This study sought to identify factors associated with SARS-CoV-2 infection among HCW to improve interventions aimed at reducing the risk in these front-line workers. After considering a wide range of characteristics, we found sociodemographic, lifestyle, psychological factors, and PPE use in and outside the workplace were the most relevant to have a positive RT-PCR. In particular, the results highlight the effectiveness of PPE in this population, demonstrating a greater protective effect of high-efficiency masks when compared to other types of masks when use in and outside the workplace. However, we observed unexpected results regarding the impact of PPE usage training as well as the effectiveness of some PPE elements such as surgical hats, face shields/googles and gloves.
In this study, all HCW wore some type of facemask at work. Based on our results, those always-wearing high-performance filtering masks had a better protection when compared to those wearing them occasionally or wearing other types of facemasks. This is in line with a previous meta-analysis suggesting that N95 mask could be more strongly associated with protection from viral transmission than surgical masks (20). However, low certainty evidence suggests that surgical masks and N95 respirators offer similar protection against coronavirus in non-aerosol-generating care (21). One study even reported an increased risk of N95 compared to surgical mask, but that study only adjusted for age as a potential confounder (22). Different types of masks, manufacturer standards, and the evaluation of potential confounders may explain discordances between studies.
There is not clear recommendation for the type of mask that HCW need to wear in “low risk” areas in the hospitals or even outside the workplace. For example, the evidence comparing the efficacy of N95 respirators to surgical masks in the outpatient setting is scarce (19, 23). Our results suggest that fabric and surgical masks performed similarly, while wearing high-performance filtering masks or a combination of fabric plus surgical mask reduces the risk of infection compared to the use of surgical mask exclusively (24). A recent experiment demonstrated that the combination of the fabric mask covering the surgical mask blocked 92.5% of the cough particles, which double the filtration capacity of each mask alone (25). Therefore, HCW could be advised to wear high-performance mask even when they are not directly taking care of COVID-19 patients, or in case of a shortage, low resource settings or high cost of high-performing masks, a combination of fabric plus surgical mask as an alternative. In any case, emphasis needs to be given to the proper use of PPE as previously stated (15, 26, 27). PPE fit is an important component in their functional efficacy. Surgical masks and fabric masks do not require special fitting. Nonetheless, it has been shown that even non-fit-tested N95 respirators were significantly more protective than surgical masks (28).
It is expected that the use of PPE including face masks, hats, gowns, shoe covers, gloves, or face shields are associated with decreased risk, but there are not necessarily so when included in multiple regression models (29). In general, the evidence for the effectiveness of face shields in preventing transmission of viral respiratory diseases is limited (23). Controversially, our study reported a greater risk among those who always wore face shields/goggles, gloves and surgical hats. We believe that certain behaviors for example, sharing reusable PPE e.g; face shields, without appropriate cleaning and disinfection protocols, or relaxing their use while taking rest, might contribute to the transmission. PPE could also provide a false sense of security resulting in self-contamination during patient´s care and doffing process (30). It has been reported that half of HCW correctly remove their PPE, and very few dispose them in the proper location (31, 32). Common reported errors are doffing gown from the front, removing face shield, and touching potentially contaminated surfaces during doffing (33). Glove and gown removal simulations showed that self-contamination of skin or clothing occurred in 46% of simulations (34). Then, respiratory hygiene programs including adequate training activities and on-site PPE monitoring strategies, are required to decrease the risk of inappropriate use of PPE. In this regard, an unexpected finding of our study was the almost significant association between being trained and being positive. More than 50% or HCW had received PPE training for less than 2 hours, which could be insufficient and might explain, at least in part, this result.
We also showed a differential effect of gender and hypothesized about the potential role of hormonal contraceptives intake in infection. Our results support a greater risk of having a positive RT-PCR among men. Testosterone suppresses the innate immune responses (35), and seems to increase renin-angiotensin protein system and angiotensin converting enzyme (ACE) activity, while estrogen decreases them (36). Additionally, ACE2 is also present in testicles (37). These findings may suggest that a differential expression of ACE2 between males and females and could explain, at least in part, the gender differences in COVID-19 susceptibility. On the other hand, a previous meta-analysis showed that women are about 50% more likely than men to adopt and practice non-pharmaceutical interventions (38), suggesting a better compliance with biosafety measures among women. The greater risk of SARS-CoV-2 among men was also reported in HCW in India (26); however, some studies in the general population have reported a greater risk among women (17, 39). Differences in the amount and type of contact patterns could explain these discrepancies, as could be the case of women having predominant roles as caregivers in the general population.
Notably, we observed a differential but no significant risk among women according to the use of hormonal contraceptives. Like testosterone, progesterone generally inhibits inflammatory innate immune responses by promoting an anti-inflammatory state and suppressing the activation of macrophages and dendritic cells (40). This anti-inflammatory environment contributes to a greater bacterial burden of Mycobacterium tuberculosis (41) (40) and might play a similar role in SARS-CoV-2. However, estrogens at low dose have an immune stimulatory activity, which is the case of oral contraceptives (42). Therefore, combine birth control pills, might contribute to symptomatic infections (43). This could explain the higher prevalence of cases particularly among those symptomatic women who had used hormonal contraceptives. However, we did not evaluate whether the contraceptive was progestogen or combined based. Therefore, this specific result should be considered as exploratory and requires further evaluation.
Psychological factors such as feeling scared or nervous showed a protective effect. Notwithstanding stress is a negative consequence of the pandemic, our study supports that certain amount of stress might confer protection. Despite we did not evaluated the source of stress, anxious individuals are less confident in their abilities to managing threated situations (44). Therefore, they are more sensitive to feedback and to be hyper vigilant in monitoring their surroundings and themselves which leads to strategic actions to avoid harm (45). However, it is not clear whether this apparent protective effect observed before the first peak of cases could persist through the duration of the pandemic. On the other hand, the greater risk among less-educated adults compared to university graduated is consistent with a previous report (46). We also observed a trend to a lower risk among those living with more than two persons in the household. One reason could be that HCW adhere more to biosafety measures if they feel responsible for their family. On the contrary, comorbidities previously associated with severity (47–49) were not associated with the infection in our study (50).
Despite nursing assistants represented the largest proportion of HCW among those who tested positive (14), our study reports a greater risk among nurses when compared to them; however, the precision of this estimation was low. Nurses directly care positive patients and practice procedures including bronchial aspiration. Other healthcare professionals such as respiratory therapists and physicians are also at high risk of exposure. Administrative staff are not expose to such procedures; however, they could be less supported concerning the use of PPE and biosafety measures. Other occupational characteristics previously reported as risks factors were not independently associated, such as sharing spaces for food consumption (51), night shift (52) or daily work hours (18).
Strategies to minimize potential biases included blindness of interviewers to prevent the observer bias, and confirming the participant status to avoid misclassification of cases and controls were implemented. However, we acknowledge that recall bias could have been present, though we anticipate its impact to be non-differential given that the time between the RT-PCR results and the interview were similar between groups. The proportion of proper PPE use may have been overestimated in both cases and controls, since this variable was self-reported. The quality of training, whether donning and doffing procedures were both included, or whether training prioritize certain PPE, was not evaluated and warrants further studies. This study was performed close to the first peak of cases in Colombia when personal and institutional biosafety measures were reinforced to prevent contagion. This might explain the homogeneity of cases and controls in certain COVID-19 related exposures and PPE use. In this context, the sample size could be insufficient to detect small differences between the groups. Residual confounding could also be present for certain variables such as the type of alcoholic beverage or number of mask layers. Other variables as the reuse or sharing PPE, doffing practices, or the prevalence of the infection in the place of residence were not evaluated. The employed workforce tends to have fewer sick people and may behave differently regarding the PPE use, which limit the extrapolation of our result to the general population.
In conclusion, high-performance filtering masks used in and outside workplace were protective for SARS-CoV-2 infection among HCW independent of the level of exposure. Gender, level of education together with occupational and personal characteristics, influence the risk of infection and need to be considered when planning public health and hospital infection prevention strategies. As the pandemic progresses, the quality of training and monitoring strategies of PPE use becomes more relevant for HCW both in and outside the workplace. Further research is warranted 1) to identify the types of errors that might occur in using PPE, 2) to evaluate and improve adherence to recommended protocols for PPE donning and doffing, 3) to identify effective means, contents and quality standards of training, and 4) to identify the potential risk of hormonal contraceptives in symptomatic SARS-CoV-2 infection.