Estimation of delay-adjusted all-cause excess mortality in the USA: March-December 2020

Abstract

We estimate the delay-adjusted all-cause excess deaths across 53 US jurisdictions. Using provisional data collected from September through December 2020, we first identify a common mean reporting delay of 2.8 weeks, whereas four jurisdictions have prolonged reporting delays compared to the others: Connecticut (mean 5.8 weeks), North Carolina (mean 10.4 weeks), Puerto Rico (mean 4.7 weeks) and West Virginia (mean 5.5 weeks). After adjusting for reporting delays, we estimate the percent change in all-cause excess mortality from March to December 2020 with range from 0.2 to 3.6 in Hawaii to 58.4 to 62.4 in New York City. Comparing the March-December with September-December 2020 periods, the highest increases in excess mortality are observed in South Dakota (36.9-54.0), North Dakota (33.9-50.7) and Missouri (27.8-33.9). Our findings indicate that analysis of provisional data requires caution in interpreting the death counts in recent weeks, while one needs also to account for heterogeneity in reporting delays of excess deaths among US jurisdictions.

Keywords: COVID-19; excess mortality; pandemic; reporting delay.

Fig. 1.

(a) Mean reporting delay by jurisdiction using different estimation approaches (legend). Error bars indicate the 95% CI for individually estimated reporting delays using a parametric model. Dashed line indicates a common mean delay inferred from the partial pool model. The entire shaded area indicates the 95% CI for the common mean delay, whereas the dark shaded area covers the interquartile range of the posterior. (b) Relationship between the fraction of deaths reported within the first 10 days and the mean reporting delay by jurisdiction obtained from non-parametric estimation of the reporting delay distribution. Dashed line indicates an estimate of 61% cited in technical notes of CDC [5]. (c) Correlation between number of reported COVID-19 deaths per 100 000 from September to December 2020 and the mean reporting delay by jurisdiction. Solid line is obtained from a linear regression model. Shaded area indicates 95% CI.

Fig. 2.

Comparing the nowcasted all-cause excess deaths by week of 2020 with expected deaths. Black line and grey shaded area show the median and 95% CI of the nowcasted death count in 2020. Blue line and blue shaded area indicate the median and 95% CI derived from posterior distributions of the expected weekly deaths. Individual grey lines indicate the reported deaths in 2014–2019.