Estimating pain visual analogue scale from health assessment questionnaire for rheumatoid arthritis with beta mixture models

Area of Science:

• Rheumatology
• Biostatistics
• Health Economics

Background:

• Rheumatoid arthritis (RA) management requires understanding patient-reported outcomes.
• Health Assessment Questionnaire disability index (HAQ) and pain visual analogue scale (VAS) are key metrics.
• Accurate mapping between HAQ and pain VAS is needed for evidence synthesis and resource allocation.

Purpose of the Study:

• To develop and validate a statistical model to predict pain visual analogue scale (VAS) scores from the Health Assessment Questionnaire disability index (HAQ) in rheumatoid arthritis patients.
• To support resource allocation and evidence synthesis by providing reliable pain estimations based on disability data.

Main Methods:

• Utilized beta mixture models with HAQ, its square, age, and sex as predictors.
• Employed Bayesian Information Criteria for model selection and k-fold cross-validation for performance estimation.
• Validated the preferred model using an independent Rheumatoid Arthritis Medication Study cohort.

Main Results:

• A two-component beta mixture model with HAQ as the main predictor demonstrated good fit.
• Visual plots showed similar cumulative distributions for observed and predicted pain VAS values.
• External validation confirmed the model's robust performance on an independent dataset.

Conclusions:

• Beta mixture models accurately provide non-linear estimates of pain VAS from HAQ scores in rheumatoid arthritis.
• This mapping facilitates improved evidence synthesis and decision-making for patient care and resource allocation.