[Operationalization of Quality Indicators with Routine Data Using the Example of the Evaluation of "Integrated Care...

As part of the 10-year evaluation of Gesundes Kinzigtal Integrated Care (IVGK, Innovation Fund Project 01VSF16002), a multidisciplinary group of experts agreed on 101 quality indicators (QI) to evaluate the quality of regionally integrated care with its focus on health and prevention programs. One criterion was that the selected QI should in principle be suitable for mapping using routine data. The aim of the study was to investigate how many and in what way the QI developed can actually be mapped in Germany with routine data and for what reasons operationalization was restricted or not possible.

The operationalization of the QIs was performed using pseudonymized billing data of the AOK Baden-Württemberg from 2006 to 2015, which the Scientific Institute of the AOK (WIdO) provided to the evaluation team. All operationalized indicators were binary coded (criterion fulfilled yes/no). The diagnoses, procedures, or drugs named in the numerator and denominator definitions were operationalized using ICD-10 codes (inclusion and exclusion diagnoses), EBM codes, OPS codes, ATC codes. Indicator prevalences were examined over time to check for abnormalities as an indication of possible misscoding.

Ninety of the 101 indicators were operationalizable with routine data. Fourteen of the 90 indicators could only be operationalized with restrictions, as corresponding service codes were only introduced or existing codes were changed during the observation period. Seventy-six of 90 indicators could be operationalized without restrictions. In this context, 15 of these 76 indicators required pre- and follow-up periods, which meant that they could not be presented for all years. Eleven of 101 QIs could not be operationalized because EBM codes were only introduced after 2015 or were not recorded as individual services for all physician groups (e. g., spirometry and long-term ECG). Striking trends in indicator prevalences could be explained.

Routine data enable resource-saving quality monitoring. A change in the data basis during the observation period, for example through the introduction or deletion of billing codes, makes the longitudinal, routine data-based quality assessment more difficult, but enables further or new indicators to be operationalized for later periods.