Weng W, Arnold GK, Holmboe ES, Lipner RS. — American Board of Internal Medicine
Presented: AcademyHealth Annual Research Meeting, June 2008
Research Objective: The focus of pay-for-performance (P4P) programs is to link compensation to measures of quality of patient care. While clinical measures dominate most of the P4P programs, measures from patient surveys (e.g., CAHPS patient survey) and practice system survey (e.g., National Commission for Quality Assurance’s Physician Practice Connections [PPC] survey) are sometimes used. Most P4P programs, however, typically do not include performance measures from all three data components (clinical, patient satisfaction and system). The goal of this study is to compare physician performance rankings among different combinations of the three data components.
Study Design: The hypertension practice improvement module (PIM) developed by the American Board of Internal Medicine provides a comprehensive assessment of a physician's practice using three data collection components – patient survey, chart audit and practice system survey. The patient survey provides five indicators including overall patient satisfaction and satisfaction with aspects of hypertensive care. The chart audit provides nine clinical measures, including two outcomes measures – blood pressure control, LDL control – and seven processes and treatment measures. The practice system survey is loosely based on the PPC with additional disease-specific questions. The patient satisfaction and chart composites are the means of the individual measures. The 89 system questions are aggregated into six domains and the mean of the domain scores constitute the system composite. Each of the three composite measures is then standardized. Correlations and changes of physician rankings for six composites are compared. The composites are chart, patient, system, chart + patient, chart+ system, and chart + patient + system; the data components that are combined are weighted equally.
Population Studied: Data include practice system survey of 668 physicians, 407 general internists and 261 subspecialists, who completed the Hypertension PIM between October 2004 and October 2007; 16,880 chart audits; and 18,742 patient surveys.
Principal Findings: Mean physician age was 44.7 years (SD = 6.5), most physicians (71%) were in group practice, and 26% were female. Correlations among physician rankings of chart, patient and system composites are low (0.18 to 0.25). Physician rankings on the six composite measures vary significantly. Fifteen percent of physicians rank more than one-quartile (167 positions) higher on the chart than on the patient composite, while 13% rank more than one-quartile lower. The differences in ranks between chart and system composite and between patient and system composite are similar in magnitude. Ten percent of physicians rank more than one-quartile higher on the chart than on the three-part composite, while 11% rank more than one-quartile lower.
Conclusions: There are only very moderate correlations among patient satisfaction, chart composite and system composite. The rankings of physician performance change considerably depending on which combinations of performance measures are used in the composite.
Implications for Policy, Practice or Delivery: These results empirically demonstrate the complexity of measuring multiple dimensions in the quality of patient care. In assessing physician’s quality of patient care, a profile that incorporates all three aspects of patient care (i.e., clinical measures, patient satisfaction and practice systems) may be preferable to clinical measures alone.
Funding Sources: ABIM, NQF
For more information about this presentation, please contact Research@abim.org.