gms | German Medical Science

Introduction: ME/CFS is a severe and complex condition with a long period of suffering for those affected and a delayed official diagnosis [1], [2]. Besides a lack of essential knowledge of the disease, a wide range of symptoms impedes the diagnosability, leading to a delayed, under- and misdiagnosis [3]. We assumed that symptoms indicative of ME/CFS could be detectable in routine data via ME/CFS associated ICD codes before the officially coded diagnosis of ME/CFS. Hence, we used 1:5 matched routine data from the Techniker Krankenkasse (TK) from 2019 to 2022 to investigate potential early ICD-markers of ME/CFS by analyzing differences in ME/CFS-associated ICD codes in young people aged 6 to 27 years with ME/CFS in the year prior the official diagnosis compared to those without.

Methods: In this exploratory approach, we divided the dataset in an 80:20 ratio to first train models and validate in the test data. The training data set included data of 4.869 cases and 24.242 controls matched in age, living area and sex. Potential candidate ICD-10 codes were pre-selected based on clinical expertise. Conditional multiple logistic regression modelling was used to calculate Odds Ratios for later ME/CFS with 95% CI. Combining information from statistical tests of the odds ratios, prevalence levels of the ICD-10 codes (5-10%, ≥10%) and clinical expertise clusters of ICD-10 codes were defined which might serve as early indicator variables.

Results: In the year prior to diagnosis, 59% of cases had at least one ICD-10 code, compared to 39% of the controls. In more than 40% of the cases, compared to 27% of the controls, these were coded at least 7 to 12 months prior to ME/CFS diagnosis. Two ICD clusters were identified: symptoms of ME/CFS and comorbidities of ME/CFS. In more than 10% of the cases, e.g., fatigue (R53; OR 2.65, 95% CI: 2.38-2.94), migraine (G430-39; OR 1.48, 95% CI: 1.32-1.65), or abdominal pain (R104; OR 1.37, 95% CI: 1,22-1,53) were coded, and in 5-10% of the cases, e.g., neuropathy (F480; OR 2.05, 95% CI: 1.71-2.54), dizziness (R42; OR 1.68, 95% CI: 1.54-1.95), asthma (J451, J458, J459; OR 1.42, 95% CI 1.26-1.60) or post-traumatic stress disorder (1,34 (1,20-1,50). Other models suggested that there were four other candidate ICD clusters to consider: infections preceding ME/CFS, increased utilization of laboratory and consultation in the medical system, additional symptoms or conditions with potential, and differential diagnoses of ME/CFS.

Conclusion: We showed that for young people with ME/CFS, symptoms based onICD-10 codes in certain clusters are associated with later ME/CFS and might serve as early indicators, in a period when ME/CFS itself is not yet diagnosed. These ICD-codes with increased OR reflect the clinical picture and precursors of ME/CFS and thus could be considered as potential markers to identify potential ME/CFS cases not yet diagnosed in routine data. These preliminary results will be completed by additional models and will finally be validated using also the test data set. Combinations of the identified ICD clusters and their potential for the development of predictive models will be explored.

The authors declare that they have no competing interests.

The authors declare that a positive ethics committee vote has been obtained.