A large multicenter study shows an increased mortality risk to trauma patients during a time when NYC was considered the epicenter for COVID-19. Non-COVID patients with ISS<15 were more likely to die compared to the same subgroup in 2018/2019.

BACKGROUND

During early spring 2020, New York City (NYC) rapidly became the first US epicenter of the COVID-19 pandemic. With an unparalleled strain on health care resources, we sought to investigate the impact of the pandemic on trauma visits and mortality in the United States' largest municipal hospital system.

PATIENTS AND METHODS

RESULTS

A total of 9,196 injured patients were cared for at NYC H + H's five level 1 trauma centers during the study periods for 2018, 2019, and 2020. Compared with January and February 2018 and 2019, there was no significant difference in median number of trauma visits per month in January and February 2020. There was, however, a significant decrease in median number of trauma visits, by approximately 25%, in the early pandemic period of March through May 2020 as compared with the same time frame in 2018 and 2019 (Table1).

Patients presenting to our trauma centers were similar in most regards before and after the beginning of the COVID19 pandemic. There were several small statistically significant differences; many of which were not clinically important (Table1). The age, race, and ethnicity of patients were similar for all periods examined. Moreover, patients were not substantially more severely injured in the early pandemic compared with prior years, as there were no important differences in hypoxia, hypotension, GCS, need for thoracostomy tube, endotracheal intubation, and need for blood transfusion. In addition, the ISS was similar across all periods.

Compared with prior years, however, patients presenting for penetrating injuries increased significantly by approximately 50% in the early pandemic period (15% in March to May 2018 and 2019 vs. 22% in March to May 2020). Table1 shows the breakdown by specific mechanism. The cut/pierce and gunshot wound mechanisms represented 15% and 7% of injuries during the early pandemic period of March to May 2020, respectively. In all other periods analyzed, these mechanisms represented about 11% and 3% of injuries, respectively.

The pandemic brought some important changes to our trauma centers (Table2). While there were no substantial differences in the thoracotomy or laparotomy rates, or the hospital and ICU LOS, there were significantly fewer patients transferred out and less patients discharged to rehabilitation facilities in March to May 2020 compared with the 2 years prior. Of note, in the years before the pandemic, the in-hospital overall mortality for trauma patients was 2.6%. During the early pandemic period in the spring of 2020, however, the in-hospital morality rate increased significantly by doubling to 5.3% (Table2).

To account for potential confounding in the association of the pandemic and mortality, we performed inverse probability weighting using the propensity score for treatment and compared patients treated in March to May 2018 and 2019 to those in March to May 2020. Before propensity weighting, standardized differences and variance ratios for the baseline covariates ranged from −0.08 to 0.18 and 0.6 to 1.7, respectively. After propensity score weighting, differences observed for the baseline covariates that were evident prior in the sample were no longer present. Moreover, after propensity weighting, standardized differences and variance ratios for the baseline covariates were much improved and within the recommended limits, ranging from −0.01 to 0.02 and 1.0 to 1.1, respectively.

Table3 demonstrates the mortality risk for injured patients in the early pandemic months compared with prior years after adjusting for important confounders using propensity score weighting. Here, we note that, although there was a significant difference in mortality overall (prepandemic: 79 [2.8%] vs. early pandemic: 127 [4.6%];p = 001), once the 353 patients with COVID-19 were excluded from the analysis, the mortality difference overall was attenuated and no longer statistically significant (prepandemic: 69 [2.6%] vs. early pandemic: 88 [3.3%];p = 0.17). In the early pandemic months, the unadjusted mortality risk was significantly higher for injured patients with COVID-19 (37 [10.5%]) compared with those without COVID-19 (39 [3.6%];p < 0.001).

We performed several subgroup analyses. When evaluating patients by ISS, we found that the adjusted mortality risk was significantly higher for those patients with lower ISS but not in those with higher ISS (Table3). This relationship persisted even after excluding the COVID-19 patient cohort. The in-hospital mortality rate for non–COVID-injured New Yorkers with an ISS of <15 cared for during March to May 2020 (2.0%) was 2.7 times higher than that of patients admitted during the same time frame in 2018 and 2019 (0.8%;p = 0.005). In many of the other subgroups, we also found significant differences in the adjusted mortality risk when comparing the early pandemic to prior years, but after excluding COVID-19 patients, most of these differences were attenuated and no longer statistically significant. We did find, however, that, after excluding the COVID-19 cohort, patients with penetrating injuries were significantly less likely to die in the early pandemic compared with prior years. Specifically, penetrating trauma victims without COVID-19 admitted during the early spring of 2020 were 40% less likely to die in the hospital compared with the same time frame in 2018 and 2019.

Comorbidities varied slightly among patients in the early pandemic compared with prior years (Table4). Importantly, mental health disorders and substance abuse were significantly increased. The proportion of patients experiencing an in-hospital complication, however, for the most part remained unchanged (Table5). Of note, adjusting for patient comorbidities and in-hospital complications had little effect on the mortality risk experienced by trauma patients during the early pandemic (Table6).

DISCUSSION

In the first few months of the COVID-19 pandemic, SARS-CoV-2 exhausted NYC's health care system and resources, peaking at more than 5,000 new cases per day.² Reported fatality rates neared 10% during this same period with about a third of hospitalized patients succumbing to COVID-19 infection.^2,3 In response to this burden, elective surgeries were canceled, and hospital beds and staff were reallocated to the care of COVID-19 patients throughout NYC. As trauma surgeons and surgical critical care physicians were redeployed to COVID-ICUs, an early report from the Greater New York American College of Surgeons Committee on Trauma (ACS COT) recognized and acknowledged the new burden on the NYC trauma system.¹⁷ This report recommended frequent institutional assessments of the effect of the crisis on trauma center functions. Given the lack of governmental or consensus guidance for trauma system maintenance at that time, the Greater New York ACS COT advocated to preserve trauma infrastructures and capacity whenever possible in each region.

In this study of the NYC's municipal trauma system of five level 1 trauma centers, we found that overall trauma visits declined significantly during the initial COVID surge in Spring 2020 compared with the two preceding years of the same period, consistent with our primary hypothesis. Despite this overall decline in trauma visits, the rates of penetrating trauma significantly increased by almost 50% during this same period. Unexpectedly, we found that there was an increase in the mortality rate compared with prior years when accounting for important confounders. However, after excluding COVID-19–positive patients, the increased risk of death persisted only in the subset of patients with low ISS. Specifically, non–COVID-19 patients who were less severely injured were more than twice as likely to die during the early pandemic compared with prior years.

The effect the pandemic and “stay-at-home” orders had on trauma care in this country has been evaluated by authors in three other major cities across the United States. One of the first reports came from Philadelphia where, similar to our experience, there was a 20% decrease in trauma admissions but a 35% increase in penetrating trauma.¹⁰ A subsequent large study from San Francisco showed no change in overall violent traumas but did demonstrate a small statistically significant increase in homicides in association with their shelter-in-place order.¹¹ However, these two studies did not examine the effect of the pandemic on trauma mortality. A recently published study from Los Angeles County, however, did examine mortality and found an overall decrease in both injury severity and a significant decrease in the mortality rate of trauma patients during the early pandemic compared with the prior years.¹² This led the authors to suggest that “in future lockdowns … not only [is] there no need for trauma closures or additional critical care resources but … areduced need for these resources, may be taken into account.” There remains the question, however, of why did the NYC municipal trauma system, in contrast to the LA County experience, demonstrate an increase in mortality during the early pandemic among its less severely injured patients?

To begin to answer this question of increased mortality, it is important to note that NYC was the first large metropolitan area in the United States to be affected by the COVID-19 pandemic. Without any prior experience to draw from, NYC quickly became the most severely affected city and the US epicenter for the pandemic during the early spring 2020.^2,5,18,19 Elmhurst Hospital, one of the five trauma centers included in our study, was frequently noted to be one of the worst affected hospitals in the country and was called the “epicenter of the epicenter” by the media.²⁰ The Bronx, where two of our trauma centers are located, had the highest hospitalizations and fatalities due to COVID-19 per 100,000 population in the city and, at that time, in the world.²¹ In short, NYC H + H was at the forefront of the US COVID-19 response. Many of our surgical ICUs and most trauma surgeons who practice surgical critical care throughout the NYC municipal trauma system were “redeployed” to primarily care for COVID-19 patients during this time. This redeployment occurred even as the trauma system still needed to care for the injured. With much of our human resources reallocated and our supplies and equipment frequently depleted, providing optimal care for trauma patients became challenging.^22,23

The Greater New York ACS COT from the beginning encouraged frequent reassessments of the burden the pandemic had on the trauma system and advocated for its protection.¹⁷ Under this advisement and our trauma system leadership, the trauma teams were never fully dismantled, and these resources were protected for high-priority trauma activations in our emergency departments. Without outpatient clinic requirements and elective surgical caseloads, general surgeons, some subspecialists, and operating rooms were available for patients requiring emergent surgical interventions. Despite these recommendations and our best efforts to maintain trauma care at our centers, our study suggests that the pandemic placed a substantial and unprecedented strain on the NYC municipal trauma system. The enormous burden of COVID-19 disease that our trauma system faced, which was unparalleled in the United States at the time, may have contributed to the increase in mortality of trauma patients in several possible ways.

First, critically ill COVID-19 patients, often with serious comorbidities and associated complications (e.g., acute kidney injury requiring renal replacement therapy) were a substantial diversion from patients thought to be “less sick” or less severely injured.^24,25 One can easily understand how this distraction could potentially compromise the usual standard of care devoted to trauma patients in normal unstrained circumstances, given the demanding needs of critically ill COVID-19 patients. Second, the COVID-19 burden depleted material resources, in particular critical care beds normally reserved for trauma patients. For instance, one of our system's surgical ICUs, which typically has 55% to 60% of beds occupied by trauma patients, was filled to 100% capacity with critically ill COVID-19 patients after the first week of the surge. As a whole, the entire NYC municipal trauma system experienced similar circumstances necessitating the conversion of alternative spaces (e.g., waiting areas, hallways) into makeshift critical care units. At the same time, the COVID-19 burden depleted human resources, as many providers at all levels and specialties were overworked with little to no time off early in the pandemic. Many house staff and subspecialists were reallocated, and providers themselves became ill from COVID-19, removing them from the workforce temporarily or, worse yet, permanently.

Despite the increase in mortality, our study elucidated an interesting and unexpected finding, specifically, a significant, although small, survival benefit to non–COVID-19 patients with penetrating injuries admitted during the early pandemic. This was especially interesting because we also found no significant difference in the number of thoracotomies or laparotomies, or the ICU LOS as compared with prior years. Although this phenomenon may be the result of unmeasured confounders in our study, there are also other reasons one may speculate as to why victims with penetrating traumas may have seen improved survival despite the numerous strains on the system described. For one, the pandemic placed extraordinary strain on emergency medical systems (EMS) worldwide and resulted in prolonged response times as COVID-19 cases increased.²⁶ One study cited a 3-minute increase in NYC EMS response time in the early days of the pandemic, corresponding with the period of this study.²⁷ While quantifying EMS response times is beyond the scope of this study, it follows that likely increased response times could impact the survival profile of patients with penetrating trauma. Were patients with enough physiologic reserve to withstand the prolonged EMS transport times as compared with prior years selected for survival while the other more severely injured patients died in the field?²⁸ While further study is certainly required to confirm this theory, this could be one explanation for the increased survival seen in patients with penetrating trauma during the pandemic period.

As surrogates for resource utilization, our study looked at several measures including transfers-out and discharges to rehabilitation facilities as secondary outcomes. Many of our centers had to go on brief periods of diversion to accommodate the critically ill COVID-19 patients. Thus, we found an increased number of trauma transfers out to other acute care facilities in this study. Our study also showed significantly fewer discharges to acute rehabilitation facilities during the study period, suggesting that these centers were likely equally overwhelmed and at capacity with patients. This may have led to a “bottleneck” and hindered our ability to discharge trauma patients to these facilities during the study period. Notwithstanding, overall hospital LOS was significantly lower by 1 day, but this trend was also noted to be significant in the months leading up to the pandemic.

There are several key limitations to our study that should be taken into account when interpreting our results. First, the design of our study did not allow for an in-depth analysis of material or human resources that may have shed more light on how their depletion may have affected patient mortality. It is possible that the lack of available beds may have changed triage and discharge practices at many of our facilities such that those with minor injuries who may have otherwise been admitted were now discharged from the emergency department, thus increasing the average injury severity of remaining patients. Determining which resource depletion contributed the most to the increase in mortality, whether it be human resources or ventilators or blood products, may help other cities in preventing deaths due to trauma during future pandemics. In addition, during the early pandemic, it is important to note that COVID-19 testing was not widespread and the tests that were available were not very sensitive.^29,30 Thus, patients were often labeled as carrying a clinical diagnosis of COVID-19 despite negative testing. This creates an unavoidable potential for bias and error. Specifically, it is possible that those patients with perceived minor injuries may not have been tested as frequently as those with severe injuries, partially accounting for the increased mortality risk we detected. Lastly, like most retrospective cohort studies, our study was subject to confounding by particular variables related to both the exposure and the outcome. To account for potentially important confounders, we created a propensity score for treatment and used it to balance these confounders across periods. While we attempted to include as many important confounders in the propensity score as reasonably possible using a large multicenter data set, residual confounding may still exist.

In summary, we present the first study describing the impact of the COVID-19 pandemic on trauma patient mortality within the multicenter municipal trauma system of NYC during its time as the US epicenter. Our unanticipated increase in mortality in non-COVID patients with low ISS should not be seen as a failure of the trauma system as many appreciate the nothing short of heroic efforts displayed by NYC health care workers during this time. Instead, this report should be used to demonstrate the importance of supporting and safeguarding locoregional trauma systems and their resources in the event of another global pandemic or other large-scale event, such as a natural disaster or terrorist attack, which may lead to a surge in critically ill and critically injured patients. More study beyond this report is warranted to further elucidate how the pandemic has affected trauma care and to inform how trauma systems can best communicate and share resources during catastrophic events, including future pandemic surges.

AUTHORSHIP

A.L. contributed in the literature search, study design, data interpretation, data collection, data analysis, manuscript writing, and critical revision. M.E.S. contributed in the literature search, study design, data interpretation, manuscript writing, and critical revision. H.M. contributed in the data collection. J.A.M. contributed in the literature search, study design, data interpretation, data collection, data analysis, manuscript writing, and critical revision. G.A., M.E.S., M.B., V.R., E.C., S.H.R., and S.H.T. contributed in the manuscript writing and critical revision.

ACKNOWLEDGMENTS

We thank Carrie Garcia, RN; Margaret Ewen, MS, PA-C; Fred Harry, RPA-C; Mark Anderson, RN; Joanne Scarpinato, LPN; and all the rest of the Trauma Program Office staff.

DISCLOSURE

The authors declare no conflicts of interests.

Footnotes

Contributor Information

Melvin E. Stone, Jr., Email: melvin.stone@nychhc.org.

Harley Markel, Email: markelh1@nychhc.org.

George Agriantonis, Email: agriantg@nychhc.org.

Marko Bukur, Email: marko.bukur@nyulangone.org.

Sherry Melton, Email: meltons@nychhc.org.

Valery Roudnitsky, Email: valery.roudnitsky@nychhc.org.

Edward Chao, Email: edward.chao@nychhc.org.

Srinivas H. Reddy, Email: srinivas.h.reddy@nychhc.org.

Sheldon H. Teperman, Email: sheldon.teperman@nychhc.org.

James A. Meltzer, Email: james.meltzer@nychhc.org.

REFERENCES