New York State Dept. of Social Services, DAB No. 1134 (1990)

DEPARTMENTAL APPEALS BOARD

Department of Health and Human Services

SUBJECT: New York State Dept.

DATE: February 15, 1990
of Social Services Docket Nos. 89-51, 89-86, and 89-87 Decision No.1134

DECISION

The New York State Department of Social Services (State) appealed
determinations by the Health Care Financing Administration (HCFA)
disallowing federal financial participation (FFP) claimed under title
XIX of the Social Security Act (Act). The State claimed FFP in payments
which it determined were made on behalf of individuals eligible for
Medicaid on the basis of age (under 21), pregnancy, or eligibility for
benefits under title IV-A of the Act (AFDC). 1/ The payments were
originally charged solely to the State-funded medical assistance (MA)
program. The State later took a statistical sample of MA cases and
found that some individuals had been eligible for Medicaid during the
periods in question. Based on the extrapolation of the results of the
sample to the universe of MA payments, the State made retroactive claims
for the amounts in question here. 2/ HCFA disallowed the claims on the
ground that they were impermissibly based on a statistical sample.

The Board issued a Preliminary Analysis in these cases which tentatively
concluded that HCFA had not shown a valid basis for prohibiting these
particular claims solely because they were developed using statistical
sampling procedures. We have carefully considered the parties'
responses to our Analysis. For the reasons set forth below, we adopt
our preliminary conclusion. However, as the parties agreed, this
resolves only the threshold issue in this case. Accordingly, we remand
to HCFA for further consideration, as HCFA deems appropriate, of
additional issues presented by the appeals but not reached here (such as
whether the statistical methodology used was valid and other matters
discussed below).

Summary of Decision

The following points are developed in our decision:

- There is no regulation which prohibits claims based on
statistical sampling.

- The prohibition on sampling in HCFA's State Medicaid Manual does
not have the force and effect of law. Accordingly, it must be found
reasonable in order to bind the State.

- The State's use of sampling to establish the amount of its
Medicaid claims, which HCFA has prohibited, parallels HCFA's reliance on
statistical sampling to establish the amounts of disallowances of
Medicaid claims, and calls into question the reasonableness of HCFA's
prohibition.

- Sampling was not used here as a substitute for the State's basic
eligibility determination process in which a determination as to each
applicant's eligibility for Medicaid or MA is made.

- There is no evidence that sampling substantially impinges on
HCFA's ability to audit the State's claims or otherwise ensure that the
claims covered payments on behalf of Medicaid eligibles for covered
services and are not duplicative of other claims. (HCFA is not,
however, required to actively audit the claims in order to determine
whether there has been duplication, but may require the State to show
that there has been no duplication.)

- The State had a statutory right to claim FFP in payments on
behalf of Medicaid eligibles, and was not precluded from using sampling
to make such claims simply because it may not have effectively utilized
other means of identifying payments for Medicaid eligibles.

- The prohibition on sampling would force the State to identify
payments for Medicaid eligibles through case- by-case reviews, thus
substantially increasing Medicaid costs to both the State and HCFA
without increasing the accuracy of the State's claims or serving other
program purposes discernable from HCFA's argument here.

- This case is unlike New York State Dept. of Social Services, DAB
No. 1012 (1989), in which the Board held that HCFA's prohibition on
sampling was reasonable where sampling circumvented specific
requirements for documenting Medicaid claims based on disability. HCFA
has not shown that similar requirements are applicable to the
non-disability-based claims in question here.

- While the Board finds the prohibition on sampling unreasonable as
applied to the types of claims in question here, the costs may not be
allowable if the State does not establish that it correctly reclassified
the individuals in the sample cases as Medicaid eligibles who received
services covered by Medicaid and that its overall sampling and
statistical methodology was valid.

Background

The expenditures in question were made for individuals originally
determined to be ineligible for Medicaid but eligible for the
State-funded MA program. MA is available for individuals who meet
Medicaid financial eligibility requirements but do not fall into any of
the categories of Medicaid eligibles. Some individuals who were
eligible for Medicaid on the basis of age (under 21), pregnancy, or
eligibility for AFDC were nevertheless classified as MA due to
caseworker or clerical error or because evidence of Medicaid eligibility
was not received until after the eligibility determinations were made.
These individuals were later reclassified as Medicaid- eligible based on
an audit performed by the State's Office of Audit and Quality Control.
The auditors selected a sample of MA cases which they reviewed in order
to identify payments to individuals who were in fact eligible for
Medicaid at the time the eligibility determinations were made. The
sample findings were extrapolated to a defined universe of MA payments
to determine the amount for which title XIX funding should have been
claimed. The State reduced the gross expenditure amount by amounts
claimed through another State project which involved the
reclassification of MA expenditures to Medicaid, the Shares
Reclassification System, in order to avoid duplicate claims for the same
expenditures. The claims based on statistical sampling were filed
within two years of the date the expenditures were incurred, as required
by section 1132 of the Act.

Whether the Disallowance Is Authorized by Regulations Prohibiting Claims
Based on Estimates

HCFA took the position that the express prohibition on the use of
estimates in 42 C.F.R. 430.30(c)(2) applied to prohibit those claims
based on statistical sampling which were filed after its effective date
(October 21, 1988) because such claims were in fact estimates. Section
430.30(c) is captioned "Expenditure reports," and provides as follows:

(1) The State must submit Form HCFA-64 (Quarterly Medicaid
Statement of Expenditures for the Medical Assistance Program) to
the central office (with a copy to the regional office) not later
than 30 days after the end of each quarter

(2) This report is the State's accounting of actual recorded
expenditures. The disposition of Federal funds may not be reported
on the basis of estimates.

(Emphasis added.) HCFA asserted that the underscored language applied
to prohibit the claims in question here since one definition of the term
"estimate" is "a numerical value obtained from a statistical sample and
assigned to a population parameter." Webster's Seventh New Collegiate
Dictionary 284 (1967). HCFA also noted that the statistician who helped
the State to develop its claims specifically referred to the figures
arrived at through his procedures as estimates.

We are not persuaded that the term "estimate" is used here as HCFA
argued, however. Since the regulation implements the statute, the
regulation should use the term "estimate" in the same sense as the
statute. Section 1903(d)(1) of the Act provides that--

Prior to the beginning of each quarter, the Secretary will estimate
the amount to which a State will be entitled . . . for such
quarter, such estimates to be based on (A) a report filed by the
state containing its estimate of the total sum to be expended in
such quarter. . . .

This provision obviously uses the term "estimate" to mean what a state
expects to expend during a quarter when it draws down funds at the
beginning of the quarter. It does not refer to a scientifically valid
statistical extrapolation of what was actually expended, as the term
"estimate" is used in statistical parlance.

Moreover, HCFA's interpretation of the prohibition on estimates fails to
take account of the context in which it appears in the regulation. The
prior sentence indicates that a state's claim is an accounting of actual
expenditures. Thus, a better reading of the sentence in question may be
that claims shall not be based on other than actual expenditures. (This
simply turns the characterization of a claim in the prior sentence
around for emphasis.) The claims in question here represent a
statistical estimate of the amount of actual MA payments which were
incurred on behalf of individuals who were eligible for Medicaid. Since
the claims were thus "based on" actual expenditures, we conclude that
the State complied with the regulation here.

Additional support for this conclusion comes from the fact that the
preamble to 42 C.F.R. Part 430 states that the regulation was
promulgated to "update and simplify" 45 C.F.R. Part 201 and not to make
substantive changes. 53 Fed. Reg. 36569, 36570 (September 21, 1988).
The portion of Part 201 from which section 430.30(c) derives provided:

The State agency must also submit a quarterly statement of
expenditures for each of the public assistance programs under the
Act. This is an accounting statement of the disposition of the
Federal funds granted for past periods and provides the basis for
making the adjustments necessary when the State's estimate for any
prior quarter was greater or less than the amount the State
actually expended in that quarter. . . .

45 C.F.R. 201.5(a)(3) (1986). HCFA pointed to nothing in the language
of this regulation which could be interpreted to prohibit statistical
sampling (although HCFA argued that the regulation had this effect). 3/
Since section 430.30(c) made no substantive change in this regulation,
there is no basis for reading such a prohibition into section 430.30(c).

Whether HCFA Properly Relied on the State Medicaid Manual as a Basis for
the Disallowance

HCFA also relied on Transmittal No. 30 of the State Medicaid Manual,
issued by HCFA in December 1984, as a basis for the disallowance. 4/
The transmittal provides, in pertinent part, that "[c]laims developed
through the use of sampling, projections, or other estimating techniques
are considered estimates and are not allowable." There is no dispute
that this language would prohibit FFP in the type of claims submitted by
the State here. It is also undisputed that the State had notice of the
transmittal not only before it filed the claims but also before the
expenditures were incurred. HCFA agreed, however, that the prohibition
on sampling was binding only if it "is based on a reasonable
interpretation of the regulations." HCFA's 6/1/89 brief, p. 12. 5/
This is consistent with both Board and court holdings that an
interpretative rule, unlike a regulation published in accordance with
the Administrative Procedure Act, 5 U.S.C. 551 et seq., does not have
the force and effect of law and is binding only if there was timely
notice and the interpretation is reasonable. State of New Jersey v.
Dept. of Health and Human Services, 670 F.2d 1262, 1282- 1283 (3rd Cir.
1981), and cases cited therein; New York State Dept. of Social Services,
DAB No. 1023 (1989); New York State Department of Social Services, DAB
No. 788 (1987); Maine Dept. of Human Services, DAB No. 712 (1985); and
Nebraska Dept. of Health, DAB No. 373 (1982).

Accordingly, we consider below whether HCFA's prohibition on sampling
may be upheld as a reasonable interpretative rule with respect to the
costs in question here. 6/ The State argued that the prohibition on
sampling was a substantive rather than an interpretative rule, and was
not binding because it was not published pursuant to notice and comment
rulemaking as required by the Administrative Procedure Act. In view of
our conclusion that the prohibition on sampling was unreasonable as
applied here, we do not reach the question whether the rule was
something other than an interpretative rule. However, we note that the
decision cited by HCFA -- St. Mary's Hospital v. Blue Cross and Blue
Shield Ass'n, 788 F.2d 888 (2nd Cir. 1986) -- does not resolve this
question in HCFA's favor. That decision held that provisions of the
Medicare Provider Reimbursement Manual were interpretative rules exempt
from notice and comment rulemaking. It does not follow from the court's
holding that the altogether different State Medicaid Manual provision at
issue here constituted an interpretative rule. 7/

Whether HCFA's Prohibition on Sampling by States Is Reasonable in Light
of HCFA's Own Use of Sampling

There is no dispute that HCFA itself regularly relies on sampling in
order to support disallowances of claims under title XIX, and includes
sampling as an essential component of its procedures to impose
disallowances pursuant to quality control reviews of Medicaid claims.
Both the Board and the courts have upheld HCFA's use of sampling in
these contexts, finding that valid statistical sampling was a reasonable
and reliable method of determining unallowable amounts. The Board has
specifically noted, and HCFA here acknowledged, that valid statistical
sampling produces a result which has a high degree of probability of
being at least as accurate as a case-by-case review. Tennessee Dept. of
Health and Environment, DAB No. 898 (1987); see also State of Georgia v.
Califano, 446 F. Supp. 404 (N.D. Georgia 1977); Illinois Physicians
Union v. Miller, 675 F.2d 151 (7th Cir. 1982); and Tape Recording of
1/4/90 Telephone Conference. The State's use of sampling to claim FFP
in the expenditures here parallels HCFA's use of sampling to determine
how much of a state's claim is unallowable in that both uses are
designed to determine, with precision, economy and efficiency, whether
eligibility determinations are or are not supported by the case record.
Thus, HCFA's prohibition on sampling by the State in this case is
unreasonable unless there is some specific legal requirement or
compelling policy consideration which distinguishes the State's and
HCFA's use of sampling. As discussed in the remainder of the decision,
we conclude that there is no meaningful distinction.

Whether the Rationale for Upholding the Disallowance in DAB No. 1012
Applies Here

DAB No. 1012 distinguished the State's use of sampling from HCFA's on
the ground that the State used sampling in lieu of making individual
disability determinations which the regulations require in each case for
which payment is made. There, the State made timely disability
determinations in the sample cases after the sample was drawn; however,
the remaining cases in the universe lacked timely disability
determinations. DAB No. 1012 contrasted this with HCFA's routine use of
sampling to identify errors in eligibility determinations which are
already fully documented. We conclude that a similar distinction does
not exist here, and that the rationale in DAB No. 1012 for upholding the
disallowance of the disability-based claims does not apply here.

In the present appeals, the State asserted -- and we assume for purposes
of this decision -- that all case records in the sample drawn by its
auditors contained evidence concerning whether or not the individuals
were under age 21, pregnant, or eligible for AFDC. We further assume
that the universe from which the sample was drawn was a homogeneous,
defined universe, as the State asserted. Accordingly, unlike the
situation in DAB No. 1012, documentation in support of the total amount
claimed existed prior to the State's audit. 8/ Sampling was not used
here in lieu of developing the evidence required to show that the claims
covered payments to Medicaid eligibles; it merely tested the evidence
already present. 9/

HCFA argued, however, that this case was like DAB No. 1012 because the
State was attempting here to use sampling instead of making positive
eligibility determinations, i.e., determinations that individuals were
Medicaid-eligible, in each case. This argument appears to assume that
there is a specific requirement in the statute or regulations for a
positive eligibility determination, just as DAB No. 1012 found that
there is a specific requirement for a disability determination. We find
no requirement comparable to the requirement for a disability
determination, however. A disability determination is a judgment made
by qualified individuals, based on specific documents, that an
individual is disabled. No such judgment is necessary to determine
whether an individual is eligible for Medicaid based on age, pregnancy
or AFDC eligibility. The only regulations applicable here require
simply that the state Medicaid agency "dispose of each application by a
finding of eligibility or ineligibility. . . ." 42 C.F.R. 435.913(b)
(1979). The underlying intent of this regulation is to assure that
prompt action is taken on applications for medical assistance so that
eligible individuals receive timely payments. A determination that an
individual is eligible for MA although ineligible for Medicaid satisfies
the regulation since the individual receives the same coverage
regardless of the program to which his payments are charged.
Accordingly, unlike the situation presented in DAB No. 1012, the State
here complied with the requirement for making eligibility determinations
in each case even though its original eligibility determinations were
not positive ones.

HCFA also asserted that, to the extent that documentation of Medicaid
eligibility did not exist when the original eligibility determinations
were made, then a disallowance was required under the Board's reasoning
in New York State Department of Social Services, DAB No. 854 (1987).
However, HCFA did not identify, and we are unable to discern, any basis
for this position. In DAB No. 854, the Board held in part that a claim
which is inadequately documented when filed cannot be supported by
documentation obtained after the filing deadline has expired. Thus, the
fact that the case record would not have supported a determination of
Medicaid eligibility at the time the original eligibility determination
was made is not significant; the relevant inquiry is whether adequate
documentation existed when the claims were filed. (There is no dispute
here that the claims in question here were timely filed.) Since we
conclude in the next section that the documentation which existed at
that point was adequate, DAB No. 854 does not require a disallowance
here.

Whether Other Documentation Requirements Provide a Reasonable Basis for
the Prohibition on Sampling

HCFA argued that its prohibition on sampling was reasonable because the
State did not comply with the requirement articulated by the Board that
documentation must be related to the particular expenditures being
claimed. New York State Dept. of Social Services, DAB No. 433 (1983).
HCFA took the position that this requirement meant, in this context,
that the State must identify the specific individuals in the universe
who were Medicaid-eligible.

HCFA appears correct that the State would not meet its documentation
burden by simply providing case records for all individuals in the
universe of MA recipients; the State must in addition establish the
relationship between these individuals and its claims, which represent
payments to only some of these individuals. We conclude such a
relationship can be established by statistical sampling procedures.
That is, sampling could have established the extent to which there
existed, within the finite universe of MA cases, Medicaid-eligible
individuals for whom payments properly reimbursed under title XIX were
made. We emphasize, however, that this requires that the State have
correctly applied a sound, generally-accepted sampling methodology, and
that HCFA is not precluded from requiring the State to show that it did
so.

HCFA also relied on New York State Dept. of Social Services, DAB No. 982
(1988); Illinois Dept. of Public Aid, DAB No. 715 (1986); and Oklahoma
Dept. of Human Services, DAB No. 484 (1983). In each of these
decisions, the Board rejected attempts by the states to substitute
presumably allowable but unclaimed expenditures for other expenditures
which had been timely claimed but had subsequently been disallowed.
HCFA argued that it was implicit in these decisions "that a state's
claim for FFP is a claim for the federal share of particular
expenditures." HCFA's 6/1/89 brief, p. 21. The State's claims here did
not violate this principle, however. The particular expenditures
represented by the claims were those made on behalf of Medicaid-eligible
individuals within a finite universe of already- identified MA cases
sampled by the State. We see nothing in the cited decisions which would
require a further particularization of expenditures.

HCFA argued further that there was a statutory basis for requiring
documentation which specifically identifies the individuals to whom a
claim relates. In HCFA's view, since, under section 1903(a)(1) of the
Act, FFP is available only in expenditures for "medical assistance," and
that term is defined in section 1905(a) as payment of the cost of
certain care and services provided to individuals meeting specified
eligibility criteria, it follows that FFP is not available in
expenditures incurred on behalf of individuals whom the State has never
determined to be eligible for Medicaid. This argument ignores the fact
that, if the State's sampling methodology was valid, the State would
have determined the total amount paid on behalf of eligible individuals
in the universe sampled. The identification of the particular
individuals in the universe who met the eligibility criteria would do
nothing further to assure that FFP was paid only for medical assistance
to eligible individuals.

Whether Federal Oversight or Accountability Concerns Justify the
Prohibition on Sampling

HCFA argued that the prohibition on sampling was reasonable because the
use of claims based on extrapolation from a statistical sample would
make it impossible (1) to identify duplicate claims, (2) to perform a
quality control review with respect to the claims, or (3) to audit the
claims. HCFA stated that "[t]he possibility that these problems may
well prove to be amenable to solution at some time in the future . . .
does nothing to detract from the reasonableness of the Agency's
prohibition of estimated claims at the present time." HCFA's 6/1/89
brief, p. 27.

Nothing in the record shows that these problems are as substantial as
HCFA argued, or that they make the prohibition on sampling reasonable.
HCFA acknowledged that the State had reduced its estimate of Medicaid-
eligible expenditures to prevent duplicate claiming. HCFA's 6/1/89
brief, p. 23, n. 6. There is no evidence so far that this reduction did
not in fact account for all duplicate claims. We note that it is
appropriate here for HCFA to demand reasonable assurances that the
State's claims do not duplicate prior claims and that steps will be
taken to avoid future duplication of the claims in view of the potential
for duplication which arises because the claims based on extrapolation
from a sample are not tied, one-on-one, to particular individuals in the
universe. Moreover, if HCFA identifies a separate claim by the State
for any expenditures within the universe sampled in this case, that
claim would simply be unallowable unless the State could somehow
demonstrate that there was no duplication.

There is also no evidence that the use of sampling here has any
potential to affect the quality control system. HCFA asserted that no
reliable error rate could be established based on a quality control
review of the small number of sample cases. However, error rates are
determined based on Medicaid payments during an annual assessment
period. 42 C.F.R. 431.804(d) (1986). We are unaware of any provision
for adjustment when retroactive claims are filed. Thus, the cases
involved in the retroactive claims in question here would not be subject
to a quality control review even if they were based on a case-by-case
review rather than on extrapolation from a sample.

Moreover, the quality control system was designed to measure a state's
error rate in making initial determinations of eligibility, a different
type of error from errors potentially involved here. The State's
auditors might have made errors in reclassifying MA cases as
Medicaid-eligible, but HCFA could disallow that part of the State's
claims resulting from such errors (without applying the tolerance level
provided for by the quality control system for errors made by
caseworkers in determining eligibility).

HCFA nevertheless suggested that if sampling were permitted, nothing
would preclude the State from using sampling to make all its claims, not
just claims involving erroneously classified MA payments which could not
be individually reclassified before the statutory deadline for filing
claims. HCFA argued that, were the State routinely to make Medicaid
eligibility determinations only in sample cases, this would wreak havoc
on the quality control program since there would be few eligibility
determinations subject to review.

It is clear, however, that the State was not attempting here to
substitute statistical sampling for its basic eligibility determination
process. It is also highly unlikely that a state would cease making
Medicaid eligibility determinations in individual cases (where the state
ran its own medical assistance program in addition to Medicaid) since
all medical assistance would have to be paid with state funds until
retroactive claims based on sampling were made. The approach to
claiming envisioned by HCFA may in addition be contrary to the
requirement for a finding of eligibility or ineligibility for Medicaid
in each case. In any event, our decision that the use of sampling is
permissible in the facts of this case does not necessarily limit HCFA's
authority to prohibit the use of sampling under the circumstances which
it described, nor, indeed, HCFA's ability to limit sampling, as used
here or in other situations, in any reasonable manner it chooses through
rulemaking.

HCFA also failed to support its assertion that the use of sampling here
deprives HCFA of the ability to audit the State's claims. As the State
pointed out, HCFA can review the sample cases on which the State's
Medicaid claims were based to see if they were properly reclassified as
Medicaid-eligible. (HCFA's complaint that the pool of expenditures from
which its auditors could in turn draw a sample would be too small could
presumably be dealt with by requiring the State to use a larger sample
size.) Alternatively, HCFA can take its own sample from the universe of
MA cases to verify the accuracy of the claims. In addition, HCFA can
review the State's statistical methodology to determine whether it was
valid. (The use of an invalid methodology would render the claims
unallowable unless the State were willing and able to recalculate the
claims pursuant to a valid methodology.) HCFA did not show why these
steps would not offer effective audit tools.

Whether the Existence of Other Means of Identifying Cases Eligible for
FFP Is a Valid Basis for Prohibiting Sampling

HCFA argued finally that the State had means other than sampling of
identifying cases which were improperly classified as ineligible for
Medicaid, and that HCFA should not be required to allow yet another
effort to do so. HCFA noted that the State's own regulations required
it to make redeterminations of eligibility for State public assistance
programs at least once every six months, so that the State "had a
minimum of four separate opportunities to catch its original error and
to submit a timely claim for FFP." HCFA's Response to the Board's
Preliminary Analysis, p. 10. 10/ HCFA also noted that the State's
Shares Reclassification System (SRS) was designed to identify
expenditures which were erroneously determined to be ineligible for
Medicaid. HCFA asserted that "it would be patently unreasonable to
require HCFA to redefine its longstanding position as to what
constitutes a properly documented claim simply to afford the state one
more opportunity to claim FFP which remains unclaimed through no fault
of HCFA's but rather due to the State's own administrative
inadequacies." Id., p. 12.

We are not persuaded that the existence of alternative means of
identifying amounts which could have been claimed as Medicaid
expenditures is controlling here. Neither the Board nor the courts have
expressly limited the use of statistical sampling to determine
unallowable amounts to situations where a case-by-case review was not
feasible. Moreover, as the State pointed out, a grantee under title XIX
has a statutory and regulatory right to claim federal funding for
expenditures made on behalf of eligible individuals. While we
sympathize with HCFA's apparent sense of frustration with what it views
as inefficient State procedures for making claims, any such inefficiency
would not render the claims themselves invalid. 11/

It should be noted, moreover, that the State's need to reclassify
expenditures arose because the State was administering a complementary
state-only medical assistance program, MA, as well as Medicaid. Such
complementary programs presumably are to be encouraged since they
further the purposes of the Medicaid program and may reduce the demand
for federal Medicaid funding. Thus, a state should not be needlessly
penalized for errors which might not occur but for the complexities of
administering two programs in tandem.

HCFA also argued that statistical sampling was not a cost-effective way
of identifying amounts which should have been claimed under title XIX
where the State was already identifying cases which were misclassified
as ineligible for Medicaid through periodic redeterminations and its
SRS. The Board stated in the preliminary analysis issued in this case
that the cost implications of sampling were a factor having a bearing on
the reasonableness of the prohibition on sampling. HCFA reasoned that
there were no cost savings because sampling was not done in lieu of, but
rather in addition to, a case-by-case review in the form of the State's
redetermination process and SRS. If SRS and the redetermination process
did not adequately identify misclassified cases, however, then the State
was entitled to perform a further review. 12/ Thus, the cost of
sampling was properly compared to the cost of an additional case-by-case
review. HCFA did not dispute the State's assertion that sampling was
the more cost- effective of the two procedures. Any cost savings would
accrue to HCFA as well as the State to the extent that the
administrative costs of reclassifying MA cases are chargeable to
Medicaid. 13/

HCFA argued further that the reasonableness of its position was
demonstrated by the fact that section 1132(b) of the Act provides that a
state's administrative inadequacies do not constitute good cause for a
state's failure to file a claim within the two-year statutory period.
However, the fact that administrative inadequacies cannot be used to
justify a failure to meet an applicable time limit does not mean that an
otherwise timely and allowable claim can be properly disallowed because
of a state's administrative inadequacies. (HCFA did not argue that
section 1132(b) applied directly here, where there is no question that
the State's claims were timely filed.)

HCFA also suggested that the State should "pay a price for failing to
make a correct determination" of Medicaid eligibility in cases which
were misclassified due to caseworker or clerical error. HCFA's Response
to The Board's Preliminary Analysis, p. 15. The "price" referred to by
HCFA is the denial of the opportunity to use statistical sampling to
identify cases which were misclassified as ineligible for Medicaid.
According to HCFA, this would give the State an incentive to improve its
eligibility determinations. However, the fact that the State must use
its own funds where an individual is erroneously found ineligible for
Medicaid is a strong incentive for the State to make correct eligibility
determinations. Absent more affirmative authority in law or regulation
than we find here, HCFA cannot reasonably disallow FFP in otherwise
allowable costs merely because it feels that the State needs additional
incentives.

Further Issues

Although we conclude that HCFA's prohibition on claims based on
statistical sampling is unreasonable as applied to the
non-disability-based claims in this case, this is only a threshold
issue. The parties agreed to assume for purposes of this decision that
(1) the sample cases on which the extrapolation was based were in fact
Medicaid- eligible, and (2) the State's statistical sampling methodology
was valid. HCFA's brief dated June 1, 1989, p. 3, first full paragraph,
and n. 1. Should either of these assumptions prove unfounded, then the
allowability of some or all of the costs remains in question.
Specifically, if any of the sample cases were incorrectly found to be
Medicaid-eligible based on age (under 21), pregnancy, or eligibility for
AFDC payments, the amounts claimed based on extrapolation from the
payment made in any such case would not be allowable. Furthermore,
extrapolation would not appear to be justified if the universe from
which the sample was taken did not consist of MA cases which would have
been eligible for Medicaid if the individuals had been identified as
having one of these characteristics. In addition, if any aspect of the
State's statistical sampling methodology was invalid, the claims would
be unallowable unless the State could develop an acceptable methodology
which would support some or all of the claims. As previously noted, the
State bears the burden of establishing the validity of its statistical
sampling methods. Thus, HCFA may require it to provide a detailed
written explanation of its procedures which describes the State's
methodology and its theoretical underpinnings. HCFA should not be
required to conduct an audit of the State's statistical procedures, but
is entitled to rely on the State to provide complete information.

HCFA further stated in proceedings in DAB No. 1012 that it assumed
"solely for the sake of argument" that all MA expenditures in the
universe were made for services which would constitute covered services
under Medicaid. Docket No. 88-82, HCFA's brief dated 10/18/88, p. 5, n.
4. HCFA stated that--

. . . we are skeptical about the accuracy of the State's assertion
on this point since there are several types of services which are
covered by the state-only program of medical assistance but not by
Medicaid, e.g., abortions, inpatient mental health services for
individuals between the ages of twenty- one and sixty-four, etc. .
. . The only way for the State to inform itself of . . . whether
the services are covered under Medicaid . . . is for the State to
conduct . . . [a] case-by-case review. . . .

Id. Assuming that this remains HCFA's position, however, HCFA can
determine the extent to which the cases in the universe involved covered
Medicaid services by looking at the cases in the sample. To the extent
that unallowable services were involved, the claims would not be
allowable.

Conclusion

For the foregoing reasons, we conclude that the claims were not properly
disallowed on the sole ground that they were developed using statistical
sampling procedures. The appeals are remanded to HCFA to assess the
validity of the sampling methodology which the State used to establish
its claims or other matters left open by this decision.


____________________________ Judith A.
Ballard


_____________________________ Donald F.
Garrett


_____________________________ Norval D. (John)
Settle Presiding Board Member

1. Docket No. 89-51 involves a claim for $1,940,948 FFP for
expenditures incurred during the quarters ended September 30, 1985,
December 31, 1985, and June 30, 1986. Docket No. 89-86 involves a claim
for $106,426 FFP for expenditures incurred during the quarter ended
September 30, 1986. Docket No. 89-87 involves a claim for $860,012 FFP
for expenditures incurred during the quarter ended December 31, 1986.
There was no claim for non- disability-based cases for the quarter ended
March 31, 1986.

2. HCFA asserted, and the State did not dispute, that claims for the
expenditures made on behalf of the individuals in the sample cases were
not included in the amounts in dispute here. HCFA's brief, p. 5.

3. HCFA did argue without specific reference to this regulation that
"estimated figures should not be used for the purpose of the
reconciliation that takes place at the end of a quarter," since that
would require HCFA to base its final adjustments for a quarter "on a
comparison of one set of estimates with another." HCFA's brief dated
6/1/89, pp. 17-18. However, unlike the estimates which provide the
basis for advance payments to the states, the estimates based on
statistical sampling are derived from actual expenditures. Accordingly,
in making final payment on a claim based on statistical sampling, HCFA
would not simply be comparing two estimates which result from guesswork
on the part of the State.

HCFA also cited New York State Dept. of Social Services, DAB No. 537
(1984), as authority for its position that the estimates in question
here cannot be used to adjust a state's quarterly claim. However, the
Board there found that "the figure on the QER, even though referred to
by both the state and the Agency as an `estimate,' was not the kind of
an `estimate' reasonably prohibited by the action transmittal [a
document which contained language similar to section 230.30(c)]." DAB
No. 537, p. 14. The figure used on the State's QER in that case "was
considerably more than just a guess," since it was "determined by a
systematic analysis of expenditures designed to identify those
appropriately claimed for FFP." Id., p. 16. Thus, DAB No. 537 does not
support HCFA's position.

4. HCFA relied in addition on Transmittal No. 52, dated September
1987. The Board found in DAB No. 1012, supra, that this transmittal
constitutes a more explicit statement of the policy articulated in
Transmittal No. 30. Since Transmittal No. 52 was not substantively
different from Transmittal No. 30, we need not address the State's
contention that Transmittal No. 52 was published too late to be binding
on the State in relation to the claims involved here.

5. We reject HCFA's suggestion that the fact that it had a
"longstanding position" that claims based on statistical sampling were
unallowable showed the reasonableness of its interpretation. Clearly,
the longevity of HCFA's position alone is not sufficient to establish
its reasonableness.

6. HCFA argued specifically that the prohibition on claims developed
through sampling, projection, and other estimating techniques
"constitutes an interpretation of the term `claim' as used in section
1132 of the Act and in the corresponding regulations." HCFA's 6/1/89
brief, p. 32. We agree that this prohibition is a reasonable
interpretation to the extent that it is read as meaning that a state
cannot evade the two-year filing limit in section 1132 by non-scientific
sampling or other estimating techniques which produce only a rough
estimate rather than an auditable determination of what amounts were
actually expended for a particular purpose during a quarter. Our
decision is simply that, as applied here, the prohibition on sampling is
not a reasonable interpretation of the regulations governing the
claiming process.

7. HCFA argued in the alternative that the transmittal was a
procedural rule, also excepted from notice and comment rulemaking under
section 553(b), "since it applies only to the manner in which states
present their claims for FFP and not to the states' entitlement to FFP
in the expenditures underlying these claims." HCFA's 6/1/89 brief, p.
33. Contrary to HCFA's assertion, however, the prohibition on sampling
does not specify how or when to submit claims, but qualifies the nature
of allowable claims, resulting in a denial of reimbursement for
expenditures to Medicaid eligibles. If the prohibition were enforced,
the State would have to restructure its eligibility determination
process or conduct costly case-by-case reviews of MA cases in order to
obtain federal funding for the types of cases which were mis-classified
here. Thus, consistent with case law holding that a rule is not
procedural when it "is likely to have the intent or effect of
substantially altering party behavior," (American Hospital Assoc. v.
Bowen, 834 F.2d 1037, 1050 (D.C. Cir. 1987)), the transmittal is not a
procedural rule binding on the State.

8. If either of the assumptions on which this finding is based is
incorrect, then the State's sampling approach might be impeached.

9. Given these facts, it is not clear, as it was in DAB No. 1012, that
the State failed to comply with the requirement at 42 C.F.R.
431.17(b)(1) (1979) for the maintenance of individual records that
contain information on "[f]acts essential to determination of initial
and continuing eligibility. . . ."

10. The four redeterminations come at six-month intervals during the
two years after an expenditure is incurred during which the State may
file a claim. There would be fewer redeterminations if a case was
closed because an individual became ineligible for MA during this
period, however.

11. While the State acknowledged that there were inadequacies in its
redetermination process, the State denied that its overall
administration of the Medicaid/ State medical assistance program was
inadequate, noting that the reclassified expenditures at issue here
"represent a relatively small percentage of the Appellant's FNP-MA
expenditures for the time periods in question." State's supplemental
reply brief, p. 13, n. 10.

12. It does not necessarily follow that FFP in the costs of the review
itself is allowable. In New York State Department of Social Services,
DAB No. 1102 (1989), the Board held that the administrative costs of
reclassifying State Home Relief cases as AFDC-eligible were properly
charged as costs of the AFDC program where the reclassification activity
did not duplicate regular eligibility determination procedures and
HCFA's determination that the costs were not necessary was made on a
retroactive rather than prospective basis. We do not determine whether
the administrative costs of conducting the sampling here are analogous
to the costs at issue in DAB No. 1102 in this respect.

13. HCFA asserted that the State's calculation that the cost of a
case-by-case review would be $21,556,640 as compared to the $120,000
cost of the sampling performed here overstated the cost savings since it
failed to recognize the cost to HCFA of employing statistical experts to
evaluate the claims based on sampling. However, since the State has the
burden of establishing that it used valid statistical procedures, HCFA's
review of the State's procedures should not require significant
manpower. (We note in any event that HCFA already has some statistical
support in making disallowances based on sampling which might be
available for the type of review in question