Health, Administration

Rule R380-400

Use of Statistical Sampling and Extrapolation

Notice of Proposed Rule

(New Rule)

DAR File No.: 35216
Filed: 09/01/2011 03:21:53 PM

RULE ANALYSIS

Purpose of the rule or reason for the change:

In their report to the Legislature, Number 2009-12, and in subsequent reports, the Office of Legislative Auditor General recommended that the State Medicaid program obtain authority to use statistical sampling to establish overpayment recoveries due the state. This rule sets forth the methodology to adopt that sampling methodology and sets standards for its use.

Summary of the rule or change:

The Department proposed an extrapolation rule that was published in February and never made effective. A committee of interested providers, the Department, and the Office of Inspector General of the Medicaid Program have met since that time and this rule is the result of that collaboration. All agree that the rule is much improved. The definitions have greater clarity, as well as the methodology. Extrapolation may only be applied for a maximum period of 36 months and the initial sample must be random. Statistical sampling techniques are authorized when: 1) a pattern of errors are discovered during an audit such that a transaction error rate greater than 10% is discovered or a dollar error rate of 5%; 2) sampling size must establish a confidence level of 95% and a confidence interval of plus or minus 5%; and 3) the estimated error rate will then be extrapolated to the universe from which the sample is drawn to establish the overpayment. All sampling will be based on a single provider and a single billing code. Detailed provider notification requirements are set forth in Section R380-400-7 of the rule. Section R380-400-8 of the rule sets forth the claims using the rule's statistical methodology which is based on sound scientific statistical principles is sufficient to satisfy the burden of establishing a claim for recovery, unless rebutted by sound statistical evidence by the provider.

State statutory or constitutional authorization for this rule:

• Section 26-18-3
• Section 26-1-5

Anticipated cost or savings to:

the state budget:

Savings are anticipated to the state budget, but the exact amount cannot be predicted.

local governments:

Local governments provide Medicaid services and may be subject to recoveries under this rule.

small businesses:

Yes, small business provide Medicaid services and may be subject to recoveries under this rule.

persons other than small businesses, businesses, or local governmental entities:

Yes, businesses provide Medicaid services and may be subject to recoveries under this rule.

Compliance costs for affected persons:

Yes, there may be significant costs to Medicaid providers to respond to audits and recoveries performed as a result of this rule.

Comments by the department head on the fiscal impact the rule may have on businesses:

Businesses may have significant costs if audits find they received inappropriate payments, but there would be no fiscal costs to businesses as a result of the rule itself. This rule was developed with the participation of representatives from medical providers who would be affected by the rule in a negotiated rulemaking process. Although medical providers are not in agreement that extrapolation is the best way to determine the amount of recoveries they are agreed on the procedures for using statistical sampling and extrapolation when it would be applied as defined in this rule. The Department appreciate the support of these providers in the negotiated rulemaking process and for their helpful suggestions in improving the rule.

David Patton, PhD, Executive Director

The full text of this rule may be inspected, during regular business hours, at the Division of Administrative Rules, or at:

Health
Administration
CANNON HEALTH BLDG
288 N 1460 W
SALT LAKE CITY, UT 84116-3231

Direct questions regarding this rule to:

• Doug Springmeyer at the above address, by phone at 801-538-6971, by FAX at 801-538-6306, or by Internet E-mail at dspringm@utah.gov

Interested persons may present their views on this rule by submitting written comments to the address above no later than 5:00 p.m. on:

10/17/2011

This rule may become effective on:

10/24/2011

Authorized by:

David Patton, Executive Director

R380. Health, Administration.

R380-400. Use of Statistical Sampling and Extrapolation.

R380-400-1. Purpose and Authority.

This rule governs the methodology for statistical sampling and extrapolation on services covered by Title XIX of the Social Security Act. This rule is authorized by Sections 26-1-5, 26-18-3, and 26-18-605.

 

R380-400-2. Definitions.

Definitions for the purposes of this rule are as follows:

(1) "Billing Codes" means the current codes that may be billed to the Department and may consist of currently used DRG Codes, CPT Codes, HCPC Codes, or other nationally or locally accepted codes.

(2) "Confidence Interval" means a range of values within which a pattern of error is statistically estimated to lie.

(3) "Confidence Level" means the probability that the value of a parameter falls within a specified range of values.

(4) "Cost Effective" means provides the greatest estimated return of recoveries for overpayments relative to cost considering the available alternatives.

(5) "Diagnostic Related Groups (DRG)" means a group of related medical conditions used to establish reimbursement.

(6) "Dollar Error Rate" means the percentage of the total dollars in the initial sample found to be overpayments to the total dollars in the initial sample.

(7) "Error Types" means overpayments with a similar cause or result. For purposes of this rule, error types are limited to the following:

a. Insufficient or no documentation to support services billed, medical necessity, diagnosis codes, or billing codes.

b. Upcoding.

c. Incorrectly Unbundled services.

d. Incorrect billing code combinations.

(8) "Extrapolation" means an estimate of overpayments in claims that lie beyond the range of observation taken from a universe of records.

(9) "Initial Sample" means a statistically valid random sample of claims from the universe of records from a period not less than three months and not more than eighteen months, used to establish a pattern of error.

(10) "Standard deviation" means a statistical measure of variability that reflects the typical deviation from the mean of a distribution.

(11) "Overpayment" means any amount paid by the Department to a provider which is in excess of the amount allowed either through fraud, waste or abuse; a mistake; the lack of appropriate documentation; billing errors; errors caused either by the department, Reviewing Agency, provider, or a mechanized claims processing system; or payments not allowed under part 1902 of the Social Security Act or in violation of state rules or federal regulations, or Federally published policies.

(12) "Pattern of Error" means a transaction error rate of 10% or more, or a dollar error rate of 5% or more, found in the initial sample.

(13) "Random Sample" means a statistically valid sample drawn from the universe of records by chance; a sample drawn in such a way that every item in the universe of records has an equal and independent chance of being included in the sample.

(14) "Review" means the process in which the Reviewing Agency will select a universe of records to be sampled to determine the appropriateness of a claim. Factors used to assess appropriateness will include medical necessity; appropriate documentation; compliance with department, state and federal program policies, rules, regulations, statutes, and laws; and adherence to contract requirements.

(15) "Reviewing Agency" means any state agency, or other entity acting on behalf of a state agency, authorized by state or federal law to perform reviews, which include samples of claims filed for a public benefit funded with state or federal funds administered by the Department.

(16) "Sampling Methodology" means the use of the sampling tool, by certified users, developed by the Texas Department of Health and Human Services version 2009, which is hereby incorporated by reference, to select a random sample from a universe of records in order to calculate a dollar error rate for means of extrapolating an overpayment in a universe of records.

(17) "Transaction Error Rate" means the percentage of claims in the sample containing overpayments to the total number of claims in the sample.

(18) "Underpayment" means any amount paid by the Department to a provider which is less than the amount allowed under part 1902 of the Social Security Act or state rules or federal regulations, or federally published policies.

(19) "Universe of Records" means the total number of claims based on a single provider and for services for a single billing code, for dates of service up to 36 months prior to the date of the review.

(20) "Risk Assessment" means the identification of the level of risk of overpayments involved with the universe of records.

 

R380-400-3. Use of Sampling Methodology.

The Reviewing Agencies' procedures for performing reviews include the use of the sampling methodology.

 

R380-400-4. Initial Review to Determine Dollar and Transaction Error Rates and Need for Extrapolation.

(1) The Reviewing Agency, based on a review, of the initial sample of claims, will determine whether a pattern of error is present.

(2) Following a review of the initial sample, if a pattern of error was found and the Reviewing Agency, at its sole discretion, concludes it is cost effective, and that the error rate lies within 2.5 standard deviations of the mean, the Reviewing Agency may proceed with extrapolation based on reviewing the results from a random sample. If the error rate of the random sample lies outside 2.5 standard deviations of the mean of the initial sample and the error rate is lower than 2.5 standard deviations from the mean of the initial sample, extrapolation shall not be applied and only those errors discovered will be considered as overpayment.

(3) When extrapolation is applied, sampling methodology will be used to extrapolate the dollar and transaction error rate within the universe of records. The statistical random sample will be of sufficient size to achieve a confidence interval of 95% and a confidence level of plus or minus 5%. The dollar and transaction error rates will be determined based on the results of the statistical sample.

 

R380-400-5. Initial Sample Size Determination.

(1) Referrals will be processed through any federally-approved fraud and abuse detection software (FADS) tool, when access to such a tool is available.

(2) The Risk Assessment will be considered "moderate" unless the risk assessment is determined to be either "high" or "low."

(3) The Risk Assessment will be considered "high" when any of the following are true:

a. The claims being considered for review are indicated to be aberrant by the use of a FADS tool, when access to such a tool is available, or by the use of any data-mining analysis.

b. The applicable provider type is classified, as of the date of the review, as "high" risk in the CFR for initially categorizing provider risk. See Federal Register/Vol. 76, No. 22/Wednesday, February 2, 2011/Rules and Regulations, pages 5895-5896, which is incorporated by reference.

c. The provider is operating during the first 12 months after signing a provider agreement.

If the provider is considered "high" risk during any period of a review, then the provider is considered "high" risk during the entire period of the review.

(4) The Risk Assessment will be considered "low" when the risk assessment has not been determined to be "high" and when all of the following are true:

a. The applicable provider type is classified, as of the date of the review, as "low" risk in the Federal Register for initially categorizing provider risk.

b. The Reviewing Agency, based on any previous review of the same provider, assumes both the dollar and transaction errors in the initial sample are likely to be below the pattern of error.

c. The Reviewing Agency, based on any previous reviews involving the same provider type, assumes both the dollar and transaction errors in the initial sample are likely to be below the pattern of error.

(5) The statistically valid sample size table for initial samples is as follows in Table 1:

 

TABLE 1

  

  Risk                  Universe of Records      Universe of Records
  
Assessment          > = 250 Claims               < 250 Claims
  
High                       100                        80
  
Moderate                    75                        60
  
Low                         50                        40

 

R380-400-6. Overpayments and Underpayments.

The dollar amount of the extrapolated overpayment will be computed by applying the dollar error rate of the statistical random sample to the total dollar amount actually paid the provider as documented from the universe of records. If the review establishes that any claims from the universe of records should have been paid at a lesser amount, then only the difference between the total amount actually paid to the provider and the lesser amount that should have been paid to the provider will be used to calculate the dollar error rate. Any underpayments discovered during a review will offset the final total dollar amount of the overpayment. The final total dollar amount of the overpayment will constitute a debt by the provider to the Department.

 

R380-400-7. Provider Notification Requirements.

(1) When extrapolation is not applied after the initial sample, notice will be sent to the provider of the following:

a. The opportunity to request a hearing.

b. The criteria used to determine the initial sample.

c. The dollar and transaction error rates.

d. The size of the sample.

e. The specific claims sampled.

f. The reason(s) for the overpayments.

g. The actual total dollar amount of the total overpayments specifically identified to be recovered.

(2) When a statistical sample has been reviewed and extrapolation has been applied, notice will be sent to the provider of the following:

a. Items (1) a. through f. above in R380-400-7 as applied to the initial sample and the statistical sample.

b. Total underpayments noted.

c. The final total dollar amount of the overpayment based on extrapolation.

 

R380-400-8. Administrative Hearing Appeals and Burden of Proof.

If a provider appeals an action of the Department or Reviewing Agency regarding a claim based on statistical sampling using this rule's methodology, the action shall be deemed to satisfy the Department's or Reviewing Agency's burden of providing evidence sufficient to establish the claim, unless rebutted by the provider.

 

KEY: Medicaid

Date of Enactment or Last Substantive Amendment: 2011

Authorizing, Implemented or Interpreted Law: 26-1-5; 26-18-3