th Cir. 2010). “Subsidiary issues of fact are reviewed for clear error.” Id. For (much) more on these standards of review as they apply to the merits of Keohane’s Eighth Amendment claim, see infra at 27–28, 28 n.8.

11

social-transitioning accommodations—including the ability to wear long hair, makeup, and female undergarments—violate the Eighth Amendment?3

II

The Eighth Amendment prohibits the “inflict[ion]” of “cruel and unusual punishments.” U.S. Const. amend VIII. Under the Amendment, the “[f]ederal and state governments . . . have a constitutional obligation to provide minimally adequate medical care to those whom they are punishing by incarceration.” Harris v. Thigpen, 941 F.2d 1495, 1504 (11th Cir. 1991). As particularly relevant here, the Supreme Court has held that prison officials violate the bar on cruel and unusual punishments when they display “deliberate indifference to serious medical needs of prisoners.” Estelle v. Gamble, 429 U.S. 97, 104 (1976).

A deliberate-indifference claim entails both an objective and a subjective component. Brown v. Johnson, 387 F.3d 1344, 1351 (11th Cir. 2004). First, the inmate must establish “an objectively serious medical need”—that is, “one that has

3 We can dispense at the outset with the FDC’s contention that the district court’s injunction violates the Prison Litigation Reform Act. The PLRA requires injunctive relief to be “narrowly drawn,” to “extend[] no further than necessary to correct the violation of the [f]ederal right,” and to be “the least intrusive means necessary to correct the violation of the [f]ederal right.” 18 U.S.C. § 3626(a)(1)(A). The FDC asserts that the district court ran afoul of the PLRA by decreeing what the FDC calls a “blanket policy” allowing transgender female inmates in male facilities access to the privileges afforded to inmates in female prisons. In short, we just don’t see it. By its terms, the district court’s order directs the FDC to provide a particular course of treatment to Keohane specifically. See, e.g., Keohane, 328 F. Supp. 3d at 1318–19 (“Defendant must provide Ms. Keohane with hormone therapy . . . .”; “To treat Ms. Keohane’s gender dysphoria, Defendant must permit Ms. Keohane to socially transition . . . .”).

12

been diagnosed by a physician as mandating treatment or one that is so obvious that even a lay person would easily recognize the necessity for a doctor’s attention”—that, “if left unattended, poses a substantial risk of serious harm.” Id. (alteration adopted) (quotation omitted). Second, the inmate must prove that prison officials acted with deliberate indifference to that need by showing (1) that they had “subjective knowledge of a risk of serious harm” and (2) that they “disregard[ed]” that risk (3) by conduct that was “more than mere negligence.” Id.

Here, as already noted, there’s no debate about the objective component.

The FDC admits—and the parties thus agree—that Keohane’s gender dysphoria constitutes a “serious medical need.” Rather, the dispute hinges on the subjective component. Specifically, the parties disagree—at least in part—over whether the particular types of treatment that Keohane has requested are medically necessary, such that any course of care that doesn’t include them would be constitutionally inadequate.

A prisoner bringing a deliberate-indifference claim has a steep hill to climb.

We have held, for instance, that the Constitution doesn’t require that the medical care provided to prisoners be “perfect, the best obtainable, or even very good.” Harris, 941 F.2d at 1510 (quotation omitted). Rather, “[m]edical treatment violates the [E]ighth [A]mendment only when it is so grossly incompetent, inadequate, or excessive as to shock the conscience or to be intolerable to

13

fundamental fairness.” Id. at 1505 (quotation omitted). We have also emphasized—as have our sister circuits—that “a simple difference in medical opinion between the prison’s medical staff and the inmate as to the latter’s diagnosis or course of treatment [fails to] support a claim of cruel and unusual punishment.” Id.; accord, e.g., Lamb v. Norwood, 899 F.3d 1159, 1162 (10th Cir. 2018) (“We have consistently held that prison officials do not act with deliberate indifference when they provide medical treatment even if it is subpar or different from what the inmate wants.”); Kosilek v. Spencer, 774 F.3d 63, 82 (1st Cir. 2014) (en banc) (“[The Eighth Amendment] does not impose upon prison administrators a duty to provide care that is ideal, or of the prisoner’s choosing.”).

Against that backdrop, we consider whether the FDC violated the Eighth Amendment (1) by adopting and previously enforcing the since-repealed freeze- frame policy, (2) by initially declining to provide Keohane with hormone therapy, and (3) by continuing to refuse Keohane’s social-transitioning-related requests to dress and groom herself according to female standards.

A

First, the former freeze-frame policy. Keohane contends that it constituted “deliberate indifference to [a] serious medical need[],” Estelle, 429 U.S. at 104, in that it amounted to a per se rejection of any treatment that an inmate hadn’t received prior to her incarceration, without regard to (or any exception for) medical

14

necessity. The district court agreed and permanently enjoined the FDC from “reenacting and enforcing” its former policy. Were we free to reach the merits, we would almost certainly agree, as well. As already explained, the FDC has repeatedly conceded that Keohane’s gender dysphoria constitutes a “serious medical need.” It seems to us that responding to an inmate’s acknowledged medical need with what amounts to a shoulder-shrugging refusal even to consider whether a particular course of treatment is appropriate is the very definition of “deliberate indifference”—anti-medicine, if you will. Cf. Webster’s Second New International Dictionary 1527 (1944) (defining “medicine” as “[t]he science and art dealing with the prevention, cure, or alleviation of disease”). Unsurprisingly to us, other courts considering similar policies erecting blanket bans on gender-dysphoria treatments—without exception for medical necessity—have held that they evince deliberate indifference to prisoners’ medical needs in violation of the Eighth Amendment. See, e.g., Fields v. Smith, 653 F.3d 550, 559 (7th Cir. 2011); see also Hicklin v. Precynthe, No. 4:16-CV-01357-NCC, 2018 WL 806764, at *11 (E.D. Mo. Feb. 9, 2018); Soneeya v. Spencer, 851 F. Supp. 2d 228, 247 (D. Mass. 2012).

We conclude, though, that we are not free to reach the merits. Because the FDC has formally rescinded its freeze-frame policy and replaced it with a new one that properly attends to inmates’ individualized medical needs, we hold that Keohane’s challenge to the old policy is moot. There is, quite simply, no longer

15

any freeze-frame policy to challenge—nothing to enjoin, as the district court purported to do.

Mootness arises when an issue presented in a case is “no longer ‘live’ or the parties lack a legally cognizable interest in the outcome.” Powell v. McCormack, 395 U.S. 486, 496 (1969). In particular, we have held that a case must be dismissed as moot “[i]f events that occur subsequent to the filing of a lawsuit . . . deprive the court of the ability to give the plaintiff . . . meaningful relief.” Al Najjar v. Ashcroft, 273 F.3d 1330, 1336 (11th Cir. 2001). “[D]ismissal is required because mootness is jurisdictional,” in that a moot case no longer presents a live “Case[]” or “Controvers[y]” within the meaning of Article III of the Constitution. Id. at 1335–36.

Here, because the FDC repealed its freeze-frame policy following the onset of litigation—approximately two months after Keohane filed suit—we must determine whether the “voluntary cessation” exception to the mootness doctrine applies. Pursuant to that exception, a defendant’s “voluntary cessation of allegedly illegal conduct does not moot a case.” United States v. Concentrated Phosphate Exp. Ass’n, 393 U.S. 199, 203 (1968). For reasons we will explain, we hold that the exception does not apply here, and that Keohane’s challenge to the since- rescinded freeze-frame policy is moot.

16

The basis for the voluntary-cessation exception is the commonsense concern that a defendant might willingly change its behavior in the hope of avoiding a lawsuit but then, having done so, “return to [its] old ways.” Id. (quotation omitted). So when a defendant contends that a plaintiff’s claim has become moot as a result of the defendant’s own independent decision to cease some disputed action, it usually “bears the . . . burden of showing that it is absolutely clear the allegedly wrongful behavior could not reasonably be expected to recur.” Doe v. Wooten, 747 F. 3d 1317, 1322 (11th Cir. 2014) (quoting Friends of the Earth, Inc. v. Laidlaw Envtl. Servs. (TOC), Inc., 528 U.S. 167, 190 (2000)). Importantly here, though, we have explained that “governmental entities and officials have . . . considerably more leeway than private parties in the presumption that they are unlikely to resume illegal activities.” Flanigan’s Enters., Inc. of Ga. v. City of Sandy Springs, 868 F.3d 1248, 1256 (11th Cir. 2017) (en banc) (quoting Coral Springs St. Sys., Inc. v. City of Sunrise, 371 F.3d 1320, 1328–29 (11th Cir. 2004)). The reason, we have said, is that government actors are more likely than private defendants “to honor a professed commitment to changed ways.” Troiano v. Supervisor of Elections, 382 F.3d 1276, 1283 (11th Cir. 2004) (quotation omitted); see also, e.g., Nat’l Ass’n of Bds. of Pharmacy v. Bd. of Regents of the Univ. Sys. of Ga., 633 F.3d 1297, 1310 (11th Cir. 2011) (“Hence, ‘the Supreme Court has held almost uniformly that voluntary cessation by a government

17

defendant moots the claim.’” (alterations adopted) (quoting Beta Upsilon Chi Upsilon Chapter v. Machen, 586 F.3d 908, 917 (11th Cir. 2009))).

That is especially true when, as here, a government defendant has formally rescinded a challenged statute, ordinance, rule, or policy. As the en banc Court emphasized in Flanigan’s, “the repeal of a challenged statute”—or other similar pronouncement—is ordinarily “one of those events that makes it absolutely clear that the allegedly wrongful behavior . . . could not reasonably be expected to recur.” 868 F.3d at 1256 (quoting Harrell v. The Fla. Bar, 608 F.3d 1241, 1265– 66 (11th Cir. 2010)).4 “As a result, once the repeal of [a policy] has caused our jurisdiction to be questioned, the plaintiff bears the burden of presenting affirmative evidence that [her] challenge is no longer moot.” Id. (alterations adopted) (quotation omitted). “The key inquiry” is whether the plaintiff has shown a “reasonable expectation”—or, as we phrased it elsewhere, a “substantial likelihood”—that the government defendant “will reverse course and reenact” the

4 Notably, in so stating, the Flanigan’s Court was merely reiterating the conclusions of an earlier en banc decision, which itself was merely repeating the conclusions of earlier circuit decisions. See, e.g., Flanigan’s, 868 F.3d at 1259 (“This Court and the Supreme Court have repeatedly held that the repeal or amendment of an allegedly unconstitutional statute moots legal challenges to the legitimacy of the repealed legislation. A superseding statute or regulation moots a case . . . to the extent that it removes challenged features of the prior law. If the repeal is such that the allegedly unconstitutional portions of the [challenged] ordinance no longer exist, the appeal is rendered moot because any decision we would render would clearly constitute an impermissible advisory opinion.” (citations and internal quotation marks omitted) (quoting Tanner Advert. Grp., L.L.C. v. Fayette Cty., Ga., 451 F.3d 777, 789–90 (11th Cir. 2006) (en banc))).

18

repealed rule. Id.; Beta Upsilon Chi Upsilon Chapter, 586 F.3d at 917 (quotation omitted).

In Flanigan’s, we explained that, in determining whether a plaintiff has shouldered its burden, a reviewing court should look to “three broad factors”— although we hastened to add that “these factors should not be viewed as exclusive nor should any single factor be viewed as dispositive,” and that, in any event, “a mootness finding should follow when the totality of [the] circumstances persuades the court that there is no reasonable expectation that the government entity will reenact” the challenged policy. 868 F.3d at 1257. “First, we ask whether the change in conduct resulted from substantial deliberation or is merely an attempt to manipulate our jurisdiction.” Id. In this connection, “we will examine the timing of the repeal, the procedures used in enacting it, and any explanations independent of this litigation which may have motivated it.” Id. “Second, we ask whether the government’s decision to terminate the challenged conduct was ‘unambiguous’”— which, in turn, entails an inquiry into whether the government’s policy shift is fairly viewed as being “permanent and complete.” Id. (quotation omitted). Finally, “we ask whether the government has consistently maintained its commitment to the new policy.” Id.; accord, e.g., Doe, 747 F.3d at 1322–23 (articulating the same three factors).

19

Applying these factors here, we come to the same conclusion that we reached in Flanigan’s: “[T]here is no substantial evidence indicating a reasonable likelihood that” the defendant—here, the FDC—“will reenact the challenged provision”—here, the freeze-frame policy—“which it has now repealed.” 868 F.3d at 1260. With respect to the first factor, the district court concluded that the FDC’s decision to rescind its freeze-frame policy “was an attempt to manipulate jurisdiction—certainly not the result of substantial deliberation.” Keohane, 328 F. Supp. 3d at 1300. To echo a sentiment expressed in Flanigan’s, “[w]e are not unsympathetic to this argument.” 868 F.3d at 1260. We don’t doubt for a minute that the FDC’s about-face just two months after Keohane filed suit was motivated, at least in part, by a desire to rid itself of this litigation. Even so, as we took care to clarify in Flanigan’s, the timing of a government defendant’s decision to repeal a challenged policy shouldn’t be overemphasized. Id. at 1259 (“[T]he timing of repealing legislation should not be dispositive of our inquiry into whether there is a reasonable expectation of reenactment.”); id. (“[T]he timing of repealing legislation should not control the mootness inquiry.”).5 Moreover—and contrary to the dissent’s suggestion—the fact that “the FDC still hasn’t admitted that its practices violated the Constitution,” see Dissenting Op. at 51, has little, if anything,

5 The dissent fails to heed the en banc Court’s warning against overstating timing considerations. See, e.g., Dissenting Op. at 44, 45, 48, 50, 56.

20

to do with the substantial-deliberation factor, or with the voluntary-cessation analysis at all, for that matter, see Flanigan’s, 868 F.3d at 1262 (noting that “even at en banc oral argument” the government defendant there had “declined to concede that [its ordinance] was unconstitutional” but clarifying that “[w]hether the [government] defended the [o]rdinance and/or continue[d] to believe it was constitutional” had little bearing on the mootness analysis). Finally, and in any event, even if we were to give Keohane the substantial-deliberation factor, it is but one among several, and here the remaining considerations tip the scale decisively in the other direction.

What we said in Flanigan’s about the second factor applies here too: The FDC’s formal repeal of the freeze-frame policy “is plainly an unambiguous termination.” Id. at 1261. Just like the government defendant there, the FDC “has not merely declined to enforce the [freeze-frame policy] against” Keohane in particular—so as, in effect, to give her a personalized exemption. Id. Rather, “it has removed the challenged portion” of the policy “in its entirety.” Id. Indeed, the FDC has gone a step farther by replacing the old freeze-frame policy with a new protocol that provides for individualized evaluation. And as the FDC explained at oral argument, it would have to do some serious hoop-jumping to rescind the current, individual-assessment policy and reenact the former freeze-frame policy even if it wanted to do so. See Oral Argument at 4:23 (explaining the protracted

21

administrative process that accompanies a formal policy change). Moreover—and again, just as in Flanigan’s—the FDC has repeatedly “assured this Court . . . that it has no intention of reenacting” the freeze-frame policy. 868 F.3d at 1261–62; see also Br. of Appellant at 48–49; Oral Argument at 4:15, 4:55, 5:10, 5:15. “We have previously relied on such representations,” and there is no evidence or history that would cause us to doubt them here. Flanigan’s, 868 F.3d at 1262.6 Finally, as to the third factor, we conclude that the FDC has “consistently maintained” and applied its new individualized-assessment policy. Id. at 1257. There is certainly no “pattern” of broken promises here of the sort that has concerned us in the past. See, e.g., Doe, 747 F.3d at 1324. To the contrary, the FDC rescinded the freeze-frame policy in October 2016, immediately replaced it with a new policy that provides for personalized evaluation, and (so far as we can tell) hasn’t looked back. In an effort to turn the consistent-application factor to her advantage, Keohane has asserted (and the dissent repeats, see Dissenting Op. at 51) that one inmate was denied hormone-therapy treatment pursuant to the freeze- frame policy even after its formal repeal. Tellingly, though, not even the district court found that lone instance probative, saying that it would be “hard pressed to find that evidence of one mistake in applying old policies—or, perhaps, one rogue

6 Cf. Burns v. Windsor Ins. Co., 31 F.3d 1092, 1095 (11th Cir. 1994) (“Every lawyer is an officer of the court. And . . . he always has a duty of candor to the tribunal.”); Model Rules of Prof’l Conduct r. 3.3 (Am. Bar Ass’n 1983) (“Candor Toward the Tribunal”).

22

doctor acting contrary to protocol—[was] sufficient” to demonstrate inconsistency. Keohane, 328 F. Supp. 3d at 1299.

At the end of the day, we’re less concerned with the subjective question whether the initial reason for the government’s decision was sincere than with the objective question whether there is any “substantial evidence indicating a reasonable likelihood that the [FDC] will reenact the challenged [freeze-frame policy] which it has now repealed” and replaced. Flanigan’s, 868 F.3d at 1260. Evidence that the FDC realized and corrected its mistake a little late in the game in no way suggests that it would revert back to its old ways absent the injunction. All of the evidence, in fact, is squarely to the contrary. Cf. Henslee v. Union Planters Nat. Bank & Trust Co., 335 U.S. 595, 600 (1949) (Frankfurter, J., dissenting) (“Wisdom too often never comes, and so one ought not to reject it merely because it comes late.”). Accordingly, we hold that Keohane’s challenge to the FDC’s former freeze-frame policy is moot.

B

Second, hormone therapy. Keohane contends that the FDC’s initial refusal to provide her with hormone-therapy treatment violated the Eighth Amendment. The district court agreed and entered an injunction (separate from the one prohibiting the reenactment of the freeze-frame policy) requiring the FDC to

23

“provide Ms. Keohane with hormone therapy so long as it is not medically contraindicated.” Keohane, 328 F. Supp. 3d at 1318.

On the merits, the question might be a close one. The record seems to indicate that the FDC knew that denying Keohane hormone therapy threatened a serious risk of self-harm—the grievances that she filed with prison officials expressly and repeatedly linked the two. And given the circumstances, it’s possible that the FDC disregarded that risk “by conduct that [was] more than mere negligence.” Brown, 387 F.3d at 1351.

Once again, though, we find that we cannot reach the merits because we conclude that Keohane’s hormone-therapy-related challenge is moot. Approximately two weeks after Keohane filed suit—and even before it formally repealed the freeze-frame policy—the FDC referred her to an endocrinologist who prescribed her hormone therapy, and she has been receiving hormone-therapy treatment ever since. Accordingly, the FDC contends that there is no longer any live controversy concerning Keohane’s entitlement to hormone therapy.

As before, the mootness inquiry hinges on the application of the voluntary-

cessation exception. And as already explained, under that exception “governmental entities and officials have . . . considerably more leeway than private parties in the presumption that they are unlikely to resume illegal activities.” Flanigan’s, 868 F.3d at 1256 (quoting Coral Springs, 371 F.3d at

24

1328–29). And even though here we consider the FDC’s freestanding determination to provide Keohane hormone therapy—independent of its later repeal of the freeze-frame policy, which the district court enjoined separately—the governing principles remain basically the same. As we summarized in Flanigan’s, “even where the intervening governmental action does not rise to the level of a full legislative repeal . . . ‘a challenge to a government policy that has been unambiguously terminated will be moot in the absence of some reasonable basis to believe that the policy will be reinstated if the suit is terminated.’” Id. at 1256 (emphasis added) (quoting Troiano, 382 F.3d at 1285).

We find no “reasonable basis” to believe that, following a dismissal, the FDC would revert to refusing hormone therapy to Keohane. As with its repeal of the freeze-frame policy, we recognize that the timing of the FDC’s decision to provide Keohane with hormone treatment—here, a mere two weeks after she filed suit—may well suggest a desire to eliminate potential liability. It seems scarcely debatable that the FDC hoped that by acceding to Keohane’s request it could avoid litigation. But again, we have clarified that timing considerations shouldn’t be overemphasized in the voluntary-cessation analysis and, in any event, that alleged jurisdiction-manipulation is only one among several non-exhaustive factors that inform the inquiry. See id. at 1257–59. The remaining factors demonstrate that Keohane’s hormone-therapy challenge, like her freeze-frame challenge, is indeed

25

moot. Most notably, we are satisfied both (1) that the FDC’s “decision to terminate the challenged conduct”—here, its reversal of its initial denial of hormone therapy—was “unambiguous” in the sense that it was “both permanent and complete,” and (2) that the FDC “has consistently maintained its commitment” to Keohane’s new course of treatment. Id. at 1257 (quotation omitted). The FDC has given us concrete assurances—in both word and deed—that it will continue to provide Keohane’s hormone therapy. Not only has the FDC rescinded the freeze- frame policy pursuant to which it refused Keohane’s early requests for hormone treatment, but its own doctors have concluded—and testified under oath—that Keohane’s hormone therapy is medically necessary. And consistent with that view, since initially granting Keohane’s request in September 2016, the FDC has faithfully provided her with hormone-therapy treatment and has repeatedly represented to us that it will continue to do so.7 The decision on which the district court principally relied in rejecting the FDC’s mootness argument, Doe v. Wooten, actually provides a useful contrast here. There, an inmate filed suit alleging that two prison officials had acted with deliberate indifference to his serious need for protection after he assisted the Bureau of Prisons in an investigation of one of its own employees. 747 F.3d at 1321. Specifically, the officials promised the inmate that they would protect him

7 See supra n.6.

26

and transfer him to a lower-security prison in exchange for his cooperation. Id. at 1320. Although he was briefly moved to a lower-security facility as promised, he was then, over the course of several years, repeatedly transferred to other high- security prisons where he was exposed as an informant and severely assaulted. Id. at 1320–21. After years of litigation and more questionable transfers, the BOP suddenly changed course and moved the inmate to a lower-security facility just days before the trial was set to begin—and then contended that the inmate’s challenge was moot. Id. at 1321. We held that the BOP had failed to establish the unlikelihood of a recurrence for four basic reasons: (1) the BOP’s ultimate transfer of the inmate to a lower-security prison so soon before trial strongly suggested that its motivation was solely to manipulate jurisdiction; (2) the BOP had a “pattern” of breaking its transfer-related promises; (3) the “mere fact” that the BOP was (at that moment, anyway) giving the inmate what he wanted wasn’t enough to overcome its history of recurring misbehavior, and (4) the BOP “never said” that it wouldn’t transfer the inmate back to a high-security prison. Id. at 1323–25.

As already noted, the Doe Court’s first reason—that the timing indicated a desire to dispose of a lawsuit—may well apply here, too. But the other reasons are inapplicable—or more accurately, belied—in this case. There is no “pattern” of broken promises here; since acceding to Keohane’s hormone-therapy request in September 2016, the FDC has consistently provided her the treatment. And more

27

than the “mere fact” that Keohane is currently being given hormone therapy, we have reasonable assurance that the FDC won’t revert back to its previous posture; whereas in Doe the BOP had “never said” that it wouldn’t backslide, the FDC has repeatedly represented that it will continue to provide Keohane with hormone therapy so long as her team “believes the hormones are medically necessary to treat her gender dysphoria.” Br. of Appellant at 7–8.

In short, we conclude that there is no “reasonable basis” to believe that, if Keohane’s hormone-therapy claim is dismissed, the FDC will reverse course and refuse to provide the treatment. Accordingly, we hold that Keohane’s hormone- therapy-related claim is moot.

C

Lastly, social transitioning. Keohane asserts that the FDC is continuing to violate the Eighth Amendment by denying her requested social-transitioning- related accommodations—specifically, to grow out her hair, use makeup, and wear female undergarments. Unlike Keohane’s arguments concerning the freeze-frame policy and hormone therapy, her social-transitioning claim unquestionably presents a live controversy, inasmuch as the FDC (for the most part, anyway) continues, to this day, to refuse her requests. Accordingly, we proceed to consider Keohane’s social-transitioning-based challenge on the merits. In so doing, we review de novo the district court’s ultimate determination “that there was an Eighth Amendment

28

violation warranting equitable relief,” and we review for clear error any “[s]ubsidiary issues of fact.” Thomas v. Bryant, 614 F.3d 1288, 1303 (11th Cir. 2010). 8

8 We pause here to respond briefly (or perhaps not so briefly) to the dissent’s extended critique of the standards of review that we apply. The dissent accuses us—vigorously and repeatedly— of ignoring the observation in Thomas that “[a] prison official’s deliberate indifference is a question of fact which we review for clear error.” 614 F.3d at 1312. We’ve done no such thing. As we trust the text demonstrates, we haven’t endeavored to re-find any historical facts—e.g., what happened, who knew what, how did they respond? See, e.g., Farmer v. Brennan, 511 U.S. 825, 842 (1994) (“Whether a prison official had the requisite knowledge of a substantial risk is a question of fact subject to demonstration in the usual ways . . . .”); Goebert v. Lee County, 510 F.3d 1312, 1327 (11th Cir. 2007) (“Disregard of the risk is also a question of fact that can be shown by standard methods.”). Instead, we’ve simply done what Thomas commands us to do, and what the dissent itself recognizes we must do—apply “de novo review . . . to the district court’s ultimate conclusion whether the objective and subjective elements of a deliberate- indifference claim state an Eighth Amendment violation.” Dissenting Op. at 60.

The dissent seems to think that the clear-error standard’s application in a deliberate-

indifference case somehow supersedes and supplants the foundational rule that we, as an appellate court, must review de novo a district court’s ultimate determination whether an Eighth Amendment violation has occurred. On the dissent’s understanding, the de novo standard’s sole office is to ensure that the district court puts “checkmarks” in the right boxes, and then doesn’t make a truly boneheaded, asinine mistake:

[I]f the district court, despite checkmarks in both the objective and subjective boxes, still concluded that there was no Eighth Amendment violation, we would lend no deference to this error. We would review it de novo, and would no doubt reverse. And if the district court, despite holding that one of the elements was not met, still concluded that there was an Eighth Amendment violation, we would do the same. We would review this error de novo, and no doubt reverse. That is the ultimate conclusion that we review de novo.

Id. at 61–62. This mindless, mechanical box-checking assessment cannot possibly be what we’ve meant when we have repeatedly held that de novo review applies to the district court’s determination whether “there was an Eighth Amendment violation warranting equitable relief.” Thomas, 614 F.3d at 1303.

Contrary to the dissent’s suggestion, meaningful appellate review of a district court’s ultimate constitutional holding follows straightaway from Supreme Court precedent prescribing de novo review of other application-of-law-to-fact questions—including those arising under the Eighth Amendment. See United States v. Bajakajian, 524 U.S. 321, 336–37 & n.10 (1998) (excessiveness of a fine); see also, e.g., Cooper Indus., Inc. v. Leatherman Tool Grp., Inc., 532 U.S. 424, 435–36 (2001) (punitive-damages award); Ornelas v. United States, 517 U.S. 690, 699

29

Recall that a deliberate-indifference claim entails both an objective and a subjective component. As we have explained, the objective component is clearly satisfied here—all agree that Keohane’s gender dysphoria constitutes a “serious medical need” within the meaning of Eighth Amendment precedent. Brown, 387 F.3d at 1351. The dispute here centers on the subjective component, which requires the plaintiff to show that prison officials (1) had actual “knowledge of a risk of serious harm” and (2) “disregard[ed]” that risk (3) by conduct that was “more than mere negligence.” Id.

(1996) (probable cause and reasonable suspicion). District courts are undoubtedly better situated than appellate courts to make findings of historical fact, and their determinations with respect to those facts are accordingly entitled to deference. But what the Eighth Amendment means—and requires in a given case—is an issue squarely within the core competency of appellate courts. And to be clear, it’s no answer to say, as the dissent does—citing Justice Scalia’s solo dissent in Ornelas—that some issues underlying a deliberate-indifference claim may be “fact-specific and not easy to generalize.” Dissenting Op. at 70 n.13. The Supreme Court recognized as much regarding the “mixed questions” in Ornelas, Bajakajian, and Cooper—and yet applied de novo review anyway. Just so here. See generally Kosilek v. Spencer, 774 F.3d 63, 68–69, 84–85 (1st Cir. 2014) (en banc) (rejecting the very same arguments being offered by the dissent in this case and holding that de novo, rather than clear-error, review governed a district court’s ultimate determination that the Eighth Amendment required prison authorities to accommodate a transgender inmate’s medical-treatment requests).

Now, having said all that, we hasten to add that nothing here rides on the applicable standard of review. Even if the deferential clear-error standard did apply (as the dissent suggests) in such a way as to render essentially meaningless the de novo review that applies to the district court’s ultimate determination whether an Eighth Amendment violation has occurred, we would have little trouble formulating the required “firm conviction that a mistake ha[d] been committed.” Silva v. Pro Transp., Inc., 898 F.3d 1335, 1339 (11th Cir. 2018) (quotation omitted). For reasons explained in text, the district court’s determination—that the FDC “disregard[ed]” a risk of serious harm “by conduct that [was] more than mere negligence,” Brown, 387 F.3d at 1351—was not just erroneous, but clearly so.

30

Although the parties vigorously debate whether the actual-knowledge prong is satisfied here, we needn’t resolve that issue, because even assuming that FDC officials knew that Keohane was at risk of serious harm—thus satisfying the subjective prong’s first factor—there is no basis for concluding that by denying her social-transitioning requests they disregarded that risk “by conduct that [was] more than mere negligence” and thereby violated the Eighth Amendment. That is so for two reasons.

First, as already explained, unlike with respect to hormone therapy, the testifying medical professionals were—and remain—divided over whether social transitioning is medically necessary to Keohane’s gender-dysphoria treatment. Keohane’s retained expert testified that it is. By contrast, the members of Keohane’s medical-treatment team, Wexford’s staff psychiatrist, the FDC’s chief clinical officer, and the FDC’s retained expert all testified that it isn’t. The closest any of those witnesses got—not nearly close enough, it seems to us—was the FDC’s expert’s acknowledgment that social-transitioning, while not strictly medically necessary, would be “psychologically pleasing” to Keohane. Cf. Harris, 941 F.2d at 1511 n.24 (“[N]othing in the Eighth Amendment . . . requires that [inmates] be housed in a manner [that is] most pleasing to them.” (quotation

31

omitted)). 9 Keohane’s medical-treatment team further concluded that requiring Keohane to comply with the FDC’s policies regarding hair and grooming standards doesn’t put her at a substantial risk of self-harm or severe psychological pain.

At worst, then, this is a situation where medical professionals disagree as to the proper course of treatment for Keohane’s gender dysphoria, and it’s well established that “a simple difference in medical opinion between the prison’s medical staff and the inmate as to the latter’s diagnosis or course of treatment [cannot] support a claim of cruel and unusual punishment.” Id. at 1505; Waldrop v. Evans, 871 F.2d 1030, 1033 (11th Cir. 1989); accord, e.g., Lamb, 899 F.3d at 1163 (holding that “disagreement alone” does not constitute deliberate indifference); Kosilek, 774 F.3d at 90 (“The law is clear that where two alternative courses of medical treatment exist, and both alleviate negative effects within the boundaries of modern medicine, it is not the place of our court to second guess medical judgments or to require that the DOC adopt the more compassionate of

9 We needn’t dwell on the point, but we emphatically reject the dissent’s suggestion that the Eighth Amendment requires the government—which is to say taxpayers—to fund any medical treatment that is “psychologically pleasing” to an incarcerated inmate. See Dissenting Op. at 82– 85. Necessity, not pleasure, is the constitutional standard—and no amount of massaging can make those two things the same. And to be clear, it hardly renders the dissent’s would-be standard more “sensible” to note that “gender dysphoria is . . . a psychological illness.” Id. at 82. By the very same logic, the dissent would presumably conclude that the Eighth Amendment requires the provision of any treatment that is “physically pleasing”—not necessary, but pleasing—to address an inmate’s physical illness. The possibilities are endless.

32

two adequate options.” (quotation omitted)). 10 Put simply, when the medical community can’t agree on the appropriate course of care, there is simply no legal basis for concluding that the treatment provided is “so grossly incompetent, inadequate, or excessive as to shock the conscience or to be intolerable to fundamental fairness.” Harris, 941 F.2d at 1505 (quotation omitted). Here, therefore, implementing the course of treatment recommended by Keohane’s FDC medical team, and seconded by a number of other medical professionals, isn’t “so unconscionable as to fall below society’s minimum standards of decency”—and thus violative of the Eighth Amendment—merely because it conflicts with the opinion of Keohane’s retained expert. Kosilek, 774 F.3d at 96.11

10 To be clear, because the district court awarded only prospective injunctive relief—as relevant here, in the form of an order directing the FDC to begin providing Keohane social-transitioning accommodations—the deliberate-indifference question must “be determined in light of the prison authorities’ current attitudes and conduct,” i.e., “their attitudes and conduct at the time suit is brought and persisting thereafter.” Farmer, 511 U.S. at 845 (emphasis added) (quoting Helling v. McKinney, 509 U.S. 25, 36 (1993)). It is irrelevant, therefore, that there was once a time— years ago now—when the FDC was refusing Keohane both hormone therapy and social- transitioning accommodations. As matters stood at the time the parties’ witnesses testified before and during trial, and at the time the district court issued its injunction, Keohane was—and as things stand today, still is—receiving both mental-health counseling and hormone therapy, and is enjoying the use of female pronouns, safer housing accommodations, and private shower facilities. See Kosilek, 774 F.3d at 91 (“The subjective element of an Eighth Amendment claim for injunctive relief requires not only that [the plaintiff] show that the treatment she received was constitutionally inadequate, but also that the DOC was—and continues to be—deliberately indifferent to her serious risk of harm.” (emphasis added)). 11 The several decisions concluding that prison officials acted with deliberate indifference to a transgender inmate’s serious medical need are distinguishable in three important ways. First, in each of those cases there appeared to be general (and consistent) consensus among the inmate’s medical providers that a particular treatment was medically necessary. See, e.g., Soneeya, 851 F. Supp. 2d at 250 (inmate’s providers consistently recommended feminization). As explained in text, that’s not the case with respect to Keohane’s social-transitioning requests; indeed, the

33

Second, the FDC denied Keohane’s social-transitioning-related requests, at least in part, on the ground that they presented serious security concerns— including, most obviously, that an inmate dressed and groomed as a female would inevitably become a target for abuse in an all-male prison. “When evaluating medical care and deliberate indifference, security considerations inherent in the functioning of a penological institution must be given significant weight.” Id. at 83; see also Helling v. McKinney, 509 U.S. 25, 37 (1993) (“The inquiry into [the subjective] factor also would be an appropriate vehicle to consider arguments regarding the realities of prison administration.”); Evans v. Dugger, 908 F.2d 801, 806 (11th Cir. 1990) (“[P]rison officials operate[] under a mandate to provide for [physical and medical] needs while simultaneously assuring the safety and security of [inmates].”). As the Supreme Court has long recognized, “[p]rison administrators . . . should be accorded wide-ranging deference in the adoption and execution of policies and practices that in their judgment are needed to preserve

district court’s own finding—that before Keohane filed suit her treatment team couldn’t reach a consensus about the medical necessity of her social-transitioning requests—actually undermines its deliberate-indifference conclusion. Second, in those cases the prison officials denied treatment based on a blanket policy without exception for medical necessity. See, e.g., id. at 249–50. Here, while that was once the case, the FDC has since rescinded its freeze-frame policy and has clarified that it will make exceptions for social-transitioning-related requests if deemed medically necessary. Third, in those cases it was clear that the current course of treatment was insufficient because the gender-dysphoria symptoms persisted or even worsened. See, e.g., id. at 250; see also Hicklin, 2018 WL 806764, at *11–13. Here, to the contrary, the evidence indicates that Keohane’s symptoms improved after she was prescribed hormone therapy, which she continues to receive.

34

internal order and discipline and to maintain institutional security.” Bell v. Wolfish, 441 U.S. 520, 547 (1979). Accordingly, even an outright “denial of care may not amount to an Eighth Amendment violation if that decision is based in legitimate concerns regarding prisoner safety and institutional security.” Kosilek, 774 F.3d at 83. 12 That is all the more true where, as here, the inmate has not been refused care entirely, but has instead been given a meaningful course of treatment that includes many (if not all) of the components that she originally sought.13 The First Circuit’s en banc decision in Kosilek v. Spencer is especially instructive here. There, as here, a transgender inmate who was suffering from gender dysphoria—and who had attempted both to castrate herself and to commit suicide—alleged that the treatment she was receiving in prison violated the Eighth

12 To the extent that the dissent suggests that “denying treatment for non-medical reasons”— including for reasons of institutional security—always and everywhere “arises to subjective deliberate indifference,” Dissenting Op. at 72, we disagree. The lone case that the dissent cites for that proposition—McElligott v. Foley, 182 F.3d 1248 (11th Cir. 1999)—had nothing to do with prison security. And we think it inconceivable that the Eighth Amendment could be read to disable prison administrators from refusing an inmate’s treatment request on the ground that it would threaten safety and security. 13 The district court declined even to address the security issue—calling it a “red herring”—on the ground that the FDC had entered into a pretrial stipulation that it would accommodate Keohane’s social-transitioning requests if they were deemed medically necessary. On appeal, Keohane likewise asserts that the stipulation “begins and ends the discussion of security in this case.” The point, seemingly, is that if the FDC has promised to meet Keohane’s social- transitioning demands if necessary, prison security must not be as serious an issue as the FDC contends. What this argument ignores is that the FDC’s stipulation expressly clarified that even if it did become necessary to accommodate Keohane’s requests, “additional security measures [would be] taken” as needed. So, far from admitting that Keohane’s social-transitioning requests present no meaningful security concerns, the FDC has simply said that if it became necessary to do so, it would address the newly presented security issues in new and different ways.

35

Amendment. 774 F.3d at 68–69. There, as here, the inmate was getting some, but not all, of the treatment she wanted; in particular, while the prison was providing her with mental-health counseling, hormone therapy, and gender-appropriate clothing, it had persistently refused her requests for sex-reassignment surgery. Id. at 69–70. There, as here, the testifying medical professionals disagreed about whether a constitutionally adequate course of treatment required the prison to grant the inmate’s remaining request. Id. at 74–79. There, as here, prison officials had raised security-related concerns about accommodating the inmate’s demand. Id. at 79–81. And finally, there, as here, the district court had “issued an extensive opinion” concluding (1) that the inmate’s gender dysphoria constituted a “serious medical need,” (2) that “the only adequate way to treat” her condition was by granting all of her requests—including, there, for sex-reassignment surgery—and (3) that the prison officials’ “stated security concerns were merely pretextual.” Id. at 81.

The First Circuit framed the question before it in terms that apply equally here: “[W]e are faced with the question whether the [prison’s] choice of a particular medical treatment is constitutionally inadequate, such that the district court acts within its power to issue an injunction requiring provision of an alternative treatment—a treatment which would give rise to new concerns related to safety and prison security.” Id. at 68. Notwithstanding the “extensive[ness]” of

36

the district court’s determinations, the First Circuit reversed. In so doing, the en banc court emphasized that “[t]he law is clear that where two alternative courses of medical treatment exist, and both alleviate negative effects within the boundaries of modern medicine, it is not the place of [a reviewing] court to second guess medical judgments or to require that [prison officials] adopt the more compassionate of two adequate options.” Id. at 90 (quotation omitted). The First Circuit also stressed the “wide-ranging deference” to which prison administrators are entitled when making safety and security assessments. Id. at 83, 92 (quotation omitted). Concluding, the court held (1) that prison authorities “ha[d] chosen to provide a form of care that offer[ed] direct treatment” for the inmate’s gender dysphoria and (2) that they had “done so in light of the fact that provision of” the inmate’s preferred treatment “would create new and additional security concerns— concerns that do not presently arise from its current treatment regimen.” Id. at 96.

Kosilek is closely (if not quite exactly) on point here. The FDC has given Keohane some, but not all, of what she wants—although it has denied her social- transitioning requests (at least as they pertain to clothing and grooming), it has provided mental-health counseling, hormone therapy, the use of female pronouns, safer housing accommodations, and private shower facilities. And like the prison officials in Kosilek, the FDC has struck that balance both because Keohane’s treatment team has determined that her current regimen is sufficient to treat her

37

gender dysphoria and because it has rationally concluded that her social- transitioning requests—to dress and groom herself as a woman—would present significant security concerns in an all-male prison.14 Bottom line: In light of the disagreement among the testifying professionals about the medical necessity of social transitioning to Keohane’s treatment and the “wide-ranging deference” that we pay to prison administrators’ determinations about institutional safety and security, Bell, 441 U.S. at 547, we simply cannot say that the FDC consciously disregarded a risk of serious harm by conduct that was “more than mere negligence” and thereby violated the Eighth Amendment, Brown, 387 F.3d at 1351. Rather, it seems to us that the FDC chose a meaningful course of treatment to address Keohane’s gender-dysphoria symptoms—treatment that, while perhaps different from (and less than) what Keohane preferred, is sufficient to clear the low deliberate-indifference bar. For better or worse, prisoners aren’t constitutionally entitled to their preferred treatment plan or to medical care that is great, “or even very good.” Harris, 941 F.2d at 1510; see also Lamb, 899 F.3d at

14 Rather than Kosilek, the dissent embraces the Ninth Circuit’s recent decision in Edmo v. Corizon, Inc., 935 F.3d 757, 767 (9th Cir. 2019), which held that a state prisoner suffering from gender dysphoria was constitutionally entitled to state-funded sex-reassignment surgery. See Dissenting Op. at 90 (“The Ninth Circuit got it right . . . .”). In ruling the way it did, though, the Ninth Circuit emphasized that the state there had “not so much as allude[d] to” any of the sorts of security concerns that the FDC has raised here. Edmo, 935 F.3d at 794 (“Our approach mirrors the First Circuit’s [in Kosilek], but the important factual differences between cases yield different outcomes. Notably, the security concerns in Kosilek, which the First Circuit afforded ‘wide- ranging deference,’ are completely absent here. The State does not so much as allude to them.” (citation omitted)).

38

1162 (“[P]rison officials do not act with deliberate indifference when they provide medical treatment even if it is subpar or different from what the inmate wants.”). So long as the care provided isn’t “so grossly incompetent, inadequate, or excessive as to shock the conscience or to be intolerable to fundamental fairness,” then the Eighth Amendment is satisfied. Rogers v. Evans, 792 F.2d 1052, 1058 (11th Cir. 1986); Harris, 941 F.2d at 1505. We are confident that the care here passes constitutional muster.15

* * *

One final word: This is a case that stirs emotions. And understandably so—

the question whether and to what extent Florida prison officials must accommodate Keohane’s gender dysphoria is a sensitive one, on both sides of the “v.” Our dissenting colleague’s opinion is passionate and heartfelt, as is evident from its rhetoric. He accuses us, among other things, of “usurp[ing]” (and alternatively

15 A brief word about the dissent’s repeated refrain (echoing the district court) that Keohane’s treatment team was “incompetent.” See Dissenting Op. at 79–81, 84, 90; see also Keohane, 328 F. Supp. 3d at 1307, 1316. The nub of the dissent’s critique seems to be that “no member of Keohane’s treatment team had ever treated” someone in Keohane’s precise situation—i.e., “a pre-transition patient with gender dysphoria”—and that the team members hadn’t been trained specifically in the “World Professional Association for Transgender Health” standards. Dissenting Op. at 79. Respectfully, the dissent sets the constitutional bar entirely too high. To repeat what we’ve just said—because it bears repeating—the medical care provided to prisoners needn’t be “perfect, the best obtainable, or even very good.” Harris, 941 F.2d at 1510 (quotation omitted). In the same vein, we have held that “[m]inimally adequate care usually requires minimally competent physicians.” Id. at 1509. Keohane’s treatment might not have been “perfect,” but it wasn’t conscience-shocking either, and her treatment-team members might not have had specialties (or particularized experience) in caring for “pre-transition patient[s] with gender dysphoria,” but they were (at the very least) “minimally competent.”

39

“commandeer[ing]” and “annexing”) the district court’s role (Dissenting Op. at 41, 53, 86), “rearrang[ing] the record” to suit our own desires (id. at 41), strategically “ignoring” bad facts while focusing on those we “like[] better” (id. at 53), “pluck[ing]” favorable tidbits from cases (id. at 54), and—the coup de grâce— “shak[ing] the magic 8-ball until it gives us” a result we want (id. at 92). Needless to say—and with all due respect—we don’t think that we’ve done any of those things.

Make no mistake, we too have sympathy for Ms. Keohane, and we too regret her predicament. But our first obligation—our oath—is to get the law right. See 28 U.S.C. § 453. And our best understanding of the law is that—for better or worse—it simply does not entitle Ms. Keohane to additional relief. Our dissenting colleague, of course, sees things differently. But let us pause briefly to consider the implications of his position:

• First, on his view, the Constitution should be read to require prison officials to provide every convicted inmate—at taxpayer expense—with any treatment that is “psychologically pleasing.” See Dissenting Op. at 82–85.

That cannot possibly be the law.

• Second, on his view, the Constitution should be read to require prison officials to provide every convicted inmate—again, at taxpayer expense—

with doctors who have particularized experience (perhaps even a specialty)

in dealing with his or her precise condition, no matter how rare. See Dissenting Op. at 79–81. Again—inconceivable.

• Finally, on his view, the Constitution should be read to prohibit prison authorities from making a prophylactic judgment that housing a transitioning

40

woman—wearing long hair, female undergarments, and makeup—in a men’s prison simply poses too grave a threat to institutional security. See Dissenting Op. at 74–79. We just don’t think so.

This is a difficult case—no doubt. While we respect our dissenting colleague’s fervor, we find ourselves constrained to disagree with his conclusions, which, we think, would precipitate sweeping changes in the law of prison administration.

III

For the foregoing reasons, we hold that Keohane’s challenges to the FDC’s former freeze-frame policy and its initial failure to provide her with hormone therapy are moot, and we reject on the merits her claim that the FDC violated the Eighth Amendment by refusing to accommodate her social-transitioning-related requests.

We VACATE the district court’s order, DISMISS AS MOOT in part, and REVERSE in part.

41

WILSON, Circuit Judge, dissenting:

The majority has usurped the role of the district court. In a painstaking, 61-

page order, the district judge made detailed factual findings, concluding from them that Keohane’s claims were not moot and that the FDC was liable for deliberate indifference. We must review those findings with great deference, disregarding them only if clearly erroneous. But the majority does not apply ordinary clear- error review, as we might in a sentencing case or an employment dispute. Instead, the majority steps into the district court’s shoes to reweigh the facts, reassess credibility determinations, and rearrange the record to reach a different result. See Mach. Rental Inc. v. Herpel (In re Multiponics, Inc.), 622 F.2d 709, 723 (5th Cir. 1980) (“Merely because a reviewing Court on the same evidence may have reached a different result will not justify setting a finding aside.”). 1 That is not our role. The clearly erroneous standard is weighty for a reason:

It reflects the “unchallenged superiority of the district court’s factfinding ability” and its capacity “to judge . . . the credibility of the witnesses.” Salve Regina Coll. v. Russell, 499 U.S. 225, 233 (1991). We cannot simply supplant the district court’s findings with our own. And yet that is what the majority does here.

1 In Bonner v. City of Prichard, we adopted as binding precedent all decisions of the former Fifth Circuit handed down before October 1, 1981. 661 F.2d 1206, 1209 (11th Cir. 1981) (en banc).

42

Since the district court’s findings have substantial footing in the record, I would accept them. And given those findings, none of Keohane’s claims are moot, and the FDC was deliberately indifferent to her gender dysphoria when it failed to provide her with social-transitioning treatment.

One brief note before we begin: The majority cites language from my dissent to suggest that I have let the emotions surrounding this issue sway my opinion. The majority quotes me correctly; I have strong words about its analysis. But make no mistake—any fervor in the text below stems not from the facts of this case, but from the majority’s misapplication of our precedent.

I.

First, mootness. Federal courts decide only “live” controversies. See Troiano v. Supervisor of Elections in Palm Beach Cty., Fla., 382 F.3d 1276, 1281– 82 (11th Cir. 2004). A case is no longer live, and is thus moot, when a court can no longer grant “meaningful relief” to the challenging party. Id. at 1282. This can happen when allegedly unlawful conduct ceases, but a party’s “voluntary cessation of allegedly unlawful conduct ordinarily does not suffice to moot a case.” Friends of the Earth, Inc. v. Laidlaw Envtl. Servs., Inc., 528 U.S. 167, 174 (2000). This is because the defendant might simply return to its “old ways.” Troiano, 382 F.3d at 1283. A defendant “claiming that its voluntary compliance moots a case” thus bears the “formidable” and “heavy” burden of “showing that it is absolutely clear

43

the allegedly wrongful behavior could not reasonably be expected to recur.” Doe v. Wooten, 747 F.3d 1317, 1322 (11th Cir. 2014); see also Harrell v. Fla. Bar, 608 F.3d 1241, 1268 (11th Cir. 2010).

That analysis is a little different for the government, though. Government actors are more likely to “honor a professed commitment to changed ways” and get “more leeway than private parties in the presumption that they are unlikely to resume illegal activities.” See Doe, 747 F.3d at 1322; Troiano, 382 F.3d at 1283. As a result, this court has held that when the government “unambiguously terminat[es]” challenged conduct, we presume that the conduct will not begin again. Doe, 747 F.3d at 1322. It has also held that the “government actor is entitled to this presumption only after it has shown unambiguous termination of the complained of activity.” Id.

Our case law, however, has shifted slightly from this framing. See Flanigan’s Enters., Inc. of Ga. v. City of Sandy Springs, Ga., 868 F.3d 1248, 1257 (11th Cir. 2017) (en banc), cert. denied sub nom. Davenport v. City of Sandy Springs, Ga., 584 U.S. ___, 138 S. Ct. 1326 (2018). We used to ask first whether the government had “unambiguously terminated” its conduct before we would apply the presumption against recurrence; if it did, the burden would shift to the plaintiff to prove that there was a reasonable basis to believe that the challenged conduct would renew. See Doe, 747 F.3d at 1322. Now we ask only “whether the

44

evidence leads us to a reasonable expectation that the [government] will reverse course and reenact the allegedly offensive [conduct]” should the court dismiss the case. See Flanigan’s, 868 F.3d at 1256.2 As before, we rely on three broad and non-exclusive factors to help us answer this question. See id. at 1257.

“First, we ask whether the change in conduct resulted from substantial deliberation or is merely an attempt to manipulate our jurisdiction.” Id. The “timing” of the cessation is highly relevant to this inquiry. See id. So are the government’s reasons for its delayed action and change of heart. See id. at 1260; Doe, 747 F.3d at 1323. A “defendant’s cessation before receiving notice of a legal challenge weighs in favor of mootness, while cessation that occurs ‘late in the game’ will make a court more skeptical of voluntary changes that have been made.” Harrell, 608 F.3d at 1266 (citation omitted) (internal quotation mark

2 There was good reason to do away with the unambiguous-termination two-step. Our prior unambiguous-termination analysis ran confusingly parallel to other standards for deciding mootness after the government’s voluntary cessation. Compare Doe, 747 F.3d at 1322–23 (holding that the unambiguous-termination test is the proper standard for deciding whether a case is moot after the government’s voluntary cessation), with Troiano, 382 F.3d at 1283–84 (holding that whether “there is a substantial likelihood that the offending policy will be reinstated if the suit is terminated” is the proper standard for deciding whether a case is moot after the government’s voluntary cessation). The unambiguous-termination analysis was also redundant in two ways. For one, the first of the three factors for determining whether the government unambiguously terminated its conduct was, paradoxically, whether the government unambiguously terminated its conduct. See Doe, 747 F.3d at 1322–23. For another, the three factors that went into the unambiguous-termination analysis were the same factors that went into the later reasonable-basis-for-recurrence analysis, making the second step effectively redundant. See id. By streamlining the test to focus on the broad factors we’ve typically relied on, our en banc court eliminated this redundancy while reaffirming that the government must still unambiguously terminate its conduct to survive the voluntary-cessation analysis. See Flanigan’s, 868 F.3d at 1257.

45

omitted). Similarly, a “well-reasoned justification for the cessation” is critical evidence “that the ceasing party intends to hold steady in its revised (and presumably unobjectionable) course.” Id. In contrast, the government’s inconsistent and unsupported position changes tend to show a lack of substantial deliberation. See Doe, 747 F.3d at 1325–26.

“Second, we ask whether the government’s decision to terminate the challenged conduct was unambiguous.” Flanigan’s, 868 F.3d at 1257 (internal quotation mark omitted). The question here is “whether the actions that have been taken to allegedly moot the case reflect a rejection of the challenged conduct that is both permanent and complete.” Id. As with the first prong, the “timing and content of the decision” are highly relevant to whether the government has decidedly abandoned prior conduct. See Harrell, 608 F.3d at 1266. So is the government’s refusal to admit that its prior position was wrong. See, e.g., Sheely v. MRI Radiology Network, P.A., 505 F.3d 1173, 1187 (11th Cir. 2007) (holding that a case was not moot despite voluntary cessation in part because the government had not admitted that its conduct was wrong); ACLU v. Fla. Bar, 999 F.2d 1486, 1494–95 (11th Cir. 1993) (same when the government had not admitted that a prior rule was wrong). 3 As are the government’s assurances that it will steer clear of

3 The majority says that the government’s current beliefs about the constitutionality of its action have “little, if anything” to do with the voluntary-cessation analysis. Majority Op. at 19–20.

46

prior conduct. See Flanigan’s, 868 F.3d at 1261–62. But this point is key: The government’s promises are not a trump card. They can prove hollow under the weight of other evidence. See Sheely, 505 F.3d at 1184 (noting that a party’s assertion “that it has no intention of reinstating the challenged practice” does not suffice to moot a case and is merely “one of the factors” to consider).

“Third, we ask whether the government has consistently maintained its commitment to the new policy or legislative scheme.” Flanigan’s, 868 F.3d at 1257. We are also “more likely to find a reasonable expectation of recurrence when the challenged behavior constituted a continuing practice or was otherwise deliberate.” Doe, 747 F.3d at 1323.

Although these factors lend helpful guidance, they are not the be-all and end-all. When considering government cessation, including “a full legislative repeal of a challenged law—or an amendment to remove portions thereof—these factors should not be viewed as exclusive nor should any single factor be viewed as dispositive.” Flanigan’s, 868 F.3d at 1257. “Rather, the entirety of the relevant

That is an overstatement. To be sure, Flanigan’s concluded that the government’s current beliefs there provided “only weak evidence, if any” that its termination was unambiguous. 868 F.3d at 1262. Yet we did not blink this consideration out of existence (and overrule years of precedent in the process). See Sheely, 505 F.3d at 1187; ACLU, 999 F.2d at 1494–95. Rather, we noted that this evidence in Flanigan’s paled alongside strong evidence of unambiguous termination, including a public repeal and a unanimous and public adoption of a resolution supporting the repeal. See 868 F.3d at 1262. As I explain below, we have none of those safeguards here. See infra at 52–56. We have long known that a party is more likely to pursue a practice it believes is lawful than one it thinks is not. See Sheely, 505 F.3d at 1187; ACLU, 999 F.2d at 1494–95. So the FDC’s current beliefs hold weight in the unambiguous-termination analysis.

47

circumstances should be considered and a mootness finding should follow when the totality of those circumstances persuades the court that there is no reasonable expectation that the government entity will reenact the challenged [policy].” Id. In other words, though the formal repeal of a policy or practice often goes a long way toward showing that the government won’t turn heel, it is not always determinative; other facts can call the repeal into doubt. See Doe, 747 F.3d at 1323 (noting that the cessation “analysis may vary depending on the facts” of each case).

Finally, although we consider the voluntary-cessation analysis de novo, we review the factual findings that play into this analysis for clear error. Troiano, 382 F.3d at 1282. The FDC, as the appellant, must prove that these findings are clearly erroneous. See Thelma C. Raley, Inc. v. Kleppe, 867 F.2d 1326, 1328 (11th Cir. 1989). Under clear-error review, we must defer to the district court’s factual findings and credibility determinations, as that court “had the advantage of observing the witnesses and evaluating their credibility firsthand.” Hiram Walker & Sons, Inc. v. Kirk Line, 30 F.3d 1370, 1376 (11th Cir. 1994). The district court’s findings bind us “unless, in view of the entire record, we are left with a definite and firm conviction that a mistake has been committed.” Pelphrey v. Cobb Cty., Ga., 547 F.3d 1263, 1268 (11th Cir. 2008). A “mistake,” however, is not merely a difference in judgment. See Multiponics, 622 F.2d at 723. As an appellate court,

48

we cannot set aside a finding just because we would “have reached a different result.” Id.

A.

The district court held that the FDC failed to show under the factors that “the allegedly wrongful behavior could not reasonably be expected to recur.” Doe, 747 F.3d at 1322. Its analysis rested on a collection of careful factual findings. Since these findings have substantial support in the record, I accept them. And given their persuasive weight, I conclude that there is a reasonable basis to believe that the FDC will return to its old ways.

First up is substantial deliberation or jurisdictional manipulation. The district court held that the FDC retreated from the challenged conduct—both its freeze-frame policy and its refusal to treat Keohane with hormone therapy—only to manipulate jurisdiction. Many findings compel this result.

One is timing. As the majority concedes, the timing of the FDC’s termination muddies the waters around its cessation, and the timing on this point is key. See Flanigan’s, 868 F.3d at 1257. For more than two years, the FDC refused to treat Keohane with hormone therapy under the freeze-frame policy. But less than a month after Keohane filed this lawsuit, the FDC gave her hormone therapy. And within about two months, it repealed the freeze-frame policy.

49

The district court reasonably found these fourth-quarter concessions suspect—they suggest that the FDC only changed its ways to silence litigation. See Doe, 747 F.3d at 1325. Its murky motives “create[] ambiguity” about the FDC’s commitment to its changes and about whether the FDC will reoffend in the future. See Rich v. Sec’y, Fla. Dep’t of Corr., 716 F.3d 525, 532 (11th Cir. 2013).

Dubious timing was not all the district court relied on. Alongside a suspicious sequence of events, the court also doubted the FDC’s deliberation because the FDC took inconsistent positions throughout the case and could not explain its prolonged delay or sudden change of heart. See Doe, 747 F.3d at 1325– 26; Harrell, 608 F.3d at 1266.

To start, the district court found that the FDC had flip-flopped about its policy and practices throughout the litigation—a finding that no one disputes. Then, after the FDC reluctantly changed its ways, the court found that the FDC could not explain why it had done so. The FDC provided no minutes, no memoranda, and no testimony to show that it had thoughtfully considered its policy shifts. See Harrell, 608 F.3d at 1267 (holding that the government’s retreat from challenged conduct did not moot a case in part because the government failed “to disclose any basis for its decision,” making it unclear whether the decision was “well-reasoned and therefore likely to endure” (internal quotation mark omitted)). The FDC’s only explanation for its turnaround was that its general counsel had

50

found some “case law” on the subject. Yet the FDC provided no support for this justification. It did not identify what case law called for the changes, who at the FDC compelled the changes, or what procedures went into making the changes. In short, the only evidence the FDC gave to show that it had engaged in substantial deliberation was its word that it had engaged in substantial deliberation.

For these reasons, the district court found that the FDC’s justification for its policy shifts was incredible and pretextual. The court also found that the FDC had taken inconsistent positions and had been hasty in its decision-making. These findings led the district court to conclude that the FDC shifted its policy and practice for one purpose: to manipulate jurisdiction. The majority does not claim that any of these findings leave it with a “definite and firm conviction” that the district court made a “mistake.” Pelphrey, 547 F.3d at 1268. Nor could it—they have substantial record support. See supra at 48–50. Given these detailed findings, there is only one conclusion: The FDC shied away from its former ways not to make amends, but to manipulate jurisdiction.

Second is whether the FDC unambiguously terminated its freeze-frame policy and practice of denying Keohane hormone therapy. As before, the “timing and content” of the FDC’s reversal is crucial. See Harrell, 608 F.3d at 1266. The district court found the FDC’s timing suspect, its reasons dodgy, and its process a mystery. See supra at 48–50; Harrell, 608 F.3d at 1266; Doe, 747 F.3d at 1325–

51

26. It also found its obstinance telling. Because even after the FDC turned heel, it still refused to admit the error in its ways. See Sheely, 505 F.3d at 1187. It continued to argue that hormone therapy is not constitutionally required for treating Keohane’s gender dysphoria—words that contradict its actions and the “case law” that spurred the FDC into motion. In fact, the FDC still hasn’t admitted that its practices violated the Constitution, and it refused to do so at oral argument. See Oral Argument at 1:40; Sheely, 505 F.3d at 1187.

The record supports these findings. The majority does not claim them clearly erroneous. So I am hard pressed to see how anyone could conclude, given these findings of fact, that the FDC’s termination was unambiguous. Properly confined to the district court’s findings, I conclude, as that court did, that the FDC’s termination was hazy at best. 4 Third, then, is inconsistency. The district court found that, even after the FDC repealed the freeze-frame policy, the FDC withheld hormone therapy from at least one other inmate under the freeze-frame policy. 5 The court also found that,

4 In this step, the majority analyzes the repeal of the policy and the government’s promise not to revert to it old ways. Since those issues are relevant to my analysis of Doe and Flanigan’s, see infra at 52–56, I will discuss them there to avoid repetition. 5 The majority claims that the district court did not find this “lone [inconsistent application] probative” to the mootness analysis. Majority Op. at 21–22. But the majority is incorrect. The district court relied on this evidence to conclude that the FDC failed to establish an unambiguous termination. See Keohane v. Jones, 328 F. Supp. 3d 1288, 1300 (N.D. Fla. 2018) (“Given [a host of other reasons] and at least one instance of inconsistent application of the new policy, this Court finds Defendant has failed to establish an ‘unambiguous termination’ . . . .” (emphasis

52

even after the FDC began giving Keohane hormone therapy, the FDC delayed in providing her hormone therapy. This delay caused her to attempt suicide twice in three days. The record also shows that the delays would have been more frequent had she not vigilantly pursued her treatment.

Like canaries in a coal mine, these deviations warn that the FDC is not as dedicated to its new positions as the majority would have us believe. And these instances were not mere anomalies. They were new applications of the FDC’s prior practices—practices that were not mere blips, but were “continuing” and “otherwise deliberate.” See Doe, 747 F.3d at 1323. For these reasons, the district court held that this factor cut against the FDC. The majority, as before, does not question the evidence that the court relied on. Nor does it hold that the court clearly erred in finding that the FDC delayed in providing hormone therapy and applied the freeze-frame policy after its repeal. Given these uncontested facts, I agree with the district court that the last factor cuts against mootness.

B.

If we were conducting a typical mootness review, the district court’s unchallenged findings would lead us to the same conclusion that the district court

added)). True, the district court said that it would be “hard pressed to find that evidence of one mistake . . . is sufficient” standing alone to find against the government. Id. at 1299. But it noted that this evidence doesn’t stand alone; “this drop of evidence only adds to the tidal wave of other circumstances crashing down on [the FDC’s] mootness argument.” Id. So contrary to the cropped picture the majority presents, the full frame shows that the district court did rely on this information, as should we.

53

reached. But rather than lend due weight to the district court’s findings, the majority commandeers the district court’s role, ignoring that court’s conclusions while focusing us on the facts it likes better. To do so, it places a heavy emphasis on the repeal of the policy, and it takes solace in the FDC’s oral-argument assurance that it will stay on the straight and narrow. The majority then strictly stacks this case up against the facts of Doe and Flanigan’s, concluding that there is no reasonable expectation of recurrence. That analysis is wrong for four reasons.

First, the repeal of the freeze-frame policy does not deliver a de facto win for the government. A repeal, to be sure, is “often” determinative of unambiguous termination, see Doe, 747 F.3d at 1322, but the key word is “often.” When the weight of the other facts suggests that the government’s moves were ambiguous, inconsistent, and made to manipulate jurisdiction, there is reason to fear that the government will veer from its new course. Relying on a sea of red flags, the district court found that this was exactly the case—that the government’s reasons for its repeal didn’t add up, and that it was thus unclear whether its old ways were gone for good. The majority again does not hold that any of the court’s findings are clearly erroneous.6 And given these findings, the only reasonable conclusion

6 In fact, the FDC even concedes that the district court’s findings are not clearly erroneous. See Oral Argument at 13:40–14:50 (conceding in response to a question about the clearly erroneous standard that we should “just give the judge those factual findings” and move on to the legal merits of the deliberate-indifference analysis).

54

under the factors is that Keohane’s claims are not moot. See id. at 1322–23.

Second, though the majority contends that this case is a clone of Flanigan’s, it clouds key parts of the picture. The City Council in Flanigan’s “twice voted on the relevant remedial measures,” “put forth persuasive explanations that [were] not dependent upon [the] litigation,” and unanimously repealed the policy in “open session during regularly scheduled meetings.” 868 F.3d at 1260. The City also had a “long history of non-enforcement,” which, “coupled with the recent repeal, indicate[d] the commitment to [its] new legislative scheme.” Id. at 1263 n.10.

We have none of that. We have no idea how many times the FDC considered these policy shifts—it didn’t tell us. We have no persuasive explanation for its about-face—it gave us none, and it met behind closed doors. See Harrell, 608 F.3d at 1267. And we have no history of looking the other way— the FDC, until this litigation, enforced these policies to their fullest extent. Simply put, Flanigan’s was a case in which the government publicly replaced—with good reasoning—a rule that it never enforced. Ours is a case in which the government privately replaced—with no reasoning—a rule that it enforced daily. Apples to oranges; Flanigan’s to Keohane.

Third, although the majority says that our facts differ from the facts it plucks from Doe, this case is closer to Doe than the majority admits. For instance, the majority claims that, in Doe, the government had a “pattern” of breaking its

55

promises to the plaintiff, but here there is no history of broken promises. Though we may not have explicit broken promises, we have a substantial comparator: The FDC has consistently defended its old policies and inconsistently applied its new ones. Its inconsistencies and contradictions raise the same flags as broken promises—they cast doubt on the FDC’s shaky commitment to its newfound path.

Finally, the majority draws a line through this case, Doe, and Flanigan’s, because here (like Flanigan’s and unlike Doe) the government has assured us that it will not revert to its old ways. But those assurances do not deserve the weight the majority gives them. For one thing, the district court found that the FDC never made this assurance about the freeze-frame policy, see Keohane, 328 F. Supp. 3d at 1300, and taking the FDC’s lawyer’s word at the final hour over the district court’s finding again ignores clear-error review. Equally important, the City Council in Flanigan’s backed up its oral-argument statements with actual statements: It passed a resolution “expressly disavowing any intent to reenact the Ordinance or any similar regulation.” 868 F.3d at 1262 (alterations accepted) (internal quotation mark omitted). This, coupled with the City’s “alternative reasons for the repeal” and “history of non-enforcement,” assured us that the City would keep its promise. See id. at 1263. The FDC, in contrast, has offered no security to guarantee its claims. And contrary to the majority’s view, there is “evidence or history that would cause us to doubt them here”: a record teeming

56

with temperamental positioning, clandestine decision-making, all-too-convenient timing, and an adamant refusal to admit the error of its ways. See id. 7

* * *

The majority quips that wisdom “often never comes, and so one ought not to reject it merely because it comes late.” But the district court didn’t find that wisdom had come late; it found that wisdom had never come at all. It concluded that the FDC’s reversals were born of desperation, not deliberation. And its holding stood on a host of findings: that the timing of the FDC’s concessions was suspect; that the FDC had no explanation for its delay; that the FDC’s positions throughout the litigation were inconsistent; that the FDC’s decision-making was a black box; that the FDC’s prior practices were not accidental, but deliberate and historical; that the FDC refused to promise that it would not re-enact the freeze- frame policy; that the FDC still was adamant that its practices were valid, even after it claimed to change its ways; that the FDC delayed in providing Keohane’s hormone therapy, even after it agreed that she needs it; and that, on at least one

7 The majority also emphasizes the supposed hoops the FDC would need to jump through to bring back the policy, but a review of the oral-argument recording reveals that all it would take is a run-of-the-mill internal review. See Oral Argument at 4:23. Apparently, this doesn’t take awfully long, given that the FDC changed the freeze-frame policy in just two-months’ time. So the protection that these procedural hurdles offer is slight, if any at all.

57

occasion, the FDC applied the repealed freeze-frame policy to bar hormone therapy for a patient with gender dysphoria.

As the majority does not hold that any of these findings are clearly erroneous, they bind us. See Pelphrey, 547 F.3d at 1268. And these findings show that the FDC has not apologetically turned over a new leaf, but has acted to manipulate jurisdiction. I would affirm the district court’s holding that these claims are not moot. 8 II.

Next, the merits of the social-transitioning claim. The Eighth Amendment bars a prison official from being deliberately indifferent to a serious medical need. See Brown v. Johnson, 387 F.3d 1344, 1351 (11th Cir. 2004). A deliberate- indifference claim thus has two components: an objectively serious medical need, and subjective deliberate indifference to that need. See id. We review de novo the district court’s ultimate conclusion that there was an Eighth Amendment violation warranting equitable relief, and we review issues of fact supporting this conclusion for clear error. See Common Cause/Georgia v. Billups, 554 F.3d 1340, 1349 (11th Cir. 2009).

8 As for the merits of these claims, I agree with the majority that the FDC was deliberately indifferent to Keohane’s serious medical need when it subjected her to the freeze-frame policy. I would also affirm that the failure to provide hormone therapy, despite Keohane’s undisputed need for the therapy, arises to deliberate indifference. Because the majority does not dispute these conclusions, I will not analyze them further.

58

An objectively serious medical need is “one that has been diagnosed by a physician as mandating treatment or one that is so obvious that even a lay person would easily recognize the necessity for a doctor’s attention.” Farrow v. West, 320 F.3d 1235, 1243 (11th Cir. 2003). We review the legal conclusion that a medical need is objectively serious de novo. See Thomas v. Bryant, 614 F.3d 1288, 1307 (11th Cir. 2010). 9 Because both sides and the majority agree that gender dysphoria is an objectively serious medical need, only the subjective element is in dispute.

To establish subjective deliberate indifference, a plaintiff must show “(1)

subjective knowledge of a risk of serious harm; (2) disregard of that risk; and (3) by conduct that is more than mere negligence.” Brown, 387 F.3d at 1351. Unlike the objectively-serious-need element, the subjective-deliberate-indifference element presents a question of fact that we review for clear error. Thomas, 614 F.3d at 1312.

The district court found that the FDC was deliberately indifferent to Keohane’s gender dysphoria when it refused to let her wear female undergarments or use female grooming products. This finding has substantial support in the record. So I’d let it stand. And since both the subjective and objective elements of an Eighth Amendment violation are met, I would affirm.

9 We review any underlying fact issues—like what the medical need is—for clear error. See Thomas, 614 F.3d at 1307.

59

The majority reaches a different result, however, and it uses the wrong standard of review to get there. Reviewing the subjective-deliberate-indifference finding de novo during its review of the “ultimate” Eighth Amendment violation, it concludes that the district court was wrong to find that the last two subparts of the deliberate-indifference element were met. 10 I explain below why this analysis is incorrect and why we must affirm the district court’s conclusion that the FDC was deliberately indifferent to Keohane’s gender dysphoria.

A.

The majority’s first mistake comes in articulating our standards of review.

While listing the standards, the majority agrees that we apply clear-error review to questions of fact supporting the district court’s conclusion that a defendant violated the Eighth Amendment. See Majority Op. at 28. It also accepts that subjective deliberate indifference is a question of fact that we review for clear error. See id. at 28 n.8. And yet, despite these directives, the majority refuses to apply clear-error review to the district court’s finding of subjective deliberate indifference. Instead, it insists that it retains de novo review over this factual finding, citing the unremarkable rule that the “ultimate determination” whether “there was an Eighth Amendment violation warranting equitable relief” is a legal conclusion that we

10 The majority rightly assumes that the FDC knew that Keohane’s gender dysphoria put her at substantial risk of self-harm, which satisfies the first subpart. See Majority Op. at 30.

60

review de novo. Id. at 27–28. In other words, the majority has somehow read Thomas to hold that, even though the underlying finding of deliberate indifference is a question of fact reviewed for clear error, we (really) review that finding again de novo when we consider the “ultimate” Eighth Amendment violation.

That is simply wrong. We have long reviewed a finding of subjective deliberate indifference for clear error. See, e.g., Thomas, 614 F.3d at 1312. We can reverse a subjective-deliberate-indifference finding only if left with a firm and definite conviction that the district court made a mistake. See Pelphrey, 547 F.3d at 1268. Our de novo review extends only to questions of law (i.e., the objectively- serious-need element) and to the district court’s ultimate conclusion whether the objective and subjective elements of a deliberate-indifference claim state an Eighth Amendment violation. Precedent makes this clear.

Take Thomas—the case from which the majority derives its de-facto-de-

novo rule. There we explained that we review de novo the district court’s ultimate determination that there was an Eighth Amendment violation. See Thomas, 614 F.3d at 1303. We also explained that we review for clear error questions of fact supporting this conclusion. Id. Against this backdrop, we analyzed the two elements of a deliberate-indifference claim. We first reviewed the objectively- serious-need prong de novo, concluding as a matter of law that the prisoner’s medical needs were sufficiently serious under the Eighth Amendment. See id. at

61

1307–13. Then we analyzed the subjective-deliberate-indifference finding. Citing Supreme Court and Eleventh Circuit precedent, we held—unequivocally—that the subjective-deliberate-indifference element raises a “question of fact which we review for clear error.” Id. at 1312 (citing Farmer v. Brennan, 511 U.S. 825, 842 (1994); Goebert v. Lee Cty., 510 F.3d 1312, 1327 (11th Cir. 2007)). And then we did just that: We reviewed the district court’s finding of subjective deliberate indifference for clear error, holding that we could not reverse on the subjective- deliberate-indifference prong, because the finding was not “clearly erroneous.” Id. at 1313–16. Finally, we summed up by “[c]oncluding that [the plaintiff] satisfied both the objective and subjective prongs of his Eighth Amendment” claim, leading us to affirm the district court’s ultimate conclusion that there was an Eighth Amendment violation. Id. at 1317.

That is precisely how this analysis should go. We review the objectively-

serious-need element de novo, as it is a question of law. Id. at 1307. We review the subjective-deliberate-indifference finding for clear error, as it is a question of fact. Id. at 1312. And then we review the district court’s “ultimate” application of these elements de novo. Id. at 1303. So if the district court, despite checkmarks in both the objective and subjective boxes, still concluded that there was no Eighth Amendment violation, we would lend no deference to this error. We would review it de novo, and would no doubt reverse. And if the district court, despite holding

62

that one of the elements was not met, still concluded that there was an Eighth Amendment violation, we would do the same. We would review this error de novo, and no doubt reverse. That is the ultimate conclusion that we review de novo. See id.11 But that is not how the majority analyzed this case. If you look closely through its opinion, you won’t see a single attempt to analyze whether the district court’s subjective-deliberate-indifference finding was clearly erroneous. To be sure, the majority pays lip service to this standard at the end of footnote eight. But that perfunctory paragraph is no more than a fail-safe to cover itself should its de novo rule prove too much. Even a skim through its opinion shows that the majority has not applied clear-error review to the district court’s finding of subjective deliberate indifference; it has swapped the deference we typically apply with overarching de novo review. This switch allows it to reweigh the deliberate- indifference evidence as it sees fit, disregarding the ample evidence the district court relied on to make its factual findings. Contra id. at 1312.

11 Indeed, I’m curious how the majority thinks these standards work otherwise. What is the point of initial clear-error review for a finding of subjective deliberate indifference if we review that finding again de novo when we review the ultimate Eighth Amendment violation? The majority doesn’t tell us. It simply recites these two standards, with no attempt to make sense of how they work together, and chooses the one endowing it with a greater level of review.

63

B.

To justify its new standard, the majority pens a footnote treatise that reads the clear-error rule out of Thomas. See Majority Op. at 28–29 n.8. It first opines that Thomas’s clear-error rule applies only to “historical facts” supporting the district court’s finding of subjective deliberate indifference—i.e., the who, what, when, and where facts—not the determinative facts. It then concludes that Thomas compels it to review the subjective-deliberate-indifference finding de novo during its review of the ultimate Eighth Amendment violation. And it supports these claims with out-of-context Supreme Court precedent, asserting that its ultimate factual review “follows straightaway” from the Court’s application of de novo review in cases predating Thomas. For three reasons, these arguments fall short.

First—as the majority well knows—it does not matter what we think the prior panel should have held under then-existing Supreme Court precedent: All that matters is what the prior panel held. See In re Lambrix, 776 F.3d 789, 794 (11th Cir. 2015) (“[A] prior panel’s holding is binding on all subsequent panels unless and until it is overruled or undermined to the point of abrogation by the Supreme Court or by this court sitting en banc.” (emphasis omitted)). In Thomas, we held— with the benefit of all the Supreme Court precedents the majority cites—that subjective deliberate indifference is a “question of fact which we review for clear error.” 614 F.3d at 1312. There were no qualifiers—we did not say that clear-

64

error review applies to only the “historical” subparts of the district court’s deliberate-indifference finding. Nor did we double back to review the subjective- deliberate-indifference finding de novo at the end; we did just the opposite, affirming the subjective-deliberate-indifference finding because it was not clearly erroneous. See id. at 1313–17. The majority cannot reexamine the legal landscape the prior panel considered to conclude that the prior panel was wrong—such second-guessing would “undermine the values of stability and predictability in the law that the prior panel precedent rule promotes.” Smith v. GTE Corp., 236 F.3d 1292, 1303 (11th Cir. 2001). For this reason, we have “categorically reject[ed] any exception to the prior panel precedent rule based upon a perceived defect in the prior panel’s reasoning or analysis as it relates to the law in existence at that time.” Id. If the majority has a problem with Thomas’s clear-error rule, the proper place for its concerns is in a concurrence urging our court to consider the issue en banc. What it cannot do is what it does here: overrule a co-equal panel and take the en banc court’s role for itself.12

12 At any rate, the Supreme Court cases the majority cites did not arise in the context of deliberate indifference. See Majority Op. at 28–29 n.8 (citing cases in the excessive-fine, punitive-damages, and probable-cause contexts). Crafting a standard of review is a contextspecific inquiry, so we have no clue whether the Supreme Court would apply clear-error or de novo review to a deliberate-indifference finding. Thankfully, we need not read any tea leaves— Thomas already concluded that clear error is the right standard of review in our circuit. But the majority’s reliance on off-point Supreme Court precedent shows that its analysis is “less an application of existing precedent than a prediction of what the Supreme Court will hold [should] it choose[] to address this issue in the future.” United States v. Greer, 440 F.3d 1267, 1275 (11th Cir. 2006). This, we have held, the majority cannot do. See Lambrix, 776 F.3d at 794.

65

Second, Thomas foreclosed the majority’s academic notion that clear-error review extends only to historical facts. As said before, Thomas’s holding had no reservations: “A prison official’s deliberate indifference is a question of fact which we review for clear error.” 614 F.3d at 1312. The majority does not explain how this rule statement applies only to historical facts. Nor could it do so—Thomas itself applied the rule beyond historical facts. Just pages after Thomas set out the ultimate-de-novo-review rule that the majority clings to, the panel applied clear- error review to all three subparts of the subjective-deliberate-indifference element. It first held that the record supported the district court’s finding that prison officials were subjectively aware of a risk of harm to the prisoner. See id. at 1313. Then it took on the second and third prongs, holding that the “record also supports the district court’s finding that the Secretary of the DOC and the Warden of FSP recklessly disregarded the risk of psychological harm to inmates like McKinney.” Id. at 1315. If this application weren’t clear enough, the panel erased any doubt when it held that “the DOC’s refusal to modify its non-spontaneous use-of-force policy provides support for the district court’s finding of more than mere or even gross negligence on the part of the DOC.” Id. And the panel confirmed that clear- error review applies to the entire deliberate-indifference element, holding that “an examination of his entire record demonstrates that the district court did not commit clear error in finding the defendants’ deliberate indifference.” Id. at 1317. These

66

findings are not historical facts; they are the determinative facts that make up a finding of subjective deliberate indifference. See Brown, 387 F.3d at 1351. So Thomas’s application proves that clear-error review extends beyond the who, what, where, and when—it extends to the entire deliberate-indifference element.

Thomas isn’t the outlier in our precedent—it’s the norm. We have held time and again that subjective deliberate indifference is a factual finding. See, e.g., Greason v. Kemp, 891 F.2d 829, 840 (11th Cir. 1990) (holding at summary judgment that the “evidence could support a finding that the conduct reflected a deliberate indifference to [the prisoner’s] [E]ighth [A]mendment right to adequate mental health care,” and holding that whether “the conduct actually constituted deliberate indifference . . . is a factual question” (emphasis added in underline)); McElligott v. Foley, 182 F.3d 1248, 1256 (11th Cir. 1999) (holding at summary judgment that there “was sufficient evidence to permit a jury to infer that the defendants in this case knew of a substantial risk of harm to [the prisoner]” and “to draw the conclusion that [the defendants] were not merely negligent” in providing subpar care (emphasis added)). We review factual findings like these for clear error. See, e.g., United States v. Williams, 340 F.3d 1231, 1234 (11th Cir. 2003). We don’t dissect them with artificial labels, defying prior precedent.

In fact, the type of review the majority presses here echoes the sole instance where we do dissect facts with artificial labels—our de novo review of

67

“constitutional facts” in First Amendment cases. In those cases, the constitutional facts are determinative; they answer “why” a government actor suppressed certain speech and whether its motives were unconstitutional. See Am. Civil Liberties Union of Fla., Inc. v. Miami-Dade Cty. Sch. Bd., 557 F.3d 1177, 1206 (11th Cir. 2009). We review those facts de novo and accompanying historical facts for clear error. See id. But this an exceedingly narrow exception—one unique to First Amendment cases. See Flanigan’s Enters., Inc. of Ga. v. Fulton Cty., Ga., 596 F.3d 1265, 1276 (11th Cir. 2010). We have never imported this plenary factual review to an Eighth Amendment case. Cf. Thomas, 614 F.3d at 1312. And yet, despite Thomas, the majority does so now.

Third, the majority’s super-de-novo-review rule proves unjustifiable when we consider how we review subjective deliberate indifference after a jury trial (rather than after a bench trial). For when a jury finds that a defendant was deliberately indifferent, we don’t retrace this finding de novo; we review the jury’s finding for sufficient evidence. See Carswell v. Bay Cty., 854 F.2d 454, 457 (11th Cir. 1988) (affirming a jury’s finding of deliberate indifference because there was sufficient evidence to reach that conclusion). Why would we review subjective deliberate indifference for sufficient evidence after a jury trial, but de novo after a bench trial? The answer is simple: We don’t. We have never suggested in these cases that we somehow revisit the factual finding of subjective deliberate

68

indifference de novo when reviewing the “ultimate” Eighth Amendment violation. And the majority does not point us to a single Eleventh Circuit case that does so.

Turning to the second paragraph in its footnote, the majority claims that there is no way that the purportedly “mindless, mechanical box-checking” that I described above marks the extent of our de novo review over the ultimate Eighth Amendment violation. See supra at 61–62. This is a mountain made out of a molehill—of course our de novo review has less teeth here than in most constitutional cases. For one, the defendants here have conceded that the objectively-serious-need element—one we review de novo—is met, leaving us little to review on that side of the coin. For another, our precedent has set up a clear, two-part test for establishing an Eighth Amendment violation, and one of those parts is a fact question, leaving little to ultimately review de novo. See Thomas, 614 F.3d at 1312. In other cases, like an Equal Protection challenge or a Due Process case, our ultimate review of the constitutional violation driving the preliminary injunction is far more searching—the questions there are almost entirely legal. But a finding of subjective deliberate indifference is different; it is heavily fact-intensive. See Greason, 891 F.2d at 837. Though we are just as equipped as the district court to decide the legal question of whether a medical need is objectively serious, the district court is far better situated to analyze the subjective, state-of-mind question of a prison official’s deliberate indifference. See

69

Salve Regina, 499 U.S. at 233 (noting that deference is warranted when the district court is better positioned decide the issue, as it is when applying a fact-dependent legal standard). So our “ultimate” de novo review, in the deliberate-indifference context, is understandably more limited.

As its parting word, the majority notes that the First Circuit, sitting en banc, rejected clear-error review for subjective deliberate indifference, ultimately applying the de novo review the majority applies here. See Kosilek v. Spencer, 774 F.3d 63, 84 (1st Cir. 2014) (en banc). But that’s because the First Circuit concluded that deliberate indifference is a “[s]ubsidiary legal question,” not a fact question. Id. That is, word for word, the opposite of what we held in Thomas. See 614 F.3d at 1312. The First Circuit also justified its de novo review of the deliberate-indifference finding by citing our statement in Thomas that the “ultimate legal conclusion of whether prison administrators have violated the Eighth Amendment is reviewed de novo.” 774 F.3d at 84. But, as explained above, Thomas’s ultimate-de-novo-review rule only extends to the district court’s application of the objective and subjective elements; it does not empower us to rereview the subjective-deliberate-indifference-finding de novo. See supra at 60– 62, 65–69. Finally, the Kosilek court was sitting en banc; was not bound by prior precedent; and, so far as Kosilek explains, did not have precedent squarely holding that deliberate indifference is a question of fact reviewed for clear error. We, in

70

contrast, are not sitting en banc; are bound by prior precedent; and do have precedent squarely on point. See Thomas, 614 F.3d at 1312. So Kosilek holds no weight in this analysis. 13

* * *

In the end, “our first obligation—our oath” is to follow the law. Majority Op. at 39 (citing 28 U.S.C. § 453). In the Eleventh Circuit, that means following prior precedent. Thomas holds that subjective deliberate indifference is a question of fact that we review for clear error. We cannot refuse to apply this holding simply because we disagree. Nor can we reanalyze the issue for ourselves to overrule the prior panel. If the majority has concerns about our precedent, it should voice those concerns separately for our en banc court. And it may have the

13 Another important fact: The First Circuit decided Kosilek with a bare 3-2 majority, and the two dissenting judges strenuously dissented from the majority’s de novo review of the subjective- deliberate-indifference finding. Since we are not sitting en banc, I need not explain—as the Kosilek dissenters did—why a court might conclude that subjective deliberate indifference is a finding best reviewed for clear error. But there are indeed reasons why a court might do so. See 774 F.3d at 98–100 (Thompson, J., dissenting). As Justice Scalia noted in Ornelas v. United States, “[l]aw clarification requires generalization, and some issues lend themselves to generalization much more than others.” 517 U.S. 690, 703 (1996) (Scalia, J., dissenting). When the issues are fact-specific and not easy to generalize, “probing appellate scrutiny will not contribute to the clarity of legal doctrine” and deference to the district court is warranted. Salve Regina, 499 U.S. at 233. A subjective-deliberate-indifference inquiry is fact-specific, motivebased , and not due for easy generalization. One could thus say that deference to the district court is warranted, as an appellate court cannot “hope to match the district judge’s expertise in these areas.” Kosilek, 774 F.3d at 100. But we need not say that—we have already made that determination. See Thomas, 614 F.3d at 1312.

71

chance to do just that, as its disregard for our precedent has no doubt transformed this routine deliberate-indifference case into one justifying en banc review.

C.

Having minted a new standard of review, the majority applies it to reverse the district court at the last deliberate-indifference step: Whether the FDC “disregarded” a substantial risk of harm by “more than mere negligence.” The district court found that the FDC had for two reasons. First, the FDC denied social transitioning because it blindly deferred to the FDC’s clothing policy, effectively enacting a blanket ban on social transitioning without case-specific medical or security judgment. Second, the FDC denied Keohane access to medical personnel competent enough to realize that she needs to transition to avoid severe self-harm. Reviewing de novo, the majority replaces these findings with its own. It concludes that the FDC denied treatment because medical professionals disagreed with Keohane about her need to transition and because the FDC concluded that the security risks of the treatment were too great.

To reach that conclusion, though, our precedent compels the majority to hold that the district court’s findings were clearly erroneous. See Pelphrey, 547 F.3d at 1268. The majority doesn’t do so (and its cursory footnote is no substitute for true clear-error review). See Majority Op. at 29 n.8. In reality, the majority takes the issue up anew, concluding that it “simply cannot say that the FDC consciously

72

disregarded a risk of serious harm by conduct that was more than mere negligence.” Majority Op. at 37 (internal quotation mark omitted).

As I explain below, that is the wrong approach. Because both of the district court’s findings hold substantial footing in the record, we must affirm. See Pelphrey, 547 F.3d at 1268.

1.

I’ll start with the blanket-ban finding. The denial of medical care based on blind deference to a blanket rule, rather than on an individualized medical determination, violates the Eighth Amendment. See Kosilek, 774 F.3d at 91; Roe v. Elyea, 631 F.3d 843, 862 (7th Cir. 2011); Colwell v. Bannister, 763 F.3d 1060, 1068 (9th Cir. 2014); Soneeya v. Spencer, 851 F. Supp. 2d 228, 247 (D. Mass. 2012); Brooks v. Berg, 270 F. Supp. 2d 302, 310 (N.D.N.Y. 2003) (vacated in part on other grounds). So does denying treatment for non-medical reasons. McElligott, 182 F.3d at 1255. The district court found that the FDC refused to consider Keohane’s need for social transitioning because treatment team members blindly deferred to prison policy, without evaluating her specific circumstances. That conduct arises to subjective deliberate indifference. See Kosilek, 774 F.3d at 91; McElligott, 182 F.3d at 1255. And the record supports this finding.

For starters, one member of Keohane’s treatment team admitted that she “never assessed whether Ms. Keohane has a mental-health need for longer hair or

73

access to female undergarments because . . . [the FDC’s] policies prohibit these things.” She also admitted that she doesn’t think the FDC would even permit a medical pass for social transitioning, and so she focused Keohane’s therapy “on coping without access to this particular treatment.” The FDC’s regional medical director further testified that she did not consider any medical exceptions to the FDC’s clothing policy. And another member of Keohane’s treatment team even conceded that the team discussed whether Keohane needed access to female clothing, but ultimately concluded that “it is out of our hands, that we understand, but there’s nothing we can do.”

Perhaps more concerning, these blanket denials didn’t start from the bottom;

they came from the top. The district court found that the FDC’s final decisionmaker—its chief medical officer—would have refused an exception even if treatment team members had recommended it, solidifying the FDC’s blanket ban on social transitioning. Indeed, the officer testified that it would be a “hard sell” for him to grant a medical exception for social transitioning, regardless of the inmate’s particular needs. The district court also found that the officer made this decision without considering Keohane’s specific circumstances, as he “has never

74

decided this issue, nor has he been presented with any medical request for any exceptions to security policies to allow for social transitioning.”14 Given the testimony from treatment team members and the chief medical officer’s steadfast and unreasoned refusal to consider social-transitioning treatment, the record supports the district court’s finding that the FDC categorically denied Keohane treatment under a blanket policy, without considering her individual circumstances. As all agree, “responding to an inmate’s acknowledged medical need with what amounts to a shoulder-shrugging refusal even to consider whether a particular course of treatment is appropriate is the very definition of ‘deliberate indifference’—anti-medicine, if you will.” Majority Op. at 14. Since this finding is not clearly erroneous, we must affirm it, even if we would have found differently. See Multiponics, 622 F.2d at 723.

And yet the majority does find differently: It crowns security king. Citing out-of-circuit precedent, the majority holds that the FDC can shrug off Keohane’s medical need if it decides that the security risks of the treatment outweigh its necessity. See Kosilek, 774 F.3d at 92. We need not get into the limits of this rule

14 Alongside this, the district court found that the chief medical officer’s prejudgment was also born of “ignorance of gender dysphoria and bigotry toward transgender individuals in general,” further leading the court to conclude that the medical officer’s refusal to provide treatment was categorical and not based on medical need. The record supports this finding as well. See Keohane, 328 F. Supp. 3d at 1306 & n.11 (noting that the medical officer “thinks treating gender dysphoria by encouraging the transition of gender roles ‘goes against nature’” and that the officer admits that his religion plays into his views on transgender people in general).

75

today, 15 because that’s not what happened here. More specifically, that’s not what the district court found happened here. The district court did not conclude that the FDC denied treatment because it considered Keohane’s need for social transitioning and decided that the security risks outweighed her need. The court found that the FDC did not consider necessity or security at all when denying treatment, because prison officials blindly deferred to the FDC’s clothing policy. Said differently, prison officials denied treatment because of the policy, not because of their views on her need for the treatment or the security risks it presents.

That situation fails even under Kosilek, the case from which the majority draws its security exception. Kosilek held that so “long as prison administrators make judgments balancing security and health concerns that are ‘within the realm of reason and made in good faith,’ their decisions do not amount to a violation of the Eighth Amendment.” Id. A fair enough rule—prison administrators can of course balance a prisoner’s need for treatment with security concerns and fairly conclude that the risks outweigh the need. But the prison must “balance” the competing interests—something that the district court found the FDC did not do.

15 Although there must be some limits—I would be deeply troubled if the majority thought that a prison can withhold truly life-saving treatment for security’s sake. See Brown v. Plata, 563 U.S. 493, 511 (2011) (“A prison that deprives prisoners of basic sustenance, including adequate medical care, is incompatible with the concept of human dignity and has no place in civilized society.”).

76

In the district court’s view, the FDC did not balance Keohane’s medical needs with prison security and decide that security carried the day. Prison officials denied treatment solely because they thought prison policy forbade social transitioning— the strength of Keohane’s medical need had nothing to do with it. A prison cannot balance security over medical necessity if it never factored medical necessity into the analysis. So the district court’s finding that the FDC failed to consider Keohane’s medical need when denying her treatment precludes the FDC from claiming that security won the day, leaving it liable for its deliberate indifference. See id. at 91; Roe, 631 F.3d at 862; Colwell, 763 F.3d at 1068; McElligott, 182 F.3d at 1255; Soneeya, 851 F. Supp. 2d at 247.

Worse, the FDC didn’t just fail to consider Keohane’s medical need; it failed to consider security too. Prison officials did not evaluate the security risks of social transitioning and deem them too great—they assumed that FDC policy would always forbid the treatment, and that was the end of it. Based on this belief, treatment team members “couldn’t even fathom requesting an exception to [the policy] even if the inability to socially transition drives a patient to suicide.” Nor did they elevate the issue to the chief medical officer—the one who has the final say on granting medical exceptions to security policy. So far as her treatment team was concerned, “Keohane simply can’t transition because [the FDC] does not permit inmates housed in its male facilities access to the clothing and grooming

77

standards it applies to female inmates.” A weighing of security risks had no role in the matter.

Adding insult to injury, treatment team members were wrong that security policy always forbids social transitioning. Contrary to their view, the FDC’s security representative testified that medical staff—not security staff—has the ultimate say in granting a medical exception to prison security policy. The representative even listed examples of how the FDC could accommodate social transitioning, despite potential security risks. The FDC’s chief medical officer affirmed this procedure, testifying that medical (not security) makes the final call when it comes to medical exceptions. And the FDC confirmed this procedure yet again here, stipulating that it would accommodate social transitioning “if having longer hair or female undergarments or makeup were deemed to be medically necessary for an inmate with gender dysphoria.” 16 In sum, then, medical staff, not security policy, has the final say on whether security risks outweigh medical need. Yet, in a catch-22, medical staff denied

16 This policy makes sense. When “society takes from prisoners the means to provide for their own needs,” prisons must provide the “necessary medical care.” Plata, 563 U.S. at 510. If necessary medical care creates security concerns, the prison must make accommodations. The FDC does just that: If medical officials say something is medically necessary, security officials make it work. But security officials never considered how to make social transitioning work here, because the treatment team never tried to seek an exception. Even though the FDC’s own policy makes clear that security bends the knee to medicine, Keohane’s treatment team washed its hands of the matter the moment it concluded that FDC policy forbids social transitioning. That arises to deliberate indifference. See Kosilek, 774 F.3d at 91.

78

social transitioning here because they thought that security policy barred social transitioning. And while the staff pointed fingers, no one evaluated whether the security risks of social transitioning outweighed Keohane’s specific medical need.

Which brings us to what distinguishes this case from Kosilek. There, unlike here, the record showed that prison officials extensively considered the inmate’s medical need and whether the requested treatment would create security concerns, ultimately concluding that the need was too little and the concerns were too great. See Kosilek, 774 F.3d at 73–75, 79–84 (chronicling the prison’s in-depth security balancing); id. at 95 (“[T]he [prison] testified consistently that it believed the postoperative security concerns surrounding Kosilek’s treatment were significant and problematic.”). The record here shows, by contrast, that Keohane’s medical need and the security risks of her treatment played zero part in the FDC’s decision to withhold social transitioning. That decision was driven by blind deference to FDC policy, without regard for medical need, no matter how dire the straits.

In the end, the district court said it well: “What’s clear from the treatment team’s testimony is that everybody knows Ms. Keohane has harmed herself and attempted suicide, but still, nobody has requested any exceptions to [the FDC’s] male grooming and clothing policies to treat her gender dysphoria.” They failed to do so not because they balanced security risks with individual medical need, but because they (erroneously) thought that prison policy forbade social transitioning,

79

regardless of the circumstances. That situation does not fall within Kosilek’s security exception. It is a categorical, blanket ban on social transitioning and a level of disregard that rises above mere negligence. See id. at 91; Colwell, 763 F.3d at 1068; McElligott, 182 F.3d at 1255. Since this finding is well supported by the record, we must affirm. See Pelphrey, 547 F.3d at 1268.

2.

Next is the incompetent-personnel finding. A prison official disregards a substantial risk of harm by more than mere negligence when the official provides physicians who are incompetent to adequately treat a prisoner’s serious medical need. See Ancata v. Prison Health Servs., Inc., 769 F.2d 700, 704 (11th Cir. 1985). The record shows that Keohane’s treatment team was not qualified to treat her illness and thus provided subpar care. As the district court found, no member of Keohane’s treatment team had ever treated a pre-transition patient with gender dysphoria. In fact, most of her team members had never treated a patient with gender dysphoria, period. Team members were not trained in the World Professional Association for Transgender Health (WPATH) standards—standards that the district court (and many others) have found authoritative for treating gender dysphoria in prison. See also Edmo v. Corizon, Inc., 935 F.3d 757, 769 (9th Cir. 2019) (collecting cases). Keohane’s treatment team leader conceded that she was not qualified to decide whether Keohane needed social transitioning. And

80

the FDC’s chief medical officer admitted that he “doesn’t know one way or the other if social transitioning is helpful in treating gender dysphoria,” even though that treatment is standard for treating the illness.

To this adds that the FDC’s staff psychiatrist—a psychiatrist the FDC brought in post-litigation to say that Keohane doesn’t need social transitioning— admitted that he lacked knowledge about the proper standard of care for gender dysphoria, that he had only read the parts of Keohane’s record relevant to her psychiatric need (Keohane has never taken psychiatric medication while in FDC custody), and that he had never before “evaluated anyone in prison to determine a medical need for access to clothing or grooming standards to treat gender dysphoria.” This led the district court to fairly conclude that his views deserved little, if any weight.

Prison officials have a tough job, but the Constitution requires that they be prepared to treat the inmates they take into their custody. See Ancata, 769 F.2d at 704. The record here shows that the FDC was ill-equipped to treat Keohane’s gender dysphoria, which ultimately led it to withhold necessary social-transitioning treatment. The district court was therefore within its bounds to find that the FDC’s incompetence arose to disregard by more than mere negligence. And the majority’s paragraph-long footnote replacing this detailed finding with its own

81

only confirms that the majority is not reviewing for clear error, as it must under our precedent. See Thomas, 614 F.3d at 1312.

3.

Despite the district court’s findings, the majority colors this case one of disagreement, not disregard. The argument is this: The FDC did not disregard a substantial risk of harm by more than mere negligence, because “members of Keohane’s medical-treatment team, Wexford’s staff psychiatrist, the FDC’s chief clinical officer, and the FDC’s retained expert” merely disagreed with Keohane and genuinely believed that hormone therapy sufficed to treat her gender dysphoria. The majority also taps Kosilek again for help, claiming that there, as here, medical officials disagreed over the right course of treatment.

This analysis is off for a few reasons. Chief among them, and as I said before, the district court found that the treatment team’s position on Keohane’s need for social transitioning played zero role in its decision to withhold treatment. Without ruling that this finding was clearly erroneous, we cannot hold that the district court’s ruling was incorrect. See Pelphrey, 547 F.3d at 1268.

At any rate, the majority couldn’t hold that the court’s findings were clearly erroneous even if it applied the right standard. This is because there is no genuine dispute or difference in medical opinion in this record. Both experts agreed at trial that Keohane should have this treatment. Team members recognized that Keohane

82

remained in pain despite hormone therapy and knew that social conditioning could help her pain. And the district court found that the few officials who said social transitioning was unnecessary were incompetent and incredible.

Starting with the experts, Keohane’s trial expert testified that social transitioning was medically necessary to treat Keohane’s severe gender dysphoria. The FDC’s trial expert also testified that medically necessary treatment can be treatment that is “psychologically pleasing to the patient” (sensible enough— gender dysphoria is, of course, a psychological illness). The expert then agreed that letting Keohane wear female underwear and grow out her hair would be “psychologically pleasing” to her. He also noted that Keohane could be vulnerable to “acute decompensation” and would have “a suicidal ideation and crisis” if she were denied access to social transitioning. Given that the experts agreed that this treatment would be deeply helpful for Keohane’s mental state (and that she would be at great risk of self-harm without it), the court fairly found that “[e]xperts on both sides agreed at trial that [the FDC] should allow Ms. Keohane access to female clothing and grooming standards to treat her gender dysphoria.”17

17 The majority thinks that I believe an inmate is always entitled to medical care that is “psychologically pleasing.” I believe no such thing. My view—and what the law requires—is that a prison must provide an inmate with medical care that is psychologically pleasing if that care is medically necessary. See Plata, 563 U.S. at 510. Here, the FDC’s expert opined that care that is “psychologically pleasing” can be medically necessary for a patient with gender dysphoria. Given this testimony, the expert’s accompanying testimony, and other medical testimony in the case, the district court found that there was no genuine dispute: Social

83

Along with this, treatment team members knew that Keohane remained in serious pain despite receiving hormone therapy. They knew that she still suffered from suicidal ideation and severe psychological harm. They also knew the cause of this pain: Keohane’s inability to express herself as a woman. Keohane told them so in grievance after grievance. She told them again at trial. And her words weren’t her only symptoms. She also tried to kill herself twice after a string of forced haircuts—first by hanging herself with a sheet from her bunk, and then by tying a pants leg around a door handle, tying the other leg around her neck, and sitting down on the floor to cut off the blood flow.

And that’s not all treatment officials knew. They also knew that social transitioning is an effective way to treat gender dysphoria. According to one team member, “[i]t allows you to express yourself in the gender that you feel yourself to be . . . [and i]t helps with self-esteem, it helps with expression, [and] it helps with . . . emotions.” Team members also knew that an individual may need both hormone therapy and social transitioning to adequately treat the disease.

To be sure, there were a few prison officials who testified that they don’t think Keohane needs social transitioning. Putting aside the fact that this is not why

transitioning is medically necessary to treat the psychological harm flowing from Keohane’s severe gender dysphoria. So, to be clear, it does not matter if care is pleasing to an inmate; all that matters is whether the care is medically necessary. For a psychological illness like gender dysphoria, then, it makes perfect sense that medically necessary treatment might be treatment that eases the inmate’s psychological pain.

84

they denied her treatment, see supra at 72–79, the district court also found them incompetent and their views incredible. Some of these officials think that a treatment is not medically necessary unless it is a matter of life or death—a frighteningly incorrect view of medical necessity. See Keohane, 328 F. Supp. 3d at 1310; Gayton v. McCoy, 593 F.3d 610, 620 (7th Cir. 2010) (“A medical condition need not be life-threatening to be serious; rather, it could be a condition that would result in further significant injury or unnecessary and wanton infliction of pain if not treated.”). These witnesses also lack experience with WPATH standards and gender dysphoria in general. See supra at 79–81. And the court gave “little if any weight” to the FDC’s staff-psychiatrist-turned-armchair-quarterback, because he was not an expert, was not a treating physician, had spoken with Keohane for less than an hour, had not reviewed Keohane’s history of self-harm, and had reviewed only the psychiatric section of her chart (even though she has never taken psychiatric medicine while in FDC custody). So even if medical disagreement could have played into the FDC’s social-transitioning ban (it did not), there is still no credible medical testimony in this record saying that Keohane does not need social transitioning. As the reviewing court, we must respect these credibility determinations, not remake them. See Salve Regina, 499 U.S. at 233.

Which again brings us to why Kosilek does not apply. There, unlike here, the record showed that medical officials were genuinely and fervently divided on

85

the appropriateness of gender-reassignment surgery. Several prison doctors concluded, without reservation, that gender-reassignment surgery was not medically necessary. One doctor even testified that surgery was the wrong treatment for the prisoner. Kosilek, 774 F.3d at 72. There was also a decreased risk of self-harm, as the prisoner had not tried to harm herself throughout her twenty-year incarceration. Id. at 69.

We have just the opposite. Keohane’s self-destructive tendencies have only ramped up in recent years. See supra at 52, 83. Keohane’s expert says social transitioning is medically necessary. The FDC’s expert testified that social transitioning would be “psychologically pleasing” for Keohane, which, in the expert’s view, can be a medically necessary treatment for a psychological illness like gender dysphoria. And everyone on Keohane’s treatment team failed to factor Keohane’s need for the treatment into their decisions (either because of unyielding deference to prison policy or a warped view of medical necessity and Keohane’s medical situation).

Given all this, the district court was within its bounds to find that the FDC denied treatment not because of a genuine belief that social transitioning was unnecessary, but because of blind deference and medical incompetence. The majority does not explain how these findings lack record support. Nor could it, as it is ill-suited to make the credibility determinations that were key to this case. See

86

Salve Regina, 499 U.S. at 233. Yet it beats on anyway, annexing the district court’s role to reweigh the evidence, remake the credibility determinations, and thus refind de novo that there was no disregard by more than mere negligence. Contra Pelphrey, 547 F.3d at 1268.

4.

To wrap up, the majority tries to distinguish several cases holding that the failure to provide social-transitioning treatment arises to deliberate indifference. In footnote 11, it gives three reasons for why these cases do not apply. Each reason falls flat.

First, the majority says that, in those cases, medical providers all agreed that a certain medical treatment for gender dysphoria was medically necessary, while here they don’t. That is both wrong and irrelevant. It is wrong because medical providers in those cases did not all agree that a given type of treatment was necessary. See, e.g., Hicklin v. Precynthe, 2018 WL 806764, at *12 (E.D. Mo. Feb. 9, 2018) (noting that the regional medical director in that case stated that permanent hair removal was not medically necessary). And it is irrelevant because, again, the views of Keohane’s treatment team had no role in its decision- making. Team members blindly deferred to prison policy, which caused them to deny care. As shown in the cases that the majority tries to distinguish, that level of

87

neglect arises to deliberate indifference. See, e.g., Kosilek, 774 F.3d at 91; Soneeya, 851 F. Supp. 2d at 247.

Second, the majority claims that prison officials in those cases denied care under a blanket ban, but here the FDC has rescinded the freeze-frame policy and has conceded that it will grant exceptions if social transitioning is medically necessary. This is a misdirection. The FDC’s freeze-frame policy has nothing to do with its general security policy requiring all inmates to dress as their biological sex. And the district court found that the treatment team here denied treatment under a blanket ban. See supra at 72–79. Given the treatment team’s unwavering deference to FDC policy and the chief medical officer’s staunch refusal to provide social transitioning in any circumstance, the court was right to treat Keohane’s case as a blanket-ban case. See, e.g., Soneeya, 851 F. Supp. 2d at 247.

Third, the majority says that it was clear in those cases that the patient’s health was declining, but here the evidence shows that Keohane’s symptoms improved after she received hormone therapy. That distinction is hollow for two reasons.

One, it can be true that Keohane’s treatment team knew that hormone therapy was helping and also knew that she still suffered from significant distress due to her lack of social transitioning. These facts are not mutually exclusive— although Tylenol dulls the pain of a gunshot wound, the patient still needs stitches.

88

See Ancata, 769 F.2d at 704 (“Although the plaintiff has been provided with aspirin, this may not constitute adequate medical care. If, ‘deliberate indifference caused an easier and less efficacious treatment’ to be provided, the defendants have violated the plaintiff’s Eighth Amendment rights by failing to provide adequate medical care.”); Soneeya, 851 F. Supp. 2d at 246–50 (holding that a blanket ban on laser hair removal and surgery arose to deliberate indifference even though the transgender plaintiff was receiving some treatment, including psychotherapy and hormones).18 So it does not matter if Keohane’s treatment team believed that hormone therapy was helping. That evidence does not detract from the evidence showing that the FDC also knew that Keohane was still suffering from her inability to transition and yet did nothing more to treat her pain.

Two, the district court found that the FDC knew that Keohane was suffering greatly because of its refusal to provide social transitioning. See Keohane, 328 F. Supp. 3d at 1314–15; supra at 83. The court also discredited testimony from treatment team officials claiming that they thought hormone therapy was enough to treat her pain. See supra at 83–84. So the majority is again disregarding the

18 See also De’lonta v. Johnson, 708 F.3d 520, 526 (4th Cir. 2013) (“By analogy, imagine that prison officials prescribe a painkiller to an inmate who has suffered a serious injury from a fall, but that the inmate’s symptoms, despite the medication, persist to the point that he now, by all objective measure, requires evaluation for surgery. Would prison officials then be free to deny him consideration for surgery, immunized from constitutional suit by the fact they were giving him a painkiller? We think not.”).

89

district court’s finding (without engaging in meaningful clear-error review) and placing greater weight on facts it prefers. Contra Pelphrey, 547 F.3d at 1268.

Despite the majority’s framing, this case is very similar to cases across the country holding that the categorical refusal to adequately treat gender dysphoria amounts to deliberate indifference. See, e.g., Hicklin, 2018 WL 806764, at *11 (“Ms. Hicklin has presented compelling evidence that Defendants’ refusal to provide her with hormone therapy after her diagnosis is based on the Policy rather than on a medical judgment concerning Ms. Hicklin’s specific circumstances.”); Soneeya, 851 F. Supp. 2d at 248 (“The DOC cannot, therefore, claim that Ms. Soneeya is receiving adequate treatment for her serious medical needs because it has not performed an individual medical evaluation aimed solely at determining the appropriate treatment for her GID under community standards of care.”).

In fact, a recent Ninth Circuit opinion highlights the ways the majority has gone wrong. See Edmo, 935 F.3d 757. Our sister circuit there affirmed a district court’s holding that the failure to provide gender-reconstruction surgery arose to deliberate indifference, and it did so on a similar posture. There, as here, the state argued that theirs was simply a case of dueling medical opinions. And there, as here, the district court found that this was not so, crediting the prisoner’s experts’ view that surgery was medically necessary, and discrediting the state’s experts because they lacked necessary experience.

90

On appeal, the Ninth Circuit took the district court’s credibility findings as true, recognizing that, absent clear error, “it is not our role to reevaluate them.” Id. at 787. Given the district court’s findings, the appellate court agreed that surgery was medically necessary. Id. at 790. Then—again lending deference to the district court’s findings—the appellate court affirmed that the state was deliberately indifferent to the prisoner’s serious medical need because the state knew that the prisoner had engaged in substantial self-harm due to her gender dysphoria and yet continued to provide ineffective treatment. Id. at 793. Finally, the court rejected the argument that, since the state had provided at least some treatment for gender dysphoria, it was not deliberately indifferent to the prisoner’s medical needs. In doing so, the appellate court recognized that the “provision of some medical treatment, even extensive treatment over a period of years, does not immunize officials from the Eighth Amendment’s requirements.” Id. Because the state did not provide medically necessary care—there, gender-reconstruction surgery—it was liable for deliberate indifference, even if it provided other care.

The Ninth Circuit got it right, and its analysis leads us to the right result here. The district court found that FDC officials deferred to a blanket policy and were incompetent to treat Keohane’s gender dysphoria. It also found that, for both those reasons, they denied her access to social transitioning—a treatment that the district court found as medically necessary to treat Keohane’s gender dysphoria.

91

That amounts to deliberate indifference. See Kosilek, 774 F.3d at 91; McElligott, 182 F.3d at 1255. And none of these findings are clearly erroneous, because they rest on a wealth of evidence, and because they necessarily rely on the district court’s credibility determinations—determinations that we are not equipped to make. See Salve Regina, 499 U.S. at 233. As a result, I would affirm.19

* * *

Our role on appeal is not to reweigh the evidence or recreate factual findings. That is for good reason: We did not attend the hearing; we did not hear the testimony; we did not see the record develop. We must therefore defer to the district court’s findings, accepting them as true unless the record leaves us with a firm and definite conviction that the court made a mistake. It does not matter if, after reviewing the record, we would have found differently. See Multiponics, 622 F.2d at 723 (“Merely because a reviewing Court on the same evidence may have

19 The majority says that our case is more like Kosilek, and less like Edmo, because the security concerns present in Kosilek were not present in Edmo. See Majority Op. at 37 n.14. Yet that is exactly what makes this case closer to Edmo than Kosilek: The FDC did not consider security concerns here; it blindly deferred to prison policy without considering whether security concerns outweighed Keohane’s specific medical need. As the Ninth Circuit recognized, the government in Kosilek considered “significant security concerns that would arise if the prisoner underwent [gender-reconstruction surgery].” Edmo, 935 F.3d at 794. Here, as in Edmo, the FDC did no such thing.

92

reached a different result will not justify setting a finding aside.”). All that matters is that the record supports the district court’s findings of fact. See id.

The majority ignores this standard of review today, turning our court into a district court in the process. It does not explain how the district court’s findings on deliberate indifference lack support; it simply believes that its own findings are better. That is not the law. Because this record supports the district court’s findings, we should affirm them, not shake the magic 8-ball until it gives us a different result. And on those findings, the FDC is liable for deliberate indifference to Keohane’s gender dysphoria. I dissent.