THE IMPACT OF COVID-19 INCIDENCE ON EMERGENCY MEDICAL SERVICES UTILIZATION

INTRODUCTION

At the onset of the COVID-19 pandemic in March 2020, provision of health care appeared on the verge of collapse, with a system under severe strain and extending to the emergency medical services (EMS) tasked with transporting patients from their homes to hospitals (1). In addition, anecdotal reports appeared of patients with emergent non–COVID-19 illnesses avoiding necessary emergency care due to concerns about contracting COVID-19.

Initial studies examining changes in the utilization of EMS at the onset of the pandemic in the United States, United Kingdom, Germany, Finland, and Australia noted absolute reductions in the volume of ambulance calls, ascribed to measures including lockdowns and social distancing (2, 3, 4, 5). Conversely, ambulance volumes increased in certain regions of Italy that were strongly affected by the first wave of COVID-19 (6).

Further studies examining the nature of the EMS calls noted ambulance activations for injuries decreased in the United States at the onset of the pandemic before returning to previous levels (2,7). Poisoning, respiratory, and trauma ambulance calls also decreased in New Zealand during the initial 5-week lockdown period (8). In the United States, Germany, and New Zealand, mental health–related ambulance calls increased during this initial interval (4,8,9). Alternatively, obstetric activations remained unchanged in Maryland, and little evidence emerged for a change in myocardial infarction or stroke volume in the U.K.’s West Midlands (10,11). Although these studies looked at changes in EMS utilization with the onset of the pandemic, previous studies have not examined how community COVID-19 incidence affected EMS utilization.

This study sought to examine how community COVID-19 incidence correlated with EMS utilization, including ambulance and 9-1-1 volumes, timing intervals, and patients with emergent non–COVID-19 illness utilization.

MATERIALS AND METHODS

Study Design

This study was a retrospective, observational study with regression models using ambulance data at the county level from the National Emergency Medical Services Information System (NEMSIS) for 2020 and COVID-19 incidence data for 2020 at the county level. The NEMSIS is a national system that stores and shares EMS data collected from U.S. states and territories. It is a program of the National Highway Traffic Safety Administration's office of EMS and is hosted by the University of Utah. We included all data provided by NEMSIS, and so did not exclude any of it.

Study Sample, Data Sources, and Measurements

The 9-1-1 call volumes were defined as the EMS activations triggered by calls to 9-1-1 submitted by member organizations and aggregated to the county-week level, using the variable “eResponse.05” with the answer “911 Response (Scene).” Ambulance call volumes were defined as EMS activations triggered by calls to 9-1-1 submitted by member organizations that then resulted in either transport by that EMS unit or transfer of care to another EMS unit, submitted by member organizations and aggregated to the county-week level, for example, Santa Clara County for week 4 of the year; this was done through aggregating responses that had the “eResponse.05” answer “911 Response (Scene)” and the “eDisposition.12” answers “Patient Treated, Transferred Care to Another EMS Unit” or “Patient Treated, Transported by this EMS Unit.” The 9-1-1 and ambulance rates were then calculated as county-week ambulance and 9-1-1 volumes divided by county population and multiplied by 100,000. County populations were obtained using the 2020 U.S. Census population data (12). Entries where there were no ambulance call volume in the county that week were initially represented as blanks, but were converted to zeros for analysis as they indicated zero calls for entry. Analysis was also conducted with blank entries dropped and gave similar results (Supplementary Tables 1–4).

Time intervals were calculated for each ambulance call and averaged at the county-week level. Response interval was defined as the difference in minutes between the time the ambulance unit arrived on the scene and the time the unit was en route to the scene. Scene interval was defined as the difference in minutes between the time the unit arrived on scene and the time the unit left the scene. Off-load interval was defined as the difference in minutes between the time the patient arrived at the hospital and the time the unit was back in service.

The presence of cardiac arrests for calls was assessed through the percentage of patients aggregated at the county and week levels who had a cardiac arrest prior to or after arrival of EMS. We used the variable “eArrest.01,” which contains the coded answer for the presence of cardiac arrest as “No,” “Yes, Prior to Any EMS Arrival,” and “Yes.” Due to the presence of multiple categories of cardiac arrest timing, we calculated the presence of cardiac arrest at any point as one minus the percentage of patients without cardiac arrest. For each call, the primary organ system of the chief symptom was recorded, aggregated, and converted to percentages at the county-week level. These percentages were then multiplied by the ambulance volume to obtain the volume for each of these variables.

Cumulative COVID-19 case and death volumes by county were obtained via the New York Times online repository (13). New COVID-19 cases and deaths by week and county were then computed and per-capita rates were calculated using the 2020 U.S. Census population data (14).

COVID-19 case and death rates at the county-week level were submitted to NEMSIS and merged with the aggregated ambulance volume along with the metrics described above. County identifiers were masked, such that observations over time within the same county could be matched but individual counties could not be identified and returned for analysis.

Outcomes

Primary outcomes included the correlation between either volume of ambulance calls resulting in transport or volume of EMS activations due to 9-1-1 calls and COVID-19 incidence or COVID-19 death rates.

Secondary outcomes included the correlation between ambulance response, scene, and offload intervals with COVID-19 incidence. We also sought the correlation between the volumes of ambulance calls for primary organ systems and COVID-19 incidence. After initial analysis, pulmonary and general systems were combined for analysis as COVID-19–related calls, and all other primary organ systems were then combined for analysis as non–COVID-19-related calls. This was the only post-hoc analysis and the chosen outcomes were otherwise determined before analysis began.

As tertiary outcomes, we included correlation between volume of cardiac arrests and COVID-19 incidence and death rates.

Statistical Analysis

Ordinary least squares regressions were then run using STATA, version 17 (StataCorp) with either COVID-19 rates or COVID-19 death rates as the independent variable and the dependent variables as outlined above. Location-invariant differences across weeks and time-invariant differences across counties were controlled for with week and county fixed effects, respectively. Robust SEs were clustered at the week level. These regressions were run for 2020 overall as well as by quarter of that year.

Our analysis studied separate outcomes: volume of ambulance calls resulting in transport, volume of EMS activations due to 9-1-1 calls, response interval, scene interval, offload interval, pulmonary/general organ systems, all other organ systems, and cardiac arrest, each with up to two explanatory variables: COVID-19 rates and deaths from COVID-19. We therefore made a Bonferroni adjustment on the usual p < 0.05 by dividing it by 16 (8 × 2) and then rounding down to get our preferred statistical significance level of p < 0.001.

RESULTS

Ambulance data were obtained for all 3142 U.S. counties of the 50 states and the District of Columbia from the NEMSIS for the year 2020, generating up to 134.676 observations per analysis. All EMS activations for 2020, totaling 43,488,767 submitted by 12,319 EMS agencies for the 50 states and the District of Columbia were included. Primary outcomes included that ambulance call volume resulting in transport was higher by 9.5 calls for each increase in 100 cases of COVID-19 at the county level (95% CI 8.6–10.3; p < 0.001) (Figure 1 , left). This correlation remained statistically significant for quarters 3 and 4 of 2020 (Table 1 ). Overall volume of EMS activations triggered by 9-1-1 calls was higher by 14.1 calls for each additional 100 cases at the county level and was indeed strongly correlated with community rates of COVID-19 when week and county were controlled for (95% CI 12.8–15.4). Ambulance call volume resulting in transport was also higher by 116 calls for each additional 100 deaths from COVID-19 at the county level (95% CI 95.5–136.4; p < 0.001).

Figure 1.

Relative ambulance volume compared with the first quartile of COVID-19 incidence for all ambulance volume (left) and pulmonary ambulance volume (right). Coefficients and 95% CI for county-week ambulance volumes relative to the first quartile (show as 0 for comparison) beginning in week 10 of 2020.

Table 1.

Correlation of Volume of EMS Activations Prompted by 9-1-1 Calls and Volume of Ambulance Calls Resulting in Transport with COVID-19 Incidence and Death Rates

Variable	Coefficient (Volume per Additional 100 COVID-19 Cases)	95% CI	Observations, n	p Value
Correlate to COVID-19 rate
Volume of ambulance calls resulting in transport
Overall unadjusted	+6.4	3.6 to 9.1	134,676	< 0.05
With fixed effects	+9.5	8.6 to 10.3	134,676	< 0.001
Quarter 2	+6.3	3.9 to 8.8	40,716	< 0.05
Quarter 3	+10.9	9.4 to 12.4	40,716	< 0.001
Quarter 4	+6.7	5.2 to 8.1	40,716	< 0.001
Volume of EMS activations due to 9-1-1 calls
Overall unadjusted	+4.1	–0.095 to 8.2	134,676	> 0.05
With fixed effects	+14.1	12.8 to 15.4	134,676	< 0.001
Quarter 2	+9.2	5.8 to 12.4	40,716	< 0.05
Quarter 3	+13.9	12.0 to 15.8	40,716	< 0.001
Quarter 4	+9.9	7.5 to 12.3	40,716	< 0.01
Correlate to deaths from COVID-19
Ambulance volume resulting in transport
Overall unadjusted	+228.389	157.1 to 300.0	134,676	< 0.01
With fixed effects	+116.0	95.5 to 136.6	134,676	< 0.001
Quarter 2	+137.7	91.9 to 18.4	40,716	0.05
Quarter 3	+51.4	–51.0 to 15.4	40,716	> 0.05
Quarter 4	+58.1	32.6 to 83.5	40,716	< 0.05

EMS = emergency medical services.

In terms of secondary outcomes, efficiency of EMS resources was also examined through the correlation of response interval, scene interval, and offload interval (EMS time spent at the accepting hospital), with marginal increases in COVID-19 cases (Table 2 ). However, after adjusting for location-invariant differences across weeks and time-invariant differences across counties, no correlations met our predetermined statistically significant threshold of p < 0.001, including both overall and for individual quarters of 2020.

Table 2.

Correlation of Emergency Medical Services Time Intervals and COVID-19 Incidence

Variable	Coefficient (Change in Minutes per Additional 100 COVID-19 Cases)	95% CI	Observations, n	p Value
Correlate to COVID-19 rate
Response interval
Overall unadjusted	–0.027	–0.398 to 0.344	120,562	> 0.05
With fixed effects	+0.078	–0.241 to 0.397	120,562	> 0.05
Quarter 2	–0.287	–0.509 to –0.065	36,643	0.01
Quarter 3	+0.141	–0.165 to 0.447	36,733	> 0.05
Quarter 4	+0.201	–0.203 to 0.604	35,980	> 0.05
Scene interval
Overall unadjusted	–0.092	–0.339 to 0.154	120,177	> 0.05
With fixed effects	+0.191	0.037 to 0.344	120,177	< 0.05
Quarter 2	+0.769	–0.264 to 1.801	36,551	> 0.05
Quarter 3	+0.251	–0.32 to 0.822	36,586	> 0.05
Quarter 4	+0.161	0.05 to 0.272	35,844	< 0.01
Offload interval
Overall unadjusted	+0.641	0.264 to 1.018	118,821	< 0.001
With fixed effects	+0.329	–0.039 to 0.698	118,821	> 0.05
Quarter 2	+1.591	0.424 to 2.758	36,111	< 0.01
Quarter 3	+0.542	–0.189 to 1.273	36,201	> 0.05
Quarter 4	+0.180	–0.194 to 0.554	35,457	> 0.05

To assess whether COVID-19 was affecting the cause of the ambulance call and to assess whether patients with emergent, non–COVID-19 concerns were indeed avoiding care, the primary organ system of these calls was examined, and aggregated to a dichotomy of combined pulmonary and general calls, proxies for COVID-19, vs. all other systems combined, proxies for non–COVID-19 illness. The volumes of combined pulmonary and general calls were higher by 9.2 calls and strongly correlated with each additional 100 COVID-19 cases (95% CI 8.4 to 10.0; p < 0.001) (Table 3 , Figure 1, right). The volume of calls for all other organ systems combined, proxies for non–COVID-19 illness, was unaffected by additional COVID-19 incidence (0.10 more calls per 100 additional cases of COVID-19; 95% CI –0.64 to 0.85; p > 0.05).

Table 3.

Correlation of Volume of Ambulance Calls by Primary Organ System with COVID-19 Incidence

Variable	Coefficient (Calls per Additional 100 COVID-19 Cases)	95% CI	Observations, n	p Value
Correlate volume of primary organ system to COVID-19 rate
Combined pulmonary and general with fixed effects	+9.2	8.4 to 10.0	134,676	< 0.001
Combined all other organ systems with fixed effects	+0.10	–0.64 to 0.85	134,676	> 0.05

As a tertiary outcome and to examine whether the overall severity of the ambulance calls was associated with community COVID-19 rates, potential correlations between the volume of calls with cardiac arrest occurring at any point during the EMS call cycle, of both transported and nontransported patients, were looked at and both COVID-19 incidence and deaths attributed to COVID-19 (Table 4 ). The volume of cardiac arrests was not higher with additional COVID-19 cases or additional deaths from COVID-19 (0.60 more cardiac arrests per 100 cases of COVID-19; 95% CI 0.60–0.91; p > 0.05; 7.4 more cardiac arrests per 100 deaths from COVID-19; 95% CI 4.1–10.6; p < 0.05).

Table 4.

Volume of Cardiac Arrests and COVID-19 Incidence and Death Rates

Variable	Coefficient (Cardiac Arrests per Additional 100 COVID-19 Cases)	95% CI	Observations, n	p Value
Correlate to COVID-19 rate
Cardiac arrest
Overall unadjusted	+0.26	–0.014 to 0.53	134,676	> 0.05
With fixed effects	+0.60	0.29 to 0.91	134,676	> 0.05
Quarter 2	–0.049	–0.42 to 0.32	40,716	> 0.05
Quarter 3	–0.40	–0.65 to –0.16	40,716	> 0.05
Quarter 4	+0.44	0.19 to 0.69	40,716	> 0.05
Correlate to COVID-19 death rate
Cardiac arrest
Overall unadjusted	+4.0	–0.14 to 8.1	134,676	> 0.05
With fixed effects	+7.4	4.1 to 10.6	134,676	< 0.05
Quarter 2	+8.1	–7.6 to 23.7	40,716	> 0.05
Quarter 3	–8.3	–13.6 to –3.0	40,716	> 0.05
Quarter 4	+4.2	0.094 to 8.2	40,716	> 0.05

DISCUSSION

Optimal utilization of EMS during a pandemic requires an understanding of the factors contributing to its use. Higher rates of COVID-19 infection were strongly correlated with elevated ambulance use, as well as 9-1-1 call volume. Although these are intuitive results, they have not been reported previously and confirmation of this effect enables appropriate planning of emergency services utilization during this and future public health care crises. Moreover, as the COVID-19 pandemic enters a more baseline-surge status, an understanding that surges do bring increased EMS utilization can ensure critical staffing and ambulance vehicles are present to enable continued quality of care.

Conversely, the overall absence of correlation between rates and response interval suggests that despite dire images at the beginning of the pandemic of a collapsing system, the EMS system was not stressed to the point of compromise. Were that to be the case, it would be expected that higher COVID-19 rates would be associated with higher response intervals as personnel would be unavailable for new calls. Similarly, offload intervals were not statistically correlated with COVID-19 rates overall or during any phase of the pandemic, suggesting that hospitals were appropriately able to handle influxes of patients and no backing up of waiting ambulances occurred. That scene interval was not statistically associated with marginal increases in COVID-19 rates is consistent with our tertiary result that indicates that the rates of cardiac arrest during calls were not associated with COVID-19 rates, suggesting that the severity of cases was consistent throughout 2020.

That this association of ambulance rates with COVID-19 incidence is linked to COVID-19 illness itself is furthered by analysis in the correlations of primary organ system for the calls with the COVID-19 infection rates. The strong correlation between both the percentage of calls related to “pulmonary” and “general” concerns and COVID-19 infection rates suggests that COVID-19 infection, and not illness rates overall, were associated with the increase in ambulance rate. Because COVID-19 is primarily a respiratory illness, with 50% reporting cough and pneumonia as the most frequent severe manifestation, these patients would fall under the “pulmonary” category (15,16). Beyond respiratory symptoms, fever and myalgias are common presentations and would be characterized under the “general” category. Together, these account for the predominance of initial COVID-19 presentations (15).

Crucially, the ominous concerns dramatically reported at the start of the pandemic that patients with emergent non–COVID-19 illness were avoiding care were not borne out by the data and rates of EMS utilization for these conditions with changes in COVID-19 incidence, including during the second quarter of 2020 at COVID-19’s onset (17). This is consistent with the few studies that have examined this topic, which have showed no change in volume for obstetric activations in Maryland and myocardial infarction or stroke volume in the U.K.’s West Midlands (10,11).

Limitations

Limitations to this study include assessing these potential correlations across counties but only for the entire United States and not at a stratified, more granular sublevel of only specific regions or states. The data use agreement for the NEMSIS data set precluded its utilization if it would identify the counties. Consequently, given that the association with more granular regional information in conjunction with the weekly COVID-19 rates would likely enable identification of specific counties, the extent these observed correlations varied at the regional or state level could not be further characterized.

The study was also limited by the NEMSIS database itself. Although the NEMSIS database has grown significantly since its inception in 2009, it does not include all 9-1-1 activations, thus rendering it a convenience sample. That said, this is mitigated by the rapid expansion of the database and by 2020, the NEMSIS database included approximately 19.5 million 9-1-1 activations (18). Comparatively, 2019 data from the National Hospital Ambulatory Medical Care Survey reported 23.3 million ambulance transfers to hospitals (19). Together these suggest that the NEMSIS database captures the vast majority of 9-1-1 activations. Selection bias is also possible in the NEMSIS database due to differential levels of training among contributing reporters (17).

In addition, the data were drawn solely from 2020, and as the pandemic matured and its presence further integrated into daily life during 2021 and 2022, it is plausible that the correlation with COVID-19 rates and ambulance rates evolved as well. Similarly, as variants of the initial virus emerged throughout 2021, it is also possible that their respective severity and transmissibility, and in turn strain on EMS personnel, altered these correlations as well. National vaccination campaigns only started in the final weeks of 2020, so any role this played in altering the ambulance and infection rate calculus could not be assessed.

CONCLUSIONS

With the COVID-19 pandemic continuing and future pandemics looming, understanding how EMS are used during these times remains critical. Our work found that utilization is correlated with infection rates, suggesting a need to be cognizant of staffing and ambulance vehicle availability during pandemic surges. Although response and offload intervals showed some correlation during the initial phase of the pandemic. The absence of a correlation with COVID-19 incidence and response, scene, and offload intervals suggests that the EMS system was indeed able to absorb these varying levels of activity well, and did not strain under the weight of the pandemic, despite initial anecdotal evidence to the contrary. Patients with emergent non–COVID-19 illness did not avoid medical care either. Future studies are needed to assess how these correlations continue to evolve through the more current eras of the pandemic, including the roles of both vaccinations and virus variants.

Article Summary

1. Why is this topic important?

Understanding how COVID-19 incidence affected emergency medical services (EMS) utilization during the COVID-19 pandemic is crucial for ongoing and future pandemic planning responses.

2. What does this study attempt to show?

This study correlates COVID-19 incidence with EMS utilization to examine the relationship between these two variables.

3. What are the key findings?

Ambulance call volume overall and calls due to pulmonary or general concerns, proxies for COVID-19 symptoms, increased with additional cases of COVID-19. The volume of calls due to all other organ systems, proxies for emergent non–COVID-19 illness, did not significantly change with higher COVID-19 incidence.

4. How is patient care impacted?

Knowledge that use of EMS scales with COVID-19 incidence is crucial to prehospital care planning. That patients with emergent non–COVID-19 illness continued using EMS also means that hospitals must plan for larger patient volumes without compromising care.

Declaration of Competing Interest

XX report no conflict of interests. No funding was received for this work.