UCSF

Background

Are 9-1-1 ambulances relatively late to poorer neighborhoods? Studies suggesting so often rely on weak measures of neighborhood (e.g., postal zip code), limit the analysis to particular ambulance encounters (e.g., cardiac arrest responses), and do little to account for variations in dispatch priority or intervention severity.

Methods

We merged EMS ambulance contact records in a single California county (n = 87,554) with tract-level data from the American Community Survey (n = 300). After calculating tract-level median ambulance response time (MART), we used ordinary least squares (OLS) regression to estimate a conditional average relationship between neighborhood poverty and MART and quantile regression to condition this relationship on 25th, 50th, and 75th percentiles of MART. We also specified each of these outcomes by five dispatch priorities and by three intervention severities. For each model, we estimated the associated changes in MART per 10 percentage point increase in tract-level poverty while adjusting for emergency department proximity, population density, and population size.

Results

Our study produced three major findings. First, most of our tests suggested tract-level poverty was negatively associated with MART. Our baseline OLS model estimates that a 10 percentage point increase in tract-level poverty is associated with almost a 24 s decrease in MART (-23.55 s, 95% confidence interval [CI] -33.13 to -13.98). Results from our quantile regression models provided further evidence for this association. Second, we did not find evidence that ambulances are relatively late to poorer neighborhoods when specifying MART by dispatch priority. Third, we were also unable to identify a positive association between tract-level poverty and MART when we specified our outcomes by three intervention severities. Across each of our 36 models, tract-level poverty was either not significantly associated with MART or was negatively associated with MART by a magnitude smaller than a full minute per estimated 10 percentage point increase in poverty concentration.

Conclusion

Our study challenges the commonly held assumption that ambulances are later to poor neighborhoods. We scrutinize our findings before cautiously considering their relevance for ambulance response time research and for ongoing conversations on the relationship between neighborhood poverty and prehospital care.