Recently, there has been an increased push for the adoption of alternative responses to policing. With mental health calls, these efforts take the form of co-responder models, in which mental health professionals respond to calls alongside police, or more recently behavioural health teams that respond alone. For any alternative, it is important to understand the prevalence of the problem, here measured as the volume of calls. However, measuring calls involving persons with mental illness (PwMI) has been historically problematic due to data ambiguity and missing context. We partner with the Burlington Police Department in North Carolina to better identify the full universe of calls involving PwMI. Using supplementary data sources and mental health keyword identification, we find that less than half of all calls involving PwMI are identified as such in the original call. Understanding the full universe of calls has implications for resource allocation, decisions to adopt alternate responses, and providing context for the current state of interaction between police and the population of PwMI.
