While the applied surveillance was intended in the first place to serve the RHA to manage the crisis and mitigate the epidemic, it allowed us to quantify the COVID-19 pandemic in Belgian NHs prior to the vaccination campaign. Unfortunately and as notified in many countries, Belgian NHs were severely hit by this pandemic. Only a few NHs (3%) never reported any possible or confirmed COVID-19 case among residents. Most NHs (82%) reported cases in both waves. Moreover, absenteeism data show that a considerable number of NHs suffered from staff shortage due to COVID-19 infections among staff members. This is an underestimation of the real absenteeism as other reasons for absence such as chronic conditions like burn-out were not taken into account.
From 12 March 2020 onwards, no visitors were allowed in the NHs for at least 40 days. Although this measure was implemented in the very beginning of the pandemic, it could not avoid the spread of SARS-CoV-2 in the NHs during W1. This suggests that the virus was introduced in this period in many NHs by staff. Asymptomatic cases of staff might have played a role here . Only since 16 April (due to the limited supply of masks in the beginning of the pandemic), the use of surgical masks became recommended for contacts with residents. The initial lack of personal protective equipment (PPE) and infection prevention and control (IPC) training suggests that the majority of cases in W1 were healthcare-associated. Between the end of April and half of May (depending on the regional health authority), a limited number of visitors was allowed again and this remained so until the end of the study period. During the study period, IPC measures of different kind have been implemented at regional and local level (e.g. quarantine measures, cohorting of (possible) cases among residents and staff, use of additional PPE, preventive and rapid testing, ventilation). Each RHA provided detailed instructions for NHs how to deal with COVID-19, but also human resources support initiatives, pro-active calling centres and professional education support. Depending on the available resources and based on the results of the surveillance, outbreak investigations were set up for the largest outbreaks and in NHs with the highest need for local support. Due to important regional and local differences over time it is difficult to assess the impact of the different mitigation measures taken. We can nevertheless reasonably assume that without these measures, W2 would have been much worse .
In both waves, the peak in prevalence occurred two weeks after the peak in incidence numbers. On 12 May, it had been clarified in the protocol that a confirmed case remained a case for at least 14 days (or longer depending on symptoms). The cumulative incidence of possible/confirmed cases among residents during W2 was almost twice as high compared to the end of W1, giving the impression that W2 was even worse than W1. One should be careful when comparing these waves. In the beginning of W1, extensive and systematic testing of possible cases in NHs was not possible as free tests were essentially reserved for hospitalized patients. When it became clear that many NHs reported high COVID-19 mortality rates and more tests became available, a massive testing campaign was set up between 8 April and 18 May 2020. This cross-sectional mass testing revealed that about 75% of those that tested positive (residents as well as staff) were asymptomatic at the moment of testing . After this campaign, screening of all residents continued in the context of outbreak investigations, which had an impact on the results. Moreover, over time, NH staff became more familiar with the surveillance, leading possibly to better data registration and quality. Additionally, incidence data were missing for Wallonia in W1, which may have introduced selection bias. All these reasons explain why the results for W2 give a more complete picture and are more robust than those for W1. Nevertheless, despite enough PPE being available in W2 (in contrast with W1), still a high number of COVID-19 cases were reported.
Relatively more residents were hospitalised in W2 as compared to W1. It should be noted that information on residents that refused to be admitted to the hospital and residents that were refused by the hospital due to capacity issues is missing. Moreover, in the beginning of the epidemic, the geriatric society advised openly that residents with COVID-19 should only be referred to a hospital if they would substantially benefit from this hospitalization. Hence, the number of hospitalizations of residents might not be the best indicator to evaluate the severity of the infection. The number of deaths can be considered as a more stable indicator. With 57% of the COVID-19 deaths in 2020 being NH residents, this population (only 1.1% of the total population) was the main group affected. At the beginning of the pandemic, Belgium was acknowledged for its excellent registration including possible COVID-19 deaths in the COVID-19 death statistics, in contrast to many other countries who reported only confirmed cases or even only hospital deaths . The number of deaths of NH residents in W1 was not exceeded in W2 (if counted until 3 January). W2 lasted longer than W1, but peaked less high in terms of deaths. In November 2020, the increase of COVID-19 cases in LTCFs in Belgium and other European countries led to a rapid risk assessment of ECDC. This assessment indicated that several countries were suffering from a second COVID-19 wave in this setting with many fatal cases . Since the NH setting and population can be highly variable across countries, comparing with other countries is difficult.
Since July 2020 (sufficient testing capacity) the 7-day incidence of confirmed COVID-19 cases among NH residents shows a similar (but higher) trend as this incidence in the community. In the community, a sharp increase was observed in week 40 (end of September) , whereas in NHs this was observed in week 41. Both the possible introduction by staff (and visitors) and the week delay in the increase of the trend, suggest that the virus was, again in W2, silently introduced from the community. Similar findings have been reported in other countries (UK, Canada, United States) [23,24,25]. On top, the fact that 51% of the NHs reported a prevalence of at least ten possible/confirmed cases at one moment in time during W2, suggests that once the virus was introduced in the facility, it was still difficult to prevent spread and outbreaks.
Strengths and limitations
The Belgian COVID-19 surveillance in NHs has been set up very quickly in the context of the pandemic, leveraging the existing experience with infectious disease surveillance in hospitals and LTCFs. Only 14 days after the first COVID-19 death, the surveillance was operational in all regions. The surveillance is part of the national action plan for COVID-19. Several complementary surveillances (e.g. COVID-19 surveillance of confirmed cases in the general population, of cases admitted to Belgian hospitals  and of mortality ) were put in place and data are linked where possible .
Including possible cases permitted to detect a possible outbreak early and intervene quickly with focussed testing (maximal five tests per NH when an outbreak was suspected in the first months of the epidemic), assistance with PPE, extra staff, psychosocial support and infectious disease management courses. Looking at lab-confirmed cases only would have led to an important underestimation of the incidence/prevalence, especially in the beginning of the COVID-19 crisis. However, it should be taken into account that possible cases might have included respiratory tract infections other than COVID-19 and might have led to an overestimation of the incidence/prevalence of COVID-19.
A high weekly participation rate (median 95%) was reached over the whole study period. Downscaling the registration in some regions from daily to at least once a week in the beginning of July ensured that it remained feasible for all NHs to keep registering data. By collecting also denominator data of residents and staff, relative numbers could be presented making the interpretation over time with varying participation rates easier. A good cooperation between the RHA and Sciensano made it possible to perform analyses on a national level and to visualize the COVID-19 epidemic in Belgian NHs. Since the beginning of the pandemic, detailed weekly reports of the surveillance were made available to the authorities and general public . The COVID-19 surveillance in Belgian NHs has been used as an example in ECDC guidelines .
The surveillance also has its limitations. Firstly, as mentioned earlier, the surveillance was set up very quickly in the context of the pandemic by the different RHA, independently from each other. Changes in the protocol, case definitions, data collection tools and variables to be collected were necessary during the first months to respond to the evolving insights in this emerging crisis and to harmonise the data collection. Therefore data of W1 are less robust. Secondly, due to the speed with which this surveillance was set up, there was no time for extensive testing of the different tools. In addition, NH staff were not familiar nor trained to collect surveillance data. Especially in this older population with often vague complaints, it might have been difficult to identify possible cases based on symptoms. Moreover, the surveillance is based on aggregated and not case-based data (except for deaths), which limits the possibilities for data validation (e.g. identification of possible double reported cases, missing cases or delayed reporting), but also for the interpretation of the results. Despite these limitations, it was possible to monitor the epidemic guiding the policy makers in their decisions. However, for more in-depth analyses (e.g. cluster analyses) and to investigate other determinants that played a role (e.g. comorbidities), a case-based registration would be needed. Finally, analyses were limited to the first and second COVID-19 wave and only performed at national level. In future studies, an impact analysis of the vaccination campaign (which started in January 2021) and regional differences could be considered.