Can the use of Electronic Health Records in General Practice reduce hospitalizations for diabetes patients? Evidence from a natural experiment

Disease management programmes (DMP) in the general practice sector are increasingly used to improve health of chronically ill patients, reduce hospitalizations and thereby costs. The aim of this paper is to estimate the causal effects of the enrolment of general practices (GP) in a DMP based on Electronic Health Records (EHR) on diabetes patients total hospitalizations, diabetes related hospitalizations and hospitalizations with diabetes and cardiovascular related Ambulatory Care Sentive Conditions (ACSC). We use a rich nationwide panel dataset (2004-2013) with information of stepwise enrolment of GPs in the EHR program. As a control group we use GPs who never enrolled. Following the recent literature on causal inference with panel data, we use a standard propensity score matching estimator where we also match on pre-treatment outcomes. This allows controlling for all the unobservable confounders which were already present in the pre-treatment outcomes. Alternative, we use a difference in difference as well as a parametric model with a continuous treatment specification and find similar results. Our results show that enrolment in EHR reduced diabetes patients’ risk of hospitalizations by more than 10%. The results are comparable with studies on EHR programs from California and the magnitudes of the effects are comparable to DMPs including both EHR and financial incentives.