Back

Underdiagnosis of myalgic encephalomyelitis/chronic fatigue syndrome-like illness in a large integrated healthcare system -- Kaiser Permanente Northern California, 2022-2023

Wood, M.; Halmer, N.; Bertolli, J.; Amsden, L. B.; Nugent, J. R.; Lin, J.-M. S.; Rothrock, G.; Nadle, J.; Chai, S. J.; Champsi, J. H.; Yang, J.; Unger, E. R.; Skarbinski, J.

2024-12-06 epidemiology
10.1101/2024.12.04.24318508 medRxiv
Show abstract

BackgroundSurveillance of myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS), a chronic, debilitating multisystem illness, is challenging because ME/CFS can be under-recognized in healthcare settings. MethodsUsing a population-based panel study of 9,820 adult members of Kaiser Permanente Northern California (KPNC), a large, integrated healthcare system, we compared survey-defined ME/CFS-like illness with presence of an ME/CFS diagnosis in the electronic health record (EHR) to evaluate ME/CFS underdiagnosis. ResultsOf those with survey-defined ME/CFS-like illness, an estimated 97.8% (95% confidence interval [CI] 97.1%-98.4%) did not have an ME/CFS diagnosis in the EHR. The group without EHR diagnosis was younger, less likely to identify as white non-Hispanic, and more likely to have developed fatigue in the past three years than the EHR diagnosed group. Both diagnosed and undiagnosed ME/CFS-like illness groups had significantly impaired physical, cognitive, and social functioning, and significantly worse mental health and anxiety than those without ME/CFS-like illness. ConclusionME/CFS is underdiagnosed in the Kaiser Permanente Northern California healthcare system. Enhanced syndromic surveillance that characterizes patients with ME/CFS who have not been diagnosed has the potential to increase timely recognition of ME/CFS.

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.