CRA shares practical insights gained by helping clients voluntarily disclose self-discovered evidence of potential fraud, under the guidance contained in the US Health and Human Services’ Office of Inspector General’s Health Care Fraud Self-Disclosure Protocol (SDP), as updated on November 8, 2021.
Part 1: The basics
What is a federal payer?
Disclosing parties typically choose to include Medicare, TRICARE, and Medicaid in their definition of federal payers. However, some disclosing parties have made the principled decision to exclude Medicare and Medicaid Advantage plans from the definition of federal payers because of the capitated payment structure associated with these managed care plans.
Do I have to sample every federal payer?
The SDP allows a disclosing party to draw one sample from the total population of all Federal paid claims of interest, and then allocate a pro rata share of the resulting consolidated damages to each payer in a systematic and rational way.
What patient groups should I include?
Given the different reimbursement methodologies used for inpatient claims (e.g., MS-DRG, or APR DRG codes billed for a specific diagnosis on a per-discharge basis under the Inpatient PPS) versus outpatient claims (i.e., a separate CPT/HCPCS code for each documented, medically-necessary service), it may be appropriate to include only outpatient claims in the study population.
How far back should I go?
Depending on what may have given rise to the potential non-compliance, or how the issue came to the attention of the institution, it may be defensible to consider various alternatives for the start date of the damages period.
Which coverage standards apply?
After the random sample of paid claims has been drawn, disclosing parties typically seek to locate the most applicable reimbursement rules to apply when reviewing each sampled paid claim for compliance.
Part 2: Selecting a sample
Confirm that the non-compliant conduct has ended
Before finalizing the date range for the study population and selecting a statistically valid sample of paid claims, consider performing a robust validation process to confirm that the non-compliant conduct has ended and that any new mitigating controls are effective as designed.
Select a population
A sample is typically selected from the population of paid claims, including payers and patients, during the date ranges the disclosing party determines to be most defensible.
Validate the completeness of the population
Disclosing parties will often reconcile the population selected with audited financial statements or other available “control totals” that can be triangulated in order to reasonably demonstrate that the population selected for analysis is complete.
Determine if you want the results of your sample to be statistically meaningful
The SDP requires that a sample contain at least 100 units, but that is not necessarily large enough to yield statistically-defensible results. The SDP does not require a minimum precision level or precision range, but a higher level and a tighter range will yield a more statistically-defensible set of findings.
Select a sampling unit
The sampling unit used (e.g., a claim, a claim line, a patient) will be impacted by a variety of factors (e.g., the type of service of interest, how the data are available).
Deal with missing sample items
The SDP does not allow alternate sampling units. It states that missing sample items should be treated as errors, citing federal healthcare program rules which require the retention of supporting information for submitted claims.
Part 3: Dealing with damages
A dollar saved is $1.50 earned
The Office of Inspector General has stated that its “general practice in [civil monetary penalty] settlements of SDP matters is to require a minimum multiplier of 1.5 times the single damages.” For every dollar by which single damages can be reduced, the disclosing party can expect to save at least $1.50.
Demonstrate that the sample is reasonably representative
To ensure the defensibility of the damages analysis, the disclosing party should consider analyzing its statistically valid sample to demonstrate that the sampled units are reasonably representative of the population as a whole.
Pick the best approach
Disclosing parties often consider “but-for” reimbursements as an offset to estimated total damages. Government representatives may be open to this approach as a matter of equity if it is 1) disclosed and presented in a transparent manner and 2) consistent with the underlying facts and circumstances of the situation.
Keep track of underpayment
The SDP prohibits a “…reduction, or ‘netting’ for any underpayments discovered in the review.” However, if a significant number of underpayments are noted during the review process, disclosing parties may choose to include them in the SDP report for context as the information may be relevant in their negotiations with the government regarding penalties.
Avoid double payment
The disclosing party does not need to repay improper claims that were identified during the probe sample if 1) they were included in the study population from which a statistically valid sample was drawn and 2) repayment was made based on that sample. This would result in a double payment to the government.
