Combining multiple indicators of clinical quality: an evaluation of different analytic approaches.

Objective: To compare different methods of combining quality indicators scores to produce composite scores that summarize the overall performance of health care providers.

Methods: Five methods for computing a composite quality score were compared: the "All-or-None," the "70% Standard," the "Overall Percentage," the "Indicator Average," and the "Patient Average." The first 2 "criterion-referenced" methods assess the degree to which a provider has reached a threshold for quality of care for each patient (100% or 70%). The remaining "absolute score" methods produce scores representing the proportion of required care successfully provided. Each method was applied to 2 quality indicator datasets, derived from audits of UK family practitioner records. Dataset A included quality indicator data for 1178 patients from 16 family practices covering 23 acute, chronic, and preventative conditions. Dataset B included data on 3285 patients from 60 family practices, covering 3 chronic conditions.

Results: The results varied considerably depending on the method of aggregation used, resulting in substantial differences in how providers scored. The results also varied considerably for the 2 datasets. There was more agreement between methods for dataset B, but for dataset A 6 of the 16 practices moved between the top and bottom quartiles depending upon the method used.

Conclusions: Different methods of computing composite quality scores can lead to different conclusions being drawn about both relative and absolute quality among health care providers. Different methods are suited to different types of application. The main advantages and disadvantages of each method are described and discussed.