UNITED STATES DISTRICT COURT FOR THE DISTRICT OF COLUMBIA __________________________________ ) UNITEDHEALTHCARE INSURANCE ) COMPANY, et al., ) ) Plaintiffs, ) ) v. ) Civil Action No. 16-157 (RMC) ) ALEX M. AZAR II, Secretary of the ) Department of Health and Human ) Services, et al., ) ) Defendants. ) __________________________________ )
MEMORANDUM OPINION
This Court vacated a final rule issued by the Centers for Medicare & Medicaid
Services (CMS) to determine when certain private insurers were overpaid by Medicare because it
did not comply with the statutory requirement of “actuarial equivalence.” UnitedHealthcare Ins.
Co. v. Azar, 330 F. Supp. 3d 173, 176 (D.D.C. 2018). The government moves for
reconsideration. Although the government does not ask to reinstate the rule, it does ask the
Court to narrow its decision based on new empirical analysis. Because the data underlying that
analysis has long been in CMS’ possession but was not litigated and because the analysis does
not persuade, the Court will deny the motion.
I. BACKGROUND
A more robust description of the statutory scheme, regulatory scheme, and facts
of this case can be found in the Court’s previous decision. See id. at 176-83. A brief recap is
necessary for context.
Under the Medicare Advantage program, Medicare-eligible beneficiaries can elect
to receive health insurance coverage through private insurance companies instead of through
1 traditional Medicare programs administered by CMS. CMS reimburses hospitals participating in
traditional Medicare a fixed amount based on each patient’s diagnosis at discharge, and it
reimburses doctors a fixed amount based on the specific services provided. By comparison,
CMS reimburses insurers participating in Medicare Advantage a fixed amount for each patient
they enroll, based in part on various risk factors including diagnosis on discharge.
Although different reimbursement schemes are at play, by statute CMS must pay
Medicare Advantage insurers in a manner that ensures “actuarial equivalence” with payments to
traditional Medicare providers. See 42 U.S.C. §1395w-23(a)(1)(C)(i). CMS accomplishes this
feat by using a complex risk-adjustment model, the CMS Hierarchical Condition Category
(CMS-HCC) model, to regress total traditional-Medicare expenditures onto traditional-Medicare
beneficiaries’ risk factors. The output of this model is a marginal dollar cost associated with
each risk factor, reduced to a “normalized” risk coefficient that takes as its starting point the
“average beneficiary.”1 Medicare Advantage insurers are paid based on the cumulative risk
scores of their patients.2 The underlying logic is that developing risk coefficients from
traditional Medicare data, and then adjusting a Medicare Advantage beneficiary’s risk score, will
render the cost to CMS under traditional Medicare and the cost to the insurer under Medicare
Advantage actuarily equivalent.
As part of its oversight of the Medicare Advantage program, CMS audits a sample
of reimbursement requests submitted by Medicare Advantage insurers. Costs associated with
1 For example, the model might determine that the average beneficiary receives $10,000 per year in reimbursable expenses and that the marginal cost of a given risk factor is $2,000. By definition the average beneficiary has a risk score of 1.0, so the risk factor would have a normalized risk coefficient of 0.2. 2 For example, a patient with a cumulative risk score of 1.2 costs 20% more than the average beneficiary and the Medicare Advantage insurer would be reimbursed 120% the average benchmark rate.
2 unsupported diagnoses must be reported to CMS. But reimbursement is not limited to only those
audited cases. As of 2008, CMS applies a “Risk Adjustment Data Validation” (RADV) audit to
extrapolate the error rate in the audited sample across an entire insurance contract, and the
insurer is responsible for returning all overpayments calculated based on that extrapolated rate.
RADV audits introduce a complication in this payment scheme. RADV audits
extrapolate an error rate based on audited data from a Medicare Advantage insurer, but Medicare
Advantage payment rates are based on data drawn from traditional Medicare, which is itself
unaudited and admittedly prone to some degree of error. This has the effect of making
traditional Medicare patients appear healthier, and cost less per diagnosis code, than their
Medicare Advantage counterparts.3 For years CMS counterbalanced this effect by implementing
a fee-for-service adjuster (FFS Adjuster), which estimated the error rate present in traditional
Medicare diagnoses; insurers were only responsible for repayment of RADV audit errors
exceeding the estimated traditional Medicare error rate. In early 2014, however, CMS finalized
a rule which eliminated the FFS Adjuster and upset this balance. See 79 Fed. Reg. 29,844 (May
23, 2014) (Overpayment Rule). UnitedHealthcare challenged the Overpayment Rule in January
2016. See Compl. [Dkt. 1].
This Court made three findings relevant to the instant motion when it ruled on
summary judgment. First, the Court determined that “two figures are actuarially equivalent
when they share the same set of actuarial assumptions.” UnitedHealthcare, 330 F. Supp. 3d at
186 (citing Stephens v. U.S. Airways Grp., Inc., 644 F.3d 437, 440 (D.C. Cir. 2011)). “Different
assumptions behind the elements of a calculation would, necessarily, result in actuarially non-
3 This is because the CMS regresses total Medicare expenditures onto both audited and unaudited diagnosis codes. Put another way, costs are spread out among a larger set of diagnoses, such that each individual diagnosis takes up a smaller share of the costs.
3 equivalent results.” Id. Thus, an “inevitable” result of relying on unaudited data to set payment
rates but audited data to determine overpayment is that CMS “will pay less for Medicare
Advantage coverage because,” unlike traditional Medicare settings, “essentially no errors would
be reimbursed.” Id. at 187. This violates the actuarial equivalence requirement of 42 U.S.C.
§ 1395w-23(a)(1)(C)(i).
Second, the statutory scheme requires CMS to establish risk factors for Medicare
Advantage patients “using the same methodology as is expected to be applied in making
payments under” traditional Medicare. 42 U.S.C. § 1395w-23(b)(4)(D). However, beneficiary
risk factors in traditional Medicare were developed using unaudited diagnoses. So, for the same
reason, the Court determined that CMS failed to use the “same methodology” and violated this
statutory requirement when it subsequently applied RADV audits to Medicare Advantage
payments without accounting for the “crucial data mismatch” between audited and unaudited
data. UnitedHealthcare, 330 F. Supp. 3d at 187.
Third, CMS stated as part of prior rulemaking that the FFS Adjuster was
necessary to “account[] for the fact that the documentation standard used in RADV audits to
determine a contract’s payment error (medical records) is different from the documentation
standard used to develop the [Medicare Advantage] risk-adjustment model (FFS claims).” Id. at
188 (quoting RADV Final Methodology at AR5314-15) (emphasis and internal quotations
omitted). Although “Medicare Advantage insurers should [not] be permitted knowingly or
recklessly to bill CMS for erroneous diagnosis codes,” id. at 189, the Court determined that this
concern did not adequately explain why, as a technical matter, the FFS Adjuster was no longer
necessary and why the agency had changed its position. This absence of adequate explanation
4 rendered the Overpayment Rule arbitrary and capricious. Id. (citing Republic Airline Inc. v. U.S.
Dept. of Transp., 669 F.3d 296 (D.C. Cir. 2012)).
For each of these three reasons, the Court vacated the Overpayment Rule. Id. at
192. Sixty days later, the government moved for partial reconsideration under Federal Rule of
Civil Procedure 60(b). See Defs.’ Rule 60(b) Mot. for Partial Recons. (Mot.) [Dkt. 76]. The
government does not dispute that the Overpayment Rule failed to explain the shift in policy, was
arbitrary and capricious, and should remain vacated. Id. at 1. But the government notes that just
weeks after the Court’s decision, CMS finalized an FFS Adjuster Study which concluded that, as
an empirical matter, “diagnosis error in FFS claims data does not lead to systematic payment
error” in the Medicare Advantage program. CMS, Fee for Service Adjuster & Payment Recovery
for Contract Level Risk Adjustment Data Validation Audits at 6 (Oct. 26, 2018) (FFS Adjuster
Study), available at https://tinyurl.com/ve3737d; see also 83 Fed. Reg. 54,982 (Nov. 1, 2018)
(publishing FFS Adjuster Study and soliciting comments on its conclusions). The government
contends that this conclusion calls into question the Court’s own findings regarding the
“inevitable” consequences of a data mismatch between audited and unaudited records, and
further asks the Court, as a matter of judicial prudence, to reconsider its opinion and reserve a
decision on the necessity of the FFS Adjuster until CMS has an opportunity to further investigate
the issue through regular rulemaking processes.
UnitedHealthcare signaled its intent to oppose, but briefing was stayed pending
the release of the data underlying the FFS Adjuster Study. See 12/20/2018 Minute Order. CMS
publicly released some data on April 25, 2019. Shortly thereafter, CMS noticed its intentions to
release “[a]dditional data . . . to all parties who have entered in an applicable data use agreement”
and to “replicate” the FFS Adjuster Study and publish the results. 84 Fed. Reg. 18,215, 18,216
5 (Apr. 30, 2019). CMS published that replicated study—which explained certain methodological
decisions and confirmed the FFS Adjuster Study’s conclusions—on June 28, 2019, and further
extended the comment period for the FFS Adjuster Study. See Defs.’ Reply in Supp. of Their
Rule 60(b) Mot. for Partial Recons. (Reply), Ex. A, FFS Adjuster Study Addendum [Dkt. 97-1];
see also 84 Fed. Reg. 30,983 (June 28, 2019) (2019 FFS Adjuster Study Rule). Although that
rulemaking process has not yet completed, briefing resumed and the motion is now ripe for
review.4
II. LEGAL STANDARD
The government asks the Court for relief pursuant to Federal Rule of Civil
Procedure 60(b). Rule 60(b) provides as follows:
On motion and just terms, the court may relieve a party or its legal representative from a final judgment, order, or proceeding for the following reasons:
(1) mistake, inadvertence, surprise, or excusable neglect;
(2) newly discovered evidence that, with reasonable diligence, could not have been discovered in time to move for a new trial under Rule 59(b);
(3) fraud (whether previously called intrinsic or extrinsic), misrepresentation, or misconduct by an opposing party;
(4) the judgment is void;
(5) the judgment has been satisfied, released, or discharged; it is based on an earlier judgment that has been reversed or vacated; or applying it prospectively is no longer equitable; or
(6) any other reason that justifies relief.
4 See Mot.; United’s Brief in Opp’n to Defs.’ Rule 60(b) Mot. for Partial Recons. (Opp’n) [Dkt. 91]; Reply [Dkt. 97].
6 Fed. R. Civ. P. 60(b). The government specifically moves for relief under provisions (b)(2) and
(b)(6).
“In considering a Rule 60(b) motion, the district court ‘must strike a delicate
balance between the sanctity of final judgments . . . and the incessant command of a court’s
conscience that justice be done in light of all the facts.’” PETA v. HHS, 901 F.3d 343, 354-55
(D.C. Cir. 2018) (quoting Twelve John Does v. District of Columbia, 841 F.2d 1133, 1138 (D.C.
Cir. 1988)) (internal quotations omitted). To that end, a district court considering a Rule 60(b)
motion “is vested with a large measure of discretion.” Id. at 355. Notwithstanding, “[m]otions
for reconsideration are ‘disfavored,’” Walsh v. Hagee, 10 F. Supp. 3d 15, 18 (D.D.C. 2013)
(citation omitted), and the D.C. Circuit has cautioned that Rule 60(b) “should be only sparingly
used.” PETA, 901 F.3d at 355.
III. ANALYSIS
A. Rule 60(b)(2)
The government first argues that its Rule 60(b) motion is appropriate because the
FFS Adjuster Study constitutes “new evidence” that would have been relevant to the Court’s
decision. UnitedHealthcare responds that the data underlying the FFS Adjuster Study has been
in CMS’ possession for many years now and that the agency has not shown that with “reasonable
diligence” the study “could not have been discovered” before judgment. In turn, the government
does not contest the age of the data but asserts that the study is the culmination of an extended
review, that it is commonplace for agencies to review previous decisions, and further that the
agency is entitled to a presumption of regularity when performing such a review. See Allied
Mech. Servs., Inc. v. NLRB, 668 F.3d 758, 770-71 (D.C. Cir. 2012).
The government’s response misses the mark. Although CMS is entitled to a
presumption of regularity in its review of prior decisions, the development of facts central to this
7 litigation does not call merely for regularity—it calls for the exercise of “reasonable diligence.”
Compare Reasonable Diligence, Black’s Law Dictionary (11th ed. 2019) (“A fair degree of
diligence expected from someone of ordinary prudence under the circumstances like those at
issue.”), with Ordinary Diligence, Black’s Law Dictionary (11th ed. 2019) (“The diligence that a
person of average prudence would exercise in handling his or her own affairs.”). Here, CMS
was aware of actuarial criticisms of the Overpayment Rule when it first responded to comments.
See Overpayment Rule at 29,844. It was similarly aware that those criticisms were at the heart
of this lawsuit when the Complaint was filed in early 2016. See generally Compl. And it
remained aware of the importance of this issue through over two years of litigation in this Court.
By contrast, the underlying data sets CMS used for its FFS Adjuster Study were
developed in 2004, 2005, 2008, and 2011, respectively. See 2019 FFS Adjuster Study Rule at
30,983. The FFS Adjuster Study itself is only sixteen pages long. There is no indication in the
record or from the government that the timeline for completion of the FFS Adjuster Study was
informed in any way by its potential evidentiary value in this litigation. After CMS received
criticisms of the FFS Adjuster Study, it only took some four months for the agency to replicate
that study. See 2019 FFS Adjuster Study Rule. Given the years available to CMS, the Court
cannot conclude that the completion of the FFS Adjuster Study after the 11th hour is the result of
“reasonable diligence” under the circumstances. See In re Neurontin Mkg. & Sales Practices
Litig., 799 F. Supp. 2d 110, 114-15 (D. Mass. 2011) (holding a new meta-analysis of existing
scientific studies was not new evidence because the defendant “could have performed a similar
meta-analysis prior to the trial”); see also Good Luck Nursing Home, Inc. v. Harris, 636 F.2d
572, 577 (D.C. Cir. 1980) (“[A] party that . . . has not presented known facts helpful to its cause
when it had the chance cannot ordinarily avail itself on [R]ule 60(b) after an adverse judgment
8 has been handed down.”); cf. Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579, 596-97
(1993) (“Yet there are important differences between the quest for truth in the courtroom and the
quest for truth in the laboratory. Scientific conclusions are subject to perpetual revision. Law,
on the other hand, must resolve disputes finally and quickly.”). 5
B. Rule 60(b)(6)
The government argues that relief is nonetheless warranted under Rule 60(b)(6),
which permits relief from judgment for “any other reason that justifies relief.” Fed. R. Civ. P.
60(b)(6). But not just any reason will do; such relief is reserved for “extraordinary
circumstances.” Ackermann v. United States, 340 U.S. 193, 199 (1950). This is a weighty
burden that is best satisfied when “the interest that litigation must someday end [is] only slightly
impinged, while the countervailing interest that justice be done [is] seriously at stake.” Good
Luck Nursing Home, 636 F.2d at 577-78. For example, “[w]hen a party timely presents a
previously undisclosed fact so central to the litigation that it shows the initial judgment to have
been manifestly unjust, reconsideration under [R]ule 60(b)(6) is proper even though the original
failure to present that information was inexcusable.” Id. at 577 (emphasis added). And for the
reasons below, the Court finds these criteria are not satisfied.
1. Unjust Outcome
In response to the government’s motion, UnitedHealthcare has gone to great
lengths to explain why the conclusions of the FFS Adjuster Study are incorrect. For its part, the
government has done little to substantiate the findings of the FFS Adjuster Study to the Court.
Without getting too much into the weeds, UnitedHealthcare argues:
5 As the government suggests, provision (b)(2) is an odd fit with the administrative record and rulemaking process because there is no “evidence”; analysis under provision (b)(6) may be more appropriate. See Reply at 10-11.
9 First, that the FFS Adjuster Study answers the wrong question. The FFS Adjuster
Study concluded that “errors in FFS claims data do not have any systematic effect on the risk
scores calculated by the CMS-HCC risk adjustment model, and therefore do not have any
systematic effect on the payments made to [Medicare Advantage] organizations.” FFS Adjuster
Study at 5. But problems with the Overpayment Rule arise because it operates in two steps: (1)
payment to insurers; and (2) recoupment of overpayment by CMS. The Court determined that
the Overpayment Rule created a “crucial data mismatch” between the first step and the second.
UnitedHealthcare, 330 F. Supp. 3d at 187. That is, unaudited data was used to develop risk
coefficients for the first step, but audited data was used to determine when insurers had been
overpaid. The FFS Adjuster Study addresses only the effect of audited data on the development
of risk factor coefficients for payments, i.e., only the first step. It does not examine the effect of
using only audited data to determine overpayment amounts to Medicare Advantage insurers, i.e.,
the second step, and so does not speak to the “crucial data mismatch” identified by the Court.
Second, and more fundamentally, that the FFS Adjuster Study mixes audited and
unaudited data when analyzing payments to insurers and so actually negates its authors’
conclusions. Simplified, the CMS-HCC risk model also proceeds in two steps: (1) regression of
total Medicare Parts A and B expenditures for each beneficiary onto all risk factors, producing a
marginal dollar cost for each risk factor; and (2) normalization of those marginal costs against
the average beneficiary cost, producing a risk coefficient. While the FFS Adjuster Study used
audited data to generate the marginal dollar costs in step one, it then normalized those
coefficients against unaudited data. See CMS, Fee for Service Adjuster and Payment Recovery
for Contract Level Risk Adjustment Validation Audits - Technical Appendix at 13 (Oct. 26,
2018), available at https://tinyurl.com/rte2b6l (“In the next step, we take the new coefficients
10 and apply them on the original FFS data set.”). This was accomplished by mathematically
correcting—i.e., adjusting downwards—the results using audited data to conform to the results
using unaudited data. Without this correction, the coefficients over predict total Medicare costs,
which is exactly what one would expect if higher marginal dollar costs generated from only
audited diagnoses were applied to a beneficiary population that includes both audited and
unaudited diagnoses. But this correction plays the same role as the FFS Adjuster, only proving
the FFS Adjuster’s necessity in the payment scheme. See Opp’n, Ex. 7, Decl. of Julia Lambert
[Dkt. 91-7] ¶¶ 40-42.
The Court need not linger on the details of these arguments. On a motion to
reconsider, it is sufficient to say that the arguments are fully explained and the government does
not adequately respond. Indeed, the government asserts that it would be improper to “fully
address United’s criticisms outside of the rulemaking process.” Reply at 23. Instead, the
government offers the FFS Adjuster Study not “for the validity of its conclusions, which are still
tentative, but rather as evidence that the Court’s conclusions may not necessarily be accurate,
[and] to demonstrate the technical complexity of the questions that the study addresses.” Id. at
24.
In the regular course of rulemaking pending comments, the government is entitled
to withhold its final conclusions until its review process is completed. But the government
cannot be coy when it seeks extraordinary relief. Having already argued and lost its case after
two years of litigation and careful consideration by the Court, merely hinting at possible
inaccuracies and suggesting technical complexity is not enough to now convince the Court that
the interests of justice are “seriously at stake” or that the outcome was “manifestly unjust.” True,
this case is technically complex, but it did not somehow become more technically complex after
11 the Court’s decision than it was before. In the face of robust argument that the Court’s initial
decision was correct—which itself followed only after extensive briefing from both parties—the
government’s new arguments to the contrary must be both convincing and definitive.
2. Interest of Finality
On the other hand, the stakes for Plaintiffs are high. As the government itself
previously argued, merely vacating the Overpayment Rule without addressing the merits of the
CMS methodology “would provide plaintiffs no relief” because it “would not necessitate any
change to the Secretary’s risk adjustment methodology.” Defs.’ Mem. of P. & A. in Supp. of
Their Mot. to Dismiss for Lack of Subject Matter Jurisdiction [Dkt. 12-1] at 19. The government
motion for reconsideration demonstrates the problem: without the finality of a decision, the
government seems intent on re-litigating the Court’s findings.
C. Deference to the Regulatory Process
The government nonetheless counsels deference to the CMS administrative
process and asks the Court to give UnitedHealthcare “time . . . to ‘convince the agency to alter a
tentative position’” and provide the agency “‘an opportunity to correct its own mistakes and to
apply its expertise,’ potentially eliminating the need for (and costs of) judicial review.” Am.
Petroleum Inst. v. EPA, 683 F.3d 382, 387 (D.C. Cir. 2012) (quoting Pub. Citizen Health
Research Grp. v. FDA, 740 F.2d 21, 30-31 (D.C. Cir. 1984)). But the factors discussed in
American Petroleum Institute do not support reconsideration when applied to this case.
In American Petroleum Institute, EPA promulgated a final rule that exempted
some hazardous materials from regulation, but not others. The petitioners argued that the final
rule should have exempted a broader range of materials. During the pendency of litigation,
however, EPA backtracked and proposed a rule eliminating the exemption entirely which, if
adopted, would have mooted the petitioners’ claims. Alternatively, the comment process on the
12 proposed rule gave the petitioners another avenue to argue their case to the agency, before any
judgment by the court. Accordingly, the D.C. Circuit held the case in abeyance, reasoning that
“waiting to resolve this case allows EPA to apply its expertise and correct any errors, preserves
the integrity of the administrative process, and prevents piecemeal and unnecessary judicial
review.” Id. at 388.
Essentially all those facts cut in the opposite direction here. For one, CMS is no
longer writing on a blank slate: UnitedHealthcare had plenty of time to convince CMS and the
Court of its position; CMS had plenty of time to consider and finalize its interpretation; and
whatever the costs of judicial review, after two years of litigation, multiple rounds of briefing,
and three decisions by the Court, they have already been expended. For another, if the Court
modifies its decision and CMS adopts its proposed rule, the effect would be to expand the scope
of litigation, not contract it. That is, the proposed rule is not a “complete reversal of course” by
the agency that might otherwise end this litigation. Id. To the contrary, CMS is doubling down
on its position. Thus, instead of mooting UnitedHealthcare’s claims, CMS seeks to re-open a
matter which has already been decided. Further, this expansion would occur even if the Court
modifies its decision and CMS does not adopt the proposed rule. Under those circumstances, the
most that could be said is that the regulatory landscape would revert to the same condition as
before any of this litigation began, setting the parties up for another four years of conflict.
“Put simply, the doctrine of prudential ripeness ensures that Article III courts
make decisions only when they have to, and then, only once.” Id. at 387. The government does
not contest that this matter was properly ripe when it was litigated or when it was decided. The
Court carefully considered the matter and issued its decision, and that decision was crafted to
give practical, not merely nominal, relief to the prevailing party. Having lost, the government
13 now seeks to reset the process. But “an agency [cannot] stave off judicial review of a challenged
rule simply by initiating a new proposed rulemaking that would amend the rule in a significant
way.” Id. at 388. By that same token, an agency clearly cannot undo judicial review of a
challenged rule by initiating proposed rulemaking after an adverse decision has already been
handed down.
IV. CONCLUSION
For the reasons stated, the Court will deny the government’s Rule 60(b) Motion
for Partial Reconsideration, Dkt. 76. A memorializing Order accompanies this Memorandum
Opinion.
Date: January 27, 2020 ROSEMARY M. COLLYER United States District Judge