HEALTH CARE UTILIZATION AFTER A ROAD TRAFFIC ACCIDENT: Prediction modeling and identifying risk factors.

Previous research showed that, after trauma, patients indicate lack of information on the perspectives about their rehabilitation trajectory and recovery, while at the same time physicians state not to answer suchlike questions due to uncertainty. This study aims to test if secondary databases can be used to predict health care utilization after a road traffic accident. Linkage of claims data with hospital data was built at patient-level. A set of outcomes was selected for the acute- and post-acute phase ( up to one year after the traffic accident). For the acute phase we could demonstrate the association between injury severity and the probability of in-hospital mortality and the length of hospital stay until discharge with medical advice. This model was built while accounting for time-dependent variations and competing endpoints. In the post-acute phase, the incremental healthcare compared to the period before the accident was mapped. With a specific focus on rehabilitation we developed a prediction model estimating the probability of receiving rehabilitation and if so, the probability to belong to a certain rehabilitation trajectory. For those who had in-hospital rehabilitation, the length of admission was analyzed. As a synopsis of all these studies, it can be concluded that secondary data can be used to answer questions of health care utilization. However, further research on increasing the model performance is mandatory. Further attention needs to be given on how these model-algorithms could be translated in practice in support of patient communication and how to facilitate implementation in daily practice.