key: cord-0852075-i1xsdcg4
authors: Ioannidis, J. P. A.; Axfors, C.; Contopoulos-Ioannidis, D. G.
title: Second versus first wave of COVID-19 deaths: shifts in age distribution and in nursing home fatalities
date: 2020-11-30
journal: nan
DOI: 10.1101/2020.11.28.20240366
sha: b510d8d6125ef02beaa99c3f3b6f7a71325cd6cd
doc_id: 852075
cord_uid: i1xsdcg4

OBJECTIVE: To examine whether the age distribution of COVID-19 deaths and the share of deaths in nursing homes changed in the second versus the first pandemic wave. ELIGIBLE DATA: We considered all countries that had at least 4000 COVID-19 deaths occurring as of November 25, 2020, at least 200 COVID-19 deaths occurring in the first wave period, and at least 200 COVID-19 deaths occurring in the second wave period; and which had sufficiently detailed information available on the age distribution of these deaths. We also considered countries with data available on COVID-19 deaths of nursing home residents for the two waves. MAIN OUTCOME MEASURES: Change in the second wave versus the first wave in the proportion of COVID-19 deaths occurring in people <50 years old among all COVID-19 deaths and among COVID019 deaths in people <70 years old; and change in the proportion of COVID-19 deaths in nursing home residents among all COVID-19 deaths. RESULTS: Data on age distribution in eligible locations were available for 11 countries. Individuals <50 years old tended to have a larger share in the total COVID-19 deaths in the second wave than in the first wave in western European countries and the USA, but the absolute difference did not exceed 0.5% in any country. The proportion of deaths in individuals <50 years old was higher in Turkey and Ukraine, but it decreased in the second wave. Separate data on nursing home COVID-19 deaths for first and second waves were available for 9 countries. With the exception of Australia, the share of COVID-19 deaths that were accounted by nursing home residents decreased in the second wave, and the decrease was significant and substantial (relative risk estimates: 0.28 to 0.78) in 7/9 countries. CONCLUSIONS: In the examined countries, age distribution of COVID-19 deaths has been fairly similar in the second versus the first wave, but the contribution of COVID-19 deaths in nursing home residents to total fatalities has decreased in most countries in the second wave.

CONCLUSIONS: In the examined countries, age distribution of COVID-19 deaths has been fairly similar in the second versus the first wave, but the contribution of COVID-19 deaths in nursing home residents to total fatalities has decreased in most countries in the second wave.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020. 11.28.20240366 doi: medRxiv preprint What is known on this topic * COVID-19 risk of death has a very steep age gradient * Many COVID-19 deaths occur in nursing home residents * Many countries have seen a pattern of two separate waves of COVID-19, but it is unknown whether these two waves differ in the age distribution of COVID-19 deaths and in fatalities in nursing home residents What this study adds * Age distribution of COVID-19 deaths has been fairly similar in the second versus the first wave in most countries, with some exceptions. * Deaths in people <50 years old remain a small minority of COVID-19 deaths.

* The contribution of deaths in nursing home residents to total fatalities remains high in absolute magnitude, but it has decreased in most countries in the second wave.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.28.20240366 doi: medRxiv preprint

Many countries around the world, in particular in Europe, have seen a pattern of the coronavirus disease 2019 (COVID- 19) pandemic where a first wave occurred in the spring that substantially subsided during the summer, and a second wave emerged in the fall of 2020. A key question is whether the age distribution of COVID-19 fatalities in these locations has remained steady between the two waves or not.

COVID-19 is known to have an extremely steep risk gradient for death across age groups (1) (2) (3) . If the relative share of infections is larger among young people, the infection fatality rate in the total population is expected to be lower. Conversely, if the relative share of infections is larger among older people, the infection fatality rate is expected to be overall higher.

Data from a considerable number of seroprevalence studies done in different countries have suggested that not all age groups may have been equally infected during the spread of the virus in the first wave (4). In several countries that did very well in the first wave, e.g. Singapore, Australia, or Iceland (5) evidence suggests that elderly people were less likely to be infected and in particular nursing homes were not contributing many fatalities (4). The opposite pattern was seen in countries that had high rates of death, where nursing homes were massively infected during the first wave (6) . One wonders whether preferential protection of older, higher-risk people and in particular of nursing home residents is possible and whether this can happen more efficiently in the second wave (7) .

In order to answer these questions, we assessed the age distribution of COVID-19 fatalities in the two waves in countries with a substantial burden of fatalities as of November 2020, and we also assessed the relative contribution from deaths of nursing home residents in the two waves.

We considered data from publicly available situational reports of countries that had a large number of deaths in both the first and second waves, so as to allow meaningful inferences in comparing the age . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.28.20240366 doi: medRxiv preprint distributions, and where the trough between the two waves had happened between May 15 and September 15.

Specifically, we considered all countries that had least 4000 COVID-19 deaths, at least 200 COVID-19 deaths occurring in the first wave, and at least 200 COVID-19 deaths occurring in the second wave; and which had information available on the age distribution of these deaths separately. Searches were done on November 13 and updated on November 25 to see if any additional countries had become eligible. In order to separate the two wave periods in a consistent manner, we used the date between the two peaks that had the trough (lowest number of deaths) for a 7-day average according to Worldometer data (8) . When two or more dates were tied for trough values, we picked the earliest one.

For each eligible country, we found the most recent situational report that mentioned age distribution of deaths; and the situational report that mentioned age distribution of deaths as of the trough date (or a separation date for the two waves that was as close as possible to the trough, when data were not available specifically up to the trough date). It is acknowledged that a few deaths reported after the trough/separation dates may have happened earlier, but this is likely to pertain to very small numbers and it would not change the overall comparison.

Age cut-offs of interest were pre-specified to be 50 years and 70 years. In European countries that have seen two waves and in the USA, almost two-thirds of the population are younger than 50 years old and the age stratum of 50-69 accounts for another quarter of the population. In a few other countries that have also seen two waves, the proportion of the population younger than 50 years old even exceeds 80%, e.g. in Turkey or Pakistan. However, the lion's share of COVID-19 deaths happen in people above 70; moreover, in countries where elder care facilities are common, a large share of COVID-19 deaths occur in nursing home residents. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.28.20240366 doi: medRxiv preprint occurring in people <50 years old among the COVID-19 deaths occurring in people <70 years old changed between the two waves. When data were not provided for the cut-offs of 50 and 70 years for COVID-19 deaths, we used the closest cut-offs available provided it was not more than 5 years off (45 and 65 years).

We also extracted data on COVID-19 deaths occurring in nursing home residents. We preferred data on all COVID-19 deaths of such residents (occurring in the nursing homes or in hospitals) unless information was available only for COVID-19 deaths happening in nursing homes. We also preferred data on both confirmed and probable COVID-19 deaths, unless only data on confirmed deaths were available. For these data to be eligible, the definition of nursing home resident should not have changed in the two waves.

We compared proportions of these nursing home residents' deaths among all COVID-19 deaths in the second versus first wave periods. The International Long-Term Care Policy Network has issued reports based on available country-level official data for the number of deaths among nursing home residents linked to COVID-19. Their latest report (October 14, 2020) (6) and previous report (June 26, 2020) (9) were considered.

Given the substantial spread of COVID-19 occurring in many countries after their latest report, we restricted our analysis to the countries for which we could find updates that covered at least until November 6, 2020 using the official sources cited in the International Long-Term Care Policy Network reports. Latest update for these data was on November 23, 2020. We used as cutoff for the first wave the dates given in the June 26 report (between June 1 and June 23).

All comparisons of proportions of COVID-19 deaths per country between the two periods used risk ratios and 95% confidence intervals thereof. Meta-analysis of risk ratios used a random effects model.

Heterogeneity was expressed with the I 2 statistic and tested with the chi-squared-based Q test. P-values are two-tailed. Statistical analyses were run in STATA (10).

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.28.20240366 doi: medRxiv preprint

Thirty-six countries had at least 4000 deaths until November 13. Of those, 17 countries (Belgium, Canada, Czechia, France, Germany, Italy, Netherlands, Pakistan, Poland, Romania, Russia, Spain, Sweden, Turkey, Ukraine, United Kingdom, USA) also had at least 200 deaths in each wave and had a trough between May 15 and September 15. Another 3 countries (Hungary, Portugal, Switzerland) passed 4000 deaths by November 25 and would also be eligible, for a total of 20 eligible countries. We could retrieve some data on the age distribution separately for the two waves for 15 of the 20 countries. However, Canada and Sweden did not show deaths in the age group <50 (or <45) years old; and Portugal and Romania provided only graphs with percentages per age group and resolution was not sufficiently accurate in the <50 years old age group.

Eventually, data from 11 countries were finally included ( Table 1 ). As shown in Table 1 , the trough separating the two waves occurred between June 1 and August 21 in all locations. The number of deaths in the second wave as of the date of the analysis was lower than the number of deaths in the first wave in 9 countries, whereas it was already higher in the second wave in Switzerland and Ukraine. Sources of information for the 11 countries appear in Supplementary Table 1 . Table 2 shows the age distribution of COVID-19 deaths in the first and second wave in each eligible country with data.

The proportion of COVID-19 deaths <50 years among all COVID-19 deaths in the first wave did not exceed 1.5% in any western European country, but it was higher in the USA (2.8% for a cut-off of <45 years) and it was even higher in Turkey (6.8%) and in Ukraine (12.7%). In the second wave, the proportion did not change by more than 0.5% in western European countries and in the USA, but the absolute change was larger in Ukraine and Turkey. In most countries, the proportion tended to increase, but it decreased in France, Ukraine and Turkey (Figure 1a) , creating substantial between-country heterogeneity in the risk ratios . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.28.20240366 doi: medRxiv preprint comparing the proportions in the two waves (I 2 =88%, p<0.001 for heterogeneity). Given this heterogeneity, single summary estimate (summary risk ratio 0.96, 95% CI 0.80-1.16) may not be very reliable.

When estimated only among the COVID-19 deaths in people <70 years old, the proportion of COVID-19 deaths in individuals <50 years old tended to increase in the second wave versus the first wave, but again differences were relatively small (Figure 1b) . There was still large between-country heterogeneity (I 2 =79%, p<0.001 for heterogeneity) and the summary estimate of the risk ratio (1.04, 95% CI 0.91-1.20) was thus also tenuous. Excluding Turkey and Ukraine, there was limited between-country heterogeneity among the remaining 9 high-income countries (I 2 =14%, p=0.32 for heterogeneity) and the summary risk ratio suggested a small increase in the relative representation of individuals <50 years old among COVID-19 deaths in people <70 years old (summary risk ratio 1.10, 95% CI, 1.03-1.19).

Eligible data on nursing home residents' COVID-19 deaths in the first versus second wave could be obtained for 9 countries ( Table 3) . Sources of data appear in Supplementary Table 2 . Notably, the definitions of COVID-19 deaths and nursing home institutions differed across the different countries. In 3 countries, nursing home deaths data included only deaths that occurred in the nursing home environment (i.e. in-hospital deaths of nursing home residents were excluded). In 4 countries, only confirmed COVID-19 nursing home deaths were considered. Definitions of nursing home facilities and other included facilities appear also in Table 3 .

With one exception (Australia), the proportion of nursing home COVID-19 deaths among all COVID-19 deaths was lower in the second wave than in the first wave ( Figure 2 ). There were large and statistically significant relative risk reductions in 7 countries (risk ratios 0.28 to 0.78), no significant different in Denmark where few deaths were recorded in both waves, and a major increase in the proportion of nursing home COVID-19 deaths in Australia (risk ratio 2.87) in the second wave. This resulted in extreme between-country heterogeneity (I 2 =98%, p<0.001 for heterogeneity) making a summary risk ratio tenuous to obtain.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.28.20240366 doi: medRxiv preprint

The age distribution of COVID-19 deaths did not change much within each examined western European country or the USA between the first and second waves. A small, non-significant trend for a higher share of younger COVID-19 deaths was seen in most of these high-income countries in the second wave.

Concurrently, there was a strong pattern for a decreasing share of COVID-19 deaths of nursing home residents in the second wave versus the first wave, with Australia being an exception to this pattern. Data from two middle-income countries (Turkey and Ukraine) showed overall much higher proportions of COVID-19 deaths among people <50 years old, with a relative decrease in the second wave.

The decreasing share of nursing home resident COVID-19 deaths in the second wave may reflect multiple factors. Higher awareness of the extreme fatality risk of nursing home residents and raised efforts to protect nursing homes (learning from experience) (11, 12) may be responsible for this pattern. Perhaps hygiene measures, infection control, testing of personnel and residents, and avoidance of staff working across multiple nursing homes have become more routine during the second wave. Another potential contributing factor may be better treatment and management. For example, dexamethasone (13) has been available in the second wave;

conversely, use of mortality-increasing hydroxychloroquine (14) and suboptimal mechanical ventilation practices may have been reduced in some countries. Cohort effects are also possible: e.g. the first wave may have killed some of the most frail residents and/or may have already preferentially devastated nursing homes with poorer standards and thus higher infection risk.

Some additional factors need to be considered. Within the same country, the second wave may have spread to different communities with different shares of nursing home populations compared with the first wave. Also some data artefact cannot be fully excluded, in particular if some nursing home COVID-19 deaths in the second wave suffer reporting delays. Finally, we tried to ensure that definitions of deaths and nursing . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.28.20240366 doi: medRxiv preprint home facilities were similar in the two waves, but subtle changes cannot be excluded. In all, the difference between the two waves in overall COVID-19 deaths in most countries is quite large, thus unlikely to be only a data or methods artefact.

Despite the clear improvement, the proportion of COVID-19 deaths that occurred among nursing home residents remained very large in western European countries and the USA. Moreover, not all countries have seen improvements in the second wave. Australia witnessed a major increase, and this may be explained by non-remedied dysfunctions in its elder care (15) . The Australian aged care system has long been criticized for understaffing, and using low-pay staff with poor skills who work across multiple facilities. Most facilities are privately run for-profit enterprises. Similar problems with nursing home care inefficiencies have been described also in other countries with a large share of deaths in nursing homes, e.g. in Canada (16); however, in other countries, COVID-19 already hit nursing homes rapidly during the first wave.

The relative stability of the age distribution of COVID-19 deaths in the same country between the two waves suggests that country-specific population demographics are the key driver of the age distribution of infections, which then get reflected also in the age distribution of deaths (17) . If anything, we observed a possible increase in the share of COVID-19 deaths among people <50 years old in the second wave in most high-income countries. The absolute difference is small, however; and this may well be a chance finding.

Otherwise, it may reflect that older individuals got relatively more protected so far during the second wave than during the first wave. Of interest, COVID-19 deaths in the examined countries during the first wave largely happened during lockdown conditions. Conversely, COVID-19 deaths in the second wave until November reflect infections that happened until October under mostly less restrictive measures without implementation of aggressive lockdown measures. Possibly young people are relatively more frequently infected when there are few restrictive measures, while lockdown may be associated with a more even infection risk across age groups.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.28.20240366 doi: medRxiv preprint One would also expect shifts in the age distribution of COVID-19 deaths in the second wave if it is spreading to new epicenters and populations with different demographics. This may be the main reason for the paradoxical shift seen in Turkey where relatively more COVID-19 deaths in young people were seen in the first wave. During the first wave, the highest numbers of cases per million population were seen in Istanbul and southern Turkey (18). Both epicenters are characterized by a very young population, younger than the inhabitants of other locations that were hit more in the second wave.

The observed patterns of diminished COVID-19 deaths in nursing homes and a small, less certain shift with more COVID-19 deaths in younger individuals in high-income countries may both act in the direction of decreasing the infection fatality rate of the pandemic in the second wave. For an equal number of total infections, when younger people are infected and when nursing homes are spared, the number of total COVID-19 deaths will be less (19) . Paradoxically, having a larger share of young deaths is thus not necessarily a bad sign, but may indirectly be an indicator of better protection of vulnerable elderly individuals. Conversely, for other countries where the pandemic shifts from urban centers with young population to areas with older populations, infection fatality rate may increase.

More data would need to be accumulated as the second wave progresses to validate whether the patterns detected so far continue to hold true or are reversed. For example, in particular, use of more aggressive lockdown measures in several countries as the second wave evolved in late fall might affect the demographic composition of COVID-19 fatalities even within the same wave over time. We did not have sufficient data to investigate the evolution of demographic features among fatalities during the first wave in the current analyses. However, some countries like Italy have presented data where the median age of fatalities during the first wave started at lower values, increased during lockdown, and then declined again as restrictions were relieved (20) .

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.28.20240366 doi: medRxiv preprint Some limitations need to be discussed. First, age information was missing on some deaths, but the missingness is unlikely to have created systematic bias in the comparison of first versus second waves.

Second, many nursing home deaths have substantial ambiguity in their attribution to COVID-19, especially when there is no test confirmation. We tried to use data with consistent definitions and approaches in the two waves, but the two periods may still differ, e.g. typically more testing was done in the second wave. If anything, this would usually tend to increase the number of confirmed COVID-19 deaths in nursing homes in the second wave. Third, it would be useful to understand whether there are differences between the two waves in the share of deaths in people without underlying conditions and/or specific risk profiles. This could give additional insights about the relative exposure and protection of these groups in the two time periods.

However, such data are very sparse in currently available situational reports (e.g., 21, 22). Fourth, we did not find sufficiently detailed data on age distribution of COVID-19 deaths even for some high-income countries (in contrast to countries that make even their historical reports easily available). This is a deficiency that could be quickly corrected in country-level situational reports. We found very sparse data from middle-income countries and no data from low-income countries. However, most low-income countries have not had a trough separating two waves. Instead, they have typically seen continuous epidemic activity comprising a single wave at least to-date.

Acknowledging these caveats, demographic profile changes in the future evolution of the second waves and of any other surges of COVID-19 pandemic activity should be monitored. Moreover, the effects of different non-pharmaceutical measures, as well as prospective vaccinations during 2021, on the distribution of deaths warrants close attention. It would be useful to understand the extent to which the epidemiological footprint of the epidemic waves is largely intrinsically dependent on the population structure and on the virus (and thus relatively constant) or, conversely, can be efficiently modulated by appropriate interventions.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.28.20240366 doi: medRxiv preprint

Santé publique France. COVID-19. Point épidémiologique hebdomadaire du 12 novembre 2020. https://www.santepubliquefrance.fr; 2020.

Rijksinstituut voor Volksgezondheid en Milieu -RIVM. Epidemiologische situatie COVID-19 in Nederland, 10 november 2020, 10:00. https://www.rivm.nl; 2020.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.28.20240366 doi: medRxiv preprint and in these cases the most proximal date with data available was used. **data were not always available until November 13, and in these cases the most proximal date with available data was used.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.28.20240366 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.28.20240366 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.28.20240366 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.28.20240366 doi: medRxiv preprint Figure 2 . Proportion of COVID-19 in nursing home residents in the second versus first wave.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 30, 2020. ; https://doi.org/10.1101/2020.11.28.20240366 doi: medRxiv preprint

OpenSAFELY: factors associated with COVID-19-related hospital death in the linked electronic health records of 17 million adult NHS patients

Population-level COVID-19 mortality risk for non-elderly individuals overall and for non-elderly individuals without underlying diseases in pandemic epicenters

Age-specific mortality and immunity patterns of SARS-CoV-2

Precision shielding for COVID-19: metrics of assessment and feasibility of deployment

Spread of SARS-CoV-2 in the Icelandic Population. The New England journal of medicine

Mortality associated with COVID-19 in care homes: international evidence. Article in LTCcovid.org, International Long-Term Care Policy Network

Shielding from covid-19 should be stratified by risk

COVID-19 CORONAVIRUS PANDEMIC Dover, Delaware, USA2020

Mortality associated with COVID-19 in care homes: early international evidence. Article in LTCcovid.org, International Long-Term Care Policy Network, CPEC-LSE

Evolution and effects of COVID-19 outbreaks in care homes: a population analysis in 189 care homes in one geographical region of the UK. The Lancet Healthy Longevity

COVID-19 related mortality and spread of disease in long-term care: a living systematic review of emerging evidence

Dexamethasone in Hospitalized Patients with Covid-19 -Preliminary Report. The New England journal of medicine

Mortality outcomes with hydroxychloroquine and chloroquine in COVID-19: an international collaborative meta-analysis of randomized trials

Experts criticise Australia's aged care failings over COVID-19. The Lancet

COVID-19 in longterm care homes in Ontario and British Columbia

Use of "normal" risk to improve understanding of dangers of covid-19

Global perspective of COVID-19 epidemiology for a full-cycle pandemic

Caratteristiche dei pazienti deceduti positivi all'infezione da SARS-CoV-2 in Italia

Supplementary Table 1. Sources for COVID-19 deaths in specific age groups Countries First wave Second wave (if different source) Belgium COVID19BE_MORT_20201114.xlsx

COVID-19). 01.08.2020 -AKTUALISIERTER STAND FÜR DEUTSCHLAND Robert Koch Institut

2-integrated-surveillance-data Netherlands Rijksinstituut voor Volksgezondheid en Milieu -RIVM. Epidemiologische situatie COVID-19 in Nederland

Situación de COVID-19 en España. Casos diagnosticados a partir 10 de mayo. Informe COVID-19

Casos diagnosticados a partir 10 de mayo

Turkey Ministry of Health

19-current-situation-and-case-numbers#cases-in-aged-careservices Belgium Sciensano: the Belgian Institute for Health

Täglicher Lagebericht des RKI zur Coronavirus-Krankheit-2019 (COVID-19) 13.11.2020 -AKTUALISIERTER STAND FÜR DEUTSCHLAND England and Wales (UK) publishedweek4320202.xlsx

All websites were accessed 2020-11-23. * For the first wave, the following reference was used for all locations: Comas

Mortality associated with COVID-19 outbreaks in care homes: early international evidence. Article in LTCcovid.org, International Long-Term Care Policy Network