Introduction

Because of their frailty and comorbidities, older adults living in nursing homes (NH) paid a heavy price to the COVID-19 pandemic [1, 2] despite early and often drastic prevention measures.

The evolution of the epidemic between 2020 and 2022 in NH was marked by several phases: first, the occurrence of numerous and rapidly evolving epidemics, with high incidence and mortality rates. The modes of transmission of this new virus were still poorly understood, particularly the frequency of transmission in the pre-symptomatic period, and the usual measures for controlling epidemics of acute respiratory infections were not adapted [3, 4]. As soon as it became available, the vaccine was then offered as a priority to residents of NH with a very effective vaccination campaign in France [5]. Finally, from the fifth wave in France, the OMICRON variant became the majority variant, highly transmissible but with less severe clinical forms [6].

In France, regulations require that nursing homes report outbreaks of acute respiratory infections among residents and/or professionals. This reporting system, managed by Public Health France, allows the collection of epidemiological data and provides support to establishments for the implementation of measures to manage the epidemic by specialized regional structures [7].

We used this system to compare the evolution of the pandemic in the population and in NH, and to study its impact on NH residents and professionals over 2 years.

Methods

This cross-sectional study analyzed the incidence of COVID-19 in the general population and clusters in NHs (at least 3 confirmed cases in residents and/or staff within 7 days) in Normandy region (3,235,522 inhabitants in 2020, 348 NHs with 32,115 residents and 23,147 health care workers), France. Data from two national databases were used to monitor test results (“SI-DEP”, French Ministry of Solidarity and Health) and cluster case reports in medico-social institutions (“Voozanoo”, Public Health France, data as of March 1, 2022) [7].

The study period covered 2 years, from March 2020 to February 2022, divided in three sub-periods: 1/Period 1: 2020/03/02 to 2021/01/19 corresponding to waves 1 and 2 of COVID-19 (during this period, population incidence was not available until 2020/05/18, as no testing was performed); 2/Period 2: 2021/01/20 to 2021/12/19 corresponding to waves 3 and 4 and COVID-19 vaccination rate in residents of nursing homes ≥ 50%; 3/Period 3: 2021/12/20 to 2022/02/27 corresponding to wave 5 and Omicron variant ≥ 50% in tests.

The study was approved by the local ethic review board. The analyses were performed in R version 4.1.1 (R Development Core Team). Time series and cross-correlation analysis methods were used to study the relationships between series. The durations of clusters (date of last case—date of first case) and rates (with 95% confidence intervals—95%CI) for each sub-period were compared with the Kruskal–Wallis or the Chi2 test. To take into account the decrease in the number of residents due to deaths in previous clusters, and not to bias the estimation of attack and mortality rates, a sensitivity analysis limited to the first cluster episode in each NH was performed. A p value < 0.05 was considered significant.

Results

Overall, 84% of NH reported at least one cluster during the study period. Clusters were more frequent in period 1 compared to period 2 and 3 (p < 10–3) (Table 1). The weekly proportion of NH with clusters was strongly correlated with population incidence (Fig. 1B): in period 1, the peak of clusters in NH occurred in the same week as the peak of population incidence (r = 0.926, p < 10–3), in period 2, 1 week before (r = 0.700, p = 0.006) and in period 3, 2 weeks before (r = 0.727, p < 10–3).

Table 1 Number of cases, hospitalizations, and deaths in nursing homes with COVID-19 clusters. Normandy, France, 2020–2022
Full size table
Fig. 1
figure 1

COVID-19: study periods, population incidence, and impact in nursing homes. Normandy, 2020–2022. Graphs show weekly resident vaccination rates (first dose) and proportion of omicron variants in screening tests (A), COVID-19 population incidence and proportion of nursing homes with clusters (B), and COVID-19 attack rates and case fatality rates among residents of nursing homes with clusters (observed rates and moving averages) (C). Period 1: 2020/03/02 to 2021/01/19 (COVID-19 waves 1 and 2). Population incidence not available until 05/18/2020 (no testing performed). Period 2: 2021/01/20 to 2021/12/19 (COVID-19 waves 3 and 4, COVID-19 vaccination rate in residents of nursing homes ≥ 50%). Period 3: 2021/12/20 to 2022/02/27 (COVID-19 wave 5, Omicron variant ≥ 50% in tests)

Full size image

Proportion of NH with clusters remained stable in period 3 although the population incidence was very high (Fig. 1B). In NH with clusters, attack rates among residents and professionals were lower in period 2 compared with periods 1 and 3 (p < 10–3) (Table 1 and Fig. 1C). Among residents, mortality and case fatality rates were highest in period 1 and decreased drastically in the other periods. Sensitivity analysis showed similar results, with mortality rates for residents with COVID-19 of 4.2%, 1.6%, and 0.3% in periods 1, 2, and 3, and case fatality rates of 17.8%, 13.4%, and 0.8%, respectively.

Over the entire period, a total of 17 (0.1%) professionals were hospitalized and none died.

Discussion

We found a significant correlation between population incidence and the cluster peaks in the NH, which occurred before the community peak in period 2 and 3, likely reflecting early and mass screening strategies in NH. Such a relationship has already been pointed out in France [8] and in the United States [9]: a publication of the Center for Diseases Control concluded that COVID-19 prevention strategies in NH must include a comprehensive plan to monitor local transmission of SARS-CoV-2. The virus can be introduced into facilities by visitors or staff. In a study conducted in the canton of Geneva, Switzerland in 2020 [10], a strong correlation was found between staff SARS-CoV-2 seroprevalence and resident cumulative incidence of COVID-19 cases, suggesting that SARS-CoV-2 transmission between staff and residents may contribute to the spread of the virus within nursing homes.

We observed a relative reduction in mortality of -76% during the period of diffusion of vaccination in the NH (period 2 vs. 1), confirmed and of the same magnitude ( – 62%) when considering only the first cluster episodes (sensitivity analysis), and amplified ( – 89%) when the omicron variant became predominant. Finally, during period 3, resident attack and mortality rates were similar or lower than those observed during influenza epidemics [11]. The lower severity of clinical forms of the OMICRON variant, and the decrease in the number of the most vulnerable people after previous waves, may explain at least part of these rates. During the study period, the evolution of prevention measures in the population and medical treatment of patients also contributed to decreased COVID-19 morbidity. Studies also point the early detection and management of cases [12], and the development of immunity following successive waves of COVID-19, and especially vaccination [13].

Our data are declarative, but benefit from a high level of completeness and quality because they are based on the reporting system set up by the health authorities, which ensured an important quality control of the data. In addition to other known limitations for this type of study [1], we did not have the possibility to analyze the age and profile of the residents over the three periods, in relation with the impact of COVID-19 in NH. Nevertheless, our description allows us to objectify the evolution of the incidence rate and lethality in NHs.

It is important to continue to monitor the future evolution of the epidemic in NHs, which are also places where people live and work, to guide the adjustment of prevention measure when needed.