Previous studies suggested that the prevalence of chronic respiratory disease in patients hospitalised with COVID-19 was lower than its prevalence in the general population. The aim of this study was to assess whether chronic lung disease or use of inhaled corticosteroids (ICS) affects the risk of contracting severe COVID-19.
In this population cohort study, records from 1205 general practices in England that contribute to the QResearch database were linked to Public Health England’s database of SARS-CoV-2 testing and English hospital admissions, intensive care unit (ICU) admissions, and deaths for COVID-19. All patients aged 20 years and older who were registered with one of the 1205 general practices on Jan 24, 2020, were included in this study. With Cox regression, we examined the risks of COVID-19-related hospitalisation, admission to ICU, and death in relation to respiratory disease and use of ICS, adjusting for demographic and socioeconomic status and comorbidities associated with severe COVID-19.
Between Jan 24 and April 30, 2020, 8 256 161 people were included in the cohort and observed, of whom 14 479 (0·2%) were admitted to hospital with COVID-19, 1542 (<0·1%) were admitted to ICU, and 5956 (0·1%) died. People with some respiratory diseases were at an increased risk of hospitalisation (chronic obstructive pulmonary disease [COPD] hazard ratio [HR] 1·54 [95% CI 1·45–1·63], asthma 1·18 [1·13–1·24], severe asthma 1·29 [1·22–1·37; people on three or more current asthma medications], bronchiectasis 1·34 [1·20–1·50], sarcoidosis 1·36 [1·10–1·68], extrinsic allergic alveolitis 1·35 [0·82–2·21], idiopathic pulmonary fibrosis 1·59 [1·30–1·95], other interstitial lung disease 1·66 [1·30–2·12], and lung cancer 2·24 [1·89–2·65]) and death (COPD 1·54 [1·42–1·67], asthma 0·99 [0·91–1·07], severe asthma 1·08 [0·98–1·19], bronchiectasis 1·12 [0·94–1·33], sarcoidosis 1·41 [0·99–1·99), extrinsic allergic alveolitis 1·56 [0·78–3·13], idiopathic pulmonary fibrosis 1·47 [1·12–1·92], other interstitial lung disease 2·05 [1·49–2·81], and lung cancer 1·77 [1·37–2·29]) due to COVID-19 compared with those without these diseases. Admission to ICU was rare, but the HR for people with asthma was 1·08 (0·93–1·25) and severe asthma was 1·30 (1·08–1·58). In a post-hoc analysis, relative risks of severe COVID-19 in people with respiratory disease were similar before and after shielding was introduced on March 23, 2020. In another post-hoc analysis, people with two or more prescriptions for ICS in the 150 days before study start were at a slightly higher risk of severe COVID-19 compared with all other individuals (ie, no or one ICS prescription): HR 1·13 (1·03–1·23) for hospitalisation, 1·63 (1·18–2·24) for ICU admission, and 1·15 (1·01–1·31) for death.
The risk of severe COVID-19 in people with asthma is relatively small. People with COPD and interstitial lung disease appear to have a modestly increased risk of severe disease, but their risk of death from COVID-19 at the height of the epidemic was mostly far lower than the ordinary risk of death from any cause. Use of inhaled steroids might be associated with a modestly increased risk of severe COVID-19.
National Institute for Health Research Oxford Biomedical Research Centre and the Wellcome Trust.
A subsequent systematic review found mixed evidence on asthma, with some data suggesting hospitalisation was equally probable but that COVID-19 would probably be more severe;
however, the evidence was weak. A systematic review and meta-analysis of cohorts of people hospitalised with COVID-19 suggested that chronic obstructive pulmonary disease (COPD) was associated with a four-times higher risk of severe disease (meaning intensive care unit [ICU] admission, death, or other critical outcomes).
However, although useful for prognostication in hospital, such cohorts cannot assess true risk because most people with COVID-19 are not admitted to hospital and decisions about who is and who is not admitted to hospital or to ICU are in part determined by severity of comorbid conditions. Because only a small fraction of people with COVID-19 are admitted to hospitals, selecting only hospitalised patients for cohort entry often induces bias, producing incorrect estimates of the size and even direction of effect of risk factors for deterioration or death.
Evidence before this study
We searched MEDLINE from inception of the database to Nov 21, 2020, using the terms “respiratory disease”, “COVID-19”, and “systematic review”. One review estimated that 5% of all people admitted to hospital with COVID-19 had chronic respiratory disease and a second that 8% of people who progressed to ICU care or who died of COVID-19 had respiratory disease. Two more reviews reported that the odds ratio for death from COVID-19 in people hospitalised with the disease was between two and four. One review focusing only on asthma produced an uncertain estimate. One community cohort reported little or no increased risk of severe COVID-19 in people with asthma, but reported a 63% increased risk of COVID-19 death associated with all other respiratory diseases combined. We also searched MEDLINE from inception of the database to Nov 21, 2020, using the terms “COVID-19” and “inhaled corticosteroids” (ICS). Two systematic reviews have examined the evidence on whether ICS are associated with COVID-19 outcomes. One review found no relevant studies. A second reported that in people admitted to hospital with severe COVID-19, use of ICS was associated with faster recovery and shorter duration of hospital stay, but it found no evidence of association with need for ICU care or death. A community cohort study reported evidence of lower risk of hospital admission with COVID-19 for users of ICS than for people with asthma who do not use ICS. A second community cohort suggested that steroids were associated with a modestly increased risk of death from COVID-19, but residual confounding could have explained the association.
Added value of this study
This study used a representative and unselected cohort to generate risk estimates for severe COVID-19 in people with chronic respiratory diseases. In doing so, collider bias could be avoided, which often creates distorted risk estimates when the population studied includes only people with moderately severe COVID-19—defined as admission to hospital. Unlike previous studies of people hospitalised with COVID-19, which suggested large increases in risk, this cohort provides evidence that the risk of severe COVID-19 in people with common airways diseases is only modestly raised over the risk in people without such diseases. Interstitial lung disease appears to increase the risk of severe COVID-19 by around 50%. These relative risks were similar before and after the shielding policy was introduced. Shielding supported everyone with cystic fibrosis, interstitial lung disease, and people with particularly severe chronic obstructive pulmonary disease (COPD) and asthma to remain at home and avoid all social contact. The study provides evidence that the use of ICS is not associated with a substantially increased the risk of severe COVID-19, but nor does it appear to be associated with reduced risk.
Implications of all the available evidence
Asthma appears not to be associated with a major increased risk of the severity of COVID-19, whereas COPD and interstitial lung disease are associated with a 50% increased risk of severe COVID-19. The apparent relative risks of severe COVID-19 arising from chronic respiratory disease are less than those of being male and of having diabetes, and are a small fraction of the ordinary risk of death from any cause. Advice about COVID-19 to people with chronic respiratory disease might have inadvertently generated undue anxiety. Placing these modestly increased risks in the context of other COVID-19 risk factors and the normal risk of mortality might reduce anxiety reported by individuals with respiratory diseases. Whether using inhaled steroids is associated with an increased risk of severe COVID-19 remains unclear.
People with a greater restriction of vital capacity had a somewhat higher mortality. There were few data on the different types of interstitial disease.
By contrast, there is observational evidence and evidence from randomised trials that ICS are associated with an increased risk of non-COVID-19 respiratory tract infections.
A systematic review aimed to assess the association between previous treatment with ICS and COVID-19 severity;
however, no studies were identified. Since then, the OpenSAFELY cohort
reported some evidence of increased risk in patients with COPD and patients with asthma prescribed high-dose ICS, but these differences were plausibly explained by risk factors not recorded in the available data.
Given the dearth of data on true risks of respiratory disease for severe COVID-19, we undertook a cohort study of a large, unselected community-based population to examine these issues.
Study design and participants
The data included all general practitioner consultations; illnesses, including those diagnosed in hospital that the general practitioner recorded; and prescriptions in primary care. These were linked to SARS-CoV-2 RT-PCR test records held by Public Health England. The QResearch records were linked to Hospital Episode Statistics, which gave the diagnoses of people admitted to hospitals in England, and the Intensive Care National Audit and Research Centre (ICNARC) Case Mix Programme database. This provided data from all ICUs in England. We also linked the QResearch database to death certificates provided by the Office for National Statistics.
All patients 20 years or older who were registered with one of the 1205 general practices on Jan 24, 2020, were included in the cohort. Patients were censored following the date of death, occurrence of the relevant outcome of interest (either hospital admission, ICU admission, or death), or study end date (April 30, 2020), whichever occurred first. This population cohort study was approved by the QResearch Science Committee, which has delegated power from the NHS ethics committee to approve studies using QResearch.
We coded a person as exposed to ICS if they were prescribed ICS at least twice within 150 days of the study start date.
The outcomes were hospital admission, ICU admission, and death due to COVID-19. Hospital admission was defined as a positive test for SARS-CoV-2 and appearing in the Hospital Episode Statistics dataset as an in-patient within 30 days of that test or having an International Classification of Diseases (ICD)-10 code U07.1 for confirmed COVID-19 or U07.2 for suspected COVID-19. Admission to ICU was defined as admission to an ICU with severe COVID-19 (ICD-10 code U07.1 or U07.2) in ICNARC records. Death from COVID-19 was defined as having confirmed or suspected COVID-19 (ICD-10 codes U07.1 and U07.2) on the death certificate, including deaths in and out of hospital.
Although respiratory disease might cause frailty and the need for care, the experience of this group is not representative of most people with underlying respiratory disease. In a post-hoc analysis, we analysed death from COVID-19 excluding all care home residents.
We examined the effect of the national lockdown and shielding introduced in England on March 23, 2020. Lockdown was a legal requirement for everyone except essential workers to stay at home, with dispensation for essential shopping and exercising. Shielding asked people with diseases or medications that were assumed to convey extreme risk of severe COVID-19 to stay at home at all times, with support provided to achieve this. Shielding also required practisers to minimise any social contact, even with members of the same household.
We assessed whether there was evidence that the associations between two common respiratory diseases, COPD and asthma, and severe COVID-19 were modified by age, sex, ethnicity, and smoking status.
We calculated representative absolute death rates from COVID-19 for people of various ages with and without respiratory disease, presenting the difference. To put the mortality rates into context, we compared excess deaths in people with respiratory diseases with so-called normal risk.
We used Cox’s proportional hazards models to estimate the mutually adjusted HRs for the associations between each respiratory disease and each of the severe COVID-19 outcomes and to estimate the mutually adjusted HRs between ICS use and severe COVID-19, with Stata (version 16.0). HRs were adjusted for patient age, sex, demographic factors, and comorbidities in all analyses. When adjustment variables had missing values, we included these participants in a missing category.
In the post-hoc analyses, we excluded people living in care homes and reran the analysis to examine the risk of death from COVID-19. To assess the effects of shielding, we reassessed hospitalisation, ICU admissions, and deaths, censoring outcomes on March 31, 2020, for the preshielding period, and from April 1–30, 2020, for the post-shielding period. To assess the effect of age, sex, ethnicity, and smoking status on the association between COPD and asthma and COVID-19, we added appropriate multiplicative interaction terms to the fully adjusted model described.
which was calculated with data from the Office for National Statistics.
For asthma and COPD, we present the risks assuming no interaction with age and then assuming age-specific risks.
Role of the funding source
The study was funded by the National Institute for Health Research Oxford Biomedical Research Centre and the Wellcome Trust. The funders had no role in study design; data collection, analysis, and interpretation; writing of the Article; and the decision to submit it for publication.
Table 1Risk of severe COVID-19 outcomes for people with underlying respiratory disease
Individuals could have more than one respiratory disease. COPD=chronic obstructive pulmonary disease. HR=hazard ratio. ICU=intensive care unit.
People with active asthma had at least one prescription for asthma medication. People with severe asthma were prescribed at least three different classes of medication for asthma in the year before cohort entry. COPD=chronic obstructive pulmonary disease. HR=hazard ratio.
Table 3Modification of the association between respiratory disease and severe COVID-19 by the introduction of lockdown and shielding
Data are hazard ratio (95% CI). People with active asthma had at least one prescription for asthma medication. People with severe asthma were prescribed at least three different classes of medication for asthma in the year before cohort entry. COPD=chronic obstructive pulmonary disease. ICU=intensive care unit.
Table 4Absolute difference in risk of death from COVID-19 in people with respiratory disease compared with people without respiratory disease and percentage of normal risk
People with active asthma had at least one prescription for asthma medication. People with severe asthma were prescribed at least three different classes of medication for asthma in the year before cohort entry. COPD=chronic obstructive pulmonary disease.
Table 5Association between regular use of inhaled corticosteroids and severe COVID-19
People with active asthma had at least one prescription for asthma medication. People with severe asthma were prescribed at least three different classes of medication for asthma in the year before cohort entry. HR=hazard ratio. ICU=intensive care unit.
People with pre-existing respiratory disease had a modestly increased risk of severe COVID-19 in this representative community cohort. People with asthma had an 18% increased risk of hospital admission with COVID-19, but there was no evidence that asthma was associated with an increased risk of death. Even in people with asthma who were prescribed three or more airways medications, the risk of death was modestly raised at worst compared with people without respiratory conditions. In the whole population, COPD was associated with a 54% increase in the risk of hospitalisation or death. The risks in people with bronchiectasis were modestly above one, although they were imprecisely estimated for death. Cystic fibrosis was too uncommon to estimate risks with precision, with only five of the 2081 patients in our cohort having a severe outcome, and no one with cystic fibrosis died from COVID-19. Interstitial lung disease was associated with an increased risk of severe COVID-19, while lung cancer was associated with a near doubling of severe COVID-19. People using ICS were at modestly increased risk of severe COVID-19 adjusted for respiratory disease and other comorbidities.
a factor that did not apply here. However, our exposures were restricted to the presence or absence of respiratory disease and did not record its severity. ISARIC reported that interstitial lung disease with greater restrictive lung function was associated with worse outcomes.
Our estimates of association are likely to overestimate the risks in people with mild disease and underestimate them in people with more severe disease. General practice records capture all prescriptions issued by general practitioners. Our classification of people as regular ICS users if they received at least two prescriptions in the 150 days before cohort entry might have classified some regular users as non-users, and some people fulfilling that definition might not have been using steroids at the relevant time of exposure. Exposure misassignment is unlikely to relate to an acute outcome like severe COVID-19 and thus is likely to move estimated associations towards the null. Additionally, our use of ICU care to indicate severe disease, in common with many studies, is a limitation. Most people who worsen with COVID-19 will not be admitted to ICU because their underlying health state suggests little likelihood of benefit. As such, low rates of ICU admission for people with COPD, interstitial lung disease, and lung cancer probably reflect the selection process and not the severity of COVID-19. A final limitation is our definition of severe COVID-19. So-called long COVID is an important outcome of COVID-19, but our dataset is unable to capture risk of this condition.
People with asthma might be at increased risk of longer duration of symptoms from COVID-19.
Because of an expectation of increased risk, it is possible that people with underlying respiratory disease could have more rigorously instituted behavioural measures to reduce their risk of infection from SARS-CoV-2 than people without respiratory conditions. This might mean we have underestimated the true risk of severe COVID-19. The introduction of shielding meant that people with interstitial lung disease and cystic fibrosis probably had less social contact than people without these diseases, and the risks associated with these conditions might have been expected to fall, but we saw only a small amount of evidence supporting this. Our estimates of relative risk for people with underlying respiratory disease reflect a combination of the risk conferred by that disease and the social distancing that people with respiratory disease might have instituted, which might have been more vigorous than that of the general population with or without shielding.
The odds ratio (OR) for severe COVID-19 with any form of respiratory disease was 4·2 (95% CI 2·9–6·0), with no differentiation between respiratory diseases. Selecting only people with COVID-19 that is severe enough to warrant hospital admission means collider bias could have distorted these associations. The prevalence of people with respiratory disease in patients hospitalised with COVID-19 appears to be lower than the prevalence in the general population. It is possible that those admitted with respiratory disease were more frail than those without, creating the impression that respiratory disease confers greater risk than is actually the case.
To our knowledge, ISARIC is the only study to have examined risks in people with interstitial lung disease.
In patients admitted to hospital, people with interstitial lung disease matched for age, sex, and comorbidity had an HR of 1·60 (95% CI 1·17–2·18) for death. The risks were slightly higher in people with idiopathic pulmonary fibrosis (1·74 [1·16–2·60]) than in those with other forms of interstitial lung disease (1·50 [1·02–2·21]).
These estimates are slightly higher than the estimates we produced, but we adjusted for more potentially confounding variables and included all people with interstitial lung disease in the population. The only representative population cohort we know to have reported on these associations is the OpenSAFELY study
of 17 million patients in England. It reported on risk of death for people with asthma who had not used oral steroids within the past 4 months (HR 0·99 [0·93–1·05]), those who had used oral steroids within the past 4 months (1·13 [1·01–1·26]), and all other respiratory diseases (1·63 [1·55–1·71]). Although we did not differentiate people with asthma by oral steroid use, the estimates for asthma are remarkably similar to those in our cohort. A US cohort matched people with asthma infected with SARS-CoV-2 to people without asthma with a SARS-CoV-2 infection.
They reported that the risk of hospitalisation was similar in the two groups, ICU admission non-significantly less likely in patients without asthma who had a SARS-CoV-2 infection, and death markedly less likely in people with asthma and COVID-19. Our findings on airways disease have expanded the current literature considerably, showing that asthma is associated with a very small increase in risk. The absence of a substantial association might seem surprising, given that respiratory viral infection is the main cause of exacerbations of asthma.
However, the finding is compatible with data from a small cohort of patients with asthma who were hospitalised with COVID-19, which showed no evidence that COVID-19 exacerbated asthma.
whereas men (whether they have asthma or not) are at higher risk of severe COVID-19.
It could be that women’s tendency towards severe asthma might explain why asthma appears to confer a slightly higher risk of severe COVID-19 for women than men, but any increase in risk for severe COVID-19 in women with asthma is low.
Since then, two studies in community cohorts have been reported. A Spanish cohort found that people using ICS were at lower risk of being hospitalised with COVID-19 than people who did not use ICS, OR 0·58 (95% CI 0·44–0·77).
The OpenSAFELY cohort
reported that in people with COPD, ICS use had a higher HR for death from COVID-19 of 1·39 (95% CI 1·10–1·76) than did long-acting β-agonists and long-acting muscarinic antagonist use.
In people with asthma, ICS use had an HR of 1·14 (0·85–1·54) for low or medium dose, and 1·55 (1·10–2·18) for high-dose ICS use, compared with short-acting β-agonist only. We used a different approach for the analyses, but the OpenSAFELY
estimates are similar to the ones we report for COVID-19 death associated with steroid use, adjusted for other respiratory illnesses, medication use, and other potential confounders. However, we also report estimates for the associations with hospitalisation and ICU care, which suggest similar relative hazards to those for death from COVID-19. Because ICS use will typically be higher in people with more severe disease, these data from community cohorts taken together suggest that ICS use does not influence the likelihood of severe COVID-19.
Our results do suggest that such diseases confer increased risk, but that the extent of this increase in risk, particularly for people with asthma, might be less than originally thought. COPD and interstitial lung disease appear to confer a 30–50% increased risk of severe COVID-19, while asthma confers a 10–30% increase compared with those without disease. By contrast, in the same models, being male was associated with a 46% increased risk of hospitalisation, a three-times higher risk of ICU admission, and an 86% increased risk of death compared with being female. People with respiratory disease feel particularly anxious about their disease predisposing them to severe COVID-19.
One study found that anxiety and depression increased during lockdown to a significantly greater extent in people with asthma than in those without.
We know of no evidence that being a man generates the same degree of concern about risk of COVID-19 as that reported by people with asthma, when it presents an objectively higher risk. Our use of normal risks of mortality also put these risks in context. The risk of death from COVID-19 perhaps caused by underlying respiratory disease was far lower than the ordinary risk of death in the 4 months (January to April, 2020) that included the peak of the SARS-CoV-2 epidemic in England. Although this could be partly due to behavioural responses of people with respiratory disease, the noticeably higher risk in people with diabetes,
also subject to the same warnings regarding their susceptibility to severe COVID-19, together with our analyses suggest that this is not the whole explanation.
The assumption that pre-existing respiratory disease would inevitably lead to an increased risk from COVID-19 was reasonable at the start of the pandemic, but as data continue to emerge, these assumptions might need to be revisited, and advice to people with respiratory disease from governments and charities should perhaps be substantially more reassuring.
JH-C and PW conceived of the record linkages and developed the platforms for the overall study. All authors conceived of this analysis. All authors discussed the protocol and edited it. MG and PA developed the analysis plan. MG analysed the data. PA and MG verified the underlying data. PA wrote the manuscript and all other authors revised the draft.
JH-C reports grants from the National Institute for Health Research Biomedical Research Centre (NIHR), Oxford, John Fell Oxford University Press Research Fund, Cancer Research UK ( C5255/A18085 ) through the Cancer Research UK Oxford Centre, and Oxford Wellcome Institutional Strategic Support Fund ( 204826/Z/16/Z ), during the study; and personal fees from ClinRisk (until 2019) outside of the submitted work. JH-C is an unpaid director of QResearch, a not-for-profit organisation that is a partnership between the University of Oxford and EMIS Health that supplies the QResearch database used for this work. PW reports grants from the NIHR Biomedical Research Centre, Oxford and the Wellcome Trust (221514/Z/20/Z), during the study; grants from Sensyne Health, outside of the submitted work; and personal fees from Sensyne Health, outside of the submitted work. PST reports previous consulting with AstraZeneca and Duke-National University of Singapore, outside of the submitted work. IDP reports personal fees from AstraZeneca, Boehringer Ingelheim, Aerocrine, Almirall, Novartis, GlaxoSmithKline, Genentech, Regeneron, Teva, Chiesi, Sanofi, Circassia, and Knopp; and grants from NIHR, outside of the submitted work. MG reports a scholarship from China Scholarship Council and funding from Oxford Biomedical Research Centre during the study. All other authors declare no competing interests.