Journal Watch: Lights, Sirens, and Ambulance Crash Risk

Reviewed This Month

Is Use of Warning Lights and Sirens Associated With Increased Risk of Ambulance Crashes? A Contemporary Analysis Using National EMS Information System (NEMSIS) Data. 

Authors: Watanabe BL, Patterson GS, Kempema JM, Magallanes O, Brown LH.  

Published in: Ann Emerg Med, 2018 Nov 14. 

Use of lights and sirens has been a topic of debate in our field for some time. Historically they have been used as a tool to help us arrive on scene quickly. Studies have shown there are some low-frequency, high-acuity calls and critical patient conditions in which lights and sirens can have a positive impact on patient outcomes.

However, multiple studies have also shown the use of lights and sirens does not substantially reduce either response or transport times. Further, literature suggests the use of lights and sirens increases the risk of ambulance crashes and the risks of injury and death when an ambulance crashes. Unfortunately, most of the available literature on this topic is not recent, and until now no national studies have been conducted. 

Thus there was certainly a need for this month's study, the objective of which was to “provide a contemporary, nationwide comparison of reported crash rates for U.S. ambulances responding to or transporting patients from a 9-1-1 emergency scene with or without lights and sirens.” Data for this study was obtained from the National EMS Information System (NEMSIS) 2016 public-release research data set. This is a wonderful resource those conducting EMS research should consider as a data source for addressing their questions. 

The 2016 data set included almost 30 million EMS calls submitted by approximately 10,000 agencies from 49 states and territories. For their study the authors included “all dispatches of a transport-capable ground EMS vehicle to a 9-1-1 emergency scene.” Because not all calls resulted in a patient transport, responses to and transports from scenes were evaluated separately. Only 9-1-1 ground ambulance calls were evaluated—all other types were excluded. 

The NEMSIS element response mode to scene was the main independent variable of interest. For the analysis the authors divided this element into three categories: full lights and sirens, no lights and sirens, and any lights and sirens. The category of any lights and sirens combined lights and sirens, initial lights and sirens, downgraded to no lights and sirens, no initial lights and sirens, and upgraded to lights and sirens.

The outcome variable of interest was an ambulance crash. NEMSIS didn’t directly collect ambulance crash information in the 2016 data set; however, there were two NEMSIS elements that enabled the reporting of response or transport delays, and both included a response option to indicate a delay due to a crash. These elements were used as a proxy. For calls in which both a response and transport delay were attributed to an ambulance crash, the authors considered this as only a response delay because they “presumed a response-phase crash had a carryover effect that also resulted in a transport delay.” 

The authors performed descriptive analyses and calculated the rate of crash-related delays per 100,000 responses or transports. They also built a multivariable logistic regression model. Regression modeling was performed to identify predictors of crash-related delays. While use of lights and sirens was the main independent variable of interest, to adjust for the effects of other important variables, the logistic regression model included the time of the 9-1-1 call (daytime, evening, overnight), the community size in which the event took place, the level of service, the organization type, and whether the organization was volunteer, nonvolunteer, or mixed. 

Results

The NEMSIS 2016 public-release data set included 20,465,856 9-1-1 dispatches of ground transport ambulances. The authors reported 2,539 crash-related delays. That represents an overall crash rate of 12.4 per 100,000 ambulance runs. When evaluating responses and transports separately, the crash rate was 5.3 per 100,000 responses and 9.3 per 100,000 transports.

When evaluating the response phase, after adjustment, when compared to no lights and sirens, the odds of a crash-related delay were greater in the any lights and sirens group (OR 1.50; 95% CI, 1.19–1.90) and the full lights and sirens group (OR 1.53; 95% CI, 1.21–1.94). The results were similar but more dramatic when evaluating the transport phase (after adjustment: any lights and sirens, OR 2.90; 95% CI, 2.18–3.87; full lights and sirens, OR 2.84; 95% CI, 2.12–3.80). 

As with all studies, there were some limitations. The authors noted that minor crashes that did not result in a delay would be missed by their analysis. Also, it is possible that the use of lights and sirens occurred after a crash. The authors also could not account for all potentially important variables such as weather, traffic, lighting conditions, travel distances, and others; however, they used a method called clustering when building their logistic regression model to help account for this. 

This is an interesting study that substantially adds to the literature on lights and sirens use in EMS. While the overall rates were relatively low, the associations found with an increased risk of ambulance crashes are important. I hope you can read this manuscript in its entirety; there are a lot of interesting analyses as well as tables and figures to help interpret the results.   

Antonio R. Fernandez, PhD, NRP, FAHA, is research director at the EMS Performance Improvement Center and an assistant professor in the Department of Emergency Medicine at the University of North Carolina–Chapel Hill. He is on the board of advisors of the Prehospital Care Research Forum at UCLA.