Three-year outcomes of post-acute sequelae of COVID-19

Miao Cai; Yan Xie; Eric J. Topol; Ziyad Al‐Aly

doi:10.1038/s41591-024-02987-8

Home SearchPaper Details

Three-year outcomes of post-acute sequelae of COVID-19

Miao Cai, Yan Xie, Eric J. Topol, Ziyad Al‐Aly

Nature Portfolio (2024) • Volume 30, Issue 6, Pages 1564-1573

Empirical-ClinicalObservational-CohortPDF AvailableGrade Eligible⚠️ Moderate Risk Flags

Overall Assessment

Adequate Methodological Quality

Assessment created by PaperScorers Medical AI v0.1.0 on Dec 18, 2025

C+

67/100

Key Takeaways

•Risks decline over 3 years but persist for hospitalized patients, including death (year-3 IRR 1.29).
•Non-hospitalized show small residual PASC burden by year 3 (9.6 DALYs/1,000).
•Multiplicity uncorrected across 80 outcomes; extensive sensitivity checks and negative control support robustness.

Conclusion

Robust retrospective VA cohort shows durable but diminishing long-COVID risks; persistence is substantial after severe acute disease.

Quick Actions

Read Full Paper

Quality Dimensions

Bias & Integrity

COI handling, outcome switching, selective reporting

C+68

Methods Rigour

Protocol clarity, sampling, controls

B-74

External Validity

Generalisability, setting, population

D+52

Transparency & Reproducibility

Data/code/materials availability, preregistration

C60

Significance & Novelty

Contribution vs prior art; context if known

B+80

Statistical Validity

Models, assumptions, multiplicity, CI priority

C+66

Integrity & Transparency

Integrity checks

P-hacking risk?

?

Review recommended

Signs of selective hypothesis testing or analysis choices that could inflate false positive rates.

Outcome switching?

?

Review recommended

Evidence that primary outcomes were changed during or after the study, potentially distorting results.

Conflict of interest?

?

None suspected

Authors have financial or other relationships that could bias research findings.

Data integrity issues?

?

None suspected

Concerns about the accuracy, completeness, or authenticity of the reported data.

Open science signals

Research data

?

Restricted

Raw data used in the analyses is publicly available or accessible upon request.

Analysis code

?

Open access

Code or scripts used to analyze the data are shared for reproducibility.

Study materials

?

Not applicable

Protocols, questionnaires, and other materials are publicly available.

Premise

Primary Research Question

Do death and PASC risks persist up to 3 years after SARS-CoV-2, and how do they vary by hospitalization status?

Hypothesis

Risks decline over time but may persist, especially after severe (hospitalised) acute COVID-19.

Hypothesis is Falsifiable

✓Yes

Incident outcomes with defined time windows vs controls allow refutation.

PICO Framework

Population✓Yes

VA users alive at T0, with prior VA use; infection in 2020.

Intervention✓Yes

Exposure = SARS-CoV-2 infection; stratified by hospitalization.

Comparison✓Yes

Uninfected VA users with matched T0 distribution; IPW applied.

Outcome✓Yes

Death and 80 prespecified PASC outcomes; DALYs computed.

Literature Positioning

Literature Review Balanced

Well covered

Extensive prior work cited across organ systems.

Evidence: Refs 1–33; p. 1569–1573

Research Gap Clearly Stated

Well covered

Unclear extent of risks in third year; longer follow-up limited.

Evidence: Abstract; Introduction p. 1564–1565

Stated Contribution Clear

Well covered

Comprehensive 3-year risk and DALY estimates by care setting.

Evidence: Abstract; Methods/Results

Study Provenance

Peer reviewed venueFunding disclosedConflicts disclosed3 affiliations listed

Authors & Affiliations

VA St. Louis Health Care System; Washington University in St. Louis; Scripps Research.

Funding Statement

Funded by US Department of Veterans Affairs; funder had no role.

Conflicts of Interest

Z.A.-A. consulting for Gilead; Y.X. consulting for Guidepoint; uncompensated consulting for Pfizer; others no COI.

Peer Review

✓Nature Medicine, published 30 May 2024.

Evidence: DOI: 10.1038/s41591-024-02987-8

Methodological Assessment

Observational StudySTROBEObservational

12 applicable checklist items.

Yes 10 · No 1 · Unclear 1

Study Design & Definitions

✓

Participant Selection Described

VA cohort construction and flow described; eligibility clear.

Evidence: Methods Cohort; Ext. Data Fig.5/6

✓

Variable Definitions Clear

80 outcomes via ICD-10, meds, labs; organ-system composites.

Evidence: Methods Outcomes

✓

Confounding Addressed

IPW with many covariates; HDPS sensitivity; doubly robust model.

Evidence: Methods Statistical analyses; Sensitivity (1),(4)

✓

Temporality Established

Incident outcomes from 30 days after T0 across three windows.

Evidence: Methods Statistical analyses

✓

Exposure Definition Anchored

Exposure via PCR/antigen; controls assigned T0 from case distribution.

Evidence: Methods Cohort

Bias Mitigation

✓

Balance Diagnostics Acceptable

SMD <0.1 after weighting across subcohorts.

Evidence: Extended Data Fig.1

✓

Selection Bias Addressed

VA-user design, IPW, competing risks/censoring addressed.

Evidence: Methods Statistical analyses; Sensitivity (8),(9)

✓

Misclassification Assessed

Acknowledged possible undiagnosed infection; negative control used.

Evidence: Negative outcome control analyses p. 1569

✓

Immortal Time Bias Mitigated

Follow-up starts at 30d post T0 to avoid acute phase bias.

Evidence: Methods Statistical analyses

✗

Time-varying Confounding Handled

No explicit time-varying confounding modelling.

Evidence: Methods Statistical analyses

Robustness

✓

Negative Controls or E-value

Incident neoplasm negative control near-null across years.

Evidence: p. 1569 Negative outcome control analyses

?

Model Diagnostics Reported

Model fit diagnostics not reported; zero-inflated checked.

Evidence: Sensitivity analyses (2)

Modules without applicable checklist items (9)

These designs were fully marked as not applicable but remain available for reference.

Quick Actions

Read Full Paper

Study Overview

Study Design

Observational Cohort

Primary classification used to evaluate this research.

Population

US Veterans with SARS-CoV-2 in 2020 and VA users without infection; followed 3 years; non-hospitalized and hospitalized strata.

Brief summary of who or what was studied.

Sample Size

5,341,996

Number of participants included in the study analyses.

Study Arms

3 arms

Total number of intervention or comparison groups.

Pre-registration

Not registered

Indicates whether the protocol or trial was registered before data collection.

Randomisation

None reported

How participants were allocated to groups: None (no randomisation), Simple (random order), Blocked (groups balanced in blocks), Stratified (balanced within subgroups), Adaptive (allocation adjusted based on previous enrollments).

Follow-up

3 years (1,080 days)

Duration of observation or follow-up for outcomes.

Primary Outcome

Primary outcome was pre-specified in the protocol or registry.

Effect Size

IRR: Death year-3, hospitalized: IRR 1.29 (95% CI 1.19–1.40)

Reviewer Notes

Retrospective cohort using VA EHR; IPW and GEE Poisson; infected vs uninfected controls by acute care setting.

Publication Details

DOI

10.1038/s41591-024-02987-8

Published

May 30, 2024

Citations

161 citations

FWCI

123.67

External Resources

DOI Link OpenAlex

Disclaimer: This assessment is generated by AI and should not be the sole basis for clinical or research decisions. Always review the original paper and consult with domain experts.

Suggested Papers

C

Large language models encode clinical knowledge

Karan Singhal et al.•2023

D

How iPS cells changed the world

Megan Scudellari•2016

No Report

cGAS-STING Signaling Regulates Initial Innate Control of Cytomegalovirus Infection

Chan‐Wang Jerry Lio et al.•2016

Suggested Papers

From Our Blog

How AI is Changing Peer Review: The Future of Science

AI will not replace scientists. But it will replace scientists who do not use AI. Here is how algorithms are fixing peer review.

Composite Endpoints: Mixing Apples and Oranges

Drug companies love to combine death, heart attacks, and chest pain into one score. It hides the truth.

Selection Bias: Who Are You Studying?

If your participants are not representative, your results are worthless. Selection bias ruins studies before they start.