United States District Court, E.D. Virginia, Alexandria Division
ELLIS, III UNITED STATES DISTRICT JUDGE.
a Medicare-certified home health services provider, brings
this action against defendant, the Secretary of the United
States Department of Health and Human Services ("the
Secretary"), seeking reversal of a decision by the
Medicare Appeals Council ("MAC") that plaintiff had
been overpaid approximately $1 million for Medicare claims
submitted from 2008 to 2010. Specifically, plaintiff
challenges the MAC's determinations
(i) that ten claims submitted by plaintiff were not covered
(ii) that a valid statistical sampling methodology was used
to derive through extrapolation plaintiffs total overpayment
(iii) that plaintiff was not entitled to a waiver of
liability for the overpayment amount. In response, defendant
argues that the MAC's decision passes muster under the
deferential standard of review prescribed by the
Administrative Procedures Act ("APA") because each of
the MAC's conclusions applied the correct legal standards
and is amply supported in the record.
issue in this matter are the parties' cross-motions for
summary judgment, which have been fully briefed and argued
and are thus ripe for decision.
confines judicial review of agency decisions to the
administrative record of proceedings before the agency.
See 5 U.S.C. § 706; see also Camp v.
Pitts, 411 U.S. 138, 142 (1973). Put another way,
"when a party seeks review of agency action under the
APA, the district judge sits as an appellate tribunal."
Am. Bioscience, Inc. v. Thompson, 269 F.3d 1077,
1083 (D.C. Cir. 2001). Given the district court's limited
role in reviewing the administrative record, the ordinary
summary judgment standard does not apply. The key difference
in an APA case is that "the presence or absence of a
genuine dispute of material fact is not in issue, as the
facts are all set forth in the administrative record."
Hyatt v. U.S. Patent & Trademark Office, 146
F.Supp.3d 771, 780 (E.D. Va. 2015). Therefore, in a review of
agency action under the APA, "[t]he 'entire
case' on review is a question of law." Am.
Bioscience, Inc., 269 F.3d at 1083.
administrative record pertaining to plaintiffs administrative
appeal proceeding before the MAC reflects the following
• On March 11, 2010, AdvanceMed, a Centers for Medicare
and Medicaid Services ("CMS") contractor, opened an
investigation based on a complaint that plaintiff was (i)
admitting patients who did not qualify for home health
services and (ii) continuing to provide physical therapy to
patients even after those patients reached their maximum
level of potential. In addition, AdvanceMed conducted
preliminary data analysis that showed plaintiff ranked above
average in peer comparison of Medicare billing. After
determining that a full audit of all claims paid to plaintiff
would not be feasible, AdvanceMed conducted a statistical
sampling of the Medicare claims paid to plaintiff. In this
respect, AdvanceMed reviewed the services provided by
plaintiff to 30 randomly selected beneficiaries and then
extrapolated the overpayment determinations to estimate the
total amount plaintiff was overpaid by Medicare.
• AdvanceMed selected January 1, 2008 to June 30, 2010
as the period to be reviewed and used simple random sampling
as the sample design. AdvanceMed defined the universe as all
fully and partially paid claims submitted by the provider for
the period covered. It defined the sampling unit as
individual beneficiaries, with each unit identified by a
health insurance claim ("HIC") number. AdvanceMed
created the sampling frame by identifying those sampling
units from the universe where at least one line of service on
the claim was paid greater than $0 to the provider, and then
sorted the frame by HIC number. The frame included 1, 717
• AdvanceMed then used a random number-generator
software to draw a simple random sample of 30 beneficiaries.
• Fifteen of the 1, 717 HIC numbers used in the sampling
frame to identify the beneficiaries did not match the
beneficiaries' actual HIC numbers.
• On October 14, 2010, AdvanceMed requested
documentation from plaintiff to support 60 Medicare claims
plaintiff had submitted and received reimbursement for on
behalf of those 30 beneficiaries. AdvanceMed used the
documentation to conduct a medical review of those claims and
determined that plaintiff had been overpaid for 33 claims not
covered by Medicare. Using the lower limit of the 90%
two-sided confidence interval,  AdvanceMed then extrapolated the
sampled overpayment to conclude that plaintiff had received a
total overpayment of $2, 775, 432.
• On June 6, 2012, AdvanceMed provided Plaintiff with
documentation supporting its overpayment extrapolation,
including, inter alia, documentation of the sample
design, the universe of claims, the sampling frame, the
random numbers used, the random sample generated, and the
• After receiving notice of the overpayment assessment,
plaintiff sought redetermination of AdvanceMed's
determination that the 33 claims were not covered. The
resulting decision was partially favorable, as AdvanceMed
reversed 2 of the claim denials. Plaintiff then requested
reconsideration by a separate CMS contractor, which affirmed
all 31 of the claim denials.
• Next, plaintiff requested and received a hearing
before an Administrative Law Judge ("ALJ"). At the
hearing, plaintiff both challenged AdvanceMed's
individual claim denials and argued that AdvanceMed's
statistical sampling could not be replicated based on the
documentation provided by AdvanceMed. The ALJ rejected
plaintiffs challenge to AdvanceMed's statistical
sampling, but reversed certain claim denials.
• Finally, plaintiff requested review of the ALJ's
decision by the MAC, which constituted the final stage of
administrative review. On March 29, 2018, the MAC issued a
decision that reversed 3 claim denials by the ALJ but
affirmed the ALJ's conclusions with respect to the
validity of AdvanceMed's statistical sampling and the
remaining 15 claim denials.
standards of review that govern the district court's
review of the MAC's final decision are set forth in the
Medicare statute and the APA. First, the Medicare statute
provides that the MAC's factual findings must be upheld
"if supported by substantial evidence." 42 U.S.C.
§§ 405(g), 1395ff(b)(1)(A). The scope of review of
the MAC's factual findings under this standard is quite
limited. As the Supreme Court has explained, substantial
evidence "does not mean a large or considerable amount
of evidence, but rather 'such relevant evidence as a
reasonable mind might accept as adequate to support a
conclusion.'" Pierce v. Underwood, 487 U.S.
552, 565 (1988) (quoting Consolidated Edison Co. v.
NLRB, 305 U.S. 197, 229 (1938)). And importantly, the
district court may not "reweigh conflicting evidence,
make credibility determinations, or substitute [its] judgment
for that of the [MAC]." Johnson v. Barnhart,
434 F.3d 650, 653 (4th Cir. 2005); see Jarvis v.
Berryhill, 697 Fed.Appx. 251, 252 (4th Cir. 2017)
("The duty to resolve conflicts in the evidence rests
with the [agency], not with a reviewing court.").
pursuant to the APA, the MAC's decision may be set aside
only if it is "arbitrary, capricious, an abuse of
discretion, or otherwise not in accordance with law." 5
U.S.C. § 706(2)(A). The Fourth Circuit has made clear
that "[r]eview under this standard is highly
deferential, with a presumption in favor of finding the
agency action valid." Ohio Vail. Envt'l
Coalition v. Aracoma Coal Co., 556 F.3d 177, 192 (4th
Cir. 2009). In general, an agency decision will not be
considered arbitrary and capricious as long as "the
agency has examined the relevant data and provided an
explanation of its decision that includes 'a rational
connection between the facts found and the choice
made.'" Id. at 192-93 (quoting Motor
Veh. Mfrs. Ass'n v. State Farm Mut. Ins. Co., 463
U.S. 29, 43 (1983)).
applying these standards of review to the MAC's decision,
it is important to describe briefly the statutory and
regulatory framework that governs the MAC's determination
whether a healthcare provider was overpaid by Medicare.
is a federal health insurance program for the elderly and
disabled that is administered by the Secretary through CMS.
Among other things, the Medicare program reimburses providers
of certain medical and health services for the cost of
services that are covered by the Medicare Act. See
42 U.S.C. § 1395 et seq. To promote the
integrity of the Medicare program, the Secretary is
authorized to enter into contracts with private entities to
review claims for reimbursement submitted by providers, to
determine whether Medicare payments should not be, or should
not have been, made, and to recoup payments that should not
have been made. Id. § 1395ddd; 42 C.F.R. §
405.371(a)(3). In light of the substantial volume of Medicare
claims submitted by providers, Medicare contractors are
permitted to use statistical sampling and extrapolation to
determine the extent to which a provider was overpaid by
Medicare. See CMS Ruling 86-1 at 11 (Feb. 20, 1986).
to the Medicare Program Integrity Manual ("MPIM"),
a Medicare contractor must follow six steps to conduct
statistical sampling for overpayment calculation. MPIM Ch. 8
§ 220.127.116.11 (Pub. No. 100-08, Rev. 377) (2011). First, the
contractor must select the provider to be reviewed.
Id. Second, the contractor must select the period to
be reviewed. Id. §§18.104.22.168, 22.214.171.124.
Third, the contractor must define the universe, the sampling
unit, and the sampling frame. Id. § 126.96.36.199.
The "universe" consists of all Medicare claims
submitted by the provider during the period under review.
Id. § 188.8.131.52.1. The "sampling unit"
is the element that will be reviewed (e.g.,
individual claims or beneficiaries). Id. §
184.108.40.206.2. The "sampling frame" is the group of
sampling units that remain after any limiting criteria are
applied to the sampling universe. Id. §
220.127.116.11.3. Fourth, the contractor must choose a sampling
method and implement the method to select the sample.
Id. § 18.104.22.168. The sampling method that is used
must be classified as "probability sampling," that
is, (i) the method must be capable of selecting a set of
enumerable, distinct samples from the sampling frame and (ii)
each sampling unit must have a known probability of being
selected that is greater than zero. Fifth, the contractors must
review each unit in the selected sample and determine if an
overpayment has been made. Id. §§ 22.214.171.124,
126.96.36.199. Sixth, the contractor must estimate the total
overpayment to the provider during the review period by
extrapolating the results from the selected sample to the
entire sampling frame. Id. §§ 188.8.131.52,
provider may challenge a Medicare contractor's
calculation of overpayment through the administrative appeals
process. The use of statistical sampling by the
contractor "creates a presumption of validity as to the
amount of an overpayment." CMS Ruling 86-1 at 11. It is
the provider's burden to overcome this presumption by
demonstrating either (i) that the sample is not statistically
valid or (ii) that the contractor's determinations of
overpayment with respect to specific units in the selected
sample are incorrect. Id.
the provider may challenge the statistical validity of the
sample selected by the contractor. A challenge to the
validity of the sample "must be predicated on the actual
statistical validity of the sample as drawn and
conducted." MPIM § 184.108.40.206. Accordingly, "[i]f
a particular probability sample design is properly
executed" in accordance with the six steps set forth
above, "then assertions that the sample and its
resulting estimates are 'not statistically valid'
cannot legitimately be made." Id. § 8.4.2.
Put simply, "a probability sample and its results are
always 'valid.'" Id.
the provider may challenge the contractor's determination
that certain sampling units in the selected sample are not
covered by the Medicare Act and thus resulted in an
overpayment to the provider. In this respect, home health
services qualify for Medicare coverage if such services are
"reasonable and necessary" and are provided to a
beneficiary who is (i) confined to the home, (ii) under the
care of a physician who establishes a plan of care in
accordance with 42 C.F.R. § 409.43, and (iii) in need of
"skilled services" as certified by a physician. 42
U.S.C. §§ 1395f(a)(2)(C), 1395y(a)(1)(A); 42 C.F.R.
§ 409.42. A skilled service is one that is "so
inherently complex that it can be safely and effectively
performed only by, or under the supervision of, professional
or technical personnel." 42 C.F.R. § 409.32
first argues that the MAC'S decision should be set aside
because the MAC'S decision that AdvanceMed's sampling
methodology could be accurately replicated is arbitrary and
capricious, incorrectly applies the relevant legal standards,
and is not supported by substantial evidence in the record.
Notably, plaintiff does not challenge the MAC's
determination that the sampling methodology applied by
AdvanceMed was statistically valid. Rather, plaintiffs argument,
distilled to its essence, is that AdvanceMed's
extrapolated overpayment determination must be invalidated
because it is impossible to replicate the sample based on the
materials in the record.
addition to providing contractors with instructions on the
proper execution of statistical sampling for overpayment
calculation, the MPIM also requires Medicare contractors to
document the sampling methodology, the sampling universe and
frame, and the random number selection process that were used
to estimate overpayment. MPIM §§ 220.127.116.11, 18.104.22.168,
22.214.171.124.1, 126.96.36.199.3. The purpose of these documentation
requirements is to ensure that the sampling frame and the
sample can be replicated in the event that the methodology is
challenged. Id. §§ 188.8.131.52, 184.108.40.206.1. As
previous MAC decisions have concluded, failure to supply the
provider with sufficient documentation to recreate the
sampling frame and sample effectively deprives the provider
of its right to challenge the statistical validity of the
sample and thus may constitute a ground for invalidating the
overpayment extrapolation. See William Vecchioni,
D.C., M-13-3700 (H.H.S. Nov. 20, 2013); Global Home
Care, Inc., M-11-116 (H.H.S. Jan. 11, 2011);
Podiatric Medical Associates, M-10-230 (H.H.S. June
here, as the MAC concluded, plaintiff was provided with ample
documentation to enable plaintiff to replicate the sampling
frame and the sample. The record confirms that the
statistical sampling information CMS provided to plaintiff
included, inter alia, an electronic spreadsheet of
the frame used in the overpayment review; a memorandum
explaining the universe, sampling frame, sampling unit,
sample size, and sample design; the sample that was selected
from the frame; and the exact random numbers that were
generated and used to select the sample from the frame. And
it is undisputed that applying the random numbers provided by
CMS to the sampling frame provided by CMS would generate the
same sample as the one selected and recorded by AdvanceMed.