Examination of Time Lags in Automatic Refills Kolmogorov Smirnov Statistical Test Venki Srinivasan, M.S. Lead Health Analyst MEDIC Health Integrity LLC., Easton, MD Topics Medicare Part D Program Overview MEDIC Medicare Drug Integrity Contractor for Prescription Drugs Prescription Drug Automatic Refill Proactive Analysis Kolmogorov Smirnov Statistical Test Principal Component Analysis 3 Sigma Rule Medicare Part D Program Overview
Medicare Part D Program Overview Medicare Part D is a federal program to subsidize the costs of prescription drugs for Medicare beneficiaries in the United States. It was enacted as part of the Medicare Prescription Drug, Improvement, and Modernization Act of 2003 (MMA) and went into effect on January 1, 2006. Medicare Part D Program Overview Enrollment: Beneficiaries can obtain the Medicare drug benefit through two types of private plans: beneficiaries can join a Prescription Drug Plan (PDP) for drug coverage only or they can join a Medicare Advantage plan (MA) that covers both medical services and prescription drugs (MA PD). Enrollment for most beneficiaries is voluntary. MEDIC Medicare Drug Integrity Contractor for Prescription Drugs
MEDIC Contract Overview Health Integrity, LLC, is responsible for monitoring fraud, waste, and abuse in the Medicare Prescription Drug benefit program. Fraud Detection Areas: Pharmacy Fraud Physician Fraud Beneficiary Fraud WA OR NV CA AK ID ID UT AZ MT WY CO NM VT ND ME MN SD WI SGS MI NY MA NE IA PA NJ CT RI IL IN OH DE KS MO WV VA KY OK TN NC AR Health Integrity SC MD MS AL GA TX LA FL PR HI Guam American Samoa USVI Pharmacy Fraud Areas of Concern Steering to generics instead of brand Unauthorized charge backs to patients Billing for one Medication and Dispensing another
Physician Fraud Areas of Concern Physician in collusion with a pharmacy to dispense expensive controlled substance drugs Physician prescribing expensive brand drugs rather than lower cost generic drugs. Beneficiary Fraud Areas of Concern Drug seeking and drug selling Using someone else s card Dual eligibles hopping from plan to plan to avoid limits on utilization
Automatic Refill Program is to provide convenience and ease of automatic refills and renewals to a patient who takes common maintenance medications for a chronic condition or long term therapy. Possible Fraud Scheme: A pharmacy may abuse the automatic refill system without a patient s authorization. A pharmacy can have prescriptions automatically refilled every n (n>=1) days. Typically you would see prescriptions refilled with some variance (for example, between 25 35 days for a 30 day prescription). A pharmacy that is abusing the automatic refill system would have a higher number of refills concentrated on a certain number of days than the normal pattern. Data: Medicare Part D data extracted from Integrated Data Repository (IDR) Time Lag summary extracted from IDR Mail order, Chain and Specialty pharmacies were eliminated. A pharmacy with at least 1,000 Prescription Drug Events and a maximum of 150,000 Prescription Drug Events was selected for the analysis. Time frame: January 1, 2007 to June 30, 2007
Some Definitions: Time Lag: Number of days between two consecutive prescription drug fills Three Ranges of Time Lags: Range 1: Between 5 to 15 days (mostly for Controlled Substance drugs) Range 2: Between 25 to 35 days Range 3: Between 85 to 95 days Kolmogorov Smirnov Statistical Test The Kolmogorov Smirnov Statistical Test (K S Test) is used to identify aberrant pharmacies whose automatic refill pattern is variant from the norm. Norm distribution was calculated across pharmacies in the South East Region. K S Test was used to quantify the distance between two empirical distribution functions, i.e. to compare the distribution of each individual pharmacy with the norm. K S Test is one of the most useful nonparametric methods for comparing two distributions. It is sensitive to differences in both location and shape of the empirical cumulative distribution functions. Kolmogorov Smirnov Statistic The empirical distribution function F n for n identically independently distributed observations X i is defined as where is the indicator function, equal to 1 if Xi x and equal to 0 otherwise. The Kolmogorov Smirnov statistic for a given cumulative distribution function F(x) is
Kolmogorov Smirnov Distance Graph Kolmogorov Smirnov Statistical Test Proc npar1way procedure in SAS was used to calculate the values of the Kolmogorov Smirnov statistic The option EDF outputs the values of the K S test statistic. Distances for each of the three ranges of time lags (D1,D2,D3) were obtained from the npar1way output. Principal Component Analysis Principal Component Analysis can be thought of as revealing the internal structure of the data in a way which best explains the variance in the data. This technique was applied to the three ranges of time lags to find out the most important factors. Two important factors were identified.
Pharmacy Score Uncorrected sum of squares was used to calculate a score, D, for each pharmacy. In order to produce a final statistical score for each pharmacy, 30 percent weight was given to the total number of claims for each individual pharmacy. 70% * D + 30% * Total Number of Claims per Pharmacy 3 Sigma Rule 3 Sigma Rule was applied to the pharmacy scores. It was used to select outlier pharmacies which have an aberrant automatic refill pattern that is variant from the norm. This rule states that for a normal distribution, almost all (99.7%) of the values lie within 3 standard deviations of the mean. Note: if the normality assumption is not satisfied, Chebyshev s inequality guarantees 89% of the values lie within 3 standard deviations of the mean. Findings: Ten pharmacies were identified as having an aberrant automatic refill pattern that is variant from the norm. Nine pharmacies were in the state of Florida and one was in Puerto Rico. The potential payment at risk for these ten pharmacies for the time period of Jan 1, 2007 through June 30, 2007 is approximately $10 million.
Investigations/ Outcomes Targets for investigation are chosen in conjunction with complaints received and other proactive analyses. Of the ten pharmacies identified, four investigations were initiated. Three investigations are currently in process and one closed. Pharmacy audits are vital to gather evidence related to the auto refill program. References http://www.cms.hhs.gov/prescriptiondrugcovgenin/ http://support.sas.com/documentation/cdl/en/statug/59654/ HTML/default/statug_npar1way_sect001.htm http://www.itl.nist.gov/div898/handbook/eda/section3/eda3 5g.htm http://www.physics.csbsju.edu/stats/ks test.html
Contact Information Venki Srinivasan Lead Health Analyst MEDIC (Office) 410 822 0697 x 11109 (Email) srinivasanv@healthintegrity.org