United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT
Argued December 15, 2017 Decided August 10, 2018
No. 17-5006
BILLINGS CLINIC, ET AL.,
APPELLANTS
v.
ALEX MICHAEL AZAR, II, SECRETARY, U.S. DEPARTMENT OF
HEALTH AND HUMAN SERVICES,
APPELLEE
Appeal from the United States District Court
for the District of Columbia
(No. 1:13-cv-00643)
Sven C. Collins argued the cause for appellants. With
him on the briefs was Stephen P. Nash.
Sydney Foster, Attorney, U.S. Department of Justice,
argued the cause for appellee. With her on the brief were
Jessie K. Liu, U.S. Attorney, and Michael S. Raab, Attorney.
Before: GARLAND, Chief Judge, and SRINIVASAN and
MILLETT, Circuit Judges.
Opinion for the Court filed by Circuit Judge MILLETT.
MILLETT, Circuit Judge: Several hospitals challenge the
2
methodology that the Department of Health and Human
Services used to calculate the “outlier payment” component of
their Medicare reimbursements for 2008, 2009, 2010, and
2011. Following this court’s decision in Banner Health v.
Price, 867 F.3d 1323 (D.C. Cir. 2017) (per curiam)—which
upheld the challenged methodology at its inception in 2007—
the primary question before us is whether the Department’s
decision to continue with its methodology after the 2007 fiscal
year was arbitrary in light of accumulating data about the
methodology’s generally sub-par performance. Because the
Department had, at best, only limited additional data for 2008
and 2009, and because the 2009 data suggested that hospitals
were paid more than expected, the Department’s decision to
wait a bit longer before reevaluating its complex predictive
model was reasonable.
On appeal, the Hospitals also challenge the Department’s
failure to publish a proposed, but later abandoned, draft rule
during the 2003 rulemaking process. As the parties now
acknowledge, Banner Health decided this issue in favor of the
Department. That prior circuit precedent controls here.
I
A
Congress first established Medicare in 1965 as part of the
Social Security Act, Pub. L. 89–97, Title XVIII, 79 Stat. 286,
291 (1965), as a “federally funded medical insurance program
for the elderly and disabled,” Fischer v. United States, 529 U.S.
667, 671 (2000). In its early years, Medicare paid its claims
much like most other insurance providers, reimbursing
hospitals for the “reasonable costs” of services provided to
Medicare patients. County of L.A. v. Shalala, 192 F.3d 1005,
1008 (D.C. Cir. 1999).
3
But over time, that system broke down. The “reasonable
cost” payment structure provided little incentive for hospitals
to husband their costs. The more they spent, the more they
would receive. County of L.A., 192 F.3d at 1008. So
healthcare costs rose, driving up the costs of the Medicare
program. See id.
In 1983, Congress confronted the problem of rising costs.
To better align the providers’ incentives, it constructed a new
“prospective” payment system that reimbursed hospitals based
on the average rate of “operating costs [for] inpatient hospital
services.” County of L.A., 192 F.3d at 1008. After adopting
this new scheme, the Department of Health and Human
Services began to reimburse hospitals “at a fixed amount per
patient, regardless of the actual operating costs they incur in
rendering [those] services.” Sebelius v. Auburn Reg’l Med.
Ctr., 568 U.S. 145, 149 (2013).
Generally speaking, this reimbursement system operates
as follows:
First, the Secretary of Health and Human Services
calculates a base payment rate. See 71 Fed. Reg. 47,870,
47,876 (Aug. 18, 2006) (codified at 42 C.F.R. §§ 412.308,
412.312). This rate contains both a labor and a non-labor cost
component. Id. The Secretary then adjusts the labor-related
component to account for labor costs in the area where the
hospital is located. Id.
Second, the Secretary develops a list of “diagnosis-related
groups.” 71 Fed. Reg. at 47,876. These groups encompass
numerous related medical diagnoses that the Secretary believes
impose a similar cost on the provider hospital. Id. To reflect
the average cost of treatment for patients in each diagnosis
group, the Secretary establishes a unique “relative weight” for
4
that group. Id.
Third, the base payment rate is multiplied by the relative
weight to create a generic payment amount for each diagnosis-
related group. 71 Fed. Reg. at 47,876. That is:
𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 × 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊ℎ𝑡𝑡 =
𝐺𝐺𝐺𝐺𝐺𝐺𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃
Fourth, qualifying hospitals can receive various payment
“add-ons.” For example, if a hospital treats a high proportion
of low-income patients, it receives a percentage increase in
Medicare payments known as the “disproportionate share
hospital (DSH) adjustment.” 71 Fed. Reg. at 47,876.
Likewise, if the hospital serves as an approved teaching
hospital, it can receive a percentage add-on payment, known as
the indirect medical education adjustment. Id. Hospitals
also can receive additional payments for cases involving the
use of new technologies. Id.
Fifth, even with those add-ons, Congress recognized that
healthcare providers would encounter patients with needs well
outside the norm. Country of L.A., 192 F.3d at 1009. To
account for those abnormally costly cases and to protect against
large financial losses for hospitals, the statute permits hospitals
to request additional “outlier payments.” See 42 U.S.C.
§ 1395ww(d)(5)(A)(ii). Hospitals may seek such payments
where “charges, adjusted to cost, * * * exceed the sum of the
applicable [diagnosis-related group] prospective payment rate
plus any amounts payable under [the payment adjustment
provisions] plus a fixed dollar amount determined by the
Secretary.” Id. In other words, hospitals are eligible for
outlier payments where:
5
𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎, 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑡𝑡𝑡𝑡 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 > (𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 +
𝐴𝐴𝐴𝐴𝐴𝐴 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 +
𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝑟𝑟 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 (𝑠𝑠𝑠𝑠𝑠𝑠 𝑏𝑏𝑏𝑏 𝑡𝑡ℎ𝑒𝑒 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆))
Any cost-adjusted charges above the applicable threshold
are eligible for outlier compensation. Charges below the
threshold are not. For that reason, the latter half of the above
formula—the generic prospective payment, adjustments, and
additional buffer (or “outlier threshold”)—is collectively
referred to as the “fixed-loss cost threshold.” 72 Fed. Reg.
47,130, 47,417 (Aug. 22, 2007) (codified at 42 C.F.R. pt. 412).
The Department can control the threshold (and thus the number
of cases eligible for outlier payments) by adjusting the
additional buffer amount up or down at the start of the year.
The first figure in the formula, “charges, adjusted to cost,”
represents the estimated cost of care for the patient at issue.
Since the Department will not know the hospital’s actual cost
of care at the time of payment, it can only estimate the
hospital’s costs using historical information about the
hospital’s past costs in relation to its prior charges. 1 The
1
Unlike other payments, outlier payments are typically made
based on the data available at the time the Department processes the
hospital’s claim. 68 Fed. Reg. 34,494, 34,500 (June 9, 2003)
(codified at 42 C.F.R. pt. 412). As noted below, due to systemic
abuse of this payment structure, the Department now retains the right
to adjust payments in certain cases through a process known as
“reconciliation,” which relies on later-acquired cost information.
Id. at 34,500–34,502. In the mine-run case, however, the
Department will not adjust the payment based on subsequently
obtained information. See 72 Fed. Reg. at 47,419. Instead, the
hospital will receive an outlier payment, if any, based on the
Department’s predicted cost of care for the patient in question given
the hospital’s actual charges for that patient.
6
Department estimates costs as follows:
𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶
𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎, 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑡𝑡𝑡𝑡 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 × 𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎
See 53 Fed. Reg. 38,476, 38,503 (Sept. 30, 1988).
The final piece of this formula—historical costs/historical
charges—is known as the hospital’s “cost-to-charge ratio.” It
reflects the percentage of that hospital’s charges attributable to
actual costs. To illustrate: If a hospital submits a bill for
$1,000, the Department will look to see whether the hospital’s
estimated costs (or, as the Department refers to them, “charges,
adjusted to cost”) exceed the fixed-loss cost threshold. To do
so, it will first need to know the costs embedded in that $1,000
charge. That is where the cost-to-charge ratio enters in. If
the hospital charged $500 for this procedure in prior years, and
its costs were $375, the hospital would have a cost-to-charge
ratio of $375/$500 or .75. Put another way, in the past, 75%
of the hospital’s charges reflected its costs of care. Knowing
that, the current costs of care can be estimated as follows:
$1,000 × .75 = $750
In this instance, the hospital’s “charges, adjusted to costs”
would be $750, and that number can be weighed against the
fixed-loss cost threshold for that patient’s diagnosis-related
group to determine whether the hospital should receive an
outlier payment.
Finally, the statute provides that the total outlier payment
for a given hospital “shall be determined by the Secretary and
shall * * * approximate the marginal cost of care beyond the
[applicable] cutoff point.” 42 U.S.C § 1395ww(d)(5)(A)(iii).
To implement this objective, the Department currently pays
80% of all costs above the applicable threshold. 42 C.F.R.
7
§ 412.84(k).
Continuing the previous example: If the fixed-loss cost
threshold was $500, but the estimated cost of a patient’s care
was $750, the hospital would be eligible for an outlier payment
of $200 (or 80% of $250, the amount falling above the $500
threshold). If, however, the threshold was $1,000, the hospital
would receive no payment at all.
B
Over the years, the Department has taken various
approaches to the cost-to-charge ratio data used to calculate the
cost-adjusted charges in this formula. Originally, it employed
a nationwide ratio for all hospitals. In the late 1980s, it shifted
to hospital-specific ratios that more accurately reflected the
costs at a given facility. 53 Fed. Reg. at 38,503, 38,507–
38,509 (codified at 42 C.F.R. pts. 405, 412, 413 & 419). But
even after making that change, the Department had difficulties
aligning its projections with actual costs of care. Prior to
2003, it used cost and charge data from the “latest available
settled cost report” without any forward projections. 68 Fed.
Reg. 34,494, 34,495 (June 9, 2003) (codified at 42 C.F.R. pt.
412).2 But cost reports take several years to settle. And that
time lag generated opportunities for abuse. Hospitals could
manipulate their outlier payments by inflating current charges
so that the historic cost-to-charge ratio employed to calculate
outlier payments did not reflect the hospital’s true costs. In
those situations, the hospital’s cost-to-charge ratio would
2
A “cost report” is an annual report submitted by each hospital
that details the hospital’s costs for treating Medicare patients during
the prior fiscal year. The Department uses these reports to
“determine total allowable inpatient Medicare costs” as well as to
calculate the hospital’s cost-to-charge ratio. 68 Fed. Reg. at 10,423.
8
overstate actual costs, resulting in an inflated cost estimate for
the current year’s claims. Id. at 34,496.
To use another example: If a hospital charged $2,000 for
a certain procedure ($1,500 of which reflected the hospital’s
costs), it would have a historic cost-to-charge ratio of 75%. If
the hospital wanted to increase its chances of obtaining an
outlier payment, it could simply increase the charge for that
procedure to $4,000. Using the now outdated cost-to-charge
ratio, the Department would calculate the hospital’s estimated
costs as $3,000 ($4,000 * .75), though, in reality, the costs were
likely much closer to the original $1,500.
To make matters worse, prior to 2003, if a hospital’s cost-
to-charge ratio fell outside a specified window, the Department
would substitute a statewide cost-to-charge ratio in lieu of the
hospital-specific ratio. 68 Fed. Reg. 10,420, 10,424 (March 5,
2003). That compounded the benefit of charge inflation
because hospitals could reap the rewards of inflated charges in
the short term without feeling the effects of a deflated cost-to-
charge ratio in subsequent payment periods.
Unsurprisingly, this approach led to rampant inflation in
hospital charges, a problem that came to be known as “turbo-
charging.” Banner Health, 867 F.3d at 1333. The
Department later identified 123 hospitals that had engaged in
the practice of turbo-charging starting in the 1990s. 68 Fed.
Reg. at 10,423.
In 2003, when the turbo-charging problem came to light,
the Department issued a regulation that addressed the problem
in two key ways. First, the Department sought to close the gap
between cost-to-charge ratios and current costs by using more
recent data—specifically by permitting the use of “either the
most recent settled or the most recent tentative settled cost
9
report, whichever is from the later cost reporting period.” 68
Fed. Reg. at 34,499. Second, the Department reserved the
right to recalculate a hospital’s eligibility using actual cost data
at the time of settlement. Id. at 34,501. Through this
process, known as reconciliation, the agency could claw-back
undue outlier payments. Id.
At that same time, the Department also considered
immediately adjusting the 2003 outlier threshold, which had
been set at the beginning of the year, to account for the effect
of the reforms on 2003 outlier payments. To that end, the
Department drafted a rule proposing to decrease the existing
outlier threshold for the remainder of the 2003 fiscal year.
The Department ultimately abandoned that effort, opting
instead to maintain the existing threshold until the year’s end
to allow rates to settle. 68 Fed. Reg. at 34,506. So the draft
rule was never published. The Hospitals later obtained a copy
of the draft through a Freedom of Information Act request.
C
To add complexity to the complexity, the Medicare
statute also limits the total amount of all outlier payments the
Department can make in a given fiscal year—setting both a
floor and a ceiling. Under the Act, the “total amount of”
outlier payments made in a fiscal year “may not be less than 5
percent nor more than 6 percent of the total payments projected
or estimated to be made based on [diagnosis-related group]
prospective payment rates for discharges in that year.” 42
U.S.C. § 1395ww(d)(5)(A)(iv). Of course, that requires the
Department to estimate certain numbers at the start of the year
before it has actual claims information. Because the statutory
target is tied to “projected or estimated,” not actual, payments,
id., the Department has interpreted the statutory directive to
mean that the fixed loss threshold must be set at a level that,
10
“when tested against historical data, will likely produce
aggregate outlier payments totaling between five and six
percent of projected or estimated [diagnosis-related group]
payments.” County of L.A., 192 F.3d at 1013 (emphasis
added); see e.g., 50 Fed. Reg. 35,646, 35,710 (Sept. 3, 1985).
For every year since 1989, the Department has aimed to set the
fixed-loss cost threshold so that the total outlier payments will
be 5.1% of all Medicare payments, or:
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 =
5.1% (𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃)
But, alas, this is a predictive enterprise. The Department
must set the outlier threshold at the start of each year before it
knows how many hospitals will actually have outlier patients.
In other words, the agency must estimate the number of outlier
cases for the upcoming year and set a threshold that it believes
will result in outlier payments of 5.1%.
As a result, to compute an appropriate outlier threshold,
the Department must estimate the total outlier costs for all
hospitals for the upcoming year. In practice, the Department
uses a formula similar to the one it uses to calculate actual
outlier payments, inputting projected and historical cost and
charge information to fill in the gaps. More specifically, the
agency takes historical charges and projects them forward to
reflect the upcoming year’s charges. The Department then
takes the historical cost-to-charge ratio from the most recent
year available and projects those figures forward to predict
current cost-to-charge ratios. 3 The basic formula is as
3
In reality, the agency calculates operating and capital ratios
separately. For simplicity, we treat this as a single-track
calculation.
11
follows:
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎, 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑡𝑡𝑡𝑡 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 ×
𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅
In 2007, the Department refined its methodology for
projecting historical cost-to-charge ratios forward to the
current year. 71 Fed. Reg. at 48,149. In the aftermath of the
2003 turbo-charging reforms, cost-to-charge ratios had fallen.
As a result, the Department’s efforts to hit its 5.1% target had
begun to land consistently short. Commenters, including
many hospital providers, asked the Department to adopt an
“adjustment factor” to update the cost-to-charge ratio formula
based on the current downward trend in hospital costs. Id. at
48,150. The Department agreed. It proposed the following
formula to estimate costs for purposes of determining the
outlier threshold:
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎, 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑡𝑡𝑡𝑡 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 =
(𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 × 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹) ×
𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹
� × �
𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹
See id. at 48,149.
This refined formula relied upon two critical metrics to
predict future costs and charges from the available historical
information: the cost inflation factor and the charge inflation
factor.
The latter was relatively simple. The agency would
update historical charge data using the average annualized rate
of change in charges per case. In other words:
𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 =
12
𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑜𝑜𝑜𝑜 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑖𝑖𝑖𝑖 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑝𝑝𝑝𝑝𝑝𝑝 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶
71 Fed. Reg. at 48,149.
The cost inflation factor, however, was more complex. It
factored in both hospital-specific cost inflation and general
inflation as measured by the change in a standard market basket
of goods and services. At the highest level, the formula is as
follows:
Cost Inflation Factor =
(Average Annual Hospital Cost Inflation for three years prior ×
Annual Inflation of Market Basket for the most recent year available)
And within this formula:
𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 =
𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑝𝑝𝑝𝑝𝑝𝑝 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎
𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼
To put this all together in a more concrete setting, the
Department would calculate the cost inflation factor for
purposes of the 2008 threshold using historical data as follows:
2008 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐼𝐼𝐼𝐼𝐼𝐼𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 =
13
2004 𝑡𝑡𝑡𝑡 2005 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑝𝑝𝑝𝑝𝑝𝑝 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎
2005 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 +
⎛⎛ ⎞⎞
⎜⎜2003 𝑡𝑡𝑡𝑡 2004 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑝𝑝𝑝𝑝𝑝𝑝 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 + ⎟⎟
⎜⎜ 2004 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 ⎟⎟
⎜⎜ ⎟⎟
⎜ 2002 𝑡𝑡𝑡𝑡 2003 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑖𝑖𝑖𝑖 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑝𝑝𝑝𝑝𝑝𝑝 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 ⎟
⎜⎝ 2003 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 ⎠⎟
⎜ ⎟
⎜ ⎟
⎜ ⎟
⎜ 3 ⎟
⎜ ⎟
⎝ ⎠
× 2006 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼
See 71 Fed. Reg. at 48,150 (outlining this process for the 2007
threshold).
Once the agency calculates both the cost inflation and
charge inflation factors, it then divides the cost inflation factor
by the charge inflation factor to obtain the “adjustment factor.”
This adjustment factor is then multiplied by the historical cost-
to-charge ratio to obtain an updated, or projected, cost-to-
charge ratio for that year.
𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑛𝑛 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹
×
𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝐶𝐶ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹
= 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅
D
Many providers supported the downward adjustment in
the cost-to-charge ratios, but not the “magnitude of that
adjustment.” Banner Health, 867 F.3d at 1355. Several
hospitals then challenged the 2007 outlier payment
methodology. This court rejected those claims, holding that
the Department’s new formula was not arbitrary or capricious
14
in light of the information available to the Department. Id. at
1355–1356 (finding “many non-arbitrary reasons for
predicting that costs and charges * * * will not continue on their
current trajectories”).
The following year, the Department used the same
methodology to calculate the 2008 outlier threshold. See 72
Fed. Reg. at 47,417. The result was a proposed fixed loss
threshold of the prospective payment rate plus any cost
adjustments plus $23,015. Id.
Several commenters thought that the buffer amount was
too high. They noted that outlier payments had been, by their
calculation, only 4.63% of overall 2007 payments. They
urged the Department to adopt a simplified cost inflation factor
based on the actual rate of change in hospital costs—the same
method already being used to estimate projected charges.
Commenters also faulted the agency for not using more recent
cost-to-charge ratio data, and suggested applying the cost-to-
charge ratio adjustment factor over different periods of time
(longer or shorter than one year) based on individual hospital’s
fiscal calendars. Others simply urged the Department to
lower the threshold without providing an alternative approach.
The Department rejected all of those proposals. 72 Fed.
Reg. at 47,418. By the time of the final rule, the Department
estimated that outlier payments for 2007 had been 4.6% of
overall payments. Id. at 47,420. That fell far short of the
5.1% payment target. But because cost reports had not been
settled for 2007, 4.6% was still just an estimate of what the
actual 2007 percentage would be. Id. (“This estimate is based
on simulations[.]”). For that reason, the Department declined
to alter its methodology, reasoning that its chosen cost inflation
factor was “more accurate and stable than the commenter’s * *
* because it takes into account the costs per discharge and the
15
market basket percentage increase[.]” Id. at 47,418. As it
had in prior years, the Department also determined that its
methodology would not account for any potential payment
claw-backs made during the reconciliation process because
those would be too “difficult to predict” and because the
amounts are generally too small to materially affect the
predictive assumptions. Id. at 47,419. The Department,
however, did update the data for its final rule, a change that
resulted in a lower buffer amount of $22,635. Id.
That process was largely repeated in 2009. See 73 Fed.
Reg. at 48,763. The Department proposed a fixed loss
threshold of the prospective payment rate plus any cost
adjustments plus $21,025, and ended with an updated, final
buffer of $20,185. 73 Fed. Reg. 23,528, 23,711 (April 30,
2008); 73 Fed. Reg. at 48,766.
Once more, commenters challenged the cost inflation
factor, calling it “unnecessarily complicated.” 73 Fed. Reg. at
48,764. They again urged the agency (i) to use more recent,
historical, and industry-wide rates of change; (ii) to vary the
cost-to-charge ratio adjustment factor to more or less than one
year; and (iii) to apply more recent hospital data to its
calculations. Id.
As before, the Department rejected the suggestions,
largely reiterating the reasons it had provided in 2008. See 73
Fed. Reg. at 48,763. The Department now estimated 2007
outlier payments as 4.6% of final payments, and it projected
2008 payments to be 4.7%. Despite widely missing the 5.1%
mark again, the Department maintained its approach to cost
inflation, asserting that the use of “the market basket in
conjunction with the cost per discharge takes into account two
sources [of] potential cost inflation and ensures a more accurate
and stable cost adjustment factor.” Id. at 48,764.
16
For 2010, the agency again employed the same formula.
74 Fed. Reg. at 44,007. Using that methodology, the agency
proposed a fixed loss threshold of the prospective payment rate
plus any adjustments plus $24,240, a 21% increase from the
previous fiscal year. With updated data, the Department later
arrived at a final buffer of $23,140.
This time, the Department’s estimates appeared slightly
more promising. By the final rulemaking, the Department
estimated that 2008 outlier payments had been 4.8% of final
payments, but 2009 outlier payments had been 5.4% of final
payments—meeting and even exceeding the 5.1% target.
Nevertheless, the proposed increase in the buffer amount
prompted renewed protest. See 74 Fed. Reg. at 44,007.
Commenters could not understand why—when the agency had
met its target in 2009—there should be any change to the
threshold amount. Id. at 44,009. Others accused the agency
of purposefully erring on the low end of the 5% to 6% target.
Id. Another asked the agency to make a mid-year adjustment
if it appeared that the existing threshold would not result in
payments in the 5% to 6% range. Id. Still others repeated the
requests to use more recent data in the final rule and to account
for reconciliation. Id. at 44,009–44,010.
For its part, the Department insisted that it had “use[d] the
most recent data available to set the outlier threshold.” 74
Fed. Reg. at 44,009. It rejected the mid-year course correction
because such adjustments would be “extremely difficult or
impracticable (if not impossible) to administer.” Id.
(incorporating 70 Fed. Reg. 47,278, 47,495 (Aug. 12, 2005),
the Department’s response to a similar request made in the
2006 rate proposal). And it denied all other suggestions for
the same reasons given in prior rulemakings. See id. at
44,010.
17
Fiscal year 2011—the last at issue in this case—proved no
different. See 75 Fed. Reg. 50,042, 50,427 (Aug. 16, 2010).
The Department proposed a buffer of $23,970. 75 Fed. Reg.
at 24,069. And it ultimately settled on $23,075 as the final
amount. 75 Fed. Reg. at 50,430. The complaints from
hospitals and the answers from the Department were repeated.
In addition, using updated data, the Department calculated
2009 outlier payments to be 5.3% of final payments, and 2010
outlier payments to be 4.7% of final payments. Commenters
took issue with the Department’s 2009 estimate, arguing that
its calculations indicated outlier payments of only 4.9% for the
2009 year.
The following chart summarizes the key data points from
each rulemaking:
18
2008 2009 2010 2011
Proposed Fixed- $23,015 $21,025 $24,240 $23,075
Loss Threshold
Final Fixed-Loss $22,635 $20,185 $23,140 $23,075
Threshold
Agency’s Target 5.1% 5.1% 5.1% 5.1%
Agency’s 4.8% 5.3% 4.7% 4.8%
Reported
Estimate
Hospitals’ 4.6% 4.9% N/A N/A
Estimate
Shortfall -0.3% +0.2% -0.4% -0.3%
(Agency
Estimate )
Shortfall -0.5% -0.2% N/A N/A
(Hospital
Reported
Estimate)
E
There is yet one final piece of this byzantine process that
bears a bit of explanation. When a hospital seeks Medicare
payments from the Department, it must first submit its request
to a fiscal intermediary—that is, a contracted entity to which
the Department has delegated payment determinations. See
42 U.S.C. §§ 1395kk-1, 1395oo(a). If the hospital is
unsatisfied with the intermediary’s final determination, it may
appeal the decision to the Provider Reimbursement Review
Board. Id. In the normal course, the Board would review the
claim, and the hospital would retain the right to seek “judicial
review of any final decision of the Board.” 42 U.S.C.
§ 1395oo(f)(1).
However, if the hospital’s claim “involves a question of
19
law or regulations relevant to the matters in controversy * * *
[that the Board] is without authority to decide,” the hospital can
ask the Board to allow it go directly to district court. 42
U.S.C. § 1395oo(f)(1); 42 C.F.R. § 405.1842. If the Board
agrees, it will certify the question for immediate judicial
review. Id. Only the hospital can challenge the Board’s
certification determination. See Allina Health Servs. v. Price,
863 F.3d 937, 941–942 (D.C. Cir. 2017).
F
Several acute care hospitals have challenged the outlier
payments received in 2008, 2009, 2010, and 2011. They
allege that the Department’s methodology for determining the
outlier threshold during this period was arbitrary and
capricious, pointing in particular to the consistent
underpayments in each of the relevant years and the failure to
account for the possibility of reconciliation. They also object
to the Department’s failure to publish the proposed but
ultimately not-adopted 2003 draft rule, which, in their view,
contains much-needed ammunition to show that the
Department should have updated its methodology in these later
years.
Because the Hospitals challenged the legality of the
applicable outlier thresholds, they requested expedited access
to judicial review from the Board. The Board granted many
of those certification requests, either initially or on
reconsideration. With respect to some Hospitals, however,
the Board concluded that it lacked jurisdiction to grant
expedited review because those Hospitals had failed to comply
with the required procedures for filing their cost reports. See
e.g., Ex. 1 PRRB Decisions, Lee Mem’l Hosp. v. Sebelius, No.
1:13-cv-00643 (D.D.C. Jan. 10, 2014), ECF No. 22-1 (citing
42 C.F.R. § 405.1835(a)(1)(ii) (2013)). For that subset of
20
Hospitals, the Board dismissed the cases for lack of jurisdiction
without deciding the certification question.
Both the dismissed and certified Hospitals filed suit in
district court. The government conceded that the Board erred
in dismissing some of the cases for lack of jurisdiction. In
light of that concession, the district court held that the Board
had jurisdiction, and that the court could likewise exercise its
own jurisdiction under 42 U.S.C. § 1395oo(f)(1).
The district court subsequently granted summary
judgment for the Department, concluding that its approach to
calculating the outlier threshold was not arbitrary or capricious.
The Hospitals moved for reconsideration, and the district court
denied the motion. The Hospitals now appeal.
II
A
We start, as we must, with jurisdiction. The Medicare
Act specifies that “[n]o findings of fact or decision of the
[Secretary] shall be reviewed by any person, tribunal, or
governmental agency” except as the Medicare Act itself
provides jurisdiction. 42 U.S.C. § 405(h). The relevant
source of jurisdiction in this case is 42 U.S.C. § 1395oo(f).
That provision allows providers to seek review of a final
decision of the Provider Reimbursement Review Board and to
seek expedited judicial review where the Board lacks
“authority to decide” a question of law relevant to the matter at
hand. Id.
As noted, for the majority of the plaintiff Hospitals, the
Board granted expedited review on the ground that it lacked
authority to override the outlier regulations. The district court
21
properly exercised jurisdiction over those claims. See Allina
Health Servs., 863 F.3d at 941–942.
As for the plaintiff Hospitals over which the Board
declined to exercise jurisdiction, the question is more
complicated. While the Secretary has since disavowed the
Board’s procedural objection to the claims in that case, that
leaves unanswered whether the district court could proceed
without first remanding for either a final decision or
certification for expedited review from the Board.
We need not resolve that jurisdictional quandary because
there are Hospitals with valid Board certifications of expedited
review for each of the years at issue, and only non-
individualized injunctive relief is sought. We accordingly
proceed to the merits on a clean jurisdictional slate.
B
Under the Administrative Procedure Act, we may only set
aside agency action if it is “arbitrary, capricious, an abuse of
discretion, or otherwise not in accordance with law.” 5 U.S.C.
§ 706(2)(A). In making that assessment, we must ensure that
the agency has “examine[d] the relevant data and articulate[d]
a satisfactory explanation for its action including a rational
connection between the facts found and the choices made.”
Motor Vehicle Mfrs. Ass’n v. State Farm Mutual Auto. Ins. Co.,
463 U.S. 29, 43 (1983) (citation and internal quotation marks
omitted).
The Hospitals argue that the cost-projection methodology
used by the Department to set the annual outlier thresholds is
arbitrary and capricious for three reasons. First, they object to
the Department’s failure to publish the 2003 draft rule, which
they allege deprived them of useful information in the
22
subsequent rulemakings. Second, they challenge the
Department’s failure to account for the possibility of
reconciliation claw-backs in setting the 2008, 2009, 2010, and
2011 thresholds. Third, they object to the Department’s
continued use of its cost-inflation methodology in the face of
repeated underpayments and the availability of a simpler
formula, which the Department was already using to calculate
inflation in hospital charges.
The first two challenges are foreclosed by circuit
precedent. See Banner Health, 867 F.3d at 1337, 1356. As
for the third, while the Hospitals’ frustration with the
Department’s frequently off-target calculations is
understandable, the methodology has not sunk to the level of
arbitrary or capricious agency action.
1
Our decision in Banner Health, which involved a similar
challenge to the 1997 through 2007 outlier payment rates,
disposes of the Hospitals’ procedural challenge regarding the
Department’s unpublished draft rule. There, this court held
that the Department did not err in failing to disclose the 2003
draft rule because it had not relied on it in crafting its final rule.
Banner Health, 867 F.3d at 1337.4
4
Because Banner Health controls disposition of this claim, we
need not address the Department’s alternative standing argument.
See Ruhrgas AG v. Marathon Oil Co., 526 U.S. 574, 584–585 (1999)
(“It is hardly novel for a federal court to choose among threshold
grounds for denying audience to a case on the merits.”). In any
event, the Department’s concerns speak more to the merits of the
Hospitals’ ability to obtain their desired relief with the aid of the
2003 draft rule—a question not relevant to a threshold standing
inquiry.
23
Banner Health also largely answers the Hospitals’
argument that the Department had to factor reconciliation
claw-backs into its threshold predictions. Banner Health
rejected that exact same challenge to the 2005 outlier
thresholds. 867 F.3d at 1351–1352, 1356. We held that the
Department “was under no obligation” to “account for the
possibility of reconciliation in setting the fixed-loss threshold.”
Id. at 1356; see also District Hosp. Partners, L.P. v. Burwell,
786 F.3d 46, 61 (D.C. Cir. 2015) (concluding that the 2003
reforms “corrected the flaw in the outlier payment system that
created the opportunity—and incentive—to turbo-charge,” and
thus the need for large reconciliations).
That conclusion applies with equal force to the 2008
through 2011 outlier thresholds. As in Banner Health, the
Department reasonably concluded “that [the] charging
practices would not fluctuate significantly enough to justify
accounting for reconciliation[.]” 867 F.3d at 1352
(alterations, citation, and internal quotation marks omitted);
see e.g., 72 Fed. Reg. at 47,419 (rejecting the need to consider
reconciliation for the 2008 threshold for these same reasons).
Nothing in the current record supports a different answer here.
The Office of the Inspector General’s 2012 Report, to which
the Hospitals direct us, does not indicate that these payments
occur with more regularity now than the Department suggested
at the time of Banner Health.
Finally, Banner Health sanctioned the agency’s 2007
methodology for calculating cost inflation, at least to the extent
that the Hospitals challenge its facial validity. See 867 F.3d at
1356 (“[T]he Hospitals have provided no reason to doubt that
the market basket percentage increase correlated reasonably
well with cost-per-case inflation.”). On top of that, just as
Banner Health recognized, there is good reason to believe that
a model that accounts for macro-level change may better
24
predict future costs than one that does not. Id.
The Hospitals’ proposal of a simpler method does not
make the Department’s method arbitrary. “A model’s
complexity, by itself, reveals little about its rationality.”
Banner Health, 867 F.3d at 1356. In the wake of the turbo-
charging scandal and persistent cost containment problems, it
was not unreasonable for the Department to be wary of an
industry-specific inflation metric. Plus, as the Department
explained, a method accounting for general inflation “is more
accurate and stable” than the industry-specific alternative. 72
Fed. Reg. at 47,418.5
2
That leaves one final question: Did the Department act
arbitrarily in maintaining its cost inflation methodology after
the 2007 fiscal year in the face of numerous underpayments?
In light of the short pattern of missed targets, the limited and
inconsistent data available to the Department at the time of its
rulemakings, and the lengthy time lag in finally determining
the actual payments made for a preceding year, we conclude
that it did not.
To be sure, a methodology used for prediction “can look
5
The Hospitals also argue that the Department acted arbitrarily
in treating under- and over-payments differently. Specifically, they
claim that the agency responded to underpayments that fell short of
the 5.1% target with indifference, but promptly increased the outlier
threshold in 2010 after what it believed was an overshoot of the 5.1%
target. That argument misunderstands what happened in 2010.
The Department did not change its methodology after it hit what it
believed to be a 5.3% mark. It employed the same methodology it
had been using since 2007. That methodology simply produced a
higher threshold for the 2010 fiscal year.
25
more arbitrary the longer it is applied.” American Petroleum
Inst. v. EPA, 706 F.3d 474, 477 (D.C. Cir. 2013). But that line
was not crossed for the years at issue here. In 2008, the
Department had only one year’s worth of yet unsettled data to
test the validity of its new model. By the time the Department
had to develop its 2008 outlier payment amount, it had not
definitively settled on the 2007 results, a cost-resolution
process that can take several years. The tentative 2007 results
were not so far off base as to suggest the need for immediate
abandonment of the newly adopted system. Even by 2009, the
Department had only a slightly more settled 2007 estimate and
a tentative number for 2008. The time-lag inherent in
obtaining accurate payment data underscores the
reasonableness of the Department’s deliberative and cautious
approach to evaluating the operation and accuracy of its
methodology.
Putting some proof in the pudding, the Department’s
calculations indicated that it not only met, but exceeded, the
5.1% mark in 2009, resulting in payments to the Hospitals that
exceeded the Department’s 5.1% target. Even considering the
Hospitals’ contrary estimation of 4.9%, the 2009 payments
nearly reached the Department’s intended target. For that
reason, we cannot conclude that the Department acted
arbitrarily in continuing to employ its predictive model for the
next two years while accumulating additional data points.
*****
For all its complexity and labyrinthine mathematical
formulae, this case turns on a simple concept: Some things
take a bit of time to sort out. The Department’s efforts to
predict Medicare costs for patients across the Nation each fiscal
year is fraught with variables, estimates, and uncertainties.
The Medicare statute recognizes that difficulty by requiring the
26
Department to model results that fall between 5% and 6% of
total projected payments, without mandating that the
Department actually hit the bullseye each year. See 42 U.S.C.
§ 1395ww(d)(5)(A)(iv). Though the Department has an
obligation to act reasonably and to account for the actual results
of its decisions, the need for time both to study the results and
to determine how to improve accuracy must inform any
evaluation of the appropriateness of the Department’s actions
in these years.
For those reasons, we affirm the judgment of the district
court.
So ordered.