Who counts and how: Retrospective analysis of point-in-time homeless counts in the United States

It is well documented that the homeless suffer from disproportionate high rates of health issues that have an impact on broader society. On average, their life expectancy is almost 15 years less than the life span for the general population. Understanding the incidence of homelessness is imperative in the development and implementation of public health interventions as well as the more equitable and effective distribution of resources. Beginning in 2004, the US Department of Housing and Urban Development (HUD) required that regional or local planning bodies, referred to as Continuums of Care (CoC), receiving funds intended for homeless services and programs conduct point-in-time counts of sheltered and unsheltered homeless individuals, children and families every other year. Our study assesses the relationship between homeless estimates and enumeration methods as they are employed across the United States by conducting an in-depth retrospective analysis of public data from HUD and CoCs. We explore nationwide patterns in estimate variation among and within the local CoCs. Our preliminary analysis of nationwide homeless estimates reveals an inexplicably wide range of within-CoC unsheltered homeless estimates, from a minimum of 0% to a maximum of 71,641% from 2007-2013. Our preliminary work also suggests that CoCs are frequently changing the geographic areas sampled, the methods of inferring from sampled data to total estimates, the number of volunteers dedicated to enumeration, and the contracting agencies in charge of enumeration. All of these inconsistent estimate methods lead to variable homeless counts. A better understanding of homelessness enumeration can promote better design and implementation of public health programs and more equitable and effective resource distribution.