Mortality in COVID-19: Further Evidence for a Sex-Based Difference in the OpenSAFELY Study

Both sex- and gender-related factors impact disease susceptibility and outcomes. Although the SARS CoV-2 virus is estimated to have infected close to 40 million people globally (8,127,522 U.S.), with 1,111,714 deaths as of October 18, 2020, ¹ sex-disaggregated data on infection rates, disease severity, and mortality is limited. An analysis of patients in Wuhan, China, who died of COVID-19-related pneumonia in late January, 1 month after the first case of COVID-19, showed a higher incidence of death in men. ² Subsequently, national and global databases reported proportionately more COVID-19 deaths in men compared with women. Sex-disaggregated country level data from Global Health 50/50, based upon government sources, show that in most countries, including the United States, the proportion of deaths in men exceed those in women, whereas infection rates overall appear similar. ³ Data collected as of October 6, 2020, in 77 countries that reported fully sex-disaggregated data on cases and mortality at some time point, showed that of 595,135 deaths in confirmed cases of COVID-19, the male to female ratio averaged 1.44 (range 0.58–2.51), with a ratio of 1.26 in the United States. These numbers, however, do not represent all confirmed cases or deaths in individual countries, because only sex-disaggregated data are reported. In the United States, state-specific data on COVID-19, reported by the Harvard Gender Science Lab, as of November 18, 2020, shows higher age-adjusted mortality in men in 37 of 43 states with data. ⁴

Until recently, strong evidence for a sex-based difference in mortality for COVID-19 was lacking. It was not clear whether higher deaths in men were related to biological (sex-based) differences, comorbidities that are more frequent in men, ethnicity, socioeconomic conditions, or gender-related factors (such as exposure to hazardous environmental factors, occupation, willingness to access health care services, and quality of health care), or to a combination of these variables. ⁵

New data from OpenSAFELY, a cohort study of 17 million adult patients of primary care practices in England, including 10,926 who died from COVID-19 complications, provides support for sex-based differences in mortality even after adjusting for comorbidities, clinical risk factors, and social determinants of health. ⁶ In an analysis adjusted for age only, male sex was associated with increased mortality (estimated age-adjusted hazard ratio [HR] 1.78, 95% confidence interval [CI] 1.71–1.85). Importantly in a fully adjusted analysis, accounting for age, geographical area, deprivation (low income and lack of resources), smoking, and pre-existing diseases, mortality remained significantly higher in males (estimated fully adjusted HR 1.59, 95% CI 1.53–1.65). In a separate analysis adjusted for all the previous factors plus ethnicity, and reported in the same article, the results were similar. Factors other than sex that were associated with increased risk of mortality were age, deprivation, South Asian or black ethnicity, obesity (defined as body mass index [BMI] > 30), diabetes (higher risk with uncontrolled blood sugar), chronic cardiac disease, severe pulmonary disease, including asthma requiring oral corticosteroids, hematological malignancy, cancer, renal impairment (estimated glomerular filtration rate <60 mL/[minute ·1.73 m²]), liver disease, stroke and dementia, other neurological diseases, specific autoimmune diseases (rheumatoid arthritis, lupus, and psoriasis), and immunosuppression. Increased mortality was not associated with high blood pressure or hypertension, nor with smoking. The effects of smoking were largely accounted for by other diseases that increased mortality.

The association of male sex with increased risk for COVID-19 death, even after adjustment for all these variables, suggests that sex as a biological variable influences mortality from COVID-19, regardless of social or clinical risk factors, or comorbidities. However, various diseases such as diabetes (HbA1C ≥ 7.5%: age, sex-adjusted HR 2.61; fully adjusted HR 1.95), obesity (BMI 30–34.9, BMI 35–39.9, BMI ≥40: age, sex-adjusted HR 1.23, 1.81, 2.66, respectively; fully adjusted HR 1.05, 1.40, 1.92, respectively), and chronic heart disease (age, sex-adjusted HR 1.57, fully adjusted HR 1.17) confer a higher risk of COVID-19-related death, independent of sex. This is similar to the relationship between diabetes and cardiovascular disease in women: although premenopausal women have a lower risk of cardiovascular disease than men, the onset of diabetes reduces this protective effect. ⁷

The vast data set in OpenSAFELY is possible because a National Public Health Service (NHS) provides medical care to virtually the entire population in England. The size of the study enables detection of social factors and diseases that influence COVID-19 severity, thus providing information that could impact patient care, and methods and policies for disease prevention. Open access to all patient records, rather than the last visit or only hospital records, and inclusion of data from nonhospitalized patients are additional strengths of the OpenSAFELY database. The study did not address all variables that might influence COVID 19 mortality, such as access to personal protective equipment, employment, social distancing, care-seeking behavior, and pregnancy.

The effects of SARS-CoV-2 in the pregnant woman and newborn child have been investigated in other studies. Stillbirths, miscarriages, preterm deliveries, and possible maternal–fetal transmission ⁸ have been reported, although causality cannot be determined. ^{9 –12} A retrospective systematic review of 86 studies of pregnant women with COVID-19 found low maternal and neonatal mortality (<1%), admission to the intensive care unit (ICU) in 7%, which was more common in women with comorbidities and older age, and a low rate of spontaneous preterm delivery that appeared consistent with what would be expected in pregnant women without COVID-19. ¹³ A systematic review of 77 cohort studies in 13,118 pregnant women and 83,486 nonpregnant women with COVID-19 found that cough, fever, and myalgia were less common in pregnant women; however, pregnant women were more likely to be admitted to the ICU, develop pneumonia, and deliver preterm, but did not have increased mortality. ¹² More severe disease was associated with comorbidities such as age and obesity. In 17 of the studies, 25% of infants were admitted to the neonatal unit. Limitations of these reports include lack of common questions for clinical assessment, the absence of a comparator group, such as pregnant women without COVID-19, or nonpregnant women with COVID-19 in many studies, and the inability to use individual participant data for a meta-analysis.

One of the largest investigations, by the U.S. CDC, analyzed data from about 325,000 surveillance reports in women of reproductive age with COVID-19. ¹⁴ Pregnancy status was ascertained in 28% of the reports: 91,412 women (8,207 pregnant, 83,205 not pregnant). Pregnancy was associated with increased rates of ICU admission (adjusted risk ratio [RR] 1.5, 95% CI 1.2–1.8), and mechanical ventilation (adjusted RR 1.7, 95% CI 1.2–2.4), without increased mortality. However, a major limitation of this study is missing data, on adverse outcomes and on pregnancy status. More research is needed to assess effects of COVID-19 in all trimesters of pregnancy, neonatal outcomes, the risk of maternal–fetal transmission, and the risk of noniatrogenic preterm delivery and C-Section.

Although the mechanism of the sex difference in mortality from COVID-19 is not yet known, biological differences in the immune system may have a role. ¹⁵ Studies in humans and animals have found that females have stronger innate and adaptive immune responses to infection ⁹ and that estrogen enhances the immune response. ¹⁶ Both cellular responses to infections and antibody production have been shown to be greater in females, providing an advantage in overcoming infections but also a predilection to autoimmune diseases. ^17,18 Genes regulating immune function may be found on the X chromosome, theoretically giving females an advantage over males who have only one X chromosome. ¹⁹ Genes for angiotensin converting enzyme 2 (ACE2), which has a critical role in the entry of the SARS-CoV-2 virus into cells, and the Toll-like receptor 7 (TLR-7), which has a role in host defense, are located on the X chromosome. ²⁰ Other X-linked genes regulating immune function are TLR-8 and IRAK1, a molecule in the toll-like receptor pathway, ¹⁸ and nuclear factor kappa B modulator, which is involved in many immunological pathways. ¹⁹

Research shows that the SARS Co-V2 virus enters cells in the lung, nasal mucosa, and other tissues, through the binding of its spike protein to the ACE2 receptor. Before binding, the spike protein is activated by a transmembrane serine protease, TMPRSS2. ²¹ Preclinical studies suggest that ACE2 expression is regulated by sex steroids, ²² with downregulation in females, which could potentially hinder SARS-CoV-2 entry into host cells. ¹⁵ Females may have another advantage over males in defense against SARS-CoV-2 because TLR-7 is believed to escape X chromosome inactivation, leading to greater expression in females. ¹⁵ Furthermore, TMPRSS2 has been shown to be upregulated by androgens in the prostate. ²³

Antibody levels in response to some viruses and vaccinations, such as the 2004–2005 trivalent influenza vaccine (2004–2005), ¹⁸ may be higher in women compared with men. It is not known whether a sex difference in antibody production in response to SARS-CoV-2 exists, although one study in Iceland found lower SARS-CoV-2 antibody levels in women who had recovered from COVID-19. ²⁴ More research on sex-based immune responses to SARS CoV-2 is needed, and could potentially provide new targets for prevention, and/or reducing disease severity.

The observation of a higher death rate in men with COVID-19 points to the importance of the collection and examination of sex-disaggregated data in preclinical and clinical research and public health surveillance. For the COVID-19 pandemic, analyses by sex and gender may aid the development of strategies for prevention. As gender is nonbinary, it would be ideal to have data available for women, men and gender-diverse people. Such evidence is anticipated to improve knowledge of sex and gender differences in the pathophysiology of disease, as well as the safety and effectiveness of vaccines, new medications, and other forms of treatment.