SARS-CoV-2 Transmission in the Military during the Early Phase of the Pandemic—A Systematic Analysis

Militaries worldwide have been affected by COVID-19 pandemic. However, the impact and epidemiological characteristics of transmission during the early phase of the pandemic is not well-studied. This study aims to systematically estimate the baseline incidence of COVID-19 in the military worldwide and identify the potential risk factors of transmission and clinical characteristics of the cases. English and Chinese literature reporting COVID-19 cases in military worldwide published on four electronic databases (PubMed, Scopus, EMBASE, and CKNI) through 28 May 2021 were systematically screened and synthesized qualitatively. Forty-six studies involving at least 711,408 military personnel in 17 countries were synthesized. Low incidence of cases was observed in the military with pooled COVID-19 incidence of 0.19% (95%CI: 0.00–9.18%). We observed a higher incidence among those (1) with overseas exposure (39.85%; 95%CI: 0.00–95.87%) rather than local exposure (3.03%; 95%CI: 0.00–12.53%), (2) who were on either local/overseas military deployment (26.78%; 95%CI: 0.00–71.51%) as compared to those not deployed (4.37%; 95%CI: 0.00–17.93%), and (3) on overseas military deployment (39.84%; 95%CI: 0.00–95.87%) as compared to local military deployment (3.03%; 95%CI: 2.37–3.74%). The majority of the cases were symptomatic (77.90% (95%CI: 43.91–100.00%)); hospitalization and mortality rates were low at 4.43% (95%CI: 0.00–25.34%) and 0.25% (95%CI: 0.00–0.85%), respectively; and headache, anosmia, ageusia, myalgia, nasal congestion, and cough were the most commonly observed symptoms. Overseas and local deployment were observed to have higher risk of SARS-CoV-2 transmission. Sustainable, active SARS-CoV-2 surveillance strategies are crucial to detect and contain transmission early during military deployments.


Introduction
Reports on an outbreak of pneumonia caused by an unknown etiological agent first broke out in Wuhan, China, during late December 2019. While the World Health Organization declared the outbreak a Public Health Emergency of International Concern on 30 January 2020, the coronavirus disease 2019 (COVID-19), due to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), was only declared a pandemic on 11 March 2020 [1].
The military is characterized by attributes that are advantageous for managing such disease outbreaks: crisis-management capacities, ability to execute missions in sub-optimal environments, logistic resources for deployment, and the ability to mobilize large forces in risky situations within or outside the country [2]. While some militaries have service members with substantial emergency-response background and public health expertise, those without can also utilize its national command network, pool of disciplined manpower, and logistical support to supplement civilian frontline services [3].

Data Analysis
Apart from qualitative synthesis of the data, meta-analysis was explored to gain additional insights. Incidence of the exposure, outcomes and clinical characteristics, and their corresponding 95% confidence intervals were pooled using the inverse variance heterogeneity (IVhet) model [13]. The model is a modification of the fixed-effects model that accounts for between-study heterogeneity while retaining the individual weight of studies [14]. Freeman-Tukey double arcsine transformation was used to avoid giving weight to studies with estimates that are too skewed. Forest plots were generated for graphical representations. I 2 statistic values were calculated to quantify degree of heterogeneity among studies that was not attributable to chance; values of 25-50% suggest heterogeneity, and values of >50% indicate substantial heterogeneity. All meta-analyses were conducted with MetaXL meta-analyses software (version 5.3, EpiGear International, Sunrise Beach, Australia).
Studies including case reports and case series without military population for denominator were excluded from the meta-analyses; those constituting less than five patients were also excluded from all meta-analyses of clinical characteristics [15]. Subgroup analysis of incidence was conducted for (1) deployment or no deployment, (2) reporting of case from local or overseas setting, and (3) deployment to a local or overseas setting.

Published Literature
The initial database searches identified 3307 studies. Following the removal of 547 duplicates, 2760 studies were screened for their titles and abstracts. From 141 studies shortlisted in the primary screen (Figure 1), 46 studies in English language were selected and synthesized in this systematic analysis (Tables 1-3, Supplementary Tables S1 and S2). Only 36 of the included studies were used in various meta-analyses (Tables S3-S24, Supplementary Figures S1 and S2), as the remaining ten had a potential overlap in cases due to the period of reporting. [16][17][18][19][20][21][22][23][24][25].
After accounting for potential overlap in cases, this study utilized 36 articles from 17 countries-the U.S. (n = 17), Israel (n = 2), Switzerland (n = 2), and the United Kingdom (n = 2) and one each from Belgium, Bolivia, Brazil, Canada, Djibouti, France, India, Italy, Norway, Philippines, South Korea, Sri Lanka, and Tunisia . The Israel-based study by Talmy et al. was solely used for the meta-analysis of specific symptoms since it was the only Israeli study reporting symptoms breakdown [58]. There were at least 24,930 males reported by 19 studies, comprising 50.20% to 93.10% of these study populations ( Table 1). The mean/median age reported by nine studies ranged from 19.1 to 45.1 years old; eight were below 34 years old, while the sole study with a higher mean of 45.1 years old was based in a hospital setting. Five case reports had patients ranging from 21 to 36 years old. Settings where positive cases were detected were available for 28 studies-military hospitals/treatment facilities (n = 8), aircraft carriers (n = 2), recruit schools (n = 2), army bases (n = 4), quarantine facilities (n = 5), deployment at field hospital/hospital ship (n = 2), and arrival testing-on return from deployment (n = 2)/at deployment site (n = 1), deployment at long-term care facilities (n = 1), and air evacuation to medical facilities (n = 1).
After accounting for potential overlap in cases, this study utilized 36 articles from 17 countries-the U.S. (n = 17), Israel (n = 2), Switzerland (n = 2), and the United Kingdom (n

COVID-19 Incidence and Demographic of Cases
A total of 8635 cases were reported from a known military population of 711,408 between January 2020 and early May 2021 ( Table 2). The majority of the population was attributed to Oh et al., who reported a denominator of 599,000 military personnel constituting both active servicemen and civilian employees in South Korea. In those confirmed to have received COVID-19 testing, positivity by RT-PCR and serology were 5817/51,083 (11.39%) and 81/3538 (2.29%), respectively. Most studies had reported low incidence of cases, and high incidence (>0.50%) was observed in only three studies-Joshi et al., Paleiron et al., and Sasongko et al. We observed a pooled COVID-19 incidence of 0.19% (n = 22; 95% CI: 0.00-9.18%), with high heterogeneity among the studies (I 2 : 99.93%) (Figure 2a). There were 3400 (76.73%) males out of 4431 cases with known gender (Table S1). Race was reported for 1414 cases by five studies from the U.S. (Table S1). The majority of cases occurred in White Americans (628; 44.41%), followed by Hispanic/Latino Americans (274; 19.38%), African Americans (274; 19.38%), Asian/Pacific Islanders (106; 7.50%), Native Americans (34; 2.40%), and others (98; 6.93%).     there could be more cases, but breakdown was not provided. j Cases regarded as true positive using serology test did not report 11 regarded as false positive by PCR. Figure 2. Meta-analyses of COVID-19 incidence (a) COVID-19 in military populations, (b) COVID-19 incidence by populations with possible exposure locally or overseas, (c) COVID-19 incidence by deployment, (d) COVID-19 incidence by local or overseas deployment, (e) males in cases (using maximum case numbers provided by studies that were not included in original study analysis).

Deployment and Possible Exposures
Location of possible exposure was reported by 15 studies; a higher incidence was observed for possible exposure overseas (n = 3; 39.85%; 95%CI: 0.00-95.87%) as compared to possible local exposure (n = 12; 3.03%; 95%CI: 0.00-12.53%) (Figure 2c). Incidence of cases was higher in those who were deployed (n = 6; 26.78%; 95%CI: 0.00-71.51%) as compared to those not deployed (n = 9; 4.37%; 95%CI: 0.00-17.93%) (Figure 2d). Areas of deployment outside of the cases' home country include Djibouti, Morocco, South Korea, Niger, Europe, countries in the U.S., Central and Europe Command (CENT-COM/EUCOM), and aircraft carriers ( Table 3). The duration of deployment ranged from 1 day to 5 months in seven studies. Pooled incidence of COVID-19 from six available studies was higher among overseas deployment (n = 3; 39.84%; 95%CI: 0.00-95.87%) than local deployment (n = 3; 3.03%; 95%CI: 2.37-3.74%) (Figure 2e). The three overseas deployments recorded were on aircraft carriers USS Theodore Roosevelt, Charles de Gaulle, and Belgium forces to a military training institute in Niger [38,49,51]. The three local deployments were to Javits New York Medical Station (a field hospital in New York City), USNS Comfort (a U.S. Navy hospital ship), and long-term care facilities in Thirteen studies reported rank information for 4332 cases-1418 cases (32.73%) were in the command line (officers and non-commissioned officers), and 2914 cases (67.27%) were in enlisted soldiers (Table S1). Of cases in enlisted soldiers, 1386 occurred in newly conscripted recruits. Five studies reported information on the job nature of 578 cases-twenty-three (3.98%) cases had medical responsibilities as registered nurses, physician/physician assistants, corpsman, and medical support with patient interaction or medics; five (0.87%) were trainers who returned from a deployment to a military education and training center in Niger. Kasper 14.16%), and deck (4/86, 4.65%). The odds of COVID-19 disease were 1.73 to 2.70 times significantly higher in the weapons, supply, and engineering and reactor divisions with reference to air division, which the study attributed to the divi-sions' (reactor, engineering, supply, weapons and combat support) predisposition to more confined working spaces as compared to the air and deck departments.
A plausible transmission origin was recorded for 579 cases in 12 studies (Table 3). Thirty cases (5.18%) had recent recreational activity before testing positive, such as a vacation trip, visited clubs/beaches, or returned from leave, while 538 cases (92.92%) were reportedly contacts. Twenty-four cases, specifically, had infected roommates. Belgium military personnel were found to be infected with a strain with a recent African common ancestor, indicating local transmission during their deployment to Niger [51]. Separately, there were cases that were likely infected by the public during transit or in their residential area [40,45]. Possible healthcare-associated exposure was reported in at least 16 cases (2.76%) by three studies-14 cases were directly involved in the care of COVID-19 patients, five of whom had conducted aerosol generating procedure, and one had reported a break in personal protection equipment [29,42,48].

Discussion
Potential risk factors identified from this analysis include confined working spaces aboard ships, performing aerosol-generating procedures during COVID-19 missions, and having vacations. Payne et al. further identified greater odds of COVID-19 in those who had (1) contact with known cases (OR: 2.5; 95%CI: 1.1-5.8) and (2) shared sleeping berth with cases (OR: 3.3; 95%CI: 1.8-6.1) as compared to those without [20]. These factors coalesced into close physical proximity favoring the spread of respiratory diseases commonly transmissible via airborne or droplet routes, which aligned with evidence showing that communication lasting at least 30 min and sharing a bedroom were associated with higher risk of transmission among household contacts [62]. Medical procedures such as tracheotomy, non-invasive ventilation, and manual ventilation prior to intubation emit respiratory particles [63]. A systematic review found 6.60 times greater odds of SARS-CoV-1 in healthcare personnel who had exposure to aerosol-generating procedures and an absolute increase of 10-15% in risk of SARS-CoV-1 transmission [64]. As for COVID-19, 10.7% of healthcare workers reported suspected or confirmed diagnosis within 32 days of conducting tracheal intubation [65].
Despite heightened susceptibility to disease transmission, we observed low incidence of COVID-19 infection, hospitalization, and mortality of cases in the military population. This was not surprising given that their healthy base profile might have been protective against infection and clinical progression. The military population was young and predominately male (most studies reported a mean age between 20-30 years old), and intakes were heavily screened in terms of medical fitness [47]. Furthermore, the vulnerability of militaries to infectious diseases was compensated by swift, strict, and regimented containment strategies to which the compliant community adhered [22,39,48,54]. As of 3 March 2020, the PLA had reported no confirmed cases due to successful and strengthened prevention measures [66]. The USFK, United States Forces in Italy, South Korean military, and Israel Defense Forces echoed early curtailing of soldiers' lifestyles and movement, development of extensive testing and contact tracing capacities, and stringent quarantine standards since the early stages of the pandemic [22,39,48,54]. Algorithms guided systematic and efficient flow of response, from screening to immediate quarantine of units with confirmed cases and identification of hotspots on post and close contacts. Additionally, greater access to healthcare services could have contributed to low mortality since military personnel typically receive healthcare services within the military installation/framework and are entitled to comprehensive medical coverage in most countries [67].
A higher pooled incidence of COVID-19 was observed for populations (1) with possible exposure overseas, (2) on deployment, and particularly, (3) overseas deployments in the early phase of the pandemic. This corresponded with higher incidence of disease infection during deployment historically. Higher rates of food borne disease outbreaks were recorded in French soldiers deployed overseas (26.7 outbreaks/100,000) as compared to those in France (2.4 outbreaks/100,000) from 1999 to 2009 [11]. Separately, respiratory illness was newly reported in 14% of deployed soldiers to Iraq and Afghanistan as compared to 10% of non-deployed soldiers [68]. Heightened infection risk during deployment could be due to harsh conditions in foreign environments, possible inadequacy in hygiene and sanitation, sleep deprivation, and physical and mental strain [11]. Amid uncertainties surrounding the novel virus during early stages of the outbreak, these stressors could have synergistically altered the physiological and immunological states of the military, increasing their vulnerability to infections [11]. Airborne and droplet transmission of viruses are likely to be exacerbated by congregated living arrangements and impaired hygiene practices [69]. Interaction with the local populace and environment [69] can further obscure infection control in foreign land due to dependence on local policies and community response, exposing soldiers to the risk of acquiring disease circulating in the local community [51]. The disparity in COVID-19 incidence between local and overseas deployment in this study could be due to differences in deployment purpose and time period. Studies on overseas deployment-USS Theodore Roosevelt, Charles de Gaulle, and Niger military institutehad commenced prior to or in the early stage of the pandemic when COVID-19 information was sparse and received little attention. It was likely that prevention may have been neglected in view of a higher perceived risk from rarer diseases [70], such as diseases endemic to the deployed setting. Chassery et al. also noted that Charles de Gaulle had a stopover in Brest for crew changeover and technical purposes [17]. On the other hand, studies on local deployments were to field hospital, hospital ships, and long-term care facilities during the first wave of the pandemic. The preparedness behind deployments to support medical and care facilities was likely more robust in terms of logistic supply of personal protective equipment, protocol, and heightened awareness of precautionary hygiene measures.
Contrarily, service members may also act as vessels and introduce viruses to the deployed region. The transmission of pathogens across geographical boundaries was well-documented by Zemke et al. in 67 studies dating between 1955 and 2018 [71]. Just as the first influenza A(H1N1)pdm09 cases in Kuwait and Iraq were imported by U.S. military personnel who were infected during pre-deployment trainings [71,72], Nepalese peacekeepers infected from their mission in Kathmandu served as the source of a major cholera outbreak in Artibonite River Valley in 2010 and the subsequent endemicity of the pathogen in Haiti [71,73]. Likewise, COVID-19 cases have been detected in U.S. military personnel after arriving in South Korea and Japan [74].
Understanding the common clinical traits of COVID-19 in military personnel can improve detection of infected personnel. While the high incidence of symptomatic patients observed in this study aligned with other studies [75,76], the significant proportion of asymptomatic cases could have resulted from scheduled testing practices and social distancing [27,44]. Fever (78%), cough (57%), fatigue (31%), and hyposmia (25%) were the most prevalent symptoms in patients with a mean age of 49 years old from nine countries [77], while fever (46%), cough (37%), and diarrhea (19%) were the most commonly reported in pediatric patients [78]. In this study, headache was the most prevalent symptom, followed by anosmia, ageusia, myalgia, nasal congestion, and cough. It is possible that active testing in the military enabled identification of non-distinctive symptoms such as headache, which are more likely to be overlooked as compared to fever and other respiratory symptoms. The association between headache and anosmia/ageusia and their likely occurrence in the early stage of the disease has been well-established beyond this study [79,80]. Alternatively, differences in disease presentation could plausibly be attributed to infection with different strains that surfaced over the pandemic course. In the United Kingdom, the proportion of asymptomatic cases did not change significantly with the increasing incidence of B.1.1.7 variant [81]. Elsewhere, associations between B.1.1.7 and hospitalization and death were mixed [82]. The delta variant, on the other hand, showed 2.61 times and more than 2 times higher risk of hospitalization compared to alpha variant in England and Scotland, respectively [83].

Preventive Measures
With resonance in the scientific community on the possible endemicity of SARS-CoV-2, vaccination and regular testing as cardinal components of military strategies can facilitate the near future management of COVID-19. The effectiveness of vaccination in military was evident in Singapore during the 2009 H1N1 vaccination program, where the incidence rate was reduced by 54% compared to the unvaccinated prediction [84]. While Pfizer-BioNTech's vaccine was approved in December 2020, the impact of vaccination against COVID-19 in the military was beyond the focus of this study. Nevertheless, there is sound evidence backing the efficacy of vaccination, inducing protection against infection and reduction in the severity of COVID-19 [85,86]. Vaccination can alleviate the impact of COVID-19 infections resultant of high exposure, preserving military strength and operational readiness. This ought to be complemented with regular testing to identify asymptomatic cases and individuals in their early phase of illness, allowing early intervention to break the transmission chains. In April 2021, the U.S. military reported that one percent of service members in priority groups will be tested fortnightly as testing capacity ramps up [87]. The Singapore Armed Forces has also made adaptations to operate in a COVID-19 endemic environment with regular testing of frontline and service personnel and entry tests for all military events [88,89]. Rapid antigen tests, oropharyngeal/middle turbinate, and deep-throat saliva testing were trialed for adoption in the surveillance program [88,89].

Limitations
While the meta-analyses served as a statistical indicator of our findings, the results were inconclusive due to significant heterogeneity of studies and limited reporting of data by the militaries. The high heterogeneity across all meta-analyses was likely because studies encapsulated different time points of the pandemic geographically and involved different fractions of the military population in varied settings. As Oh et al. reported COVID-19 cases including civilian employees as the full military strength of South Korea [48], we could not discern if this inclusion was consistently practiced worldwide. Similarly, Escalera-Antezana et al. reported infections in terms of all military personnel registered in Bolivia at an unspecified time point [33]. There was a lack of uniformity in included studies, as we attempted to capture the maximum amount of literature related to COVID-19 infections in the military to provide a baseline incidence. Most studies (n = 27) had figures cut-off before June 2020, some by December 2020 (n = 13), and only a single study was from 2021. The incidence may not be generalizable to the current situation, which has worsened with recurrent waves caused by variants of concern. We also note the saturation of literature published by the developed nations, predominately the United States. The varied degree of transparency exercised by the military worldwide may have caused reporting bias and skewed representation of COVID-19 incidence. The true impact of COVID-19 on military population was further obscured by publication bias, as authors could not discern if the lack of relevant publication identified from the Chinese database (CKNI) and English databases was due to true absence of cases attributed to effective prevention control or publication bias.
Another limitation is the soldier's autonomy and different testing practices. The accuracy of incidence concluded by individual studies in the early phase of the pandemic could have been diminished as soldiers could refuse sample taking or data usage [28]. While testing rigor was low and typically reserved for symptomatic individuals with defined exposure [29], testing frequency varied across setting depending on risk sentiment, needs, and accessibility to test kits [90,91]. In the U.S. Air Force basic military trainees, the testing prerequisite changed from symptomatic individuals with possible exposure to being symptomatic with mere presentation of symptoms [45] and eventually to universal testing in subsequent cohorts enrolled between May and August 2020 [46]. Given the likelihood of exposure from deployments, it is conceivable that incidence of COVID-19 was heavily underestimated in the early phase of the pandemic.
Lastly, studies were not assessed for quality and publication bias. Since the study types were not restricted, the authors did not deem fit to conduct a quality assessment on literatures that were not intended to be nor held to the rigor of a full-fledged investigational study. The paucity and high heterogeneity of studies investigating COVID-19 outbreaks within the military highlights the need for more attention on this essential but highly susceptible population.

Conclusions
Despite the low pooled incidence of hospitalization and mortality rates, which were likely attributed to the young, healthy demographic of the military, there was high pooled incidence of symptomatic cases. Active SARS-CoV-2 surveillance strategies is critical for early detection and containment to reduce risk of transmission during military deployments at the early phase of the pandemic.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/ijerph19127418/s1. References  are cited in the Supplementary Materials.