All three approaches for the definition had been applied in works about comorbidity measures. Thirdly, complications could be separated from preexisting diseases and those summarized as comorbidity. Secondly, comorbidity could be defined as the coexistence of several diseases, without the necessity to exclude a disease of interest. In routine data of inpatient care, the principle diagnosis is frequently taken as the disease of interest, the set of secondary diagnoses as comorbidity. Firstly, comorbidity can be regarded as diseases that exist additionally to one or several diseases of interest. If morbidity is the set of diseases of an individual, comorbidity might be different. The developed new measures may be used to control for confounding.įor this work, definition three is applied. The sum scores showed a comparable performance. The best result was achieved with the measure based on ICD-10-groups with an area under curve of 0.910 (95% confidence interval = 0.907–0.913). According to the receiver operating characteristic, the quality of the measures based on the structure of the ICD-10 was superior compared with the Charlson Index and the Elixhauser Comorbidities. Charlson Index and Elixhauser Comorbidities were mapped to the German ICD-10.
Logistic regression was used to derive the new models. Models and scores were compared with the Charlson Index and the Elixhauser Comorbidities using the receiver operating characteristic. The models were transformed into sum scores using whole-number weights. Two different risk models were developed, one based on ICD-10 chapters, the other based on ICD-10 groups. Routine data from three German hospitals with inpatients discharged 2008 were used for model development routine data from 36 German hospitals with inpatients admitted and discharged 2010 were used for model evaluation. Objective of this work was to elaborate the power of the ICD structure for defining morbidity and comorbidity measures. However, a measure that makes use of the ICD structure is missing. Several proposals for those measures are defined with ICD-coded diagnoses available in hospital routine data. Measures of morbidity and comorbidity are frequently used for the control of confounding, particularly in health services research.
0 kommentar(er)
