Abstract
BACKGROUND: Statutory health insurance claims data are used for risk adjusted quality assessment of inpatient treatments. This study evaluated the impact of clinical data on quality assessment of the treatment of acute myocardial infarction (AMI), and whether the content of clinical factors can be approximated by claims data. MATERIAL AND METHODS: In this observational study, claims data of statutory health insurance in Germany were linked retrospectively with clinical data from 14 hospitals. The hybrid study population encompassed 3148 cases with inpatient treatment of AMI who underwent coronary angiography. Quality indicators for five outcomes were developed by means of stepwise logistic regression. Using claims data from a nationwide study population of 165 130 AMI cases and a subsample of the same size as the hybrid group, the performances of the risk adjustment models were compared by receiver operating characteristic-area under the curve (ROC-AUC) and standardised mortality/morbidity ratios (SMR). RESULTS: As clinical risk factors, a haemoglobin value of ≤10 g/dL and an estimated glomerular filtration rate of ≤60 mL/min/1.73qm were included, but did not result in a gain of ROC-AUC (w/o clinical variables: 0.74-0.86; with: 0.74-0.87). By approximating the content of these clinical factors by means of claims data, an increase in model performance of up to 4.7% was achieved (with surrogates: 0.77-0.89), but did not influence the final quality assessment by SMR. CONCLUSIONS: While replication of our findings is necessary, our models for risk adjustment and surrogates for two clinical factors show that comparative quality reporting using claims data is feasible, although with acknowledged limitations. This would minimise the data set for quality assessment of AMI treatment in accordance with the principle of data minimisation and avoid the need for additional manual documentation.