Measuring Medication

Do Interviews Agree with Medical Record and Pharmacy Data?

Published in: Medical Care, v. 40, no. 12, Dec. 2002, p. 1270-1282

Posted on on January 01, 2002

by P. Todd Korthuis, Steven M. Asch, Martha J. Timmer, Martin F. Shapiro, W. Christopher Mathews, William Cunningham, J. Allen McCutchan, Allen Gifford, Martin L. Lee, Samuel A. Bozzette

Read More

Access further information on this document at

This article was published outside of RAND. The full text of the article can be found at the link above.

BACKGROUND: Medication measurement is crucial in assessing quality for chronic conditions yet agreement of alternate data sources remains uncertain. OBJECTIVES: To evaluate medication agreement between interviews, medical records, and pharmacy data; to assess data source contribution to attributing medication exposure; and to describe the impact of combining data sources on models that predict medication use. RESEARCH DESIGN: Prospective cohort study. SUBJECTS: Probability sample of HIVinfected participants in the HIV Cost and Services Utilization Study. MEASURES: Medications reported in 2267 interviews, 1936 medical records, and 457 pharmacy records were compared using crude agreement, kappa, and the proportion of average positive and negative agreement. The percent of medications reported in each source alone was used to assess their relative contribution to attributing exposure status. The authors performed weighted logistic regression in alternate data sources. RESULTS: Kappa varied from 0.38 for nucleoside reverse transcriptase inhibitors to 0.70 for protease inhibitors, when comparing drug classes in interview versus medical record, interview versus pharmacy data, and medical record versus pharmacy data. The percentage of medications reported in medical records was greater than that reported in interviews or pharmacy data. Pharmacy data contributed little to the attribution of medication exposure. Adding medication data to interview data did not appreciably change analytic models predicting medication use. CONCLUSIONS: For specific medications, agreement between alternative data sources is fair to substantial, but is lower for key drug classes. Relying on one data source may lead to misclassification of drug exposure status, but combining data sources does not change the results of analytic models predicting appropriate medication use.

This report is part of the RAND Corporation External publication series. Many RAND studies are published in peer-reviewed scholarly journals, as chapters in commercial books, or as documents published by other organizations.

Our mission to help improve policy and decisionmaking through research and analysis is enabled through our core values of quality and objectivity and our unwavering commitment to the highest level of integrity and ethical behavior. To help ensure our research and analysis are rigorous, objective, and nonpartisan, we subject our research publications to a robust and exacting quality-assurance process; avoid both the appearance and reality of financial and other conflicts of interest through staff training, project screening, and a policy of mandatory disclosure; and pursue transparency in our research engagements through our commitment to the open publication of our research findings and recommendations, disclosure of the source of funding of published research, and policies to ensure intellectual independence. For more information, visit

The RAND Corporation is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions of its research clients and sponsors.