In this large population-based study, we have observed a substantial male sex disadvantage in severe morbidity and mortality due to COVID-19. In general, men’s excess risk was most pronounced among middle aged people (50–59 years), in particular for the risk of COVID-19 related death. Adjusting for several important comorbidities and socioeconomic factors did not attenuate the observed associations. Further, adjusting for unobserved factors shared by individuals in the same household in a conditional Cox analysis did not provide any further explanation to the sex disparity in severe COVID-19 outcomes. In addition, among patients hospitalized due to COVID-19, men had an increased risk for complications associated with COVID-19 such as viral pneumonia, acute respiratory distress syndrome, acute respiratory insufficiency, acute kidney injury, and sepsis.

Our finding that men are at higher risk of severe COVID-19 infection and death compared to women, and that the increased relative risk varies across age-groups is consistent with previous studies.[1, 3, 19] The age-sex pattern of a pronounced higher relative risk for men around middle age has been hypothesized to be (at least partially) caused by the differential distribution of comorbidities across sexes and age strata. Women are generally reported to have a higher comorbidity burden, especially at older ages. However, women tend to suffer from more non-fatal chronic conditions such as migraine, depression, autoimmune and musculoskeletal diseases, whereas men have more life-threatening conditions such as cardiovascular diseases, hypertension, chronic lung diseases and type-2 diabetes,[28, 29] which are associated with a worse progression of COVID-19.[5] Thus, a higher burden of specific comorbidities in men may explain the markedly increased risk for severe COVID-19 outcomes around middle age, whereas a reduction of the increased relative risk for men at older ages may be explained by a survival effect leaving only the healthiest men to survive to old age. However, adjustments for comorbidities and socioeconomic factors did not attenuate the association between sex and severe COVID-19 outcomes in any age group. This finding is in line with previous studies on case fatality among patients with a confirmed infection reporting an increased risk among men independent of comorbidities, demographics, and health behaviors.[19, 30].

Gender-related differences in occupations, health-behaviors and attitudes have also been suggested to partly explain men’s increased risk for COVID-19 outcomes.[1] Studies have shown that women are more likely to perceive COVID-19 as a serious health problem and to agree and comply with restraining public policies.[6] However, such behaviors are primarily associated with risk for infection, whereas results from large meta-analyses and epidemiological surveillance demonstrate that COVID-19 infection rates are similar between sexes.[2, 4] Further, the test positivity rate among asymptomatic individuals has also been shown to be similar between sexes,[19] or even higher in women.[31] Therefore, the increased risk for morbidity and mortality in COVID-19 among men does not seem to be explained by higher risk of SARS-CoV-2 exposure or susceptibility. Our finding that among hospitalized patients with COVID-19, men had an increased risk for COVID-19 related complications (i.e., viral pneumonia, acute respiratory distress syndrome, acute respiratory insufficiency, acute kidney injury, and sepsis) independent of previous comorbidities and socioeconomic factors, also suggests that men have a higher risk of severe disease progression, rather than just increased virus exposure or susceptibility. It has also been suggested that gender-related social norms may cause men to postpone seeking out medical treatment, thereby experiencing more severe COVID-19 due to this delay. However, previous studies have found indications both for and against this tendency.[19, 32] Since we have not been able to adjust for these types of behaviors, we cannot rule out that delayed treatment among men partially explains our results. However, data from countries worldwide show that the sex disparity in COVID-19 case fatality and intensive therapy unit admissions is a global phenomenon with relatively homogenous relative risk estimates,[1, 4, 10] whereas gender differences in social behavior, life-style factors and comorbidities vary across countries. In addition, results from the within-household design provided no further explanation to the sex disparity, despite accounting for housing-related living conditions, including many social and lifestyle related factors to a large extent, as well as exposure to SARS-CoV-2, given that much of the transmission of SARS-CoV-2 occur within households.[33] Thus, the within-household design presumably partly accounts for gender-related differences, especially those who are related to risk of infection, even though gender-related differences in attitudes and behavior exist within households. Thus, we find it unlikely that gender differences in social behaviors is a driver of men’s increased risk in severe COVID-19 outcomes.

The observed sex disparity in COVID-19 outcomes could be a consequence of several biological sex differences, such as sex steroid hormone levels, sex chromosomes, differential genetic expression, and differences in the immune function, whereof some may interact with age.[10] Sex steroid hormones have known immunomodulatory functions; whereas estrogens may have both pro- and anti-inflammatory effects, androgens mainly act immunosuppressive.[34,35,36] Experimental animal studies support a beneficial immunomodulatory effect of estrogens towards coronaviruses,[37, 38] and epidemiological evidence indicates that hormone replacement treatment with estradiol may reduce COVID-19 mortality in post-menopausal women.[39] Although less studied, there are also some epidemiological indications of a potential negative effect of androgens on risk of COVID-19 infection, as a positive effect of anti-androgen treatment among male prostate cancer patients was observed.[40].

Women may also benefit from having two X-chromosomes that encode a large number of immune-related genes.[41, 42] As carriers of a single X-chromosome, men are more vulnerable for X-linked mutations, compared to women who are mosaic for X-linked genes. Due to random X-inactivation in females, there is usually no difference in dosage of X-linked gene products between sexes. However, incomplete inactivation of X-linked genes may occur at varying degree between genes and individuals,[43] and have been associated with higher prevalence of autoimmune diseases in women. For example, enhanced Toll-like receptor 7 (TLR7) gene expression has been associated with increased risk for SLE and other autoimmune disorders in women.[44] TLR7 is a pattern recognition receptor used to detect single stranded RNA, including coronaviruses. Ligand binding to TLR7 results in an increased production of type I interferons, which is an important mechanism of the innate immune system towards virus infections found to be stronger in women.[45, 46] In case studies, supposed loss-of-function mutations in TLR7 have been identified in young men with severe or fatal outcomes in COVID-19, without any known predisposing risk factors.[47, 48] A subsequent study screening for X-linked mutations in men, identified deleterious TLR7 variants in men with unexplained severe COVID-19, whereas none were identified among the mildly or asymptomatically infected men.[49] Angiotensin-converting enzyme 2 (ACE2) is the entry receptor for SARS-CoV-2 and the ACE2 gene is also coded on the X-chromosome and has been found to be regulated by estrogens.[50]ACE2 has a crucial role in the renin-angiotensin-aldosterone system (RAAS) which regulate blood pressure and electrolyte homeostasis. Thus, it has also been speculated that sex-derived differences in ACE2 expression and regulation may in part explain differences in COVID-19 outcomes.[51, 52].

Severe COVID-19 is associated with a state of dysregulated immune responses associated with excessive production of inflammatory cytokines and chemokines, inflammatory cell infiltration in the lungs, a so-called cytokine storm, causing lung injury and respiratory distress, together with lymphopenia. Deaths due to COVID-19 usually result from acute respiratory distress syndrome, acute respiratory failure, coagulopathy, organ failure and septic shock.[53] We found that men hospitalized due to COVID-19 were at higher risk of acute cardiac injury and cerebral infarcts compared to women, but this effect seemed to be confounded by underlying comorbidities. In contrast, we also found that male COVID-19 patients had a higher risk of developing pneumonia, acute respiratory distress syndrome, acute respiratory insufficiency, acute kidney injury, and sepsis compared to female patients, which seemed to be independent of underlying comorbidities and socioeconomic factors.

A major strength of this study is the total population-based setting with complete coverage of COVID-19 related hospitalizations and deaths, why this study is not subjected to selection bias. Another important strength is the use of administrative register data on socioeconomic factors. For example, these included information on housing arrangements that enabled the novel use of within-household comparisons to address unknown and unmeasured confounders shared by cohabitants in the study of sex disparities in severe COVID-19 outcomes. Further, the unique Stockholm Regional Healthcare Data Warehouse, which contains information on hospitalizations as well as data from primary and secondary care, improved our capability to identify individuals with important comorbidities, compared to using only hospital-based data.

There are also limitations to consider. This is an observational study, and as such we cannot determine the causal relationships that underlies the male sex disadvantage in morbidity and mortality due to COVID-19. This study could have been subjected to surveillance bias as men generally receive more specialist inpatient care, which could increase the chance of being detected with COVID-19 in health registers. To minimize this issue, we defined COVID-19 hospitalizations based on main diagnosis only. The accuracy of COVID-19 diagnoses in Swedish health and cause of death registers is presumably high but have not been formally validated. Sensitivity of hospital diagnoses as a measure of severe COVID-19 morbidity among older patients may be an issue since age discrimination in access to care has occurred.[54] Although it has not been shown that this age discrimination differs by sex, results of hospitalizations and ICU admissions among elderly should be interpreted more carefully. Some cases of comorbidity may have been misclassified. However, this is likely nondifferential with regard to sex. On the other hand, under-diagnosis, especially of less severe comorbidities, may differ between sexes and introduce bias. Residual confounding by life-style factors and comorbidities, especially mild cases not identified by registers, may be a limitation. However, these factors are likely captured by adjustments for socioeconomic factors and factors shared within households to some extent, though adjustments for these factors did not notably change the observed estimates. Emerging evidence seems to support an association between smoking and more severe COVID-19 outcomes.[55, 56] However, more women than men smoke daily in Sweden, although the difference is marginal.[57] Therefore, it is not likely that smoking status is a major driver of the observed sex disparity. Another consideration is that the follow-up period for hospitalizations and ICU admissions (ending on May 16, 2021) overlaps with the introduction of vaccination against SARS-CoV-2, which started in week 52, 2020. The vaccination coverage increased slowly during the first quarter of 2021 and thereafter started to increase more steeply. At the end of our study period, almost 10% had received two doses and about 30% had received at least one dose. Thus, vaccination status could potentially affect our results for hospitalizations and ICU admissions. However, results from our sensitivity analysis showed that HRs were the same during the first (before vaccination was initiated) and second wave (when the vaccination started) of the pandemic. Thus, it is unlikely that vaccination status interferes with our results. It is also unlikely that other factors that vary over time or between pandemic waves, such as stochastic transmission of the virus to different groups of people and new mutations and strain variants emerging, affect the sex ratio observed in our results.

Our results confirm that men have an excess risk of hospitalization, ICU admission and death due to COVID-19. This sex disparity varies across age groups and is most pronounced among middle aged men compared to women of the same age. Further, among hospitalized COVID-19 patients, men are more likely to be diagnosed with COVID-19 related complications. Different explanations have been suggested to explain men’s vulnerability to COVID-19, including gender related health behaviors, differential distribution of comorbidities and biological sex differences such as sex steroid hormones, genetic constitution, or differential gene expression. The reason for men’s excessive risk is likely multifactorial with a complex interplay between several factors. In this study we adjusted for multiple confounders, including important comorbidities, socioeconomic factors as well as unmeasured household-related factors. These adjustments had no substantial impact on men’s excess risk of severe COVID-19 outcomes. Thus, our results motivate a focus in future studies on biological mechanisms that influence the immune response towards SARS-CoV-2 to help explain women’s advantage in COVID-19. This study also highlights the importance to consider the combined effect of age and sex when defining those at highest risk of severe COVID-19 outcomes.