D I S S E R T A T I O N R The Potential of Claims Data to Support the Measurement of Health Care Quality Jennifer Hicks RAND Graduate School This document was prepared as a dissertation in March 2003 in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Policy Analysis at the RAND Graduate School. The faculty committee that supervised and approved the dissertation consisted of Steven Garber (Chair), John Adams, Sandra Berry, and Elizabeth McGlynn. The RAND Graduate School dissertation series reproduces dissertations that have been approved by the student’s dissertation committee. RAND is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND® is a registered trademark. RAND’s publications do not necessarily reflect the opinions or policies of its research sponsors. © Copyright 2003 RAND All rights reserved. No part of this book may be reproduced in any form by any electronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from RAND. Published 2003 by RAND 1700 Main Street, P.O. Box 2138, Santa Monica, CA 90407-2138 1200 South Hayes Street, Arlington, VA 22202-5050 201 North Craig Street, Suite 202, Pittsburgh, PA 15213-1516 RAND URL: http://www.rand.org/ To order RAND documents or to obtain additional information, contact Distribution Services: Telephone: (310) 451-7002; Fax: (310) 451-6915; Email: order@rand.org -iii - PREFACE While widespread quality measurement relying extensively on medical records would provide more broad-based and accurate information, the associated costs currently appear prohibitive to most organizations. This economic reality provides the key motivation for this dissertation, which explores the capacity of claims data, a less costly, and albeit less informative data source, to expand the scope of quality measurement. This research examines (a) the dimensions of clinical quality that can be measured with claims data and (b) the types of supplemental information that would most increase our capacity to measure health care quality with electronic data. It is intended to be of interest to researchers, policy makers, and practitioners who are currently engaged in quality measurement activities or considering future investments in administrative or clinical information systems. This study was submitted as a dissertation to the RAND Graduate School in March 2003 in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Policy Analysis. The research was funded by the Agency for Healthcare Research and Quality under dissertation grant #1-R03-HS11721-01. -v- CONTENTS PREFACE ............................................................. iii CONTENTS .............................................................. v LIST OF FIGURES ...................................................... ix LIST OF TABLES ....................................................... xi ABSTRACT ........................................................... xiii ACKNOWLEDGEMENTS ..................................................... xv 1. INTRODUCTION ...................................................... Data Sources and Quality Measurement ............................ Purpose of the Dissertation ..................................... Organization of the Dissertation ................................ 1 2 4 5 2. CONCEPTUALIZING AND MEASURING QUALITY OF CARE .................... 7 What is quality and how is it measured? ......................... 7 What is a quality indicator? .................................... 9 Data sources for quality measurement ........................... 11 Availability of claims data .................................... 13 Whose information is included in claims data?................ 13 What information is included in claims data?................. 14 Summary: Availability of Claims Data for Quality Measurement. 20 Accuracy of CLAIMS data ........................................ 20 Agreement between medical record and claims data............. 21 Sources of Inaccuracy in Claims Data......................... 23 Summary: Accuracy of Claims Data............................. 26 Applications of CLAIMS data to measure quality ................. 27 3. ASSESSING THE FEASIBILITY OF MEASURING QUALITY WITH CLAIMS DATA . 29 Selected indicator Set: QA Tools ............................... 30 Constructing quality of care indicators ........................ 32 Conceptual model for characterizing quality measures ........... 33 Primary Means of Indicator Classification: Modality.......... 34 Secondary Means of Indicator Classification.................. 37 Summary...................................................... 43 Methods ........................................................ 43 Step 1: Identifying Data Elements............................ 43 Step 2: Classifying the Data Elements........................ 44 Step 3: Assessing the Availability of Data Elements in Claims Data ...................................................... 47 Step 4: Feasibility of Constructing Indicators with Claims Data48 Example 1: An indicator that can be constructed with claims data ...................................................... 48 Example 2: An indicator that cannot be constructed with claims data ...................................................... 50 Summary of approach to feasibility analysis.................. 52 Findings: Data element analysis ................................ 53 Availability of information on laboratory tests.............. 55 Availability of information on procedures.................... 55 -vi- Availability of information on medications................... Availability of information on diagnoses..................... Findings: indicator analysis ................................... Summary of Feasibility Assessment............................ Increasing the capacity of CLAIMS data for quality measurement . How Could Additional Information be Obtained?................ Methods...................................................... Findings and Discussion...................................... Description of effects of additional information on quality measurement. .............................................. Description of effects of additional information on quality measurement. .............................................. Summary...................................................... 4. 56 57 59 64 65 65 69 70 75 76 77 THE ACCURACY OF QUALITY MEASUREMENT WITH CLAIMS DATA ............ 79 Description of the Data ........................................ 79 HMO Data..................................................... 80 Study approach ................................................. 81 Selected Indicators.......................................... 82 Constructing the QA Tools Indicators with Claims Data........ 84 Measuring Agreement.......................................... 88 Description of Agreement about Eligibility and Scoring....... 91 Theory of agreement ............................................ 95 The Effects of Errors on Quality Measurement................. 95 Factors Affecting Agreement Between Claims and Medical Records Data ...................................................... 96 Summary of Hypotheses....................................... 102 Agreement Analysis ............................................ 103 Independent Variables....................................... 103 Dependent Variables......................................... 107 Distribution of Variables................................... 108 Bivariate Analysis of Agreement............................. 111 Multivariate Analysis....................................... 116 Scoring - Multivariate Analysis about Agreement............. 122 Summary of Multivariate Analysis of Agreement about Eligibility and Scoring .............................................. 127 Agreement About Performance Rates........................... 131 Conclusions ................................................... 136 5. MOVING QUALITY MEASUREMENT FORWARD ............................... Untapped Potential of Claims Data for Quality Measurement... What can and cannot be Measured with Claims Data?........... Building a Health Information Infrastructure................ What Additional Data Elements Contribute the Most to Quality Measurement? ............................................. 137 138 139 140 142 APPENDIX A – TABULAR RESULTS: INCREASING CAPACITY FOR QUALITY MEASUREMENT .................................................... 145 APPENDIX B – INDICATORS USED TO ASSESS AGGREEMENT ................... 147 APPENDIX C – CLAIMS DATA SPECIFICATIONS ............................. ASTHMA ........................................................ CORONARY ARTERY DISEASE ....................................... DIABETES ...................................................... 154 154 164 189 -vii- HEART FAILURE ................................................. 196 PNEUMONIA ..................................................... 211 PREVENTIVE CARE ............................................... 228 APPENDIX D – KAPPA STATISTICS ....................................... 240 APPENDIX E – SEMI-STANDARDIZED REGRESSION COEFFICIENTS .............. 242 REFERENCES .......................................................... 245 -ix- LIST OF FIGURES Figure 3.1 Figure 3.2 Figure 3.3 Figure 3.4 Figure 3.5 Figure 3.6 Figure 3.7 Figure 3.8 Figure 3.9 Figure 3.10 Figure 3.11 Figure 4.1 Figure 4.2 Relationship of Function and Modality Among 553 QA Tools Indicators Distribution of QA Tools Indicators by Function and Payment Status Type and Modality of QA Tools Indicators Distribution of QA Tools Indicators by Type and Payment Status Illustrating the 4 Steps of the Feasibility Analysis: Example 1 Illustrating the 4 Steps of the Feasibility Analysis: Example 2 Distribution of Indicators that can be Constructed with Claims Data – by Payment Status Potential Increase in Capacity for Quality Measurement – All Indicators Potential Increase in Capacity for Quality Measurement – by Modality Potential Increase in Capacity for Quality Measurement – by Function Potential Increase in Capacity for Quality Measurement – by Type of Care Schematic Overview of Analysis of Agreement between Medical Record and Claims Data Data for Agreement Analyses about Eligibility and Scoring 38 39 41 42 50 52 62 72 73 74 75 89 93 -xi- LIST OF TABLES Table 2.1 Table 2.2 Table 2.3 Table 3.1 Table 3.2 Table 3.3 Table 3.4 Table 3.5 Table 3.6 Table 3.7 Table 3.8 Table 3.9 Table 3.10 Table 3.11 Table 4.1 Table 4.2 Table 4.3 Table 4.4 Table 4.5 Table 4.6 Table 4.7 Table 4.8 Table 4.9 List of Standard Data Elements: Hospital Claims Forms List of Standard Data Elements: Outpatient Claims Forms List of Standard Data Elements: Prescription Claims Forms Adult Clinical Areas in the QA Tools System Modalities of Care: Definitions, Distribution in QA Tools, and Relationship to Billing Categories Used to Classify Data Elements Distribution of Data Element Types across Components of Indicators Number of Data Elements by Classification and Proportion Available in Claims Data Assessment of Data Elements in Claims Data: Prescription Sub-Classifications Assessment of Data Elements in Claims Data: Diagnosis Sub-Classifications Number and Proportion of Indicators that can be Constructed with Claims Data – by Modality Number and Proportion of Indicators that can be Constructed with Claims Data – by Function of Care Number and Proportion of Indicators that can be Constructed with Claims Data – by Type of Care Potential Increase in Capacity for Quality Measurement – All Indicators Descriptive Statistics of Study Sample Description of Indicators Used in Agreement Analysis Standardized Codes Used to Identify People Meeting Eligibility and Scoring Criteria – Example Claims and Medical Records Data Agreement about Eligibility and Scoring Independent Variables for Agreement Equations Distribution of Covariates Across the Eligibility Models Distribution of Covariates Across the Scoring Models Levels of Agreement about Eligibility – Bivariate Comparisons Level of Agreement about Scoring – Bivariate Comparisons 15 16 17 31 36 45 47 54 57 58 60 63 64 71 81 84 87 94 105 109 111 113 115 - xii- Table 4.10 Table 4.11 Table 4.12 Table 4.13 Table D.1 Table D.2 Table E.1 Table E.2 Odds Ratios from Logistic Regressions for Agreement between Claims Data and Medical Records about Eligibility Odds Ratios for Agreement between Claims Data and Medical Records about Scoring Results of Hypothesis Testing Comparing Performance Rates from Claims and Medical Records Data Overall Agreement about Eligibility: Kappa Statistics Overall Agreement about Scoring: Kappa Statistics Semi-Standardized Regression Coefficients for Agreement Models about Eligibility Semi-Standardized Regression Coefficients for Agreement Models about Scoring 121 126 130 134 242 243 244 245 - xiii- ABSTRACT Annual health expenditures in the United States exceed $1.5 trillion. Despite a growing number of efforts to measure and report levels of health care quality resulting from these expenditures, useful information is neither uniformly nor widely available. However, the information that is available suggests that millions of Americans fail to receive adequate quality care. Consequently, there are large potential gains in welfare associated with improving the performance of the health care system. There would be enormous value in finding reliable ways to measure the quality of care in a routine and affordable manner. The goals of this dissertation were to (1) examine whether the capacity of claims data for quality measurement is being fully recognized and (2) identify the types of information that, if available, would increase the potential for quality measurement with electronic data. The strengths and limitations of quality measurement with claims data were examined using two separate analyses. First, more than 550 quality of care indicators were analyzed for whether they could be assessed with claims data. This analysis was performed to evaluate the dimensions of quality that can be evaluated with claims data and to identify the types of information that would increase the capacity for quality measurement with electronic data. A separate analysis compared agreement between claims and medical records data about quality measurements to assess the accuracy of quality measurement with claims data. Of the 553 indicators reviewed, 186 (34%) could be assessed with claims data. This represents a significant increase in the number of indicators commonly used to measure quality. However, because the primary purpose of claims data is to facilitate payment, they do not include sufficiently detailed clinical information to assess many quality of care indicators. As investments in electronic health information increase, it would be helpful to know, from the perspective of quality measurement, which -xiv - types of information would add the most value. The results of this study found that the addition of any one type of information would not have a very large contribution to the number or type of indicators that could be constructed. While more detailed information about patients’ presenting signs and symptoms would have the largest contribution to quality measurement, linking existing data sources, such as laboratory test results, to claims data may be a more practical next step in increasing the capacity for quality measurement with electronic data. When performance rates for using claims data were compared to rates constructed with medical records data, the medical records rates were not consistently higher than the claims based rates. This finding contradicts the assumption that claims data will consistently underestimate the rate of performance. While widespread quality measurement relying extensively on medical records would provide more broad-based and accurate information, the associated costs currently appear prohibitive to most organizations. There are four broad options for quality measurement over the next several years: (1) the status quo; (2) expanded use of claims data, possibly augmented with the kinds of information analyzed in chapter 3; (3) expanded use of medical records; and (4) expanded use of both claims data and medical records. The benefits and costs of various versions of the last three options relative to the status quo, and relative to each other, are important, but open, questions. -xv- ACKNOWLEDGEMENTS I am grateful for the support and guidance that many individuals provided me during this study. I extend special thanks to my committee members – Steve Garber, Beth McGlynn, John Adams, and Sandy Berry – for their contributions to my research as well as to my professional training. I have benefited greatly from working with these researchers and am grateful for their patience and guidance. Professor David Nerenz, from Michigan State University, read an earlier draft of the dissertation and offered constructive comments and insights from which I have benefited. I would also like to extend special thanks to Joan Keesey and David Klein for their analytic support. The programming expertise they provided and their willingness to teach me the basics of SAS were integral to this study. I am also grateful to the staff at the HMO who not only provided the data for this study but also accommodated questions throughout my research. I cannot begin to express my gratitude to my family and friends for their continued support and encouragement in my academic and professional choices. This work was supported by a dissertation grant from the Agency for Healthcare Research and Quality.