Table 2. Framework for considering which patient characteristics to include in case mix adjustment
Methodological Considerations in Generating Provider Performance Scores
|Guiding questions||Examples of patient characteristics|
|Is the patient characteristic considered to be beyond control of providers?a||Yes||Yes||Controversial|
|Is the relationship between the characteristic and performance considered acceptable?a||Yes, for some performance measures (e.g., mortality rates)||Controversial||Controversial|
|Include characteristic in case mix adjustment?||Yes, for some performance measures||Controversial||Controversial|
|As an alternative to case mix adjustment, present stratified results?||A reasonable option, but no advantage over case mix adjustment||Yes, may be a good alternativeb||Stratification still controversialc|
b. If stratified performance reports display scores on the same scale in each stratum, then performance disparities will be reported. However, if stratified reports display scores on different scales (e.g., a ranking within each stratum), then stratification will have no advantage over case mix adjustment. Both techniques will make performance disparities appear to vanish.
c. Like case mix adjustment, stratifying by adherence (i.e., separately reporting performance for "adherent" and "nonadherent" patients) eliminates the incentive to improve adherence.