Improving myalgic encephalomyelitis population sampling: Applying an online respondent-driven method to address biases in G93.3 register data. — CFSMEATLAS
Improving myalgic encephalomyelitis population sampling: Applying an online respondent-driven method to address biases in G93.3 register data.
Kielland, Anne, Liu, Jing, Tyldum, Guri et al. · Journal of health psychology · 2026 · DOI
Quick Summary
This study looked at how well official medical records capture ME/CFS cases and who gets diagnosed. Researchers found that medical register data misses many cases and disproportionately underrepresents people from lower-income backgrounds. They used a new online method to find ME/CFS patients more accurately and checked whether official diagnoses matched recognized diagnostic criteria.
Why It Matters
Accurate data about who has ME/CFS and their characteristics is essential for understanding disease burden, directing research resources, and advocating for patient needs. This study reveals that official diagnoses systematically undercount cases in vulnerable populations, which has implications for clinical recognition, research recruitment, and health equity in ME/CFS care.
Observed Findings
G93.3 register data significantly underrepresent people from socially deprived backgrounds compared to those meeting Canada Consensus Criteria
Not all individuals with G93.3 codes would meet validated ME/CFS diagnostic criteria
Sociodemographic factors (including social deprivation) significantly predicted G93.3 status independent of medical factors
Online respondent-driven sampling identified ME/CFS cases missed by traditional health register approaches
Inferred Conclusions
Health register code G93.3 alone cannot serve as an unbiased sampling frame for ME/CFS research or prevalence estimation
Selection bias in diagnostic coding creates systematic underrepresentation of socially disadvantaged ME/CFS patients
Novel sampling methodologies are necessary to obtain representative data on ME/CFS populations and demographics
Validated diagnostic algorithms should be applied to register data to verify case definitions
Remaining Questions
What specific barriers prevent people from lower socioeconomic backgrounds from receiving G93.3 diagnoses—healthcare access, clinician awareness, or other factors?
How do prevalence estimates change when using register-independent sampling methods versus traditional health register approaches?
What This Study Does Not Prove
This study does not prove that socioeconomic status causes ME/CFS, only that diagnostic coding is biased by social deprivation. It also does not establish prevalence figures for ME/CFS overall—it identifies methodological problems in obtaining unbiased prevalence data. The findings reflect diagnostic and registration patterns, not necessarily true disease distribution.