Unravelling the true death toll of COVID-19
Article information and share options
How deadly is it?
This has been the burning question since the earliest days of the COVID-19 outbreak. As more credible data is collected, epidemiologists, actuaries and modelers continue to analyse the virus and unravel the true death rate.
Beyond the emotional toll of each life lost, the collective number of COVID-19-related deaths is important for many reasons. The mortality rate influences vital decisions around public health policy and interventions. The economic fallout of mobility, trade and other restrictions affects both life and non-life insurers. Life insurers are more directly impacted by the death toll, by a change in the level of claims and changes that may be required for underwriting and pricing practices.
So, the big questions then are: How accurate is the death count? How does it vary among people who were already sick with another illness? How many more people died as a result of the pandemic (beyond what would normally be expected?) And what will be the effect on life insurers?
We aim to give you a better understanding of these issues with the best information available thus far. As the pandemic has not fully run its course, we will continue to see new developments and findings.
The studies and data used in this paper reference a wide range of sources and countries. The conclusions drawn are however generally broad enough to apply across countries and societies affected by the pandemic, with some exceptions which are noted where relevant.
Section 1: How accurate are the death counts for COVID-19?
It used to happen once a year. Government statisticians would release a flurry of tables that summarise deaths from more than a year ago, followed by a barrage of actuarial analyses to identify trends in mortality. Now it happens live on-air, around the world, every day as various institutions record and report COVID-19 deaths. But how accurate are these "real-time" numbers?
Daily death mis/counts and delays
The truth is that the number of deaths cannot be tracked so easily or immediately, nor do the numbers fit neatly into media deadlines. As well as being a distressing time for families, registering the death of a person can be a protracted process with serious legal implications. Daily reports may include deaths that happened the day before – or weeks before. Test results to determine official cause of death are often delayed over weekends and public holidays and processing death certificates and other documentation can also be hindered. All of this, combined with the high number of deaths stressing the system during this pandemic, mean the patterns of reported numbers don't always correspond to reality.
Comparisons across countries and US states are complicated by differences in how long it takes to file death notifications (as highlighted by a recent series on excess deaths by the Financial Times)1. Depending on the country and reporting dates, delays in data processing can range from a week, to more than two weeks, with first reported numbers lower than the final tally in all cases. For example, in the UK, the weekly reported number of deaths is typically 80% of the final number. In the US, the initial number is estimated to be 60% of the final number according to the Centers for Disease Control and Prevention (CDC).
Another reason for under-reporting is that not all deaths will necessarily be captured in official COVID-19 statistics due limited testing and post-mortem diagnosis. If no reference is made in the death certificate to COVID-19 (for example pneumonia may be cited as the official cause of death even though it was caused by the virus), then that death might not be included in the official COVID-19 death count.
Clinical guidance and case definitions vary by country
Comparisons are complicated by differences in how countries attribute a death to COVID-19. Early in the epidemic, the clinical guidance varied from country to country. Consistency has improved as more countries base their clinical guidance on the "cause of death" classifications from the World Health Organisation's (WHO) "International Classification of Diseases" guidelines. For example, both the UK and US provide similar guidelines for doctors that the cause-of-death statement is an informed medical opinion that should be based on sound medical judgment drawn from clinical training and experience, as well as knowledge of current disease states and local trends2,3. CDC guidelines in the US are explicit: COVID-19 is to be recorded as the underlying cause of death if it is believed to have been part of the disease progression that led to death (even if it's not the direct cause). If the deceased exhibited symptoms of being infected, the doctor could apply clinical judgement to list COVID-19 as the cause of death – a lab test is not always required.
In some countries, more stringent rules require a mandatory laboratory confirmation to register a death as COVID-19. This means it's more likely deaths are officially attributed to another disease or organ failure as in the case with Russia4. At the other end of the spectrum, Belgium has chosen to report every death in nursing homes that is potentially attributable as COVID-19, which could be one reason why its COVID-19 mortality rate is among the highest in the world5.
Another complication is the fact that the pandemic introduced a new and unknown virus that has not yet been studied in sufficient detail. As the disease affects more people, and as diagnostics improve, we learn more about emerging symptoms and settings. As a result, the clinical case definition and guidance for medical practitioners who record deaths continues to evolve.
The differences in the death count from three publishing bodies in the UK illustrates many of these factors. From 29 April 2020, COVID-19 deaths in England were reported using the Public Health England (PHE) data series, which combines data on deaths occurring in hospitals, deaths notified to local PHE Health Protection Teams in the course of outbreak management and laboratory reports where a person had a confirmed COVID-19 test linked to death reports from electronic hospital records. The National Health Service (NHS) counts COVID-19 deaths as those occurring in a hospital with a positive test for the virus, while the Office of National Statistics (ONS) records COVID-19 deaths as those where COVID-19 is mentioned in the death certificate. Figure 1 depicts the difference in total death counts across these three sources.
Courtesy: COIOS Research
Similarly, China expanded its case definition in mid-February for Hubei Province only, in light of its limited testing capacity. In Hubei, a chest X-ray imaging was allowed to confirm "probable COVID-19" instead of only RNA testing which was required in the rest of the country. As a result, there was a spike in the number of reported COVID-19 new cases.
Estimated completeness of data
The question of how complete provisional data is will vary by jurisdiction, reporting frequency, age of the deceased, and cause of death. Until data for a calendar year is finalised, (which typically happens sometime in the following year) the completeness of the data cannot be confirmed.
To address this, actuaries have developed techniques to estimate outcomes that have not yet been reported. By calculating development factors, actual "incurred but not reported deaths" (IBNR) can be estimated. In the UK, the NHS provides a daily break down of all reported deaths by age, region, ethnicity and date of death. This dataset allows actuaries and demographers to estimate COVID-19 deaths in NHS hospitals that have likely happened but have not been recorded. This has now fallen significantly as the number of formally diagnosed cases has also decreased. Figure 2 shows the pattern between the steady trend in estimated deaths, versus the pattern of reported deaths in NHS hospitals.
Not perfect, but good enough
All this tells us that the death count on any given day may not fully reflect the true death toll and doesn't lend itself to a straightforward comparison across countries. Despite this, official death counts remain the best source to determine mortality and whether lockdowns and other social interventions have been effective and when they might be phased out. Hospital occupancy rates can be a more immediate measure of a change in infection rates, allowing for interventions to be taken earlier.
Experts believe that since we know much more now than at the start, changes in clinical guidance won't be as significant and won't affect death counts to the same degree as early in the pandemic. Some comfort can be gained that a reasonable level of consistency has been established across countries that were hit early based on WHO clinical standards, although we may continue to see different clinical practices in countries where the pandemic escalated more recently.
Section 2: Who is at greater risk?
At the onset of the pandemic it became quickly apparent that some groups were more vulnerable than others. While the epidemiological implications of the virus will take time to fully understand, there are clinical learnings we can already apply. Understanding who is vulnerable and why is especially critical to design protective public policy responses.
Statistics consistently show that men are more likely to die than women, even though in most countries the proportion of infected men and women is largely the same. We explored this more in our paper on immunity testing. There is also a clear link between age and fatality rates, with older ages dying at much higher rates across countries and regions. This is comparable to the 2018 influenza season in the US, which saw a 2-6 fold increase in hospitalisation rates for those 85+, compared to those in the 65 – 74 age range6. Among juveniles, the fatality rate is not comparable to flu as COVID-19 appears to have a milder effect on children than flu.
Source: Swiss Re COVID-19 response
Source: Swiss Re COVID-19 response
Older people tend to have more pre-existing chronic conditions, and many of these are comorbidities. Early data and studies did not distinguish the relative contributions of age, gender and comorbidities to fatality rates separately, but it quickly became apparent that unhealthier lives were paying a disproportionate price. As the pandemic expanded across countries and continents, more data and research confirmed the positive link between comorbidities and COVID-19 death, among other risk factors.
While age and gender are the most obvious risk factors, we can also link pre-existing health conditions to an elevated fatality risk from COVID-19. An early study on 72,000 patients in China suggested a relative risk of 6 to 11 times among people with at least one pre-existing condition including hypertension, diabetes, cancer, cardiovascular disease, and chronic respiratory disease vs. healthy lives without comorbidities.
Adjusted for age, smoking status and other comorbidities, another early Chinese study on 1,590 patients suggested conditions such as chronic obstructive pulmonary disorder (COPD), diabetes, high blood pressure and cancer would increase the fatality rates from 1.58 to 3.50 times more compared to those without these conditions, as calculated by the hazard ratios7.
The most comprehensive study published to date included records from 17 million UK patients from 1 February to 25 April 20208. This study adjusted for confounding factors in the risk factors it analysed. This indicated that people who were overweight, impoverished, had one of a range of pre-existing conditions (including diabetes, recent diagnosis of cancer, organ transplant, stroke, dementia, etc.) had a higher risk to contract the disease and die from it. With good treatment for their underlying conditions, these people could have lived longer were it not for contracting COVID-19.
This hazard ratio of smokers versus non-smokers raised the provocative thought that maybe smokers are at a lower risk of COVID-19 death. Could it be that smoking actually reduces your risk? Opinions are mixed as there is confounding evidence. From a biological standpoint, it is possible that the chemical pathway that COVID-19 uses to enter the body's cells could be impacted by the presence of nicotine in the blood. We explain more about this process (known as "ACE2 receptors") in this article on developing drugs to treat COVID-19. A French study even suggested that there could be a protective effect from nicotine against the virus, leading to a lower risk of death9. A UK study also suggested a potential protective effect for current smokers10. This is far from conclusive, and other research suggests the opposite. Given the well proven serious health effects from smoking, it's certainly not a good reason to start or continue a smoking habit.
Why are nursing home deaths so high?
From the very first noted outbreak in the US, at The Life Care Center in Kirkland, Washington, outbreaks in nursing homes have been a key factor in the high death rates. In the US alone, estimates show 30 to 40% of the deaths occurred in nursing homes, and in the UK estimates are that close to 50% occurred in care homes. This has been seen across many western countries, including France, US and the UK.
Source: UK Office of National Statistics
The International Long Term Care Policy Network (ltccovid.org) published an update that shows through 21 May, the percentage of COVID-19-related deaths as compared to all deaths among care home residents ranged from 24% in Hungary to 82% in Canada.
There are a host of reasons why frail, elderly people in nursing homes are particularly susceptible to dying from COVID-19. Older people in general have reduced immune response due to fewer and weaker immune cells as explained in our article on antibody testing. Another key reason is that in many of the nursing homes impacted, infection control was poorly implemented, and was not rigorous enough to protect the occupants. Nursing and care homes were key places for super-spreading events. In these areas, workers were poorly screened for infection and had limited access to Personal Protective Equipment (PPE), and infected patients were readmitted into care homes rather than being isolated. In contrast, those countries with better isolation requirements for patients returning from hospitals have demonstrated better infection control in nursing homes11.
The elderly in nursing homes are more likely to have terminal illnesses or other underlying chronic conditions that make it harder for their bodies to recover. Nursing homes also concentrate the elderly and most vulnerable in one place. It is likely that the particularly frail individuals who live in nursing homes (or at home) died because of the overwhelming nature of the multi-system organ damage that COVID-19 inflicts. In the UK, there are just over 410,000 people living in nursing homes, of whom approximately 25% die every year. For many in this group, COVID-19 may have precipitated the end of life by a few months to a year.
Others were either healthy or adequately treated before contracting COVID-19 but were more susceptible to infection and more likely to develop severe disease and die for reasons that remain as yet unknown. There are likely to be many contributing factors, such as pre-existing conditions, genetic factors and the timing and availability of hospital and emergency care. An increasing number of studies are already identifying changes to the immune response that might make individuals more likely to contract the virus and less likely to fight off the virus12. Despite having relatively poor health, these individuals would not have been expected to die imminently, as we will see in our next section.
Section 3: What are the excess deaths attributable to the virus?
It is ultimately excess mortality that determines both the impact of COVID-19 and the rationale for movement and contact restrictions imposed on individuals and businesses. A better understanding of current and future levels of excess mortality has always been a key focus for life insurers. In this section we explore the different reasons and origins of excess mortality due to COVID-19.
Excess deaths directly attributable to COVID-19
The pandemic has not fully run its course, but the data we already have makes it clear that COVID-19 is not like seasonal flu. A key question is whether the pandemic is materially increasing death rates compared to other diseases. Or is it too early for the available data to yield conclusive results?
It is easy to forget that COVID-19 is the latest in a long line of infectious diseases that threaten us. In response to past pandemics, public health officials have set up early warning systems to focus on unexpected surges in the total number of deaths. Public health observatories around the world provide provisional weekly estimates of the number of death notifications.
These estimates show that in a typical year, we see a sinusoidal wave pattern with more deaths in the winter months. In previous years, the focus has been on influenza and pneumonia and reports of excess winter deaths may be driven by exceptionally cold weather and the effectiveness of that year’s seasonal vaccine. The 2017-18 winter influenza season13 was particularly severe, in part because public health officials were caught by unexpected changes in the dominant circulating flu virus, and that year's vaccine was only 40% effective.
The number of excess deaths therefore provides the first metric that can be used to make comparisons between countries about the impact of COVID-19 on mortality rates. The pattern from the last five years is used by groups such as EuroMOMO (European Mortality Monitoring), as a moving reference point to quantify the level of excess deaths in any given year. The chart below depicts the extra mortality up to week 24 (ending 12 June) of 202014.
A look at England and Wales weekly deaths published by the ONS shows that COVID-19 caused significant excess deaths compared to the average of the last five years. A comparison of excess deaths and COVID-19 deaths shows that despite the fact that the two are highly synchronized, there are many more excess deaths than deaths directly attributed to COVID-19. There can be many reasons for this:
- Deaths that occurred in nursing home or other places that were COVID-19 related but not counted in the COVID-19 death toll.
- Premature deaths indirectly related to COVID-19 because treatments schedules were interrupted due to the pandemic.
- Negative effects of mental health impacts.
The statistics show the profound social impact of COVID-19 with its wider disruption of social, economic and health activities.
The largest impact spans the period when weekly death counts were at their highest, when the virus was spreading most freely, and medical resources were most strained. This can be seen clearly from the EuroMOMO chart presented as Figure 12, which shows that when COVID-19 deaths peak, there are also large number of excess non-COVID-19 deaths, but the excess drops rapidly soon after.
The ONS data depicted in figure 13 similarly shows that England and Wales' excess non-COVID-19 deaths reduced quickly after the peak number of deaths in hospitals and care homes had passed. The weeks ending 15 May and 22 May already had very little excess non-COVID-19 deaths15.
Source: UK Office of National Statistics
The Economist presents data on excess mortality based on several public health databases (see figure 14). For most countries it appears that the peak is well past. We are seeing that the non-COVID-19 deaths were also higher than normal levels during the critical few weeks when the pandemic peaked in these countries.
Source: reproduced from data made available by The Economist
In the US, Brazil and other Latin American countries, we cannot yet say that the big impact will only be over a short and limited time – it is already clear that in the US the first wave of the pandemic will be more protracted than in Europe and Asia.
Can we expect reduced deaths from other causes?
A potential positive side effect of the lockdowns, business closures and other mobility restrictions is a reduced number of violent deaths – particularly from motor vehicle accidents. These accidental deaths are lower in 2020 than in a normal year, with country data showing negative excess deaths below the age of 40. However, the reduction in these deaths is far outweighed by the increased deaths from COVID-19.
Indications are that the hope of reduced deaths from non-COVID-19 causes may not materialize in many major Western countries – or if it does materialize then the offset will be minor. For example, COVID-19 death counts in the US are already far higher than any potential offset by reduced motor vehicle deaths. The total number of US motor vehicular fatalities in an entire year is roughly 33,000, while the US death toll from COVID-19 has already crossed 125,000 with the pandemic still to run its full course.
Even if there were a substantial reduction in motor vehicle accidents during the peak period of restrictions, as restrictions ease, it's likely that the fatality rate will revert to normal levels for the rest of the year. A study on excess deaths in Italy that compares the total number of deaths since lockdown was imposed on 9 March to average mortality in South Italy (which had a low infection rate) showed no reduction in mortality, which implies a negligible effect on other violent deaths16.
The impact of the pandemic on mental health will play out over the longer term as we discussed in this report, as has been reported in other studies17,18. Even, now there are increased reports of depression, anxiety and other disorders, supported by data from the ECLB-COVID-19 Multicentre Study19. Loss of income and lack of certainty about future employment prospects will only intensify and compound these effects. Other effects, such as rising reports of domestic violence are another consequence of a protracted lockdown.
Other factors to consider
On the flipside, there are also benefits experienced from the lockdown and changes in routines. Reduced air pollution, fewer accidents on the road and positive changes to behaviour in the face of a common adversary are just a few. The increased focus on personal hygiene and hand washing will not only slow the spread of COVID-19, but if maintained should also reduce future deaths from seasonal influenza20.
The full picture of all the effects is highly complex with many factors. The challenge is that so far most have focused on limited sources of hard data – namely number of deaths and number of positive tests. However, this is changing as existing datasets are made more widely available, and new datasets are constructed at breakneck speed.
As the medical and scientific community shifts its attention from emergency clinical care to research, our understanding of the implications around excess mortality will deepen.
Section 4: What should life insurers watch for?
As life insurers try to anticipate claims impact, it's not straightforward to draw direct conclusions from the general population experience. The longer term impact of COVID-19 is also of keen interest as this will impact claims on in-force insured portfolios and inform medical underwriting practices.
Life insurance portfolios may deviate from the broader general population along several key attributes that impact the death rates from COVID-19.
As age is the number one driver of the COVID-19 fatality rate, so a portfolio's age mix may imply very different fatality rates. Note that while COVID-19's impact on mortality rates is much more significant at older ages, indications are that there is a meaningful percentage increase in mortality at younger ages too – particularly for individuals with underlying comorbidities. This is an important consideration when reviewing underwriting guidelines for new sales.
There is strong evidence men have a higher fatality rate than women when infected by COVID-19. Portfolios with a significantly higher percentage of men are likely to be worse affected than those more balanced or skewed toward women.
The insured population typically undergoes an underwriting selection process and thereby represents a healthier group on average than the general population, although the health of some policyholders will have deteriorated since policy inception. As comorbidity is an important COVID-19 fatality risk factor, it is reasonable to expect that insured portfolios would experience better mortality rates due to COVID-19 compared to the general population, even if overall mortality is likely to be elevated in the insured population compared to pre-pandemic times. A direct inference on the experience of in-force blocks from population experience is therefore difficult to draw. For new business, the underwriting process also allows insurers the opportunity to assess comorbidities in the context of COVID-19.
Urban vs. suburban and rural
Comparisons of age-adjusted COVID-19 death rates by geographic areas suggests that major metropolitan areas have much higher COVID-death rates than non-metropolitan areas (particularly major international travel hubs), presumably because there is a higher exposure to overseas travellers transmitting the virus and higher population density which would hasten the spread of the virus. Such factors most like drove the concentration of COVID-19 infections in London, New York City and Lombardy in Italy.
While much of Europe and countries of the Far East are well past the peak in cases and deaths, COVID-19 is still running its course in the US, Latin America and South Asian countries so it is still too early to conclude that non-metropolitan areas will always fare better than the big cities.
However, to the best of the available data and experience so far in the pandemic, indications are that portfolios that are more concentrated in large metropolitan areas (e.g., New York, London, Milan) will likely fare worse than portfolios with a less of a focus on urban populations.
Product split and sum-assured profile
Age, gender and location are difficult to generalize across the insurance industry. The product type, sum assured, geographic regions and customer base a company focuses on will influence the company experience and may differ from the industry wide impact.
Years of life lost
Early in the pandemic, death statistics showed that older people and those with multiple comorbidities and underlying conditions were more at risk. Some speculated that the people dying of COVID-19 were nearer the end of their lives, and that the virus accelerated an impending death. The implication being the spike in death rates would later be offset by reduced number of deaths in subsequent months.
For older age groups the prevalence of comorbid conditions in the general population is not too dissimilar to that observed among those dying from COVID-1921, 22. These studies suggest that many of the people who died were not necessarily at the end of their life by estimating "years of life lost". It has been suggested that on average, hospitalised COVID-19 patients who died lost about 12-14 years of life compared to a person with similar comorbidities23. This is not surprising from a medical perspective as the prognosis for many comorbidities is generally quite positive, with only the very oldest seeing a dramatic cut to their lifespans. Underwriting tables imply that people who suffer from conditions like hypertension and diabetes tend to survive for quite a long time.
Source: UK Office of National Statistics and Public Health Scotland
The effects of COVID-19 itself and the associated lockdowns are likely building an excess of morbidity that may take a long time to fully manifest itself. For those who need to develop future models of mortality and morbidity, COVID-19 has imposed itself as a breakwater that will disrupt historical trends and muddy future projections. There will likely be a need to consider these different cohorts in greater granularity going forward, bringing together multi-disciplinary insights and data from electronic health records as we assess the future volatility of our demographic assumptions.
So, what have we learned? In this article we explored the leading factors around COVID-19:
- In the broadest sense, excess mortality remains the most appropriate reflection of COVID-19 deaths.
- The key risk factors are clear: age and comorbidities. So far, the evidence on high blood pressure and smoking as a driver of COVID-19 is inconclusive.
- Data is key: electronic health records are the way forward in building a better understanding of the disease in general, and excess mortality specifically. More research into this area is needed,
- Extra years of life lost are significant and above what was expected. Only the most elderly victims of COVID-19 are not seeing a significant loss.
- Nursing and care homes are vulnerable for a variety of reasons. For everyone's sake, this population must be especially protected.
Each week we gain more insight from a growing collection of accessible data and studies, but the journey into understanding the virus and its impact has only just begun. A cross functional team of Swiss Re experts continue to monitor and analyse this in our collective quest to fully understand this threat and its impact. Our next publication will more deeply explore the long-term impact of COVID-19 on health and longevity based on an analysis of electronic health records. Stay tuned.
Contributing authors: Adam Strange and Dan Ryan, Chief Scientific Officer, COIOS Research
Managing editor: Susan Imler
- FT Visual & Data Journalism team. Coronavirus tracked: has the epidemic peaked near you? 2020 [cited 2020 24/6/20].
- Centers for Disease Control and Prevention (CDC). Guidance for Certifying Deaths Due to Coronavirus Disease 2019 (COVID–19). 2020 [cited 2020 24/06/20].
- Office of National Statistics (ONS). Guidance for doctors completing Medical Certificates of Cause of Death in England and Wales. 2020 [cited 2020 24/06/20].
- Ministry of Public Health - Russian Federation. Профилактика, диагностика и лечение новой коронавирусной инфекции (covid-19). 2020 [cited 2020 24/06/20].
- Politico. Why is Belgium’s death toll so high? 2020 [cited 2020 24/06/20].
- Czaja, C.A., et al., Age-Related Differences in Hospitalization Rates, Clinical Presentation, and Outcomes Among Older Adults Hospitalized With Influenza—U.S. Influenza Hospitalization Surveillance Network (FluSurv-NET). Open Forum Infectious Diseases, 2019. 6(7).
- Guan, W.-j., et al., Comorbidity and its impact on 1,590 patients with COVID-19 in China: A Nationwide Analysis. 2020. Preprint-Preprint.
- Williamson, E., et al., OpenSAFELY: factors associated with COVID-19-related hospital death in the linked electronic health records of 17 million adult NHS patients. medRxiv, 2020: p. 2020.05.06.20092999.
- Changeux, J.-P., et al., A nicotinic hypothesis for Covid-19 with preventive and therapeutic implications. Qeios, 2020.
- Simons, D., et al., Smoking and COVID-19: Rapid evidence review for the Royal College of Physicians, London (UK). Qeios, 2020.
- The Guardian. MPs hear why Hong Kong had no Covid-19 care home deaths. 2020.
- Nguyen, A., et al., Human leukocyte antigen susceptibility map for SARS-CoV-2. medRxiv, 2020: p. 2020.03.22.20040600.
- Nielsen, J., et al., European all-cause excess and influenza-attributable mortality in the 2017/18 season: should the burden of influenza B be reconsidered? Clin Microbiol Infect, 2019. 25(10): p. 1266-1276.
- EuroMOMO. EuroMOMO Bulletin, Week 24, 2020. 2020 [cited 2020 24/06/20].
- Office of National Statistics (ONS). Deaths registered weekly in England and Wales, provisional: week ending 5 June 2020. 2020 [cited 2020 24/06/20].
- Modi, C., et al., How deadly is COVID-19? A rigorous analysis of excess mortality and age-dependent fatality rates in Italy. medRxiv, 2020: p. 2020.04.15.20067074.
- Pfefferbaum, B. and C.S. North, Mental Health and the Covid-19 Pandemic. New England Journal of Medicine, 2020.
- Banerjee, S., et al., Social Isolation as a predictor for mortality: Implications for COVID-19 prognosis. medRxiv, 2020: p. 2020.04.15.20066548.
- Ammar, A., et al., Emotional consequences of COVID-19 home confinement: The ECLB-COVID19 multicenter study. medRxiv, 2020: p. 2020.05.05.20091058.
- Liu, M., et al., Protective Effect of Hand-Washing and Good Hygienic Habits Against Seasonal Influenza: A Case-Control Study. Medicine (Baltimore), 2016. 95(11): p. e3046.
- The Actuary. The co-morbidity question. 2020 [cited 2020 24/06/20].
- Hanlon, P., et al., COVID-19 – exploring the implications of long-term condition type and extent of multimorbidity on years of life lost: a modelling study. Wellcome Open Research, 2020. 5.
- Woolf, S.H., et al., Excess Deaths From COVID-19 and Other Causes. March-April 2020. JAMA, 2020.