Reducing the US Covid Fatality Rate

Here’s an article referencing a few recent empirical studies that have found lowered covid fatality rates since the beginning of the pandemic; that is, a reduction in the percentage of covid-infected people who die from the infection. Most persuasive are the studies that control statistically for age and comorbidities and that extrapolate beyond inpatients to population-wide estimates.

For the first months of the pandemic I estimated the US covid fatality rate at 0.6%; in mid-October, based on well-designed serology surveys, I raised it to 0.85%. Now I’m lowering back down to 0.7%, which is right around the IHME’s updated estimate.

Estimates of fatality and infection rates are cobbled together from multiple local studies available data. Population-based diagnostic testing surveys, repeated at regular intervals, would provide a more direct and accurate estimate. These surveys are still not being conducted in the US.

Reduced fatality rates are attributed to reduced overburdening of hospitals and to improved treatment. The current surge in cases is again starting to overload the healthcare providers and treatment facilities, so care might begin to deteriorate and fatality rates to go back up.

One thought on “Reducing the US Covid Fatality Rate

  1. Based on a quick-and-dirty everse engineering of their charts, it looks like the IHME has now lowered the estimated covid fatality rate a bit more, from 0.7% to 0.6%. From IHME’s 12 November briefing:

    We have substantially revised the infection-fatality rate (IFR) used in the model. To date, we had used an IFR that was derived from an analysis of population-representative antibody surveys where we disaggregated prevalence by age and matched COVID-19 death rates. The age-specific IFR from this analysis was assumed to be the same across locations and time.

    We have now accumulated considerable empirical evidence that suggests that 1) the IFR has been declining since March/April due to improvements in the clinical management of patients, and 2) the IFR varies as a function of the level of obesity in a community.The evidence supporting these observations includes:

    • An analysis of detailed clinical records of more than 15,000 individuals from a COVID-19 registry organized by the American Heart Association. This registry covers patientsin more than 150 hospitals. Our analysis suggests that after controlling for age, sex,comorbidities, and disease severity at admission, the hospital-fatality rate has declinedby about 30% since March/April.

    • An analysis of more than 250,000 individuals admitted to hospitals in Brazil withCOVID-19 shows that after controlling for age, sex, obesity, and oxygenation atadmission, the hospital-fatality rate has declined by about 30% since March/April.

    • An analysis of age-standardized IFRs from more than 300 surveys also suggests thatthe population-level trends in the IFR are consistent with a 30% decline sinceMarch/April. These data also suggest that the prevalence of obesity at the populationlevel is associated with a higher IFR and that the magnitude of the effect is similar tothat found in the individual-level analysis.

    Based on these empirical findings, we have switched to a new estimated IFR. The new IFR varies over time (declining since March/April by approximately 0.19% per day until the beginning of September), varies across locations as a function of obesity prevalence, and varies across locations (as before) as a function of the population distribution by age. The implication of lower IFRs over time is that for a given number of observed deaths there are more cumulative infections.

    Again, I’m wary that the current spike in deaths could compromise quality of care, causing fatality rates to increase. So for now I’ll stick with the 0.7% estimate.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s