RAC Audits, Extrapolation and Defensive Strategies

Transcription

1 RAC Audits, Extrapolation and Defensive Strategies RAC University, powered by edutrax February 18, 2010 Cornelia M. Dorfschmid, PH.D. Executive Vice President Strategic Management 5911 Kingstowne Village Parkway Suite 210 Alexandria, VA James R. Proctor Director KPMG LLP 303 Peachtree Street NE Suite 2000 Atlanta, GA Dinetia M. Newman, Esquire Partner Balch & Bingham LLP 401 E. Capitol Street Suite 200 Jackson, MS

2 Objectives RACs Audit Approaches and Potential Need, Opportunity or Conflict with Extrapolation Methodology Types of RAC Audits Potentially Involving Extrapolated Recovery Amounts and Provider Factors Basics of Statistical Sampling and Extrapolation Questions Providers and Suppliers Should Raise when Faced with Extrapolated Repayment Requests Use of External Resources in Preparing for and Defending Against RAC Extrapolation Attempts 2

3 Recovery Audit Contractors (RACs) In section 306 of the Medicare Prescription Drug, Improvement, and Modernization Act of 2003 (MMA), Congress directed the Department of Health and Human Services (DHHS) to conduct a 3-year demonstration program using Recovery Audit Contractors (RACs) to detect and correct improper payments in the Medicare FFS program. The Recovery Audit Contractor (RAC) demonstration program was designed to determine whether the use of RACs will be a cost-effective means of adding resources to ensure correct payments are being made to providers and suppliers and, therefore, protect the Medicare Trust Fund. The demonstration operated in New York, Massachusetts, Florida, South Carolina and California and ended on March 27, CMS Website ( 3

4 RACs: A Quick History CMS started the RAC program in 2005 in NY, CA, and FL and expanded to 3 other states in 2007 with the goal of identifying Medicare overpayments and underpayments Between March 2005 and March 2008, RACs corrected $1.03 billion improper payments, over 96% of which were overpayments RAC Program is now permanent and nationwide. RAC Program was made permanent by the Tax Relief and Health Care Act of 2006 with Nationwide implementation mandated in RAC Scope of Work Revised: Claim paid dates may never be earlier than October 1, 2007 Cases may be reviewed at Day 1 Request volume must be reasonable. RACs only paid on denials that survive all levels of appeal. RACs may use extrapolation to collect overpayments. 4

5 RACs Audit Approaches and Potential Need, Opportunity or Conflict with Extrapolation Methodology Complex Review (with medical record) Written Medicare policy/article or Medicaresanctioned coding guidelines exist Allowed No written Medicare policy/article or Medicaresanctioned coding guidelines exists Allowed (often called Individual Claim Determinations ) Automated (without medical record) Coverage/Coding Determinations Written Medicare policy/article or Medicaresanctioned coding guidelines exists Certainty exists Allowed NO Certainty exists Not allowed No written Medicare policy/article or Medicaresanctioned guidelines exists Certainty exists Allowed with prior CMS approval (often called clinically unbelievable situations) NO Certainty exists Not allowed Other Determinations (duplicates, pricing mistakes, etc.) Certainty exists Allowed NO Certainty exists Not allowed 5

6 Types of RAC Audits Potentially Involving Extrapolated Recovery Amounts and Provider Factors Extrapolation used if: Sustained or high level of payment error Documented education efforts failed to correct the error Error rate applied to entire universe of claims Determination of sustained or high level payment error is not subject to administrative or judicial review Areas of Concern Medical Necessity One-day stays Inappropriate setting, e.g. Inpatient Rehabilitation Coding Errors Excisional Debridement Complex DRGs, e.g. Septicemia, Respiratory Disorders, OR procedures Education, preparation and execution are critical to success. 6

7 Basics of Statistical Sampling and Extrapolation Regulatory Framework Program Memorandum, Transmittal B , CMS Pub. 60B Statistical procedure standards; less specific than Sampling Guidelines Appendix, Medicare Carriers Manual, Part B, Part 3, Claims Processing (effective prior to January 8, 2001) Medicare Prescription Drug, Improvement and Modernization Act of 2003 (MMA), 935 Limits use of statistical sampling and extrapolation Contractors may use only when determination that high level of payment error exists or documented educational intervention has failed to correct the payment error Program Integrity Manual (PIM)(CMS Pub. No ), Chapter 3, Section Use of Statistical Sampling for Overpayment Estimation Instructions to PSC and ZPIC Benefit Integrity units and contractor medical review units on using statistical sampling to calculate and extrapolate overpayment amounts; less specific than Sampling Guidelines Appendix RACs follow PIM as required by applicable Statement of Work PIM, Chapter 1, 1.1 7

8

9 Basics of Statistical Sampling and Extrapolation When May Contractors Use Statistical Sampling? Threshold Determination: High Level of Payment Error (PIM, ) Error rate determinations by Medical Review (MR) unit, Program Safeguard Contractor (PSC), ZPIC Probe samples Data analysis Provider/supplier history Information from law enforcement investigations Allegations of wrongdoing by provider/supplier employees OIG audits or evaluations 9

10 Basics of Statistical Sampling and Extrapolation Statistical Sampling vs. Claim-by-Claim Analysis (PIM, ) Number of claims in universe and claims dollar amounts Resources available Sampling s cost effectiveness 10

11 Basics of Statistical Sampling and Extrapolation Requirements for Contractors (PIM, ) Consult with statistical expert Requirement unless sampling methodology routinely and repeatedly used Use of other statistically valid audit sampling methodologies used by law enforcement permissible Follow procedure resulting in probability sample Probability sample and results are always valid ; some may have higher precision level Select provider or supplier Select period for review E.g., the time period number of days, weeks, months or years Define universe All the relevant claims (e.g., all claims coded a specific MS-DRG) Define sampling unit May be individual claim or individual line claim or clusters of claims for a beneficiary Define sampling frame Listing of all possible sampling units covering completely the target universe 11

12

13

14

15

16 Basics of Statistical Sampling and Extrapolation Sample Selection - PIM Sampling Designs Simple Random Sampling Uses random selection method to draw a fixed number of sampling units from the frame without duplication (example: using deck of cards and dealing certain number of cards) Systematic Sampling Involves numbering sampling units, starting with a random unit and selecting units at a fixed interval (e.g., every 10 th unit starting with 8 : 8,18,28,38, etc. until the universe is exceeded) Stratified sampling Classifying sampling units in non-overlapping groups such as where overpayment amounts in each strata are as similar as possible Cluster sampling Involves random sample of clusters; all the claims for a beneficiary would be a cluster Precision of estimated overpayment may be reduced Design combinations Two or more of above methods 16

17

18

19

20

21 Basics of Statistical Sampling and Extrapolation Sampling Designs (cont.) Random Number Selection PSC, ZPIC, RAC contractor documents random numbers and documents the program May use reputable software such as RAT-STATS Sample Size Important but not only determinant of overpayment estimate s precision Documentation of Sampling Methodology PSCs, ZPICs and other contractors must maintain documentation of the methodology used 21

22

23

24 Basics of Statistical Sampling and Extrapolation Example of Statistical Sampling and Extrapolation from OIG review of Medicare Part B claims processed by National Government Services for New York providers during period January 1, 2003 through December 31, 2005 Dated April ts/region2/ pdf 24

25 Basics of Statistical Sampling and Extrapolation Sample Design and Methodology Population Part B claims valued at $10,000 or more paid with DOS in CY Sampling Frame Access file with 942 Part B claims; total reimbursement - $11,720,424 Sample Unit Part B claim to Medicare beneficiary Sample Design - simple random sample Sample Size 100 high-dollar Part B claims Source of Random Numbers: RAT- STATS 2007 Method of selecting items: claims sequentially numbered, selected claims correlated to random numbers and list of 100 items selected 25

26 Basics of Statistical Sampling and Extrapolation 26

27 Questions Providers and Suppliers Should Raise when Faced with Extrapolated Repayment Requests Anticipating and Managing the Risk Of recovery audit itself, of spill over risk Thinking about extrapolation and statistics comes at two stages A: Record Request Stage B: Demand Letter Triggered by Recovery Audit Understanding the underlying statistical concepts and potential financial exposure is part of a good risk management strategy & successful appeal strategy. Don t assume statistical portion of the government audit is necessarily correct and justifies the recovery amount. 27

28 Questions Providers and Suppliers Should Raise when Faced with Extrapolated Repayment Requests A: Claims and Record Requests The Beginning? 85% of overpayments in Demo RAC for inpatient claims CMS believes RAC will continue focusing on hospitals and health systems with auditing activities: Inpatient claims in complex reviews are most susceptible to need/threat of extrapolation Record Request Limits (10%/max 200) and Sample Sizes Complex Review Record Request can be the beginning of Overpayment Extrapolation. Questions to Ask: What Universe? Random Sample Involved? Could records be composed to make up a Full Sample? How many would that be? 28

29 Questions Providers and Suppliers Should Raise when Faced with Extrapolated Repayment Requests B: Recovery Audit Demand Letter Arrives The Continuation In post payment audits CMS will audit a small sample and if it finds overpayment, CMS will extrapolate to population. MMA sets limits when statistical extrapolation may be used CMS sets guidelines for its Medicare contractors when performing an audit based on probability sampling: Guidelines can work in your favor, ask the right questions. If guidelines are not followed and extrapolation is flawed, total dollars may be only actual alleged overpayment and not extrapolated overpayment. If a sample is not valid, extrapolation is flawed. 29

30 Questions Providers and Suppliers Should Raise when Faced with Extrapolated Repayment Requests A probability sample is required by CMS, and statistical expertise is needed to ensure that a statistically valid sample is drawn. Centers of Medicare & Medicaid, Medicare Program Integrity Manual, Chapter 3 Verifying Potential Errors and Taking Corrective Actions, 3.10 Use of Statistical Sampling for Overpayment Estimation. Probability sample: fairness and condition for valid projection. Probe Sample/Discovery Sample -> Full Sample->Extrapolation Be aware of OIG Financial Error Rate 5% and ramifications. Challenging sample validity is key. Challenging the sample: Can be fastest and simplest step toward a successful first level of appeal (re-determination), i.e., stop recoupment, but May be out of immediate comfort zone of providers and attorneys. 30

31 Questions Providers and Suppliers Should Raise when Faced with Extrapolated Repayment Requests A simple rule: If a sample is not valid, any projection is invalid and hence not owed with good reason, and would need further evidence. Don t confuse sample size and sample validity. Your Response Team (Medical/Coding; Counsel; Statistical Expert) should be ready to respond Assess the statistical validity of the audit. Assess the validity of the random sampling method and estimation technique. Assess the validity of the sample actually drawn. Assess the criteria and characteristics applied compliance with coverage rules of the audit. Assess the overpayment estimate, re-estimate. 31

32 Questions Providers and Suppliers Should Raise when Faced with Extrapolated Repayment Requests Don t assume statistical portion of the government audit is necessarily correct and justifies the recovery amount. Be mindful of broader exposure: other time periods, facilities, strata, secondary payors,. Spillover effects. Ask the RAC or government auditor for clarification on methods and attempt verification via replication first. Be specific and formal in your requests, otherwise you may get general and not necessary helpful answers. Involve counsel early, ask if an independent SME provide same answers? 32

33 Questions Providers and Suppliers Should Raise when Faced with Extrapolated Repayment Requests 1. Request Information, formally and writing: Data/info needed to replicate and state that in the request. Sampling Plan (incl. Sampling Type and Unit, Sample Size) Universe and Sampling Frame Random Numbers & RN Generator, software. 2. Verify the confidence interval, point estimate & bounds. 3. Are you assessed at lower limit? This is a strategic question. 4. If sample valid and confirmed, examine claim-by-claim. 5. Re-analyze claim by claim using an independent auditor with appropriate credentials and credibility. 6. If sample valid, but claim-by-claim analysis refuted, re-estimate w/ RAT-STATS appropriate statistical formula. (Additional documentation) 7. When responding to a overpayment request based on sampling, check all statistical aspects and get expertise. 33

34 Use of External Resources in Preparing for and Defending Against RAC Extrapolation Attempts RAC Response Team Development Phase Self-asses and assemble Readiness Team that provides you access to all needed resources (Clinician, Coder & HIM, Billing Office, Compliance, Counsel, Statistical expert) Mind who the project manager will be (outside/inside) Document Request Phase Demand Letter Arrival Phase Engage Team, request their feedback to develop strategy Stages of Appeal Phase Outside legal expert, statistical, coding experts 34

35 Things to Remember Note: No RAC has used extrapolation yet, but RACs CAN and we expect that they WILL. The determination of a sustained or high level of payment error is not subject to administrative or judicial review. RACs are CMS contractors and follow CMS Program Integrity rules. Statistical portion of government audits are NOT necessarily correct. Remain vigilant. If RACs can detect high levels of payment error so can you. Keep mining your own data. Engage outside expertise early. Think of a Team Ready To Go. Small sample overpayment amounts can translate into big dollars when projected to the Universe. Records request > 30 records means increased risk of extrapolation. How far to go in appeal process is a matter of how sure you are of being right and cost/benefit analysis. 35

36 References and Useful Materials Recovery Audit Contractors and Medicare Audits: What Can Hospitals and Health Systems Expect as the RAC Program Expands Nationwide?, Member Briefing, January 2009 (see ndracscanexpectasracprogramexpandsnationwide.aspx) Office of the Inspector General, Department of Health and Human Services, Provider Self-Disclosure Protocol (1998) (see Centers for Medicare & Medicaid, Medicare Program Integrity Manual (CMS Pub ), Chapter 3 Verifying Potential Errors and Taking Corrective Actions, 3.10, Use of Statistical Sampling for Overpayment Estimation (see Office of Inspector General, Department of Health and Human Services, Review of High-Dollar Claims for Medicare Part B Payment Processed by National Government Services for New York Providers For the Period January 1, 2003 through December 31, 2005 (April 2008)(see Centers for Medicare & Medicaid, Recovery Audit Contractor, Final RAC Statement of Work (see 36

37 Thank You!! Cornelia M. Dorfschmid, PH.D. Executive Vice President Strategic Management 5911 Kingstowne Village Parkway Suite 210 Alexandria, VA James R. Proctor Director KPMG LLP 303 Peachtree Street NE Suite 2000 Atlanta, GA Dinetia M. Newman, Esquire Partner Balch & Bingham LLP 401 E. Capitol Street Suite 200 Jackson, MS