Our data suggest that the differences in risk of adverse outcomes diminish as men and women age; men older than 75 had the least magnitude difference in relative risk of hospitalization and death compared to age matched women. With increasing age, levels of testosterone are known to decrease, implying a role for testosterone levels in the outcomes of those infected by COVID-19.23 Hypogonadism, the clinical state of testosterone deficiency, is associated with numerous comorbidities known to worsen COVID-19 outcomes; these include obesity, type 2 diabetes mellitus, and obstructive sleep apnea.24 Low testosterone states are associated with elevated levels of proinflammatory cytokines such as IL-6, IL-1, and TNF-alpha; this may potentiate the COVID-19 related cytokine storm observed in severe disease states.24 Other data implicated pro-inflammatory cytokines including IL-8, IL-10, IL-17a, and IL-23 in more severe disease course.25 It remains to be seen whether these cytokines may be clinically useful in terms of predicting disease outcomes, or may manifest differently in men as opposed to women. Importantly, the largest absolute number of deaths occurs in both men and women of age > 75, arguing against any protective effect of an isolated age-related hormone decline.
It is possible that some of the observed age and sex based differential effect is not due to increased risk in men alone, but instead derives from a protective effect in women <75 as reflected by CFRs in this subgroup. If this effect is hormonally driven, it may become diminished when a women experiences menopause. The rheumatologic and immunologic literature suggests that women may have enhanced immune responses to a wide variety of pathogens, potentially due to the impact of progesterone and estrogens on the immune system (estrogen receptor expression occurs throughout immune system cell types), but also due to altered gene dosage of sex chromosomes.26 Upon entrance into menopause, women may experience an “immune-senescence” resulting from an immune decline due to reduced hormone levels.27 Additionally, estrogen is a known driver of ACE2 expression.28 Nevertheless, an increase in estrogen levels may not alone offset the risk of worsened disease in men; males with obesity have increased circulating estrogen levels due to peripheral aromatization of testosterone to estrogen, yet continue to have worsened outcomes with COVID-19.29 Prospective trials are underway assessing the impact of administration of female sex hormones including progesterone and estrogen to men hospitalized with COVID-19 (ClinicalTrials.gov NCT04365127 and NCT04359329).
Finally, the observed sex disparity in COVID-19 outcomes may relate to comorbidity differences. The Center for Disease Control lists potential risk factors for worsened outcomes with COVID-19.30 Many of these comorbidities occur in greater rates amongst men.31 Men have more severe disease profiles, reduced rates of healthcare utilization, and riskier lifestyle habits.31 However, when analyzing these data, rates of comorbidities among those dying in NYC were comparable between sexes. Differences in comorbidities between men and women may not provide an adequate explanation for the observed sex disparity in COVID-19 outcomes.
Limitations of our study include the surveillance-based nature of this evolving dataset. Ethnographic data were missing in a large percentage of citywide data, prohibiting inclusion of ethnicity in our analysis. However, given our study size, including population-level adjustments, we can assume our findings are sufficiently powered, generalizable and externally valid. Data on antibody testing may permit more meaningful analysis, but at this time antibody testing remains suboptimal. While specific data on comorbidities in men and women would be ideal, these data were similarly not available given our cohort size and the rapid course of disease in our region. Furthermore, we were unable to temporally stratify our data, introducing possible bias from under sampling of earlier cases as testing was initially limited. This may overemphasize the absolute number of hospitalizations and deaths compared to cases, but is unlikely to be different between sexes. Case fatality rates in our series were calculated utilizing a broad definition of death (probable and confirmed), therefore the true magnitude of this effect is likely overestimated. Data were obtained in aggregate for age grouping and sex, thereby preventing further