Case mix adjustment
Case mix adjustment refers to the use of statistical procedures to permit comparison of treatment outcomes between providers with differing mix of patients with regard to diagnoses, severity of illness, and other variables associated with the probability of improvement with treatment.
A case mix variable may be any variable that is predictive of improvement on the outcome measure. The intake score on the outcome questionnaire is almost always the strongest predictor of the final score and by extension the change score on the same questionnaire. The fact that the repeated administrations of the questionnaire are correlated over time assures that this will be the case (see also
Regression to the Mean).
Other variables such as diagnosis, prior treatment history, drug/alcohol use, age and sex may also be predictive of the change score.
Case mix adjustment is best achieved through the use of a
general linear model which uses the intake score and other case mix variables to predict the change score for each patient. The difference between the predicted score and the patient's actual change score is referred to as the
residualized change (or gain) score (or the
Benchmark Score). This residual score conveys the extent to which a specific patient's outcome differed from that of all other patient's in the sample after matching for intake score and other variables.
Primary topics:
You may also view all topics in alphabetical order in
WebTopicList.
--
JebBrown - 09 Jan 2007