Comparison of COVID-19 outcomes among shielded and non-shielded populations: A general population cohort study of 1.3 million

Background Shielding (extended self-isolation) of people judged, a priori, to be at high-risk from COVID-19 has been used by some countries to protect the individuals and reduce demand on health services. It is unclear how well this strategy works in either regard. Methods A general population study was conducted using linked primary care, prescribing, laboratory, hospital and death records up to end of May 2020. Poisson regression models and population attributable fractions were used to compare COVID-19 outcomes by overall risk category, and individual risk criteria: confirmed infection, hospitalisation, intensive care unit (ICU) admission, population mortality and case-fatality. Results Of the 1.3 million population, 32,533 (2.47%) had been advised to shield, a further 347,374 (26.41%) were classified as moderate risk. Testing for COVID-19 was more common in the shielded (6.75%) and moderate (1.99%) than low (0.72%) risk categories. Referent to low-risk, the shielded group had higher risk of confirmed infection (RR 7.91, 95% 7.01-8.92), case-fatality (RR 5.19, 95% CI 4.12-6.53) and population mortality (RR 48.64, 95% 37.23-63.56). The moderate risk had intermediate risk of confirmed infection (RR 4.11, 95% CI 3.82-4.42) and population mortality (RR 26.10, 95% CI 20.89-32.60), but had comparable case-fatality (RR 5.13, 95% CI 4.24-6.21) to the shielded, and accounted for a higher proportion of deaths (PAF 75.27% vs 13.38%). Age [≥]70 years made the largest contribution to deaths (49.53%) and was associated with an 8-fold risk of infection, 7-fold case-fatality and 74-fold mortality. Conclusions Shielding has not been effective at preventing deaths in those with highest risk. To be effective as a population strategy, shielding criteria would need to be widely expanded to include other criteria, such as the elderly.

Introduction 1 2 Early in the COVID-19 pandemic, a major concern was that the demand on health services 3 would exceed capacity in terms of hospitalisations, intensive care unit (ICU) admissions, and 4 ventilation 1 . It was assumed that sub-groups of the population would have worse prognosis 5 and, therefore, contribute disproportionately to adverse outcomes and healthcare demands. 6 7 Asian countries generally relied on population-wide strategies 2 . Early, widespread 'test, 8 trace, isolate' strategies were made possible by higher testing capacity and greater 9 willingness to monitor and enforce compliance. In contrast, Europe and the USA adopted a 10 two-pronged approach; 2 general population interventions, such as physical distancing and 11 hand hygiene, designed to reduce transmission in the population as a whole, supplemented by 12 shielding of those assumed to be at higher risk. Notably, Sweden, an outlier in not applying 13 lock-down, nonetheless mandated shielding 3 . 14 15 In the UK, a Vulnerable Patient List (Supplementary Table 1) 6 was produced comprising two 16 categories labelled high risk, highest risk or clinically extremely vulnerable and moderate 17 risk, at risk or clinically vulnerable by various UK organisations. In this manuscript, they are 18 referred to as shielded and moderate risk respectively, with the remaining population labelled 19 low-risk. In the UK, the shielded group received individual letters strongly recommending 20 they self-isolate over a protracted period -not leaving their homes and avoiding non-essential 21 contact with household members -and were provided with support at home such as delivery 22 of food packages. The moderate risk category was simply advised to be vigilant in adhering 23 to general advice. 24 25 . 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 18, 2020. ;https://doi.org/10.1101https://doi.org/10. /2020 The category definitions were based largely on expert opinion informed by our understanding 1 of previous viruses and the need for better definitions has been highlighted 7 . Studies are 2 emerging of the risk factors associated with COVID-19 outcomes. Among two million UK 3 community-based app users self-reported heart disease, kidney disease, lung disease, diabetes 4 and obesity were associated with self-reported hospital admission and respiratory support for 5  . Similarly, linkage of family practitioner records of 17 million people in 6 England reported a wide range of long-term conditions associated with in-hospital death from 7 COVID-19 including: respiratory, heart, liver and kidney disease, diabetes, cancers, stroke 8 and organ transplantation 9 . Unfortunately, the investigators did not have access to deaths in 9 the community. COVID-19 risk scores are being developed in an attempt to improve 10 identification of high risk individuals who could be advised to shield 10 but attempts to 11 investigate the potential contribution of a shielding strategy to population-level outcomes and 12 healthcare demands have so far been limited to mathematical modelling 11-19 . 13 14 The aims of this study were to compare those classified, a priori, as high risk (and therefore 15 advised to shield) and those classified as moderate and low-risk, in terms of their individual 16 risk of COVID-19 infection and outcomes and the extent to which they accounted for 17   Overall, 15,865 (1.21%) people were tested for COVID-19. The likelihood of being tested 11 increased with age, was higher in women and the moderate-risk category and highest in the 12 shielded group (Table 1). Overall, 3,348 (0.25%) people had confirmed COVID-19 infection. 13 The likelihood of laboratory-confirmed COVID-19 infection followed similar patterns as 14 testing. It increased with age, was higher in women, was highest in the shielded group and 15 lowest in the low-risk category (Table 2). After adjustment for sex and deprivation quintile, 16 the risk of laboratory-confirmed infection remained higher in the moderate-risk category and 17 highest in the shielded group (Table 3).  18   19 Overall, 1,661 people were hospitalised for COVID-19. Within the general population, 20 hospitalisations increased with age but were comparable between men and women (Table 2). 21 Hospitalisations were more common in the moderate-risk category and most common in the 22 shielded group (Table 2), remaining so after adjustment for sex and deprivation (Table 3). 23 Overall, 122 people were admitted to ICU wards for COVID-19. ICU admissions were 24 significantly more common among people aged 45-64 years of age than among older people 25 . 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 18, 2020. ;https://doi.org/10.1101https://doi.org/10. /2020 (Table 2). Compared with the low-risk category, the shielded group were 18 times more 1 likely to be hospitalised but only 4 times more likely to be admitted to ICU (Table 3).  2 Overall, 1,027 (0.08%) people died from COVID-19. Within the general population, 3 mortality increased with age but was similar in men and women (Table 2). Population 4 mortality was higher in the moderate-risk category and highest in the shielded group (Table  5 2) and remained so after adjustment for sex and deprivation (Table 3). 6 7 Among the sub-group with laboratory-confirmed (test-positive) COVID-19 infection, 1,661 8 (49.6%) were hospitalised. Hospitalisations increased with age but were comparable between 9 men and women ( Table 4). The moderate-risk category was more likely to be hospitalised 10 and the shielded group most likely (Table 4), remaining so after adjustment for age and 11 deprivation (Table 5). Among those with laboratory-confirmed infection, ICU admissions 12 were more common in men and more common in people aged 45-64 years than those older 13 (Table 4). Low-risk cases were more likely to be admitted to ICU than the moderate-risk and 14 shielded groups (Tables 4 & 5). Among the sub-group with clinically-confirmed (test-15 positive or COVID-19 related death) COVID-19 infection, 1,027 (26.70%) died (Table 4). 16 Case-fatality increased by age and was higher in men than women. It was lowest in the low-17 risk category but not significantly different between the moderate-risk and shielded groups 18 (RR shielded/moderate [95% CI] 1.12 [0.96-1.31], p=0.14) ( 28.8% of the population would have had to receive the current level of shielding including 25 . 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 18, 2020. ; https://doi.org/10. 1101/2020 those with five criteria classified as moderate-risk at the time of the study (Supplementary 1 Figure 1). 2 3 Individual risk criteria 4 5 Due to insufficient numbers, the individual risk criteria models could not be run for pregnant 6 women with severe heart disease or for COVID-19 related ICU admission in the shielded 7 category. All the remaining individual risk criteria were associated with higher likelihood of 8 being tested for COVID-19 (Table 1), laboratory-confirmed infection (Table 2), 9 hospitalisation, population mortality (Table 3) and case-fatality (Table 5) independent of sex 10 and deprivation. Among the moderate-risk category criteria, age ≥70 years and weakened 11 immune system had risks of population mortality (Table 3) and case-fatality (Table 5) at least 12 as high as the overall shielded group. Apart from the 0.13% of people with relevant rare 13 diseases or inborn errors of metabolism and 1.78% on renal dialysis, the strongest 14 associations were observed for those aged ≥70 years who were eight times as likely to have 15 confirmed infection (Table 3); seven times as likely to die following confirmed infection 16 (Table 5); and 74 times as likely to die overall (  Table 2). Among those hospitalised for COVID-19, the likelihood of ICU 20 admission was significantly lower for all individual risk criteria in the moderate-risk 21 category, other than diabetes (Table 5). In particular, hospitalised patients ≥70 years of age 22 were 14 times less likely to be admitted to ICU than low-risk hospitalised patients (Table 5). 23 24 Discussion 25 . 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 18, 2020. ;

1
The 2.03% of people advised to shield were, nonetheless, eight times more likely to have 2 confirmed infections than the low-risk category, five times more likely to die following 3 confirmed infection and 49 times more likely to die from COVID-19 overall. Whilst selective 4 testing might explain the first outcome, it does not explain higher overall mortality which 5 suggests that the shielding strategy was not as effective as was hoped. 6 7 One quarter of the population were classified as moderate-risk and not advised to shield. 8 Nonetheless, they were four times more likely to have confirmed infections than the low-risk 9 category, five times more likely to die following confirmed infection and 25 times more 10 likely to die overall, suggesting that the shielding criteria should be expanded. In particular, 11 older age needs to be considered since the elderly are both at high individual risk and 12 contribute significantly to population burden due to their relatively high numbers. 13

14
In spite of people in the shielded and moderate-risk categories having poorer prognosis, they 15 were less likely to be admitted to ICU following hospitalisation for COVID-19, especially 16 patients ≥70 years. This finding reinforces the importance of protection in those with the 17 worst prognosis. 18

19
Our finding that 26.85% of people satisfy moderate-risk criteria is consistent with limited 20 existing evidence. A study linking English primary and secondary care records on 3.9 million 21 people reported that 20% of population satisfied similar criteria 20 . Similarly, analysis of the 22 Global Burden of Diseases Study estimated that 22% of the global population are at increased 23 risk of severe COVID-19 disease 21 . A USA study using data from the Behavioral Risk Factor 24 Surveillance System reported that 45.4% of 444,649 adults had one or more of a longer list of 25 . 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 18, 2020. ; morbidities that may be associated with higher risk from COVID-19 22 . Another USA study 1 estimated that 14.2% of participants in the National Health Interview Survey had more than 2 two-fold risk and 1.6% had more than 10-fold risk 23 . 3   4 The evidence on COVID-19 related complications among those classified as high risk, and 5 therefore advised to shield, has mainly come from case series and expert opinion. Case series 6 found higher COVID-19 related complications among organ transplant recipients 24,25 , 7 patients receiving chemotherapy, radiotherapy or immunotherapy for cancer 26,27 , and patients 8 with haematological cancers 28 . Systematic review suggested higher COVID-19 complication 9 risk among COPD patients, but the effect of COPD severity was not investigated 29 . Patients 10 with cystic fibrosis and sickle cell disease were classified as high risk based on expert 11 opinion 30,31 . While pregnant women with COVID-19 were found to have higher risk of poor 12 maternal and perinatal outcomes 32,33 , outcomes were not investigated specifically for 13 pregnant women with heart disease. There was no evidence of worse COVID-19 related 14 complications among patients on immunosuppressants 34 . A large community study in 15 England found strong association between severe asthma (hazard ratio 1.25) and COVID-19 16 related mortality but did not investigate the risk of COVID-19 infection or hospitalisation 9 . 17

18
In common with previous studies, we demonstrated that age was a major individual-level risk 19 factor for death. Additionally, we showed it is important at the population level with 49.55% 20 of deaths attributable to age ≥70 years. The higher mortality in the elderly was mediated in 21 part by higher case-fatality but they also had a higher incidence of infection, possibly due to 22 transmission within care homes. Lower ICU admissions following hospitalisation for 23 COVID-19 may have contributed to their higher case-fatality. Previous studies have reported 24 that men are at higher risk of  Our study demonstrated they are less likely to be 25 . 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 18, 2020.
tested for COVID-19, have confirmed infection, and be hospitalised. They have comparable 1 overall mortality from COVID-19, due to their lower incidence, but their case-fatality is 2 higher. 3 4 This study adds to the existing evidence of the possible effectiveness of a shielding strategy 5 which is currently limited to mathematical modelling of population effects based on 6 assumptions 11-19 . Ours was a large-scale, unselected general population study. The data 7 cover a period when shielding was in place. Linkage of family practitioner, laboratory, 8 hospital and death data enabled us to examine a range of COVID-19 outcomes and study a 9 range of exposure variables including the overall risk categories and their individual criteria. 10 The datasets were linked using exact, rather than probabilistic, matching. We were able to 11 adjust for potential sociodemographic confounders. The exposure data were collected prior to 12 the outcomes occurring avoiding potential reverse causation and recall or recording bias. Our 13 analysis of potential risk factors was restricted to those used as criteria for shielding and 14 moderate-risk at the time of the study. The shielding and moderate-risk criteria were correct 15 at the time of extracting data but may be revised over time. 16 17 Our findings suggest that our attempts to shield those at highest risk have not been as 18 successful as hoped, with those advised to shield experiencing higher rates of infection and 19 death. Since this group was also less likely to be admitted to ICU, protecting them from 20 infection is essential. For shielding to be effective as a population level strategy, the current 21 criteria would need to be expanded since three-quarters of deaths were associated with 22 moderate-risk criteria for which shielding has not hitherto been recommended. In our study, 23 more than one-quarter of the general population would have needed to be effectively shielded 24 to prevent over 80% of deaths. Since this is unlikely to be acceptable at a time when 25 . 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 18, 2020. ; governments are under pressure to avoid further lock-downs, shielding is probably best 1 viewed as an individual-level intervention to be used alongside other population-wide 2 interventions such as physical distancing, face coverings and hand hygiene. The CHI register provided sociodemographic information (age, sex, area socioeconomic 16 deprivation). Deprivation was measured using the Scottish Index of Multiple Deprivation 17 (SIMD), derived from seven domains -income, education, health, employment, crime, 18 housing, and access to services -and categorised into general population quintiles. ECOSS 19 collects laboratory data on infectious diseases, including test date and result. Albasoft 20 software extract data from the family practitioner electronic health record systems EMIS and 21 Vision, and PIS collects data on medications prescribed by family practitioners. SERPR 22 records data on renal replacement therapy and transplantation. RAPID collects real-time data 23 on hospitalisation, including dates of admission and discharge, and type of ward, and the 24 Scottish Morbidity Record 01 (SMR01) subsequently records the relevant disease codes. 25 . 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 18, 2020. ; Death certificates provide the date and cause of all deaths, whether in-hospital or in the 1 community. Follow-up data were available until the end of May 2020, before the shielding 2 recommendation was lifted. 3 4 Supplementary Table 1 lists the criteria for the shielded and medium risk categories applied 5 at the time of data extraction. All remaining patients are categorised as low-risk. The Scottish 6 list of high-risk individuals is compiled centrally, and regularly updated, using family 7 practitioner, hospital admission, disease registry and medication data. Family practitioners 8 check the completeness and accuracy of the list before letters, recommending shielding, are 9 sent to patients. The NHS GGC Shielding List we used contains the validated data including 10 the criterion satisfied. We ascertained moderate risk individuals using Albasoft extraction of 11 EMIS and Vision data, and PIS data. 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 18, 2020. ; was defined as an SMR01 hospitalisation record with an ICD code U07.1 or U07.2 or, for 1 more recent admissions, a RAPID hospitalisation record plus positive PCR test taken 2 between two weeks before and two days after hospitalisation. ICU admission during such 3 hospitalisations was assumed to be COVID-related. 4 5 Sociodemographic characteristics were compared by risk category using chi-square tests. 6 Poisson regression models with robust standard errors were used to compare risk ratios (RR) 7 for the shielded and moderate-risk categories referent to the low-risk category. The models 8 were run univariately; then adjusted for sex and SIMD quintile as potential confounders. Age 9 was not included as a covariate because it was a moderate-risk criterion. The models were re-10 run using the individual criteria for the shielded and moderate-risk categories as the exposure 11 variables, referent to the low-risk category. 12 13 Population attributable fractions (PAFs) were calculated, from prevalence and adjusted RR,14 to determine the proportion of each outcome that could be attributed to being shielded and 15 moderate-risk, as well as the proportion due to each individual criterion. The PAFs of 16 individual criteria were proportionally calibrated so that their sum equated to the overall PAF 17 of the relevant risk category. PAF confidence intervals were estimated using bootstrapping (x 18 1000). 19 20 Ethical approvals 21

22
The study was approved by the NHS GGC Primary Care Information Sharing Group and the 23 NHS GGC Local Privacy Advisory Committee (Reference GSH/20RM005) and was covered 24 . 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 18, 2020. ; by the generic Safe Haven Research Ethics Committee approval 1 (GSH20RM005_COVID_Community). 2 3 4 . 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 18, 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. wrote the first draft of the manuscript. All other authors interpreted the data and critically 10 revised the manuscript. All authors approved the final submitted version of the manuscript. 11 BJ, FH, and JP serve as the guarantor of the manuscript and accepts full responsibility for the 12 work and/or the conduct of the study, had access to the data, and controlled the decision to 13 publish. The corresponding author attests that all listed authors meet authorship criteria and 14 that no others meeting the criteria have been omitted. 15 16

Conflicts of interest 17
All authors have completed the ICMJE uniform disclosure form at 18 www.icmje.org/coi_disclosure.pdf and declare: no support from any organisation for the 19 submitted work; no financial relationships with any organisations that might have an interest 20 in the submitted work in the previous three years; no other relationships or activities that 21 could appear to have influenced the submitted work. 22

23
Funding 24 No external funding sources. 25 . 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 18, 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 18, 2020. ;