LAUSR

but after thorough review of hundreds of incidents, we came to two conclusions. First, there was no consistent method of accurately measuring this time, and second, the women and men who respond to these shootings should be commended for their bravery and altruism under very challenging conditions. The analyses of transit time differences were adjusted for “distance to hospital” using distance as a continuous covariate in an analysis of covariance. The distance covariate was statistically significant for all analyses and was included along with type of transport in the models. For the analysis of mortality, single and multiple variable logistic regression models were used to compare differences (raw and adjusted) between methods of transport and ShotSpotter groups. In the multiple variable analysis, differences in mortality were adjusted for distance to hospital, transit time, and injury severity score (ISS) as continuous covariates. The covariates included were statistically significant. The adjusted p value for the mortality of the ShotSpotter group when correcting for distance ISS, trauma and injury severity score (TRISS), and shock index was 0.1224. Aside from the sensitivity/specificity guarantee from the manufacturer of this equipment, we do not have specific data on the incidence of false positives. As we describe in the article, the auditory sensors for this system are highly sensitive but have the ability to record nongunshot sounds. Each activation of a sensor is reviewed by remote human listeners who are trained to distinguish the difference between a car backfiring or firecracker, for example. While there are still likely “false positives” that occur, there is no way to accurately delineate a false positive from a true positive in many cases because the lack of an injured victim at the point of acoustic triangulation does not mean there was no gunshot which took place. Our feeling is that this can be viewed as an “over triage” in a way, where the most likely incidents are investigated by the police to search for a potential victim and, when such a victim is identified, the technology has done its job. Due to confidentiality reasons, local law enforcement has asked our group not to share information, such as precise location of sensors, number of sensors, number of acoustic activations, or geographic areas where the majority of acoustic activations occurred. We crossed referenced time stamps for every shooting which resulted in an injured victim being brought to our trauma center and identified several incidents where multiple sensors were activated for the same event. As far as incidents with multiple victims, a review of our data showed that 11% of incidents had more than one injured victim arriving at the hospital. Any patient who presented to our facility, whether as a trauma activation or a minimally injured patient arriving in the emergency department, was reviewed as part of this study.