Using floating catchment area (FCA) metrics to predict health care utilization patterns

BMC Health Serv Res. 2019 Mar 4;19(1):144. doi: 10.1186/s12913-019-3969-5.

Abstract

Background: Floating Catchment Area (FCA) metrics provide a comprehensive measure of potential spatial accessibility to health care services and are often used to identify geographic disparities in health care access. An unexplored aspect of FCA metrics is whether they can be useful in predicting where people actually seek care. This research addresses this question by examining the utility of FCA metrics for predicting patient utilization patterns, the flows of patients from their residences to facilities.

Methods: Using more than one million inpatient hospital visits in Michigan, we calculated expected utilization patterns from Zip Codes to hospitals using four FCA metrics and two traditional metrics (simple distance and a Huff model) and compared them to observed utilization patterns. Because all of the accessibility metrics rely on the specification of a distance decay function and its associated parameters, we conducted a sensitivity analysis to evaluate their effects on prediction accuracy.

Results: We found that the Three Step FCA (3SFCA) and Modified Two Step FCA (M2SFCA) were the most effective metrics for predicting utilization patterns, correctly predicting the destination hospital for nearly 74% of hospital visits in Michigan. These two metrics were also the least sensitive to changes to the distance decay functions and parameter settings.

Conclusions: Overall, this research demonstrates that FCA metrics can provide reasonable predictions of patient utilization patterns and FCA utilization models could be considered as a substitute when utilization pattern data are unavailable.

Keywords: Access to health care; Floating catchment areas; Health care use; Hospitalizations; Spatial accessibility; Utilization patterns.

MeSH terms

  • Catchment Area, Health*
  • Health Services Accessibility*
  • Hospitals / statistics & numerical data*
  • Humans
  • Michigan
  • Models, Statistical
  • Patient Acceptance of Health Care / statistics & numerical data*