Discussion

In the province of Quebec, during the first three COVID-19 pandemic waves, occupational exposure was perceived by SARS-CoV-2-infected HCWs as the most frequent source of infection. Its relative importance decreased over time dropping by half from 91% to 40%, while the contribution of household-acquired infections increased eightfold from 4% to 33%.

Since the beginning of the pandemic, reports worldwide have described that HCWs were at greater risk of infection than the general population, attributing this to their occupational exposure when caring for patients. ^{Reference Gómez-Ochoa, Franco and Rojas1} Similar to our results, the proportion of HCWs among all diagnosed cases in Canada and in the USA was highest in the first wave (19% and 16%, respectively). It was lower but stable between July 2020 and January 2021 and decreased to 7% in Canada and 5% in the USA from January to May 2021, indirectly suggesting that their risk was approaching that of the general population. ^13,14 During the first pandemic wave, mitigation measures limited the contacts and exposures to infection of the general population. However, PCR testing was limited and more readily available to HCWs, likely overestimating the relative proportion of HCWs among all cases. After the first wave, PCR testing became widely available to the population reducing the importance of this bias. The weekly number of reported HCW infections was, however, still high during the peak of the second wave, but decreased sharply from February to May 2021 despite the peak in community cases during the third wave.

Parallel to the decrease in SARS-CoV-2 infections, our study shows that the most common perceived source of infection shifted from the workplace to the community/household from February 2021 onward. This coincided temporally with the expected effect of vaccination against COVID-19 in HCWs, ^{Reference Daniel, Nivet, Warner and Podolsky15–Reference Rhee, Wang and Ye17} but also with the expected effect of the vaccination among LTCF residents and other vulnerable persons at risk of hospitalization, who were also prioritized at the beginning of Quebec immunization campaign. ^{11,Reference Fortin, De Wals and Talbot18} Other factors, such as the improvement of IPC measures, may have also contributed to this relative reduction of SARS-CoV-2 occupational transmission. ^{Reference Vitrat, Maillard and Raybaud4,Reference De Serres, Carazo and Villeneuve19} The end of the lockdown with the loosening of strict public health measures and reopening of schools was associated with a higher community and household transmission of SARS-CoV-2 over the waves. ²⁰

Identification of the source of infection is challenging. Although household exposures are more obvious, occupational exposures leading to an infection may be more difficult to demonstrate or confirm. Multiple encounters may happen in the workplace during the incubation period with known or unknown contagious patients and/or coworkers during which different levels of IPC measures may have been applied. Similarly, exposures in the community are often difficult to identify. A study in two US hospitals using structured interviews to determine the infection source reported that only 6% of infections among HCWs were hospital-acquired, but classified as “unknown source” for those whose only known exposure was a contact with a COVID-19 patient while wearing PPE. ^{Reference Rhee, Baker and Tucker21} Reports about the importance of the workplace exposure in hospitals are contradictory ^{Reference Wilkins, Gray and Wallia3,Reference Jacob, Baker and Fridkin5,Reference Baker, Nelson and Overton6,Reference Belan, Charmet and Schaeffer22} and no studies compared the evolution before/after the COVID-19 vaccine deployment. Although employer-based data may underestimate occupational exposure (e.g., undetected PPE lapses), self-reporting of workplace-acquired infections may overestimate this exposure if attributed by HCWs who did not identify any alternative source of infection or if associated with financial advantages. This latter bias should have had a minimal impact in our study, since the survey was confidential and participants did not get any benefit by reporting occupational exposure.

In our study, proportions of self-reported source of infection varied with time but also by occupation and facility. HCWs with closer and prolonged contacts with patients (auxiliary nurses and patient support assistants) perceived more often the workplace as an infection source, consistent with employer-generated data and other studies reporting a higher frequency of SARS-CoV-2 infections among these categories. ^{Reference Wilkins, Gray and Wallia3,Reference Jespersen, Mikkelsen and Greve23–Reference Modenese, Casolari and Rossi27} During the first pandemic wave, similar high proportions (∼90%) of staff from hospitals and LTCF reported workplace-acquired COVID-19. The relative decrease in the subsequent waves was more pronounced in hospitals than in LTCF and more pronounced for patients than for coworker exposure. These observations are consistent with the reduction in the number of COVID-19 hospitalizations and outbreaks in healthcare facilities following the vaccination campaign. ^{Reference Fortin, De Wals and Talbot18,20} Few studies have compared the risk of transmission in ACH versus LTCF. Akinbami et al reported higher seroprevalence at the end of the first wave among staff in nursing homes (13%) than in hospitals (5%), ^{Reference Akinbami, Chan and Vuong7} while our previous study in Quebec showed that working in LTCF or private seniors’ homes increased the risk of SARS-CoV-2 infection compared to ACH. ^{Reference Carazo, Villeneuve and Laliberté10} Conversely, the relatively higher proportion of HCWs with household-acquired infections among ACH workers might be a reflection of this lower workplace risk.

HCWs participating in the study and not living alone infected 30% of their household members, increasing from 29% during the first months of the pandemic to 42% during the period of Alpha-variant predominance. This is higher than the overall household SAR of 19% (95% CI, 16%–22%) reported by a meta-analysis summarizing 87 studies published through June 2021. ^{Reference Madewell, Yang, Longini, Halloran and Dean28} Similar to our results, they described an increasing pattern overtime; the household SAR was 20% in studies from March to June 2020, compared with 31% in studies from July 2020 to March 2021. This may be attributable to better ascertainment of secondary cases over time as PCR testing became widely available, ^{Reference Schwartz, Achonu and Buchan29,Reference Paul, Daneman and Brown30} as well as changes in variant circulation. We identified secondary cases through both epidemiological link (during first wave) and laboratory confirmation (during the second and third waves). Consistent with our data, an updated meta-analysis stratified by variant circulation reported higher household SAR for Alpha variant (38%) than for the original wild-type strain (19%), ^{Reference Madewell, Yang, Longini, Halloran and Dean31} probably partially attributable to the higher transmissibility of this variant. ^{Reference Davies, Abbott and Barnard32,Reference Lindstrøm, Engebretsen and Kristoffersen33}

The self-reported the presence of symptoms in the index HCW case was associated with a twofold increase in the household SAR (31% for symptomatic versus 16% for asymptomatic index cases), in line with studies showing that symptomatic COVID-19 cases are more contagious than asymptomatic cases. ^{Reference Madewell, Yang, Longini, Halloran and Dean28,Reference Byambasuren, Cardona and Bell34–Reference Tibebu, Brown, Daneman, Paul and Buchan38}

We found a 25% risk reduction of SAR when the index case was vaccinated with at least one dose versus unvaccinated after adjustment for the period of variant circulation. As vaccination coverage increased in parallel with the Alpha-variant circulation, punctual SAR estimates reflected both factors and are difficult to disentangle. An analysis restricted to the Alpha-dominated period would have been useful but was not possible due to the limited sample size. One meta-analysis that included 12 studies, from January 2021 to January 2022, reported a nonsignificantly lower risk of household transmission when the index case was vaccinated (23% for fully vaccinated versus 36% for unvaccinated), while the summary SAR including 4 studies during Alpha-variant circulation was significantly lower with fully vaccinated (11%) versus unvaccinated (36%) index cases. ^{Reference Madewell, Yang, Longini, Halloran and Dean31}

Our study has several limitations. Like most studies about occupational COVID-19 acquisition, ours is a series of three cross-sectional surveys based on self-reported perceived source of infection. During the third pandemic wave, fewer HCWs were infected by COVID-19 and participated in our study, limiting our power to estimate the impact of vaccination. Finally, our results are not representative of the situation after May 2021, with the circulation of new emerging and more transmissible variants, a very high proportion of adults vaccinated with two or more doses, and a large proportion of adults already infected at least once. ^{Reference Skowronski, Kaweski and Irvine39}

In conclusion, HCWs and their household members were largely affected during the first pandemic waves of COVID-19, but the relative and the absolute occupational risk lowered contemporaneously with the start of vaccination and the improvement of ICP measures. These findings underscore the importance of pandemic preparedness in healthcare settings since ICP measures may be the only feasible intervention to protect HCWs from emerging biological hazard exposures until targeted vaccination becomes available.