key: cord-0797733-iq8npra7
authors: Wu, Jianhua; Mafham, Marion; Mamas, Mamas; Rashid, Muhammad; Kontopantelis, Evangelos; Deanfield, John E.; de Belder, Mark A.; Gale, Chris P.
title: Place and Underlying Cause of Death During the COVID-19 Pandemic: Retrospective Cohort Study of 3.5 Million Deaths in England and Wales, 2014 to 2020
date: 2021-02-16
journal: Mayo Clin Proc
DOI: 10.1016/j.mayocp.2021.02.007
sha: 939390b331c04adfa3e20373b3e68e2510ebec18
doc_id: 797733
cord_uid: iq8npra7

Objective To describe the place and cause of death during the COVID-19 pandemic to assess its impact on excess mortality. Methods This national death registry included all adult (aged≥18 years) deaths in England and Wales between 1st January 2014 and 30th June 2020. Daily deaths during COVID-19 pandemic were compared against the expected daily deaths estimated using Farrington surveillance algorithm for daily historical data between 2014 and 2020, by place and cause of death. Results Between 2nd March and 30th June 2020, there was an excess mortality of 57860 (a proportional increase of 35%) compared with the expected deaths, of which 50603 (87%) were COVID-19 related. At home, only 14% of 16190 excess deaths were related to COVID-19, with 5 963 deaths due to cancer and 2485 deaths due to cardiac disease, few of which involved COVID-19. In care homes or hospices, 61% of the 25611 excess deaths related to COVID-19, 5539 of which were due to respiratory disease and most of these (4315 deaths) involved COVID-19. In hospital, there were 16174 fewer deaths than expected which did not involve COVID-19, with 4088 fewer deaths due to cancer and 1398 fewer deaths due to cardiac disease than expected. Conclusion The COVID-19 pandemic has resulted a large excess of deaths in care homes, which were poorly characterised and likely the result of undiagnosed COVID-19. There was a smaller, but important and ongoing excess in deaths at home, particularly from cancer and cardiac disease, suggesting public avoidance of hospital care for non-COVID-19 conditions.

Globally, as of August 6 th 2020, coronavirus disease (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) accounted for 702,642 deaths. 1 In the UK, this has been evidenced by an abrupt increase in the number of deaths above that expected for the historical average. 2 3 However, the basis for this excess mortality is poorly defined with limited information about the causes of death during the pandemic. This is important because whilst the SARS-CoV-2 virus is known to result in an acute respiratory syndrome for whom the highest risk of death is among the elderly and those with pre-existing medical conditions, 4 5 but also people may have died from other causes due to restructuring of medical services during this period or avoidance of health care settings.

Moreover, we and others have reported a dramatic decline in admissions to hospitals with medical emergencies. [6] [7] [8] [9] [10] [11] [12] Consequent to delays to seeking help for life-threatening illnesses, many deaths are likely to have occurred in the community. Equally, there has been an increase in the number of deaths among those living in care homes. [13] [14] [15] Here, the vulnerability of residents to infection as well as changes to health behaviour may have played a role in their demise. Should there have been a displacement in the place of death as a result of the pandemic, then lessons may be learnt to be better prepared in case of a second increase in COVID-19 cases. 16 Thus, a systematic characterisation of the cause and place of death associated with the COVID-19 pandemic and how this changed compared to pre-pandemic era is necessary and may offer insights into the susceptibility of the public to the virus as well as the impact of health and public guidance aimed at reducing the spread of the virus. We report the underlying causes of all adult deaths during the COVID-19 pandemic in England and Wales, the location of deaths (e.g. hospitals, home or care homes) and their relation to the COVID- 19 infection. This information is vital for the understanding of healthcare policy during the emergence from lockdown and to assist Governments around the world reorganise healthcare services now that incident rates of COVID-19 are in decline and social isolation policies are relaxed.

The analytical cohort included all certified and registered deaths in England and Wales ≥18 years of age, between 1 st January 2014 and 30 th June 2020 recorded in the Civil Registration Deaths Data of the Office for National Statistics (ONS) of England and Wales. 17 

The primary analysis was based upon the ICD-10 code corresponding to the underlying cause of death registered, as stated on the medical certificate of cause of death (MCCD). The MCCD is completed by the doctor who attended the deceased during their last illness within 5 days unless there is to be a coroner's post-mortem or an inquest. Underlying causes of death were then categorised as detailed in the Office for National Statistics (ONS) short list for causes of death, 16 with additional aggregation of causes for cancer, cardiac and respiratory pathologies (Supplemental table 1 ). ICD-10 codes 'U071' (confirmed) and 'U072' (suspected) listed in any position on the MCCD were used to identify whether a death involved COVID-19 infection.

For the purposes of this investigation, the ICD-10 code corresponding to the underlying cause of death was used. Pre-existing conditions or other diseases that contributed to, but did not directly lead to death were excluded from the analyses. We found that about one in ten MCCDs reported COVID-19 as the underlying cause of death, and for such cases we used the pathology named as directly leading to death to select the underlying cause, which is also the approach taken by the ONS. 16 The place of death as recorded on the MCCD was classified as home, care home or hospice, and hospital.

Baseline characteristics were described using numbers and percentages for categorical data.

Data were stratified by COVID-19 status (suspected or confirmed COVID-19 recorded, not mentioned), age band (<50, 50-59, 60-69, 70-79, 80+ years)), sex and place of death. The number of daily deaths was presented using a 7-day simple moving average (the mean number of daily deaths for that day and the preceding 6 days) from 1 st February 2020 up to and including 30 th June 2020, adjusted for seasonality.

The expected daily deaths from 1 st February 2020 up to and including 30 th June 2020 were estimated using Farrington surveillance algorithm for daily historical data between 2014 and 2020. 18 The algorithm uses overdispersed Poisson generalised linear models with cubic spline terms to model trends in counts of daily death, accounting for seasonality. The number of non-COVID-19 deaths each day from 1 st February 2020 were subtracted from the estimated expected daily deaths in the same time period to create a zero historical baseline. Deaths above this baseline may be interpreted as excess mortality, which were calculated as the difference between the observed daily deaths and the expected daily deaths. The proportion of excess deaths was estimated by dividing the excess mortality by the sum of the expected deaths between 2 nd March 2020 and 30 th June 2020. To compare the impact on mortality from COVID-19 pandemic and flu epidemic, information about influenza and pneumonia (ICD-10 code: J09-J18) was extracted for the two months either side of the date of the peak death rate each year between 2015 and 2020. The averaged daily deaths over 6 years in the 'influenza season' were compared with the averaged daily deaths in the trough period (two months before and after 1 st July each year). All tests were two-sided and statistical significance considered as P<0.05. Statistical analyses were performed in R version 4.0.0.

Ethical approval was not required as this study used fully anonymised routinely collected civil registration deaths data. The data analysis was conducted through remote access to NHS Digital Data Science Server. The peak in deaths during the COVID-19 pandemic was much greater than for any of the influenza seasonal peaks in the years between 2015 and 2020 ( Figure 1 ). Following the first COVID-19 death on the 2 nd March 2020 to 30 th June 2020, there was an excess mortality of 57,860 (a proportional increase of 35%) compared with the expected daily deaths estimated using Farrington surveillance algorithm for daily historical data between 2014 and 2020 ( Figure 2 , Table 2 ). The number of excess deaths was higher for men than women (29,956; a proportional increase of 36% vs. 27,839; a proportional increase of 33%), and was the highest among people aged over 80 years (37,244; a proportional increase of 40%) ( Table 2 ). London had the largest absolute number of excess deaths (9,001 deaths; a proportional increase of 55%). Almost half the excess deaths occurred in care homes and hospices (25,611 deaths) where deaths were 55% higher than expected. One quarter of the excess deaths occurred in hospital (15,938 deaths; a proportional increase of 21%) with the remainder occurring at home (16, 190 deaths; a proportional increase of 39%) ( Table 2) .

Among the excess deaths, 50,603 (87%) were COVID-19 related ( Figure 2 , Table 2 ). There Table 2 ). There were smaller numbers of excess deaths due to cardiac disease (2,225 deaths [including 1,050 related to COVID]; a proportional increase of 9%) and cancer (687 deaths; a proportional increase of 1% [1,127 related to COVID, but 440 fewer cancer deaths than expected after subtracting the COVID related deaths]) ( Figure 3 , Table 2 ).

Between 2 nd March 2020 and the 30 th June 2020, there were 50,603 COVID related deaths, one quarter of the deaths occurring during this period (Table 1) . About two thirds of the COVID-19 related deaths occurred in hospital, about one third occurred in care homes and hospices, and less than 5% occurred at home (Table 1) . Fifty-five percent of COVID-19 related J o u r n a l P r e -p r o o f deaths occurred in men and about two-thirds occurred in those aged 80 years or older with less than 2% occurring in those younger than 50 years. In around half of the COVID-19 related deaths, the condition recorded as leading directly to death was recorded as COVID-19 (21,935 deaths) or an ill-defined cause of death (2,718 deaths) with a further 39% of COVID-19 related deaths (19,681 deaths) in which a respiratory disease lead directly to death (639 involving asthma, COPD or another chronic lung disease and 18,264 involving respiratory failure or respiratory infection) ( Table 1) (Table 2 ). There were 5,963 excess deaths at home due to cancer and 2,485 excess deaths from cardiac disease, very few of which involved COVID-19 (Table 3) .

There were 25,611 excess deaths in care homes and hospices, of which about two thirds (15,966 deaths) were related to COVID-19 ( Figure 2 ). Of the excess deaths in care homes and hospices, 5,539 were due to respiratory disease and most of these (4,315 deaths) involved COVID-19. There were 6,267 excess deaths due to dementia, and 2,358 excess deaths from ill-defined conditions, in care homes or hospices, of which only 783 and 1,003 respectively were recorded as COVID-19 related. There were 1,495 fewer deaths in care homes and hospices due to cancer than expected and 1,211 excess deaths in care homes due to cardiac disease (Table 3 ).

In hospital after 2 March 2020, there were 32,112 COVID-19 related deaths, but 16,174 fewer deaths than expected which did not involve COVID-19, meaning that the total number of excess deaths in hospital was 15,938 ( Figure 2 , Table 3 ). There were 4,088 fewer deaths in hospital due to cancer and 1,398 fewer deaths in hospital due to cardiac disease than expected (Table 3) . 

The numbers of excess deaths from influenza and pneumonia in previous years occurred with the greatest magnitude in hospital, an increase of 28% during influenza epidemics compared with a 21% increase during COVID-19 pandemic (Supplemental Table 2 ). Of 40 223 excess deaths during influenza epidemics, 12 929 (32%) excess deaths were due to respiratory causes (Supplement Figure 2 and Supplement Table 2 ). In comparison, 87% of excess deaths were recorded as COVID-19 related during the pandemic ( Table 2 ).

We report, for the first time, in a complete analysis of all adult deaths in England and Wales, the extent, site and underlying causes of the increased mortality during the COVID-19 pandemic compared with previous years. This shows that, most of the 58,000 excess deaths during this period involved COVID-19 and, in most of these, COVID-19 appeared to be the direct cause of death. However, there was a substantial increase in the absolute numbers of deaths occurring at home, especially from cancer and cardiac disease, while deaths from these causes in hospital were lower than expected. In care homes and hospices there was an abrupt increase in the absolute numbers of deaths from dementia, Alzheimer's disease and ill-defined causes, in addition to COVID-19 related deaths.

We found evidence for the displacement in the place of death from hospital to community setting during the pandemic. Historically in England and Wales, almost half of all adult deaths J o u r n a l P r e -p r o o f occur in hospital, but during the pandemic only a quarter did. During the pandemic about 26,000 excess deaths (almost half of the total excess deaths) occurred in care homes and hospices. Residents of care homes frequently died from respiratory disease (mostly involving COVID-19), but also from 'symptoms and signs of ill-defined conditions' (which typically indicates old age and frailty 18 ) and dementia and Alzheimer's disease. While it is not possible to be certain about the factors leading to the substantial excess in deaths from these less well-defined causes, undiagnosed COVID-19 is likely.

The efficient person-to-person transmission of the SARS-2-CoV virus, 19 its asymptomatic incubation and transmission period 20 21 and propensity to death in the elderly and co-morbid will have been major contributing factors to the excess mortality in care homes. In March 2020, a report detailing an outbreak of COVID-19 infection at a long-term care centre that was associated with high mortality rates recommended proactive steps by such places to identify and exclude potentially infected staff and visitors and implement infection prevention and control measures to reduce the introduction of the virus to residents. 22 Yet, in the UK, patients were discharged from hospitals to care homes without information of their infective status, where the virus could easily spread 13 and actions to effectively reduce the spread of the virus in social care were not implemented early in the pandemic. 23 Early in the pandemic testing of suspected cases was only available in hospital while routine testing of staff and residents in care homes was not implemented until May 2020 24 potentially leading to under-diagnosis of COVID-19. 25 Additionally, it is possible that care home residents who became unwell during the pandemic were not referred or decided not to go to hospital for fear of becoming infected -a notion which aligns to the substantial reduction in hospital attendances for medical emergencies following the UK lockdown. 6 At home, the largest number of excess deaths were from cardiac disease and cancer and few deaths involved COVID-19. This may be explained by infection serving as a trigger to acute decompensation of a pre-existing disease 27 (and which may be under-reported due to nonsystematic testing 28 ) but is more likely to be related to a reluctance by the public to attend hospital when unwell because of fear of becoming infected with the SARS-CoV-2 virus.

Another possible explanation is that early in the course of the pandemic hospitals prepared for a potential mass influx of patients by expeditious hospital discharge of in-patients to the community, which may have resulted in a number of deaths.

This study only assessed the excess deaths during the first wave of the COVID-19 pandemic.

As the infection was better controlled and hospital pressure eased during the Summer, the excess deaths were less or returned to normal and shown by a plateauing of the excess curves towards July 2020. As Winter approached and infection rates increased during the second wave of the pandemic, there was similar pattern of excess deaths. 14 Whilst previous reports have described an elevated risk of death among the elderly and people with cardiovascular disease during the COVID-19 pandemic, none have characterised the underlying causes and place of death in an unselected national cohort. 5 J o u r n a l P r e -p r o o f The number of daily deaths is presented using a 7-day simple moving average (indicating the mean number of daily deaths for that day and the preceding 6 days). The blue line represents daily deaths in all places, the red line represents daily deaths in hospital, the purple line represents daily deaths at care homes and hospices and the green line represents daily deaths at home. The number of daily deaths is presented using a 7-day simple moving average (indicating the mean number of daily deaths for that day and the preceding 6 days) from 1st February 2020 up to and including 30th June 2020, adjusted for seasonality. The number of non-COVID-19 excess deaths each day from 1st February 2020 were subtracted from the expected daily death estimated using Farrington surveillance algorithm in the same time period. The green line is a zero historical baseline. The red line represents daily COVID-19 deaths from 2nd March to 30th June 2020, the blue line represents daily non-COVID-19 deaths from 2nd March to 30th June 2020. The number of daily deaths is presented using a 7-day simple moving average (indicating the mean number of daily deaths for that day and the preceding 6 days) from 1st February 2020 up to and including 30th June 2020, adjusted for seasonality. The number of non-COVID-19 excess deaths each day from 1st February 2020 were subtracted from the expected daily death estimated using Farrington surveillance algorithm in the same time period. The green line is a zero historical baseline. The red line represents daily COVID-19 deaths from 2nd March to 30th June 2020, the blue line represents daily non-COVID-19 deaths from 2nd March to 30th June 2020.

Place Vector-borne diseases and rabies A20, A44, A75-A79, A82-A84, A85.2, A90-A98, B50-B57 29 Supplemental Figure 1 Time series of daily deaths by underlying cause of death and place of death.

The number of daily deaths is presented using a 7-day simple moving average (indicating the mean number of daily deaths for that day and the preceding 6 days) from 1st February 2020 up to and including 30th June 2020, adjusted for seasonality. The number of non-COVID-19 excess deaths each day from 1st February 2020 were subtracted from the expected daily death estimated using Farrington surveillance algorithm in the same time period. The green line is a zero historical baseline. The red line represents daily deaths in hospital, the purple line represents daily deaths at care homes and hospices and the blue line represents daily deaths at home.

J o u r n a l P r e -p r o o f

COVID-19) dashboard. Geneva: World Health Organization

?areas=usa&areas=gbr&areas=bra&areasRegional=usny&areasRegional= usca&areasRegional=usfl&areasRegional=ustx&cumulative=0&logScale=1&perM illion=0&values=deaths Accessed 6th

Deaths involving COVID-19, England and Wales: deaths occurring in

Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study

Estimating excess 1-year mortality associated with the COVID-19 pandemic according to underlying conditions and age: a population-based cohort study

The Covid-19 Pandemic and the Incidence of Acute Myocardial Infarction

Collateral Effect of Covid-19 on Stroke Evaluation in the United States

Emergency hospital admissions and interventional treatments for heart failure and cardiac arrhythmias in Germany during the Covid-19 outbreak Insights from the German-wide Helios hospital network

COVID-19 pandemic and admission rates for and management of acute coronary syndromes in England

Patient response, treatments and mortality for acute myocardial infarction during the COVID-19 pandemic

Second Decline in Admissions With Heart Failure and Myocardial Infarction During the COVID-19 Pandemic

Let's be open and honest about covid-19 deaths in care homes

Deaths registered weekly in England and Wales, provisional: week ending 17

Mortality Associated with COVID-19 Outbreaks in Care Homes: Early International Evidence

Preventing more deaths in care homes in a second pandemic surge

User guide to mortality statistics

An improved algorithm for outbreak detection in multiple surveillance systems

SARS-CoV-2: virus dynamics and host response

Presymptomatic SARS-CoV-2 Infections and Transmission in a Skilled Nursing Facility

Suppression of a SARS-CoV-2 outbreak in the Italian municipality of Vo'

Epidemiology of Covid-19 in a Long-Term Care Facility in King County, Washington

Staggering number" of extra deaths in community is not explained by covid-19

Tackling UK's mortality problem: covid-19 and other causes

Impact of COVID-19 on cardiac procedure activity in England and associated 30-day mortality

COVID-19 and the cardiovascular system

Covid-19: Lack of capacity led to halting of community testing in March, admits deputy chief medical officer

The authors acknowledge Chris Roebuck, Tom Denwood, Tony Burton and CourtneyStephenson and data support staff at the NHS digital for providing and creating the secure environment for data hosting and for analytical support. The authors acknowledge Ben Humberstone at the Office for National Statistics for providing the civil registration deaths in England and Wales and taking responsibility for the integrity of these data. The authors acknowledge Prof Colin Baigent from University of Oxford for constructive suggestion.The programme was endorsed the British Heart Foundation collaborative, which also includes Health Data Research UK, HSC Public Health Agency, National Institute for Cardiovascular J o u r n a l P r e -p r o o f

Cirrhosis and other diseases of liver 213 (+8%) 411 (+16%) a The numbers do not add up to the total deaths due to missingness (1.9%). b Excess deaths in subgroups may not add up to total excess deaths due to rounding errors when comparing to the historical baseline data.