What peer‑reviewed estimates exist for all‑cause excess mortality in the U.S. by month from 2020–2024?
Executive summary
Peer‑reviewed, month-by-month estimates of U.S. all‑cause excess mortality exist for the bulk of the acute pandemic period (roughly March 2020 through mid‑2022) but peer‑reviewed monthly series that extend through 2023–2024 are limited; much of the publicly used monthly data for 2022–2024 comes from datasets and government tools that are not traditional peer‑reviewed journal articles (Our World in Data, CDC methods pages) [1] [2] [3] [4]. Where peer‑reviewed work is available it is typically produced with Bayesian or regression baselines and reports county‑ and month‑level estimates for 2020–2021 or through August 2022, with total excess‑death tallies that differ by methodological choice [1] [5] [2].
1. Science Advances and county‑month estimates for March 2020–February 2022
A high‑profile peer‑reviewed analysis published in Science Advances produced monthly, county‑level all‑cause excess mortality estimates covering March 2020 through February 2022 using Bayesian hierarchical models trained on 2015–2019 data, and reports roughly 634,830 excess deaths in year 1 (Mar 2020–Feb 2021) and 544,194 in year 2 (Mar 2021–Feb 2022) aggregated across counties [1].
2. PNAS peer‑reviewed county series through August 2022
A separate peer‑reviewed PNAS study extended monthly Bayesian county‑level estimates through August 2022, estimating about 1.19 million excess natural‑cause deaths from March 2020 to August 2022 and noting that a nontrivial portion of those excess natural‑cause deaths were not recorded as COVID‑19 on death certificates (estimated ~162,886 unreported to COVID‑19, with wide posterior interval) [2].
3. Aggregated peer‑reviewed tallies for 2020–2021 and methodological variations
Multiple peer‑reviewed and near‑peer outputs converge that the first two pandemic years saw between roughly 0.9 and 1.1 million excess deaths depending on baseline and model choice: one county‑level hierarchical estimate reports ~936,911 excess deaths in 2020–2021 with ~171,168 not assigned to COVID‑19; another spatial Bayesian model gives annual national estimates of ~463,187 excess deaths in 2020 and ~544,105 in 2021 (point estimates and uncertainty intervals are reported) [5] [6]. Differences reflect model training periods, geographic granularity, and how expected deaths are projected [5] [6].
4. National agencies and public datasets for 2022–2024: useful but not always peer‑reviewed
For monthly series extending into late 2022–2024 researchers and analysts commonly rely on public datasets and processed series (Our World in Data, Human Mortality Database, World Mortality Dataset) and CDC excess‑death tools; these sources provide continuous monthly comparisons through 2024 but are methodological products or curated datasets rather than traditional peer‑reviewed journal articles (Our World in Data documents its use of Karlinsky & Kobak baselines and HMD/WMD inputs) [3] [7] [8] [4].
5. Why peer‑reviewed monthly series stop in 2022 and why numbers diverge
Peer‑reviewed monthly county analyses commonly end in Feb–Aug 2022 because authors used historical training windows that stop before later pandemic waves and because manuscript review and publication lag mean newer months appear first in datasets and working papers; divergences across peer‑reviewed estimates stem from baseline construction, model family (hierarchical Bayesian vs regression), training years used to project expected deaths, and decisions about inclusion of provisional data and reporting lag adjustments [1] [2] [4].
6. What is solid, what remains uncertain, and the reporting tradeoffs
What is robust: multiple independent, peer‑reviewed county‑level studies agree that excess mortality in 2020–2021 was large (hundreds of thousands of deaths) and that a measurable fraction of excess deaths were not assigned to COVID on certificates [1] [5] [2]. What is uncertain: month‑by‑month peer‑reviewed series for 2023–2024 are scarce in the literature provided here, so analysts rely on curated national datasets and CDC adjusted estimates for late‑2022 onward; methodological choices (baseline years, handling of pandemic weeks in model fitting) materially change month‑level excess estimates and can reflect implicit agendas—public‑health groups favoring more conservative baselines versus independent datasets aiming for comparability across countries [4] [3] [9].