Taquet · The lancet. Psychiatry 2021 · retrospective propensity score-matched cohort study · n=236379

6-month neurological and psychiatric outcomes in 236 379 survivors of COVID-19: a retrospective cohort study using electronic health records.

Cited 2271 times in the scientific literature.

Level 3 - non-randomized controlled study

Retrospective propensity score-matched cohort study using electronic health records

PubMed 33836148 · doi:10.1016/S2215-0366(21)00084-5 · record verified 2026-08-30

What was done

Using the TriNetX electronic health records network comprising over 81 million patients, researchers conducted a retrospective cohort study and time-to-event analysis in 236,379 patients aged over 10 years who survived a confirmed diagnosis of COVID-19 on or after January 20, 2020. The COVID-19 cohort was compared using Cox proportional hazards modeling to propensity score-matched cohorts diagnosed during the same timeframe with influenza or other respiratory tract infections, as well as four non-respiratory comparator conditions. The primary outcomes were the 6-month incidence and hazard ratios (HR) for 14 neurological and psychiatric diagnoses (including stroke, dementia, intracranial hemorrhage, parkinsonism, anxiety, psychosis, and mood disorders), stratified by illness severity proxies (hospitalization, ITU admission, and encephalopathy).

What was found

The estimated 6-month incidence of any neurological or psychiatric diagnosis after COVID-19 was 33.62% (95% CI 33.17–34.07), with 12.84% (95% CI 12.36–13.33) receiving their first such diagnosis. In patients admitted to an ITU, 6-month incidence rose to 46.42% (95% CI 44.78–48.09) for any diagnosis and 25.79% (95% CI 23.50–28.25) for a first diagnosis. Specific estimated incidences overall versus ITU-admitted were: anxiety disorder 17.39% (95% CI 17.04–17.74) vs 19.15% (95% CI 17.90–20.48); psychotic disorder 1.40% (95% CI 1.30–1.51) vs 2.77% (95% CI 2.31–3.33); ischemic stroke 2.10% (95% CI 1.97–2.23) vs 6.92% (95% CI 6.17–7.76); intracranial hemorrhage 0.56% (95% CI 0.50–0.63) vs 2.66% (95% CI 2.24–3.16); dementia 0.67% (95% CI 0.59–0.75) vs 1.74% (95% CI 1.31–2.30); and parkinsonism 0.11% (95% CI 0.08–0.14) vs 0.26% (95% CI 0.15–0.45). Diagnostic rates were higher following COVID-19 than matched influenza (any diagnosis HR 1.44, 95% CI 1.40–1.47; first diagnosis HR 1.78, 95% CI 1.68–1.89) and other respiratory infections (any diagnosis HR 1.16, 95% CI 1.14–1.17; first diagnosis HR 1.32, 95% CI 1.27–1.36). Severe COVID-19 requiring ITU care further amplified risks compared to non-ITU cases (any diagnosis HR 1.58, 95% CI 1.50–1.67; first diagnosis HR 2.87, 95% CI 2.45–3.35).

Why it matters

This large-scale study demonstrates substantial post-acute neurological and psychiatric morbidity following COVID-19 infection, showing elevated relative risks compared to other common respiratory infections that scale steeply with initial acute illness severity.

Limits

The study relies entirely on diagnostic coding within electronic health records, which may miss undiagnosed cases, misclassify diagnoses, or reflect diagnostic billing practices rather than validated clinical assessments. Potential surveillance and ascertainment bias exist if COVID-19 patients received more rigorous clinical follow-up than control patients. The analysis is limited to a 6-month follow-up window and restricted to patients surviving until December 13, 2020, potentially introducing survivor bias regarding early acute fatal outcomes.

Cited by