key: cord-0949274-baikjy1h
authors: Stokes, Andrew C.; Lundberg, Dielle J.; Bor, Jacob; Elo, Irma T.; Hempstead, Katherine; Preston, Samuel H.
title: Association of Health Care Factors With Excess Deaths Not Assigned to COVID-19 in the US
date: 2021-09-13
journal: JAMA Netw Open
DOI: 10.1001/jamanetworkopen.2021.25287
sha: 3c25d9d65bb7727185b61e264a33e7db4661771c
doc_id: 949274
cord_uid: baikjy1h

This cross-sectional study assesses health care factors associated with excess deaths not assigned to COVID-19 in US counties in 2020.

= Changes in mortality that are independent of Covid-19 and common across regions 1 = Represents the extent to which past levels of all-cause mortality in a county are replicated in 2020 2 = Represents the extent to which mortality from Covid-19 affects all-cause mortality in 2020 after adjusting for historical mortality patterns ε = Error term Interpretation of Coefficients:

1. Combinations of and 1 indicate how mortality changes that are not associated with Covid-19 vary with the level of all-cause mortality in 2013-18. If 1 = 1.0, this would indicate that all-cause mortality in 2020 was on average equal to that in 2013-18, plus or minus the value of .

2. 2 reflects the number of excess deaths that occurred for every 1 directly assigned Covid-19 death. For example, if 2 were equal to 1.34, this means that for every 1 directly assigned Covid-19 death that occurred, 0.34 excess deaths occurred that were not assigned to Covid-19. If 2 were less than 1, this would mean that Covid-19 is over-recorded as a cause of death or reductions in mortality occurred for other causes.

To examine whether characteristics at the county or state-level (health care factors) were associated with having a higher percentage of excess deaths not assigned to Covid-19, we fully stratified our model and compared the

We calculated the percent of excess deaths not assigned to Covid-19 for each stratum using the 2 coefficient produced by our stratified models. The formula was as follows: This implies that 26% of excess deaths were not assigned to Covid-19 in these counties.

We determined the average observed directly assigned Covid-19 death rate for each stratum by calculating the weighted mean of observed directly assigned Covid-19 mortality. We then predicted the excess death rate not assigned to Covid-19 using the 2 coefficient as follows:

Excess Death Rate Not Assigned to Covid-19 = Mean Directly Assigned Covid-19 Rate • ( 2 Coefficient -1)

Excluded Counties Missing All-Cause Mortality Data for 2013-2018 (n=2) -Limited to Counties with 20+ Covid-19 Deaths (n=1,042) -Final County Analytic Sample b (n=2,096) a. To account for possible lags in mortality reporting, we used data generated on