Informed Consent Action Network v. Centers for Disease Control and Prevention

Opinion

UNITED STATES DISTRICT COURT FOR THE DISTRICT OF COLUMBIA

__________________________________________

)

INFORMED CONSENT ACTION ) NETWORK, )

)

Plaintiff, )

v. ) No. 1:23-cv-747 (GMH)

)

CENTERS FOR DISEASE CONTROL AND ) PREVENTION, et al., )

)

Defendants. )

__________________________________________)

MEMORANDUM OPINION AND ORDER Plaintiff Informed Consent Action Network has sued the Centers for Disease Control and Prevention (the “CDC” or the “Agency”) and the Department of Health and Human Services (to- gether, “Defendants”) under the Freedom of Information Act (“FOIA”), 5 U.S.C. § 552. The dis- pute centers on Defendants’ withholding of records pursuant to FOIA’s Exemption 5, which pro- tects against disclosure of “inter-agency or intra-agency memorandums or letters that would not be available by law to a party other than an agency in litigation with the agency,” id. § 552(b)(5), an exception that has been interpreted to include documents shielded by the so-called deliberative process privilege, which Defendants claim here. The parties’ cross-motions for summary judg- ment are currently pending. 1 For the reasons that follow, both motions are denied without preju- dice.

1 The documents most relevant to this Memorandum Opinion and order are: (1) Defendants’ motion for summary judgment and its exhibits, ECF Nos. 31 through 31-4; (2) Plaintiff’s cross-motion for summary judgment and opposition to Defendants’ motion for summary judgment and its exhibits, ECF Nos. 32 through 32-2; (3) Defendants’ reply in further support of their motion for summary judgment and opposition to Plaintiff’s motion for summary judgment and its exhibit, ECF Nos. 35 through 35-1; and (4) Plaintiff’s reply in further support of its cross-motion for summary judgment, ECF No. 37. After the motions were fully briefed, the parties consented to magistrate judge jurisdiction for all purposes and the case was randomly assigned to the undersigned. See ECF No. 38; Minute Order (Apr. 22, 2025); Minute Entry (Apr. 22, 2025). The page numbers cited herein are those assigned by the Court’s CM/ECF system.

I. BACKGROUND

On June 30, 2022, Plaintiff submitted a FOIA request to the CDC seeking

[a]ll records related to the Proportional Reporting Ratio (PRR) analyses performed “to identify [adverse events, or ‘AEs,’] that are disproportionately reported relative to other [adverse events]” pursuant to Sections 2.0, 2.3, and 2.3.1 of the [Vaccine Adverse Events Reporting System, or ‘VAERS,’] Standard Operating Procedures for COVID-19. This should include, but not be limited to, all communications concerning PRR analyses including communications concerning any decision not to conduct PRR analyses.[2]

ECF No. 1-1 at 2. The request suggested that such records would likely be found in the CDC’s Immunization Safety Office. See id. On July 6, 2022, the CDC National Center for Emerging and Zoonotic Infectious Diseases (known as NCEZID) informed the CDC’s FOIA office that no search had been conducted in response to Plaintiff’s request. 3 See ECF No. 35-1, ¶ 4. On July 29, 2022, Roger Andoh, the CDC’s FOIA Officer, sent the Agency’s final response letter to Plaintiff, which attempted to explain why no search had been conducted for the PRR analyses. Id.; ECF No. 1-1 at 16–17. In it, the Agency relayed that NCEZID had reported that

[t]here are no written communications regarding the use of [Empirical Bayesian data mining, also known as EB,] over PRR for purposes of signal detection.

The VAERS Standard Operating Procedures (SOP) is a planning document for internal use by CDC with collaborating partners. It is a dynamic document this is used, revised, and implemented based on the current science of the COVID-19 pandemic . Please see the “Disclaimer” excerpt from the SOP below.

“Disclaimer: This document is a draft planning document for internal use by the Centers for Disease Control and Prevention, with collaborating contractors. Numerous aspects (including but not limited to specific adverse events to be monitored, timeframes for

2 Defendants explain (and Plaintiff does not dispute) that the VAERS is a “passive reporting system to which medical professionals, pharmaceutical companies and others can report an adverse event occurring after administration of a vaccine, and [Proportional Reporting Ratio] measures how common an adverse event for a particular vaccine is compared to how common the event is in the overall VAERS database.” ECF No. 32-1 at 4, ¶ 5. 3 The Immunization Safety Office was apparently housed within the CDC’s National Center for Emerging and Zoonotic Infectious Diseases. See Organizational Structure, National Center for Emerging and Zoonotic Infectious Diseases , CDC (Jan. 26, 2024), https://www.cdc.gov/about/divisions-offices/ncezid.html [https://perma.cc/CY47- 62YB].

3

report processing, data elements to be reported, and data analysis)

are dynamic and subject to change without notice.”

Therefore, it was determined that the Proportional Reporting Ratio (PRR) analyses would not be performed. Instead, the U.S. Food and Drug Administration (FDA)

performs Empirical Bayesian (EB) data mining with VAERS data. EB data mining is a statistical method of detecting disproportionate reporting and is considered the “gold standard” for disproportionality analysis.

There are no written communications regarding the use of EB over PRR for purposes of signal detection. EB has been used for years for this purpose. It is widely accepted as the choice method for detecting potential safety signals (with passive pharmacovigilance data, at least), and thus was assumed to be the preferred method of detecting safety signals among COVID-19 vaccines. PRR is included in the [VAERS Standard Operating Procedures] as a potential alternative or adjunct method, but EB was always understood to be the superior method.

ECF No. 1-1 at 16–17. That same department later explained that “no search was conducted for any communications concerning PRR analyses including communications concerning any deci- sion to not conduct PRR analyses” because Empirical Bayesian data mining “has been the acknowledged ‘gold standard’ of detecting disproportionality for pharmacovigilance, which pre- dates COVID-19.” ECF No. 31-3, ¶ 9. Accordingly, “[n]o formal decision-making process was involved to use EB data mining for COVID-19 vaccines.” Id. PRR was used not “as the primary means of detecting disproportionality in VAERS,” but as “a quick, crude means of detecting dis- proportionality . . . to corroborate EB data mining findings.” Id.

Plaintiff appealed the CDC’s final response on October 7, 2022, asserting that the Agency had failed to conduct an adequate search for the requested records. ECF No. 1-1 at 2, 4. The appeal characterized the relevant request as contain[ing], at least, three major elements:

(1) All records related to Proportional Reporting Ratio (PRR) analysis performed to identify AEs that are disproportionately reported relative to other AEs[] pursuant to Sections 2.0, 2.3., and 2.3.1 of the VAERS Standard Operating Procedures for COVID-19;

(2) All communications concerning PRR analyses; and

4

(3) Communications concerning any decision to not conduct PRR analyses.

Id. at 4. Among other things, Plaintiff contended that the CDC’s response failed to sufficiently address Plaintiff’s request for “records related to PRR analysis performed to identify ‘AEs that are disproportionately reported relative to other AEs’ . . . for COVID-19” without reference to the VAERS Standard Operating Procedure and failed to address at all the second element, which sought production of “any communications concerning PRR analysis.” Id. at 4–5 (alteration in original).

On March 30, 2023, having received no substantive response to its appeal, Plaintiff filed this action alleging that the CDC had failed to make a determination as to Plaintiff’s appeal within the time limits prescribed in FOIA, which require a decision on an appeal to be made within twenty working days with a maximum extension of ten working days in unusual circumstances, see 5 U.S.C. § 552(a)(6)(A)(i), (a)(6)(B)(i); and had failed to establish that it had engaged in an adequate search. ECF No. 1, ¶¶ 16–17, 20. Thereafter, the CDC conducted a search and located responsive documents. According to a joint status report filed on November 22, 2023, the CDC produced eighteen pages of emails and 51 Excel files on April 3, 2023; as relevant here, Plaintiff objected to the CDC’s redactions under Exemption 5. See ECF No. 18 at 1–2; ECF No. 35-1, ¶ 9. A supplemental electronic search by the CDC returned an additional seven records. See ECF No. 32-1 at 5, ¶ 8. Another search—the final one, apparently—described by Defendants as “manual,” resulted in the release of 341 pages of records and seven Excel spreadsheets, for a total production of 366 pages of records and 58 Excel spreadsheets, some with redactions. See id. at 5, ¶¶ 9–10. Defendants produced a draft Vaughn index on July 29, 2024. See ECF No. 35-1, ¶ 10; ECF No. 31-4. That index contains entries for two email threads comprising Defendants’ original eighteen-

5

page production 4 and 21 email threads plus seven Excel spreadsheets from their final production. See ECF No. 31-4. All entries cite the deliberative process privilege as the basis for withholding; all entries other than the two relating to the original eighteen-page production assert that “[t]here is no reasonably segregable, non-exempt information that has been withheld.” Id.

Defendants filed their motion for summary judgment on November 1, 2024. See ECF No.

31. Relying on a declaration from FOIA Officer Andoh (the “Andoh Declaration”), Defendants argue that they have withheld only material protected by the deliberative process privilege, which shields from disclosure inter- or intra-agency communications that are pre-decisional 5 and delib- erative, such as preliminary findings and drafts, to safeguard the decision-making process and prevent injury to the quality of agency decisions. 6 See ECF No. 31-2 at 5–12. They further assert that they have properly determined that no reasonably segregable non-exempt information has been withheld. See id. at 12–13. Plaintiff filed its cross-motion on December 17, 2024. See ECF No. 32. It makes three overarching arguments: (1) that not all of the withheld material consists of inter- or intra-agency communications because one of the interlocutors on some emails— Bicheng (Tony) Zhang—is an individual listed on the CDC’s website as “non-government”; (2)

4 For a reason that is unclear, the Vaughn index states that production was made on February 16, 2024, see ECF No. 31-4 at 1, although elsewhere Defendants indicate the production was made on April 3, 2023, see ECF No. 18 at 1; ECF No. 31-3 at 4. However, Plaintiff identifies it as “the original 18-Page Production,” a characterization that Defendants do not contradict (or address), ECF No. 32-2 at 24, and, in any case, the precise date of the production is immaterial here. 5 The parties and the caselaw use two variant spellings of this word: “predecisional” and “pre-decisional.” This Court prefers the form “pre-decisional” but does not alter the alternative spelling when it is included in a quotation. 6 With one exception, Defendants’ Vaughn index adds nothing material to the Andoh Declaration. Indeed, the two documents often use materially identical language. Compare, e.g., ECF No. 31-3, ¶ 31 (Andoh Decl.) (asserting as to Email 1, that “[d]isclosing such thought processes and communications would inhibit open and frank discussions between CDC employees on matters related to COVID-19 statistical analyses and internal processes. This would result in a chilling effect on intra- and inter-agency communications on critical matters related to how CDC better assess health risks associated with a vaccine and reduce the ability of agency officials to deliberate in a meaningful manner during an ongoing public health emergency”), with ECF No. 31-4 at 2 (Vaughn Index) (using identical language to assert a basis for redactions in Email 1). The single exception is that only the Vaughn index—not the Andoh Declaration, nor Defendants’ briefs—describes the bases for withholding material from the documents in the first production, which are discussed in Section III.D.10, infra.

6

that the withheld material cannot be pre-decisional because the Agency’s final response letter as- serted that “there was no formal decision-making process involved in deciding whether to use EB data mining over PRR analyses for COVID-19 vaccines”; and (3) that Defendants’ repeated claim that disclosure of the withheld material would lead to confusion among the public is not an appro- priate criterion supporting non-disclosure under Exemption 5. ECF No. 32-2 at 13–14, 15–16. Plaintiff then goes on to address each of the email chains and Excel spreadsheets for which it is challenging Defendants’ Exemption 5 withholdings, including withholdings from the first 18-page production, which it asserts Defendants did not address in their motion for summary judgment. See id. at 16–25 (addressing email chains 1 through 21, five of the seven Excel spreadsheets from the final production, and the documents in the first production). 7 It also challenges Defendants’ assertion that there is no reasonably segregable non-exempt information in the withheld materials. See id. at 25–26.

II. LEGAL STANDARDS FOIA presumes that an informed citizenry is “vital to the functioning of a democratic so-

ciety, needed to check against corruption and to hold the governors accountable to the governed.” NLRB v. Robbins Tire & Rubber Co., 437 U.S. 214, 242 (1978). It was enacted to “pierce the veil of administrative secrecy and to open agency action to the light of public scrutiny,” and it favors “full agency disclosure.” Dep’t of the Air Force v. Rose, 425 U.S. 352, 360–61 (1976) (quoting Rose v. Dep’t of the Air Force, 495 F.2d 261, 263 (2d Cir. 1974)). “After an agency receives a [FOIA] request that ‘reasonably describes’ records being sought, the agency must ‘conduct a search reasonably calculated to uncover all relevant documents.’” 100Reporters LLC v. Dep’t of

7 The parties identify each of the 21 email chains in the final production by the row it occupies on the Vaughn index. Thus, the email chain in the first row is identified as Email 1; in the second row, Email 2; in the third row, Email 3; and so on. For the sake of clarity, the Court will also use that nomenclature while noting that the documents appear to be email chains, rather than individual emails.

7

Just., 248 F. Supp. 3d 115, 131 (D.D.C. 2017) (citation modified) (first quoting 5 U.S.C. § 552(a)(3)(A), then quoting Weisberg v. DOJ, 705 F.2d 1344, 1351 (D.C. Cir. 1983)). The agency must then release all responsive records to the requester unless it demonstrates that one of FOIA’s nine exemptions to disclosure applies. Assassination Archives & Rsch. Ctr. v. CIA, 334 F.3d 55, 57 (D.C. Cir. 2003). “To carry that burden, the agency ‘must provide a relatively detailed justifi- cation, specifically identifying the reasons why a particular exemption is relevant and correlating those claims with the particular part of a withheld document to which they apply.’” 100Reporters, 248 F. Supp. 3d at 135 (quoting Mead Data Cent., Inc. v. Dep’t of the Air Force, 566 F.2d 242, 251 (D.C. Cir. 1977)).

FOIA cases are usually resolved on motions for summary judgment. Brayton v. Off. of the U.S. Trade Rep., 641 F.3d 521, 527 (D.C. Cir. 2011). “If an agency’s affidavit describes the jus- tifications for withholding the information [under a FOIA exemption] with specific detail, demon- strates that the information withheld logically falls within the claimed exemption, and is not con- tradicted by contrary evidence in the record or by evidence of the agency’s bad faith, then summary judgment is warranted on the basis of the affidavit alone.” ACLU v. DOD, 628 F.3d 612, 619 (D.C. Cir. 2011). In short, an agency may be awarded summary judgment if its justification for invoking a FOIA exemption “appears ‘logical’ or ‘plausible.’” Davidson v. Dep’t of State, 206 F. Supp. 3d 178, 189 (D.D.C. 2016) (quoting Wolf v. CIA, 473 F.3d 370, 374–75 (D.C. Cir. 2007)).

To qualify for Exemption 5, “a document must . . . satisfy two conditions.” Dep’t of Inte-

rior v. Klamath Water Users Protective Ass’n, 532 U.S. 1, 8 (2001). First, it must be an inter- or intra-agency document, meaning that “its source must be a [federal] [g]overnment agency.” Id.; see also 5 U.S.C. § 551(1) (defining an “agency” for the purposes of FOIA as an “authority of the Government of the United States,” with certain exceptions). Second, the document “must fall

8

within the ambit of a privilege against discovery under judicial standards that would govern liti- gation against the agency that holds it.” Klamath, 532 U.S. at 8. In other words, it must be one that would be shielded from disclosure in civil litigation pursuant to one or more of the “protections traditionally afforded certain documents pursuant to evidentiary privileges,” such as “the attorney- client privilege, the attorney work-product privilege, or the executive deliberative process privi- lege.” Ctr. for Biological Diversity v. U.S. Army Corps of Eng’rs, 405 F. Supp. 3d 127, 140 (D.D.C. 2019) (quoting Dow Jones & Co. v. DOJ, 917 F.2d 571, 573 (D.C. Cir. 1990)). The deliberative-process privilege protects information that is (1) pre-decisional—that is, “it was ‘pre- pared in order to assist an agency decisionmaker in arriving at his decision,’ rather than to support a decision already made” and (2) deliberative—that is, “it is ‘a part of the agency give-and-take . . . by which the decision is made.” New Orleans Workers’ Ctr. for Racial Justice v. ICE, 373 F. Supp. 3d 16, 50 (D.D.C. 2019) (first quoting Petrol. Info. Corp. v. Dep’t of Interior, 976 F.2d 1428, 1434 (D.C. Cir. 1992); then quoting Abtew v. DHS, 808 F.3d 895, 899 (D.C. Cir. 2015)). In addition, an agency that withholds material pursuant to the deliberative process privilege must demonstrate that “it is reasonably foreseeable that release of those materials would cause harm to an interest protected by that privilege.” Reps. Comm. for Freedom of the Press v. FBI, 3 F.4th 350, 361 (D.C. Cir. 2021). “[T]he agency bears the burden of showing that [a] privilege properly applies.” Bloche v. DOD, 414 F. Supp. 3d 6, 26 (D.D.C. 2019).

III. DISCUSSION

The following discussion first addresses Plaintiff’s three “overarching” arguments: (1) that any communications including Zhang cannot be covered by the deliberative process privilege be- cause he was not a government employee at the time the communications were made; (2) that none “of the redacted conversations between individuals at CDC concerning PRR can[] be deliberative

9

and/or pre-decisional” because in the CDC’s final agency response to Plaintiff’s FOIA request, NCEZID asserted that “there was no formal decision-making process involved in deciding whether to use EB data mining over PRR analyses”; and (3) that Defendants inappropriately rely on claims that release of withheld material would lead to confusion among the public. ECF No. 32-2 at 13– 14, 15–16. It then addresses the specific withholdings being challenged from the final production in the order that they are raised in both the Andoh declaration and Plaintiff’s cross-motion for summary judgment: (1) Emails 1–3, 6–7, 17, and 19; (2) Emails 4 and 20–21; (3) Email 5; (4) Emails 8 and 13; (5) Email 9; (6) Emails 10–12; (7) Emails 14 and 18; (8) Emails 15–16; and (9) the Excel spreadsheets. Thereafter, Plaintiff’s objections to withholdings from the first production, followed by the issue of segregability are addressed.

A. Inter- or Intra-Agency Communications Plaintiff claims that communications including Zhang cannot be inter- or intra-agency communications because he was listed as a “non-government” employee on the CDC’s website. See ECF No. 32-2 at 13; ECF No. 37 at 4–5. At the outset, the Court observes that Plaintiff has provided no evidence to support its factual assertion. As Defendants note in their response to Plaintiff’s statement of undisputed material facts, the link Plaintiff indicates establishes that Zhang was a non-government employee leads to a different individual. See ECF No. 35-1, ¶ 11; see also ECF No. 32-2 at 13 n.5 (citing HHS Employee Details, U.S. Dep’t of Health & Hum. Servs., https://directory.psc.gov/hhsdir/eeKey.asp?Key=75006&Format=Table [https://perma.cc/RXM5- P4W4]). Plaintiff cites the same link in its reply brief. See ECF No. 37 at 5 n.1. Nevertheless, Defendants appear to concede that Zhang was a contractor working for the CDC. See ECF No. 35 at 3 (characterizing Zhang as a “CDC contract employee”); ECF No. 35-1, ¶ 11 (characterizing Zhang as an “employee contractor”). They contend, however, that Zhang “‘function[ed] just as an

10

employee would be expected to,’ in the sense that Mr. Zhang did not ‘represent an interest of [his] own, or the interest of any other entity’”; rather “Zhang’s ‘only obligations [were] to truth and its sense of what good judgment calls for [] and in those respects the consultant functions just as an employee would be expected to do.’” ECF No. 35 at 4 (alterations in original) (first quoting Am. Oversight v. Dep’t of Health & Hum. Servs., 101 F.4th 909, 916 (D.C. Cir. 2024); and then quoting McKinley v. FDIC, 744 F. Supp. 2d 128, 137 (D.D.C. 2010)). As such, Defendants seem to be invoking the “consultant corollary” doctrine to show that communications including Zhang should be deemed intra-agency communications.

The consultant corollary doctrine “can protect certain communications between an agency and an outside consultant.” Lawyers’ Comm. for Civil Rights Under Law v. U.S. Dep’t of Just., No. 18-cv-167, 2020 WL 7319365, at *20 (D.D.C. Oct. 16, 2020) (quoting 100Reporters, 248 F. Supp. 3d at 146), report and recommendation adopted, 2021 WL 1197730 (D.D.C. Mar. 30, 2021). The doctrine

“interpret[s] the phrase ‘intra-agency’ to go beyond the text and include U.S.

agency records authored by non-agency entities if those records were solicited by a U.S. agency in the course of its deliberative process” and the non-agency consultant was not pursuing its own self-interest, but rather “played essentially the same part in an agency’s deliberative process as agency personnel.”

Id. (citation modified) (quoting Am. Oversight v. Dep’t of Health and Hum. Servs., 380 F. Supp. 3d 45, 53 (D.D.C. 2019)). The D.C. Circuit has recently clarified that “the consultant corollary applies only if ‘the outsider is disinterested’—that is, only if the outsider ‘comes to the table with no obligation or stake in the outcome of an agency’s process other than a duty to provide good advice to the agency, just as the agency’s own personnel is expected to.’” Wheatland v. U.S. Dep’t of Health & Hum. Servs., No. 21-cv-3126, 2025 WL 1761697, at *3 (D.D.C. Apr. 24, 2025) (quot- ing Am. Oversight, 101 F.4th at 922).

11

If true, the assertions made here—that Zhang was a “contract employee” who functioned just as an employee would, representing only the interest of the CDC, ECF No. 35 at 3—would meet the relevant standard. The problem is that Defendants fail to support those assertions with evidence. The Andoh declaration provides no details about Zhang’s status as a “contract em- ployee,” calling him a “Data Analyst,” ECF No. 31-3, ¶ 10; a “VAERS Data Analyst,” id., ¶ 16; and a “Statistician,” id., ¶ 26. The explanation provided concerning Zhang comes exclusively from counsel, but “unsworn averments of [] counsel” are insufficient to meet a party’s burden to show that a communication is privileged. EEOC v. George Washington Univ., 342 F.R.D. 161, 170 (D.D.C. 2022) (quoting FTC v. Boehringer Ingelheim Pharm., Inc., 180 F. Supp. 3d 1, 16 (D.D.C. 2016)). For its part, Plaintiff does not address the consultant corollary doctrine at all. On this record, the Court cannot determine whether Zhang functioned “as the [CDC’s] own personnel is expected to.” Am. Oversight, 101 F.4th at 922. Accordingly, as is the practice in FOIA cases, if Defendants mean to continue to withhold communications involving Zhang, they may renew their motion for summary judgment and supplement their Vaughn index and/or declarations to support their invocation of the deliberative process privilege and the consultant corollary doctrine. 8 See Section IV, infra.

B. Assertions of the National Center for Emerging and Zoonotic Infectious Diseases

Plaintiff contends that, because NCEZID asserted that “there was no formal decision-mak-

ing process involved in deciding whether to use EB data mining over PRR analyses for COVID- 19 vaccines . . . , then any of the redacted conversations between individuals at CDC concerning PRR cannot be deliberative and/or pre-decisional in nature.” ECF No. 32-2 at 14 (emphasis

8 If Defendants seek to argue that the consultant corollary doctrine is unnecessary because Zhang was not a third-party consultant but a CDC employee—which is not an argument that has been made explicit here—they may also do so in any renewed motion for summary judgment.

12

omitted). It reasons, “[T]here was already a standard operating procedure in place for how CDC was evaluating this data”; that is, “there was already a decision made by CDC to use EB data mining and not PRR analysis. Consequently, there was never any deliberative process carried out by CDC about whether to distribute the data publicly, to utilize the PRR system as a primary source of information, or to standardize the process.” Id. at 14–15. Put somewhat more simply, that “‘[n]o formal decision-making process was involved’ inherently forecloses the use of an exemp- tion that must show decision-making.” ECF No. 37 at 7 (quoting ECF No. 31-3, ¶ 9).

The statement at issue cannot bear the weight Plaintiff heaps upon it. First, Plaintiff mis-

characterizes what NCEZID said, asserting that the department made clear “there was already a decision made by CDC to use EB data mining and not PRR analysis.” ECF No. 32-2 at 14. Rather, the statements indicate that PRR analysis was used (even if only as a rough check on EB data mining findings). See ECF No. 31-3, ¶ 9. More importantly, Plaintiff’s argument assumes that the only decisions related to PRR that would be responsive to its FOIA request would be decisions about whether to use EB data mining or PRR. That might be the case if its request asked exclu- sively for “communications concerning any decision not to conduct PRR analyses.” 9 ECF No. 1- 1 at 2. But, as Plaintiff explained in its administrative appeal of the CDC’s final response, the request also demanded records related to PRR analysis “performed to identify [adverse events] that are disproportionately reported relative to other [adverse events],” as well as “communications concerning PRR analyses.” Id. at 4. There is no reason to believe that such records could not reflect decision-making on issues other than the superiority of EB data mining to PRR analysis— which is the only issue to which the statements of NCEZID related. Indeed, the Andoh Declaration

9 Or it might not. The Court notes that the statements at issue disclaim only a “formal” decision-making process. But “[a]gency decision-making processes can be more or less formal, and they can take many different forms.” Juul Labs, Inc. v. FDA, 731 F. Supp. 3d 46, 61 (D.D.C. 2024).

13

indicates that there were myriad other decisions taken in connection with PRR analysis. See, e.g., ECF No. 31-3, ¶¶ 26–29 (indicating that decisions were being made regarding what items to in- clude in PRR tables and in what order); id. ¶ 35 (indicating that decisions were being made regard- ing the process by which analytical findings would be generated and compared); id. ¶ 36 (indicat- ing that decisions were being made regarding the organization of tables); id. ¶ 41 (indicating that decisions were being made regarding the process by which analytical findings would be generated and compared); id. ¶¶ 47–49 (same); id. ¶ 60 (indicating that decisions were being made as to how to respond to an inquiry for the press); id. ¶ 66 (indicating that decisions were being made as to analytic approaches to PRR); id. ¶ 71 (indicating that decisions were being made about topics for a forthcoming meeting and issues with the reporting decision). Accordingly, the Court rejects Plaintiff’s argument that, based on the statements of NCEZID, none of the withheld material can be pre-decisional or deliberative. That is not to say, however, that the material is protected as a matter of law by the deliberative process privilege. As discussed below, the Agency still has work to do on that front.

C. Public Confusion The Andoh Declaration repeatedly cites concerns that release of preliminary tables and similar material would create confusion among the public on “sensitive matter[s] that th[e] delib- erative process [engaged in] is designed to avoid,” which in turn “would result in a chilling effect on intra- and inter-agency communications on matters regarding the use of CDC statistical analysis techniques and reduce the ability of agency officials to deliberate in a meaningful manner.” Id., ¶¶ 32, 39, 44, 50, 57, 64, 73. Defendants also cite confusion among the public as to the contents and format of the final PRR tables, which are different than the contents and format of the prelim- inary tables that have been withheld. See id., ¶¶ 26, 29, 32, 36, 39, 45, 51, 58, 81.

14

Plaintiff states, categorically, that “‘[c]onfusion . . . is simply not a ground on which to assert this privilege as FOIA does not permit the government to serve as the gatekeeper of infor- mation it thinks is too confusing or that could be misinterpreted.” ECF No. 32-2 at 16. That is not the law in this jurisdiction. The D.C. Circuit has long recognized that public confusion is a cog- nizable harm against which the deliberative process privilege protects. In Coastal States Gas Corp. v. Department of Energy, the court listed as one of the purposes of the privilege “to protect against confusing the issues and misleading the public by dissemination of documents suggesting reasons and rationales for a course of action which were not in fact the ultimate reasons for the agency’s action.” 617 F.2d 854, 866 (D.C. Cir. 1980); see also, e.g., Reps. Comm. for Freedom of the Press, 3 F.4th at 361 (same). Petroleum Information Corp. v. Department of the Interior rec- ognized the risk of confusion from release of “erroneous or incomplete” information as a rationale for the deliberative process privilege. 976 F.2d 1429, 1436 & n.10 (D.C. Cir. 1992).

That said,

[w]hen an agency asserts a concern about public confusion, courts should proceed with caution. They must consider whether the agency is merely attempting to avoid the public vetting of “information marred by errors,” or information that could cause “embarrassment.” The Court must also bear in mind that Congress adopted the foreseeable harm requirement, in significant part, to address “increasing agency overuse and abuse of Exemption 5 and the deliberative process privilege.”

Juul Labs, Inc. v. FDA, 731 F. Supp. 3d 46, 72 (D.D.C. 2024) (internal citations omitted) (first quoting Petroleum Info. Corp., 976 F.2d at 1436 n.10; and then quoting Reps. Comm. for Freedom of the Press, 3 F.4th at 369). Courts should also consider whether the agency can mitigate any potential confusion with an explanation to FOIA requesters that, for example, the material is pre- liminary or “unofficial” and the agency “disclaims responsibility for any errors or gaps.” Petro- leum Info. Corp., 976 F.2d at 1437; see also, e.g., Colo. Wild Pub. Lands v. U.S. Forest Serv., 691 F. Supp. 3d 149, 167 (D.D.C. 2023) (noting that where “it would be a simple task for the [agency]

15

to clear up any confusion” by explaining errors in draft appraisals “were corrected and removed from the final appraisals that the agency used,” the “harm requirement—and the overall logic of FOIA—forbids an agency from opting to keep the public in the dark”).

Here, Defendants’ confusion rationale is supported largely by boilerplate that fails to “‘ar-

ticulate a specific link between the specified harm—public confusion—and the nature of the with- held documents,’ that is the ‘context and purpose’ of those materials.” Juul Labs, 731 F. Supp. 3d at 73 (citation modified) (first quoting Pub. Emps. for Env’t Resp. v. Dep’t of Homeland Sec., 575 F. Supp. 3d 34, 51 (D.D.C. 2021); and then quoting Reps. Comm. for Freedom of the Press, 3 F.4th at 371). 10 Indeed, the court in Americans for Fair Treatment v. U.S. Postal Service rejected a justification similar to one advanced by Defendants here: that disclosure of drafts of a statement could confuse the public because the drafts differed from the final version of the statement. 663 F. Supp. 3d 39, 60 (D.D.C. 2023). The court reasoned that “if stating simply that a draft version differs from the final version of a policy or statement were enough to show reasonably foreseeable harm, agencies would never need to provide a situation-specific reason for withholding a nonfinal draft.” Id. at 60–61 (citation modified). Defendants do not, for example, explain how “the differ- ences are substantive in nature,” id. at 61, or whether the material in the preliminary drafts is “erroneous or incomplete” and, if so, how those differences or errors would lead to public confu- sion that is not reasonably mitigable, Petroleum Info. Corp., 976 F.2d at 1436; see id. at 1433.

10 Defendants assert that Juul Labs stands for the proposition that “[p]ublic confusion in the public health context carries the required ‘link between the specified harm and specific information contained in the material withheld’ and, most significantly, it satisfies the statutory nexus requirement by connecting the foreseeable harm to the ‘interest protected by [the FOIA] exemption’ at issue.” ECF No. 35 at 8 (quoting Juul Labs, 731 F. Supp. 3d at 71–72). Juul Labs announced no such rule. The court’s decision that the FDA had sufficiently connected the harm from public confusion with the materials withheld was based on the specifics of the agency’s explanation—which was significantly more detailed than Defendants’ here, see Juul Labs, 731 F. Supp. 3d at 71–74—and the fact that release of preliminary, internal recommendations on an issue on which the agency had not yet formulated a final policy “would risk misleading the public about the status of the agency’s ongoing review and about the significance of those internal, staff-level recommendations,” id. at 73, which does not appear to be the case here.

16

They have also not wrestled with the precept that “public confusion applies primarily to premature exposure to discussions occurring before the policies affecting it had actually been settled upon.” Juul Labs, 731 F. Supp. 3d at 73 (citation modified) (quoting Nat’l Ass’n of Minority Veterans v. U.S. Dep’t of Veterans Affs., No. 21-cv-1298, 2024 WL 810032, at *8 (D.D.C. Feb. 27, 2024)).

As above, if Defendants mean to continue to rely on the risk of confusion as a harm at-

tendant on disclosure, they may renew their motion for summary judgment and supplement their Vaughn index and/or declarations to support their argument. 11 See Section IV, infra.

D. Specific Withholdings Plaintiff challenges redactions in ten groups of documents: eight groups of email chains and five spreadsheets from the CDC’s final production and two email chains from the first pro- duction. Several themes recur—for example, Plaintiff argues that Defendants have inappropriately redacted purely factual information and (again) that the justifications for withheld material are mere boilerplate. Plaintiff does not challenge, however, Defendants’ redaction of “internal hyper- links, draft responses of discussions of emails or letters, [or] suggested responses to media/re- questers,” ECF No. 32-2 at 17 n.8; ECF No. 37 at 9–10 (same), and the Court therefore takes it as

11 The sufficiency of Defendants’ risk of confusion rationale would be merely an academic question if their other foreseeable harm justifications were adequate. However, as Plaintiff points out—albeit in its reply brief—Defendants’ assertions that release of the withheld material would inhibit agency decision-making by chilling communication among officials are “perfunctory.” ECF No. 37 at 11; see, e.g., ECF No. 31-3, ¶ 27 (“The release of this information would compromise the deliberative process of the agency requiring open, frank discussions on decision-making among subject-matter experts concerning PRR statistical approaches. The result would be a chilling effect on intraand interagency communications and reduce the ability of agency officials to deliberate in a meaningful manner during a public health emergency.”); id., ¶ 31 (“Disclosing such thought processes and communications would inhibit open and frank discussions between CDC employees on matters related to COVID-19 statistical analyses and internal processes . This would result in a chilling effect on intra- and inter-agency communications on critical matters related to how CDC better assess health risks associated with a vaccine and reduce the ability of agency officials to deliberate in a meaningful manner during an ongoing public health emergency.”). Such a justification’s “fatal flaw . . . is that it is essentially a restatement of ‘the generic rationale for the deliberative process privilege itself.’” NPR v. U.S. Dep’t of Homeland Sec., No. 20-cv-2468, 2022 WL 4534730, at *8 (D.D.C. Sep. 28, 2022) (quoting Reps. Comm. for Freedom of the Press, 3 F.4th at 370).

17

conceded that such material is properly redacted and addresses it only where useful for context. 12 Ultimately, neither side has shown an entitlement to summary judgment, and so both motions are denied without prejudice. 13 1. Emails 1–3, 6–7, 17, and 19 According to the Andoh Declaration, these email chains contain discussions taking place between February 23, 2022, and March 3, 2022, among two statisticians (one of them Zhang); John Su, Immunization Safety Office Deputy Director; and Pedro Moro, Acting Team Lead. ECF No. 31-3, ¶ 25–26; see also id., ¶ 10. The discussions “pertain to the preliminary findings of the PRR statistical analysis and reflect the deliberative process agency subject-matter experts used to finalize the specific items to be included within the tables that were most relevant and beneficial for use by relevant people at the agency,” and include “questions, comments, and proposed changes.” Id., ¶¶ 26, 28. The attachments to the emails are “preliminary versions of a COVID-19 PRR Table,” which “include various iterations of table layout (e.g., column order, field sorting), how certain items were displayed, how often tables were to be [run], and how the multiple tables

12 In light of this admission, Plaintiff chastises Defendants for arguing in their opposition to its cross-motion that such material—that is, internal hyperlinks, draft responses, and the like—were appropriately withheld under Exemption 5. See ECF No. 37 at 9–10. However, Defendants can perhaps be excused for overlooking or misunderstanding the breadth of this concession, as it appears only in a footnote in Plaintiff’s opening brief and only in the section discussing Emails 1–3, 6–7, 17, and 19. See ECF No. 32-2 at 17 n.8. The main text of Plaintiff’s reply makes clear that the concession applies across the board. See ECF No. 37 at 9–10.

On the other hand, the Court feels compelled to note that Defendants’ response to the arguments Plaintiff mounts to these more targeted challenges is lackluster at best, relying on “examples” of assertedly appropriate redactions from only five email chains—Emails 5, 9, 10, 11, and 12. See ECF No. 35 at 6–7. 13 This Court cleaves to the “principle of party presentation,” by which courts “rely on the parties to frame the issues for decision” and assume that “the parties know what is best for them[] and are responsible for advancing the facts and arguments entitling them to relief.” Maalouf v. Islamic Republic of Iran, 923 F.3d 1095, 1109 (D.C. Cir. 2019) (quoting Greenlaw v. United States, 554 U.S. 237, 243–44 (2008)). Here, that means that if, hypothetically, an argument was made about Email 5 but not about Email 9, the Court would address that argument only in connection with Email 5, even if it is an argument that arguably could have been made about Email 9.

18

were organized when compiled.” Id., ¶ 29. The final tables were released in connection with a prior FOIA request and have been disseminated to the public. See id.

Plaintiff maintains that discussions including Zhang are not intra- or inter-agency commu-

nications, an argument addressed above. See Section III.A, supra. It also contends that “[p]ure numerical data” and factual information is not protected by the deliberative process privilege, even if it is included in a draft document. ECF No. 32-2 at 17–18. Plaintiff asserts that “tables and discussions regarding the data itself cannot be said to be predecisional, deliberative, or even inter- pretational due to the factual nature of numerical data.” Id. at 18; see also ECF No. 37 at 8 (“Throughout these emails, the discussion and redactions appear to include purely factual and nu- merical data, which is reported in tables and charts, which generally must be disclosed. . . . There- fore, ultimately, the information largely redacted throughout these emails is not exempt from dis- closure . . . .”).

Plaintiff again errs by making an absolute statement about a more nuanced legal proposi-

tion. To be sure, decisions in this Circuit have stated that pure factual material is generally unpro- tected by the deliberative process privilege. Plaintiff cites Petroleum Information Corp. for that proposition, which asserts that, “[u]nder the deliberative process privilege, factual information generally must be disclosed, but materials embodying officials’ opinions are ordinarily exempt.” 976 F.2d at 1434. But the court proceeded to complicate that categorical statement: “The fact/opinion distinction, however, is not always dispositive; in some instances, ‘the disclosure of even purely factual material may so expose the deliberative process within an agency’ that the material is appropriately held privileged.” Id. (quoting Mead Data Cent., Inc. v. Dep’t of Air Force, 566 F.2d 242, 256 (D.C. Cir. 1977)). Indeed, it “caution[ed] against reflexive fact/opinion characterization as the way to decide the full range of Exemption 5 cases.” Id. at 1435. That

19

caveat is borne out in each of the other cases Plaintiff cites supporting its seemingly unqualified fact/opinion distinction. See ECF No. 32-2 at 17–18 (citing Quarles v. Dep’t of the Navy, 893 F.2d 390 (D.C. Cir. 1990); BuzzFeed Inc. v. Dep’t of Just., 419 F. Supp. 3d 69 (D.D.C. 2019); and Hardy v. Bureau of Alcohol, Tobacco, Firearms & Explosives, 243 F. Supp. 3d 155 (D.D.C. 2017)); Quarles, 893 F.2d at 392 (“Even when requested material is found to be factual, the courts have held it exempt where they were convinced that disclosure ‘would expose an agency’s deci- sionmaking process in such a way as to discourage candid discussion within the agency and thereby undermine the agency’s ability to perform its functions.’” (quoting Dudman Commc’ns Corp. v. Dep’t of the Air Force, 815 F.2d 1565, 1568 (D.C. Cir. 1987))); BuzzFeed, 419 F. Supp. 3d at 75 (“[T]he privilege is not strictly limited to opinions. It is also ‘broad enough to protect those facts intertwined with the opinions or tending to reveal the deliberative process.’” (quoting Brinton v. Dep’t of State, 636 F.2d 600, 606 (D.C. Cir. 1980))); Hardy, 243 F. Supp. 3d at 164 (“[T]he D.C. Circuit has taken a functional approach to application of the deliberative process privilege, instructing that ‘the legitimacy of withholding does not turn on whether the material is purely factual in nature or whether it is already in the public domain, but rather on whether the selection or organization of facts is part of an agency’s deliberative process.’” (quoting Ancient Coin Collectors Guild v. U.S. Dep’t of State, 641 F.3d 504, 513 (D.C. Cir. 2011))).

Inexplicably, Defendants do not engage with those cases at all. They merely parrot the Andoh Declaration’s description of the redacted documents and its assertion that release of the redacted material would inhibit open discussion among Agency officials on critical matters related to assessing health risks associated with a vaccine during a public health emergency. See ECF No. 35 at 3 (citing ECF No. 31-3, ¶¶ 28–29, 31). To be sure, the Court can glean some support for an argument that the redacted material, if disclosed, “would reveal something about the agency’s

20

deliberative process.” Hardy, 243 F. Supp. 3d at 165. It appears that the CDC was deliberating about the specific factual material to include in the tables it was creating. See ECF No. 31-3, ¶ 26. As the D.C. Circuit recognized in Ancient Coin Collectors, “the selection or organization of facts” may be “part of an agency’s deliberative process.” 641 F.3d at 513; but cf. Judicial Watch, Inc. v. U.S. Dep’t of State, 241 F. Supp. 3d 174, 185 (D.D.C. 2017) (asserting that “case law does not support [an] argument that the mere selection of facts to be incorporated in a summary is enough in and of itself to satisfy the requirement that it be deliberative”). More, “iterative analyses are hallmarks of records to which the privilege often applies.” Sierra Club v. U.S. Fish & Wildlife Serv., 523 F. Supp. 3d 24, 37 (D.D.C. 2021). But Defendants have not provided sufficient infor- mation for the Court to determine whether disclosure of the material redacted from these emails chains and attachments would “expose the deliberative process within an agency,” Petroleum Info. Corp., 976 F.2d at 1434 (quoting Mead Data Cent., 566 F.2d at 256); they merely rely on the Andoh Declaration’s ipse dixit. That is insufficient.

Accordingly, as discussed in Section IV, infra, the parties’ cross-motions for summary judgment on these redactions will be denied without prejudice.

2. Emails 4 and 20–21 The Andoh Declaration describes these email chains as discussions taking place between April 8 and April 20, 2022, among “CDC Vaccine Safety subject matter experts concerning Table 5 COVID 19 Weekly.” ECF No. 31-3, ¶ 33. There is no explanation of what, precisely, “Table 5 COVID 19 Weekly” is, but it apparently contains “preliminary COVID 19 PRR results” that “com- par[e] the proportion of a specific adverse event.” Id., ¶ 36. The emails chains also include “de- liberative discussions between CDC staff concerning the creation of SAS[14] Proportional

14 Presumably, “Statistical Analysis System.”

21

Reporting Ratio (PRR) analysis or PRR outputs” and “discussions among CDC subject-matter experts concerning preliminary statistical analysis and the process by which agency subject-matter experts may generate and compare their analytical findings prior to publication.” Id., ¶¶ 34–35. The attachments are two different preliminary copies of Table 5 containing preliminary COVID 19 PRR results that include “various iterations of table layout (e.g., column order, field sorting), how certain items were displayed, how often tables were to be [run], and how the multiple tables were organized when compiled.” Id., ¶ 36. The final tables were released in connection with a prior FOIA request and have been disseminated to the public. See id.

Plaintiff makes similar arguments as to these email chains as it did regarding the email chains discussed above, focusing on Zhang’s participation in the emails and the withholding of numerical data. 15 See ECF No. 32-2 at 18–19; see also ECF No. 37 at 8 (“Throughout these emails, the discussion and redactions appear to include purely factual and numerical data, which is re- ported in tables and charts, which generally must be disclosed. . . . Therefore, ultimately, the in- formation largely redacted throughout these emails is not exempt from disclosure . . . .”). Defend- ant responds the same way. ECF No. 35 at 3. The Court’s decision is the same: Defendants have not provided sufficient information for the Court to determine whether disclosure of the material redacted from these emails chains and attachments would reveal anything about the CDC’s delib- erative process.

15 Plaintiff appears to concede that “internal discussions regarding the release of that data” were appropriately withheld . ECF No. 32-2 at 18–19.

22

3. Email 5

Andoh asserts that this email chain comprises discussions on May 13, 2021, “between CDC Vaccine Safety subject matter experts concerning PRR for myocarditis”; it includes discussions about “preliminary statistical analysis and how CDC subject-matter experts may generate and compare the analytic findings prior to publication,” as well as questions, comments, and proposed analysis. ECF No. 31-3, ¶¶ 40–41.

In its opening brief, Plaintiff homes in on an email from Tom Shimabakuro that apparently states, “This isn’t super urgent and I would be open to however you think is the best way to proceed without taking up too much time and person power.” ECF No. 32-2 at 19. Relying on that single sentence, Plaintiff maintains that “CDC cannot be justified in redacting records that discuss ad- verse events of a mandated vaccine under a large umbrella of ‘predecisional and deliberative’ re- dactions when the lead scientist is stating the material discussed was not ‘super urgent’ and where Defendant has not shown Shimabukuro to be a decision-maker.” Id. at 19–20. The Court rejects this argument. As Defendants point out, Plaintiff does not explain why the fact that something is not urgent means it cannot be pre-decisional and deliberative. See ECF No. 35 at 6. As for Shima- bakuro’s status as a decision-maker, the Andoh Declaration makes clear that he was a vaccine safety subject matter expert and the Director of the Immunization Safety Office. See ECF No. 31-3, ¶¶ 10, 40. Even Plaintiff describes him as the “lead scientist.” ECF No. 32-2 at 19–20. And the sentence Plaintiff highlights itself indicates that Shimabukuro was soliciting input in aid of a decision-making process. Without further development of Plaintiff’s argument, the Court finds those facts sufficient to establish that Shimakaburo was a “decision-maker.” ECF No. 32-2 at 19–20.

23

Perhaps realizing the weakness of its first attempt, Plaintiff changes tack in its reply. There, it asserts that this email chain is discussing “factual data”—specifically, “the data produced within the withheld tables and analysis of that factual data”—and that “discussions regarding public re- lease of numerical data cannot be used as a caveat to withhold large charts of factual data and information from the public view.” ECF No. 37 at 9. Thus, Plaintiff returns to a now familiar refrain—that Defendants cannot withhold purely factual material. And again, the Court finds that Defendants have not, at this juncture, provided sufficient information for the Court to determine whether disclosure of the material redacted from these emails chains and attachments would reveal anything about the CDC’s deliberative process.

4. Emails 8 and 13 These email chains were composed on July 22, 2022, “between CDC Vaccine Safety sub-

ject matter experts concerning discontinuing COVID-19 automatic tables” and deal with “gener- ating PRR outputs on a specific age group.” ECF No. 31-3, ¶¶ 46–47. They “include[] deliberative discussions concerning how CDC subject-matter experts may generate and compare their analytic findings,” as well as “comments and the proposed statistical approach.” Id., ¶ 47.

Plaintiff asserts that “this exemption does not [protect] all discussion on data findings and statistical analysis. Purely numerical data generated by this method of research should not be withheld in its entirety.” ECF No. 32-2 at 20; see also ECF No. 37 at 9. Thus, it appears that— again—Plaintiff is objecting only to the withholding of factual data. Defendants do not specifically address these email chains. 16 And so—again—Defendants have not shown how release of such data would reveal the workings of the CDC’s deliberative process.

16 In its opening brief, Plaintiff complained that the Andoh Declaration’s averments attempting to establish that the redacted material is pre-decisional and deliberative is “circular,” stating “that the material is predecisional because it predates a decision and deliberative because it is predecisional.” ECF No. 32-2 at 20–21. However, in its reply, it

24

5. Email 9

This email chain from May 13 through 18, 2022, concerns, according to the Andoh Decla-

ration, “Questions on Pulmonary Embolism.” ECF No. 31-3, ¶ 52. It contains “communications regarding whether the CDC was monitoring signals for Pulmonary embolism,” which the Andoh Declaration explains “illustrates the confusion that would result should this information be re- leased publicly: PRRs do not prespecify outcomes for which to search (something not understood in this email exchange).” Id., ¶ 54. That declaration asserts that “[t]he withheld material is prede- cisional because it predates any agency final decision or output” and is “deliberative because it consists of pre-decisional, deliberative communications about CDC monitoring policy.” Id., ¶¶ 55–56. The declaration also explains that hyperlinks leading to internal CDC shared drives housing files that provide access to “internal data and preliminary COVID-19 statistical tables” were redacted lest “an outsider” use them to gain access to the draft documents therein. Id., ¶ 53.

Plaintiff argues that Defendants have failed to demonstrate that the redacted material—

other than the hyperlinks, which it does not seek—is pre-decisional or deliberative. See ECF No. 32-2 at 21; see also id. at 17 n.8; ECF No. 37 at 9. “To find that a document is predecisional, the court must be able ‘to pinpoint an agency decision or policy to which the document contributed,’ or was intended to contribute.” Heartland All. for Human Needs & Human Rights v. U.S. Dep’t of Homeland Sec., 291 F.Supp.3d 69, 79 (D.D.C. 2018) (quoting Senate of the Commonwealth of Puerto Rico v. U.S. Dep’t of Just., 823 F.2d 574, 585 (D.C. Cir. 1987)). “To show that a record is deliberative, an agency must demonstrate that the record was ‘generated as part of a definable

appears to back off from that argument in connection with Emails 8 and 13, stating only that its “point regarding the release of purely factual data and statistical analysis also applies to Emails 8, 13, 14, and 18 . . . . [O]nce again, purely numerical data generated by this method of research should not be withheld in its entirety.” ECF No. 37 at 9. Plaintiff ’s argument about circularity is discussed below in connection with Email 9 and Emails 10, 11, and 12. See Sections III.D.5 and III.D.6, infra.

25

decision-making process,’” which typically requires identification of “(1) the nature of the specific deliberative process involved, (2) the function and significance of the document in that process, and (3) the nature of the decisionmaking authority vested in the document’s author and recipient.” Id. at 80–81 (first quoting Gold Anti–Trust Action Comm. v. Bd. of Governors of Fed. Reserve Syst., 762 F.Supp.2d 123, 135–36 (D.D.C. 2011); and then quoting Nat’l Sec. Counselors v. CIA, 960 F.Supp.2d 101, 189 (D.D.C. 2013)). Plaintiff maintains that the description of the content of the emails—“communications regarding whether the CDC was monitoring signals for Pulmonary embolism,” ECF No. 31-3, ¶ 54—“is factual, not decisional.” ECF No. 32-2 at 21. It further charges that redacted portions of “the email from [epidemiologist] Elaine Miller on page 154 of the production” do “not appear to be predecisional or deliberative.” Id. More generally, it main- tains that “Defendants did not explain the nature of the specific deliberative process involved, the function and significance of the withheld information in that process, or the nature of the decision- making authority vested in the information’s author(s) and recipient(s).” Id.

Defendants’ response is cryptic:

With regard to email no. 9, Plaintiff suggests that factual information should not be redacted if it is not predecisional. Furthe[r] Plaintiff speculates that the remainder of page 154 from email no. 9 “does not appear to be predecisional or deliberative.”

Actually, “email 9 (page 154-157) contains emails between May 13 through May 18, 2022, between CDC Vaccine Safety subject matter experts concerning Questions on Pulmonary Embolism.” The Andoh Declaration clearly states that “[t]he withheld material consists of internal hyperlinks that leads subject-matter experts to internal CDC shared drives. These share[d] drives house files that provide access to internal data and preliminary COVID-19 statistical tables. Release of the hyperlinks might allow an outsider to access the CDC draft documents in those drives[,]”

but Plaintiff appears to pretend this justification does not exist and instead states that “Defendant did not explain the nature of the specific deliberative process involved .” Again, the Andoh Declaration is unrebutted in any meaningful way.

ECF No. 35 at 6–7 (internal citations to the record omitted). That paragraph would make some sense if the only redacted material were hyperlinks. Cf. note 12, supra. But the Andoh Declaration

26

makes clear that “[t]he redactions also contain communications regarding whether the CDC was monitoring signals for Pulmonary embolism.” ECF No. 31-3, ¶ 54 (emphasis added). And De- fendants’ responsive argument that it has established that material other than hyperlinks is properly redacted relies merely on Andoh’s description of the contents of Email 9 as “emails between May 13 through May 18, 2022, between CDC Vaccine Safety subject matter experts concerning Ques- tions on Pulmonary Embolism,” ECF No. 35 at 6 (quoting ECF No. 31-3, ¶ 52), a statement that appears to describe the entire email chain rather than only the material that was withheld. Other statements in the Andoh Declaration offer precious little help to Defendants. It asserts that “[t]he withheld material is predecisional because it predates any agency final decision or output,” ECF No. 31-3, ¶ 55, but fails to reveal what relevant “decision or output” is at issue. And, in an expla- nation that Plaintiff elsewhere accurately calls “circular,” see ECF No. 32-2 at 20–21, 22, the dec- laration insists that the withheld material is “deliberative because it consists of pre-decisional, de- liberative communications about CDC monitoring policy,” ECF No. 31-3, ¶ 56. That comes no- where near meeting Defendants’ burden.

6. Emails 10–12

The Andoh Declaration states that “[t]he redact[ed] portions [of these emails from June 22 through June 23, 2022,] contained on pages 158–161, 162–166, 167–172, and 173–179 consist of pre-decisional and deliberative discussions on a response to an inquiry from the [A]ssociated [P]ress . . . regarding the recommended response to a FOIA inquiry submitted by a reporter.” Id., ¶¶ 59–60. It continues:

The Vaccine Safety Team Lead provided feedback and clarification regarding the statement previously provided by a staff member and explained to the Health Communications Specialist the exploratory work that was taking place at the time regarding the statistical analyses and the proportional reporting ratio. The language provided by the vaccine safety subject matter experts went through additional

27

discussions with Communications staff, leading to additional edits and internal review . Communications staff provided final versions of statements to respond to the Reporter’s inquiry.

Id., ¶ 60.

Plaintiff concedes that the proposed responses to the reporter are properly withheld but challenges “the remaining (b)(5) redactions,” arguing (again) that the Andoh Declaration’s expla- nation— that “the withheld material is predecisional because it predates any agency final decision” and that “the withheld material is deliberative because it consists of pre-decisional, deliberative communications” is circular and conclusory. ECF No. 32-2 at 21–22 (quoting ECF No. 31-3, ¶¶ 61–62); ECF No. 37 at 10 (same).

The Court, not having seen the email chains at issue, is flummoxed. It is unclear if the redactions identified in the Andoh Declaration (on “pages 158–161, 162–166, 167–172, and 173– 179,” ECF No; 31-3, ¶ 60) are the entire universe of redactions and Plaintiff simply does not be- lieve that all of them are related to proposed responses to the press inquiry. If that is the case, the Court would reject Plaintiff’s argument given that agency “affidavits and declarations are accorded ‘a presumption of good faith, which cannot be rebutted by purely speculative claims about the . . . discoverability of other [material].’” Ctr. for Biological Diversity, 405 F. Supp. 3d at 137 (quoting SafeCard Servs. v. SEC, 926 F.2d 1197, 1200 (D.C. Cir. 1991)). If, on the other hand, Plaintiff’s point is that there are additional redactions not identified as related to proposed re- sponses to the press inquiry, then its contention that the justification for those redactions is “circu- lar” is well-taken, as explained above. Again, the Court does not have sufficient information to find that either party has established its right to summary judgment as to these email chains.

28

7. Emails 14 and 18

The Andoh Declaration describes these email chains from February 13, 2022, as containing “pre-decisional and deliberative thoughts from [a] CDC subject-matter expert to [a] CDC statisti- cian regarding proposed PRR analyses [of available VAERS data] for a short paper [on the safety of booster data in pregnant women]. This deliberation explored different analytic approaches, including different ways of performing PRR analysis.” ECF No. 31-3, ¶ 65–66. Plaintiff cites “withh[e]ld portions of the email from [Acting Team Lead] Pedro Moro on pages 181 and 259 of the production,” which “seem[] to be discussing factual and concrete data and the interpretation of such data.” ECF No. 32-2 at 22. It argues that “[s]imply because these researchers may place this data in a short paper does not exempt these data and statistical findings” from disclosure. Id. Again, then, Plaintiff is challenging the withholding of purely factual material. See also ECF No. 37 at 9 (“[P]urely numerical data generated by this method of research should not be withheld in its entirety.”). Defendants do not specifically address these email chains and have not shown how release of such data would reveal the workings of the CDC’s deliberative process. It must do so, if the information these email chains contain is to be protected from disclosure.

8. Emails 15–16

According to the Andoh Declaration, these email chains from April 26 through 27, 2022, include “pre-decisional and deliberative discussions between CDC subject-matter experts regard- ing topics for discussion during a forthcoming meeting [of the Composite Working Group], in- cluding technical problems with the reporting system. These discussions included exploratory discussions of ideas, some of which were not pursued.” ECF No. 31-3, ¶¶ 70–71.

Plaintiff makes three arguments as to these email chains. First, it insists that “[t]echnical problems are not pre-decisional.” ECF No. 32-2 at 23. Perhaps not, but “exploratory discussions

29

of ideas” about technical problems, ECF No. 31-3, ¶ 71, might well be pre-decisional and deliber- ative. Second, it asserts that “[t]he email from [statistician] Paige Marquez on page 184 is almost fully redacted and does not appear to include ‘deliberative data’ or be predecisional and delibera- tive.” ECF No. 32-2 at 23. Without further detail, that appears to be unsupported speculation.

Third, and finally, Plaintiff contends that “Defendants did not explain the nature of the specific deliberative process involved, the function and significance of the withheld information in that process, or the nature of the decision-making authority vested in the information’s author(s) and recipient(s).” Id. That critique has some teeth. As noted, although a discussion of technical problems could be pre-decisional and deliberative, there are no details about the decision that was being deliberated—except the vague assertion that it was “regarding the Composite Report Work- ing Group.” ECF No. 31-3, ¶ 72. Nor does the Andoh Declaration provide sufficient information for the Court to determine “the function and significance of the withheld information” in the de- liberative process or the decision-making authority of the interlocutors. Again, Defendants fail to address Plaintiff’s objections to these redactions in their response. 17 The Court lacks sufficient information to determine whether either side is entitled to sum-

mary judgment as to these email chains.

9. Excel spreadsheets Defendants withheld five spreadsheets in their entirety. See ECF No. 32-2 at 24. The Andoh Declaration asserts that the withheld tables are “preliminary versions of Table 5 PRR of PTs for COVID19 MRNA compared to 2009–2022, and early versions of the Table X PRR COVID-19 Moderna compared to Pfizer and NON-COVID19.” ECF No. 31-3, ¶ 79. They

17 In its reply, Plaintiff groups these email chains with others in which “the discussion and redactions appear to include purely factual and numerical data, which is reported in tables and charts, which generally must be disclosed.” ECF No. 37 at 8. The Court does not read Plaintiff’s argument in its opening brief to contend that the redactions to these email chains are purely factual or numerical data.

30

“include various iterations of table layout (e.g., column order, field sorting), how certain items were displayed, how often tables were to be [run], and how the multiple tables were organized when compiled.” Id., ¶ 80. The final versions were released through a prior FOIA request and have been disseminated to the public. Id. Plaintiff maintains that “data and factual information are not exempt from FOIA, even if they are in draft form” and that the fact that “these tables include various iterations of table layout, different displays of items, etc. do not render them ex- empt.” ECF No. 32-2 at 24–25.

As the Court ruled in Section III.D.1 in connection with Emails 1–3, 6–7, 17, and 19, the CDC seems to have been deliberating on the specific factual material to include in the tables it was creating. See Ancient Coin Collectors, 641 F.3d at 513 (stating that “the selection or organization of facts” may be “part of an agency’s deliberative process”); but cf. Judicial Watch, 241 F. Supp. 3d at 185 (asserting that “case law does not support [an] argument that the mere selection of facts to be incorporated in a summary is enough in and of itself to satisfy the requirement that it be deliberative”). And “iterative analyses are hallmarks of records to which the privilege often ap- plies.” Sierra Club, 523 F. Supp. 3d at 37. However, Defendants have not provided sufficient information for the Court to determine whether disclosure of the material redacted from these email chains and attachments would “expose the deliberative process within an agency,” Petro- leum Info. Corp., 976 F.2d at 1434 (quoting Mead Data Cent., 566 F.2d at 256), and Plaintiff has not shown that the material is unprotected as a matter of law. Therefore, the Court will not grant summary judgment to either party.

31

10. First Production The Andoh Declaration does not address the redactions from documents in the first pro-

duction. The Vaughn index describes the redacted material from pages 1–5 as deliberative “[e]mail discussions from February 23 to March 3, 2022, between CDC Vaccine Safety subject-matter ex- perts concerning [preliminary] Proportional Reporting Ratio (PRR) [analysis or] output.” ECF No. 31-4 at 1. It describes the redacted material from pages 13–18 as deliberative “[e]mail discus- sions on March 1, 2022, between CDC Vaccine Safety subject-matter experts concerning [prelim- inary data collected on] 5–11[] year olds.” Id.

Plaintiff points to two examples of redactions it considers inappropriate. The first it de-

scribes as “a March 1, 2022, email from A. Hause that asks, ‘Who’s taking a look at flu reports for this age group?’” 18 ECF No. 32-2 at 24. The same email appears a second time in the production, but the words “this age group” are redacted. Id. Plaintiff believes the other redaction covers “the number of reports of anaphylaxis,” which is “factual information [that] must be disclosed.” Id. Plaintiff argues that these two unjustified redactions “cast[] doubt on all other redactions within the productions.” Id.

The fact that Plaintiff’s argument is overblown does not distract from Defendants’ failure to address these redactions at all and the obvious insufficiency of the information in the Vaughn index, which fails “‘to pinpoint an agency decision or policy to which the document contributed,’ or was intended to contribute,” Heartland All., 291 F. Supp. 3d at 79 (quoting Senate of the Com- monwealth of Puerto Rico, 823 F.2d at 585), and fails to identify “(1) the nature of the specific deliberative process involved, (2) the function and significance of the document in that process, and (3) the nature of the decisionmaking authority vested in the document’s author and recipient.”

18 It is not clear from the record who A. Hause is.

32

Id. at 80–81 (quoting Nat’l Sec. Counselors, 960 F.Supp.2d at 189). It further fails to provide anything but boilerplate about the “chilling effect on intra- and inter-agency communications” and public confusion that would likely be caused by release of the redacted material. See ECF No. 31- 4 at 1–2.

If Defendants intend to persist in withholding the redacted information from these docu-

ments, they must do better.

E. Segregability Determining the applicability of FOIA exemptions invoked by an agency does not end the inquiry. Indeed, “the Court errs if it ‘simply approve[s] the withholding of an entire document without entering a finding on segregability or lack thereof.’” Am. Immigration Lawyers Ass’n v. U.S. Dep’t of Homeland Sec., 852 F. Supp. 2d 66, 80 (D.D.C. 2012) (alteration in original) (quoting Powell v. U.S. Bureau of Prisons, 927 F.2d 1239, 1242 n.4 (D.C. Cir. 1991)). “It has long been the rule in this Circuit that non-exempt portions of a document must be disclosed unless they are inextricably intertwined with exempt portions.” Id. (quoting Mead Data Cent., 566 F.2d at 260). While agencies are not required to “provide such a detailed justification that would itself compro- mise the secret nature of potentially exempt information” they must “provide the reasons behind their conclusions” and “describe what proportion of the information in [the] document[s] is non- exempt and how that material is dispersed through the document[s].” Mead Data Cent., 566 F.2d at 261. Defendants have failed to do so. The Andoh Declaration asserts that “[f]or each document that was withheld or redacted, a line-by-line review was completed to determine whether any por- tions of the documents could be released. CDC has released all non-exempt material, unless the material is inextricably intertwined with exempt material” and that “[f]or each document that was withheld in its entirety, it was determined that there was no reasonably segregable, non-exempt

33

information in this document, because any non-exempt information would leave only meaningless words and phrases.” ECF No. 31-3, ¶¶ 82–83. Courts have deemed similar explanations “no more than boilerplate, conclusory language” describing the segregability determination, noting that they fail to “complete [even] the relatively simple task of ‘describ[ing] what proportion of the infor- mation in [the] document[s] is non-exempt and how that material is dispersed through the docu- ments.’” 19 Kolbusz v. FBI, No. 17-cv-319, 2021 WL 1845352, at *27 (D.D.C. Feb. 27, 2021) (quoting Mead Data Cent., 566 F.2d at 261), report and recommendation adopted, 2023 WL 2072481 (D.D.C. Feb. 17, 2023).

IV. REMEDY

Generally, where, as here, “it is likely that [an agency’s] withholdings can be sustained, at least in . . . part, with further explication,” the agency is “[p]ermitt[ed] a second bite at the apple.” Shteynlyuger v. Ctrs. For Medicare & Medicaid Servs., 698 F. Supp. 3d 82, 134 (D.D.C. 2023). Thus, in situations “where a court has found an agency failed to justify its claimed FOIA exemp- tions[,] courts have denied the defendant’s and plaintiff’s motions for summary judgment without prejudice and given the agency the choice to ‘disclose those documents or file supplemental sub- missions indicating in sufficient detail why withholding is proper.’” Kolbusz, 2021 WL 1845352, at *27 (quoting Shurtleff v. EPA, 991 F. Supp. 2d 1, 20 (D.D.C. 2013)) (collecting cases). That is the appropriate outcome here. However, “‘this is not a process that can continue indefinitely’ . . . [;] ‘FOIA does not permit an agency to make a half-hearted effort with the expectation that, if unconvincing, it can simply “do it over”—again and again, until the Court is satisfied.’”

19 There might seem to be some similarity between determining that factual information is so “intertwined with” deliberative material that disclosing it would “tend[] to reveal the deliberative process,” BuzzFeed, 419 F. Supp. 3d at 75 (quoting Brinton, 636 F.2d at 606), and the segregability analysis. However, the question in making the former determination is whether factual material (which would generally be disclosable) is actually pre-decisional and deliberative ; the latter analysis considers only whether material that is admittedly not pre-decisional and deliberative may nonetheless be withheld.

34

Shteynlyuger, 698 F. Supp. 3d at 135 (quoting Mountgordon v. U.S. Coast Guard, 691 F. Supp. 3d 72, 79–80 (D.D.C. 2023)).

V. CONCLUSION

In this case, Defendants, who have the burden to establish that the information they have withheld is protected by the deliberative process privilege and therefore may be withheld under Exemption 5, have failed to engage with relevant caselaw and failed even to attempt to counter many of Plaintiff’s arguments. In such circumstances, the presumption of good faith that is gen- erally afforded agency declarations and affidavits in FOIA cases is not enough to establish that Defendants are entitled to summary judgment. Similarly, Plaintiff’s arguments, which often ig- nore nuance in favor of absolutism, have not shown that disclosure of some or all of the withheld material is warranted. Accordingly, it is hereby ORDERED that Defendants’ motion for summary judgment, ECF No. 31, and Plaintiff’s cross-motion for summary judgment, ECF No. 32, are both DENIED WITHOUT PREJUDICE. It is further ORDERED that the parties shall meet and confer and, on or before October 24, 2025, propose a schedule for further proceedings. Before submitting that proposal, the parties shall con- sider whether, in light of the guidance provided in this Memorandum Opinion and Order, they can

narrow their disputes over the withheld material.

G. Michael

Date: Sept. 24, 2025 Harvey

_____________________________

G. Michael Harvey

United States Magistrate Judge