New Paper: COVID-19 sex disparities differ dramatically across U.S. states and over time, pointing to social factors
By Ann Caroline Danielsen and Mimi Tarrant
New out from the GenderSci Lab this week in the journal Social Science and Medicine is a comprehensive paper characterizing extensive heterogeneity in COVID-19 sex disparities over time and across states in the U.S. The paper—the first longitudinal study to quantify variation in COVID-19 gender/sex disparities across U.S. states—presents an initial analysis of the data gathered by the GenderSci Lab’s COVID-19 project, which began in April 2020. Analyzing 55 weeks of sex-disaggregated data derived from the Lab’s US Gender/Sex COVID-19 Data Tracker, we show that sex disparities in COVID-19 mortality are not large, stable, or consistent across time and place. As the paper concludes:
“The data that we present… [suggests] little reason to expect that interventions centering sex-related biological factors will play a primary or sizable role in explaining and ameliorating sex disparities.”
Key Findings: How do sex disparities differ across U.S. states and over time?
1. Cases are mostly higher among women, with variation over time and place.
Over 55 weeks of observation the total cases recorded by the Tracker for men and women were 14,889,007 and 15,383,226, respectively. The higher number of cases among women is consistent with the broader literature. Likely reasons for this disparity include that healthcare workers and pregnant individuals, both of whom are overwhelmingly women, are more likely than others to undergo regular testing. Our analysis shows that the number of U.S. states in which case rates were higher for women increased over time. During the first week of data collection, case rates were higher for women in 37 jurisdictions; by the final week of data collection this number increased to 45. For a deeper dive into how testing rates may exaggerate apparent sex disparities in case fatality rates, see this GenderSci Lab commentary published in Women’s Health Issues and the accompanying post on our blog.
2. Deaths are often higher among men, with variation over time and place.
Figure 1: Cumulative mortality rates for women (blue line) and men (red line) plotted over the entire period of observation. Graphs showing cumulative mortality rates for all other states, as well as mortality rates by week, are available in our paper. Rates per 100,000 population.
Over the 55 weeks of observation, the total deaths recorded by the Tracker were 273,455 and 227,863 for men and women, respectively. If the sex disparity in COVID-19 mortality were predominantly caused by innate, sex-linked biological factors – a central hypothesis pursued by biomedical researchers – we would expect these disparities to be somewhat consistent in magnitude across states and relatively stable over time. This does not bear out in the U.S. The sex disparity in COVID-19 cases and deaths varies over time and between U.S. states.
In Connecticut, for example, the mortality rate for women (blue line) and men (red line) were largely similar over the entire period of observation, with mortality being higher among women between June and December 2020. In New York, the mortality rates for women and men track similarly, but appear as parallel due to a large sex disparity accrued early in the pandemic, as explored in a second article by the GenderSci Lab available in preprint. In Texas, the gap in mortality between women and men consistently increased over time, while in Michigan it only became more pronounced early in the winter of 2021 (Figure 1).
3. Socio-contextual factors shape sex disparities in COVID-19 mortality
To better understand how much of the variation that we found between states and over time can be explained by socio-contextual factors, we fitted a multilevel regression model to the data. This analysis showed that 30% of the variation in the sex disparity in the U.S. can be attributed to differences between states and 10% to differences over time. (The cause of the other 60% in variation is unknown.)
4. Overall male excess mortality in the U.S. is far less than generally reported
While many reports have asserted a 2:1 sex ratio in mortality, the model shows that in a typical state the odds of a death being a man compared to a woman was 1.14 over the entire period of observation. The mortality rate ratios estimated by the model are visualized in figure 2, where each line represents a state. Figure 2 compellingly illustrates the extent to which, in each U.S. state, sex disparities changed in magnitude over time, with each state exhibiting unique patterns in the sex disparity. For example, the estimated male:female mortality rate ratio in Texas (orange line) ranged from 1.02 to 1.65 and never fell below 1 (i.e. men consistently had higher rates), while in Connecticut (green line) the rate ranged from 0.57 to 1.2 and fell below 1 for 22 weeks.
Figure 2: Predicted weekly mortality rate ratio by state from a multilevel crossed-effects conditional logistic binomial regression model. The black trendline is the average across states at that week. Texas, New York, and Connecticut are emphasized in orange, blue, and green, respectively, to highlight trends.
Why it matters
Our findings demonstrate that researchers investigating sex disparities in COVID-19 must engage with both sex and gender as factors that tangibly shape the biology of individuals and, therefore, COVID-19 outcomes.
Social factors likely influence the sex disparity in COVID-19 outcomes. The heterogeneity in sex disparities documented by our research is consistent with the interpretation that gender-related social factors, which can vary across states, contexts, and social groups, drive COVID-19 sex disparities. These include gendered health behaviors, gendered occupational stratification, and gendered pre-existing conditions. Such factors will not be homogenous within gender categories, but will operate interactionally with other social factors such as race and class, as demonstrated by our prior work which demonstrated large variation in the sex disparity across racialized social groups.
Biological sex-related factors are not adequate to explain the variation in sex disparities we observed. Assigning primary causality to biological sex influences the research streams funded, public health interventions prioritized, and populations designated as at risk, shifting the focus from social-structural variables to molecular ones. When differences in COVID-19 outcomes between women and men are observed, as they are in our paper, interpreters should beware of rushing to conclude that the sex disparity is primarily caused by sex-related biological factors.
Note that our analysis describes differences in COVID-19 between people categorized as men/males and women/females in U.S. state COVID-19 data. This binary classification reflects the categories for collecting and reporting sex-disaggregated data at the state level, which, as we’ve discussed elsewhere, results in the underinvestigation of the impacts of COVID-19 on gender-diverse individuals, despite evidence suggested that these populations are at elevated risk. We call on public health agencies to structure data collection in an inclusive way so as to close the critical gaps in knowledge about how COVID-19 has affected intersex, trans, nonbinary, and gender-expansive people.
We hope that future researchers will cross-reference the publicly available US Gender/Sex COVID-19 Data Tracker with other sources of state and federal data to further characterize the etiology of sex disparities in COVID-19.
Paper: Danielsen, A.C., Lee, K.M.N., Boulicault, M., Rushovich, T., Gompers, A., Tarrant, A., Reiches, M., Shattuck-Heidorn, H., Miratrix, L.W, Richardson, S.S. “Sex disparities in COVID-19 outcomes in the United States: Quantifying and contextualizing variation”. Social Science & Medicine, 294. 2022. Available at: https://www.sciencedirect.com/science/article/pii/S0277953622000193#bib31
How to cite this blog post
Danielsen, A.C. and Tarrant, M. “New Paper: COVID-19 sex disparities differ dramatically across U.S. states and over time, pointing to social factors”. GenderSci Blog, 19 Jan. 2022. genderscilab.org/blog/covid-19-sex-disparities-differ-dramatically-across-us-states
Statement of intellectual labor
Ann Caroline Danielsen and Mimi Tarrant wrote and edited the blog post. Marion Boulicault, Heather Shattuck-Heidorn, and Sarah Richardson provided edits.