Inside the Latest Medicaid Work Experiment Decisions: Stewart v Azar And Gresham v Azar
On March 27th, Judge James Boasberg of the United States District Court for the District of Columbia handed down his two latest decisions in the ongoing legal saga of § 1115 Medicaid work experiments. The two decisions, Stewart v Azar and Gresham v Azar, came about 2 weeks after oral arguments, 4 months after the administration reapproved Kentucky’s experiment with virtually no changes, and 9 months after the Arkansas experiment launched in 2018, leading to coverage loss for over 18,000 people by year’s end. Arkansas’s demonstration clock reset in January 2019. But on April 1, the state would once again begin to eliminate people who failed for 3 months to report sufficient work; in addition, Kentucky was on the verge of implementing its own Kentucky HEALTH work demonstration. It was no surprise that the court acted fast.
In both cases, the court issued a vacatur (i.e., vacated the HHS approval). In Arkansas’s case, the court went a step further, suspending the ongoing experiment and thus halting further coverage losses on experimental grounds, at least for the time being. The administration has not yet announced its next move, although an appeal seems most likely, likely along with a request to the appeals court that Arkansas be permitted to continue its demonstration while an appeal is pending.
In the meantime, the administration remains undeterred. Its FY 2020 budget calls for a work mandate as an eligibility condition for working-age adults, saving an estimated $130 billion over 10 years, thereby underscoring the magnitude of work mandates on coverage. And on March 29th the Centers for Medicare and Medicaid Services announced its approval of Utah’s partial Medicaid expansion linked to a work requirement. This was a striking development not only because of its timing (48 hours after the judge’s decisions) but because the state’s demonstration directly flies in the face of its own, 2018, voter-approved referendum calling for a simple expansion to reach the entire ACA-eligible population, without added eligibility conditions.
In the meantime, advocates have filed litigation to vacate federal approval of New Hampshire’s work demonstration. This case also has been assigned to Judge Boasberg.
From the Court’s perspective, the die likely was cast in Stewart once it became evident that the administration simply had ignored its clear instructions when it remanded the case in Stewart I. The Court expected the Secretary to explain how an experiment projected by the state to remove 95,000 beneficiaries from Kentucky’s Medicaid program (and potentially hundreds of thousands more according estimates prepared by plaintiffs’ amici) would promote Medicaid’s statutory objective of furnishing medical assistance to eligible people. Instead, the Secretary sidestepped matters in favor of a strategy of once again reframing Medicaid program objectives. It was not the reframing itself that proved fatal the second time through; Judge Boasberg actually left the door open to other possible objectives that the Secretary might consider when evaluating whether to proceed with a Medicaid experiment. What the administration failed to do was demonstrate that it had engaged in an evidence-based, reasoned decisionmaking process that weighed coverage losses and their consequences against measurable, realistic gains in coverage and care that could justify the risk to Medicaid eligibility.
Furthermore, Stewart I inevitably sealed the fate for the Arkansas work experiment challenged in Gresham; as far as the Court was concerned, the administration’s approval in that case suffered from the same fundamental flaws as Stewart I – failure to analyze how the experiment would promote Medicaid’s central objective of furnishing medical assistance for eligible people. Once again, these decisions drive home a basic principle of administrative law: even if a court is justified in deferring to an agency’s actions, those actions must be based in evidence and must be supported by analysis that considers all relevant factors.
Stewart II
Judge Boasberg began by recapping his previous ruling -- that the administration’s “signal omission” was its failure to “adequately consider[] whether Kentucky HEALTH would in fact help the state furnish medical assistance to its citizens.” Now, Judge Boasberg announced, “[t]he bell rings for round two” – namely, whether the defendants had done what he instructed them to do.
The Secretary argued that the second comment period following remand in Stewart I had “cured any crucial omission,” although the second approval relied “on somewhat different reasoning.” According to the Court, this new reasoning rested principally on the threat to withdraw coverage for the entire expansion population if Kentucky’s experiment did not proceed. In other words, defendants’ approval boiled down to the fact that while the experiment “may cause nearly 100,000 people to lose coverage, that number will be dwarfed by the approximately 450,000 people who would suffer that fate if Kentucky ends its coverage entirely of those who have joined the Medicaid roles via the Affordable Care Act, as it has threatened to do if the project is not approved.”
Taking a cue from the “gun to the head” image memorably captured by Chief Justice Roberts in declaring unconstitutional the mandatory Medicaid expansion in National Federation of Independent Businesses v Sebelius (2012), Judge Boasberg observed that:
Kentucky, it seems, has now picked up that gun by threatening to de-expand Medicaid. [Because of the threat,] [d]efendants urge the Court to adopt the proposition that the Secretary need not grapple with the coverage loss implications of a state’s proposed project as long as it is accompanied by a threat that the state will de-expand – or indeed, discontinue all Medicaid. By definition, so this argument goes, any number of people covered by an experimental Medicaid program would be greater than the number if there were no Medicaid at all; as a result, any demonstration project that leaves any individual on a state’s Medicaid rolls promotes coverage. The Court cannot concur that the Medicaid Act leaves the Secretary so unconstrained, nor that the states are so armed to refashion the program Congress designed any way they choose.
In other words, according to the Court, the administration’s logic provides no stopping point to experimental powers – a position that renders the 1115 statute, along with the underlying statutory purpose of Medicaid, meaningless, and the results, absurd.
As before, the Court left unaddressed certain claims raised by the plaintiffs, including, among other matters, a claim that the original January, 2018, community engagement 1115 demonstration solicitation violated the APA because the administration never went through a rulemaking process. Nor, as before, did the court conclude that work experiments are inherently contradictory to Medicaid’s purpose. Instead, Judge Boasberg devoted his lengthy opinion to a close examination of the procedural dimension of the approval – that is, to whether, as instructed, the Secretary built an evidence-based case for his actions.
Significantly, Judge Boasberg acknowledged that the fiscal sustainability of the Medicaid program is a factor that can be considered in the design and approval of 1115 experiments. But the administration failed two ways in this regard – both in mis-framing the role that sustainability considerations could play in the § 1115 review process and in failing to produce the evidence that would support his decision to rely on sustainability as a reason to proceed with an experiment that involves considerable reduction in eligibility.
The Secretary’s Reasoning And The Court’s Response
The Court found that the Secretary’s second approval rested on three key points: (1) commenters did not understand the nature of the coverage loss and thus mischaracterized its magnitude; (2) the experiment’s guardrails would minimize loss; and (3) the experiment was simply a modified expansion and a warranted response to the reality of Kentucky’s power to entirely eliminate coverage of the expansion population. Better to protect 450,000 rather than lose 95,000, in other words. In this reframing, the Secretary did concede that coverage remains an “important” objective. But, he argued, “there is little intrinsic value in paying for [those] services if they are not advancing the health and wellness of individuals receiving them, or otherwise helping the individual attain independence.” For this reason, coverage took a back seat to wellness, independence, and arguably, sustainability.
In considering this framing, the Court first held that health and financial independence “are not stand-alone objectives of the statute” and cannot make up for the failure to consider coverage impact. This conclusion was especially true, Judge Boasberg wrote, “where the Secretary made no attempt to weigh any of those three aims against the coverage-loss consequences of” the demonstration. The Court then turned to the various objectives as articulated by the administration, along with the Secretary’s process for weighing them in relation to the evidence.
Coverage Losses
With respect to coverage – which the administration conceded remained on the table as a factor – Judge Boasberg once again concluded that the Secretary never provided a bottom line estimate of coverage loss. Whether one accepted Kentucky’s 95,000 estimate at the low end or amici scholars’ far higher estimates, the court noted, the loss was “substantial.” The Secretary argued that fiscal sustainability was a reason to accept losses and that losses should be weighed against the state’s power to end the expansion entirely. This argument, the judge observed, rested on an unreasonable reading of Section 1115 itself and the power it confers, a point the judge would come back to later in his opinion. It is true, the Court determined, that §1115 demonstrations can affect eligibility; indeed, the statute contemplates demonstrations that have an impact on eligibility. But this doesn’t mean that the Secretary is relieved “of his obligation to consider the magnitude of the coverage loss” and determine how such losses “promote the objectives” of the Act.
The Secretary attempted to argue that he had no obligation to assess coverage impact with precision and that, in any event, the losses were inaccurately stated by the commenters as individuals rather than the more technical “member months” by which enrollment impact estimates are often expressed. Judge Boasberg pointed out that regardless of whether one counts impact by annualizing losses or by the more technical approach of calculating “member months,” the loss had to be reckoned with:
There are, as the Secretary acknowledges, multiple ways to slice the number of lost months: 95,000 may lose coverage for a year; a larger number may lose coverage for shorter periods of time; or fewer people may be deprived of coverage for lengthier periods. Regardless of how the number of lost member months is distributed among Medicaid beneficiaries, it indisputably reflects that a substantial number will lose coverage. As such, the Secretary cannot avoid addressing that number. This is especially so where commentators detailed the widespread predicted nature of coverage loss and its devastating effects.
The failure to consider coverage loss was compounded by the Secretary’s assertion that people would acquire commercial insurance, without offering evidence to back up the assertion – a particularly glaring omission given the fact that Kentucky HEALTH’s community engagement requirement can be satisfied through school attendance, volunteer work, and job search, none of which carry health benefits. To the Secretary’s arguments that the Kentucky projections did not account for “guardrails,” in fact, the Court concluded, the guardrails had been “baked into” the estimates. The bottom line from Judge Boasberg’s perspective was that the Secretary “granted the waivers with no idea how many people might lose Medicaid coverage.”
The Secretary also attempted to argue that the requirement would in fact promote coverage through the “My Rewards” accounts whose purpose, according to Defendants, is to help people become better health care users. But, the Court found, “[w]hether or note the program generally will lead to an uptick in preventive care, the Secretary makes no effort to quantify that uptick or to weigh it against coverage losses for those whom Kentucky HEALTH may deprive of all access to care.”
Health
The Court found that even were it to defer to the Secretary’s declaration that health is an objective of Medicaid, the administration’s position would fail. As Judge Boasberg previously had noted, “[t]reating health – rather than the furnishing of medical services – as the Act’s ultimate goal is ‘nothing more than a sleight of hand.’” Such an interpretation of the statute falls outside “the bounds of reasonableness” that agencies must adhere to when interpreting a law, since, in the Court’s view, the statutory construction offered by Defendants fits with neither statutory language nor statutory purpose. It was health care affordability that led to Medicaid, along with the ACA amendments. The Court held that “[t]he Secretary is not free instead to extrapolate the objectives of the statute to a higher level of generality and pursue that aim in the way he prefers.”
The Court rejected this position, not only because of its lack of basis in law, but because doing so would “have bizarre results. Were this the case, nothing would prevent the Secretary from conditioning coverage on a special diet or certain exercise regime.” For its part, Kentucky reiterated its argument that the ACA expansions are to be judged by standards that differ from those used for the traditional population and that for the expansion population, Medicaid is simply “transitional.” The Court, however, found no evidence that Congress intended this.
Financial Independence
According to the Court, the Secretary did not explain the statutory basis of his assertion that financial independence was a Medicaid objective, and instead erroneously pointed out that there is “little intrinsic value in paying for services if those services are not improving. . . health or otherwise helping individual[s] attain independence.” To the extent that the Secretary previously had pointed to language in Medicaid’s reference to rehabilitation coverage, this reference was not relevant to the populations affected by the work demonstrations. As the Court previously had noted, “the text … quite clearly limits the objectives to helping States furnish rehabilitation and other services that might promote self-care and independence so that it does not follow that limiting access to medical assistance would further that same end.”
Kentucky argued that financial independence justified previous work demonstrations under the Aid to Families with Dependent Children (the predecessor to Temporary Aid to Needy Families (TANF)), but the court noted that AFDC and Medicaid had completely different statutory purposes and thus, the justifications on which rested AFDC work demonstrations were irrelevant to Medicaid.
Finally, the Court concluded, even if financial independence were a “proper consideration” for approving a Medicaid work experiment, the Secretary’s approval on this basis must fail, since he never weighed the number who would gain financial independence and move out of the program against those who would lose coverage. “Even if some number of beneficiaries were to gain independence, the Secretary does not weigh the benefits of their self-sufficiency against the consequences of coverage loss, which would harm and undermine the financial self-sufficiency of others. These deficiencies render his determination arbitrary and capricious.”
Financial Sustainability
Judge Boasberg began this portion of his opinion with the following: “Long diverted into myriad other byways, the Court now arrives at the broad avenue that constitutes Defendants’ key position.” Thus began the financial sustainability analysis.
Essentially, as Defendants’ argument went, Kentucky could pull out of the expansion at any time, meaning that allowing the demonstration to proceed became essential to saving the expansion for the rest. The Court concluded that Defendants were less than clear as to whether the Secretary considered fiscal sustainability an independent objective or simply the means by which coverage for most of the expansion population could be saved and for that reason, dealt with both aspects of the argument.
The Court first assumed that the term “objectives” as used in the 1115 statute carries some ambiguity, thereby leaving room for judicial deference to the agency’s legal interpretation of fiscal sustainability as an objective. In that vein, it proceeded to weigh the defense’s arguments.
The Court thus permitted the Secretary to identify sustainability as a Medicaid program objective. But once he did so, the Secretary was obligated to explain “why Kentucky HEALTH advances that objective and why, if it is adverse to other Medicaid objectives, he could reasonably conclude that, on balance, it promotes the objectives of the Act as required by § 1115.” “On these fronts he fell short”:
First, he made no findings, supported by substantial evidence, that Kentucky HEALTH would improve the sustainability of Kentucky’s Medicaid program – either by accruing savings to the state or by any other mechanism. Second, he unreasonably prioritized program savings without weighing those against the consequences of lost coverage, rendering his determination arbitrary and capricious.
The Court further held that were no findings regarding how much money the experiment would save, and indeed, the record suggested that the state had vastly overstated expected savings ($331 million versus $2.2 billion). This failure to estimate savings is especially notable where “the record contains some evidence to believe that full administration of the expansion will save the Commonwealth money, while reducing coverage of the expansion population will cost the state.” The Court stressed that the point was not to evaluate “how Kentucky ought to spend its money.” And indeed, the Court noted, Kentucky ultimately could eliminate the expansion. But Judge Boasberg’s central point was that in using 1115 authority, the Secretary must make a “finding about the savings that Kentucky HEALTH could be expected to yield” and that without such findings “the Secretary could not make a reasoned decision that it would promote fiscal sustainability.”
Furthermore, the Court determined, the Secretary’s “reliance on fiscal sustainability was arbitrary and capricious because he did not compare the benefit of savings to the consequences for coverage,” and noted that the Court of Appeals for the Ninth Circuit has “twice invalidated similar approvals.” The Secretary attempted to distinguish these cases by arguing that these cases involved “simple benefit cuts enacted to save money.” Unconvinced by this distinction, the Court concluded that the problem with the Secretary’s reliance on sustainability here was that:
[H]e did not adequately consider the significant number of people for whom the program would entail a loss of all benefits. Their loss of coverage appears, from the record in this case, to be how the Commonwealth would save money. This is precisely what the Ninth Circuit said states cannot do with a § 1115 waiver.
The Court rejected as relevant precedent other cases cited by the Secretary as establishing precedent for allowing benefit reductions as a means of making Medicaid more sustainable. In the Court’s opinion, those cases involved reductions merely incidental to coverage and not a “significant impediment to Medicaid services.” Indeed, the Court found, these previous cases make “eminently clear that a project that enhances financial sustainability may not advance the objectives of Medicaid if it significantly impedes or curtails Medicaid services or coverage.” None of the cases cited “threatened the entirety of beneficiaries’ Medicaid coverage . . . in the name of cost savings.” They merely upheld “incidental” burdens.
In response to Defendants’ argument that Kentucky HEALTH promotes coverage because of the state’s de-expansion threat in the absence of approval, the Court dismissed the position as “inconsistent with the Medicaid act and arbitrary and capricious.” First, the Court noted, the government was incorrect that states have additional discretion to “diminish or condition eligibility for the expansion – as opposed to the traditional – population”:
[T]he privilege Kentucky seeks to exercise here is not to de-expand but to implement the ACA expansion as an a la carte exercise, picking and choosing which of Congress’s mandates it wishes to implement. [NFIB v Sebelius] did not sanction that. Rather the Court was very clear: Nothing in [NFIB] precludes Congress from offering funds under the Affordable Care Act to expand the availability of health care, and requiring that states accepting such funds comply with the conditions on their use. . . . Nothing in [NFIB] allows for “additional discretion” in how the states comply with Medicaid requirements for the expansion population as compared to the traditional one.
In other words, in the Court’s view, the ACA expansion affords HHS no additional experimental discretion in connection with the expansion population over and above what the agency would have in relation to the traditional population.[1] Both populations are equally entitled to medical assistance, in other words.
Labeling the administration’s argument a “red herring,” the Court stressed that the entire Medicaid program is optional for states; as a result, “[t]he Court does not see why – if Defendants are correct that threats to terminate the expansion program can supply the baseline for the Secretary’s 1115 review – that argument would not be equally good as applied to traditional Medicaid.”
Taken to its logical conclusion, the Secretary’s position thus makes little sense. Under his reasoning, states may threaten that they wish to de-expand, or indeed do away with all of Medicaid – for fiscal reasons or no reason at all – if the Secretary does not approve whatever waiver of whatever Medicaid requirements they wish to obtain. . . . [A]ny waiver would be coverage promoting compared to a world in which the state offers no coverage at all. Remarkably, when asked for a limiting principle to this proposition during oral argument, Defendant did not give one.
Not only does Defendants’ position entail radical results, but it is also inconsistent with the text of § 1115. The statute requires the Secretary to evaluate whether the project will promote the objectives of the Act. Against what baseline is he supposed to evaluate the project? The structure of the waiver provision assumes the implementation of the Act. It confirms that the relevant baseline is whether the waiver will still promote the objectives of the Act as compared to compliance with the statute’s requirements, not as compared with a hypothetical future universe where there is no Act. . . . The Court, furthermore, need not exclude the possibility that fiscal considerations are ever permissible in any context to reject the staggering breadth of the argument that Defendants present here. [Their] argument does not depend on dealing with the expansion population; it is equally applicable to traditional Medicaid. It does not depend on a state’s being in a fiscally precarious position, because it does not take into account the reason the states want to discontinue participating in the Medicaid program. It is not subject to any kind of limiting principle [and] constitutes an impermissible construction of the statute . . . because [it]is utterly unreasonable in its breadth.
The Court concluded by stressing that it was not ruling “that the agency may never consider the fiscal sustainability of the Medicaid program.” Finding more efficient ways to furnish medical assistance would be consistent with the “prime objective” of the Act. “But that is not the exercise the Secretary engaged in here.”
With that, all that was left was the remedy. Here, Judge Boasberg had no difficulty:
Stewart I offered clear guidance that Section 1115 mandated that coverage considerations be a central part of the analysis. Rather than follow that direction, the Secretary doubled down on his consideration of other aims of the Medicaid Act. Given a second failure to adequately consider one of Medicaid’s central objectives, the Court has some question about HHS’ ability to cure the defects in the approval.
With that, the Kentucky II approval was vacated.
Gresham
Déjà vu. The decision regarding the Arkansas approval was the easier of the two, since the approval itself tracked Stewart I in its approach and logic. Indeed, Judge Boasberg noted, “As Opening Day arrives, the Court finds its guiding principle in Yogi Berra’s aphorism, ‘It’s déjà vu all over again.’” In other words, the failures of Gresham were virtually identical to Stewart.
Following his stage-setting review of Stewart I, with its holding regarding the central role of coverage as one of Medicaid’s objectives, the judge again found that HHS simply never determined whether Arkansas Works would further that objective. The same objectives articulated by the Secretary in Stewart I (improving health outcomes, incentivizing better behaviors) undergirded the Arkansas approval as well. Indeed, the court noted, that Defendants (Arkansas and the administration) conceded during the oral argument that they had done no better in Arkansas than they had in Kentucky at getting to the heart of the matter. The decision did not measure gains against losses; indeed, the Secretary ignored all comments related to coverage loss in his analysis. To the Secretary’s assertion that he did acknowledge the comments, the court responded that “stating that a factor was considered . . . is not a substitute for considering it.” Yet nowhere did the Secretary respond to the coverage predictions, thereby signaling a failure to consider an important aspect of the problem, in contravention of key principles of administrative law.
Remarkably, Defendants offered another argument: the Secretary did not have to consider coverage losses in Arkansas because the state itself had offered no estimates.[2] But such a failure “does not alter HHS’s inquiry”:
Whether a state gives the Secretary excellent data or no data at all about coverage, his duty remains the same: to determine whether the proposed project will promote the objectives of the Act, including whether it advances or hinders the provision of health coverage to the needy.
Promote Coverage
As to the Court’s holding that the Secretary must consider the impact of the experiment on coverage, the Defendants pointed to the fact that HHS had considered the state’s request for a waiver of retroactive eligibility and concluded that elimination of retroactive coverage will encourage beneficiaries to enroll earlier and maintain continuous coverage. But, the Court pointed out, “[l]ittle needs to be said on this score… It is well established that conclusory or unsupported suppositions do not satisfy the agency’s obligation to engage in reasoned decisionmaking.”
Defendants tried to bootstrap Arkansas into approval territory by arguing first that the demonstration served other health promoting purposes and second that the deficiency was cured through the administration’s second reapproval of the Kentucky waiver. Inevitably, neither argument was successful, in light of the Kentucky decision. When the Defendants also attempted to argue fiscal sustainability as a basis for lawful action, the Court stressed that nowhere in the record was there evidence that in Arkansas, as in Kentucky, the Governor had threatened to eliminate the expansion.[3]
The Remedy
In the case of Kentucky, the remedy was pretty straightforward – a simple vacatur, since nothing had been implemented. Arkansas obviously presents a different picture. Noting that the question of remedy was not “small beer,” Judge Boasberg considered several factors in deciding whether to halt the experiment or allow it to continue.
- Would HHS be able to cure the defect on remand? Based on the Secretary’s handling of the remand in Stewart I, the Court held out little hope for remanding Arkansas and holding off on relief to the plaintiffs to see what the defendants would do. “Because the agency failed to provide a legally sufficient rationale upon remand from Stewart I, the Court is even less sanguine that it will be able to do so in this case than when it vacated the Secretary’s Kentucky approval the first time.” Indeed, the Court concluded that approving Arkansas might be even more difficult, since the state did not argue that the demonstration was needed on fiscal sustainability grounds. Since the Court had not foreclosed sustainability as a possible justification (it merely concluded that HHS had not met its burden), this in theory could be a plausible basis for reapproval. But in fact, the state had never raised cost as a factor; indeed, evidence in the record suggested that expansion actually saved Arkansas money.
- How serious would the disruption be? This, the Court said, was a “closer call.” Both the state and the administration argued that interrupting the experiment would be disruptive because it would interfere with the state’s data collection efforts and undermine its “extensive efforts” to educate beneficiaries about the demonstration. But, the Court concluded, while it was not insensitive to the problem of disruption, the fact was that the project had been allowed to begin without even an approval of an objective evaluation design (suggesting that data collection was not all that critical to anyone), and furthermore, that data collection and beneficiary education could always begin again. The more serious harm was to not pause the experiment, thereby leading to a new round, potentially, of coverage losses.
Thus, concerns over disruption, coupled with skepticism about the administration’s ability to cure the problems with its approval, favored a vacatur and therefore an end to the experiment for the time being.
What Happens Next?
In light of the Court’s dismissal of the administration’s effort to fundamentally reframe Medicaid’s objectives and thus the standard against which Medicaid eligibility reduction experiments should be evaluated -- at least in the absence of a strong evidentiary record justifying coverage losses – one should not expect the administration to mount a third attempt at gaining District Court approval. An appeal is more likely. In the meantime however, New Hampshire is on deck before Judge Boasberg, and of course, more may follow given the fact that the administration is now up to nine approvals.
Should the federal government appeal, another issue is whether the District Court will hold off on ruling in New Hampshire pending the outcome of an appeal. This is probably unlikely, given the consequences of allowing New Hampshire to implement a demonstration whose terms are even harsher than those imposed by Arkansas (for example, 100 hours of work per month compared to 80 in Arkansas).
To be continued.
[1] Note that Kentucky HEALTH would apply to both expansion and traditional populations of low income residents, although the Court does not make this point.
[2] The work experiment was presented to HHS as a modification of that state’s existing “private option” demonstration.
[3] Indeed, statements from the Governor following the decision, including the court’s refusal to allow the experiment to continue, nowhere suggest that the state is contemplating ending its expansion demonstration.
