DEPARTMENTAL APPEALS BOARD
Department of Health and Human Services
SUBJECT: Louisiana Department of Health
and Human Resources
Docket No. 87-96
Decision No. 989
DATE: October 14, 1988
DECISION
The Louisiana Department of Health and Human Resources (State) appealed
a
determination by the Health Care Financing Administration (HCFA,
Agency)
disallowing $772,431 in federal financial participation (FFP)
claimed under
Title XIX (Medicaid) of the Social Security Act. The
amount claimed
represented the federal share of expenditures for
services provided during
the period October 1, 1982 through March 31,
1983.
HCFA examined a sample of the State's Medicaid caseload and found that
the
State had claimed FFP in the cost of services to
ineligible
individuals. Specifically, HCFA found that the State had not
made
required determinations that these individuals met financial
eligibility
requirements. The State argued that the disallowance was
barred by the
Tax Equity and Fiscal Responsibility Act of 1982 (TEFRA) (Pub.
L.
97-248, 42 U.S.C. 1396b), and, even if not specifically barred,
the
disallowance was nonetheless precluded by the provisions of the
"quality
control" (QC) system under TEFRA. The State also questioned in
general
the use of statistical sampling as well as the particular
sampling
methodology HCFA used, submitted documentation to show the
eligibility
of some of the sample cases, and argued that other cases should
not have
been counted toward the disallowance. HCFA accepted the
State's
arguments as to some of the cases, but argued that the State had
not
provided proper documentation that it had determined the eligibility
of
the remaining cases.
Based on the analysis below, we have made the following determinations:
- We uphold HCFA's position on the effect of TEFRA
and,
therefore, we sustain the disallowance in principle (but not
in
amount).
- We find that Louisiana proved that HCFA erred in counting
all
of the payments in six cases as a basis for the
disallowance.
We find that HCFA was correct concerning all the
remaining cases
except for 26 cases in which the dispute centered on
the
adequacy of State Form 770 as documentation of eligibility;
we
uphold HCFA's determination on these cases
provisionally,
subject to an opportunity for the State to present an
item of
evidence discussed below.
- We find that HCFA was not precluded from using
statistical
sampling techniques to establish the amount of
this
disallowance. However, in the facts of this case, we find
the
HCFA erred in using the value of the mid-point estimate.
Discussion
A. The disallowance is not affected by TEFRA.
The State contended that by enacting a QC system in TEFRA,
Congress
confined within that system HCFA's authority to disallow for
erroneous
Medicaid payments. Below, we conclude that HCFA was
reasonable in
determining that Congress, by delaying for six months the
effective date
of the disallowance sanctions in the TEFRA-based QC system,
left HCFA
free to impose non-QC disallowances during that period, which
coincides
with the period at issue here.
1. Background: the QC disallowance system
In general terms, QC systems are concerned with lowering the incidence
of
errors in high-volume actions affecting payments in public
assistance
programs. These errors cumulatively are costly. As
part of the QC
systems adopted by Department regulations for the Aid to
Families with
Dependent Children (AFDC) program and the Medicaid program,
each state
is required on an ongoing basis to examine monthly samples taken
from
the state's entire program caseload in order to identify case
errors.
The federal agency then reviews a subsample of the state's QC
sample,
and error rates are established for each state program.
In 1979, HCFA adopted procedures and rules for imposing disallowances
of
FFP for eligibility errors detected through the Medicaid QC
system,
similar to disallowance provisions in AFDC. Because of
the
comprehensive nature of the QC system, as well as its other
attributes,
disallowances under the QC provisions were subject to
several
safeguards: 1) disallowances were taken only for errors in
excess of a
specified error rate; and 2) a state could obtain a waiver of a
QC
disallowance in certain circumstances. The effect of the
first
safeguard, known as a "tolerance level," was to authorize FFP
in
erroneous payments up to the tolerance level, in spite of the fact
that
FFP was not otherwise available in such erroneous payments
under
applicable statutory and regulatory requirements. For a more
detailed
discussion of the history and nature of Medicaid quality
control
systems, see Consolidated Appeals of Medicaid Quality
Control
Disallowances, DGAB No. 948 (1988).
In TEFRA, Congress established a new quality control system at
section
1903(u) of the Act, replacing the previous Medicaid QC system as
of
October 1, 1982. TEFRA, section 133. Quality control
monitoring under
the new standards was to begin immediately, but Congress
required
disallowances under the new system only after March 31, 1983.
Id. at
(a) and (b); section 1903(u)(1)(A) of the Act. The
intervening
six-month period was to allow time to gather information on
the
experience of the Agency with the existing quality control system.
S.
Rep. No. 494, 97th Cong., 2d Sess. 38-40 (1982). TEFRA did not
address
how states would account for erroneous payments during the
intervening
period. TEFRA did not include any provision for tolerance
levels, or
other special treatment for eligibility errors during that
intervening
period.
2. TEFRA did not preempt this disallowance.
The QC system established by TEFRA immediately replaced the
previous
Medicaid QC system, but disallowances based on excessive error
rates
detected by the QC system were not authorized unless a state failed
to
meet the target error rate for quarters after March 31, 1983.
Section
133 of TEFRA provided:
(a) [F]or the period consisting of the third and
fourth
quarters of fiscal year 1983, or for any full fiscal
year
thereafter . . . the Secretary shall make no payment for
such
period or fiscal year with respect to so much of such
erroneous
excess payments as exceeds such allowable error rate of
0.03.
* * * * *
(c) No provision of law limiting Federal financial
participation with respect to erroneous payments . . ., other
than the
limitations contained in section 1903 of such Act,
shall be effective
with respect to payments . . . for quarters
beginning on or after
October 1, 1982 . . . .
HCFA argued that section 133(c) suspended the authority to
impose
disallowances under the Medicaid QC system, but left intact
the
provisions of the Act (in sections 1903 and 1905) which permit FFP
only
in payments to eligible recipients. Thus, HCFA contended,
a
disallowance outside the QC system is authorized for payments
for
ineligibles. The State did not dispute that HCFA was authorized
to
disallow FFP for some kinds of errors during the period in question,
but
argued that payment errors for allegedly ineligible recipients
were
covered under the quality control system authorized by TEFRA and
thus
HCFA's authority to disallow FFP was preempted. HCFA did not
dispute
that disallowances for eligibility errors are governed generally by
the
QC disallowance system authorized by TEFRA (as well as the QC
system
which existed prior to TEFRA). HCFA contended, however, that
the
TEFRA-based preemption was not effective until April 1, 1983, the
date
beginning the first quarter for which disallowances were required
under
the TEFRA-based QC system, and thus was not controlling during
the
six-month period involved here.
The State cited legislative history which emphasizes that the
TEFRA-based
QC system immediately replaced the previous QC system. The
cited
history also underscores that the effective date for disallowances
for states
exceeding the target error rate was delayed until April 1,
1983:
The Senate amendment [which was adopted] deletes the
Medicaid error
rate provision and penalties
incorporated in the 1980
Appropriations Act and
substitutes language establishing a 3
percent target
error rate for quarters beginning after March 30,
1983. . . . The [QC] provision is effective on enactment.
H.R. (Conf.) Rep. No. 760, 97th Cong., 2d Sess. 439 (1982), State's
Ex.
9.
The legislative history indicates that the purpose of the delay
in
imposing disallowances was to allow more time to study the
experience
which the Agency has had with QC disallowances:
The [Senate Finance] committee is aware that many
questions remain
to be resolved relative to the
matter of sanctions for excessive
rates of error. .
. .
The committee has delayed the effective date for
imposition of
fiscal sanctions until April 1983 in
order to allow it time to
study the existing quality
control system.
S. Rep. No. 494, 97th Cong., 2d Sess. 38-40 (1982), State's Ex. 10.
We find that HCFA was reasonable in concluding that neither the
plain
language of TEFRA nor the intent of Congress evidenced by
the
legislative history precludes the type of non-QC disallowance
taken
here. In TEFRA, Congress established an error rate and a system
for
achieving that level of tolerance applicable to all Medicaid
eligibility
errors, not just the relatively limited category addressed by
this
disallowance. By making the new Medicaid QC system
effective
immediately, Congress ensured that the states would continue the
QC
review process without interruption. By delaying the disallowance
part
of the new QC system, Congress allowed time to study the effect
of
disallowances for excessive error rates detected by the QC
system
without disrupting the ongoing effort to identify and reduce
error.
This does not mean that Congress intended to give the states a
six-month
period during which they would not be accountable at all for
erroneous
Medicaid payments. To the contrary, Congress specifically
left intact
the "limitations contained in section 1903" of the Act;
section
1903(a)(1) provides FFP only for "medical assistance under the
State
plan" and section 1905(a) of the Act defines "medical
assistance"
essentially as payment for covered services provided to
eligible
individuals. Section 1903(d) of the Act authorizes HCFA to
adjust for
overpayments of FFP.
The QC tolerance level allows reimbursement of otherwise clearly
erroneous
payments, but does so only as part of an overall QC system the
purpose of
which is to lower error rates in the first place. HCFA
reasonably took
the position that it would not participate in payments
for ineligibles during
a period when no provision authorizing HCFA to do
so was in effect. The
State's reading of the TEFRA postponement
provision would stand the Medicaid
QC system on its head, in effect
requiring HCFA to participate in all
unauthorized payments merely
because Congress delayed implementation of
sanctions for excessive rates
of error detected by the QC system.
The State's position is also contrary to the decision of this Board
in
Florida Dept. of Health and Rehabilitative Services, DGAB No.
955
(1988). There the Board concluded, based on its analysis of TEFRA,
the
legislative history of TEFRA, other parts of the Social Security
Act,
and prior Board decisions, that there was "no support, either in
the
history of quality control disallowance systems or in TEFRA, for
the
State's argument that the policies inherent in the quality
control
disallowance systems require HCFA to fund erroneous recipient
payments
during periods in which no quality control disallowance system was
in
effect." Id. at 7.
3. A disallowance was authorized under the Act.
As stated above, we have concluded that the disallowance is not barred
by
TEFRA. In addition, the State questioned the authority of HCFA to
take
a sample-based disallowance outside of TEFRA.
HCFA based the disallowance on its long-standing authority under
section
1903 of the Act, which calculates FFP as a percentage of the
State's
medical assistance expenditures for eligible recipients.
Sections
1903(a)(1), 1903(d)(2), and 1905(a) of the Act. HCFA cited
Oklahoma
Dept. of Human Services, DGAB No. 417 (1983), p. 3, where the Board
held
in effect that these statutory provisions authorized a disallowance
of
FFP in payments to ineligible individuals.
The State argued that HCFA's authority under section 1903 of the Act
to
disallow FFP in payments for ineligibles was preempted by TEFRA,
which
is section 1903(u) of the Act. The State also contended that
Oklahoma
was distinguishable because there the disallowance was based
on
individual errors, not a sample.
We find here, as we have in previous cases, that sections 1903 and 1905
of
the Act do authorize disallowances of FFP in erroneous payments,
including
those due to eligibility errors. The factual differences
between this
case and the Oklahoma and California cases do not make our
holdings there
inapplicable to this case. Our holdings in those cases
did not depend
on whether the disallowances were calculated by totalling
the individually
identified errors or were calculated by the use of a
statistical sampling
methodology (an issue discussed separately below).
Moreover, as we discussed
above, non-QC disallowances for the kinds of
errors covered by section
1903(u), added by TEFRA, were not preempted
during the period in
question. Thus, HCFA had a statutory basis for
denying FFP in payments
for ineligibles.
The State also argued that if Congress had intended that
disallowances
continue without interruption, it would have specifically
provided for
that, as it did with AFDC. Congress did not delay
implementation of
comparable parts of the TEFRA-based QC system for AFDC, and
specified
that the QC regulations then in effect for AFDC would remain in
effect
until new regulations reflecting the TEFRA changes were promulgated
and
in effect. TEFRA, section 156(c); State Ex. 7. However, this
does not
mean that by delaying Medicaid QC disallowances Congress intended
to
prohibit non-QC Medicaid disallowances. Simply because Congress did
not
provide for a hiatus in AFDC QC disallowances does not mean
that
Congress gave the states a six-month period in Medicaid during which
FFP
was required in all erroneous payments.
4. HCFA is not prohibited by TEFRA from imposing a
sample-based
disallowance which does not incorporate an error tolerance level
and
opportunity for a waiver.
The State argued, in effect, that even if HCFA were authorized
to
disallow FFP in the individually identified cases of error, it could
not
base its disallowance on a sample without the type of "safeguards"
that
applied to a sample-based disallowance under either the
previous
Medicaid QC system or the new TEFRA-based QC system. These are
(1) an
error tolerance level; and (2) the opportunity for a waiver if the
State
was unable, despite a good faith effort, to maintain error rates
below
the tolerance level. The State maintained that a
sample-based
disallowance without these safeguards was a more severe sanction
than
Congress authorized either prior to or subsequent to the
six-month
period in question, and thus was unreasonable, in addition to
being
contrary to the presumed intent of Congress in delaying the
imposition
of QC disallowances during these six months. The State also
argued that
disallowances without these safeguards could be based only
on
individually identified errors, citing several Board
decisions.
However, the cited cases, with one exception, involved the AFDC
program,
not Medicaid. See California Dept. of Social Services, DGAB
No. 319
(1982); Pennsylvania Dept. of Public Welfare, DGAB No. 485 (1983);
and
Louisiana Dept. of Health and Human Resources, DGAB No. 580 (1984).
The
AFDC cases all involved a unique factual circumstance: the Board
found
that the HHS component involved had established as a matter
of
affirmative policy that it would not use sample-based disallowances,
and
could not disavow that policy without providing timely notice. No
such
history is found here. (See our discussion in Part E below)
This Board has consistently held that a sampling-based
disallowance
without an error tolerance and other QC safeguards is proper
outside of
the QC system, as here, unless there is an agency policy
restricting
such disallowances. Statistical sampling, properly applied,
is a
reliable, cost-effective audit technique which may more
accurately
establish the rate and dollar amount of errors than a 100
percent
review. California Dept. of Social Services, DGAB No. 816
(1986), pp.
4-5.
As we have seen, because this is a non-QC disallowance, the
error
tolerance level and waiver provisions of TEFRA do not apply.
Despite
the State's arguments to the contrary, HCFA was reasonable in
not
incorporating similar safeguards in this disallowance. Tolerance
of
errors is a aspect of QC systems, in which the states are required
to
monitor on an ongoing basis their entire caseload. It is arguable
that
some degree of error is unavoidable, and that HCFA's approach here
was
more severe than it had to be. However, that involves a
policy
determination beyond the authority of this Board; HCFA could
reasonably
determine under the Act that it was not required to have a
tolerance
level or to consider a waiver when it reviewed payments to one of
the
many categories of Medicaid cases. See California v. Settle, 708
F.2d
1380 (9th Cir. 1983).
Moreover, the rationale for applying these safeguards simply would
not
support reversal of the disallowance here. The QC tolerance level
is
based on a recognition that, because of the complexity of
eligibility
requirements and a bias toward finding eligibility, it is
impossible to
run an error-free eligibility determination process. See
Maryland v.
Mathews, 415 F.Supp. 1206 (D.D.C. 1976). Similarly, the QC
waiver
provision protects states whose error rate is high due to
circumstances
beyond their control. Here, the errors were caused not by
caseworker
mistakes in applying eligibility requirements, nor by
circumstances
beyond the State's control, but by the fact that the State
admittedly
was not devoting sufficient resources to the program. State
August 7,
1987 brief, p. 2. Here, the State was as a matter of practice
ignoring
the financial eligibility requirements and placing or retaining on
the
Medicaid rolls members of this foster care class, without performing
all
the required eligibility determinations. Even the Board's decisions
in
the AFDC cases relied on by the State recognized that systemic
errors
caused by a state's policy or practice may fall outside of
agency
disallowance policies established for caseworker mistakes.
B. Certain cases and amounts should not be counted as a basis for
the
disallowance.
After HCFA informed the State of the disallowance based on the sample
of
284 cases, Louisiana raised various objections to HCFA's
counting
certain cases and payment amounts as being ineligible for
Medicaid.
Both prior to and during the pendency of this appeal, HCFA agreed
that
it would not count a number of these cases as ineligible,
finally
reducing the number of cases HCFA considered ineligible to 149.
HCFA
March 31, 1988 Brief, pp. 9-13.
The State responded with argument and documentation purporting to
show
that an additional 33 cases should not be counted as
ineligible.
State's Second Supplemental Appeal File, Affidavit of Carmen
Weisner,
(Affidavit), pp. 5-25 (May 6, 1988). We find that Louisiana
has proven
that payments in six of these were improperly counted toward
the
disallowance; they are discussed in the following paragraphs.
The
remaining 27 are analyzed in Part D of this decision. This basis
for
the State's position on the six cases was primarily that these
cases
were not properly included in the universe of foster care cases
which
was the subject of the sample. HCFA had itself recognized this
problem
for certain sample cases and had treated the cases as eligible
by
assigning them zero dollar value in calculating the
extrapolated
disallowance. (We refer to this as the parties did, as
"not counting
the payments as a basis for the disallowance.")
The case of N.F.
The State argued that payments for services to N.F. should not be
counted
because N.F. was actually on the rolls of the AFDC program, and
thus not in
the foster care category which was the subject of HCFA's
survey.
Affidavit, pp. 19-20; State Ex. 50. According to the State, a
foster
care case is shown by "06" as the third and fourth digits of the
Claim
Recipient Number (C.R. No.) on the Recipient Data Sheet, Report
No.
P-O-09. The Report on N.F. has "03" as the third and fourth digits
of
the C.R. No., which the State contended was the code for AFDC
cases.
Therefore, the State contended, it was improper for HCFA to use
this
case to extrapolate errors to the universe of foster care children.
HCFA filed two briefs following the State's May 6 Brief, but did
not
specifically dispute or comment on the State's allegations
concerning
this or the other six cases discussed in this part. We
therefore find,
on the basis of the State's unrebutted evidence, that the
payments of
$2,142.67 to N.F. should not be counted as a basis for the
disallowance.
The cases of J.H. and T.W.
The State argued that payments for services to J.H. and T.W. should not
be
counted because for the period at issue these children were not
foster care
children, like the other cases under review. The State
demonstrated,
and HCFA did not subsequently deny, that during the period
at issue, J.H. and
T.W. were "receiving active treatment as inpatients
in psychiatric
facilities," which is a category of non-AFDC individuals
potentially eligible
for Medicaid but separate from "individuals in
foster homes or private
institutions for whom a public agency is
assuming a full or partial financial
responsibility," the foster care
children under review. 42 C.F.R.
435.222(b)(4) and (b)(1). Affidavit,
pp. 21-22, 24; State Ex. 52,
56. In its September 1985 letter
describing the review and in the
disallowance letter, HCFA referred to
the subject of its review as foster
care individuals. HCFA Ex. H.
We therefore conclude that, for the period at issue, payments for
services
to J.H. and T.W. were not properly included in the category
being reviewed by
HCFA, and HCFA should not have counted these payments
of $20,868.07 and
$6,350.07 as a basis for the disallowance. Even if
the financial
eligibility requirements for the two categories are
identical, HCFA was not
reasonable in using these two payments in
extrapolating to a universe of
payments for foster care individuals.
The case of J.J.
The State argued that payments for services to J.J. should not be
counted
because J.J. was eligible for Medicaid under Title IV-E of the
Social
Security Act as of December 1982, retroactive to September 1982,
the date of
a court order giving the State custody of J.J. Affidavit,
p. 22; State
Ex. 53. As with the other cases in this group of seven,
HCFA did not
dispute the State's allegation, although HCFA refused to
exclude J.J.'s
payments, without giving a specific reason. HCFA March
31, 1988 Brief,
p. 13.
We therefore conclude, on the basis of the State's undisputed
showing,
that the payments for services to J.J. ($529.88) should not
have been
counted toward the disallowance.
The case of G.R.
The State argued that although G.R. was eligible for Medicaid as
a
Supplemental Security Income (SSI) recipient during the period at
issue,
three of the Medicaid payments at issue had been claimed under
the
foster care code "06" because the SSI application was
pending.
Affidavit, p. 23; State Ex. 54. The State alleged that G.R.'s
SSI
eligibility was retroactive to the date of application and thus
covered
these payments as well as those claimed under the SSI code
"04". As
with other cases, HCFA did not counter the State's
evidence.
We therefore conclude, on the basis of the State's undisputed
showing,
that the payments of $4485.66 for services to G.R., like those to
other
SSI eligibles, should not have been counted in the disallowance.
The case of S.T.
The State alleged that the amount allegedly paid in error for S.T.
was
$33.38, not the $383.14 used by HCFA to calculate the
disallowance.
Affidavit, p. 24; State Ex. 55. The State indicated that
the difference
of $349.76 consisted of payments for which no FFP was claimed
or
payments for services during a period when federal auditors
determined
that S.T. was eligible. HCFA did not dispute this.
We conclude, on the basis of the above, that the correct amount
was
$33.38. The basis on which the disallowance was calculated should
be
decreased by $349.76.
Part D. The State did not document the eligibility of the
remaining
cases.
As stated, HCFA accepted some of the State's documentation and reduced
the
number of cases of erroneous payments to 149. The State
offered
documentation on 33 of the 149. We agreed with the State on six
(see
analysis above). However, as we discuss below, we find that the
State
did not establish that payments to the remaining 27 (of the 33)
were
made with a proper determination of eligibility, but should have
a
limited opportunity to do so in 26 of the 27 cases.
The 26 Form 770 Cases
The State argued that the determination of eligibility in 26 foster
care
cases which were the subject of the HCFA review could be documented
by
State Form Number 770, styled the "Contribution Assessment Form."
See,
e.g., State Ex. 24. That form elicits certain financial
information
concerning the child and has space for the signatures of the
child, a
parent or a guardian, and the caseworker.
HCFA asserted that the State Plan and federal regulations required
the
State to use a Form 50-M for the initial determination of
eligibility,
and that whatever form the State used, the required signatures
were to
be executed under penalty of perjury. HCFA Ex. E, F, G; 42
C.F.R.
435.907.
The State countered that the State Plan required the Form 50-M only
for
long-term care cases, and that the Form 50-M was inappropriate
for
foster care cases. The State contended that only the
caseworker's
signature was required on a Form 770. The State did not
dispute that
federal regulations required that the assistance application has
to be
signed under penalty of perjury, but contended that the requirement
had
no meaning where the applicant was a child and that the
required
signatory, the caseworker, had no motive to lie and faced a far
greater
consequence than an accusation of perjury (loss of a job)
for
falsification of the information on the form. The State also
pointed to
the instructions for the Form 770, which provide that, by signing
the
form, the caseworker "attests" to the correctness of the
information
provided.
We are not persuaded that the excerpts from the State Plan and the
State
Medical Assistance Manual relied on by HCFA require the State to use
the
Form 50-M for these foster care cases, but we also find that the
State
did not establish that the Form 770, as used here, was sufficient
to
meet the applicable regulatory requirements. While we find that
the
Form 770 was a State-prescribed form, the record is unclear as
to
whether the caseworker signed under "penalty of perjury." The
State
Plan requires the State to meet all requirements of 42 C.F.R. Part
435.
HCFA Ex. E. That regulation specifies at section 435.907 that
an
application must be on a form prescribed by the State (although it
does
not mention any particular form) and signed under penalty of
perjury.
The State Medical Assistance Manual states that long term care
records
"shall . . . contain" Form 50-M and that Form 50-M is "designed
for
completion by the client of the facts pertinent to his eligibility
for
medical assistance in order that a decision as to eligibility may
be
made," but does not specifically refer to application forms for
foster
care or other kinds of cases. The instructions for the Form
50-M
specify that its purpose is to gather data on resources and income
in
order to determine eligibility for long term care, home and
community
based services, and retroactive Medicaid (SSI or SSI
related). State
Ex. 58. The instructions for Form 50-M do not
mention foster care
cases. Thus, on the basis of this record, HCFA did
not establish that
the Form 50-M was the application form prescribed by the
State.
The State contended in effect that, during the period in question,
Form
770 was the prescribed application form. Supp. Affidavit, pp.
1-2;
State Ex. 23. In all 26 cases the State had executed a Form 770 on
the
child, mostly within the 12 months preceding or following the date
on
which the service was provided on which the Medicaid claim was
based.
Affidavit, pp. 5-19; State Ex. 24-49. All of these forms were
signed by
the caseworker; many were also signed by a parent or guardian
(State Ex.
24, 26, 32, 35, 36, 37, 38, 39, 40, 43, 44, 49). One was
signed by the
18-year-old child recipient. State Ex. 42. None
contain a statement
indicating that the form is signed under penalty of
perjury. The
instructions for Form 770 state that, by signing, the
parent or guardian
or applicant "agrees that the requirement for contribution
toward the
cost of evaluation(s) and residential care has been explained and
the
amount computed according to information furnished by the
applicant."
State Ex. 61a. Although the absence of a signature by a
parent or
guardian or the applicant is characterized as an "unusual
circumstance"
which should be noted by the caseworker in a "Comments" section
of the
form, it also is not necessary for anyone other than the caseworker
to
sign "[w]hen Form 770 is being completed in order to report
income
and/or resources available to a child for the purpose of
Medicaid
eligibility determination." State Ex. 61b, 23; see Supp.
Affidavit pp.
4-5. The State's Medical Assistance Manual states that a
caseworker's
signature on a form attests to the factual correctness of
the
information recorded, but there is no mention of a penalty of perjury
or
any other specific sanction. State Ex. 57.
In Louisiana, a child may be removed from the home and placed under
the
care of the State (foster care) by court order or by voluntary
agreement
of the parent(s) having custody of the child. State Ex.
60. The State
in turn arranges to place the child with foster parents
or with a
child-care facility. The State provides foster care services
without
regard to the amount or source of family income, although parents
who
are able to do so are required to make support payments to the State
on
behalf of their children. State Ex. 60, 61. In determining
such a
child's eligibility for Medicaid, the State must apply the
financial
eligibility requirements of its AFDC plan. 42 C.F.R. 435.222,
435.711.
The State must consider income and resources of the parent(s) only
if
they are actually contributed to the child from the parent(s).
42
C.F.R. 435.712(b).
The State argued that the requirement of an application signed
under
penalty of perjury has no meaning here because the caseworker has
no
incentive to lie, since the foster care children will receive
the
benefits even if not Medicaid eligible. The State also argued that
the
threat of the caseworker being fired is a more meaningful sanction
than
a possible perjury action, and thus even greater incentive, for
the
caseworker to tell the truth. The State's argument has a certain
logic.
It would be unreasonable to expect a young child to sign the form,
and
the caseworker, representing the State, is a likely person to verify
the
information on the form. Neither the regulation (42 C.F.R. 435.907)
nor
the State plan mandate that the child or a parent must sign, and
there
apparently is no bar to the caseworker being the sole
signatory.
However, the regulation clearly states that whoever signs
the
application for assistance must do so under "penalty of perjury."
The
instruction to the caseworker that, by signing, the caseworker
is
attesting to the correctness of the information suggests a
certain
formality, but the State did not establish that, by falsely
attesting,
the caseworker was subjecting her/himself to the type of penalty
the
regulation refers to. The phrase "penalty of perjury" at the very
least
suggests a penalty of a very serious nature. We cannot say
that
potential job loss satisfies this requirement simply because it is
a
possible outcome of a caseworker providing false information.
On the other hand, HCFA's response to the State's argument was
conclusory,
merely affirming that HCFA considered the Form 770
inadequate. Also,
the statement on the bottom of the Form 50, designed
by HCFA, suggests the
existence of a State law against providing false
information on a public
assistance application. Such a law might apply
to a caseworker acting
on behalf of a foster care child.
Thus, while we consider it the State's burden to establish that
it
complied with the regulation, we think that the State should have
a
limited opportunity to show either that, by signing the form,
the
caseworker subjected her/himself to a possible perjury action
or
equivalent penalty under State law, or that job loss was a
sufficiently
serious and inevitable consequence of signing an incorrect Form
770 to
meet the regulatory "penalty of perjury" requirement.
Thus, we conclude that, unless the State provides documentation (within
30
days of receiving this decision, or such longer period as HCFA
allows) to
show that the caseworker's signature was under "penalty of
perjury," the
State's documentation in the 26 cases was insufficient to
establish
eligibility and the disallowance should be upheld to the
extent it is based
on those 26 cases.
We further note, however, that the acceptability of the Form 770 as
an
application may not be determinative of eligibility with respect to
all
of the payments at issue in the 26 cases. Other factors which may
have
some bearing on the ultimate issue of whether FFP is available in
the
payments include 1) the allowable retroactive period of eligibility
for
cases where a May 1983 Form 770 functions as the initial application;
2)
the fact that HCFA has recognized that a failure to make a
timely
redetermination of eligibility would not necessarily result in
an
erroneous payment (Agency brief 3/31/88); and 3) the specific
situations
of those children for whom Form 770's were completed earlier, as
well as
in May 1983, where eligibility appears to be established absent
any
indication of a change in circumstances during the disallowance
period.
The case of E.H.
The State argued that payments for services to E.H. should not be
counted
because E.H. was found eligible in a QC sample taken for March
1983 and thus,
the State contended, E.H. was eligible when $630.87 of
the $680.87 in claims
at issue were filed in April 1983. Affidavit, pp.
20-21; Ex. 51.
The State also argued that because of the March 1983
eligibility, which HCFA
did not dispute, E.H. must have been eligible in
January 1983 when the
remaining $50 was claimed.
The State's reliance on the dates the claims were submitted
is
misplaced. Under the federal regulations, FFP is available in the
cost
of services to recipients who were eligible in the month in which
the
services were provided, not the month in which the provider
submitted
the claims for payment. 42 C.F.R. 435.1002(b). Thus,
the State's
determination in May 1983 that E.H. was eligible for March 1983
did not
support the State's claim for FFP in the cost of services rendered
in
the months from October 1982 through February 1983 because the State
did
not determine E.H. eligible for those months. The $680.87 was
properly
counted as a basis for this disallowance.
Part E. HCFA did not have a policy prohibiting
sampling-based
disallowances for these kinds of errors.
The State argued that internal HCFA memoranda contemporaneous with
the
period at issue reveal a policy restricting HCFA from calculating
the
amount of a disallowance on the basis of a sample. All of the
memoranda
emphasized that HCFA was responsible for taking disallowances
during the
period October 1982 - March 1983, on an "individually" or
"specifically"
identified or "case specific" basis. State Ex. 16, 17,
18. The State
contended that the HCFA regional administrator who issued
the
disallowance at issue "apparently misunderstood or ignored
this
directive," and cited two other examples of disallowances during
this
period, both of which were calculated on the basis of the
individual
cases and were not extrapolated to a larger universe.
State's May 6,
1988 Brief, p. 7. One of these was the Florida case, in
which the Board
upheld HCFA's authority to disallow FFP in payments for
ineligibles
during this period (discussed in Part A, above). HCFA
denied that it
had a policy prohibiting sample-based disallowances during the
period at
issue and showed that the disallowance letter had been reviewed at
the
national level. HCFA Ex. J.
The State's reliance on these memoranda is misplaced, for
several
reasons:
o Unlike the action transmittals in the AFDC program which
established
the applicable policy in DGAB No. 319 and related decisions,
these
memoranda were internal documents, not officially adopted
policy
statements.
o The State did not argue that it had been told that HCFA policy
was
not to use sampling during this period or that the State's failure
to
properly determine financial eligibility for foster care
children
resulted from reliance on such a policy.
o The memoranda focus on the fact that, during the period in
question
here, HCFA was required to disallow all individually
identified
eligibility overpayment errors found in its administration of
the
program. (The context in which this directive was issued is one in
which
HCFA was precluded from taking such disallowances during
periods
immediately before and after this period when the comprehensive
QC
disallowance systems were in effect.)
o Finally, and most important, the memoranda simply do not address
the
question here: whether HCFA could use sampling as a means
of
identifying the extent of errors resulting from a state's failure
to
make required financial eligibility determinations for an entire
class
of individuals.
Since the memoranda focus on HCFA's duty to act on erroneous
eligibility
determinations, we find reasonable HCFA's position that it did
not
intend to preclude use of audit sampling to measure the extent of
errors
caused by a state systematically failing to properly
determine
eligibility. As we noted above, a sound sampling process
produces a
result which is as least as accurate as a case-by-case count, and
which
is a more efficient and economic means by which HCFA can fulfill
its
responsibility to ensure that Medicaid funds are properly spent. As
a
practical matter, it is preferable to use a valid statistical
sampling
methodology to determine the total amount of erroneous payments
existing
in individual cases within a limited subset of Medicaid cases than
to
conduct a 100 percent case review, which is far more subject to
the
vagaries of reviewer mistakes.
Thus, we conclude that HCFA could use a valid statistical methodology
to
identify errors by extrapolation. However, as we discuss below,
HCFA's
methodology had a flaw which must be corrected for the methodology to
be
valid in this case.
Part F. The use of the point estimate is not justified here where
the
sampling error is substantially more than ten percent of the
point
estimate.
The State argued that HCFA could not validly base the disallowance
amount
on the point estimate (mid-point value) because the use of the
point estimate
violated a guideline set forth by the Acting Assistant
Inspector General for
Auditing in a memorandum to all Regional Audit
Directors, dated April 22,
1980. HCFA contended that the use of any
value between the upper and
lower limits of the 90 percent confidence
interval in this case would be
valid and thus its use of the mid-point
value was reasonable, because it did
not favor HCFA (as the upper limit
would) or the State (as the lower limit
would).
Absent any other consideration, we might agree with HCFA. After
all,
assuming the sampling methodology is otherwise valid, using the
point
estimate means the parties are equally sharing the risk that the
unknown
true value is higher or lower. However, we cannot ignore the
general,
Department-wide, audit policy which, if applied to the facts here,
would
result in use of the lower limit value (or other recalculation).
The Office of Inspector General (OIG) in HHS frequently performs audits
of
HHS-funded programs, and often uses statistical sampling. Based on
our
experience with many disputes involving sampling-based
disallowances, we take
notice of the fact that OIG auditors have
considerable experience and
substantial expertise in the application of
sampling methodology. It is
undisputed that the standards for
determining amounts of disallowances
enunciated in 1980 remain in
effect. To be sure, our findings that
certain high-value cases in the
sample may not be counted as ineligible for
purposes of this
disallowance (see our discussion in Part C, above) affect
these values,
but probably not enough to moot this issue.
Faced with a similar situation in a recent case involving these
same
parties, the Board found that HCFA had not justified the use of
the
point estimate:
The OIG standards do not appear to be specifically
binding on the
Agency here, and we suppose the
Agency could announce a policy in
regulations or
guidelines which established and explained a
different policy. But in the absence of any such general
official
stance, it would appear arbitrary for one
part of the Department to
ignore long-standing
standards of the part of the Department with
the
most experience and expertise in the use of
statistical
sampling. Thus, while the OIG
standards may not provide the only
answer to the
question of what the disallowance amount should be,
the existence of the OIG standards effectively placed a burden
on
the Agency to justify a deviation from those
standards. This in
part involves a factual
issue of how much deviation there is; here,
we note
that the amounts of the sampled claims ranged from $1.50
to
over $200.00 and that the sampling error was
substantial (as noted,
the sampling error may have
been as much as 45 percent of the point
estimate, or
over four times as much as the OIG standards would
allow). In this context, the Agency's justification for use of
the
point estimate -- which essentially was only
that the point
estimate seemed more fair -- is too
insubstantial to withstand a
challenge of
arbitrariness.
Louisiana Dept. of Human Resources, DGAB No. 979 (1988), p. 13.
Here, too, as in the earlier Louisiana case, there is
considerable
range in the amounts in the sample (from $6.24 to
$3,572.56). This
suggests that even after HCFA recalculates the upper
and lower limits,
the sampling error is likely to be substantially in excess
of ten
percent of the sampling error. If it is, our finding is that it
would
not be reasonable to base the disallowance on the point estimate
unless
HCFA reduced the sampling error to less than 10 percent of the
point
estimate. HCFA offered the same justification here as it did in
the
earlier case, and here, too, we find that justification inadequate.
If
HCFA is unable or chooses not to reduce the sampling error to
a
reasonable percentage, we uphold a disallowance calculated on use of
the
lower limit.
Part G. We uphold HCFA on all other issues.
The State also raised other issues, involving allegedly
double
disallowances, the rights of recipients, and the effect of cases
dropped
from the sample. We uphold HCFA on all three issues.
1. In discussing the disallowance policy enunciated in its
1983
internal memoranda, HCFA stated that it could disallow for
individual
cases identified by the QC review, as well as cases identified
outside
of a QC review. The State argued that if HCFA were allowed to
make
sampling-based disallowances for both types of cases (classified
by
source), this would result in disallowances being imposed
unlawfully
twice for the same error. HCFA denied this. We are not
persuaded that
the State has shown there was any real danger that HCFA would
disallow
twice for the same error, and, in any event, the State did not
allege
that HCFA did so here. HCFA's denial indicates that it is
unlikely that
the issue will ever arise, but if it does the Board will
consider it in
the circumstances of a real, not a hypothetical, case.
2. The parties discussed whether certain federal regulations
governing
the rights of recipients support the State's position that
the
recipients in the sample cases were eligible and thus the payments
were
not erroneous. These regulations (42 C.F.R. 435.911, 919, 930) set
out
procedures which the State must follow if it has unduly delayed
a
determination or terminates, discontinues, or suspends eligibility.
Here
the State did not document that it made the required determinations
of
eligibility (with the possible exception of the Form 770 cases).
Moreover,
these regulations were intended to protect recipients from
arbitrary actions
of a state, and do not excuse the State from the
consequences of erroneous
payments or authorize the payment of FFP in
erroneous payments.
3. The State originally argued that HCFA should have dropped cases
from
the universe in the same proportion as certain cases which
HCFA
allegedly dropped from the sample because they were not in the
foster
care category. However, HCFA explained that the cases were not
really
"dropped," but instead were counted as eligible. HCFA March 31,
1988
Brief, pp. 10-11. Moreover, as HCFA showed, the State was favored
by
HCFA's handling of these cases because they were counted as eligible
and
only ineligible cases influenced the calculation. If the cases
had
truly been dropped from the sample, the sample size would have
been
smaller, increasing the average erroneous payment. The State did
not
raise this issue again and we consider it no longer in the case.
Conclusion
For the reasons above stated, we uphold the disallowance in principle,
but
not in amount. The amount is subject to the following:
1. We reverse or partially reverse HCFA on six of the cases
on
which the disallowance was based.
2. We uphold HCFA provisionally on 26 other cases, subject
to
the State's submission of certain documentation, as
explained
above.
3. We uphold HCFA unconditionally on the remaining 117 cases.
4. We uphold HCFA on the use of statistical sampling
techniques
to estimate the amount of the disallowance, but reverse HCFA
on
the use of the mid-point estimate.
Thus, HCFA should recompute thee disallowance and advise the State of
the
corrected amount. Louisiana may return to the Board within thirty
(30)
days after receipt of HCFA's redetermination, if Louisiana disputes
the
amount.
________________________________ Judith A. Ballard
________________________________ Cecilia Sparks Ford
________________________________ Norval D. (John)