Precision of individual and composite performance scores: the ideal number of indicators in an indicator set

BACKGROUND: In many countries, quality indicators are used to assess the quality of care of family practice. Such assessments need to have an adequate precision, so that the results can be interpreted correctly. However, a small sample size per practitioner can lead to inadequate precision. A possible solution could be to create composite performance scores. OBJECTIVES: To evaluate the relationship between sample size and precision. We examine whether a composite performance score has an increased precision and how many indicators are needed minimally to achieve this level of precision. RESEARCH DESIGN: We performed a descriptive statistical study on data from the medical records of 455 Dutch practices. We included 3 different conditions: diabetes (12 indicators), chronic obstructive pulmonary disease (4 indicators), and Cardiovascular Disease and Risk Management (9 indicators). RESULTS: For individual quality indicators, patient samples close to 100 are required to achieve even moderate precision (10 percentage points) on the performance scores. This number decreases substantially when a composite score is used. A composite derived from combining 5 to 7 indicators can provide much the same precision of measurement as one made up from a much larger number of indicators. CONCLUSIONS: The added value of a composite score depends on the a priori reasons for measuring quality. Our results indicate that especially for formative quality improvement a small number of carefully selected indicators can provide a sufficiently precise composite measure