Introduction

The current SARS-CoV-2 (COVID-19) pandemic has posed unprecedented challenges for the organisation of healthcare systems and their financing. The rapid increase in hospitalisations required the suspension of elective medical services across a number of jurisdictions and reduced resources to hospital units that compete for the same resources as those used to treat COVID-19 [1]. In addition, resources including equipment and medical personnel were diverted in anticipation of managing outbreaks [2]. Trauma care and all its components were one of the most affected by the rearrangement of health care services during the first months of the pandemic [2, 3]. The justification for diverting resources and personnel from trauma units was driven by expectations that the contextual policies implemented to stop the spread of the virus such as lockdown policies, stay at home orders (SAH), and suspension of social activities (i.e. sports, entertainment, closure of pubs and restaurants, etc.) [4,5,6] would significantly reduce major trauma presentations. Nevertheless, researchers and clinicians have expressed concerns regarding the negative impact of COVID-19 outbreaks on the ability of health care systems to provide timely assessment and acute therapies to patients with major trauma [7, 8].

In this systematic review and meta-analysis, we aimed to investigate changes in the epidemiology of major trauma presentations during the implementation of movement restriction measures (such as lockdowns) to manage the first wave of the COVID-19 pandemic. This setting provides a relatively more homogeneous setting for a cross-country analysis, as subsequent waves were likely to be more heterogenous due to country-specific determinants such as additional investments in hospital resources, the impact of the first wave, vaccine availability.

We hypothesised that: (a) social restrictions will lead to a reduction in major trauma admissions; (b) there will be an increase in the severity of major trauma presentations during periods of social restriction, compared with pre-pandemic; and (c) there will be a reduction in traffic-related injuries and a corresponding increase in trauma at home.

Methods

Search strategy

The protocol for this review is available online (PROSPERO; CRD42020224827). This systematic review and meta-analysis aligns with the PRISMA guidelines [9]. The systematic search was initially performed in six databases on 19 January 2021, MEDLINE, Embase, CINAHL, the Cochrane Library, the WHO COVID-19 and LitCovid. A search of grey literature was also conducted via Google in the same timeframe, adopting the same search strategy used in Medline. We defined terms for SARS-CoV-2, policy restrictions, trauma, and hospitalisations/caseloads based on cohort studies reporting on major trauma presentations both before and after the COVID-19 pandemic outbreak [7, 10,11,12], and via review of governmental documents and media sources. The search strategy was created by the authors and peer-reviewed by a senior librarian. The search was updated on 26 July 2021 prior to submission. The full search strategy is available in the supplementary Appendix (Table A1–A5).

Eligibility criteria

Major trauma was defined as per patients requiring trauma resuscitation based on institutional criteria on arrival to the emergency department. Although this definition may not be uniform across studies like an Injury Severity Score (ISS) based inclusion criteria, it allows consistency for comparison of pre-pandemic and pandemic volumes by using the same criteria in individual institutions.

Cohort studies were included if they reported differences on the number of admissions, between patients admitted due to major trauma after the COVID-19 pandemic outbreak and patients admitted due to major trauma before the COVID-19 pandemic outbreak in the respective health care settings (see Table A6 in the supplementary material). Also articles not published in English language and commentaries or editorials were excluded.

Study selection and data extraction

We performed de-duplication in EndNote [13] and all records were exported to Covidence [14] for screening. Two reviewers (MA, MH) independently screened the titles and abstracts for relevance, and then extracted and selected relevant full-text records. Discrepancies were resolved through discussion at each stage, with any disputes of eligibility resolved by a third author where required (ZJB).

Two authors (MA and MH) developed the data extraction form in Excel. The form records bibliographic information, the number of admissions, aetiology, and pre- and post-social restrictions due to COVID-19, as well as demographics, location, and the severity of the social restrictions implemented. The extraction form was piloted using a sample of six randomly chosen studies and revised after discussion amongst authors. One author (MA) extracted the information of interest from the included studies using the final version of the data extraction form, and a second author (MH) double-checked 20% of the included items. In the case of missing or unclear data, we contacted the corresponding author of the study to provide additional information. Two studies were excluded from the meta-analysis during this process (both because relevant data was not available). For studies that reported multiple comparison periods, admissions in previous years were considered for the control period to control for potential seasonality effects.

Data analysis

A meta-analysis was conducted with ReviewManager 5.3 and Stata16. We conducted meta-analyses of proportions, based on the pooled differences between the intervention and control conditions for each hypothesis. We conducted 19 separate meta-analyses with forest plots, for overall mortality, severity, mechanisms and location on injuries. The odds ratios (ORs) for mortality, mechanism of trauma, and location of trauma were calculated using the Mantel–Haenszel method with random effects model regardless of heterogeneity. We used I2 statistics and Cochran’s Q test to evaluate inter-study heterogeneity, which was deemed to be significant if I2 > 50% or p < 0.10. Descriptive statistics were used for demographic variables. To investigate the impact of differences in COVID-19 prevalence at the time of the studies, a subgroup analysis was conducted at the continent level. Results of this analysis are reported in the Appendix.

Outcomes

Primary outcomes were number of admissions, mortality and clinical characteristics of patients admitted (i.e., average length of stay (LOS), patients requiring intensive care (ICU), patients requiring mechanical ventilation, ISS, Glasgow Coma Scale (GCS)), and trauma aetiology versus baseline. Secondary outcomes were patients’ demographics characteristics.

Bias assessment

The quality of the observational studies was assessed using the Newcastle–Ottawa Score (NOS) [15]. The GRADE approach was used to assess the quality of evidence of estimates as high, moderate, low, or very low based on considerations of study design (observational studies are rated low confidence) [16]. See also Table A7 in the Appendix for a summary of all the checks conducted.

Results

The database search returned 3719 records in total. After removal of duplicates and limiting to language and publication type, 2777 records underwent title and abstract screening and 522 underwent full-text screening. 35 cohort studies were included in the analysis (see Fig. 1 [17]).

Fig. 1
figure 1

PRISMA flow diagram of studies selection process

Full size image

Table 1 reports the characteristics of the studies included in the analysis and the associated level of policy restrictions. The 35 studies included patients from 14 countries. The most represented countries are the US (n = 9), the UK (n = 4) and South Africa (n = 4). There were significant between-country differences in terms of the severity of the pandemic and the consequent stringency of the restrictions at the time of the study, both of which might affect the outcomes considered [18]. Therefore, the policy restrictions imposed at the time of the study were retrieved from the included articles, and collated as reported in the studies.

Table 1 Characteristic of the included studies
Full size table

Major trauma admissions and clinical characteristics of patients admitted

During the implementation of social restriction orders in response to the COVID-19 outbreak, there was a statistically significant reduction in major trauma presentations overall compared with control periods (mean −24%; p < 0.01; 95%CI [−0.31; −0.17]). The results are reported in Fig. 2. Only 4 studies out of 35 reported a positive percentage increase compared to the pre-restriction period [19,20,21,22] (see Figure A1 in the Appendix for a breakdown at the continent level).

Fig. 2
figure 2

% variation of major trauma admissions pre-COVID-19 versus COVID-19

Full size image

Key clinical characteristics of the hospitalised cases are summarised in Table 2. Seventeen studies reported a measure of ISS for patients admitted to hospital [8, 10, 11, 22,23,24,25,26,27,28,29,30,31,32,33,34,35]. The median Injury Severity Score (ISS) remained unchanged in most of the studies between the two periods. We also collated the number of patients with an ISS > 12, which was reported in six studies [8, 11, 19, 23, 36, 37] (Figure A2 in the Appendix). The OR of the number of patients in this category before and during social movement restrictions derived from the meta-analysis was 1.17 (95% CI [0.77, 1.79]), suggesting no change in the number of severely injured patients. Morris et al.[38] reported similar results based on the South African Trauma Scale. There were two exceptions. Riuttanen and colleagues [29] reported a significant increase in the median ISS in four Finnish hospitals after the COVID-19 restrictions (18 vs 21, p = 0.008). In a study conducted in Israel, Rozenfeld et al. [39] found statistically significant variations by ISS groupings; a decrease for trauma admissions with an ISS 1–8, and an increase for ISS 9–14 and ISS 16–24, and no change for highly severe injuries (ISS 25–75).

Table 2 Clinical characteristics of major trauma
Full size table

In terms of the LOS during movement restrictions, overall there was no significant difference in values compared to the pre-pandemic period (n = 10 studies; mean 0.52, p = 0.12). Two studies [27, 32] reported an increase in the LOS after the introduction of the restrictions. There was no statistically significant difference between the two periods in terms of patients admitted to the ICU (n = 12 studies; mean: −0.006, p = 0.25) (Figure A3 in the Appendix). Six studies reported information concerning patients requiring mechanical ventilation [8, 22, 23, 33, 35, 37]. A statistically significant reduction was reported for three studies [22, 23, 37], while the remaining three reported a statistically significant increase compared to the previous period [8, 33, 35]. No statistically significant difference was found for the number of days on a ventilator [8, 22,23,24, 27, 33, 35] or for the overall GCS score in any study [8, 23, 24, 26, 31, 33, 35].

Mortality

Eighteen of the included studies reported the mortality of admitted patients summarised in Fig. 3. The studies provide no evidence for a change in mortality during the COVID-19 period (OR:0.94, 95%CI [0.80,1.11]). Three papers [24, 31, 33], all run in the US, had a weight greater than 10% in the analysis. When we ran the analysis excluding these three publications, the results were consistent with the primary analysis (OR:0.87, 95%CI [0.68,1.11]). The subgroup analysis by continent confirms this, excepting countries in Asia, where three studies [20, 26, 37] reported a statistically significant reduction in the period considered (OR:0.67, 95%CI [0.46, 0.98]) (Figure A4 in the Appendix).

Fig. 3
figure 3

Mortality of major trauma admitted patients

Full size image

Mechanisms of injury

Table 3 reports the distribution of injury mechanisms in the included studies. There was an overall statistically significant reduction in trauma due to motor vehicle/road traffic collisions (OR:0.70;95%CI [0.61,0.81])[8, 10, 11, 22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43]. A sensitivity analysis showed that this outcome was not statistically significant in South Africa and Asian countries (see Figure A5 in the Appendix). There was no statistically significant difference between the two periods for motorbike (OR:0.89; 95%CI [0.73,1.08]) [10, 11, 22,23,24, 29, 33,34,35,36, 39, 40, 42] or bicycle collisions (OR:1.08; 95%CI [0.80, 1.47]) [10, 11, 22, 23, 29, 30, 34,35,36, 39, 40, 42]; however, there was a statistically significant reduction observed for pedestrian collisions (OR:0.63; 95%CI [0.51,0.78]) [10, 11, 22,23,24,25, 29, 30, 33,34,35,36, 38,39,40]. An overall statistically significant reduction (OR:0.66; 95%CI [0.46, 0.94]) was observed for major trauma admissions due to assault during the period of restrictions in sixteen studies [8, 10, 11, 20, 22, 24, 25, 27,28,29,30, 33, 36, 37, 40, 44]. Among these, three studies [8, 20, 37] reported a statistically significant reduction, while the others reported a non-statistically significant reduction or a stable trend over time [10, 11, 22, 24, 25, 27,28,29,30, 33, 35, 36, 40]. The subgroup analysis showed that a statistically significant reduction was observed only in countries in Asia (Figure A6 in the Appendix). There was an overall statistically significant increase in admissions due to firearms and gunshot wounds (OR:1.34; 95%CI [1.11,1.61]) across thirteen studies [10, 22, 24, 25, 29, 31, 33, 35, 38, 40, 41, 43, 44]. Among these, only one study [38] carried out in a single centre hospital in South Africa reported a statistically significant reduction in firearm and gunshot wounds (Figure A7 in the Appendix). There was an increasing trend for stab wound admissions in nine studies, but the value was not statistically significant (OR:1.12; CI [0.98,1.28]) [22, 24, 25, 32, 33, 41,42,43,44]. There was no statistically significant difference for falls (see also Figure A8 in the Appendix) and other mechanisms which comprises minor injury mechanisms such as dog bites, found down, and finger trapped. An overall statistically significant increase in admissions due to suicide attempts or self-harm was observed across thirteen studies (OR:1.41; 95%CI [1.05, 1.89]) [8, 10, 11, 20, 22, 24, 28,29,30, 35, 36, 40, 42] (Figure A9 in the Appendix). Among these, only four studies [7, 11, 30, 36] reported a decreasing trend, which was not statistically significant in any study.

Table 3 Injury mechanisms
Full size table

In terms of location, there was a statistically significant reduction in road trauma across twelve studies (OR:0.70; 95%CI [0.56, 0.89]) [10, 11, 20, 23, 28, 30, 36, 39, 41, 42] and in other locations that imply individuals to be outside the home, such as outdoor or workplace, grouped under the label “Other” due to the high heterogeneity in the way in which they were reported (OR:0.73; 95%CI [0.59,0.89]) [10, 20, 23, 26, 30, 36, 40,41,42] (Figure A10 in the Appendix). There was a statistically significant increase in admissions for trauma occurring at home across eleven studies (OR:1.51; 95%CI [1.17, 1.94]) [10, 20, 23, 24, 26, 28, 30, 36, 39, 40, 42] (Figure A11 in the Appendix). Among these, only one study [30] reported a statistically significant reduction during the societal restrictions period. The increase is not only the result of a larger relative proportion of total presentations with respect to the reduction in presentations due to road collisions, but also an increase in the absolute numbers of trauma occurring at home compared to the previous period across studies [20, 28, 36, 39, 40]. By comparing the average home trauma admissions across studies before and after the pandemic, we found an overall average increase in the pandemic period of approximately 7%.

Demographics

We found no statistically significant differences for the average age and the share of females on the total number of admitted patients (p > 0.2) (see Table 4). Five studies [11, 23, 36, 37, 44] reported a significant increase in the proportion of females on the total, while three studies [25, 29, 34] reported a statistically significant reduction. Only one study [29] reported a clear and significant decline in the average age of patients admitted during the COVID-19 period compared to the pre-pandemic period.

Table 4 Demographic information of patients admitted due to major trauma
Full size table

Risk of bias assessment

Table A8 (supplementary appendix) reports the risk of bias assessment. Overall, 26 out 35 (74%) of the included studies were considered to be at low risk of bias [8, 10, 11, 20, 22, 24,25,26,27,28,29,30,31,32,33, 35,36,37,38,39, 41,42,43,44,45,46]. Two studies [12, 47] were classified as being at a high risk of bias, primarily due to a lack of comparability with other studies and definition of the outcomes of interest, and were therefore excluded from the meta-analysis. These studies were included in the descriptive figures and in the demographic table. The remaining seven studies reported a low/moderate risk of bias [19, 21, 23, 34, 40, 48, 49]. With further informal screening to assess the comparability of outcomes and potential biases in the population included, these were included in the meta-analysis. Less than half of the studies used at least two years of prior data to ascertain a control period [8, 10, 11, 20, 25,26,27, 29, 32, 36, 38, 39, 44, 47, 48]. The average length for the control period in the included studies was 2.2 years (± 1.8), with 45% of the studies included using a historical control based on data from more than 1 year, and 31% more than 2 years.

Discussion

From our analysis, two of our three a priori hypotheses were supported: overall, we observed a significant reduction in major trauma presentations and a reduction in traffic-related injuries alongside an increase in trauma occurring at home during the COVID-19 restrictions compared to the pre-pandemic period. However, we did not observe an increase in severely injured patients with the introduction of social restrictions. Further, we did not observe a significant change in mortality of the admitted patients. Findings from the subgroup analysis signal that most of the significant variations were recorded in the continents broadly most affected by the virus, which may signal an incremental marginal effect of the virus on major trauma epidemiology.

As expected, policy restrictions resulted in reduced road collisions and those in other locations such as outdoor places and workplaces [20, 36, 44] and an overall statistically significant increase in trauma occurring in the homes.

Social and movement restrictions have previously been shown to disproportionally affect low-income families and workers in terms of reduced income and increased social isolation [50, 51]. Others have suggested that the loss of financial stability and forced social isolation policies might lead to an increase in intentional injury and violence [24, 25, 52]. Our results show an overall statistically significant reduction in assault trauma and mixed results in terms of stab wounds [22, 24, 25, 32, 33, 35, 41,42,43,44]. However, we did find a statistically significant increase in trauma presentations resulting from firearms and gunshot wounds, in line with previous findings from the literature [25, 52]. This divergence with assault and stab trauma can be explained with additional analysis. First of all, it is a relative increase, meaning that the total number of events has remained stable compared to the pre-pandemic period while other trauma typologies dropped in all countries. Secondly, the result is mainly driven by data from the US (46% of all the studies), where firearm legislation is relatively liberal, and from South Africa (30% of the total), where interpersonal violence and firearms traumas are a significant issue [35, 38, 40]. In the other countries, where firearm injuries pre-COVID-19 were rare, we did not find any significant variation between the two periods. We were unable to systematically examine traumas due to domestic violence as only three studies [24, 28, 40] reported this measure, all of which reported a similar proportion compared to the pre-pandemic period.

A number of authors have previously raised concerns relating to the potential impact of prolonged lockdowns and social restrictions on mental health outcomes, such as suicide rates and self-harm [53, 54]. Previous findings reported mixed evidence compared to the pre-pandemic period [55,56,57]. We observed a statistically significant increase in trauma presentations due to suicide attempts and self-harm during social movement restrictions. We observed both a relative (53% of the total cases in the COVID-19 period, compared to 33% in the pre-pandemic period) and an absolute increase (average increase of 2 per study) across thirteen studies [8, 10, 20, 22, 24, 28,29,30, 35, 36, 40, 42]. The four studies that reported a decreasing (not statistically significant) trend were conducted in New Zealand and Australia, where the impact of COVID-19 has been minimal, as well as the corresponding stringency of containment measures for much of 2020 [4]. Additional research is required to infer whether the effects of the pandemic itself, or the containment measures was more relevant for the trend observed.

Several limitations need to be acknowledged in our analysis. Firstly, most of the included studies provided data on a relatively short time interval both pre- and post-comparison, which may not be adequate observation time to ascertain the true effects of a pandemic on trauma presentations. Extensive evidence reports that trauma epidemiology changes over a short period of time must be interpreted in the setting of normal seasonal variations [58, 59]. Secondly, this meta-analysis also demonstrated a high degree of statistical heterogeneity in the primary outcomes (I2 statistic ranging from 39 to 91%) which most likely originates from the clinical heterogeneity within trauma cohorts, the different time periods considered, the local intensity of SARS-CoV-2 infection, and the extent of social movement restrictions within the included studies. We attempted to explore this heterogeneity through a subgroup analysis at the continent level.

Additionally, most of the included studies did not report whether a hospital was a designated COVID-19 facility. Only two studies [28, 31] reported a change in the trauma destination protocols in response to public health priorities. Self-presentations from injury may decrease in a COVID-designated facility. This lack of information may bias our conclusions, as 72% of the studies included in the meta-analysis are single-centre studies and are therefore more susceptible to such changes. Finally, we only included articles published in English, and therefore, some relevant references published in other languages will have been excluded. Future research should fill this gap and confirm our findings, possibly comparing major trauma admissions across different waves of COVID-19.

This review is the first article that systematically analyses the impact of policy restrictions for pandemic control, and the pandemic itself, on major trauma epidemiology. The results of this systematic review provide relevant information for policy-makers about the implications of disassembling or reducing trauma services, redeploying staff to other tasks, and hospital financing needs, to inform responses to future public health crises when a novel virus is emerging, and vaccines are not yet available. Although new information continues to develop, this review reports evidence of an absolute volume reduction in major trauma admissions with unchanged severity and mortality during the first wave of COVID-19 movement restriction policies. Current data based on the first wave of the COVID-19 do not support the reallocation of highly specialised trauma professionals. If trauma professionals were redeployed to pandemic management at a higher rate than the corresponding reduction in major trauma admissions, there may be a greater burden for trauma professionals with a potential increase in the risk of morbidity and mortality for patients. The described epidemiological changes are essential to inform resource allocation decisions in future waves of viral pandemics and to identify the trauma mechanisms for which hospital and community investments and prevention programs are most needed during public health emergencies.