NEW YORK (NYTIMES) - It's one of the most well-known takeaways of the pandemic: Men die of Covid-19 more often than women do.
Early on, some scientists suspected the reason was primarily biological, and that sex-based treatments for men - like oestrogen injections or androgen blockers - could help reduce their risk of dying.
But a new study analysing sex differences in Covid-19 deaths over time in the United States suggests that the picture is much more complicated.
While men overall died at a higher rate than women, the trends varied widely over time and by state, the study found. That suggests that social factors - like job types, behavioural patterns and underlying health issues - played a big role in the apparent sex differences, researchers said.
"There is no single story to tell about sex disparities during this pandemic, even within the United States," said Dr Sarah Richardson, director of the GenderSci Lab at Harvard University, which studies how biological sex interacts with cultural influences in society.
Dr Richardson's team began collecting sex data on Covid-19 cases and deaths early in the pandemic, before the Centres for Disease Control and Prevention began collecting and sharing this information. Her research group logged on every Monday morning and checked each state's data, maintaining it on a tracker on the lab's website.
The tracker, which stretches from April 2020 through December 2021, is the only source of sex-based weekly Covid-19 data by state.
That data enabled the researchers to analyse Covid-19 case rates and deaths across all 50 states and Washington DC over a period of 55 weeks.
Nationally, they found no significant differences in case rates between men and women. But the death rates - the number of deaths among men or women divided by the state's total population of each sex - were often higher among men than women.
Just how much higher depended on the state and the date. In Texas, for example, men died at a notably higher rate in every week the research group analysed. In New York, men died at a higher rate than women - although the gap was not quite as large as in Texas - during all but three weeks. But in Connecticut, women died more than men in 22 of the weeks analysed.
"You can have states right next door to each other, like Connecticut and New York, that have a totally different pattern but yet experienced the same wave," Richardson said.
Cumulatively over 55 weeks, mortality rates were slightly higher for women in two states, Rhode Island and Massachusetts. In nine states, including Connecticut, the rates were roughly equal. And in the rest of the country, death rates were higher for men.
Sex differences in genes, hormones or immune responses are not likely to explain these differences, the researchers said.
"There would be no reason for biology to be that variable across time and space," said Katharine Lee, a biological anthropologist and engineer at Washington University in St. Louis and an author of the new study.
But social and behavioural factors, the researchers said, could help explain many of these patterns.
For example, men are more likely to have jobs in transportation, factories, meatpacking plants, agriculture and construction - occupations with higher rates of Covid-19 exposure and fatalities. Men are also more likely to be incarcerated and to experience homelessness, increasing their risk of virus exposure.
Women are more likely than men to report hand washing, mask wearing and complying with social distancing restrictions, all of which may lower their risk of contracting the virus. And women are more likely to be vaccinated.
The researchers speculated that states with more public health restrictions might see reduced sex differences. In New York, which saw a significantly higher number of male deaths in the first six weeks of the pandemic, mortality rates evened out once restrictions were put in place. The observed differences in New York could also be partly explained by better data collection, as well as under-reporting of deaths in long-term care facilities, where the majority of residents are women.
Dr Richardson's research group did not have access to age data for each sex, an important factor since older people are more likely to die of Covid-19 and different states have different age distributions.
Even before Covid-19, men had a lower life expectancy, possibly driven by higher rates of certain chronic conditions, more risk-taking behaviours and more dangerous jobs.
That "pre-existing mortality gap", rather than a specific male vulnerability to the virus, could help explain the disparity with Covid-19, Dr Richardson said.
Still, independent experts said the new findings should not lead researchers to entirely discount the role of biology.
"You can't attribute observations about things like mortality from a complex disease like Covid and say it's all biology," said Dr Sabra Klein, a microbiologist and co-director of the Johns Hopkins Centre for Women's Health, Sex and Gender Research. "But I also don't think you can say it's all social and it's all behavioural, either."
Using electronic health records from Johns Hopkins hospitals in Maryland and Washington DC, Dr Klein found that there were higher rates of severe Covid-19 illness and death among men. But biostatistical modelling showed that this disparity could be substantially accounted for by greater inflammatory responses among men, suggesting a biological difference.
And in experiments looking at the effects of Covid-19 in hamsters by sex, which can be useful since they don't include the social factors present in humans, Dr Klein's group showed that males fared worse. Other studies have also shown that women produce a stronger immune response than men.
Other experts said that having access to more granular data - on factors like race, income and education level - would enable the researchers to take a more nuanced look at the observed variations in sex differences.
"I think they're doing a lot with a little," said Dr Derek Griffith, a public health psychologist and co-director of the Racial Justice Institute at Georgetown University.
"The data sources that document these differences don't usually have the capacity to help explain them," he said, pointing to the health effects of stress, financial burdens and discrimination that might underlie racial or gender differences in health outcomes but are difficult to quantify.
Dr Griffith said that racial differences in Covid outcomes were similarly complex. In the earliest stages of the pandemic, scientists speculated that Black people had a lower risk of contracting the virus and possibly had some biological protections, he said.
But when the data started to show that Black people in the US had a higher risk of dying of Covid-19 than white people, the pendulum swung the other way, with some scientists speculating about innate genetic differences.
Now, Dr Griffith said, there's greater recognition of the many socioeconomic factors that influence health disparities. "And yet in both cases, with sex and with race, the knee-jerk assumption is that it must be biological" he said.
Dr Richardson's Harvard group is hopeful that other researchers will use its data set to analyse the effects of states' varying public health policies.
But other data gaps remain: Studies have shown that long Covid-19, for example, disproportionately affects women, yet the disease isn't tracked consistently at the state level. And the researchers didn't have data on transgender or gender-nonconforming Covid-19 patients.
"These are not as rich of data as we would like to have to characterise the full gendered impact of the Covid pandemic," Dr Richardson said.