The purpose of the payment error rate measurement (PERM) program is to measure and report an unbiased national improper payment rate for Medicaid and the State Children’s Health Insurance Program (CHIP) as required under the Improper Payments Elimination and Recovery Improvement Act of 2012 (IPERIA, P.L. 112-248). Because it would be impossible to review the accuracy of every Medicaid and CHIP payment, the Centers for Medicare & Medicaid Services (CMS) uses a statistically valid methodology approved by the Office of Management and Budget (OMB) to select small samples of payments from each state’s Medicaid and CHIP universes and then extrapolates findings from these samples to estimate the improper payment rate for the program universes.[¹]
The PERM program uses a 17-state rotational approach to measure the 50 states and the District of Columbia over a three-year period. While CMS measures each state once every three years, the national improper payment rates include findings from the most recent three-year cycle measurements.[²] With each new cycle, CMS utilizes the new findings and removes the respective cycle’s previous findings.
For the 17 states in a given cycle, CMS’s contractor selects a stratified random sample of payments from each state’s universe of payments for the review year (July through June, to align with most states’ fiscal years).[³] If a state has both fee-for-service (FFS) and managed care payments, the contractor draws samples from each universe and estimates separate component improper payment rates; these are then weighted according to expenditures. The review contractor reviews all claims and managed care payments sampled to determine if each state’s payment decisions complied with applicable federal regulations and state policies.
Beginning with the FY 2019 PERM cycle, a federal contractor also conducts eligibility reviews of beneficiaries associated with sampled FFS and managed care claims. The contractor assesses the states’ application of federal rules and the state’s documented eligibility policies and procedures.[4]
As noted above, PERM counts a payment as an error if the payment or eligibility decision did not comply with applicable federal regulations and state policies.[5] This means that improper payments include both expenditures that should not have occurred, but also include instances where there is insufficient or no documentation to support the payment or eligibility decision as proper. In addition, both the absolute value of underpayments and overpayments are considered in the estimate.
Finally, the PERM program cites as improper any amount of federal share that it is incorrect, even if the total computable amount is correct. States are required to return the federal share of any overpayment and implement corrective action plans for all errors and deficiencies. In addition, CMS develops its own corrective action plan to reduce improper payments, which can include technical assistance and additional audits.
PERM findings can be used to identify potential problem areas that can inform corrective actions. Program staff use detailed error findings to conduct root cause analysis and then develop a multi-faceted corrective action plan. OMB approves all corrective action plans and reduction targets published in the agency financial report. (Medicaid and CHIP will not be subject to reduction targets until all states have been measured under the new eligibility component.)
It is important to note that the intent of the PERM program is to measure and report an unbiased estimate of the improper payment rate for Medicaid and CHIP; it is not designed to find fraud and it is not risk-weighted towards program areas known to be more vulnerable to fraud, waste, and abuse. Program vulnerabilities and root causes identified as part of the error rate measurement can be used to inform program integrity activities and PERM findings can be used to measure the effectiveness of program integrity efforts over time.
[¹] IPERIA requires an estimated national improper payment rate bound by a 90 percent confidence interval of 2.5 percentage points in either direction of the estimate. That is, the sample must be large enough that, given standard statistical assumptions, one can be 90 percent confident that the improper payment rate for the sample is within plus or minus 2.5 percentage points of the true improper payment rate for the universe. Selecting a larger sample size can increase the confidence that the sample improper payment rate is closer to the universe improper payment rate and/or decrease the size of the range around the estimate. CMS has chosen, as an additional goal for PERM (although not required by IPERIA), to draw samples at the state level that allow an estimated state improper payment rate with a 95 percent confidence interval of 3 percentage points in either direction.
[²] To calculate a national rolling error rate using improper payment rates for individual states, each state is benchmarked to its reported payments from the year it was sampled. Then, the error and payment amounts by component are combined across all 51 states to calculate the national rolling component improper payment rates for FFS, managed care, and eligibility. The component improper payment rates are combined to form the overall national rolling improper payment rate using the same method used to combine the component-level error rates into state-level error rates. The national eligibility improper payment rate includes the findings from the 17 states in PERM cycle 1 and a proxy estimate for the 34 states (in PERM cycles 2 and 3) that have not yet been measured under the new PERM process. The seventeen states in cycle 1 that have been subject to the new PERM audit include: Arkansas, Connecticut, Delaware, Idaho, Illinois, Kansas, Michigan, Minnesota, Missouri, New Mexico, North Dakota, Ohio, Oklahoma, Pennsylvania, Virginia, Wisconsin, and Wyoming.
[³] Based on each state’s Medicaid expenditures and historical FFS and managed care improper payment data, the FFS sample size was between 302 and 1,570 claims per state, and the managed care sample size was between 38 and 242 payments per state, the eligibility FFS sample size was between 102 and 298 per state, and the eligibility managed care sample size was between 105 and 380 per state. Based on each state’s CHIP expenditures and historical FFS and managed care improper payment data, the FFS sample size was between 172 and 974 claims per state, and the managed care sample size was between 22 and 270 payments per state, the eligibility FFS sample size was between 76 and 324 per state, and the eligibility managed care sample size was between 43 and 317 per state.
[4] To catalog how states will verify eligibility, states must develop a verification plan and submit it to CMS, which will assess the plan for compliance with the eligibility regulations established following the ACA. The verification plan will serve as the basis for eligibility audits and must include information pertaining to what data sources will be used and the timing and frequency of such data checks.
[5] As the purpose of PERM is to identify improper payments, an eligibility review is conducted for the beneficiary associated with each claim sampled for the review of fee-for-service or managed care payments; PERM can only find accurate eligibility determinations or false positives. CMS does not require states to develop for review a sample of cases in which eligibility was denied (i.e., negative cases) and cannot identify false negatives as part of the PERM eligibility review. In years when states are not subject to a PERM audit, they will conduct an Medicaid Eligibility Quality Control (MEQC) pilot focused on areas not addressed through PERM, including a review of negative Medicaid and CHIP cases (42 CFR 431.812(c)).