Kathy Terry, Ph'D', Sr' Director, Data Analysis

1
Advanced TrainingReap the Bene?ts of PEPPER
Beyond Red and Green
January 10, 2006

• Kathy Terry, Ph.D., Sr. Director, Data Analysis
  Evaluation
• and
• Richard Lee, MA, MPH, Sr. Data Analyst
  Medicare/Federal Healthcare Assessment, IPRO

2
Todays Presentation

• Appropriate for administrators, analytic staff,
  compliance and financial officers, HIM
  directors, and utilization directors.
• Goal is to build upon issues and topics
  discussed in the general session and offer a
  more complex analysis and discussion of PEPPER.

3
Todays Agenda

• General training synopsis
• CMS target areas in depth
• Additional reports
• Outlier value and formula
• Suggested actions for outliers and non-outliers
• The payment error rate
• Q A

4
General Training Recap

• PEPPER The Program for Evaluating Payment
  Patterns Electronic Report.
• HPMP and PEPPER an effort to reduce the
  national payment error rate.
• PEPPER
• Presents the pattern of payments made to your
  hospital from CMS compared to the rest of the
  hospitals in your state.
• Focus is only acute care, prospective payment
  system (PPS), short stay inpatient hospitals.

5
General Training Recap

• PEPPER and auditing are part of an Office of the
  Inspector General (OIG) recommended compliance
  program.
• seeking correctly documented and billed Medicare
  charges.
• PEPPER helps hospitals prioritize their on-going
  auditing tasks
• guiding current and future auditing,
• monitoring (identification and prevention of
  payment errors).

6
General Training Recap

• PEPPER is a report on past administrative claims
  data.
• PEPPER does not identify your hospitals payment
  errors.
• PEPPER indicates which hospitals within the state
  are outliers in terms of the volume of claims
  paid by CMS.
• Concentrates on CMS target areas that are
  at risk for payment errors.

7
CMS Target Areas

• CMS has identified 13 target areas that have
  evidenced payment errors.
• national analysis found these areas high in
  either
• dollars in error or,
• proportion of payment errors.

8
Target Areas

• Each target area was selected for errors in
• DRG/Coding
• Utilization
• Billing
• Combination

9
Target Areas

• DRGs 182 and 183 one-day stays
• DRGs 296 and 297 one-day stays
• DRG 127 one-day stays
• DRG 143 one-day stays
• Seven-day readmit to same facility or elsewhere
• One-day stays excluding transfers
• Complication/Co-morbidity (CC) pairs
• Three-day skilled nursing facility
  (SNF)-qualifying admissions

• DRG 089
• DRG 014
• DRG 079
• DRG 243
• DRG 416

10
Target Area Groups

• DRG/Coding target areas
• DRG 79
• DRG 89
• DRG 416
• Complication/Co-morbidity
  (CC) pairs

11
Target Area Groups

• Utilization and/or billing error target areas
• One-day stays excluding transfers
• Seven-day readmit to same facility or elsewhere
• Three-day skilled nursing facility
  (SNF)-qualifying admissions

12
Target Area Groups

• Combination of errors target areas
• (i.e., errors in DRG/Coding, Utilization,
  billing /-)
• DRGs 182 and 183 one-day stays
• DRGs 296 and 297 one-day stays
• DRG 127 one-day stays
• DRG 143 one-day stays
• DRG 014
• DRG 243

13
Target Areas
Rationale for Selection

• Thirty-nine percent of admission errors were
  associated with one-day stays.
• DRGs 079 and 416 were selected due to the high
  dollars in error for DRG changes.
• DRG 089 was selected due to high dollars and
  volume of error for DRG changes.

14
Target Areas
Rationale for Selection

• Readmissions have been associated with payment
  errors due to billing errors, premature
  discharge, incomplete care and/or inappropriate
  readmission.

15
Target Areas
Rationale for Selection

• Three-day SNF-qualifying admissions have been
  found to have issues with medical necessity in
  several states, and data indicate that three-day
  SNF-qualifying admissions have a higher
  incidence of unnecessary admissions than other
  three-day admissions.

16
Target Areas
Rationale for Selection

• The May 4, 2005 Federal Register (Vol. 70, No.
  85, pages 23332-23333) describes the increase in
  discharges billed to the DRG with a CC in a pair
  from 61.9 in 1986 to 79.9 in 2004
  additionally, QIO reviews have indicated that
  there have been coding errors related to the
  addition of a CC that was not substantiated in
  the medical record.

17
Users Guide Containsnumerator and denominator
details for each target area.
E.g., One-day stays excluding transfers
Numerator count of discharges with length of
stay less than or equal to one day excluding
patient status of 20 (expired), 07 (left against
medical advice), or 02 (transfer to another
short-term general hospital for inpatient care).
Denominator count of all discharges excluding
patient status 02.
18
New Target Area numerator and denominator
details
Complication/comorbidity (CC) pairs
Numerator count of discharges for medical DRGs
with a CC, excluding DRGs 079/089.
Denominator count of discharges for all medical
DRG pairs, excluding DRGs 079/080/089/090.
19
New Target Area--
Three-day skilled nursing facility (SNF)
qualifying admissions.
Numerator count of discharges to a SNF with a
three-day length of stay.
Denominator count of all discharges to a SNF
(identified by patient status code of 03
(discharged or transferred to a SNF) or 61
(discharged or transferred to a swing bed)).
20
Additional Worksheets

• 1-day stay top DRGs reports
• Statewide Top 20 DRGs for One-Day Stay
  Discharges in Q3 FY 2005.
• i.e., most recent FY to date
• Hospital Top 20 DRGs for One-Day Stay Discharges
  in Q3 FY 2005.
• i.e., most recent FY to date

21
Statewide Top 20 DRGs

• What are your top 20 DRGs?
• What do they mean to you?
• How can I read this particular report in detail?
• Making the most of the data?

22
(No Transcript)
23
(No Transcript)
24
Advanced Statistical Concept Outlier Value and
Formula

• As used in the PEPPER program, the value assigned
  indicates the unusualness of a hospitals
  proportion relative to other inpatient hospitals
  within your state (acute care PPS hospitals).

25
Outlier Value and Formula

• Outliers are defined as those findings that are
  at or above the statewide 75th percentile or at
  or below the statewide 10th percentile.

Cut-points can be assigned by your QIO other than
the 75th or 10th percentile, e.g., 90th
percentile.
26
Outlier Value and Formula

• PEPPER outlier values can range from 10 to
  3.2, at the low end and from 2 to 10 at the
  high end.
• The above negative outlier values (-10 to 3.2)
  represent percentile values at or below the 10th
  percentile.
• The above positive outlier values (2 to 10)
  represent percentile values at or above the 75th
  percentile.

27
Outlier Value and Formula

• Remembering that the PEPPER outlier values can
  range from 10 to 3.2, at the low end and from 2
  to 10 at the high end we can identify the 5th and
  90th percentile by using just this outlier value.
• Specifically a value of 3.2 for a target area is
  the 90th percentile.
• And a outlier value 4.5 represents the 5th
  percentile.
• More interpretative data is on the How
  worksheet.

28
Outlier Value and Formula

• The technical formula
• Outlier Untransformed Value 100/(100
  percentile) for percentile values between 50 and
  100.
• Outlier Untransformed Value minus(100 /
  percentile) for percentile values between 1 and
  49.

29
Outlier Value and Formula

• The technical formula
• Outlier Value Square Root (Outlier Value
  Untransformed)retaining the sign of the Outlier
  Value
• e.g., Your hospital has a percentile of 55
• 100-5545
• 100/452.22
• Sq rt 2.221.49
• 1.49 is your hospitals outlier value.

30
Outlier Value and Formula

• An outlier value of 1.49 means that your hospital
  is not that unusual from the statewide
  norm/median (i.e., 50th).
• The closer to /- 10, the farther away from the
  statewide norm and thus more unusual.

31
Outlier Value x Discharge Count

• The outlier value is then multiplied by the
  discharge count for the target area
• Outlier value 1.49
• Discharge count 100
• Outlier value times discharge count 149.

32
Outlier Value and Formula

• Compare this number to other outlier value times
  discharge counts to prioritize auditing efforts
  among percentile outliers.
• This is another way of measuring the magnitude
  of the potential for payment errors for your
  hospital.
• Demonstration.

33
Suggested Actions for Outliers

• The compare worksheet and outliers.
• Sorting by outlier value.
• Or, outlier value by discharge counts column,
  secondary.
• Determine which target areas need further
  investigation and, are of most
  signi?cance for your hospital.

34
Suggested Actions for Outliers
Example One-day stays above the 75th percentile

• There could be unnecessary admissions related to
  inappropriate use of admission screening criteria
  or outpatient observation.

35
Suggested Actions for Outliers
Example One-day stays above the 75th percentile

• A sample of one-day stay cases should be reviewed
  to determine if inpatient admission was necessary
  or if care could have been provided more
  efficiently on an outpatient basis (e.g.,
  outpatient observation).

36
Suggested Actions for Outliers
Example One-day stays above the 75th percentile

• Hospitals may wish to use their own data to
  profile data to identify one-day stays (or any
  target area) sorted by DRG, physician, or
  admission source to assist in identification of
  any patterns related to one-day stays.

37
Suggested Actions for Outliers
Example One-day stays above the 75th percentile

• Hospitals may wish to evaluate whether one-day
  stays are identified for procedures that are
  designated by CMS as inpatient only or whether
  one-day stays are preceded by an outpatient
  observation stay.

38
Using PEPPER for Target Areas That Are Not
Outliers

• Subsequent to investigating outlier target areas
• Identify areas that may be future outliers,
• Track trends over time,
• Flying below the radar.
• Demonstration.

39
Auditing Methods and Root Cause Analysis

• De?ne your question,
• Take a random sample,
• Review the data,
• If errors exist, determine the source.
• Not scapegoating or witch-hunting.
• A quality improvement process.

40
National Payment Error Rate
FY 2005 estimates Net 3.63, Gross 4.85
41
National Payment Error Rate
Your hospital, the state, the nation.
Collectively, we can make a difference!
42
Contact Information

• Kathy Terry, Ph.D., Sr. Director, Data Analysis
  Evaluation, IPRO
• Richard Lee, MA, MPH, Sr. Data Analyst, IPRO
• Email kterry_at_nyqio.sdps.org
  rlee2_at_nyqio.sdps.org
• Web site(s)
• http//pepperinfo.org/
• http//jeny.ipro.org/forumdisplay.php?f53