Date of Award
Open Access Thesis
Medical Doctor (MD)
Lori A. Post
Public health, Demography
Objectives. To apply a novel statistical method to create a comprehensive estimate of incidence of firearm related injuries.
Methods. A database of firearms injuries in New Haven, Connecticut, during a five- month period was created with records from law enforcement, emergency departments, emergency medical services (EMS), news media, and the medical examiner. The overlap of these various sources was operationalized in a capture-recapture model to generate an estimate of uncounted firearms injuries, and log linear modeling was used to control for positive and negative dependencies.
Results. The combined data sources revealed 49 firearms injuries occurring during the study period within our defined geographical area. No single source recorded more than 43 of these injuries. Log-linear capture recapture methods estimated that the actual number of injuries was 49.7 (95% CI 49-52.3).
Conclusions. No single source reaches complete case ascertainment for firearms injures. Combining multiple sources improves the estimate of injury incidence, but still results in an undercount. Log-linear capture-recapture methods can be used to improve the estimate of firearms injuries.
Balsen, Zev, "Counting Uncounted Gunshot Injuries: A Capture‐recapture Study Of People Minding Their Own Business" (2015). Yale Medicine Thesis Digital Library. 1946.