Medicaid Program and Children's Health Insurance Program (CHIP); Revisions to the Medicaid Eligibility Quality Control and Payment Error Rate Measurement Programs, 48816-48852 [2010-18582]

Agencies

• Centers for Medicare & Medicaid Services
• DEPARTMENT OF HEALTH AND HUMAN SERVICES

[Federal Register Volume 75, Number 154 (Wednesday, August 11, 2010)]
[Rules and Regulations]
[Pages 48816-48852]
From the Federal Register Online via the Government Printing Office [www.gpo.gov]
[FR Doc No: 2010-18582]



[[Page 48815]]

-----------------------------------------------------------------------

Part III





Department of Health and Human Services





-----------------------------------------------------------------------



Centers for Medicare & Medicaid Services



-----------------------------------------------------------------------



42 CFR Parts 431, 447, and 457



Medicaid Program and Children's Health Insurance Program (CHIP); 
Revisions to the Medicaid Eligibility Quality Control and Payment Error 
Rate Measurement Programs; Final Rule

Federal Register / Vol. 75 , No. 154 / Wednesday, August 11, 2010 / 
Rules and Regulations

[[Page 48816]]


-----------------------------------------------------------------------

DEPARTMENT OF HEALTH AND HUMAN SERVICES

Centers for Medicare & Medicaid Services

42 CFR Parts 431, 447, and 457

[CMS-6150-F]
RIN 0938-AP69


Medicaid Program and Children's Health Insurance Program (CHIP); 
Revisions to the Medicaid Eligibility Quality Control and Payment Error 
Rate Measurement Programs

AGENCY: Centers for Medicare & Medicaid Services (CMS), HHS.

ACTION: Final rule.

-----------------------------------------------------------------------

SUMMARY: This final rule implements provisions from the Children's 
Health Insurance Program Reauthorization Act of 2009 (CHIPRA) (Pub. L. 
111-3) with regard to the Medicaid Eligibility Quality Control (MEQC) 
and Payment Error Rate Measurement (PERM) programs. This final rule 
also codifies several procedural aspects of the process for estimating 
improper payments in Medicaid and the Children's Health Insurance 
Program (CHIP).

DATES: Effective Date: These regulations are effective on September 10, 
2010.

FOR FURTHER INFORMATION CONTACT:
Elizabeth Lindner, (410) 786-7481.
Jessica Woodard, (410) 786-9249.

SUPPLEMENTARY INFORMATION:

I. Background

A. Medicaid Eligibility Quality Control Program

    The Medicaid Eligibility Quality Control (MEQC) program is set 
forth in section 1903(u) of the Social Security Act (the Act) and 
requires States to report to the Secretary the ratio of States' 
erroneous excess payments for medical assistance to total expenditures 
for medical assistance. Section 1903(u) of the Act also sets a 3-
percent threshold for improper payments in any fiscal year and the 
Secretary may withhold payments to States based on the amount of 
improper payments that exceed the threshold. The traditional MEQC 
program is based on State reviews of Medicaid cases identified through 
a statistically reliable Statewide sample of cases selected from the 
State's eligibility files and excludes separate CHIP programs. These 
reviews are conducted to determine whether the sampled cases meet 
applicable Medicaid eligibility requirements.

B. The Improper Payments Information Act of 2002

    The Improper Payments Information Act of 2002 (IPIA), (Pub. L. 107-
300, enacted on November 26, 2002) requires the heads of Federal 
agencies to annually review programs they oversee to determine if they 
are susceptible to significant erroneous payments. If any programs are 
found to be susceptible to significant improper payments, then the 
agency must estimate the amount of improper payments, report those 
estimates to the Congress, and submit a report on actions the agency is 
taking to reduce erroneous expenditures. The IPIA directed the Office 
of Management and Budget (OMB) to provide guidance on implementation. 
OMB defines ``significant erroneous payments'' as annual erroneous 
payments in the program exceeding both 2.5 percent of program payments 
and $10 million (OMB M-06-23, Appendix C to OMB Circular A-123, August 
10, 2006). For those programs found to be susceptible to significant 
erroneous payments, Federal agencies must provide the estimated amount 
of improper payments and report on what actions the agency is taking to 
reduce them, including setting targets for future erroneous payment 
levels and a timeline by which the targets will be reached.
    The Medicaid program and the Children's Health Insurance Program 
(CHIP) were identified as programs at risk for significant erroneous 
payments. The Department of Health and Human Services (DHHS) reports 
the estimated error rates for the Medicaid and CHIP programs in its 
annual Agency Financial Report (AFR) to Congress.

C. Regulatory History

1. Medicaid Eligibility Quality Control Program
    Sections 431.800 through 431.865 set forth the regulatory 
requirements for States to conduct the annual MEQC measurement. 
Currently, the MEQC program consists of the following:
     MEQC traditional--Operating MEQC under Sec.  431.800 
through Sec.  431.865 and selecting a random sample of all Medicaid 
applicants and enrollees and reviewing them under guidance in the State 
Medicaid Manual.
     MEQC pilots--Operating MEQC under a special study or a 
target population and providing oversight to reduce and prevent errors 
and improve program administration.
     MEQC waivers--Operating MEQC as a part of a CMS approved 
section 1115 waiver and reviewing beneficiaries included in the 
research and demonstration project.
2. Payment Error Rate Measurement (PERM) Program
    Section 1102(a) of the Act authorizes the Secretary to establish 
such rules and regulations as may be necessary for the efficient 
administration of the Medicaid and CHIP programs. The Medicaid statute 
at section 1902(a)(6) of the Act and the CHIP statute at section 
2107(b)(1) of the Act require States to provide information that the 
Secretary finds necessary for the administration, evaluation, and 
verification of the States' programs. Also, section 1902(a)(27) of the 
Act (and Sec.  457.950 of the regulations) requires providers to submit 
information regarding payments and claims as requested by the 
Secretary, State agency, or both. Under the authority of these 
provisions, we published a proposed rule in the August 27, 2004 Federal 
Register (69 FR 52620) to comply with the requirements of the IPIA and 
the OMB guidance. The proposed rule set forth provisions for all States 
to annually estimate improper payments in their Medicaid and CHIP 
programs and to report the State-specific error rates for purposes of 
our computing the national improper payment estimates for these 
programs.
    In the October 5, 2005 Federal Register (70 FR 58260), we published 
an interim final rule with comment period (IFC). The IFC responded to 
public comments on the proposed rule, and informed the public of our 
national contracting strategy and of our plan to measure improper 
payments in a subset of States. Our State selection process ensures 
that a State is measured once, and only once, every 3 years for each 
program.
    In response to the public comments from the October 5, 2005 IFC, we 
published a second IFC in the August 28, 2006 Federal Register (71 FR 
51050). The IFC reiterated our national contracting strategy to 
estimate improper payments in both Medicaid and CHIP fee-for-service 
(FFS) and managed care, and set forth and invited further comments on 
State requirements for estimating improper payments due to errors in 
Medicaid and CHIP eligibility determinations. We also announced that a 
State's Medicaid and CHIP programs would be reviewed in the same year.
    In the August 31, 2007 Federal Register (72 FR 50490), we published 
a final rule for the PERM program, which implements the IPIA 
requirements. The August 31, 2007 final rule responded to the public 
comments on the August 28, 2006 IFC and finalized State requirements 
for submitting claims to the Federal contactors that conduct FFS

[[Page 48817]]

and managed care reviews. The August 31, 2007 final rule also finalized 
State requirements for conducting eligibility reviews and estimating 
payment error rates due to errors in eligibility determinations.

D. Children's Health Insurance Program Reauthorization Act of 2009

    On February 4, 2009, the Children's Health Insurance Program 
Reauthorization Act of 2009 (CHIPRA) (Pub. L. 111-3) was enacted. 
(Please note, as a result of this legislation, the program formerly 
known as the ``State Children's Health Insurance Program (SCHIP)'' is 
now referred to as the ``Children's Health Insurance Program (CHIP)''). 
Sections 203 and 601 of the CHIPRA relate to the PERM and MEQC 
programs.
    Section 203 of the CHIPRA establishes an error rate measurement 
with respect to the enrollment of children under the Express Lane 
Eligibility option. The law directs States not to include children 
enrolled using the Express Lane Eligibility option in data or samples 
used for purposes of complying with the MEQC and PERM requirements. 
Provisions for States' Express Lane Eligibility option will be set 
forth in a future rulemaking document.
    Section 601(a) of the CHIPRA provides for a 90 percent Federal 
match for CHIP expenditures related to PERM administration and excludes 
such expenditures from the 10 percent administrative cap. (Section 
2105(c)(2) of the CHIP statute gives States the ability to use an 
amount up to 10 percent of the CHIP benefit expenditures for outreach 
efforts, additional services other than the standard benefit package 
for low-income children, and administrative costs.)
    The CHIPRA requires a new PERM rule and delays any calculation of a 
PERM error rate for CHIP until 6 months after the new PERM rule is 
effective. Additionally, the CHIPRA provides that States that were 
scheduled for PERM measurement in fiscal year (FY) 2007 may elect to 
accept a CHIP PERM error rate determined in whole or in part on the 
basis of data for FY 2007, or may elect instead to consider its PERM 
measurement conducted for FY 2010 as the first fiscal year for which 
PERM applies to the State for CHIP. Similarly, the CHIPRA provides that 
States that were scheduled for PERM measurement in FY 2008 may elect to 
accept a CHIP PERM error rate determined in whole or in part on the 
basis of data for FY 2008, or may elect instead to consider its PERM 
measurement conducted for FY 2011 as the first fiscal year for which 
PERM applies to the State for CHIP.
    The CHIPRA requires that the new PERM rule include the following:
     Clearly defined criteria for errors for both States and 
providers.
     Clearly defined processes for appealing error 
determinations.
     Clearly defined responsibilities and deadlines for States 
in implementing any corrective action plans.
     A provision that the payment error rate for a State will 
not include payment errors based on a State's verification of an 
applicant's self-declaration if a State's self-declaration verification 
policies meet regulations promulgated by the Secretary or are approved 
by the Secretary.
     State-specific sample sizes for application of the PERM 
requirements to CHIP PERM.
    In addition, the CHIPRA shall harmonize the PERM and MEQC programs 
and provides States with the option to apply PERM data resulting from 
its eligibility reviews for meeting MEQC requirements and vice versa, 
with certain conditions.

E. CMS Response to the CHIPRA

    As required by the CHIPRA, we proposed revised MEQC and PERM 
provisions in the proposed rule published in the July 15, 2009 Federal 
Register (74 FR 34468).
    Section 601(b) of the CHIPRA states that ``the Secretary shall not 
calculate or publish any national or State-specific error rate based on 
the application of the payment error rate measurement (in this section 
referred to as `PERM') requirements to CHIP until after the date that 
is 6 months after the date on which a new final rule (in this section 
referred to as the `new final rule') promulgated after the date of the 
enactment of this Act and implementing such requirements in accordance 
with the requirements of subsection (c) is in effect for all States.'' 
The CHIP error rate for the FY 2008 cycle was scheduled to be published 
in the FY 2009 Agency Financial Report (in November 2009), which was 
less than 6 months after the expected promulgation and effective date 
of this new final rule. Therefore, the publication of any CHIP error 
rates for FY 2008 (for States that elect to accept FY 2008 as their 
first CHIP measurement under PERM) is delayed until at least 6 months 
after the effective date of this final rule implementing the CHIPRA 
requirements for PERM.
    As noted previously, section 601(d) of the CHIPRA provides that 
States that were scheduled for PERM measurement in FY 2007 may elect to 
accept a CHIP PERM error rate determined in whole or in part on the 
basis of data for FY 2007, or may elect instead to consider its PERM 
measurement conducted for FY 2010 as the first fiscal year for which 
PERM applies to the State for CHIP. In addition, the CHIPRA provides 
that States that were scheduled for PERM measurement in FY 2008 may 
elect to accept a CHIP PERM error rate determined in whole or in part 
on the basis of data for FY 2008, or may elect instead to consider its 
PERM measurement conducted for FY 2011 as the first fiscal year for 
which PERM applies to the State for CHIP.
    Accordingly, a State measured in the FY 2007 cycle that elects to 
accept the PERM error rate for its CHIP program determined in whole or 
in part on the basis of data for FY 2007 is required to notify us of 
its intentions through an acceptance form to be provided to all States 
in a forthcoming State Health Official letter. Similarly, a State 
measured in the FY 2008 cycle that elects to accept the PERM error rate 
for its CHIP program determined in whole or in part on the basis of 
data for FY 2008 is required to notify us of its intentions through an 
acceptance form to be provided to all States in a State Health Official 
letter. If a State measured in the FY 2007 or FY 2008 cycles elects to 
reject the CHIP PERM rate determined during those cycles, they do not 
need to notify CMS of this decision. However, information from those 
cycles will not be used to calculate the State-specific sample sizes 
and we will rely on the standard assumptions for determining sample 
size.
    It should be noted that immediately after the enactment of CHIPRA, 
we suspended all CHIP measurement cycles (FY 2008, FY 2009, and FY 
2010). Due to the timing of the publication of this final rule for 
PERM, we decided that CHIP PERM will begin again with the FY 2011 
measurement cycle and no retroactive reviews will be done for FYs 2009 
and 2010. For this reason, States measured in FY 2007 will not have FY 
2010 measured, but will be measured again in FY 2013 and will have the 
option to consider FY 2013 as their first or second measurement cycle 
for CHIP PERM as described previously.
    In order for section 601(d) of the CHIPRA to be read in harmony 
with the IPIA, which requires a PERM error rate to be calculated 
annually, we believe that the appropriate reading of section 601(d) of 
the CHIPRA, construing the law as a whole and giving effect to all 
language of the CHIPRA, is that a State may only elect to reject the 
PERM error rate for the State's CHIP program for FY 2007 or FY 2008. A 
State scheduled for PERM measurement in FY 2008 still had

[[Page 48818]]

its PERM error rate for its Medicaid program measured.
    Additionally, the FY 2009 and FY 2010 Medicaid measurements are 
proceeding with no delays as a result of the CHIPRA. The FY 2009 
Medicaid measurement was conducted according to the policies in the 
August 31, 2007 final rule (72 FR 50490) because the measurement 
process was complete prior to the publication of this rule. The FY 2010 
Medicaid measurement is currently underway; therefore, parts of the 
measurement process that have already taken place prior to the 
publication of this final rule (that is, universe submission and sample 
size determination) will not be repeated once the final rule is 
effective. However, for parts of the measurement that have yet to be 
completed (that is, medical and data processing review, error rate 
calculation, corrective action plans, etc) the policies of this final 
rule will apply. We do not intend to recalculate any Medicaid error 
rates already calculated or published prior to the effective date of 
this final rule.

II. Provisions of the Proposed Regulations and Analysis of and 
Responses to Public Comments

    As a result of the CHIPRA, we proposed a nomenclature change to 
parts 431, 447, and 457. We revised current regulatory language to 
reflect the change made by the CHIPRA to refer to the program formerly 
known as the ``State Children's Health Insurance Program (SCHIP)'' as 
the ``Children's Health Insurance Program (CHIP).'' We also proposed 
the following revisions to the current PERM provisions:

A. Sample Sizes

    Section 601(f) of the CHIPRA requires us to establish State-
specific sample sizes for application of the PERM requirements with 
respect to CHIP for fiscal years beginning with the first fiscal year 
that begins on or after the date on which the new final rule is in 
effect for all States, on the basis of such information as the 
Secretary determines appropriate. In establishing such sample sizes, 
the Secretary shall, to the greatest extent practicable: (1) Minimize 
the administrative cost burden on States under Medicaid and CHIP; and 
(2) maintain State flexibility to manage such programs.
    To comply with the IPIA, the PERM program must estimate a national 
Medicaid and a national CHIP error rate that covers the 50 States and 
District of Columbia. Consistent with OMB's precision requirements 
defined in its IPIA guidance, the estimated national error rate for 
each program must be bound by a 90 percent confidence interval of 2.5 
percentage points in either direction of the estimate. Since States 
administer Medicaid and CHIP and make payments for services rendered 
under the programs, we collect State-level information at a high level 
of confidence (the estimated error rate for a State should be bound by 
a 95 percent confidence interval of 3 percentage points in either 
direction). To estimate the national error rate, as well as State-
specific error rates, reviews are conducted in three areas for both the 
Medicaid and CHIP programs: (1) FFS; (2) managed care, and (3) program 
eligibility. The FFS and managed care reviews are referred to jointly 
as the ``claims review,'' while the program eligibility review is 
referred to as the ``eligibility review.''
    Samples of payments made on a FFS and managed care basis for the 
claims review and samples of beneficiaries for the eligibility review 
are drawn each year in order to calculate a national error rate that 
meets the precision requirements described in OMB Guidance (OMB M-06-
23, Appendix C to OMB Circular A-123, August 10, 2006). The preferred 
method is to achieve the precision goal with the smallest sample size 
possible, so as to reduce the burden on States, the Federal government, 
beneficiaries, and providers. We determined that the most efficient 
method, statistically, is to draw a sample of States and then draw a 
sample of payments from the payments made by the sampled States. The 
process for drawing a sample of States is described in detail in the 
preamble to the August 31, 2007 final rule (72 FR 50490). We did not 
propose modifications to the current approach, which samples 17 States 
per year for a PERM measurement cycle. The proposed rule addressed the 
State-specific sample sizes for samples of claims and beneficiaries 
within a State.
    In response to the new CHIPRA requirements, we proposed to add new 
Sec.  431.972, to describe more fully the claims sampling procedures 
used for the claims review. In addition, we proposed to more fully 
describe the process for establishing State-specific sample sizes for 
PERM, although we note that the execution of these responsibilities 
would remain with CMS and the Federal contractors, not with the States. 
Under the Secretary's authority at section 1102(a) of the Act and in 
order to effectively implement the IPIA, we also proposed that these 
sampling procedures apply to both Medicaid and CHIP.
    We proposed to revise Sec.  431.978 to provide additional guidance 
on State Medicaid and CHIP eligibility sample sizes by clarifying the 
process for establishing State-specific sample sizes.
1. Fee-for-Service (FFS) and Managed Care
a. Universe Definition
    In order to implement the IPIA and related requirements (OMB M-06-
23, Appendix C to OMB Circular A-123, August 10, 2006) that require 
Federal agencies to estimate the amount of improper payments in 
programs with significant erroneous payments (which includes Medicaid 
and CHIP), in the current Sec.  431.970(a)(1) we require States to 
submit ``[a]ll adjudicated FFS and managed care claims information, on 
a quarterly basis, from the review year,'' so that a sample of payments 
can be reviewed and from the review findings we can estimate the amount 
of improper payments in each program. We proposed to remove the word 
``all'' from Sec.  431.970(a)(1) because certain types of payments are 
excluded from PERM sampling and review for technical reasons. The 
methodology developed by us to measure improper payments in Medicaid 
and CHIP focuses on payments made on behalf of or for individual 
beneficiaries. Accordingly, PERM has excluded certain payments for 
services not provided to individual beneficiaries such as 
Disproportionate Share Hospital (DSH) payments to facilities, grants to 
State agencies or local health departments, and cost-based 
reconciliations to non-profit providers and Federally-Qualified Health 
Centers (FQHCs) because the basis of the payment cannot be traced back 
to an individual beneficiary. This exclusion from PERM sampling was 
further clarified through instructions issued by CMS to the States at 
the beginning of each measurement cycle starting with FY 2006.
    For the PERM claims review component, the ``claims universe'' is 
defined in the new Sec.  431.972 as including payments that were 
originally paid (paid claims) and for which payment was requested but 
denied (denied claims) during the Federal fiscal year, and for which 
there was Federal financial participation (FFP) (or would have been if 
the claim had not been denied) through Title XIX of the Act (Medicaid) 
or Title XXI of the Act (CHIP). Depending on the context in which it is 
used, the claims universe may refer to either the adjudicated FFS 
claims during the fiscal year under review, or the managed care 
capitation payments made during the fiscal year under review, for 
Medicaid or CHIP. We are reiterating our long standing

[[Page 48819]]

position that, for PERM purposes, managed care claims are payments made 
by the State to entities with comprehensive risk contracts that assume 
full or partial risk for enrolled beneficiaries. FFS claims are claims 
other than managed care claims. CMS and our contractors may assign 
certain payments to the PERM FFS or managed care universe in order to 
ensure consistency across States and across cycles. Given the wide 
range of payment methodologies employed by States for similar programs, 
as well as the fact that State definitions of FFS and managed care may 
not align with PERM definitions as described previously, CMS and our 
contractors must maintain some flexibility in assigning payments to 
either FFS or managed care.
    Due to the significant variation in State systems for processing, 
paying, and claiming reimbursement for medical services under Medicaid 
and CHIP, we did not propose to include a more specific claims universe 
description in regulation. Rather, States should refer to more detailed 
claims universe specifications that will be published by us in separate 
instructions at the beginning of each PERM measurement cycle. However, 
we proposed that States must establish controls to ensure that the FFS, 
managed care, and eligibility universes are complete and accurate. For 
example, this would include the comparisons between the PERM universes 
and the State's Form CMS-64 and Form CMS-21 financial reports. We are 
placing this requirement in the regulatory text at Sec.  431.972(a)(2).
b. Stratification
    In FY 2006, we measured only the error rate for the FFS component 
of Medicaid. To obtain the required precision levels while minimizing 
the sample size, and therefore reducing the burden on States, the 
claims universe for FFS payments for Medicaid was stratified by service 
category and a stratified random sample was drawn for each State. In FY 
2007 and beyond, we measure the error rates for Medicaid FFS, Medicaid 
managed care, CHIP FFS, and CHIP managed care separately (to the extent 
that a State has each of these programs). We also stratify each 
universe by dollars rather than service category.
    Under this stratification and sampling approach, all payments in 
each universe are sorted from largest to smallest payment amounts. The 
payments are then divided into strata such that the total payments in 
each stratum are the same. For example, if five strata are used, the 
total dollars in each stratum would equal 20 percent of the total 
dollars in the universe. The first stratum would contain the highest 
dollar-valued payments, and the last stratum would contain the smallest 
dollar-valued payments, including all zero-paid and denied claims 
(denials have a zero dollar amount, and therefore, would appear in the 
stratum with the smallest dollar values). An equal number of FFS claims 
or managed care payments are then drawn from each stratum, which means 
the sample would include proportionately more high-dollar payments and 
proportionately fewer low-dollar payments and denials, compared to 
their representation in the universe. This overweighting of higher-
dollar payments (which is taken into account when calculating error 
rates) enables us to draw a smaller sample size that has a reasonable 
probability of meeting the precision requirements, compared to a 
perfectly random sample or a sample stratified by service type. 
Similarly, it reduces the risk that a single very large claim will have 
a dominant effect on the error rate. In this manner, we reduce burden 
on States, the Federal government, beneficiaries, and providers.
c. Sample Size
    In order to establish State-specific sample sizes, we proposed that 
the annual sample size in a State's first PERM cycle (referred to as 
``initial sample'' or ``base sample'') would be 500 FFS claims and 250 
managed care payments.
    We determined this initial sample size based on the experience of 
the PERM pilot study, PERM measurement in FY 2008 and FY 2009, and our 
requirement that the estimated error rate for a State should be bound 
by a 95 percent confidence interval of 3 percentage points in either 
direction. Specifically, the sample size is calculated assuming that 
the universe is ``infinite'' and the error rate for FFS is 5 percent 
and the error rate for managed care is 3 percent. (Once the universe 
contains more than approximately 10,000 sampling units, it can be 
treated as if it were infinite. Statistically speaking, beyond a 
universe of approximately 10,000 sampling units, universe size does not 
affect sample size.) Using these assumptions and historical information 
on payment variation in FFS and managed care from previous PERM cycles, 
we have determined that an annual sample of 500 FFS and 250 managed 
care payments per State per program should meet our State-level 
precision requirements with reasonable probability.
    However, States with Medicaid or CHIP PERM universes under 10,000 
line items or capitation payments can notify us in order to have an 
annual sample size smaller than the base sample size in the initial 
PERM year or future years. While the universe can be treated as if it 
were infinite if its size exceeds 10,000 sampling units, if the total 
universe from which the total (full year) sample is drawn is less than 
10,000 sampling units, the sample size may be reduced by the finite 
population correction factor. The finite population correction is a 
statistical formula utilized to determine sample size where the 
population is considered finite rather than infinite. Starting with the 
FY 2011 measurement cycle, a State that anticipates that the total 
number of payments in the FFS or managed care universe for either 
Medicaid or CHIP will be less than 10,000 payments over the Federal 
fiscal year may notify us before the fiscal year being measured and 
include information on the anticipated universe size for their State. 
Our contractor will develop a modified sampling plan for that program 
in that State.
    The State-specific annual sample size in the base PERM year is 
based on an assumed error rate of 5 percent. If a State's actual PERM 
error rate in a cycle reveals that precision goals can be achieved in 
future PERM cycles with either lower or higher sample sizes than 
indicated by the original assumptions, sample sizes after the first 
PERM cycle may vary among States according to each State's demonstrated 
ability, based on PERM experience, to meet desired precision goals.
    In subsequent years, we will provide our contractor with 
information on each State's error rate and payment variation in the 
previous cycle. Our contractor will review each State's prior PERM 
cycle claims error rate and payment variation to determine if a smaller 
or larger claims sample size will be required to meet the precision 
goal established for that PERM cycle. Our contractor will develop a 
State-specific sample size for each program in each State. If 
information from a previous cycle is not available for a particular 
State or program within the State, the contractor will use the ``base 
sample'' size of 500 FFS claims and 250 managed care payments. For 
States measured in the FY 2007 or FY 2008 cycle that elect to accept 
their State-specific CHIP PERM error rate determined during those 
cycles, FY 2007 or FY 2008 would be considered their first PERM cycle 
for purposes of sample size calculation for CHIP. Therefore, these 
States would be considered for an adjusted sample size

[[Page 48820]]

in their next year of measurement after the publication of the new 
final rule. For States measured in the FY 2007 or FY 2008 cycle that 
elect to reject their State-specific CHIP PERM error rate determined 
during those cycles, information from those cycles would not be used to 
calculate the State-specific sample sizes, and the ``base sample'' size 
of 500 FFS claims and 250 managed care payments would be used.
    We proposed to establish a maximum sample size for Medicaid or CHIP 
FFS or managed care of 1,000 claims. Additionally, as discussed 
previously, a State with a claims universe of less than 10,000 sampling 
units in a program may notify us and the annual sample size will be 
reduced by the finite population correction factor for any PERM cycle. 
We believe that by taking into consideration prior cycle PERM error 
rates, as well as the finite population correction factor in 
establishing State-specific sample sizes, the States' administrative 
cost burden will be reduced and the program will be more manageable at 
the State level.
    We received the following comments regarding our proposed revisions 
to the FFS and managed care universe, stratification, and sample sizes.
i. Universe Definition
    Comment: Some commenters raised concerns about the proposed 
definition of the universe for the claims review component 
(``adjudicated fee-for-service (FFS) or managed care claims information 
or both, on a quarterly basis, from the review year''), referencing the 
change that removes the word ``all'' from the definition used in prior 
PERM regulations. The commenters expressed concern that this change 
materially alters the definition of the universe and of a claim, while 
others stated that the change does not go far enough in excluding 
certain types of payments, such as non-emergency medical transportation 
payment records that are not maintained at the beneficiary level, 
beneficiary-specific payments that are neither FFS or managed care, and 
offline claims from payment sources other than the Medicaid Management 
Information System (MMIS). Other commenters raised concerns that 
allowing a more comprehensive universe definition to be included in 
annual program instructions rather than regulation will lead to 
inconsistency across cycles.
    Response: The IPIA requires payments matched with Title XIX or 
Title XXI funds to be included in the PERM universes. Because CMS 
designed the PERM methodology to sample and review individual, 
beneficiary-level claims and payments, we have excluded from PERM 
certain Medicaid and CHIP payments that States do not pay at the 
beneficiary level. For example, DSH payments to facilities, grants to 
State agencies or local health departments, and cost-based 
reconciliations to non-profit providers and FQHCs are excluded from 
PERM because they cannot be directly tied to an individual beneficiary. 
These payments will continue to be excluded from PERM sampling and 
review. However, in addition to these payments, State Medicaid and CHIP 
programs may make a variety of payments for services provided to 
individual beneficiaries outside of typical FFS or capitated managed 
care arrangements, which CMS considers part of FFS or managed care 
arrangements for purposes of PERM. This language change is intended to 
give CMS the flexibility to provide clarifying guidance when working 
with individual States that have unique or complex payment structures 
for certain types of beneficiary services, while continuing to meet the 
requirements of IPIA.
    We have issued updated versions of the PERM universe and claims 
detail instructions each year in order to provide States with 
clarifying guidance on meeting the PERM statutory and regulatory 
requirements. We have not changed the fundamental definition of a PERM 
universe, and do not intend to do so through this rulemaking, as PERM 
must continue to comply with IPIA. Because State programs and payment 
structures continue to evolve, we would like to maintain the 
flexibility to continue to refine the data submission specifications to 
make them easier for the States to interpret and apply, within the 
constraint of a consistent PERM universe definition.
    Regarding the comment on measurement of aggregate payments such as 
non-emergency medical transportation payments, the regulations at Sec.  
431.958 define ``payment'' as ``any payment to a provider, insurer, or 
managed care organization for a Medicaid or CHIP beneficiary for which 
there is Medicaid or CHIP Federal financial participation.'' In some 
cases, it is appropriate and possible to break aggregate payments down 
to the beneficiary level. Additionally, because some States make more 
aggregate payments or payments not stored in the MMIS than others, 
excluding these payments would result in unequal measurement across 
States. Accordingly, we are not excluding these payments from the 
claims universe. However, we will consider developing a methodology for 
sampling and review of these payments that can be applied consistently 
across States, taking into account the many variations in State payment 
systems.
    Comment: One commenter questioned what the impact would be of 
removing the word ``all'' from the universe and raised concerns as to 
whether this change could potentially mean additional work for the 
State in producing the universe.
    Response: We appreciate the commenter's concerns. Certain types of 
payments are excluded from PERM sampling and review for technical 
reasons. Therefore, the word ``all'' was removed from Sec.  
431.970(a)(1) to more accurately reflect what States are required to 
submit. States are not required to submit all adjudicated FFS and 
managed care payments. Rather, certain types of payments, such as 
adjustments, are excluded. We do not anticipate that this change will 
have an impact on what States are required to submit for the PERM 
universe.
    Comment: One commenter expressed concern over PERM regulations, 
guidelines, and communications to providers that use language related 
to ``medical services'', ``medical documentation'' and ``medical 
review'' including ``medical necessity'' despite the fact that there 
are a variety of Medicaid and CHIP services that do not fit within the 
medical review model. The commenter stated that this discrepancy causes 
confusion for State staff and providers when identifying what 
documentation is required. The commenter believed this issue is also 
confusing due to the use of the word ``claim'' throughout documentation 
pertaining to FFS samples when a variety of services that are included 
in the review are not generated from a ``claim'' but rather considered 
a ``payment.'' The commenter recommended that PERM guidance should 
reflect this consideration and the terminology should be changed from 
``medical record review'' to ``medical and service record review'', 
including revision of communication to providers around the use of the 
word ``claim'' to include ``payment''.
    Response: The purpose of all documentation that we develop and 
provide to States and providers is intended to clarify what is required 
for the PERM reviews. If improvements can be made to further provide 
clarification, we will attempt to address these issues. In addition, we 
have added the following clarification in section II.A.1.a of this 
final rule, ``for PERM purposes, managed care claims are payments made 
by the State to entities with comprehensive risk contracts that assume 
full or partial risk for enrolled

[[Page 48821]]

beneficiaries. FFS claims are claims other than managed care claims. 
CMS and its contractors may assign certain payments to the PERM FFS or 
managed care universe in order to ensure consistency across States and 
across cycles.'' Further, we will consider reviewing current guidance 
and communications to assess where further changes should be made.
ii. Provider Fraud
    Comment: We received several comments regarding the current policy 
on claims from providers under fraud investigation. Commenters 
recommended dropping these claims from the sample. It was observed by 
the commenters that beneficiaries under fraud investigation are dropped 
from the eligibility review and dropping claims from providers under 
investigation would be consistent policy. Furthermore, commenters noted 
that certain records may no longer be available if they have been 
subpoenaed, and that the PERM request for documentation may complicate 
an investigation.
    Response: The IPIA requires Federal agencies to measure ``improper 
payments'' and does not distinguish between different types of improper 
payments (for example, unintentional errors versus fraud). Our current 
policy is to maintain claims that are from providers who are under 
fraud investigation in the universe and in the sample when those claims 
are randomly selected from the universe. If States opt to have the CMS 
contractor not request supporting documentation for the claims, so as 
not to disrupt the investigation, the claim is found to be paid in 
error.
    While we appreciate the commenter's concern, we are not adopting 
the recommendation to drop claims from providers under fraud 
investigation from the sample. We do not believe that the PERM review 
will compromise or complicate an investigation because requests for 
medical records are an expected and routine part of a provider's 
participation in the Medicaid and CHIP programs. In addition, when a 
provider is the subject of a fraud investigation, it does not 
necessarily mean that all of the claims he or she submits are the 
subject of the investigation. By dropping every claim submitted by the 
provider from the PERM review, it would mean dropping claims that 
legitimately should be considered in the error rate.
iii. Universe Stratification
    Comment: Some commenters raised concerns about the current 
stratification process adopted by CMS, in which payments are stratified 
by dollar. One commenter remarked that dollar stratification has 
resulted in an oversampling of high dollar claims and an undersampling 
of low dollar claims. Another commenter raised the concern that 
stratification by dollar value will lead to an unbalanced sample of the 
various service categories and all providers will not have an equal 
chance of being selected due to variances in the dollar value of claims 
submitted by service providers.
    Response: In addition to meeting overall national IPIA precision 
requirements, we have established criteria for the precision of the 
State-level estimates. Because of the need to measure each State's 
error rate accurately, sample sizes for the States will not be 
proportional to the State's program. Statistical theory suggests that, 
for the purpose of obtaining a given level of precision, the sample 
size is independent of the universe size once the universe exceeds 
about 10,000 units. Beginning with the FY 2007 cycle, we changed to a 
dollar stratification approach (from a service stratification approach) 
to improve the precision of the error rate estimate. By intentionally 
oversampling high dollar claims and undersampling low dollar claims, we 
were able to reduce the FFS sample size from 1,000 claims to 500 claims 
and still project error rates with a level of precision that meets OMB 
requirements. Oversampling the high dollar claims also reduces the risk 
that a single high dollar claim will have a dominant effect on the 
error rate. Although claims are sorted by dollar and divided into 
strata, a random sample is drawn from each stratum so that every claim 
has a chance of being sampled. Our primary goal in adopting the dollar 
stratification approach was to develop an efficient sampling plan that 
would allow calculation of an error rate that meets OMB precision 
requirements with the smallest possible sample, to reduce the burden on 
States, providers, and the Federal government. Because PERM estimates 
an overall payment error rate for FFS, it is not necessary or desirable 
to design a stratification approach that ensures equal representation 
of every provider or service type, as long as all payments have some 
chance of being sampled.
iv. State-Specific Sample Size
    Comment: Several commenters discussed our proposed approach to vary 
the PERM sample size by State as required by the CHIPRA. Some 
commenters interpreted the CHIPRA requirement that the Secretary 
establish State-specific sample sizes for application of the PERM 
requirements to mean that a fixed sample size for each State should be 
established, and stated that the proposed rule was in conflict with the 
CHIPRA as it did not establish a fixed sample size for any State. Some 
commenters questioned whether the maximum FFS sample size (1,000 claims 
for Medicaid and CHIP respectively) was appropriate or necessary. Other 
commenters raised concerns about the administrative challenges of 
planning around uncertain and changing sample sizes. One commenter 
suggested that the overall sample sizes should be proportional to 
program size (in most cases CHIP programs are much smaller then 
Medicaid programs, but the same number of claims and eligibility cases 
are sampled for review under PERM).
    Response: As indicated previously, we are governed not only by the 
CHIPRA but also by the IPIA and OMB guidance, which does not mandate 
certain minimum or maximum sample sizes but does require CMS to 
estimate national error rates for Medicaid and CHIP that meet certain 
precision requirements. The formula for estimating a sample size highly 
likely to meet OMB precision requirements takes three factors into 
consideration: Population size; variation in payments in the universe; 
and expected error rate. Each of these factors can be determined on a 
State-specific basis using information from a prior measurement cycle. 
Therefore, we believe that the proposed approach of calculating a 
State-specific sample size prior to the beginning of each cycle, using 
information from the prior cycle, meets the CHIPRA goals. This approach 
is consistent with the CHIPRA provision that provides the Secretary 
with flexibility to determine which information is appropriate to use 
in determining sample sizes.
    State sample sizes will be calculated to result in an unbiased 
estimate of the error rate within a certain level of precision. The 
State-level rates will be combined to calculate a national error rate 
within the IPIA-required level of precision. Variation in State sample 
sizes will not affect the calculation of the national error rate or 
comparison of the national or State rates over time (both fixed and 
State-variable sample sizes are designed to result in an unbiased 
estimate of the error rate). Smaller sample sizes will reduce the 
precision of the estimates at the State level somewhat but should have 
less effect on the precision of the national error rates (it will be 
slightly lower but it will not be a substantial change). The

[[Page 48822]]

variance in the estimates will also be slightly greater at the lower 
sample sizes.
    As the State error rates are built up from the independent 
component rates, sample sizes would be calculated for all six 
components (for example, Medicaid FFS, Medicaid managed care, Medicaid 
eligibility, CHIP FFS, CHIP managed care, and CHIP eligibility), and 
the maximum and minimum sample sizes would apply to each component 
independently (there is no overall program maximum or minimum). 
Information specific to each program and component would be used to 
estimate the State-specific sample size. That is, information from the 
Medicaid FFS error rate measurement in the previous cycle would be used 
only to calculate the sample size for Medicaid FFS measurement in the 
subsequent cycle. Therefore, a State with a high FFS error rate and a 
low managed care error rate in one cycle could see a larger FFS sample 
size and a smaller managed care sample size in the next cycle.
    The possibility of a larger than ``standard'' sample size 
(currently, 500 for FFS and 250 for managed care) is necessary because 
these sample sizes are not likely to meet the precision requirements if 
a State's rate is significantly higher than expected. (In FY 2007, 3 
Medicaid programs and 8 CHIP programs did not meet the precision 
requirement with the standard sample sizes.) Failure to meet the State-
level precision goals jeopardizes the precision of the national error 
rate. Thus, if we are to establish State-specific sample sizes it must 
evaluate all three determinants of sample size (that is, population 
size, variation in payments in the universe, and expected error rate) 
for each State and increase the sample size if the error rate is 
expected to be higher than average, based on the prior cycle findings.
    Because reviewing claims requires both staff and monetary 
resources, a maximum sample size puts a limit on expenditure. 
Statistical tests suggest that if State-level precision cannot be met 
with a sample size of 1,000 claims, it is unlikely to be met with any 
reasonable sample size (the slight increases in precision that could be 
achieved would be outweighed by the significant expense associated with 
reviewing thousands of additional claims). However, a substantial 
increase in the probability of reaching precision goals can be gained 
by increasing the sample size from 500 to 1,000, so we believe this 
maximum to be reasonable and prudent.
    Finally, while CHIP programs are typically much smaller than 
Medicaid programs, from a sampling perspective there is generally no 
difference between a small and large population (number of payments for 
claims sample, number of beneficiaries for eligibility sample). 
Specifically, a property of sampling is that, once the population size 
exceeds about 10,000, it can be treated as if it were an infinite 
population. Nearly every Medicaid and CHIP program has at least 10,000 
payments or 10,000 beneficiaries across a fiscal year, so they are all 
treated as ``infinite'' in terms of population size. As a result, the 
PERM sample sizes are driven primarily by the variation in payments in 
the universe and the expected error rate, not by program size. If a 
program does have fewer than 10,000 payments or 10,000 beneficiaries 
across a fiscal year, the expected population size can be substituted 
into the calculation to determine an appropriate sample size that will 
probably be smaller than the ``standard'' sample size.
    We recognize that sample sizes, particularly for eligibility, drive 
State resource needs. Because all of the information necessary to 
develop a State-specific sample size will be available to CMS once the 
State's error rate for the prior cycle is calculated, when CMS sends a 
State notice of its error rates at the end of a cycle, it will include 
in that notice the calculation of the sample size for the next cycle. 
This will provide States with the greatest advanced notice possible. We 
are considering developing a calculator that States can use to estimate 
potential sample sizes under a variety of scenarios.
    Comment: Several commenters asked questions about our proposed 
approach regarding base years. Commenters stated that in a base year, 
the sample size for a State will be that specified in the regulation, 
not a State-specific sample size calculated using information from a 
prior cycle (the ``base year'' is, by definition, the first cycle). 
Some commenters asked if the Medicaid error rate from FY 2007 or FY 
2008 could be used to determine State-specific sample sizes for CHIP in 
the next measurement cycle, if the State decided not to accept its CHIP 
error rate from FY 2007 or FY 2008.
    Response: The commenters are correct in that the ``base year'' 
refers to a State's first cycle, and therefore, the State would have 
sample sizes as provided in the regulation.
    The CHIPRA gives States that participated in the PERM CHIP 
measurement in FY 2007 and FY 2008 the option of accepting the payment 
error rate from that cycle or not accepting that rate and treating 
their next cycle as the first fiscal year for which the PERM 
requirements apply to the State (in effect, a new ``base year''). We 
believe it is likely that a State with a low CHIP error rate would 
choose to accept that rate, and would be likely to have a sample size 
the same as or lower than the base sample size in the next cycle. We 
believe it is likely that a State with a high CHIP error rate would 
choose not to accept that rate, and would be allowed to use the base 
sample size (500 FFS claims and 250 managed care payments), rather than 
risk having a larger sample size. As a result, for States that have 
previously participated in PERM, Medicaid and CHIP program sample sizes 
could vary from the ``base year numbers.''
    The CHIPRA does not provide a similar option for States to accept 
or reject their Medicaid error rates from previous cycles. Therefore, 
sample sizes for a State's Medicaid program will be based on the 
State's error rate from their previous cycle.
    Results from FY 2007 (the only year for which CHIP error rates were 
calculated) indicate that State CHIP rates are not necessarily closely 
correlated to State Medicaid rates: that is, 7 of the 17 States had 
Medicaid and CHIP rates that were more than three percentage points 
apart. Because of differences in error rates and payment variation 
between Medicaid and CHIP programs, information on Medicaid error rates 
cannot be used to generate sample sizes for CHIP programs.
    Comment: Several commenters inquired as to whether CMS would 
implement a minimum sample size given that the proposed regulation 
offers a maximum sample size. The commenters recommended that CMS set a 
minimum sample size in regulation in order to assist States in planning 
for resource needs.
    Response: We appreciate the commenter's recommendation to adopt a 
minimum sample size for PERM, but we are not accepting this 
recommendation at this time. To comply with the IPIA, the PERM program 
must estimate a national Medicaid and a national CHIP error rate that 
covers the 50 States and District of Columbia. Consistent with OMB's 
precision requirements defined in its IPIA guidance, the estimated 
national error rate for each program must be bound by a 90 percent 
confidence interval of 2.5 percentage points in either direction of the 
estimate. By setting a minimum sample size, we risk having sample sizes 
that are too small for States that had higher error rates in their 
subsequent PERM cycles. If the realized variation for the State is not 
as

[[Page 48823]]

favorable as the earlier history, the State's error rate will not meet 
State-level precision requirements and may, in some cases, jeopardize 
meeting national precision goals. However, the States will still have 
the potential to reduce their sample sizes based on prior years' data. 
It is our intention to work closely with our contractor and the States 
to ensure States are informed well in advance of the measurement cycle 
of their sample size for planning purposes.
    Comment: Commenters expressed concern about the amount of work and 
time it takes to complete a comparison between the PERM universe and 
the Form CMS-64 and Form CMS-21 reports. Furthermore, commenters noted 
that the differences between what States include in the Form CMS-64 and 
Form CMS-21 reports (for example, adjustments, non-beneficiary specific 
payments) and how they report the information differs greatly from the 
individual beneficiary-level claims and payment data provided in the 
PERM universes.
    Commenters also offered suggestions for changes that could be made 
to the comparison, such as adopting a threshold above which a 
comparison would be considered valid, or to use the same quarter of 
data for comparison (which would require a short delay in the PERM 
universe submission).
    Response: The Form CMS-64 and Form CMS-21 comparison is a component 
of the quality control review process to validate PERM universes, 
which, like other quality control processes, is discussed in more 
detail in the PERM universe submission instructions provided to States 
at the start of each cycle.
    The purpose of the comparison, along with the rest of the quality 
control checks States are asked to complete, is to ultimately provide 
the most accurate and complete universe of Medicaid and CHIP payments 
as possible to ensure an unbiased and accurate error rate calculation. 
The comparison is not expected to be a dollar for dollar match but 
rather a means for the State and CMS to identify if, in certain areas, 
there are significant discrepancies that could indicate that payments 
were not properly included or excluded. We have found over the previous 
PERM cycles that States often overlook Medicaid or CHIP programs which 
are processed and paid outside of MMIS and/or managed by other agencies 
and divisions when developing the PERM universes. The Form CMS-64 and 
Form CMS-21 comparison serves as a tool for both States and CMS to 
determine if all payments for services provided to individual 
beneficiaries for which the State claims Title XIX or Title XXI match 
are included. As we have found that this quality control step has 
identified potential problems with the PERM universes, we are not 
adopting any recommendations to eliminate this process. However, we 
will work with States to explore options regarding how this process can 
be more effective for States and CMS. Additionally, we will consider 
for future cycles how to provide the most detailed information possible 
about this process so States can plan and prepare accordingly. As a 
result, we are modifying Sec.  431.972 to include the requirement that 
States establish controls to ensure the FFS and managed care universes 
are accurate and complete and to require a comparison of the PERM 
universes to the Forms CMS-64 and CMS-21.
    Comment: We received a number of comments related to universe 
development and sampling issues including the following:
     One commenter stated that CMS should utilize Medicaid 
Statistical Information System (MSIS) data for the Medicaid universe 
submission and if the data is not robust enough, make changes to the 
MSIS data so it can meet PERM requirements;
     One commenter stated that CMS should only require a 
universe submission and review if the universe exceeds a pre-
established minimum threshold in terms of number of claims or total 
dollar amount;
     One commenter stated that CMS should review the current 
sampling methodology which oversamples high dollar claims to determine 
if the methodology is yielding the desired results;
     One commenter stated that CMS should provide more 
technical guidance to States for the submission of the claims universe 
data to prevent differing interpretations of the requirements.
    Response: While the MSIS data will not currently fully meet the 
requirements of PERM, we understand that States are required to pull 
similar data for several CMS initiatives, resulting in redundancies 
with already limited State resources. We are currently beginning year 
two of the minimum data set pilot for PERM, in which our contractor is 
working with a small number of States, on a voluntary basis, to review 
available data fields and determine if it would be possible to create 
one data submission that meets the needs of multiple programs.
    The IPIA and OMB guidance (OMB M-06-23, Appendix C to OMB Circular 
A-123, August 10, 2006) requires that all programs that are susceptible 
to significant erroneous payments (where the annual erroneous payments 
in the program exceed both 2.5 percent of program payments and $10 
million) must participate in the error rate measurement. Only those 
programs whose annual erroneous payments fall below this threshold may 
not be subject to the error rate measurement requirements. Therefore, a 
single State universe, no matter what the size in terms of claims and 
dollars, is not eligible for omission from the national error rate 
measurement in a given cycle.
    The current sampling methodology is yielding the desired results. 
The overweighting of higher dollar payments (which is taken into 
account when calculating error rates) enables us to draw a smaller 
sample size that has a reasonable probability of meeting the precision 
requirements, compared to a perfectly random sample or a sample 
stratified by service type. In this manner, we reduce burden on States, 
the Federal government, beneficiaries, and providers.
    Finally, we appreciate the recommendation to provide States with 
more technical guidance on claims submission. We are in the process of 
developing a PERM manual, which we envision will be a single resource 
for all PERM-related guidance. As we develop the manual and update data 
submission and eligibility instructions, we will look for ways in which 
to improve technical guidance. We are also considering adding this as a 
topic for discussion with the PERM Technical Advisory Group (TAG).
2. Eligibility
    The eligibility sampling requirements are described in Sec.  
431.978. The universe for the eligibility component is case-based, not 
claims-based. The case as a sampling unit only applies to the 
eligibility component. For PERM eligibility, the ``universe'' is the 
total number of Medicaid or CHIP cases, which, as discussed in the 
proposed rule, is comprised of all beneficiaries, both individuals and 
families. The eligibility sampling plan and procedures state that the 
total eligibility sample size must be estimated to achieve within a 3 
percent precision level at a 95 percent confidence interval for the 
eligibility component of the program.
    For PERM eligibility, the initial sample size is calculated under 
the assumption that the error rate is 5 percent and the universe is 
greater than 10,000 total cases. The estimated error rate for a State 
should be at a 95 percent confidence interval of 3 percentage points in 
either direction. This means that the desired precision requirements 
will be achieved with a high probability

[[Page 48824]]

if the actual error rate is 5 percent or less. For this reason, an 
annual sample of 504 active cases and 204 negative cases should be 
selected in a State's base PERM year to meet State-level precision 
requirements with a high probability. Appendix D of the PERM 
Eligibility Review Instructions elaborates on the theory of sample size 
at the State-level for the dollar-weighted active case error rates, and 
is on the CMS Web site at https://www.cms.gov/perm/downloads/PERM_Eligibility_Review_Guidance.pdf.
    Eligibility sampling is performed by the States, and States have 
the opportunity to adjust their eligibility sample size based on the 
eligibility error rate in the previous PERM cycle. After a State's base 
PERM year, we will determine, with input from the State, a sample size 
that will meet desired precision goals at lower or higher sample sizes 
based on the outcome of the State's previous PERM cycle. The sample 
size could either increase or decrease given the results of the 
previous review year. We proposed to establish a maximum sample size 
for eligibility at 1,000 cases. States must submit an eligibility 
sampling plan by August 1st before the fiscal year being measured and 
include a proposed sample size for their State. Our contractor will 
review and approve all eligibility sampling plans. The State must 
notify CMS that it will be using the same plan from the previous review 
year if the plan is unchanged. However, we will review State sampling 
plans from prior cycles in each PERM cycle to ensure that information 
is accurate and up-to-date. States will be asked for revisions when 
necessary.
    As in the claims universe, States with PERM eligibility universes 
under 10,000 cases can propose a reduced eligibility sample size for 
either the base year or any subsequent PERM cycle.
    Additionally, section 203 of the CHIPRA describes the State option 
to enroll children in Medicaid or CHIP based on findings of an Express 
Lane agency in order to conduct simplified eligibility determinations. 
Under sections 203(a)(1) and (2) of the CHIPRA, an error rate 
measurement will be created with respect to the enrollment of children 
under the Express Lane Eligibility option. The law directs States not 
to include children enrolled using the Express Lane Eligibility option 
starting April 1, 2009, in data or samples used for purposes of 
complying with MEQC and PERM requirements. Provisions for States' 
Express Lane option will be set forth in a future rulemaking document.
    We proposed to revise Sec.  431.814 and Sec.  431.978 to reflect 
the changes and clarifications specified previously.
    We received the following comments regarding our proposed revisions 
to the eligibility sample sizes.
    Comment: A few commenters suggested that we set minimum eligibility 
sample sizes for active and negative cases.
    Response: We cannot adopt this recommendation. By setting a minimum 
eligibility sample size, we risk having sample sizes that are too small 
to meet the IPIA's precision requirements for States that had higher 
error rates in their subsequent PERM cycles. If the realized variation 
for the State is not as favorable as the earlier history, the State's 
error rate will not meet State-level precision requirements and may, in 
some cases, jeopardize meeting national precision goals. However, the 
States will still have the potential to reduce their eligibility sample 
sizes based on prior years' data. Reduced State sample sizes will 
balance the results from the PERM sampling equations with the need to 
reliably reproduce small error rates. Sample size reductions will be 
based on a State's previous eligibility error rate in PERM or MEQC 
(depending upon the method chosen by the State for PERM), the typical 
margin of error for that previous error rate, and the results from 
simulation studies on small samples. These studies examined the point 
at which small samples cease to reliably return known small error rates 
in the targeted universes. Reduced sample sizes must also meet the 
confidence and precision requirements.
    Comment: One commenter disagreed with setting a maximum sample size 
and requiring States with eligibility error rates above the 5 percent 
standard to increase their eligibility sample size. The commenter 
recommends the sample size remain constant from cycle to cycle.
    Response: We disagree with