In this study carried out during the COVID-19 lockdown period in France, 5% of nosocomial cases were reported among all patients hospitalised for COVID-19 in our institution. Considering a range of incubation time up to 14 days, acquisition was questionable for 4 nosocomial cases with an onset of symptoms between 5 and 14 days. However, only 1 case occurred before 7 days of hospitalisation, therefore the number of nosocomial cases is probably not be excessively overestimated. Most of the nosocomial cases had risk factors for severe COVID-19 and 4 died from COVID-19. Survey of hygiene measures highlighted good compliance with mask wearing a but a lack of correct hand hygiene. During the first wave, the positive test rate in the Grenoble department (Isère) was lower than the mean of France, the the highest peak at the hospital being reached at week 13 with a positive rate of 18.17%.
Several studies have analysed the relative proportions of nosocomial COVID-19 cases. A meta-analysis conducted in Wuhan on nosocomial infection of COVID-19, SARS and MERS, showed a proportion of nosocomial COVID-19 at 44% . Comparing our nosocomial rate to the literature is complex, due to the different definitions used in various studies. One study used the same definition of nosocomial cases in Japan and reported rate of nosocomial cases at 18.5% . Another hospital in Spain, using an interval of 6 days, presented a nosocomial rate of 2.5% . Percentages of nosocomial cases reported in 2 urology departments in Spain , a digestive surgery department in Paris , an orthopeadic surgery department in Spain  and a university Hospital in London  were 2.1%, 4.9%, 6.5% and 11.3% respectively. A study conducted in a large US Academic medical center found 2 nosocomial cases among 697 COVID-19 positive patients . Finally, a study in three acute hospitals in Scotland found 19 (11%) nosocomial cases among 173 COVID-19 positive patients . In comparison with influenza nosocomial cases, a meta-analysis  showed that the proportion of nosocomial cases among the total number of patients with influenza ranged from 15 to 59%. We can suggest two hypotheses to explain these differences and the low rate of COVID-19 nosocomial cases. First, the COVID-19 incidence rate in Grenoble during the first wave was rather low. However, other factors may also explain these results, including hygiene measures. During the COVID-19 outbreak, HCWs and patients systematically wore masks, which was not the case in the influenza epidemic. According to these results, universal wearing of a surgical mask by HCWs could be a way to control nosocomial transmission of respiratory viruses . Another hypothesis concerns the lockdown and the ban on visits, which was never implemented during an influenza epidemic. Some studies have evaluated the effect of non-pharmaceutical interventions including the lockdown and proved their efficiency [22, 23]. In our study, all nosocomial cases were grouped within the first 3 weeks of lockdown. As fourteen days is the longest incubation time reported in the literature , we can suppose that the lockdown was efficient and helped to control nosocomial cases. It is important to note that given the high proportion of double occupancy rooms in our hospital for patients and a HCW staff shortage, quarantine of all exposed asymptomatic patients and HCWs was not possible, a factor that may have led to an increased number of nosocomial cases.
As regards hand hygiene we observed moderate compliance (60.6%) and, above all, poor technique (13.7%) and insufficient friction time (15.5%). A lack of time or a lack of knowledge could explain these results. Whatever the reason, education on the importance of hand hygiene is essential to promote these gestures. Moreover, the disinfection of shared equipment could be improved (78.6%). The survivability of Coronaviruses on inanimate surfaces has been shown to range from hours to days . Even though the proportion of transmission from contaminated surfaces remains unknown, hand hygiene and disinfection of shared equipment are essential to avoidance of cross-transmission . The Screen deployment in double occupancy rooms is another hygiene measure needing to be improved in our hospital. Indeed, in the first cluster, all the nosocomial cases were acquired in double occupancy rooms. Transmission from one patient to another sharing the same room, especially in SARS-CoV-2 with droplet transmission, is hard to control. This constitutes a huge challenge for our institution insofar as 65% of the rooms are double occupancy. Hygiene audits have shown a patient compliance rate of 81.5% for wearing a surgical mask when they moved or during care. In France, citizens are not used to wearing masks when they present respiratory symptoms or when they are in closed and crowded spaces; use of masks is generally limited to healthcare settings. In the second cluster, patient behavior may have influenced transmission. In the Mood Disorder unit, many manual activities shared between patients and HCWs such as card games or manual artistic activities are part of the treatment. Moreover, patients admitted in the 2 units with clusters are mainly psychiatric patients or patients with social difficulties. These units are at risk for cross-transmission because of major patient turnover, patients with difficulties complying with instructions, shared spaces and wandering patients. In our hospital, though there was no hand hygiene program targeting patients, but information on barrier measures was provided on screens in halls and patient rooms. Concerning the cluster where 7 HCWs were tested, 4 were infected. All patients and HCW were tested at the same time, so we cannot draw conclusions on the chronology of transmission . Furthermore, physical proximity between HCWs during breaks could also be a way of transmission. A phylogenetic study based on sequencing could clarify these epidemiological links and confirm or not the existence of a single strain . Of note, no cluster was identified in surgery units, probably due to the drastic reduction in their activity following the ban on elective surgery [29, 30].
The strengths of our study are based on prospective data collection. The same data have been collected from electronic medical records in our institution to follow nosocomial influenza cases, and we know this is a reliable and effective system . Moreover, we were in touch with occupational health teams, infectious disease specialists and biologists in case of diagnosis uncertainty. All of these factors should help to reinforce knowledge of this emerging disease, especially concerning nosocomial cases, which are poorly described in the literature.
However this study has some limitations. First, our definition of nosocomial cases was based on 5-day incubation, whereas the incubation time described by the WHO is 1 to 14 days , meaning that we may have defined patients as nosocomial cases whereas they were not, or conversely. Given this wide range of incubation time, there is no optimum definition. We have chosen median incubation as a means of achieving early detection of potential nosocomial clusters in view of preventing further transmission, the objective being to avoid classifying a nosocomial case as community-acquired. Second, asymptomatic cases were not tested and systematic screening 14 days after discharge was not performed, certainly causing missed SARS-CoV-2 infections. Third, our institution was relatively spared and did not have as many cases of COVID-19 as other hospitals elsewhere. Finally, hygiene audits were performed after the lockdown period and did not adequately reflect the reality of lockdown period.