Knee Replacement ECR® Playbook Contents

1. Methodology
   1. ECR Definition Summary. This document presents a brief summary of key ECR definitions (e.g. types of models developed, trigger codes, episode period, exclusions, etc.). It is intended to be a quick reference for key definitions, which are described fully in the Analysis Plan.
   2. Analysis Plan. The Analysis Plan provides instructions on how to construct the typical and potentially avoidable complications (PAC) episodes using health care claims data. It also describes the statistical analysis methods used to do the typical stay and typical professional modeling.
   3. All Codes. This workbook contains code definitions (using ICD-9-CM codes, HCPCS, etc.) for the episode triggers, the risk factor variables (“RF” variables), comorbidities (“M”), procedures (“P”), PAC variables (“C”), and HAC variables (“PAC”). It does not contain pharmacy variables, which are in the Pharmacy All Codes file (listed below). The Analysis Plan refers to code definitions in the All Codes workbook.
   4. Pharmacy All Codes. This workbook lists the pharmacy variables (“PH” variables) and, for each, whether claims with the pharmacy variable should be deleted or should be considered typical or PAC. The Analysis Plan refers to the Pharmacy All Codes workbook.
2. Summary Results
   1. Decision Tree. This tree diagram shows an outline of the decisions made during the process of constructing episodes.
   2. Data flow. This workbook presents the following summary data:
      • Data Flow. This table shows the number of unique patients, average costs, and total costs for the relevant, typical, and PAC buckets. The results are broken down further by stay, professional, and pharmacy. This table is also presented in the PAC analysis below.
      • Episode Allocation Graph. This graph shows the number of episodes in the typical and PAC buckets and the overlap of the buckets.
      • Dollar Allocation Graphs. These graphs show how total costs are distributed by the typical and PAC buckets.
   3. Demographics. This file shows demographic data (age, gender) for the relevant cases bucket.
   4. Risk Factor Frequencies. This document shows the frequency of each risk factor in (a) the typical stay bucket and (b) the typical professional bucket.
3. Analysis
   1. PAC Analysis. This workbook provides information about hospital acquired conditions (HACs) and PACs:
      • Data Flow. This table shows the number of unique patients, average costs, and total costs for the relevant, typical, and PAC buckets. This table also appears in the Data Flow workbook.
      • PAC Allowance Computation & Construction of ECR – Stay and Professional Separately. This table shows how to calculate the PAC allowance and the total severity-adjusted cost of ECR. Results are shown for the stay and professional components separately.
      • PAC Allowance Computation & Construction of ECR – Stay and Professional Combined. This table shows how to calculate the PAC allowance and the total severity-adjusted cost of ECR. Results are shown for the stay and professional components combined.
      • Payer Savings Graph. This graph shows how PAC dollars can be redistributed to save money on PAC care.
      • HACs & PACs Summary. This table shows the costs for HACs and additional PACs.
      • Prevalence & Cost Graphs. These graphs show the added burden of HACs and PACs as well as the prevalence and costs for specific HACs and PACs.
   2. Typical Stay Analysis Results. This spreadsheet contains the distribution of allowed amounts in the typical stay bucket. No other analyses were performed for this bucket.
   3. Typical Professional Analysis Results. This spreadsheet contains the results of analyzing the typical professional bucket. The data are presented on the model building bucket only (which is described in the analysis plan).
      • Allowed Amount. This page shows descriptive statistics (mean, standard deviation, etc.) for allowed amount.
      • Comorbidities, etc. These graphs show the distribution of the number of comorbidities, procedures, and medications per episode.
      • t tests. To determine whether there was a significant difference in allowed amount by each predictor variable, t tests were run. Descriptive statistics (count, percent, mean, standard deviation) for allowed amount by each risk factor variable are shown in addition to the t test results.
      • Model fit. This workbook shows the adjusted R-square, number of episodes used and other model fit information for the final model.
      • Model. This shows the final model including regression coefficients, standard errors, and p-values.
      • Calculations. This worksheet allows the user to compute severity-adjusted base prices for hypothetical episodes from the final model.
   4. Length of Stay and Readmissions. This workbook contains information on length of stay for relevant stays. It also shows data on readmissions: number, average costs, length of stay, most common diagnosis and procedure codes for readmissions.