Risk adjustment for comparing outcomes across providers with multivariate models

Surgical clinics throughout Germany are under political and economical pres sure. Comparing outcomes across clinics or ranking become more and more pop ular although there are no generally accepted and scientifically valid meth ods how to compare. Using original data of the external quality assurance project of the chambe r of physicians of Westphalia-Lippia it is shown in this article that compa ring raw data is often misleading. Differences in risk factors (case mix) h ave to be taken into consideration. A model using logistic regression methods to adjust for case mix in letalit y and morbidity of hip fractures is proposed. A ranking list before and aft er adjusting for risk factors shows partly dramatic changes. One clinic for example changed from position 97 (out of one hundred) to position 26.