Medicare and Medicaid Audit Sampling Strategies Sampling Strategies - - PowerPoint PPT Presentation

presents

Medicare and Medicaid Audit Sampling Strategies

presents

Sampling Strategies

Creating Sampling Plans and Challenging Flawed CMS Audit Samples

A Live 90-Minute Teleconference/Webinar with Interactive Q&A

Today's panel features: Patricia L. Maykuth, Ph.D, President, Research Design Associates, Decatur, Ga. Anna M. Grizzle, Member, Bass Berry & Sims, Nashville, Tenn.

A Live 90-Minute Teleconference/Webinar with Interactive Q&A

, , y , ,

Wednesday, July 14, 2010 The conference begins at: The conference begins at: 1 pm Eastern 12 pm Central 11 am Mountain 10 am Pacific 10 am Pacific

You can access the audio portion of the conference on the telephone or by using your computer's speakers. Please refer to the dial in/ log in instructions emailed to registrations.

For CLE purposes, please let us know how many people are listening at your location by y

• closing the notification box
• and typing in the chat box your
• and typing in the chat box your

company name and the number of attendees attendees.

• Then click the blue icon beside the box

to send to send.

• If the sound quality is not satisfactory
• If the sound quality is not satisfactory

and you are listening via your computer speakers please dial 1-866-869-6667 speakers, please dial 1 866 869 6667 and enter your PIN when prompted. Otherwise, please send us a chat or e- , p mail sound@straffordpub.com immediately so we can address the problem.

• If you dialed in and have any difficulties

during the call, press *0 for assistance.

Sampling for Medicare and Medicaid Audits Sampling for Medicare and Medicaid Audits p g p g

Strategies for Creating Sampling Plans and Strategies for Creating Sampling Plans and Challenging Improper Samples Challenging Improper Samples

Pat Maykuth, Ph.D. Research Design Associates esea c es g ssoc ates pm@researchdesignassociates.com Anna M. Grizzle, Esq. agrizzle@bassberry com agrizzle@bassberry.com Bass, Berry & Sims PLC July 14, 2010

Overview Overview I. Use of Statistical Sampling for Overpayment Estimation p y II. Design and Implementation of Sampling and Extrapolation Plans Sampling and Extrapolation Plans III. Defending Against Audit Results IV R i i S li D f

• IV. Raising Sampling Defenses

5

Use of Statistical Sampling for Use of Statistical Sampling for Overpayment Estimation Overpayment Estimation Overpayment Estimation Overpayment Estimation

6

Applications of Sampling Applications of Sampling

• Used in different audits (Medicare, Medicaid,

tax, financial statements, etc.) A i t h d t l i

• Appropriate when records are too voluminous

for individual review Increased data mining increased sampling

• Increased data mining = increased sampling

and extrapolation

7

Purposes of Statistical Sampling Purposes of Statistical Sampling

• Determine if further investigation is

warranted

• Project onto period with unavailable

records records

• Estimate amount of error

8

Legal Authority for Use of Statistical Legal Authority for Use of Statistical Sampling for Overpayment Estimation

• Statistical sampling does not violate due

process “so long as extrapolation is p g p made from a representative sample and is statistically significant.” Chaves y g County Home Health Service, Inc. v. Sullivan, 931 F.2d 914 (D.C. Cir. 1991), , ( ), cert denied, 502 U.S. 1091 (1992).

9

Legal Authority for Use of Statistical Legal Authority for Use of Statistical Sampling for Overpayment Estimation

• A Medicare contractor may not use

extrapolation to determine overpayment amounts . . . . unless . . .

– There is a sustained or high level of t payment error; or – Documented educational intervention has failed to correct the payment error failed to correct the payment error 42 U.S.C. § 1395ddd(f)(3)

10 10

42 U.S.C. § 1395ddd(f)(3)

Legal Authority for Use of Statistical Legal Authority for Use of Statistical Sampling for Overpayment Estimation

• Sustained or high level of payment error determined by:

– Error rate determinations by MR unit, PSC, ZPIC – Probe samples – Data analysis – Provider/supplier history – Information from law enforcement investigations – Allegations of wrongdoing by current or former employees of provider or supplier – Audits or evaluations conducted by the OIG Medicare Program Integrity Manual § 3.10.1.4

11 11 11

Legal Authority for Use of Statistical Legal Authority for Use of Statistical Sampling for Overpayment Estimation

• Additional Factors to Consider

– Number of claims in universe – Variability between claims – Dollar values associated with claims – Available resources – Cost effectiveness of expected sampling results Medicare Program Integrity Manual § 3.10.1.4

12 12 12

Design and Implementation of Design and Implementation of Design and Implementation of Design and Implementation of Sampling and Extrapolation Plans Sampling and Extrapolation Plans

13 13

Numbers Numbers vs.

• vs. Statistics

Statistics

• Numbers can readily be manipulated and
• utcomes understood through the use of simple

math: addition subtraction multiplication math: addition, subtraction, multiplication, multiplication and division e.g., %s, differences, sums and averages.

• Statistics is branch of applied math concerned

with the collection and interpretation of quantitative data and the use of probability quantitative data and the use of probability theory to estimate universe parameters e.g. correlations, t-tests and point estimates.

14

Inferential Statistics Inferential Statistics

• A branch of applied statistics drawing

A branch of applied statistics drawing conclusions about a population from a random sample drawn from it random sample drawn from it

• Mathematical analyses that move beyond

mere description of research data to make mere description of research data to make inferences about the larger population from which the sample was drawn from which the sample was drawn.

15

Statistical Terms Statistical Terms

• Parametric Statistics
• Parameter
• Point Estimate
• Point Estimate.
• Confidence Interval
• Statistic

16

17

Precision and Lower Bound Precision and Lower Bound

Point estimate $1,000,000 Precision amount at 90% two-tailed confidence $ 140,000 $ Lower bound at 90% two-tailed confidence = Point estimate minus Precision amount $860,000 Precision percent = Precision amount divided by Point estimate 14%

18 18 18

19

Normal Curve Normal Curve

30 20 25 15 5 10 1 2 3 4 5 6 7 8 9 10 11

20

Calculated Statistics of Sample Calculated Statistics of Sample

Before After claim review Before

• Choice of methodology

– Simple

After claim review

• Calculate overpayment

– Per claim F l

p – Stratified

• Sample size estimation

based on

– For sample – Proportion of claims in error

• Calculate point estimate

based on

– Universe size – Standard deviation or

– Mean – Error rates – Precision for confidence interval

probe – Precision – Confidence interval

interval – Upper and lower CI

21

universe frame Inferential Statistics: universe frame following rules required Simple Sample definition Simple random sample Selection

• f seed

random number Applied to file that is unbiased Calculate sample size number table unbiased size Stratified Selected sample Independent observations Randomly selected Normally distributed sample Normally distributed Representative

22

D t N f D t N f St ti ti l R i t ti ti l R i Data Necessary for Data Necessary for Statistical Review tatistical Review

• Audit methodology and logs

Audit methodology and logs

• Readable electronic data files for universe,

frame sample selection sample and frame, sample selection, sample and

• verpayment at claim line level
• Sample selection methodology size
• Sample selection methodology, size

estimation, seed, file sorted as applied and

• utput
• utput
• Sufficient information to collapse data into

CCN strata or overpayment at every level

23

CCN, strata or overpayment at every level

24

Random Sample Random Sample Without It No Projections Possible Without It No Projections Possible

Independent claims

• Independent claims
• Randomly selected
• Yield a set of representative claims
• Yield a set of representative claims
• Large enough to reflect the statistical

characteristics of the universe

• Each selected claim is equally likely to be

selected as any other observation

25

26

27

28

29 29

45 50 40 45 30 35 20 25 10 15 5

30

Sky Water Ice Boat

Universe File

All claims for the provider for audit time period 152,480

Strat1

>$10 ‐ <$68

Strat2

$68 ‐ $200

Strat3

$200 ‐ $500

Strat4

> $500 >$10 ‐ <$68 107,466 $68 ‐ $200 30,387 $200 ‐ $500 10,969 > $500 3,658 Sample Sample Sample Sample

31

E T E T Error Terms Error Terms

• Variance

Variance

• Standard deviation
• Precision
• Precision
• Precision percent

C ffi i t f V i ti

• Coefficient of Variation
• Error rate
• Confidence Interval
• Sampling

32

Post Audit Statistics Post Audit Statistics

• Error rate - proportion of claims with

error

• Overpayment - difference between

amount paid and audited amount amount paid and audited amount

• Point Estimate

P i i (C V)

• Precision (CoV)

The first 2 are numbers, second 2 are stats

33

34

Representativeness Representativeness

Frame Claim Sample Claim Difference Of Frame Standard Sample Standard Difference of Standard Claim Mean Claim Mean Of Claim Means Standard Deviation Standard Deviation Standard Deviations 2003 $4,034 $3991 $ - 53 $2,562 $3446 $ +793 2004 $4,352 $4879 $ +527 $2,669 $3462 $ +793

# Claim Lines % Claims Mean Median sd % of Amount pd

Sample 1143 $3,242.78 $2,310.83 $2,574.43 A 572 50.0 $4,722.31 $4,604.60 $2,609.86 27.1 B 571 50.0 $1,760.65 $1,268.24 $1,434.60 72.8 Universe 9,884 $2,950.64 $1,905.83 $2,570.36 A 4,920 50.2 $4,549.08 $4,393.74 $2,678.46 76.7 B 4,964 49.8 $1,366.36 $1,161.12 $1,001.21 23.3

35

36

Other Statistics Other Statistics

• Parametric tests

make specific

• Non parametric

tests make few or assumptions about the population parameters that no assumptions about the underlying distribution of the parameters that characterize the underlying distribution of the and parameters of the population underlying distributions for that test the population

37

Numbers are like people Numbers are like people Numbers are like people Numbers are like people Torture them enough and they'll tell you anything they ll tell you anything

Defending Against A dit Res lts Defending Against A dit Res lts Defending Against Audit Results Defending Against Audit Results

39 39

Defend Individual Claim Denials Defend Individual Claim Denials

• Review EVERY claim for appeal

– Procedural -- Did contractor follow rules? Procedural Did contractor follow rules? – Substantive – Was claim appropriate?

• Raise legal defenses
• Raise legal defenses
• Consider use of outside experts

40 40 40

Look for Basic Errors in Calculations Look for Basic Errors in Calculations

• Were allowed claims included in
• verpayment sample calculation?

p y p

– Obtain claim line data at each appeal level

• Were calculations performed correctly

Were calculations performed correctly at each level?

41 41 41

Challenge Validity of Sampling Challenge Validity of Sampling Challenge Validity of Sampling Challenge Validity of Sampling Methodology Methodology

• Consider whether threshold determination

met

– No administrative or judicial review of determination of high level of payment error

F il t f ll i t i MPIM d

• Failure to follow requirements in MPIM does

not necessarily affect validity

• Not enough to argue better or more precise
• Not enough to argue better or more precise

methods are available

42 42 42

Challenge Validity of Sampling Challenge Validity of Sampling Challenge Validity of Sampling Challenge Validity of Sampling Methodology Methodology

• Can challenge validity of sampling

methodology based on “the actual gy statistical validity of the sample as drawn and conducted”

Medicare Program Integrity Manual § 3 10 1 1 Medicare Program Integrity Manual § 3.10.1.1

43 43 43

Challenge Validity of Sampling Challenge Validity of Sampling Challenge Validity of Sampling Challenge Validity of Sampling Methodology Methodology

• Is the sample representative and

statistically significant?

– Does the sample base represent the sampling frame? D th i th l b t – Do the errors in the sample base represent the errors in the sampling frame? – Does the sampling frame properly – Does the sampling frame properly represent the target population to which the sample is being projected?

44 44 44

Raising Sampling Defenses Raising Sampling Defenses

45 45

Medicare Five-Level Appeal Process

• Redetermination from the

Intermediary/Carrier

• Reconsideration from a Qualified

Independent Contractor

• Appeal to an administrative law judge
• Appeal to the Medicare Department

pp p Appeals Board

• Appeal to a federal district court

pp

Medicaid Appeals Process

• States pursue collection of
• verpayments under State law

p y

• Appeal rights under State law
• Audit MICs to provide support to States
• Audit MICs to provide support to States

during appeals process for their audits

Raising Sampling Defenses Raising Sampling Defenses

• Obtain all documentation related to sampling

calculations

– Consider provider’s prior audit history – Consider provider s prior audit history

• Know appeal timelines and requirements for

each appeal level

• Understand reasons for denial at each level
• f appeal
• Present reasons in written protest or position

Present reasons in written protest or position paper

• Prepare for oral testimony at hearing

48 48 48

Hold Auditor Accountable Hold Auditor Accountable Hold Auditor Accountable Hold Auditor Accountable

• Know what should be done
• To do it (not just say they did)
• Accurately follow methodology
• Select a random sample that is representative
• Adequately document the audit
• Provide data and documentation necessary to

replicate

• Accurately calculate statistic with appropriate
• Accurately calculate statistic with appropriate

precision

• Don’t change rules in the middle of the game

49

g g

Q ti Q ti Questions Questions

50 50