Criticism of the figures showing reported cases and deaths due to COVID-19 is rampant in the media and across social media. How deaths are recorded citing COVID-19 when someone also has diabetes, heart disease, or a respiratory illness is being criticized based on pockets of stories from nurses and doctors across the country. This has sewn doubt into the population, especially people with right-leaning political views, as they suspect COVID-19 charting is being politicized. However, numbers don’t lie, and there are ways to validate the pandemic’s effects on expected deaths that can be seen in the excess death graphs created by Tableau Public and displayed by the CDC.
Detecting reliability in underlying data
Actuaries are statisticians that calculate risks associated with healthcare costs and other liabilities. They are required by law in every state to certify reserves for insurance companies and for businesses that take on large self-insured risks, such as workers’ compensation, pension funds, health insurance, automobile coverage, or other general liabilities. Following trends, finding anomalies, and influencing policy is the overarching purpose an actuary takes on when given a project. This type of work must be looked at from multiple perspectives to not only give a professional opinion, but to analyze the data forensically, so as to discover inconsistencies in the underlying data. This is how mistakes are discovered as well as any potential fraud in reporting. According to Kathryn Fox, FSA, CPS, Senior Vice President, U.S. Mortality Markets of RGA, “actuaries do take fraud seriously”.
Regardless of the criticism of the reported COVID-19 cases and deaths, the bulk of the information appears to be reliable due to several factors outlined in the following sections.
Death distribution by age group for COVID-19
Death distribution by age group for COVID-19 cases in the U.S. is nearly identical to the distribution of all deaths. Since it is practically impossible to dispute that an actual death occurred, this chart demonstrates what we would expect to see in death-by-age distribution for all deaths and for deaths of people who had COVID-19.
Deaths are seasonal with peaks and valleys throughout the year
Peaks and valleys occur in weekly death count charts with the month of January of each year displaying the highest number of deaths followed by lower death counts the rest of the year. This chart shows the peaks and valleys from January 2017 to early May 2020. Note that the blue area shows the excess deaths that have occurred due to COVID-19 over expected deaths for the spring season in the U.S.
Seasonal death estimates are compared to actual deaths
Estimated deaths, regardless of cause, based on the orange trend line in the prior chart are compared to reported deaths for the same timeframe from the weeks ending Feb 29 to May 9. These figures are then compared to the number of COVID-19 deaths reported to date for the same timeframe. The “difference” column represents the difference in the original death projections versus reported deaths for all causes. The difference shows a significant increase in deaths over what was expected for the same timeframe. When comparing the excess deaths (87,782) to the number of COVID-19 reported deaths (80,037), one could make the assumption that COVID-19 is the cause of the vast majority of the excess deaths.
People with pre-existing conditions have a higher risk of dying with COVID-19
People with comorbidities or pre-existing conditions have been shown to be more susceptible to a hastened death when contracting COVID-19. This chart shows the distribution by cause of death from the weeks ending Feb 1 to May 16 per underlying condition. Respiratory illness, cardiovascular disease, diabetes, then obesity are the top 4 conditions reported. According to the CDC, “For 7% of the deaths, COVID-19 was the only cause mentioned. For deaths with conditions or causes in addition to COVID-19, on average, there were 2.5 additional conditions or causes per death.”
Why are provisional death counts lower than reported death counts?
Many people question the figures from the CDC because they hear about conflicting information, such as reported death counts versus provisional death counts. The difference is not due to errors in reporting but due to timing. Provisional death counts are lower because they are based on information that is charted on the actual death certificates, which are issued an average of 2 weeks after the death occurs. Death certificates and the information behind them give more documentation about the patient’s condition at the time of death. ICD-10 codes on medical charts represent diagnostic information that is entered into a database that triggers the issuance of the death certificate. This information takes longer to process.
When managing a pandemic, such as COVID-19, more immediate information is crucial to be able to make quick, informed decisions to handle the crisis. There are already delays in detecting if someone has COVID-19 or not; therefore, it is important to assess the threat more readily by reporting the deaths of people with COVID-19 as soon as possible.
Sally Hendrick of Shout Your Cause has evaluated risks for nearly 30 years as an actuarial consultant. The vast majority of her experience is in evaluating medical costs for liabilities for the healthcare industry.