Summary
Institute for Policy Research (IPR) Working Paper WP-22-03 (draft, January 2022). Uses a cohort-based approach and individual-level death certificate data from three Midwest areas (Cook County IL, Milwaukee County WI, State of Indiana) plus national Centers for Disease Control and Prevention (CDC) data to estimate population fatality rate (PFR), years of life lost (YLL), and life expectancy loss (LEL) due to COVID from March 2020 through September 30, 2021. The central finding: cohort-based national LEL is 11 days — two orders of magnitude below period-based estimates of 1+ year used in prior published work. Racial/ethnic and socioeconomic status (SES) disparities in LEL are documented and decomposed by age.
Key Claims
- Cohort vs. period LEL gap: Cohort-based national LEL = 0.029 years (11 days) through Sep 30, 2021 vs. period-based estimates of 1–2 years. Period-to-cohort ratio exceeds 1,000 for younger ages. The period-based approach assumes 2020 mortality rates persist forever — appropriate only if the pandemic is permanent.
- Mean YLL per decedent: 12.8 years (three Midwest areas), 13.71 years nationally. Both estimates reflect the age-weighted average; most COVID deaths are among the elderly (79.7% of Midwest deaths were age 65+) who have fewer expected remaining years.
- Population PFR: 0.19% (Midwest), 0.21% nationally through Sep 30, 2021.
- LEL by race/ethnicity (national): Blacks 0.041 years (15 days); Hispanics 0.045 years (16 days); Whites 0.024 years (9 days). Hispanics have lower overall PFR than Blacks but higher LEL because they die at younger ages with more years remaining.
- PFR disparities among the non-elderly: Non-elderly Hispanic men face >4× the PFR of White men of the same age. Black-White PFR ratio for men exceeds 2 until age 70. Asian-White PFR ratios are close to 1.
- SES gap: Higher PFR for residents in lower-SES zip codes (Social Deprivation Index [SDI] quintiles 2–5 vs. quintile 1); SDI quintile 5 effect ≈ being Black (average across genders). Controlling for SES reduces racial effects only modestly — SES explains only a fraction of racial disparities.
- Temporal shift: Mean age at death fell from ~80 in 2020 to <69 overall and <57 for Hispanic men by Q3 2021; mean YLL per Hispanic male decedent exceeded 26 years in Q3 2021. Younger individuals are an increasing fraction of COVID-related YLL.
- LEL by age (cohort): 30–34 year olds: 2.2 days; 85–89 year olds: 41 days. LEL rises with age because PFR rises faster than YLL per decedent falls.
- Period-based LEL age profile: Period approach shows highest LEL for young ages (assumes they face 2020 rates forever), opposite of the cohort-based result.
- Mechanism for racial disparities: Workplace exposure is the likely primary mechanism for non-elderly Hispanic/Black excess mortality; disparities in case fatality rates, hospitalization fatality rates, and age-adjusted comorbidities do not explain the pattern. Multigenerational household transmission plays only a modest role.
- Comorbidity limitation: UK/Hungary data suggest adjusting for comorbidities would reduce mean YLL by 1–2 years; nursing home resident overrepresentation among COVID deaths also inflates mean YLL.
Concepts Introduced or Extended
Entities Mentioned
Quotes
"Our estimates of life expectancy loss due to COVID to date are nearly two orders of magnitude lower than previously published estimates, which assumed that 2020 mortality rates would continue indefinitely."
"The PFR among non-elderly Hispanic men is generally more than 4 times that of White men of the same age."
"SES differences can explain only a fraction of the higher mortality of racial/ethnic minorities."
My Take
The core methodological contribution — applying the cohort approach to a time-limited event — is sound and important. Period-based LEL is the wrong tool for a pandemic that is expected to decline; using it produces estimates that alarm policymakers without informing them. The individual-level death certificate data and the quartile-by-race decomposition are the empirical assets. Two significant caveats: (1) This is a January 2022 draft covering data only through September 2021 — before Delta and Omicron waves fully played out — so the LEL estimates are incomplete; (2) the three Midwest areas may not be nationally representative in their racial/ethnic composition. The finding that SES explains only a fraction of racial disparities is consistent with workplace exposure being the primary mechanism, which has direct policy implications for occupational safety regulations during pandemics.