Administrative Memos
199908
FROM: Theodore O. Will, Chief Executive Officer
DATE: Dec 03, 1999
SUBJECT: Payment Error Prevention Program : Data Analysis Plan
IPRO CONTACTS:
Kathy Terry, Ph.D., Director, Data Analysis, Extension 316
Enclosed please find a copy of IPRO's Data Analysis Plan which describes our analytical approach for Year One of the Payment Error Prevention Program (PEPP). Please share this memorandum with all interested hospital personnel. Although IPRO plans to send future PEPP memoranda to the designated PEPP Liaison, to date, all hospitals have not responded to our request to appoint a PEPP Liaison and therefore our database is incomplete.
IPRO recognizes that the success of our plan coincides with the hospitals' ability to effect change and improve billing errors. With this end in mind, we seek your input in regard to potential areas to target for improvement.
Should you have any questions in regard to PEPP or the enclosed Data Analysis Plan, please feel free to contact Andrea Goldstein, Vice President, Health Care Assessment or Dr. Kathy Terry.
HCFA's 1999 Sixth Scope of Work
Task #4: Payment Error Prevention Program (PEPP)
IPRO Data Analysis Plan - Year One
1. Introduction:
The Sixth Scope of Work (SOW) (August 1, 1999-July 31, 2002) includes a directive by the Health Care Finance Administration (HCFA) that all Peer Review Organizations (PROs) initiate a Payment Error Prevention Program (PEPP). This new initiative was developed as a result of a 1996-1997 OIG audit, which estimated that $20 billion in incorrect payments were made for each of the reviewed years. Further, it was estimated that over 20% of these dollars were associated with inpatient services under Medicare's Prospective Payment System (PPS).
The goal of PEPP is to reduce these billing errors through a variety of methods including assessments of correct coding and determinations of medical necessity. IPRO has planned and initiated a program that focuses on both identifying these issues and resolving the source(s) of the errors. The program will operate in phases during the course of the three-year SOW. This document will present our approach to data analysis in the first year of the contract along with its associated phases. The first year program includes an assessment of the HCFA designated project areas: 1) miscoded DRG assignments and, 2) unnecessary admissions.
2. Data Analysis Plan:
The data analysis plan for Year 1 involves numerous simultaneous strategies for identifying payment errors and targeting areas for improvement. The categories below represent initial steps that will be taken as part of a comprehensive approach to PEPP analysis. These strategies are:
- Literature Review
A comprehensive literature review will be performed to identify: a) national methods and measures of assessing payment errors and, b) nationally identified problem areas (e.g. OIG reports, Medicare Compliance Reports, etc.). Current literature might point to innovative methods used for payment error detection or, specific problem MDCs, DRGs, or diagnoses. Literature review findings will be used to guide initial project selection and implementation.
- Data Mining
Data mining is an analytic method employed to produce a broad-based evaluation of data patterns. Typically data mining is performed prior to the implementation of any research program as a method of conserving resources. Markers are identified in data mining that point to areas or topics that need added evaluation. Program resources and sampling strategies then direct the topic areas that are further investigated. Data mining is especially imperative for start-up programs like PEPP, where the program directive is too broad to permit delving into all areas.
PEPP data mining will be both conservative and liberal. Conservative analyses will focus on literature identified areas to assess the evidence for similar potential problems within the NY Medicare Database. Liberal analyses will incorporate broad-based sweeps of the database to identify trends both within and across data years for New York. Traditional data mining techniques such as trend and pattern analyses will be employed.
- Innovation
The "spider" model is an innovative method of assessing DRG coding patterns. This method investigates both upcoding and downcoding by mapping DRGs within a given MDC. Conceptually, there should be an equal distribution of DRGs within an MDC except one high volume, mid-weight DRG that functions as the spider head. Associated DRGs function as equally distributed legs when accurate coding exists. Spider model analyses will be performed in areas such as cardiac MDCs to identify trends and patterns within the cardiac DRG clusters. This methodology encourages focus on both upcoded and downcoded DRGs within a cluster.
- Case Review
Historical and current case review requests and results for Medicare PPS claims will be analyzed for patterns and trends. Areas of interest include patterns within and across hospitals and DRGs. Quarterly downloads and analyses will be performed to identify individual hospitals with high error rates, specific DRGs and/or PPS hospitals with multiple requests for higher-weights with increased payment, and combinations of hospital-DRG patterns. These results may lead to complete random sample projects, probe sample abstractions or immediate educational intervention.
- Sampling Strategies
By definition, a sample is a part of the population. Data from a sample is used to make inferences and generalizations about the population from which the data was sampled. It is these inferences and generalizations that mandate that care be taken when selecting samples; the strategy itself is a crucial component of all assessments.
There are several ways to sample data with the method used contingent upon the objective behind the project. The PEPP data analysis plan will employ many sampling strategies during the course of the contract year. For projects that are speculative in nature, a probe sample will be pulled. Projects that are designed to assess patterns among hospitals will employ stratified sampling techniques. Random sampling will be used for statewide projects. The objective then is this: different conditions in the data influence the kinds of strategies used in discovering billing errors. The PEPP program will implement the most logical sampling method possible based on the individual goal for each project.
- Project Phases: Planning and Time Lines
For a program to function in the most effective manner there must be clearly defined objectives, stages, and time lines. That is, the program must function as an integrated whole where: a) each project naturally flows into the next project; b) hospitals are receiving feedback on a continual basis; and, c) interventions are ongoing. It is this integrated, multi-faceted and on-going approach that will lead to the successful completion of not only each project goal, but the entire PEPP goal.
In general all projects pass through similar stages of development, implementation, and conclusion/report generation. Incorporated within these stages are the following project phases: literature review, data mining, advisory group meeting(s), sampling, tool development and programming, reliability & validity testing, data abstraction, data cleaning and analysis, report generation, intervention and/or implementation of hospital changes, remeasurement, report generation, and finally; project assessment to define project status, e.g., completed, additional intervention strategies planned, new remeasurement date assigned, etc.
3. Preliminary Findings
- Miscoded DRG Assignments
A comprehensive literature review was performed to identify the current research in payment error detection, prevention, and resolution. Trends in the literature were identified to serve as markers for further investigation. These trends identified national patterns of upcoding for select DRG pairs. Data analysis of these pairs using the 1997-1998 NYS Medicare database will be performed to assess the possibility of upcoding at this level.
- Unnecessary/Inappropriate Admissions
The OIG's Office of Audit Services Opinion of HCFA's 1997 Financial Statement identified billing errors that were associated with unnecessary and/or inappropriate inpatient admissions for services which did not require hospitalization or which were more appropriately treated in alternate settings. Therefore, IPRO first year data analysis activities also focus on potential inappropriate/unnecessary inpatient admissions. A comprehensive analysis plan was needed to identify potential inappropriate/unnecessary admission problems within NY. A review of the current literature suggested that patterns of hospital readmissions as well as one (1) day stays are potential foci for claims data analysis.
- NYS Payment Error Data Trending Study
IPRO will use the HCFA baseline sample as a model for its data trending study. This study will include randomly selected cases from hospitals in NYS each month. IPRO will conduct data abstraction and case review (as appropriate) for issues such as admission appropriateness, discharge stability, and coding DRG validation. This ongoing sample will begin with a random pull of 750 cases representing the first ten months of fiscal year 1998-1999 (October 1, 1998-July, 1999). Each month, thereafter, there will be additional random sample pulls of 75 cases per month. These data will be aggregated at various times throughout the SOW to identify patterns in payment errors, and to assess State, as well as hospital performance improvement rates.
4. Conclusion:
IPRO is using many strategies to identify and eradicate payment errors in the Medicare inpatient claims system. Traditional methods include literature reviews, data mining and pattern analyses. Innovative methods include the spider approach.
Each project will utilize rigorous scientific practices to select samples, collect and analyze data. Additionally, all projects will follow the Abaseline-intervene-remeasure@ process to evaluate the effectiveness of the program and to bring about the desired changes.
The Year 1-PEPP Data Analysis Plan is a comprehensive, solid plan that will set the stage for the remaining two years of the 6th SOW. Additional updates and reports will be provided throughout the year.