On July 24, 2017, Administrative Law Judge Jeanne M. Cochran of the Minnesota Office of Administrative Hearings issued a recommended order concluding that the Minnesota Department of Human Services (DHS) is not entitled to recover an extrapolated $369,979.00 from The Lazarus Project (Lazarus), a provider of therapeutic services and support to children with autism. In the order, Judge Cochran recommends that the Commissioner of DHS limit the amount DHS may recover from Lazarus to $6,620.85.

On November 28, 2017, the Commissioner of DHS affirmed Judge Cochran’s Recommended Order.

The recommended order involved an audit of Medicaid claims submitted by Lazarus for services provided from December 1, 2007, through December 31, 2011. The audit was conducted by Health Integrity, a Medicaid contractor hired by the federal government and acting as DHS’s agent. To perform the audit, Health Integrity reviewed records relating to 150 claims for services Lazarus provided to medical assistance (MA) recipients. Based on its review of these records, Health Integrity identified an alleged overpayment amount of $6,620.85. Health Integrity then used statistical extrapolation to determine a total estimated overpayment of $369,979.00. Based on these results, the Surveillance and Integrity Review Section (SIRS) of DHS issued a Notice of Agency Action against Lazarus seeking to recover the full extrapolated amount. Lazarus appealed the finding, challenging the statistical methodology and extrapolation used by Health Integrity to perform the audit.

The contested case hearing centered on Minnesota Rule 9505.2220, the administrative rule governing SIRS’s ability to collect extrapolated recoveries from MA providers. This rule authorizes SIRS “to calculate the amount of monetary recovery from a vendor based upon extrapolation from a systematic random sample of claims submitted by the vendor and paid by the [MA] program.” Minn. R. 9505.2220, subp. 1. The rule incorporates by reference the statistical text Sampling Techniques, by W. Cochran (3rd Ed. 1977) and requires that all extrapolated recoveries proposed by SIRS be based on sampling techniques and statistical analyses performed according to the Cochran text.

At hearing, both DHS and Lazarus produced experts who testified regarding the statistical methods used by Health Integrity. DHS’s expert maintained that the sampling and extrapolation performed by Health Integrity were statistically valid and performed in accordance with Minnesota law. The statistical expert produced by Lazarus testified that Health Integrity failed to comply with material requirements of the Cochran text and identified a number of problems with the sampling methods Health Integrity used, including that the sample size was too small to produce a statistically valid result.

Based on the expert testimonies, the ALJ determined that DHS “failed to demonstrate that the sampling method used by Health Integrity was performed consistent with the procedures in the Cochran Treatise as required by Minn. R. 9505.2220, subp. 3C, or that the sampling and extrapolation was statistically valid.” In particular, the ALJ concluded that DHS failed to demonstrate that the sample size was sufficiently large to ensure a statistically valid estimate of the total overpayment to Lazarus. More specifically, the ALJ criticized Health Integrity’s decision to use a sample size of only 150 claims when a formula in the Cochran text suggested that a sample of at least 203 claims was needed to produce a valid result. The ALJ also took issue with the fact that Health Integrity selected a sample of 150 claims even though the statistical software used to perform the audit recommended a sample size of 206. Despite this recommendation, Health Integrity decided to proceed with a smaller sample size, stating its decision was intended to reduce the burden on the provider to produce more records. The ALJ was not persuaded and found that “Health Integrity provided no statistically valid reason for not sampling all 206 claims” as recommended by the software program or for failing to ensure that the sample size was sufficiently large to meet the requirements under Minnesota Rule 9505.2220.

Perhaps this decision will prompt DHS to be more cautious when relying on statistical extrapolations performed by Medicaid contractors. DHS and the Minnesota Legislature would also be wise to take a fresh look at the requirements for conducting statistical extrapolation to recover alleged overpayments from Minnesota providers.

Lazarus was represented by Fredrikson & Byron attorneys Sam Orbovich and Catherine London.

A copy of the July 24, 2017 ALJ Findings of Fact, Conclusions of Law and Recommendation is available here.