Brown v. Dist. of Columbia

United States Court of Appeals FOR THE DISTRICT OF COLUMBIA CIRCUIT Argued January 14, 2019 Decided July 5, 2019 No. 17-7152 IVY BROWN, IN HER INDIVIDUAL CAPACITY AND AS REPRESENTATIVE OF THE CERTIFIED CLASS, APPELLANT v. DISTRICT OF COLUMBIA, A MUNICIPAL CORPORATION, APPELLEE Appeal from the United States District Court for the District of Columbia (No. 1:10-cv-02250) Kelly Bagby argued the cause for the appellant. Maame Gyamfi, Iris Y. González, Sasha M. Samberg-Champion and Ryan Downer were with her on brief. David A. Reiser and Jonathan H. Levy were on brief for the amici curiae The Legal Aid Society for the District of Columbia, et al. in support of the appellants. Jonathan L. Marcus was on brief for the amici curiae American Association of People with Disabilities, et al. in support of the plaintiffs-appellants. 2 Sonya L. Lebsack, Assistant Attorney General, Office of the Attorney General for the District of Columbia, argued the cause for the appellee District of Columbia. Karl A. Racine, Attorney General, Loren L. AliKhan, Solicitor General, and Caroline S. Van Zile, Deputy Solicitor General were with her on brief. Stacy Anderson, Assistant Attorney General, entered an appearance. Before: HENDERSON and WILKINS, Circuit Judges, and EDWARDS, Senior Circuit Judge. Opinion for the Court filed by Circuit Judge HENDERSON. Opinion concurring in the judgment filed by Circuit Judge WILKINS. KAREN LECRAFT HENDERSON, Circuit Judge: In Olmstead v. L.C. ex rel. Zimring, the United States Supreme Court held that the unjustified segregation of disabled individuals in institutions is a form of disability discrimination barred by federal law. 527 U.S. 581 (1999). Consequently, the District of Columbia (“District”) violates the Americans with Disabilities Act of 1990 (ADA), Pub. L. No. 101-336, 104 Stat. 327 (codified at 42 U.S.C. §§ 12101 et seq.), and the Rehabilitation Act of 1973, Pub. L. No. 93-112, 87 Stat. 355 (codified at 29 U.S.C. §§ 701 et seq.), if it cares for a mentally or physically disabled individual in a nursing home notwithstanding, with reasonable modifications to its policies and procedures, it could care for that individual in the community. Plaintiffs are a class of physically disabled individuals who have been receiving care in District nursing homes for more than ninety days but wish to transition—and are capable of transitioning—to community-based care. They seek an injunction requiring the District to alter its policies and procedures in order to help them transition to the community. 3 After a nine-day bench trial, the district court entered judgment in favor of the District. We now reverse and remand. I. BACKGROUND The District funds both nursing-facility-based and community-based care for individuals with physical disabilities. In both settings, individuals are provided with assistance in eating, bathing, toileting and dressing, as well as with their mobility, medication management, meal preparation, money management and telephone use. The District does not operate nursing facilities itself; it funds care in nursing facilities certified for Medicaid reimbursement through its Medicaid State Plan. 1 There are nineteen Medicaid-certified nursing facilities in the District, which house a total of approximately 2,770 beds. Plaintiffs are physically disabled individuals in these facilities who have been receiving nursing- facility-based care for more than ninety days but wish to transition—and are capable of transitioning—to community- based care. This litigation began in late 2010, when four disabled individuals filed a class action against the District, alleging that the District’s failure to transition them to community-based care violated Title II of the ADA and section 504 of the Rehabilitation Act. The district court rejected the District’s initial argument that it was entitled to summary judgment because it had in place an effective “Olmstead Plan”—that is, 1 Medicaid is a cooperative federal-state program through which the federal government funds medical care provided by States to, among others, individuals with physical disabilities who meet certain financial requirements. States and the District submit Medicaid plans to the federal government for approval. In turn, the federal government reimburses a portion of the State’s or District’s Medicaid expenses. 4 a “comprehensive, effectively working plan for placing qualified persons with [physical] disabilities in less restrictive settings,” with “a waiting list that move[s] at a reasonable pace not controlled by the [District’s] endeavors to keep its institutions fully populated,” Olmstead, 527 U.S. at 605–06. Day v. District of Columbia, 894 F. Supp. 2d 1, 26–32 (D.D.C. 2012). It was “undisputed” that the District had not adopted a “formal Olmstead Plan,” id. at 7, and the district court rejected the District’s argument “that its existing programs and services for individuals with disabilities me[]t the requirements of an Olmstead Integration Plan,” id., pointing to undisputed figures that showed the District lacked a “measurable commitment” to the transitioning of disabled individuals to the community, id. at 28–29. In May 2012, Plaintiffs moved for class certification. The district court identified certain deficiencies in the proposed class and denied the motion without prejudice. In March 2013, Plaintiffs filed an amended complaint that revised the proposed class definition and alleged multiple deficiencies in the services the District provides to transition disabled individuals from nursing homes to the community. In March 2014, the district court granted Plaintiffs’ motion for class certification. Thorpe v. District of Columbia, 303 F.R.D. 120 (D.D.C. 2014). The certified class consisted of: All persons with physical disabilities who, now or during the pendency of this lawsuit: (1) receive DC Medicaid-funded long-term care services in a nursing facility for 90 or more consecutive days; (2) are eligible for Medicaid- covered home and community-based long-term care services that would enable them to live in the community; and (3) would prefer to live in the community instead of a nursing facility but 5 need the District of Columbia to provide transition assistance to facilitate their access to long-term care services in the community. Order, No. 1:10-cv-2250 (D.D.C. Mar. 29, 2014), ECF 129 at 1. Although the district court found class certification appropriate, it expressed doubt—in light of the lack of “readily affordable housing in the community”—that Plaintiffs would ultimately be able to establish “a causal link between any proven deficiencies in the District’s system of transition assistance and the injury associated with being ‘stuck’ in a nursing facility.” Thorpe, 303 F.R.D. at 137. At the same time, the district court denied the District’s renewed motion to dismiss based on its then-recent implementation of a formal “Olmstead Plan.” Id. at 131–32. The district court acknowledged that “the District has made some progress in the recent past” and that “this progress appears to be continuing.” Id. at 138. Nevertheless, it was “undisputed that many Medicaid residents in nursing homes have expressed a desire to receive services in a less restrictive setting in the community, but have not been able to do so.” Id. Thus, the district court held that the District had “yet to demonstrate that its Olmstead Plan is an ‘effectively working plan for placing qualified persons with . . . disabilities in less restrictive settings, [with] a waiting list that move[s] at a reasonable pace not controlled by the State’s endeavors to keep its institutions fully populated.’” Id. (first and third alterations in original) (quoting Olmstead, 527 U.S. at 606–07). In April 2014, the District petitioned this Court for leave to file an interlocutory appeal of the district court’s class certification. We denied the petition in June 2015. In re District of Columbia, 792 F.3d 96 (D.C. Cir. 2015). We held that, although “[t]he District Court’s decision to certify may or may 6 not have been an error,” “we cannot say that it was a ‘manifest error,’ which is the standard for us in this interlocutory appellate posture under Rule 23(f)” of the Federal Rules of Civil Procedure. Id. at 98. After our decision, the district court ordered discovery and Plaintiffs filed another amended complaint, which contained their proposed injunction. The proposed injunction would require the District to: 1. Develop and implement a working system of transition assistance for Plaintiffs whereby Defendant, at a minimum, (a) informs DC Medicaid-funded nursing facility residents, upon admission and at least every three months thereafter, about community-based long-term care alternatives to nursing facilities; (b) elicits DC Medicaid-funded nursing facility residents’ preferences for community or nursing facility placement upon admission and at least every three months thereafter; (c) begins DC Medicaid-funded nursing facility residents’ discharge planning upon admission and reviews at least every month the progress made on that plan; and (d) provides DC Medicaid-funded nursing facility residents who do not oppose living in the community with assistance accessing all appropriate resources available in the community. 2. Ensure sufficient capacity of community- based long-term care services for Plaintiffs 7 under the EPD, 2 MFP, 3 and PCA programs, 4 and other long-term care service programs, to serve Plaintiffs in the most integrated setting appropriate to their needs, as measured by enrollment in these long-term care programs. 2 The Medicaid Program for the Elderly and Individuals with Physical Disabilities (EPD Waiver) is a program funded by Medicaid and overseen by the District’s Department of Health Care Finance (DHCF), which provides long-term personal-care assistance to the physically disabled in community-based settings for up to sixteen hours per day. In addition to personal-care assistance, it provides individuals with case-management services, as well as a host of other services, including adult day health, homemaker, chore aide, respite, personal emergency-response system, environmental-accessibility adaptations, assisted living, participant-directed service, occupational therapy and physical therapy. 3 The Money Follows the Person (MFP) program is a federally- funded, time-limited grant program established under the Deficit Reduction Act of 2005, 42 U.S.C. § 1305 note, to help individuals transitioning from nursing facilities to the community. The program provides outreach and education, transition coordination, environmental accessibility adaptations up to $10,000, household setup costs up to $5,000, and intensive case management, both during an individual’s transition and for one full year following his discharge from a nursing facility. The District’s MFP Program will be phased out by 2020. 4 The Medicaid State Plan Personal Care Assistance (State Plan PCA) program is another Medicaid-funded, DHCF-overseen program, which provides long-term personal-care assistance to the physically disabled in community-based settings. The State Plan PCA program provides assistance for up to eight hours per day and does not include the ancillary services included in the EPD Waiver program. Depending on the individual’s needs, he may be eligible for placement in the EPD Waiver and State Plan PCA programs simultaneously, resulting in 24/7 care. 8 3. Successfully transition Plaintiffs from nursing facilities to the community with the appropriate long-term care community-based services under the EPD, MFP, and PCA programs, and any other long-term care programs, with the following minimum numbers of transitions in each of the next four years: a. 80 class members in Year 1; b. 120 class members in Year 2; c. 200 class members in Year 3; and d. 200 class members in Year 4. 4. Sustain the transition process and community-based long-term care service infrastructure to demonstrate the District’s ongoing commitment to deinstitutionalization by, at a minimum, publicly reporting on at least a semi-annual basis the total number of DC Medicaid-funded nursing facility residents who do not oppose living in the community; the number of those individuals assisted by Defendant to transition to the community with long-term care services through each of the MFP, EPD, and PCA, and other long-term care programs; and the aggregate dollars Defendant saves (or fails to save) by serving individuals in the community rather than in nursing facilities. Fourth Am. Compl., No. 1:10-cv-2250 (D.D.C. Sept. 10, 2015), ECF 162 at 31–32. The litigation then proceeded to a bench trial. The district court bifurcated the trial into a “liability” phase and a “remedy” phase. Order, No. 1:10-cv-2250 (D.D.C. May 9, 2016), ECF 9 178 at 2. It held the “liability” phase trial over nine days between September 2016 and November 2016, and, at the conclusion of that phase, ordered the parties to submit proposed findings of fact and conclusions of law. Brown v. District of Columbia, 322 F.R.D. 51, 61–62 (D.D.C. 2017). In September 2017, the district court concluded that Plaintiffs had failed to establish the District’s liability under both the ADA and the Rehabilitation Act. Brown, 322 F.R.D. 51. Thus, without proceeding to the “remedy” phase of the trial, the district court entered judgment in favor of the District. Id. at 96. It issued a lengthy opinion, explaining that “[t]his case presents the difficult legal issue of what a class of plaintiffs proceeding under an Olmstead theory of liability must prove in order to demonstrate their entitlement to relief under Rule 23.” Id. at 86. It concluded that, “under Rule 23,” “plaintiffs must prove that the District maintains a policy or practice (i.e., a concrete systemic deficiency) that has caused the class members to remain in nursing facilities despite their preference to receive long-term care in the community.” Id. at 87. It held that Plaintiffs both (1) “failed to carry their burden of proving the existence of a concrete systemic deficiency in the District’s transition services” and (2) “failed to prove that the class members’ institutionalization is caused by systemic deficiencies in the District’s transition services or that the harm can be redressed by a single injunction.” Id. As a result, it concluded that Plaintiffs “failed to carry their burden to prove that class-wide relief is appropriate under Rule 23(b)(2)” and “dismiss[ed] plaintiffs’ class-wide claims.” Id. at 96. Finding that Plaintiffs sought no individual relief, the district court entered final judgment for the District. Id. Plaintiffs timely appealed. We review the district court’s factual findings for clear error and its legal conclusions de novo. Armstrong v. Geithner, 608 F.3d 854, 857 (D.C. Cir. 2010). 10 II. ANALYSIS A. PLAINTIFFS NEED NOT IDENTIFY “CONCRETE, SYSTEMIC DEFICIENCY” The district court held that the District was entitled to summary judgment primarily because Plaintiffs failed to identify a “concrete, systemic deficiency” in the District’s transition services. See, e.g., Brown, 322 F.R.D. at 53 (“[T]he essential question before the Court is whether plaintiffs have shown concrete systemic deficiencies that harm the class and, if these deficiencies exist, whether they are redressable by a single injunction.”); id. at 87 (“[P]laintiffs must prove that the District maintains a policy or practice (i.e., a concrete systemic deficiency) that has caused the class members to remain in nursing facilities despite their preference to receive long-term care in the community. The Court . . . concludes that plaintiffs have failed to carry their burden of proving the existence of a concrete systemic deficiency in the District’s transition services.”); id. at 96 (“[P]laintiffs have failed to demonstrate the existence of a concrete, systemic failure that entitles them to class-wide relief.”). Nowhere does Olmstead talk about “concrete, systemic deficiencies.”5 5 In fact, the district court seems to have adopted that formulation on its own in a footnote. Thorpe, 303 F.R.D. at 146 n.58 (declaring, without citation, that “[t]o prevail on the merits and obtain the relief they seek, plaintiffs will have to prove concrete systemic deficiencies”). Granted, we quoted the district court’s entire footnote in denying the District’s petition for interlocutory review of the district court’s class certification decision. In re District of Columbia, 792 F.3d at 100 (noting “concrete systemic deficiencies” “could represent the sort of systemic failure that might constitute a policy or practice affecting all members of the class in the manner Wal-Mart requires for certification” (first quoting Thorpe, 303 11 Olmstead interprets the ADA, which provides that “no qualified individual with a disability shall, by reason of such disability, be excluded from participation in or be denied the benefits of the services, programs, or activities of a public entity, or be subjected to discrimination by any such entity.” 42 U.S.C. § 12132; see also 29 U.S.C. § 794(a) (parallel statutory language in Rehabilitation Act). One of the many regulations implementing the statutory mandate provides: “[a] public entity shall administer services, programs, and activities in the most integrated setting appropriate to the needs of qualified individuals with disabilities.” 28 C.F.R. § 35.130(d). At the same time, the regulation contains a caveat: “[a] public entity shall make reasonable modifications . . . necessary to avoid discrimination on the basis of disability, unless the public entity can demonstrate that making the modifications would fundamentally alter the nature of the service, program, or activity.” Id. § 35.130(b)(7)(i) (emphasis added). In Olmstead, the Supreme Court declared that, because “unjustified isolation of persons with disabilities is a form of discrimination,” 527 U.S. at 600, the ADA and its implementing regulations “require placement of persons with mental disabilities in community settings rather than in institutions” under certain circumstances, id. at 587. 6 “Such F.R.D. at 146 n.58)). But we could not have made clearer the “limited reach” of our holding, which was “only that the District Court did not manifestly err” in certifying the class. Id. at 101. We did not delineate what Plaintiffs must establish to prevail on the merits. Nevertheless, the district court on remand stated that our decision “ma[d]e clear that . . . plaintiffs must prove a uniform deprivation (or a concrete systemic deficiency).” Brown, 322 F.R.D. at 88 (emphasis added). This foundational error affected the district court’s entire opinion. 6 Olmstead dealt specifically with the ADA and the mentally disabled but its analysis applies equally to the Rehabilitation Act and 12 action is in order,” the Court said, “when [(1)] the State’s treatment professionals have determined that community placement is appropriate, [(2)] the transfer from institutional care to a less restrictive setting is not opposed by the affected individual, and [(3)] the placement can be reasonably accommodated, taking into account the resources available to the State and the needs of others with mental disabilities.” Id. Although the Court did not expressly declare that the State bears the burden of proving the unreasonableness of a requested accommodation once the individual satisfies the first two requirements, we believe it does for three reasons. First, Olmstead’s third prong originates in the above-quoted regulation, which, again, provides: “[a] public entity shall make reasonable modifications . . . necessary to avoid discrimination on the basis of disability, unless the public entity can demonstrate that making the modifications would fundamentally alter the nature of the service, program, or activity.” 28 C.F.R. § 35.130(b)(7)(i) (emphasis added). Second, interpreting this regulation, the Olmstead Court confirmed that “States [can] resist modifications” requested by segregated disabled individuals only if they “would fundamentally alter the nature of the service, program, or activity.” Olmstead, 527 U.S. at 597 (quoting 28 C.F.R. § 35.130(b)(7) (1998)). Third, interpreting both Olmstead and this regulation, other circuits have put the burden of establishing the unreasonableness of a requested the physically disabled. See Am. Council of the Blind v. Paulson, 525 F.3d 1256, 1260 n.2 (D.C. Cir. 2008) (courts have construed section 504 of the Rehabilitation Act “in pari materia with Title II of the ADA” and thus “cases interpreting either are applicable and interchangeable” (second quoting Randolph v. Rogers, 170 F.3d 850, 858 (8th Cir. 1999))); 42 U.S.C. § 12102(1)(A) (qualifying disability under ADA includes “a physical or mental impairment”); 29 U.S.C. § 705(20)(A)(i) (same under Rehabilitation Act). 13 accommodation on the State. Steimel v. Wernert, 823 F.3d 902, 914–16 (7th Cir. 2016) (if disabled individual desires community-based treatment and medical professional determines that such placement is appropriate, “[i]t is the state’s burden to prove that the proposed changes would fundamentally alter their programs”); Townsend v. Quasim, 328 F.3d 511, 517 (9th Cir. 2003) (“Because [the State] does not allow [the disabled individual] to receive the services for which he is qualified in a community-based, rather than nursing home, setting, [the disabled individual] can prove that the [State] has violated Title II of the ADA, unless [the State] can demonstrate that provision of community-based services to [him] and members of the class would fundamentally alter the nature of the services [it] provides.”); see also Frederick L. v. Dep’t of Public Welfare of Pa., 422 F.3d 151, 156–57 (3d Cir. 2005) (“[The State] is obligated by . . . federal . . . law to integrate eligible patients into local community-based settings. However, the integration mandate ‘is not boundless’ . . . [because it is] qualified by the ‘fundamental alteration’ defense, under which integration may be excused if it would result in a ‘fundamental alteration’ of the state’s mental health system . . . .” (quoting Olmstead, 527 U.S. at 603–04)). A plurality of the Olmstead Court outlined two ways in which a State can establish that the requested accommodations are unreasonable—in other words, two ways it can make out its “fundamental alteration” defense. First, the State can “show that, in the allocation of available resources, immediate relief for the plaintiffs would be inequitable, given the responsibility the State has undertaken for the care and treatment of a large and diverse population of persons with [physical] disabilities.” Olmstead, 527 U.S. at 604. Second, the State can “demonstrate that it ha[s] a comprehensive, effectively working plan for placing qualified persons with [physical] disabilities in less restrictive settings, and a waiting list that move[s] at a 14 reasonable pace not controlled by the State’s endeavors to keep its institutions fully populated,” i.e., an “Olmstead Plan.” Id. at 605–06. Other courts have treated the plurality’s approach as the starting point for analyzing the “fundamental alteration” defense. Steimel, 823 F.3d at 915; Townsend, 328 F.3d at 519 n.3. We adopt the plurality’s approach because in our view it makes good sense. It effectively requires every State that cares for disabled individuals in institutions, notwithstanding those individuals wish to, and could, be treated in the community, to have a “comprehensive, effectively working plan” for transitioning the individuals to the community and a “waiting list [for transition to the community] that move[s] at a reasonable pace,” i.e., an adequate “Olmstead Plan.” Olmstead, 527 U.S. at 605–06. Accordingly, a State that demonstrates it has an adequate “Olmstead Plan” in place meets Olmstead’s integration mandate. A State that cannot demonstrate it has such a plan in place, however, must make every modification to its policies and procedures requested by an institutionalized disabled individual who wishes to, and could, be cared for in the community, unless the modification would be so costly as to require an unreasonable transfer of the State’s limited resources away from other disabled individuals. Id. at 604. The district court’s fundamental error was looking for the existence vel non of a “concrete, systemic deficiency” in the District’s transition services. Having determined that Plaintiffs bore the burden of demonstrating the existence of a concrete, systemic deficiency, the district court considered four potential systemic deficiencies at trial. Brown, 322 F.R.D. at 89–92. At the end of the trial, the district court concluded that Plaintiffs had not proved any of the four and therefore entered judgment 15 against them. Id. at 96.7 The district court’s formulation led it to require Plaintiffs to meet a burden they should not have been made to shoulder. B. NO CLASS CERTIFICATION ISSUE A class can be modified or decertified at any point before final judgment is entered. Fed. R. Civ. P. 23(c)(1)(C). Although the district court did not decertify the class, it held that Plaintiffs failed to prove their injury “can be redressed by a single injunction,” as required by Fed. R. Civ. P. 23(b)(2), Brown, 322 F.R.D. at 87; see also id. at 92–96, and suggested Plaintiffs may not be able to satisfy Rule 23(a)(2), Brown, 322 F.R.D. at 87–89. At least at this stage, however, we accept that this litigation is a proper class action. Plaintiffs who proceed as a class must satisfy the requirements of Federal Rule of Civil Procedure 23. Under Rule 23(a): One or more members of a class may sue or be sued as representative parties on behalf of all members only if: 7 First, the district court found that Plaintiffs did not prove that the District fails to disseminate information regarding community- based long-term care options. Brown, 322 F.R.D. at 90. Second, it found that Plaintiffs did not prove that the District fails to identify individuals in nursing facilities who would prefer to receive long- term care in the community. Id. at 90–91. Third, it found that Plaintiffs did not prove that the District fails to assist individuals who wish to enroll in community-based long-term care services. Id. at 91– 92. Fourth, it found that Plaintiffs did not prove that the District fails to track class members’ individual progress toward transition or overall programmatic success. Id. at 92. 16 (1) the class is so numerous that joinder of all members is impracticable; (2) there are questions of law or fact common to the class; (3) the claims or defenses of the representative parties are typical of the claims or defenses of the class; and (4) the representative parties will fairly and adequately protect the interests of the class. Fed. R. Civ. P. 23(a). Additionally, the proposed class action must fall into one of the categories listed in Rule 23(b). Fed. R. Civ. P. 23(b). Relevant here is Rule 23(b)(2), which applies if “the party opposing the class has acted or refused to act on grounds that apply generally to the class, so that final injunctive relief or corresponding declaratory relief is appropriate respecting the class as a whole.” Id. 1. Rule 23(a)(2) The Supreme Court examined the Rule 23(a)(2) standard, also known as the “commonality” requirement, in Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338 (2011). There, three named plaintiffs sought to represent a class of 1.5 million female Wal- Mart employees who sought to sue Wal-Mart under Title VII for sex discrimination in pay and promotion across the company’s more than 3,000 stores. Id. at 342–43. The district court certified a class of “[a]ll women employed at any Wal- Mart domestic retail store at any time since December 26, 1998, who have been or may be subjected to Wal-Mart’s challenged pay and management track promotions policies and practices.” Id. at 346. The Supreme Court concluded that the class did not satisfy Rule 23(a)(2). Id. at 349–60. Although resolution of each plaintiff’s claim turned on a common 17 question—was her gender the reason she was paid less and/or not promoted?, id. at 343–45—“[w]hat matters to class certification . . . is not the raising of common ‘questions,’” id. at 350 (second alteration in original). What matters to class certification, the Court said, is “the capacity of a classwide proceeding to generate common answers apt to drive the resolution of the litigation.” Id. “That common contention, moreover, must be of such a nature that it is capable of classwide resolution—which means that determination of its truth or falsity will resolve an issue that is central to the validity of each one of the claims in one stroke.” Id. In other words, the Court said, the class must show that its “theory can be proved on a classwide basis.” Id. at 356. The problem with the Wal-Mart class action, then, was that there was no common proof leading to a common answer to the common question at the heart of each plaintiff’s claim. Indeed, local supervisors made all pay and promotion decisions; to prove that the reason for each pay and promotion decision was the same despite the diffuse decision-making structure, the plaintiffs had to show either (1) that each local supervisor used a particular company-wide decision-making procedure that incorporated sex as a consideration or (2) that Wal-Mart had a general company-wide policy of treating female employees worse than male employees. Id. at 352–53. The plaintiffs could not show either. They could not identify a common decision-making procedure that incorporated gender as a consideration. Id. at 353–55. And the only company-wide policy they could point to was that of granting local supervisors discretion to make pay and promotion decisions. Id. at 355. Absent evidence that all or substantially all local supervisors disfavored women, the policy did not amount to one that treats female employees worse than male employees. Id. at 355–56. The plaintiffs contended that certain statistical disparities demonstrated gender-based disparity in promotions. Id. at 356. 18 But the Court rejected this argument, finding that statistics alone were not enough to establish that all local supervisors, in the exercise of their discretion, would disfavor women, at least absent a “specific employment practice” to explain the statistics. Id. at 356–57. Wal-Mart establishes that Rule 23(a)(2) is satisfied if resolution of each plaintiff’s claim turns on a common question (or questions) and if common proof leads to a common answer (or answers) to that question for each plaintiff.8 We interpreted the Wal-Mart commonality requirement in the DL litigation. See DL v. District of Columbia (“DL I”), 713 F.3d 120 (D.C. Cir. 2013); DL v. District of Columbia (“DL II”), 860 F.3d 713 (D.C. Cir. 2017). There, parents of children between the ages of three and five sued the District, alleging that it violated the Individuals with Disabilities Education Act (IDEA), 20 U.S.C. §§ 1400 et seq., which imposes on the District a number of different obligations with respect to students who require special education services. DL II, 860 F.3d at 717. The District’s IDEA obligations include providing “an effective intake and referral process,” offering “adequate and timely education placements to implement individual education plans” and ensuring “a smooth and effective transition from early intervention programs to preschool 8 Although we have noted post-Wal-Mart that a “specific employment practice” could be the kind of common proof that leads to a common answer to a common question, cf. In re District of Columbia, 792 F.3d at 100 (identifying “fail[ure] to offer sufficient discharge planning” and “fail[ure] to inform and provide [nursing facility residents] with meaningful choices of community-based long-term care alternatives” as “common,” “specific deficiencies” that would support commonality (third alteration in original)), plaintiffs need not always identify a “specific employment practice” to satisfy Rule 23(a)(2). That requirement was specific to the Wal- Mart facts and the Title VII claim at issue. 19 programs.” DL I, 713 F.3d at 128. The district court originally certified a class of all three-to-five-year-olds with respect to whom the District failed to discharge any of these obligations. Id. at 124–25. We rejected the class certification in DL I because there was no “common ‘tru[e] or fals[e]’ question [that could] be answered for each of these three different claims of harm that would assist the district court in determining the District’s liability as to each group.” Id. (first and second alterations in original). We remanded the case “so the district court [could] determine whether subclasses would meet the requirements of Rule 23(a) commonality after Wal-Mart.” Id. at 129. On remand, the district court certified four subclasses of three-to-five-year-olds denied a special education: (1) those whom the District failed to identify as disabled; (2) those whom the District failed to evaluate within 120 days of referral; (3) those to whom the District failed to provide an eligibility determination within 120 days of referral; and (4) those denied a smooth transition from an early intervention program to a preschool program. DL II, 860 F.3d at 724. In DL II, we held that three of the four subclasses satisfied the commonality requirement. Id. at 724–25. 9 Subclass one was organized around a common question—did the District fail to identify certain individuals as disabled?—which was subject to a common answer—yes—based on common proof—evidence showing that the District failed to identify between 98 and 515 disabled children per month. Id. at 724. Subclass three was also organized around a common question—did the District fail to evaluate certain individuals within 120 days of their being referred for a disability evaluation?—which was subject to a common answer—yes—based on common proof—evidence 9 Subclass two’s claims were resolved before trial. DL II, 860 F.3d at 724. 20 showing that the District failed to timely evaluate 20 per cent of those referred for a disability evaluation. Id. Likewise, subclass four was organized around a common question—did the District fail to provide certain individuals a smooth and effective transition from early intervention to preschool?— which was subject to a common answer—yes—based on common proof—evidence showing that 30 per cent of toddlers were denied a smooth transition from early intervention to preschool. Id. Thus, the DL litigation followed the holding of Wal-Mart: Rule 23(a)(2) is satisfied if resolution of each plaintiff’s claim turns on a common question (or questions) and if common proof leads to a common answer (or answers) to that question for each plaintiff. In this case, resolution of Plaintiffs’ claims turns on the same series of questions. First, does the District have an adequate “Olmstead Plan” in place? If it does, the District has met its responsibilities. If there is some deficiency in the District’s plan, however, which leads the court to decide that it is not “comprehensive” and “effectively working” or that the District’s waiting list does not “move at a reasonable pace,” the District must make each accommodation Plaintiffs have requested unless it can show that an accommodation would be so costly to implement that it would be unreasonable to require the District to transfer its limited resources from other disabled individuals. Plaintiffs have requested four separate accommodations, reflected in the four provisions of the proposed injunction. Thus, the second, third, fourth and fifth common questions are: “How costly would it be for the District to implement the [[first], [second], [third], or [fourth]] provision of the proposed injunction and would it be unreasonable to require the District to transfer its limited resources from other disabled individuals to pay that cost?” If the answer to all four of these questions is “yes, it would be so costly as to be unreasonable,” the District is not liable. If the 21 answer to at least one of the four questions is “no, it would not be so costly as to be unreasonable,” however, Plaintiffs are entitled to judgment in their favor. There is no commonality problem here because common proof will lead to common answers to each of the five questions on which resolution of Plaintiffs’ claims turns. As to the first question, common proof will establish whether the District’s plan is “comprehensive” and “effectively working” and whether its waiting list for transition to the community “moves at a reasonable pace.” As to the second, third, fourth and fifth questions, common proof will establish both how costly it would be for the District to implement each provision of Plaintiffs’ requested injunction and whether it would be unreasonable to require the District to pay that cost, considering the District’s limited resources and its obligations to other disabled individuals. Consider, for example, the first provision of the proposed injunction. This provision would require the District to provide all class members with information regarding community- based long-term care options, determine whether they prefer to transition to the community and, if they do, plan their transition and assist them in accessing available resources to help them transition. Id. at 31. Common proof will establish, first, how costly it would be to provide all class members with these services and, second, whether it is reasonable to require the District to use its limited resources to pay this cost, considering the District’s obligations to other disabled individuals. The same analysis will apply to the other three provisions of the proposed injunction. Thus, on the current record, there does not appear to be a Rule 23(a)(2) deficiency. 22 2. Rule 23(b)(2) Because this litigation is a Rule 23(b)(2) class action, Plaintiffs must also show that the District “acted or refused to act on grounds that apply generally to the class, so that final injunctive relief or corresponding declaratory relief is appropriate respecting the class as a whole.” Fed. R. Civ. P. 23(b)(2). Indeed, a Rule 23(b)(2) class action is appropriate only “when a single injunction or declaratory judgment would provide relief to each member of the class,” not “when each individual class member would be entitled to a different injunction or declaratory judgment against the defendant.” Wal-Mart, 564 U.S. at 360; see also DL II, 860 F.3d at 726 (“To certify a class under [Rule 23(b)(2)], a single injunction must be able to ‘provide relief to each member of the class.’” (quoting Wal-Mart, 564 U.S. at 360)). Although the injunction must provide relief to each member of the class, the perfect need not be the enemy of the good. If a certain outcome is legally mandated and an injunction provides each member of the class an increased opportunity to achieve that outcome, Rule 23(b)(2) is satisfied. Indeed, in DL II, the District was required to provide each member of subclass three a timely eligibility determination and each member of subclass four a smooth transition from early intervention to preschool; we found Rule 23(b)(2) satisfied even though the injunction required the District to satisfy each of those obligations with respect to 95 per cent, rather than 100 per cent, of each subclass. 860 F.3d at 720, 724, 726. Although we did not make our reasoning explicit, we implied that the injunction aided every class member because it improved his likelihood of achieving the legally mandated outcome. See id. at 724 (“single injunction” requiring “District to meet its statutory deadline 95 percent of the time and to improve its performance by 10 percent in the first year and 5 percent each year thereafter” is sufficient remedy where “20 percent of preschoolers referred 23 for a disability evaluation received an eligibility determination after the statutory deadline, if [a]t all”); id. at 724–25 (“single injunction requiring annual improvement” is sufficient under Rule 23(b)(2) (emphasis added)); id at 726 (“district court’s comprehensive order,” requiring less than 100 per cent compliance with statutory mandate, can “provide relief to each member of the class” (second quoting Wal-Mart, 564 U.S. at 360)). We note, moreover, that the Supreme Court has called “[c]ivil rights cases against parties charged with unlawful, class-based discrimination” like this one, “‘prime examples’ of what (b)(2) is meant to capture.” Wal-Mart, 564 U.S. at 361 (alteration in original) (quoting Amchem Prods., Inc. v. Windsor, 521 U.S. 591, 614 (1997)); see also DL II, 860 F.3d at 726 (“Rule 23(b)(2) exists so that parties and courts, especially in civil rights cases like this, can avoid piecemeal litigation when common claims arise from systemic harms that demand injunctive relief.”). Plaintiffs claim that their transition to the community is legally mandated. Because the proposed injunction would provide, at least in part, each member of the class an increased opportunity to obtain that outcome, Rule 23(b)(2) is satisfied on the current record. C. REMAND INSTRUCTIONS Thus, this litigation boils down to resolution of the third Olmstead question: are the requested accommodations reasonable? If the answer to that question is yes with respect to the entire class, Plaintiffs are entitled to judgment in their favor.10 If the answer to that question is no with respect to the 10 As set forth above, the district court should concentrate on the accommodations that Plaintiffs in fact request—that is, the proposed injunction—when deciding the District’s liability. If liability is established, however, the district court retains its usual discretion to 24 entire class, the District is entitled to judgment in its favor. In addition, if common proof will not lead to a common answer to that question for each member of the class, the class should be modified or decertified for failure to comply with Rule 23(a)(2). And if the requested injunction will not provide relief to each member of the class, the class should be modified or decertified for failure to comply with Rule 23(b)(2). This case turns on whether the District can establish that the plaintiffs’ requested accommodations are in fact unreasonable (either because the District has an adequate “Olmstead Plan” in place, in which case every requested accommodation is categorically unreasonable, or because each individual accommodation is so costly that it would be unreasonable to require the District to transfer its limited resources from other disabled individuals). As discussed supra, pp. 13–14, the District can meet its burden to establish the requested accommodations are unreasonable in one of two ways. The District can establish that it has a “comprehensive, effectively working plan” for transitioning the individuals to the community and a “waiting list [for transition to the community] that move[s] at a reasonable pace,” i.e., an enter the appropriate declaratory or injunctive relief. See Olmstead, 527 U.S. at 590 & n.4 (“Remedies both at law and in equity are available for violations of the statute.”); see also Disabled in Action v. Board of Elections in City of New York, 752 F.3d 189, 198 (2d Cir. 2014) (“If local authorities ‘fail in their affirmative obligations’ under federal law, ‘the scope of a district court’s equitable powers . . . is broad, for breadth and flexibility are inherent in equitable remedies.” (quoting Swann v. Charlotte-Mecklenburg Bd. of Educ., 402 U.S. 1, 15 (1971))). In other words, the district court is not ultimately bound to enter the proposed injunction as a remedy; if it wishes to, it may focus its ultimate injunction on the six “key components of an effective system of transition assistance” that it has gleaned from other Olmstead cases. See Thorpe, 303 F.R.D. at 148. 25 adequate “Olmstead Plan.” Olmstead, 527 U.S. at 605–06. If it cannot demonstrate the existence of an adequate “Olmstead Plan,” the District can establish, seriatim, that each of the four provisions of Plaintiffs’ requested injunction would be so costly as to require an unreasonable transfer of the District’s limited resources from other disabled individuals. Id. at 604. The district court has not yet concluded, in clear terms and under the correct burden of proof, that the District’s “Olmstead Plan” is adequate. In fact, the district court has consistently held throughout this litigation that the District does not have an adequate “Olmstead Plan” in place. In 2012, the district court concluded that “[a] public entity cannot rely on its Olmstead plan as part of its defense unless it can prove that its plan comprehensively and effectively addresses the needless segregation of the group at issue in the case.” Day, 894 F. Supp. 2d at 27. “Given the fact that, at the time, there were at least 526 physically disabled individuals living in nursing facilities who expressed an interest in living in the community, the undisputed facts demonstrated that the District’s Olmstead Plan had not been effective.” Brown, 322 F.R.D. at 58 (citing Day, 894 F. Supp. 2d at 29). In 2014, the district court held that “the District ha[d] yet to demonstrate that its Olmstead Plan [wa]s an ‘effectively working plan for placing qualified persons with . . . disabilities in less restrictive settings, [with] a waiting list that move[d] at a reasonable pace not controlled by the State’s endeavors to keep its institutions fully populated.’” Thorpe, 303 F.R.D. at 138 (third alteration in original) (quoting Olmstead, 527 U.S. at 605–06). And in the order sub judice, the district court did not find that the District’s plan is “effectively working” or that its waiting list “moves at a reasonable pace” within Olmstead’s language. 11 In fact, it 11 The district court rejected the testimony of Plaintiffs’ expert witness, who concluded that the District “does not have an 26 stated that “[t]he District has little to be proud of regarding its historic inability to comply with Olmstead’s integration mandate.” Brown, 322 F.R.D. at 96. Thus, the District has not yet demonstrated that it has an adequate “Olmstead Plan” in place and so has not shown that all requested accommodations are categorically unreasonable. The district court is, of course, free to find on remand that the District now has an adequate “Olmstead Plan” in place.12 If the District still lacks an adequate “Olmstead Plan,” its burden is to demonstrate that each accommodation requested by Plaintiffs would be so costly as to require an unreasonable transfer of the District’s limited resources from other disabled individuals. Because the district court did not identify this showing as the District’s burden, the District did not attempt to meet it. The District will have the opportunity to do so on remand.13 effectively working system of transition assistance,” Declaration of Roger Auerbach, No. 1:10-cv-2250 (D.D.C. Sept. 20, 2016), ECF 212 at 3. Brown, 322 F.R.D. at 93 (“Plaintiffs’ expert Roger Auerbach grounds his opinion that the District fails to provide effective transition services on the faulty premise that there is affordable, accessible housing in the District that is available to class members.”). But it did not conclude that the District’s plan is “effectively working.” And although the district court found that there is no longer a waiting list for enrollment in the EPD Waiver program, Brown, 322 F.R.D. at 90 n.30, it did not find that the District’s waiting list for transition to the community moves at a “reasonable” pace. 12 The district court should consider the fact that the MFP program will be phased out next year when deciding whether the District has an adequate “Olmstead Plan” in place. 13 The district court made numerous factual findings regarding the lack of housing in the District. Brown, 322 F.R.D. at 93 (“A lack of housing and a lack of income to secure housing are the most 27 If at any point on remand, the district court concludes that the relevant questions will have different answers for different members of the class, it can modify or decertify the class under Rule 23(a)(2). Likewise, if the district court concludes that the single, requested injunction will not provide all members of the class with a better opportunity to transition to the community, it can modify or decertify the class under Rule 23(b)(2). common barriers to discharge from a nursing facility. . . . The state of affordable housing in the District is bleak. More than 80% of individuals in nursing facilities who want to move to the community need some form of public assistance to secure housing. At present, and since April 2013, the D.C. Housing Authority (DCHA) waiting list for public and subsidized housing in the District is closed. Individuals seeking public assistance with housing cannot, at present, be added to the waiting list under any circumstances. The waiting list has over 40,000 names on it, and, for some categories, it will not need to be opened for over 20 years. . . . For class members who did not add themselves to the DCHA waiting list before it closed in April 2013, the MFP voucher lottery [which will cease to be available in 2020] is essentially the only viable avenue for securing subsidized housing. With only 65 MFP set-aside vouchers, there is nowhere near enough capacity to provide housing to all class members.” (record citations omitted)). If on remand the district court reaffirms these factual findings under the correct burden of proof, it appears the third provision of the proposed injunction, supra, p. 8, requiring the District to transfer a certain number of class members to the community each year, would likely be so costly as to be unreasonable. In fact, it is hard to imagine what the District could do to transition to the community the number of individuals specified in the third provision in the face of such a low-income-housing shortage. To be clear, a lack of housing would have no bearing on other portions of the proposed injunction—for example, the fourth provision, which seeks a reporting requirement. Therefore, if the district court again finds a lack of available housing on remand, this finding alone would not resolve the litigation in the District’s favor. 28 We recognize and appreciate the significant time and effort the district court has expended on this case, which presents complicated legal issues. That time and effort has not been wasted. On remand, the district court is free to apply certain facts it has already found to the legal standards articulated in this opinion.14 It need not start over completely. III. RESPONSE TO CONCURRENCE Respectfully, we have some concerns about the concurring opinion. First, in our view, the concurring opinion miscomprehends the nature of an “adequate ‘Olmstead Plan’” and, thus, our opinion. Second, in attempting to distinguish the standard enunciated by the Supreme Court in Olmstead, the concurring opinion relies on distinctions between this case and Olmstead that are inapposite. It is important to emphasize that we view an “adequate ‘Olmstead Plan’” differently from our colleague. An “adequate ‘Olmstead Plan’” is a legal standard. Any plan that is “comprehensive,” “effectively working,” and contains a waiting list that moves at a “reasonable” pace is an “adequate ‘Olmstead Plan.’” See supra, p. 14. And the district court has discretion in applying the “comprehensive,” “effective” and “reasonable” standards. Our opinion therefore affords both the 14 As this opinion makes clear, it is the District—not, as the district court believed, Plaintiffs—that bears the burden of demonstrating that either it has an adequate “Olmstead Plan” in place or the requested accommodations are unreasonable. “[W]hen a finding of fact is based on the application of an incorrect burden of proof, the finding cannot stand.” Abbott v. Perez, 138 S. Ct. 2305, 2327 (2018). Thus, facts that were found based on an improper allocation of the burden of proof should not be reused. Some of the district court’s factual findings were based on party stipulations. Stipulated facts can, of course, be reused. 29 District and the district court far more leeway than the concurrence apparently believes. We are especially troubled by the concurrence’s suggestion that we propose “to measure success of the ADA claims based primarily on the number of completed or pending placements of disabled individuals in outside housing.” Concurring Op. 10. This is not an accurate characterization of the majority opinion. For example, the district court could find, consistent with our opinion, that, in light of the lack of available public housing, the placement of only one individual in a given year could be a “reasonable pace” of movement from the District’s waiting list. If the district court were to deem the District’s plan “comprehensive” and “effectively working,” the District would then have an “adequate ‘Olmstead Plan’” in place. The concurrence need not strain too hard to imagine a case “where a plan producing relatively few successful annual placements . . . can be called ‘effectively working,’” id. at 10– 11, for this might be such a case. In short, the concurrence’s central criticism—that we “unduly cabin the discretion that the District should have in crafting services for individuals with disabilities,” id. at 7, and that we make “speed and success of placements” the “exclusive” determinant of ADA liability, id. at 10—is mistaken. In our view, the approach suggested by the concurring opinion does not offer a viable framework for addressing the issues in this case. The concurrence first suggests that this case and Olmstead are “apples and pears.” Id. at 1. Specifically, it claims that three distinctions between this case and Olmstead make the standard set forth in that case inapplicable here. Id. at 7–9. The concurring opinion says that, in light of “three distinctions relevant to the ADA analysis,” “we must measure [the District’s] services by using a different yardstick from what the Olmstead plurality envisaged for Georgia” in order 30 “[t]o determine whether the District is satisfying its ADA obligations.” Id. at 8–9. The concurrence first points out that “we are dealing with a class action.” Id. at 8. True enough. But “general rules of practice and procedure” prescribed by the Supreme Court—such as Rule 23 of the Federal Rules of Civil Procedure—“shall not abridge, enlarge or modify any substantive right.” Rules Enabling Act, 28 U.S.C. § 2072(a)- (b). Thus, the fact that this is a class action is not “relevant to the ADA analysis.” Concurring Op. 8. Indeed, in discussing the Rules Enabling Act, the concurrence never once contends that Plaintiffs can be asked to meet a different substantive liability standard because they have chosen to proceed as a class. Id. at 17–19. Nor could it. See Tyson Foods, Inc. v. Bouaphakeo, 136 S. Ct. 1036, 1048 (2016) (“[It] would have violated the Rules Enabling Act [to] giv[e] plaintiffs and defendants different rights in a class proceeding than they could have asserted in an individual action.”). The concurrence next says that, because “the Olmstead plaintiffs sought placements in group homes, but our class members hope to reside in their own private homes[,] [w]e are . . . dealing with a whole new level of transition.” Concurring Op. 8. The concurrence provides no basis for treating the distinction between group homes and private homes as meaningful. Olmstead drew the line between “institutions” and “community settings,” 527 U.S. at 587. Olmstead said: “we confront the question whether the proscription of discrimination may require placement of persons with mental disabilities in community settings rather than in institutions.” Id. And the concurrence nowhere disputes that group homes and private homes are both “community settings.” Finally, and “most importantly,” the concurrence says, “the District does not control the housing where the plaintiff class members hope to be placed, as Georgia did in Olmstead.” 31 Concurring Op. 8. Although we agree that this fact is relevant, the Olmstead standard takes it into consideration. The lack of housing is relevant to whether the pace of movement from the waiting list is “reasonable,” which, in turn, is relevant to whether the District has an “adequate ‘Olmstead Plan’” in place. We need not fashion a new legal standard to account for a fact that the existing standard already considers. In our view, this case and Olmstead are both apples. Having eschewed the applicable Olmstead standard, the concurrence endorses the standard articulated by the district court. In particular, the concurrence says that the District must prove by a preponderance of the evidence “that there is no systemic deficiency related to the six characteristics” the district court identified as dispositive. Id. at 17. However, the concurrence does not adequately explain its preference for the district court’s six-factor approach. It concludes that the district court “reasonably derived these six characteristics from at least five settlement agreements in analogous ADA failure-to- accommodate cases.” Id. at 13. But we cannot square the standard the district court derived from its settlement- agreement-survey with the standard enunciated in Olmstead. Even assuming the six-factor approach is a useful starting point, the concurrence does not adequately explain why the District must establish that it lacks a “concrete, systemic deficiency” related to those six factors to avoid liability. Id. at 17. Ultimately, the concurrence makes a valiant effort to make sense of the litigation history of this case, but its approach finds no support in Olmstead. We therefore respectfully disagree with its suggested resolution of this case. One final note. The concurrence’s lengthy causation analysis does not represent the opinion of the court. As the concurrence recognizes, treating individuals in institutions when they wish to and could be treated in the community is 32 discrimination because of disability. Id. at 27 (citing Olmstead, 527 U.S. at 601). Members of the class have thus already proven causation. See supra, pp. 4–5 (class definition). The only remaining question is whether the requested accommodations are reasonable. See supra, Section II.C (remand instructions). *** For the foregoing reasons, we reverse the district court’s judgment and remand the case for further proceedings consistent with this opinion. So ordered. WILKINS, Circuit Judge, concurring in the judgment: I agree with the two bottom-line holdings of the majority opinion: (1) that the District Court erred in “requir[ing] Plaintiffs to meet a burden they should not have been made to shoulder”; and (2) that the class definition comports with Rule 23 of the Federal Rules of Civil Procedure. Majority Op. 15, 21, 23. But I cannot join the opinion because I disagree with how it analyzes a key precedent – Olmstead v. L.C. ex rel. Zimring, 527 U.S. 581 (1999) – and how it instructs the District Court on remand. The majority considers this class action a simple application of Olmstead. I don’t think it’s quite that simple. I recognize that the instant case and Olmstead both address community-based treatment and assert failure-to- accommodate claims under Title II of the Americans with Disabilities Act of 1990 (ADA), Pub. L. No. 101-336, 104 Stat. 327, 337-53 (codified as amended in scattered sections of 42 U.S.C.), and § 504 of the Rehabilitation Act of 1973 (Rehabilitation Act), Pub. L. No. 93-112, 87 Stat. 355, 394 (codified as amended in 29 U.S.C. § 794). But upon closer inspection, the claims are apples and pears. I believe that failing to heed these differences takes the majority opinion slightly, but materially, off course. I. I start with an overview of the relevant legal authorities: the ADA, Rehabilitation Act, their implementing regulations, and Olmstead. My colleagues and I agree that the substantive standard for the ADA and Rehabilitation Act claims is the same (with one major exception noted below), see Majority Op. 11 n.6, and thus I focus primarily on the former. Title II provides that “no qualified individual with a disability shall, by reason of such disability, be excluded from 2 participation in” government programs. 42 U.S.C. § 12132. Because Title II protects any disabled individual who can meet the “essential eligibility requirements” of those programs with the help of “reasonable modifications to rules, policies, or practices,” the ADA requires the government to provide such modifications. See id. § 12131(2). In a failure-to-accommodate claim under Title II, the plaintiff must first present a specific “reasonable accommodation.” See U.S. Airways, Inc. v. Barnett, 535 U.S. 391, 401-02 (2002); see also Barth v. Gelb, 2 F.3d 1180, 1186- 87 (D.C. Cir. 1993) (Rehabilitation Act). She may show that her accommodation is reasonable “on its face, i.e., ordinarily or in the run of cases,” or “on the particular facts.” U.S. Airways, 535 U.S. at 401, 405; see also Barth, 2 F.3d at 1187. Only after the plaintiff makes her prima facie showing does the burden shift to the defendant to prove that the accommodation would create an undue hardship or, in this case, a fundamental alteration. See U.S. Airways, 535 U.S. at 402. The burden-shifting regime is consistent with the text of the relevant Title II regulation. The regulation provides that “[a] public entity shall make reasonable modifications in policies, practices, or procedures when the modifications are necessary to avoid discrimination on the basis of disability, unless the public entity can demonstrate that making the modifications would fundamentally alter the nature of the service, program, or activity.” 28 C.F.R. § 35.130(b)(7)(i). As a textual matter, § 35.130(b)(7)(i) places the burden on the government to demonstrate fundamental alteration but does not expressly do so with respect to the showing of reasonable modifications and their necessity. I interpret the omission as intentional. The regulation therefore imposes on plaintiffs the burden of initially proffering a reasonable accommodation and 3 its necessity. Accord Frederick L. v. Dep’t of Pub. Welfare, 364 F.3d 487, 492 n.4 (3d Cir. 2004). Enter Olmstead. Two individual plaintiffs resided in Georgia-run hospitals and sought a specific accommodation: successful placement in state-contracted, community-based group homes. The Eleventh Circuit identified as an ADA violation Georgia’s failure to place them in group homes and remanded for further consideration of the fundamental alteration defense. In affirming the Eleventh Circuit’s judgment of ADA liability, Olmstead holds as a matter of law that an individual plaintiff who demonstrates (1) the state’s approval of a group-home placement and (2) her desire to receive group-home care makes a prima facie showing that successful placement is a facially reasonable and necessary accommodation. See 527 U.S. at 587, 607 (plurality opinion) (focusing on reasonableness of “placement”). My colleagues appear to agree. See Majority Op. 11-13. After affirming the prima facie determination, the Supreme Court vacated the Eleventh Circuit’s judgment only as to the fundamental alteration defense, because the Court thought that the lower court gave erroneous instructions on remand. See Olmstead, 527 U.S. at 603-07 (plurality opinion); id. at 607-08 (Stevens, J., concurring in part and concurring in the judgment). A plurality of the Supreme Court proceeded to outline the contours of Georgia’s affirmative defense. See id. at 603-06 (plurality opinion). My colleagues and I agree that the plurality opinion provides guidance as to how any government defendant may prove a fundamental alteration in practice. But we diverge on how to interpret the opinion. 4 II. According to the majority, Olmstead “effectively requires” the District to implement a so-called “Olmstead Plan” because evidence of the plan would defeat the failure-to-accommodate claim here. Majority Op. 14. The Olmstead Plan, as described in the Supreme Court case, is a “comprehensive, effectively working plan for placing qualified persons with . . . disabilities in less restrictive settings, and a waiting list that move[s] at a reasonable pace not controlled by the [government]’s endeavors to keep its institutions fully populated.” Olmstead, 527 U.S. at 605-06 (plurality opinion). Unlike the majority opinion, I do not understand the Olmstead plurality as dictating a particular type of “plan” that the government must execute to address every potential manifestation of disability discrimination in community-based treatment. The Olmstead plurality instead proffered a different, more abstract legal standard for evaluating fundamental alteration defenses across a wide range of treatment-related failure-to- accommodate claims. In any such case, the defendant must establish some “inequit[y]” that would result from “immediate[ly]” implementing the accommodation asserted by the plaintiff in the prima facie showing. See id. at 604 (plurality opinion) (“Sensibly construed, the fundamental- alteration component of the reasonable-modifications regulation would allow the State to show that, in the allocation of available resources, immediate relief for the plaintiffs would be inequitable, given the responsibility the State has undertaken for the care and treatment of a large and diverse population of persons with . . . disabilities.”). The plurality offered details as to how to apply the standard in practice. The inquiry is necessarily “case-by-case.” Id. at 606 n.16 (plurality opinion) (quoting 28 C.F.R. 5 § 42.511(c) (1998)). Just as the reasonableness of a proposed accommodation is “a contextual and fact-specific inquiry,” Solomon v. Vilsack, 763 F.3d 1, 9 (D.C. Cir. 2014), so too must the fundamental alteration defense depend on a fact-intensive inquiry. Cost to the government is relevant, but it is only one factor. See Olmstead, 527 U.S. at 604 (plurality opinion). But see Majority Op. 14 (focusing solely on whether a proposed accommodation is too “costly”). Other potentially relevant factors include the amount of government resources devoted to disability treatment in general, see Olmstead, 527 U.S. at 606 n.16 (plurality opinion), and potential harm to plaintiffs or other persons with disabilities caused by changes to current government programming, see id. at 604-05 (plurality opinion) (noting that the government should neither “phase out institutions” that would “plac[e] patients in need of close care at risk,” nor “move institutionalized patients into an inappropriate setting, such as a homeless shelter”). The plurality also emphasized that governments must be able, if they choose, to “maintain a range of facilities” and “administer services with an even hand.” Id. at 605 (plurality opinion). Despite the majority opinion’s suggestions to the contrary, see Majority Op. 13-14, 24, 28, the Olmstead Plan described by the plurality is not an intrinsic part of the “fundamental alteration” standard. The plurality was offering it as an “example” of a plan that would work in Georgia. See Olmstead, 527 U.S. at 605-06 (plurality opinion) (“If, for example, the State were to demonstrate that it had a comprehensive, effectively working plan for placing qualified persons with mental disabilities in less restrictive settings, and a waiting list that moved at a reasonable pace not controlled by the State’s endeavors to keep its institutions fully populated, the reasonable-modifications standard would be met.”); see also id. at 593-94 (defining Georgia state officials as “the State”); id. at 606 (plurality opinion) (citing to representations 6 of government counsel as support for Olmstead Plan); Frederick L., 364 F.3d at 498 (noting that the plurality was posing only a “hypothetical” example). Accordingly, the Olmstead Plan hewed closely to Olmstead’s facts. Consistent with the proposed accommodation of successful group-home placement in the state, the proposed Olmstead Plan focused on “placing qualified persons with . . . disabilities in less restrictive settings” and required a “waiting line that moved at a reasonable pace.” Olmstead, 527 U.S. at 606 (plurality opinion). The plurality emphasized the importance of a waitlist because Georgia had significant control over the group homes, all of which were maintained through state contracts. See Brief for Petitioners at 8, Olmstead v. L.C. ex rel. Zimring, 527 U.S. 581 (1999) (No. 98-536), 1999 WL 54623; Brief for Respondents at 4-5, Olmstead v. L.C. ex rel. Zimring, 527 U.S. 581 (1999) (No. 98-536), 1999 WL 144128; see also L.C. ex rel. Zimring v. Olmstead, No. 1:95-cv-1210-MHS, 1997 WL 148674, at *4 (N.D. Ga. Mar. 26, 1997) (“[T]here is no dispute that defendants already have existing programs providing community services to persons such as plaintiffs.”), aff’d in part and remanded in part, 138 F.3d 893 (11th Cir. 1998), aff’d in part and vacated in part sub nom., Olmstead v. L.C. ex rel. Zimring, 527 U.S. 581 (1999). And as the case was presented to the Supreme Court, the justices were aware that Georgia had significant authority to manage and even increase the number of beds available at those facilities. See, e.g., Oral Argument at 5 (“QUESTION: Your position is . . . that it’s up to the State to decide what voluntary facilities it will make available for the . . . [individuals with disabilities], that if the State chooses to have only institutional facilities, it may do that. And if it chooses to have, in addition, community-based facilities, it may have them in addition, but it will be up to the State how many people it will . . . allow to go into those community- 7 based facilities . . . .”), Olmstead v. L.C. ex rel. Zimring, 527 U.S. 581 (1999) (No. 98-536), 1999 WL 252681; id. at 35 (“QUESTION: Can we go back one step to have . . . this basic question clear? In your view, under the statute, is the State required to have any community-based facilities? Suppose the State says, some people we know are going to need institutionalization. We’re going to provide just one room. Is there any obligation under the ADA for the State to do more than have institutional care?”); id. at 44-45 (“QUESTION: Suppose the State said we have 500 spaces in the . . . community-based facility. There are 532 people who qualify. . . . [I]s the State then required to create another community-based facility to take care of the 32 who don’t fit into the space available?”). Because Georgia had adequate capability to ensure a well-functioning system of group homes and a quickly moving waitlist to enter them, the plurality saw “no warrant effectively to order displacement of persons at the top” of the list “by individuals lower down who commenced civil actions” and sought immediate placements. Olmstead, 527 U.S. at 606 (plurality opinion). By setting into stone an “effective[]” requirement of a plan that is identical in all respects to the Olmstead Plan in cases factually dissimilar from Olmstead, see Majority Op. 14, my colleagues unduly cabin the discretion that the District should have in crafting services for individuals with disabilities. We should not equate Georgia’s services with the District’s. This case illustrates the wisdom of providing local governments greater flexibility at the fundamental alteration stage. Compared to Olmstead, this class action presents materially different ADA issues. Our case involves more than 1,000 plaintiffs, or about two-thirds of all individuals currently in District-funded nursing facilities. See Oral Arg. Recording 0:57-1:17. The plaintiffs seek a comprehensive government 8 plan with certain transition-related services: assistance with finding housing or performing activities of daily living needed for independent living outside a nursing facility. The plaintiffs wish to move into their own homes or the homes of friends or family members. The housing targeted by the plaintiffs is controlled by either the D.C. Housing Authority (DCHA) or private citizens, and neither are parties to this lawsuit or under the District’s control. See Brown v. District of Columbia, 322 F.R.D. 51, 63, 72, 83 (D.D.C. 2017) (noting that class members have been placed into private housing or “public/subsidized housing” managed by the independently run DCHA). These facts reveal three distinctions relevant to the ADA analysis. First, we are dealing with a class action. An appropriate remedy, on the one hand, could create enormous costs for the government but, on the other hand, could appropriately bring about broad policy changes benefiting an entire population, not merely one or two litigants. Second, the plaintiffs are at different stages of community transition and deinstitutionalization; the Olmstead plaintiffs sought placements in group homes, but our class members hope to reside in their own private homes. We are thus dealing with a whole new level of transition – let’s call it “Olmstead Phase II.” Third, and most importantly, the District does not control the housing where the plaintiff class members hope to be placed, as Georgia did in Olmstead. These factual differences do not meaningfully change the prima facie analysis. The plaintiff class members here have all established that they are long-term residents of District-funded nursing facilities, that they desire to leave, and that the District has deemed it appropriate for them to enter outside care. See id. at 87 (“In this case, . . . the class includes all physically disabled individuals who have resided in nursing facilities for over 90 days, are eligible for community-based care, prefer to 9 receive long-term care in the community, and need the District’s assistance to transition to the community.”). As in Olmstead, the isolation in nursing facilities seems facially unjustified, which is enough to establish a presumption that the government violated the ADA. See 527 U.S. at 600 (“[U]njustified institutional isolation of persons with disabilities is a form of discrimination . . . .”). Thus, the plaintiffs have demonstrated an entitlement to facially reasonable accommodations. But the factual distinctions matter significantly in the fundamental alteration analysis. At bottom, they reveal the unsuitability of executing an identical Olmstead Plan requirement here. One monumental distinction in this Olmstead Phase II litigation is the reasonableness of placing a burden on the government to create a “waiting list that move[s] at a reasonable pace.” See id. at 606 (plurality opinion). The record appears to indicate that the District cannot increase the number of housing units available to the plaintiff class members, and the plaintiffs do not contend otherwise. Certainly, the District exercises materially less control over the housing being targeted in this case than Georgia did over group-home placement in Olmstead. Moreover, because we are dealing with a broad class of disabled individuals, not just two plaintiffs, the District Court reasonably may address more systemic changes than a one-off group-home placement. To determine whether the District is satisfying its ADA obligations, we must measure its services by using a different yardstick from what the Olmstead plurality envisaged for Georgia. My colleagues disagree, setting the Olmstead Plan as the primary yardstick. See Majority Op. 24; see also id. at 20 (“First, does the District have an adequate ‘Olmstead Plan’ in place? If it does, the District has met its responsibilities.”). And using the Olmstead Plan yardstick, the majority seems to 10 indicate that mere failure to move plaintiffs off a waitlist at a reasonable pace would be dispositive in showing the ineffectiveness of the District’s current plan. Majority Op. 20. But if the District has no control over the availability and suitability of the housing inventory that the plaintiffs seek, why should a reasonably paced waitlist be a suitable dispositive measurement here, as in Olmstead? My colleagues are leading the District Court astray. The majority opinion will inevitably cause the court to measure success of the ADA claims based primarily on the number of completed or pending placements of disabled individuals in outside housing. I don’t disagree that speed and success of placements could be relevant factors to the analysis, but we should not establish – as I understand the majority opinion to be doing – a bright-line rule that makes them the exclusive, or even predominant, factors. The District Court essentially would repeat the legal error it made in the opinion below, but from another direction. See Brown, 322 F.R.D. at 95 (entering judgment for the District simply because it could not envision a “single injunction that would result in the class members being transitioned to community-based long-term care”); see also Majority Op. 23 (rejecting the District Court’s analysis because the remedy need only provide “increased opportunit[ies]” for outside treatment). My colleagues contend that the District Court, in applying their Olmstead Plan standard, need not fixate on speed and success. See Majority Op. 29. For instance, the District Court could find that “the placement of only one individual in a given year” is “a ‘reasonable pace’ of movement from the District’s waiting list.” Id. The assurances of my colleagues fail to mollify me; I find the plurality’s articulation of the Olmstead Plan quite restrictive and difficult to apply liberally in the mine run of cases. I am hard-pressed to imagine a plausible situation 11 where a plan producing relatively few successful annual placements – such as one in a class of more than 1,000 – can be called “effectively working.” See Olmstead, 527 U.S. at 605 (plurality opinion); see also Effective, WEBSTER’S THIRD NEW INTERNATIONAL DICTIONARY 724 (2002) (“productive of results”). I suspect that the District, owing to no fault of its own, will be unable to show a waitlist moving faster than a glacial pace. Instead of promoting a test that needlessly renders the government unable to defend itself, I would prefer applying the flexible standard that the Olmstead plurality in fact created: whether “inequit[ies]” will result from immediate relief. See Olmstead, 527 U.S. at 604 (plurality opinion). More generally, I worry that future courts will ascertain from the majority opinion a general requirement to consider the specific Olmstead Plan analysis in all future failure-to- accommodate claims involving community-based care. But future plaintiffs may seek types of accommodations that do not fit neatly within the Olmstead mold. For instance, one plaintiff class might seek modifications to a government-aid program so that its members would remain in the community care they already receive. The standard Olmstead Plan analysis isn’t a good fit because the pace of successful community placements in the past has little relevance to such a claim seeking to prevent future re-institutionalization. Yet the majority opinion appears to make the verbatim Olmstead Plan the lodestar of all remedial analyses for failure-to-accommodate claims. In some cases, the substantial mismatch between the Olmstead Plan and the facts on the ground will ensure that the District Court’s legal analysis will cause prejudice to the local government. In others, the mismatch will harm the vulnerable population of disabled individuals seeking nondiscriminatory care. The District Court should have a freer hand in determining what constitutes adequate transition services and crafting an 12 injunction that fits within the District’s current programming and resources. Something less drastic than the speedy and guaranteed placement of more than 1,000 individuals into private homes or DCHA-controlled housing must suffice. A plan need not replicate the “Olmstead Plan” to work. III. Substantial aspects of the District Court’s legal analysis satisfy the Olmstead plurality’s fact-intensive legal standard. Although I agree with my colleagues that the District Court committed a legal error with the burden of proof, I am concerned that the remand instructions are misleading. I first commend the District Court for its dauntless (and largely faultless) work during this litigation. More than two years into the case, the plaintiffs had not specified what classwide relief they wanted, and the District Court needed a better grasp on the precise accommodations being sought. Thus, the District Court refused to certify a class until the plaintiffs provided more details. See Thorpe v. District of Columbia, 303 F.R.D. 120, 133-34 (D.D.C. 2014). The litigants filed a third amended complaint, asserting eleven particular deficiencies in the District’s services and proposing detailed injunctive relief addressing those deficiencies. See id. at 135-37 & n.40. Based on the new pleadings, the District Court correctly understood the “gravamen” of the ADA class claims to be that the District “injur[es] each and every class member by virtue of its failure to implement an effective system of transition assistance.” Id. at 146. In certifying the class, the District Court noted that at least some of the eleven deficiencies could be litigated on a classwide basis and, if proven at trial, would detract from what the court considered to be an adequate 13 government plan. See id. at 148-49; Brown, 322 F.R.D. at 90 n.30. According to the District Court, an adequate plan embraces six characteristics: (1) individual assessments upon admission and periodically thereafter for all residents to determine interest in community-based services; (2) provision of accurate information about available community-based services and eligibility requirements for those services; (3) discharge/transition planning that commences upon admission and includes a comprehensive written discharge/transition plan[]; (4) identification of what community- based services are needed and assistance in arranging for those services; (5) assistance in applying for and enrolling in available waivers or transition programs; and (6) identification of barriers to transition and assistance in overcoming those barriers to the extent possible (e.g., if housing is a barrier, providing assistance in applying for supported housing). Brown, 322 F.R.D. at 89 (quoting Thorpe, 303 F.R.D. at 148). The District Court reasonably derived these six characteristics from at least five settlement agreements in analogous ADA failure-to-accommodate cases, see Thorpe, 303 F.R.D. at 148, and the District has never genuinely contested the characteristics. Having set a useful framework for its legal analysis, the District Court at trial started to determine the “concrete[,] systemic deficienc[ies]” related to the six characteristics. Brown, 322 F.R.D. at 88. My colleagues call it a “fundamental 14 error” to seek such deficiencies, see Majority Op. 14, but I cannot agree. At the outset, I note that no party doubts the importance of finding concrete, systemic deficiencies in this litigation. Appellant Ivy Brown here argues only that the class plaintiffs had sufficiently alleged such deficiencies, not that the deficiencies lack a role in the ADA analysis. See Appellant’s Br. 40-45. Raising the issue sua sponte, my colleagues point out that “[n]owhere does Olmstead talk about ‘concrete, systemic deficiencies.’” Id. at 10. But the case does not purport to outline how every treatment-based failure-to-accommodate claim should proceed. Olmstead is but one gloss of the ADA’s failure-to-accommodate claim. As I noted above, Olmstead is not a class action, but rather a simple case involving the claims of two individual plaintiffs. The District Court has focused on concrete, systemic deficiencies in an attempt to harmonize the substantive ADA standard and our class-action precedents. Recall that the plaintiffs proposed concrete deficiencies in their pleadings. Because “reasonable” accommodations can be an elusive term, the District Court asked the plaintiffs to be more specific as to what they wanted. In response, the plaintiffs identified eleven problems, which helped substantially to clarify matters. By saying the government programming had concrete deficiencies and describing them, the plaintiffs necessarily implied that the fixes for those flaws are the accommodations they seek. Of course, these accommodations are not the same as those requested in Olmstead: successful community placements. Thus, the District Court needed to ensure that they were reasonable and necessary. See 28 C.F.R. § 35.130(b)(7)(i). The court properly did that here. The District Court found the deficiencies to relate 15 to the six characteristics of an adequate government plan, which it determined by canvassing analogous settlement agreements. Although the characteristics do not guarantee actual Olmstead Phase II placements, they still appear to provide disabled individuals with meaningful opportunities for future success. And if other local governments are consenting to providing such services, they likely are facially reasonable. Finally, because the plaintiffs established the prima facie elements of the Olmstead claim, see Brown, 322 F.R.D. at 87, they demonstrated an entitlement to accommodations fixing those specific deficiencies. But it was not enough for the District Court to focus on the concrete deficiencies identified by the plaintiffs. To faithfully follow our Rule 23 precedents, the court needed to concentrate on systemic ones. The relevant cases are Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338 (2011), and DL v. District of Columbia, 713 F.3d 120 (D.C. Cir. 2013). Interpreted by the Supreme Court in Wal-Mart, Rule 23(a)(2) requires the class action to raise a common legal or factual question that, when answered at trial, would “resolve an issue that is central to the validity of each one of the [class] claims in one stroke.” 564 U.S. at 350. DL emphasized that Wal-Mart has “changed the [legal] landscape” for class certification and, in cases where a policy or practice is challenged, requires the policy or practice to “bridge[]” all the claims through a “common harm . . . affect[ing] each class member.” 713 F.3d at 126-28. In this case, the alleged concrete deficiencies might affect various individuals in the class. But consistent with Wal-Mart and DL, the District Court decided that, for the case to warrant a classwide merits proceeding, at least one of those deficiencies must “affect[] all class members” and “provide[] the ‘glue’” combining all the claims, DL, 713 F.3d at 131 (Edwards, J., concurring) (quoting Wal-Mart, 564 U.S. at 352). Before trial, 16 the District Court determined at least two common questions relating to the eleven deficiencies (and the six characteristics of an adequate plan): (1) whether the District provides adequate discharge planning and (2) whether it informs the plaintiffs of and provides them with meaningful community-based alternatives to living in nursing facilities. See Thorpe, 303 F.R.D. at 146 n.58. (There might be others, but the District Court had no need to discuss them in pretrial proceedings. See Wal-Mart, 564 U.S. at 359 (noting that courts need find only a “single” common question (citation omitted)).) The class having been certified, the parties thus needed to litigate the answers to the common questions at trial. See 4 WILLIAM B. RUBENSTEIN, NEWBERG ON CLASS ACTIONS § 11:1 (5th ed. 2019). A negative answer would show that all class members share at least some of the same legal injuries from the District’s institutionalization, and that the injuries result from a concrete, systemic deficiency in the District’s transition services. Such classwide injury, and the correspondingly broad remedy to redress it, would distinguish this case from Olmstead. Whereas the lower court in Georgia could demand only the successful placements of the two individual plaintiffs, the District Court here could require institutional changes to the government’s transition programming. See, e.g., Lewis v. Casey, 518 U.S. 343, 360 n.7 (1996) (rejecting “systemwide relief” where the plaintiffs failed to prove that violations “pervaded the [government’s] system” (citation omitted)); Dayton Bd. of Educ. v. Brinkman, 433 U.S. 406, 420 (1977) (“[O]nly if there has been a systemwide impact may there be a systemwide remedy.”). The District Court’s legal error here was forcing the plaintiff class to establish the concrete, systemic deficiencies by a preponderance of the evidence. That cannot be squared with the elements of the ADA claim as interpreted by 17 Olmstead. Because the District Court found the requested classwide accommodations to be facially reasonable in light of the settlement agreements, and because the class at trial established a prima facie entitlement to such accommodations, the burden shifted to the government for proof of inequities. One way to show inequity is to demonstrate that the government’s programming in fact already adequately provides the requested accommodations. If it does, then the plan warrants no systemic changes. Put another way, the District may rebut the need for classwide relief by demonstrating that it lacks the concrete, systemic deficiencies raised by the class pleadings and identified by the District Court. (I note that the parties in In re District of Columbia, 792 F.3d 96 (D.C. Cir. 2015), neither briefed nor raised the issue of who had the burden of proof, and so we had no reason to discuss it at the time.) Now that we have corrected the District Court’s misunderstanding about the burden of proof, we should leave the court alone. It should largely stick to its original plan and determine on remand whether the District has proven, by a preponderance of the evidence, that there is no systemic deficiency related to the six characteristics. That would lead to a proper resolution in this Olmstead Phase II case. The majority opinion instead requires the District Court to engage in a new two-part test largely detached from the characteristics it already articulated. See Majority Op. 24-25 (requiring the District Court to determine first whether the District has an Olmstead Plan, and second whether the plaintiffs’ requested injunctive relief is too “costly”). My colleagues say my approach would lead to a violation of the Rules Enabling Act (REA), 28 U.S.C. § 2072. See Majority Op. 30. The statute provides that no Federal Rule of Civil Procedure may “abridge, enlarge[,] or modify any 18 substantive right.” 28 U.S.C. § 2072(b). Such rights include “who may sue, on what claims, [and] for what relief.” Henderson v. United States, 517 U.S. 654, 671 (1996). In particular, my colleagues submit that it would contravene the REA to look for concrete deficiencies as part of fashioning a classwide remedy. That somehow mutates the nature of the ADA claims, and my colleagues trace the change to Rule 23. I disagree. My colleagues have identified no substantive right that is abridged, enlarged, or modified. The elements of the prima facie claim remain the same in an individual or class action. So too does the fundamental alteration standard: whether inequities arise from immediate relief. If the District Court looks for systemic deficiencies, it is doing so only because they help to reveal the specific accommodations requested by the plaintiffs here and the potential inequities associated with enjoining the District. The only difference between the class action here and Olmstead would be the scope of the remedy. But even then, the District’s “aggregate” duty to provide reasonable accommodations “does not depend on whether the suit proceeds as a class action.” Shady Grove Orthopedic Assocs., P.A. v. Allstate Ins. Co. (Shady Grove), 559 U.S. 393, 408 (2010) (plurality opinion). An injunction in the individual action would provide the accommodation only to the named litigants, while an injunction in the class action would provide the same accommodation to the population at large. The latter remedy would affect more people, result in more significant policy changes, and be considered institutional reform. And in the latter case, some absent class members surely would benefit from freeriding because they would not have sought the relief themselves. Nonetheless, the substantive fix, as applied to each litigant, is the same. Seen in this way, the breadth of the class injunction is only an “incidental” effect on substantive rights, which the REA tolerates. Bus. Guides, Inc. v. Chromatic Commc’ns Enters., 19 Inc., 498 U.S. 533, 553 (1991); see also Shady Grove, 559 U.S. at 408 (plurality opinion). In the end, my colleagues allow that the District Court could “focus its ultimate injunction on the six ‘key components of an effective system of transition assistance’ that it has gleaned from other Olmstead cases.” Majority Op. 23 n.10 (quoting Thorpe, 303 F.R.D. at 148). But if that is case, we should not then opine on a broader fundamental alteration test that I believe misreads Olmstead and could create problems in future cases. We need not issue potentially misleading guidance if the District Court already knows what to do. The District Court should instead rely on the fact-intensive fundamental alteration standard devised by the Olmstead plurality: whether immediate changes to current government programming would create inequities. After reviewing what the District Court has said in its numerous pre-trial opinions, I am confident that it can apply the proper standard here. Through the six characteristics it has developed, the District Court can ensure comprehensive and effective transition services without improperly hamstringing government officials. Because I read the majority opinion as requiring the District Court to change course and apply an improper test, I cannot subscribe to it. IV. I conclude with some remarks about the elephant in the room: causation. Since the start of litigation, the District Court has expressed doubts as to whether the plaintiff class could establish a “causal link between the alleged deficiencies in the District’s system of transition assistance and the alleged 20 unnecessary segregation.” Brown, 322 F.R.D. at 60 (quoting Thorpe, 303 F.R.D. at 142). After the bench trial, the District Court sought supplemental briefing on the issue of causation and the plaintiff’s burden of proof. Id. at 62. For the plaintiffs, the court had a choice between two burdens of proving causation: (1) the traditional “but for” causation (a showing by the plaintiffs that their disability-based institutionalization “would not have occurred” had the District been providing their requested accommodations) or (2) the less onerous “motivating factor” causation (a showing by the plaintiffs that the District’s failure to provide their accommodations was “a ‘motivating’ or ‘substantial’ factor” in their disability-based isolation). See Univ. of Tex. Sw. Med. Ctr. v. Nassar, 570 U.S. 338, 343, 346-48 (2013). Contrary to what Appellant suggests, see Appellant’s Br. 27-39, the District Court said it was avoiding the issue, see Brown, 322 F.R.D. at 89 n.29. Because we are remanding the case, the District Court must return to causation. I make two points on the matter. First, the District Court expressly framed the issue of causation as whether “beef[ing] up the [transition] services” will in fact lead to success in “getting people out.” Id. at 63 (citation omitted). But the target goal should be a meaningful opportunity for a future community placement, not actual success in providing the placement. Accord Majority Op. 23. Second, whether a plaintiff has the burden of showing “but for” or “motivating factor” causation in a disability- discrimination failure-to-accommodate claim is a head- spinning legal question. For the reasons set forth below, I conclude that the plaintiffs in an Olmstead claim may establish only “motivating factor” causation in their prima facie case, and that the defendant may then prove the absence of “but for” causation to rebut liability. 21 In identifying the proper causation inquiry, we must pay close attention to the statutory text and context. See Husted v. A. Philip Randolph Inst., 138 S. Ct. 1833, 1842-43 (2018); Maslenjak v. United States, 137 S. Ct. 1918, 1929-30 (2017); Nassar, 570 U.S. at 343, 350-51; Gross v. FBL Fin. Servs., Inc., 557 U.S. 167, 175-76 (2009); Ford v. Mabus, 629 F.3d 198, 204-06 (D.C. Cir. 2010). As noted above, Title II bans government discrimination “by reason of” the individual’s disability. 42 U.S.C. § 12132. And § 504 of the Rehabilitation Act bans discrimination “solely by reason of” disability. 29 U.S.C. § 794(a); see also Lunceford v. Dist. of Columbia Bd. of Educ., 745 F.2d 1577, 1580 (D.C. Cir. 1984) (Ginsburg, J.). Our sister circuits are split as to whether the phrase “by reason of” and the absence of language authorizing a burden- shifting regime imply that the plaintiff has the burden of proving “but for” or “motivating factor” causation. Compare Haberle v. Troxell, 885 F.3d 170, 179 (3d Cir. 2018) (but for), A.H. ex rel. Holzmueller v. Ill. High Sch. Ass’n, 881 F.3d 587, 593-94 (7th Cir. 2018), and Gohl v. Livonia Pub. Sch. Dist., 836 F.3d 672, 682 (6th Cir. 2016), with K.M. ex rel. Bright v. Tustin Unified Sch. Dist., 725 F.3d 1088, 1099 (9th Cir. 2013) (motivating factor), Pinkerton v. Spellings, 529 F.3d 513, 517- 18 (5th Cir. 2008) (per curiam), and Baird ex rel. Baird v. Rose, 192 F.3d 462, 470 (4th Cir. 1999). Despite Appellant’s suggestions to the contrary, see Appellant’s Br. 29, the Second Circuit has declined to decide the issue, see Bolmer v. Oliveira, 594 F.3d 134, 148-49 (2d Cir. 2010). The circuit split reflects the difficulty of the interpretive issue. As we have explained in Ford, the causation issue comprises two distinct legal questions: which “standard of causation” does the statute at issue impose, and whether burden-shifting may occur. 629 F.3d at 204. Given recent 22 Supreme Court precedents, I think that the first question is clear. The second is not. The Supreme Court in Gross noted that the term “by reason of” ordinarily means “but for” causation, and numerous precedents have endorsed the Court’s observation. See 557 U.S. at 176; see also Burrage v. United States, 571 U.S. 204, 213 (2014); Nasser, 570 U.S. at 350; Bridge v. Phoenix Bond & Indem. Co., 553 U.S. 639, 653-54 (2008); Holmes v. Sec. Inv’r Prot. Corp., 503 U.S. 258, 267-68 (1992); cf. A. Philip Randolph Inst., 138 S. Ct. at 1842-43 (finding that “by reason of” in statute at issue implied sole causation, which is more stringent than “but for” causation). A straightforward application of Gross indicates that “but for” causation is an element of the Title II claim. Moreover, the Supreme Court in Burrage exposed the improbability of a federal statute demanding “motivating factor” causation simpliciter. The Supreme Court articulated the “motivating factor” test in Price Waterhouse v. Hopkins, 490 U.S. 228 (1989), superseded in part on other grounds by Civil Rights Act of 1991, Pub. L. No. 102-166, 105 Stat. 1071 (codified as amended at scattered sections of 2, 16, 29, 42 U.S.C.), for claims under Title VII of the Civil Rights Act of 1964, Pub. L. No. 88-352, 78 Stat. 253 (codified as amended at 42 U.S.C. § 2000e et seq.). Under the Price Waterhouse doctrine, the burden of proving causation shifts from the plaintiff to the defendant once the former proves that her protected status (in that case, gender) “played a motivating part” in the latter’s adverse action, and the defendant “may avoid a finding of liability only be proving by a preponderance of the evidence that it would have” performed the action “even if it had not taken” her status “into account.” Gross, 557 U.S. at 173-74 (quoting Price Waterhouse, 490 U.S. at 258 (plurality opinion)). But as the Court clarified in Burrage, 23 Price Waterhouse never held that Title VII required only “motivating factor” causation; the key move in the latter case was to shift the burden of proving “but for” causation to the defendant after the plaintiff made a successful prima facie showing. See 571 U.S. at 213 n.4; see also Gross, 557 U.S. at 173-74. “But for” causation remained the standard for Title VII claims until Congress expressly adopted the “motivating factor” test in the Civil Rights Act of 1991. Burrage, 571 U.S. at 213 n.4 (citing 42 U.S.C. § 2000e-2(m)). The Supreme Court also noted that courts have not yet found motivating factors simpliciter to be sufficient in practice; in cases where such a factor was identified, it either was an “independently effective” cause among other sufficient causes or would satisfy the “but for” test in any event. Id. at 215-16 (citation omitted). This is not to say that Congress is forbidden from mandating only “motivating factor” causation. See, e.g., In re U.S. Office of Pers. Mgmt. Data Sec. Breach Litig., No. 17- 5217, 2019 WL 2552955, at *15 (D.C. Cir. June 21, 2019) (per curiam); Ford, 629 F.3d at 206; cf. Burrage, 571 U.S. at 214 (noting the “undoubted reality that courts have not always required strict but-for causality”). But given the plain meaning of Title II’s statutory language and Burrage’s illumination of Price Waterhouse, I am compelled to conclude that Congress wanted “but for” causation here. But which side must prove the existence (or lack) of “but for” causation? The statutory language and precedents do not plainly lead to an answer. As Gross made clear, when “the statutory text is ‘silent on the allocation of the burden of persuasion,’” courts must “‘begin with the ordinary default rule that plaintiffs bear the risk of failing to prove their claims.’” 557 U.S. at 177 (quoting Schaffer ex rel. Schaffer v. Weast, 546 U.S. 49, 56 (2005)). 24 Title II says nothing about burden of proof, which implicates the default rule that the plaintiffs bears the burden of demonstrating “but for” causation. The Gross Court also suggested that the burden-shifting approach in Price Waterhouse is “difficult to apply,” and that the “perceivable benefit” of applying the approach to other statutory contexts may be “eliminated” by the practical problems. Id. at 179. Still, certain cues suggest that burden-shifting akin to Price Waterhouse may occur here. The parallel Rehabilitation Act provision expressly bans discrimination “solely by reason of” disability. 29 U.S.C. § 794(a). The omission of the word “solely” in Title II is a reasonable signal that Congress wanted to adopt something like Price Waterhouse burden-shifting. See Gross, 557 U.S. at 183 n.4 (Breyer, J., dissenting); see also Price Waterhouse, 490 U.S. at 241 & n.7 (plurality opinion); id. at 258-59, 268-69 (White, J., concurring in the judgment). In addition, one can distinguish this case from Gross, which required the plaintiff to establish “but for” causation for claims under the Age Discrimination in Employment Act of 1967 (ADEA), Pub. L. No. 90-202, 81 Stat. 602 (codified as amended at 29 U.S.C. § 621 et seq.). Gross emphasized that the Civil Rights Act of 1991 adopted a modified Price Waterhouse burden-shifting regime for Title VII, amended the ADEA, and declined to create such a regime for the ADEA. See 557 U.S. at 174-75. The Supreme Court interpreted the omission as an intentional decision by Congress to forego the Price Waterhouse framework for the ADEA. See id. at 174- 75, 177 n.3. But the 1991 statute never touched Title II, and so the reasoning in Gross does not squarely apply. Moreover, it is permissible to consider agency interpretations of a statutory gap. See Gross, 557 U.S. at 179 n.6 (citing NLRB v. Transp. Mgmt. Corp., 462 U.S. 393, 400- 03 (1983)); see also Olmstead, 527 U.S. at 598 (noting that the 25 ADA regulations are entitled at least to “respect” under Skidmore v. Swift & Co., 323 U.S. 134 (1944)). Unfortunately, the relevant Title II regulation on its face provides only some, but not much, help. See 28 C.F.R. § 35.130(b)(7)(i) (“A public entity shall make reasonable modifications in policies, practices, or procedures when the modifications are necessary to avoid discrimination on the basis of disability, unless the public entity can demonstrate that making the modifications would fundamentally alter the nature of the service, program, or activity.”). The regulation clearly outlines a burden-shifting framework, but it does not explicitly indicate that the burden of proving causation moves at any point. The text of the regulation does not make clear whether the government’s burden of establishing a fundamental alteration entails the burden of proving the absence of “but for” causation. As for the plaintiff’s prima facie case, the “necessary” prong of the regulation incorporates a demonstration of causation, see Wis. Cmty. Servs., Inc. v. City of Milwaukee, 465 F.3d 737, 749 (7th Cir. 2006) (en banc), but it is not obvious whether “necessary” means that the plaintiff must show “but for” or only “motivating factor” causation in her prima facie case. The prong uses the phrase “on the basis of,” see 28 C.F.R. § 35.130(b)(7)(i), but the legal meaning of that phrase has split our sister circuits in an analogous context, compare Natofsky v. City of New York, 921 F.3d 337, 348 (2d Cir. 2019) (but for), Gentry v. E. W. Partners Club Mgmt. Co., Inc., 816 F.3d 228, 235-36 (4th Cir. 2016), and Demyanovich v. Cadon Plating & Coatings, L.L.C., 747 F.3d 419, 433 (6th Cir. 2014), with EEOC v. LHC Grp., Inc., 773 F.3d 688, 702-03 (5th Cir. 2014) (motivating factor), with at least two circuits reserving the issue, see Monroe v. Ind. Dep’t of Transp., 871 F.3d 495, 504 (7th Cir. 2017); Oehmke v. Medtronic, Inc., 844 F.3d 748, 757 n.6 (8th Cir. 2016). 26 After surveying the relevant cases and authorities, I find that we have an ambiguous agency regulation filling a gap in the Title II provision. So how should we decide the burden of persuasion? Thankfully, I need not stake out any broad positions about all Title II claims in general. With respect to Olmstead claims, we have useful interpretive guidance from the Department of Justice (DOJ), which promulgated the Title II regulations. As the District Court has recognized, see Brown, 322 F.R.D. at 62 n.8, the DOJ believes that plaintiffs making an Olmstead claim “could make out a case . . . even if they could not prove [that] ‘but for’ the disability, they would have received the community-based services they sought,” U.S. DEP’T OF JUSTICE, CIVIL RIGHTS DIV., STATEMENT OF THE DEPARTMENT OF JUSTICE ON ENFORCEMENT OF THE INTEGRATION MANDATE OF TITLE II OF THE AMERICANS WITH DISABILITIES ACT AND OLMSTEAD V. L.C. 4 (June 22, 2011), https://www.ada.gov/olmstead/q&a_olmstead.pdf. I need not determine whether the DOJ’s interpretation warrants deference under Auer v. Robbins, 519 U.S. 452 (1997). See Kisor v. Wilkie, No. 18-15, 2019 WL 2605554, at *8-10 (U.S. June 26, 2019) (describing multifactor test for Auer deference). Regardless, the interpretation would merit Skidmore respect because the interpretation has not changed over time and the DOJ has amassed a considerable “body of experience and informed judgment” on the issue of disability discrimination in government treatment programs. See Olmstead, 527 U.S. at 598 (citation omitted). See generally OLMSTEAD: COMMUNITY INTEGRATION FOR EVERYONE, https://www.ada.gov/olmstead (last visited June 13, 2019). And despite having an opportunity to do so, the government has not challenged the interpretation in this litigation. See generally Reply to Plaintiffs’ Opposition, Brown v. District of Columbia, No. 1:10-cv-02250-ESH (D.D.C. filed Nov. 4, 2016), ECF No. 226. (The District Court suggested that the DOJ’s interpretation might apply only in the individual case, not in class actions. See Brown, 322 F.R.D. at 27 62 n.8. That cannot be correct; the upshot would be that individual and class actions have different proof-of-causation regimes, which would be a clear REA violation.) To reconcile Title II’s requirement of “but for” causation and the DOJ’s interpretation that plaintiffs need not shoulder the burden of proving it in the Olmstead context, I conclude that Olmstead claims must proceed under the Price Waterhouse framework. The plaintiff in an Olmstead claim must show that the government’s failure to provide the requested accommodations was a motivating factor in her institutionalization. See Thorpe, 303 F.R.D. at 148 (noting that the “lack of transition services” must “contribute[] to the lack of placements of residents into community-based services” (emphasis added)). And at the fundamental alteration stage, the government may establish by a preponderance of the evidence that the institutionalization would have occurred even with all the accommodations. In doing so, the government would sever the “but for” link between the plaintiffs’ disabilities and their isolation. The class plaintiffs clearly made their prima facie showing of causation here. By establishing that the government thinks it appropriate for them to receive community treatment, and that they desire a community placement, see Brown, 322 F.R.D. at 87, they have made the preliminary showing of their “unjustified institutional isolation,” Olmstead, 527 U.S. at 600, 602-03. As a matter of law, the Supreme Court has explained, such “[d]issimilar treatment” is discrimination by reason of their disability: “In order to receive needed medical services [from the government], [they] must, because of [their] disabilities, relinquish participation in community life they could enjoy given reasonable accommodations, while persons without [those] disabilities can receive the medical 28 services . . . without similar sacrifice.” Id. at 601 (emphasis added). On remand, the District may attempt to prove that “but for” causation does not exist – in other words, that the plaintiffs would remain in their nursing facilities even if the government were to fix the identified deficiencies. When read alongside the DOJ’s interpretation of the Title II regulation, the fundamental alteration standard must allow for such an attack on causation. This makes eminent sense; it is inequitable to require the government to change its programming if the change is futile. The District Court expressed concerns that the lack of housing might break the causal link. See Brown, 322 F.R.D. at 63. Perhaps. But that is for to the District to prove and for the District Court to conclude. As the Price Waterhouse framework makes clear, if the government succeeds in rebutting the prima facie showing of causation, the District Court should enter judgment in its favor. Accordingly, I break from my colleagues’ suggestion that a demonstrated lack of housing “alone” could never “resolve the litigation in the District’s favor.” Majority Op. 26 n.13. For that additional reason, I cannot join the majority opinion. I therefore respectfully concur only in the judgment.