An Econometric Analysis of FDNY’s Systemwide EMS Dispatch, Response and Operational Performance During the 2020 Covid Lockdown Using a Negative Binomial Model With Multiple Fixed Effects

A negative binomial longitudinal (panel) count model with multiple fixed effects in the presence of over-dispersion is used to estimate the heterogenity effects of emergency 911 incidents or call volume, year, day of the week, dispatcher-assigned medical emergency call type, priority rank, ambulance (DUAL, ALS, BLS) response and borough on average ambulance travel time. We use FDNY Emergency Dispatch Data for the periods March 20, 2019–June 13, 2019 and the corresponding lockdown period of March 20, 2020–June13, 2020 for a year over year econometric analysis. The model addresses the limitations of two-by-two contingency table analysis when there are more than two categorical variables by utilizing the underlying statistical properties of the Poisson distribution in a multivariate parametric approach by analyzing the simultaneous effects of several explanatory variables.When clustered standard errors are corrected for fixed effects and over-dispersion, we find that there was no significant statistical difference in the relationship between average travel time and call volume in the 2019 and 2020 periods. Wednesday is only day of the week that was most likely to increase travel response time. All medical emergencies showed significant declines in average travel response time at a p-value = .05, except for call types designated as allergy that was significant at p-value less than .07. Clinical priority rank 6 that required the use of BLS equipment and EMT crews used in non-medical emergencies and other cases had no statistical effect on travel time in the comparison. When compared to Manhattan, Brooklyn and Queens had no significant statistical effect on average travel response time.