Capturing Information About Multiple Sclerosis Comorbidity Using Clinical Interviews and Administrative Records: Do the Data Sources Agree?

Background/Objectives: Multiple sclerosis (MS) is often associated with comorbidities that affect clinical outcomes. Data on comorbidities can be sourced from self-reports, medical records, and administrative databases. The gold standard for collecting such data is prospective clinical collection, as in clinical trials, but this is not feasible in large epidemiological studies. This study aimed to assess the agreement between two data sources, clinical interviews and administrative records, identifying major comorbidities in people with MS (pwMS). Methods: We evaluated the agreement between clinical interview data and administrative records in pwMS enrolled at two sites (2021–2022). Seven comorbidities were investigated: depression, anxiety, diabetes, hypertension, autoimmune disease, chronic lung disease, and hyperlipidemia. We used kappa (κ), sensitivity, specificity, and predictive values to assess agreement. Results: The frequency of comorbidities varied between the sources. Administrative data often underestimated hypertension, autoimmune diseases, hyperlipidemia, and anxiety, but over-reported depression. It had high sensitivity for diabetes (80%) and moderate sensitivity for hypertension (62%). The agreement for diabetes (κ = 98.9%, PABAK = 0.98, positive agreement = 83.3%) and hypertension (κ = 89.8%, PABAK = 0.80, positive agreement = 70.8%) was high. Conclusions: The agreement between administrative data and clinical interviews was excellent for diabetes and hypertension. For other conditions, such as psychiatric, hyperlipidemia, and autoimmune comorbidities, administrative data had lower sensitivity, and often under-reported or misclassified the data.