Long COVID may be shorter and narrower than earlier studies suggested, huge Stanford study finds

Long COVID may cause fewer long-term health problems, for a shorter time, than earlier research suggested, a huge new Stanford University study finds. Researchers analysed 14.4 billion health insurance claims from 244.7 million Americans; the findings were published in Proceedings of the National Academy of Sciences (PNAS) on 27 October 2026.

The problem, the team says, is how earlier studies were designed: most compared people who tested positive with people never tested. But people who get tested differ from those who do not — they may have other conditions, or see a doctor more often — which can make COVID-19 look like the cause of problems there for other reasons.

To fix this, the team studied only people at their first PCR test, between March 2020 and March 2021, comparing those who tested positive with those who tested negative, and matched patients with similar histories using a synthetic-control method. Over two years and 614 outcomes, the stricter method found far fewer effects: 262 fell to 43 between 30 and 119 days after infection; 286 to five between 120 and 359 days; and 369 to zero between 360 and 719 days.

The team also tested outcomes COVID-19 could not cause, such as lightning strikes and dog bites. The old approach wrongly linked those to COVID-19; the new one did not. Effects it did confirm included muscle disease, non-tuberculosis pneumonia, respiratory problems and a collapsed lung. Long COVID is not necessarily less common, the team stresses — but the risks blamed on COVID-19 fade faster than earlier estimates suggested. Insurance records also miss symptoms such as fatigue and brain fog.

Why it matters

If the findings hold up, earlier long COVID studies may need revisiting with better controls — a reminder that in big-data research, size alone is not enough. Co-senior author Jure Leskovec put it plainly: “Even with 244 million patients, you can get highly precise answers to the wrong question. Study design matters as much as data size.”

What happens next

The team says its test-based design could re-check earlier long COVID studies and improve future population-health research. For doctors and patients, little changes: the authors say the results should complement a doctor’s clinical assessment, not replace it, and people with ongoing symptoms should still seek medical care. Lead author Hamed Nilforoshan’s team also published its analysis code openly so others can test the method.

This article is for general information and education only. It is not legal or financial advice.

Sources

• https://news.stanford.edu/stories/2026/10/long-covid-health-affects-study-design (Stanford University, 27 October 2026)
• https://medicalxpress.com/news/2026-10-analysis-million-americans-refines-covid.html (Medical Xpress, 28 October 2026)
• https://www.scimex.org/newsfeed/long-covid-symptoms-and-duration-revised (Scimex / PNAS, 28 October 2026)