A large multi-hospital study published in JAMA Network Open highlights a critical issue in post-COVID-19 research: the true burden of long COVID is significantly higher than what standard health surveillance systems report.
The condition, often referred to as post-acute sequelae of SARS-CoV-2 infection (PASC), continues to affect patients long after initial infection with COVID-19. According to this study, traditional diagnostic coding systems miss a large share of cases, leaving policymakers and healthcare systems with an incomplete picture of ongoing illness.
The findings suggest that long COVID is not a rare complication but a widespread chronic health issue that continues to grow over time.
This retrospective cohort study analyzed data from 457,950 adults diagnosed with COVID-19 across 58 hospitals in four US regions between 2017 and 2025. Researchers used advanced artificial intelligence based phenotyping to identify cases of long COVID more accurately than standard diagnosis codes.
Instead of relying only on billing codes, the researchers applied a machine learning approach called Precision Phenotyping for Research Cohorts (P2RC). This method examined patterns in electronic health records to detect long-term symptoms consistent with PASC.
The study’s key goal was to compare true estimated prevalence with what traditional surveillance systems capture, and to assess whether long COVID is mostly chronic in nature.
One of the most important findings is that approximately 16.28 percent of all COVID-19 patients developed long COVID when measured using AI-based phenotyping. This is more than twice the rate detected through standard diagnostic codes.
Out of 74,560 identified PASC cases, nearly 89.31 percent involved chronic conditions requiring ongoing medical care. This translates to about 1 in 6 COVID-19 patients developing long-term health issues, and about 1 in 7 developing clearly chronic disease.
Even more concerning, the study found that:
The researchers concluded that long COVID is accumulating in the population, not fading away.
The study found that long COVID affects multiple organ systems, with significant variation across regions.
Common symptom categories included:
Endocrine-related effects were particularly notable. Some regions showed thyroid-related issues, while others showed more metabolic problems such as glucose dysregulation and prediabetes-like conditions.
This variation suggests that long COVID may not be a single condition, but rather a collection of overlapping syndromes affecting different organs.
A key insight from the study is that 67.27 percent of ICD-10 codes linked to long COVID represent chronic or potentially chronic conditions. Only a small fraction were acute and self-limiting.
Among all COVID-19 patients:
These findings suggest that long COVID should be treated as a chronic disease burden, not just a temporary post-viral syndrome.
Conditions such as diabetes, thyroid disorders, and neurological symptoms were frequently observed in long COVID patients, indicating long-term systemic effects.
A major concern highlighted in the study is that standard surveillance systems dramatically underestimate long COVID cases.
Traditional coding systems rely on diagnostic labels, such as U09.9, but previous research shows these codes capture fewer than 1 in 10 cases in many settings.
In contrast, AI-based phenotyping reveals a much larger and more accurate picture.
The study shows that:
This creates a serious gap between clinical reality and public health reporting.
The study also tracked long COVID trends from 2020 to 2024 and found a steady increase in cumulative prevalence.
Key observations include:
This suggests that long COVID is not only persistent but expanding as new COVID-19 infections continue to occur.
The study found notable differences across US regions:
These differences may reflect population health differences, healthcare coding practices, or biological variation in disease expression.
The researchers suggest that long COVID may include multiple subtypes that require more personalized diagnostic and treatment approaches.
This research is important for several reasons:
If these findings are confirmed in other studies, healthcare systems may need to rethink how they track and manage post-COVID conditions.
The authors also note several limitations:
Despite these limitations, the large sample size and multi-region design strengthen the overall findings.
This large-scale analysis suggests that long COVID is not a rare complication but a widespread chronic health condition affecting millions of people.
Approximately 1 in 6 COVID-19 patients may develop long-term symptoms, and most of these involve chronic disease requiring ongoing care. Current surveillance systems appear to miss a substantial portion of these cases, leading to underestimation of the true public health burden.
The study supports a shift in perspective: long COVID should be treated as a chronic disease spectrum requiring long-term monitoring, improved diagnostic tools, and integrated healthcare pathways.
Without improved surveillance and care systems, the burden is likely to continue growing in the coming years.
Tian J, Azhir A, Decaro M, et al. Long COVID Persistence and Surveillance Gaps Across 58 US Hospitals. Published in JAMA Network Open, 2026;9(5):e2614909. doi:10.1001/jamanetworkopen.2026.14909
This article is a summary based on a published academic study. It is intended for informational and educational purposes only and does not constitute medical advice. Readers should consult qualified healthcare professionals for diagnosis, treatment, or medical guidance related to COVID-19 or long COVID conditions.
Most Accurate Healthcare AI designed for everything from admin workflows to clinical decision support.