[Evaluation of the population-based 'Integrated Health Care System Gesundes Kinzigtal' (IHGK). Findings on health care quality based on administrative data]

Background and objectives: The management company "Gesundes Kinzigtal GmbH" signed an agreement on integrated healthcare with the statutory health insurer AOK Baden-Württemberg, effective as of November 1, 2005. The aim of the evaluation was to check whether the extent of overuse and underuse of healthcare services in the area of intervention declined or increased compared with usual care.

Methods: Longitudinal study with non-randomised control group based on health insurers' claims data from the years 2004-2011. Intervention group: residents of the Kinzigtal region insured by AOK.

Control group: persons insured with the AOK in other regions of Baden-Württemberg. Healthcare quality indicators were derived from other studies and guidelines. Fifteen out of the 18 indicators related to overuse or underuse; three related to an outcome, namely avoidable hospital stays, the appearance of fractures in patients with osteoporosis, and mortality. Trend and outcome analyses rely on Poisson and Cox regressions adjusted for age, sex, the Charlson Index, and multimorbidity.

Results: Two out of 5 indicators for overuse and 2 out of 10 for underuse showed significant improvement for the intervention population relative to the control group. The risk of a fracture in patients with osteoporosis (HR: 0.809; 95 % CI: 0.740 to 0.885; p<0.0001) and mortality (HR: 0.944; 95% CI; 0.899-0.991; p=0.0194) were significantly lower in the Kinzigtal population. No negative trends were found.

Conclusions: Compared with the control group, which represents the secular trend, significant improvements of healthcare quality in the intervention group (6 out of 18 indicators) are considerably more frequent than significant changes for the worse (0 out of 18 indicators). To date, the effects are not very strong, as all insured persons from the Kinzigtal form the basis of the analysis irrespective of their participation in the integrated care program. Claims data are appIicable to indicator based evaluation, but it would be necessary to consider additional years to see whether observed positive trends could be enhanced.