P022: Use of police and SAR records to identify cases and reduce survivorship bias in prehospital care research

Abstract

Introduction: Evidence based medical practices are limited in prehospital care. A 2006 National Academies report on the state of prehospital care indicates that as little as 4% of prehospital care is evidence based. Retrospective case reviews are inexpensive studies that can effectively evaluate current practices and identify opportunities for improvement. Commonly, retrospective reviews in prehospital care rely on electronic health records from hospitals and emergency health services. These data sources suffer from three limitations; survivorship and inclusion biases, a lack of control cases, and difficulty identifying unusual etiologies in databases. Police and search and rescue records are uncommon but promising data sources for certain topics Methods: To test our methodology, we investigated outcomes of suicide attempts by jumping from bridges in Vancouver. We identified patients who threatened, attempted, or jumped from bridges >12m between 2006 and 2017. We describe the population, mortality and adverse outcomes, and identify factors differentiating survivors from fatalities. Police and Coast Guard (CG) records were searched to identify cases. Corresponding records from ambulance, hospitals, and the coroner were identified using date, time, and patient age and sex. Linked records were reviewed and key data extracted. Results: 1208 cases were identified, outcomes were positively identified for 90.3%. 273 were confirmed jumps. 78.2% of ambulance, 90.0% of hospital, and 93.6% of coroner records were identified and linked to corresponding police and CG records. By contrast, an independent search of ambulance records yielded a 99.42% false positive rate, and independent searches of hospital records were not possible due to technological limitations in patient data collection and storage tools. Further, of 197 cases where patients jumped into water, 94 were attended to by EHS, and 52 were transported to hospital. Conclusion: Police and CG records effectively identified patients. Without these data sources, identifying most cases would not have been possible. Since a majority of patients were not transported to EHS or hospital, linking data from these agencies to the hospital and EHS records limited survivorship bias. This methodology may be valuable in future prehospital and ED research, especially for topics with high likelihood of police or SAR contact like suicide attempts or avalanche burials.