As of mid-May, 90,000 Americans had been killed by COVID-19, and provisional data from the Centers for Disease Control and Prevention show that the devastation is disproportionally shouldered by racial/ethnic minorities. Nevertheless, it is way too early to assess the population effects of this deadly virus.
To be clear, COVID-19 will impact some groups more than others. We anticipate that future research will document that wealthy people living in urban areas were most at risk at the very early stages of the pandemic due to their centrality in global social networks. But over time, the disproportionate risk will likely shift to marginalized people and places. Some cannot afford to practice social distancing while others live in states that have failed to implement stay-at-home policies. Among those who become ill, the most marginalized may be more likely to die because they are already compromised by chronic conditions such as diabetes or heart disease.
Yet we caution that evaluating the full impact of COVID-19 will take time and careful analysis. Here is why.
First and foremost, data about COVID-19 exposure, infection, and mortality are inadequate and incomplete. As the pandemic unfolds in real time, the definition of what counts as a COVID-19 death is changing. For example, COVID-19 deaths spiked in mid-April in New York when the state added presumptive (suspected) cases to its official counts. Additionally, we do not yet know the numbers of people who have contracted the disease and are showing known symptoms, much less the numbers of people who are asymptomatic. As a consequence, denominators of COVID-19 mortality rates are unknown. This partially explains why estimates of the COVID-19 mortality rate have varied considerably with the degree of testing. At the heart of the problem is the lack of comprehensive testing for active infections and antibodies, thus hampering public health officials’ abilities to make basic assessments about the spread of infection and the likely risks of relaxing social distancing policies.
Data limitations also make it difficult to assess the social contours of the disease. The extent to which various infectious disease surveillance systems have captured demographic characteristics has been uneven across states and over time, even prior to the current pandemic. For example, we are only recently learning that African Americans and Latinos appear to be disproportionately dying of the disease, but we can only speculate why this is the case. Are they more likely to be exposed, or are they more likely to get seriously ill or die if infected, or both? Moreover, data on the social characteristics of COVID-19 victims, such as their occupation, education, income, and insurance status, are important for understanding potential mechanisms underlying their greater risk of exposure, illness, and death. Certainly, other groups may also face high levels of risk, but we may be blind to these inequities unless infectious disease surveillance systems systematically collect and openly distribute information on demographic characteristics. We therefore call for greater coordination in the collection and dissemination of such data.
Finally, we warn against simplistic estimates of the death toll of the pandemic. Accurate counts of COVID-19 cases and deaths are necessary but insufficient. As we learned in the aftermath of Hurricane Maria, which struck Puerto Rico in September 2017, simple counts of deaths directly attributable to the storm—such as deaths caused by high winds or storm surges—captured only a portion of the total deaths attributable to the natural disaster. Thousands more deaths occurred when injuries and medical conditions that are more manageable in ordinary times (e.g. sepsis) were neglected or unattended to following the hurricane due to inadequate basic medical care, a collapsed power grid, damaged or demolished housing, and disrupted supply chains.
We anticipate that the total number of deaths directly attributable to COVID-19 will also be an underestimate of the true number. While deaths due to traffic accidents may temporarily go down during the pandemic, mortality from other causes may increase due to the postponement of medical care for non-COVID-19 medical conditions. Deaths of despair (i.e., due to suicide, drugs, and alcohol) may also increase in the months and years following the pandemic as the social and emotional tolls (e.g. trauma, loss of friends and family) and economic costs (e.g. unemployment, health costs) of the pandemic unfold. After death certificate data are released in the years following the COVID-19 pandemic, demographers and epidemiologists will need to get to work estimating the excess mortality over and above what might have occurred in the absence of the pandemic and what policy responses could have been and will be most effective in mitigating such consequences in the future.