Estimating Prevalence, Demographics, and Costs of ME/CFS Using Large Scale Medical Claims Data and Machine Learning.
Level 4 - case-series / case-control
Retrospective cross-sectional database analysis and predictive modeling using administrative claims data.
PubMed 30671425 · doi:10.3389/fped.2018.00412
What was done
Techniques of data mining and machine learning were applied to a large database of medical and facility claims from commercially insured US patients to determine prevalence, gender demographics, and annual medical costs for patients with provider-assigned diagnosis codes for myalgic encephalomyelitis (ME) or chronic fatigue syndrome (CFS). Costs were compared to patients with lupus, multiple sclerosis (MS), and the general insured population. A machine learning model was developed to predict prevalence, and researchers assessed whether claims-based symptom codes could identify cases.
What was found
Provider-assigned diagnosis frequency was 519 to 1,038 per 100,000, extrapolating to an estimated 1.7 million to 3.38 million individuals in the US (2017 population). The relative risk for females compared to males was 1.238 for ME and 1.178 for CFS, with men comprising 35% to 40% of diagnosed patients. The machine learning model predicted a prevalence of 857 per 100,000 (p > 0.01), roughly 2.8 million individuals in the US. Average annual costs for ME or CFS were 50% higher than for lupus or MS, and 3 to 4 times higher than for the general insured population. Identifying cases solely from symptom codes could not be validated due to missing specific codes for core symptoms.
Why it matters
This study shows ME/CFS is a relatively common condition affecting millions of Americans across both sexes, incurring direct medical costs that exceed those of lupus and multiple sclerosis.
Limits
The total sample size (n) was not reported in the abstract. The cohort was restricted to commercially insured patients, excluding uninsured, Medicaid, and Medicare populations. Administrative claims rely on billing codes rather than clinical diagnostic validation, and claims data lacked codes for core ME/CFS symptoms.
Cited by
- supports An estimated 3 to 5 million people in the US, 2 to 3 million in the UK, and 20 to 24 million worldwide have either ME/CFS or long COVID.