Overview
- The study, published May 27 in JAMA Network Open, analyzed 457,950 electronic health records from 58 hospitals and estimated 16.3% of people who tested positive for COVID-19 developed long COVID.
- Researchers used a new “precision‑phenotyping” AI algorithm to identify new chronic conditions that appeared after infection and could not be explained by prior medical history.
- Authors say the ICD code for post‑COVID conditions (U09.9) appears in fewer than 7% of identified cases, and they estimate more than 10 million long COVID cases would be missed by code‑based tracking.
- The algorithm flagged a wide range of sequelae — including new prediabetes, cardiac and neurological problems, persistent fatigue, and chronic pain — with rates that varied across New England, Southeast Texas, Southern California and Western Pennsylvania.
- The findings could prompt changes in public‑health surveillance, insurer coverage and targeted clinical trials, but some experts have questioned whether the 16% figure overstates the number of people with persistent, clinically significant disease.