OBJECTIVES: Recent evidence has demonstrated a link between disease severity in rheumatoid arthritis and overall non-drug costs to the healthcare payer.  This analysis aims to estimate the likely lifetime cost savings associated with achieving better clinical response in patients treated with etanercept, compared against rituximab after the failure of two previous DMARD therapies. METHODS: An existing economic model was repopulated using data linking resource use with disease severity, as measured by the Health Assessment Questionnaire (HAQ).  The model uses Markov techniques to predict the lifetime health outcomes associated with different treatments for patients with RA in the United Kingdom.  For each treatment in the model, the initial (first six months), medium-term (six to thirty-six months) and long-term (after thirty-six months) effects on the Health Assessment Questionnaire (HAQ) score were calculated.  HAQ scores at each time period determined each patient’s utility (QALYs), resource use and mortality. Effectiveness data were derived from the results from the TEMPO trial and published literature.  Utility scores were converted from HAQ data using the EQ-5D.  Unit cost data were drawn from established national databases, and were multiplied by resource use to predict the total cost for each cohort. RESULTS: When the impact of HAQ on resource was not included in the model, the incremental cost of the etanercept arm was estimated to be £4648 over the patient’s lifetime.  When resource use was linked with response to treatment, the incremental cost fell to £3304 (a reduction of 29%).  Because the incremental effectiveness of etanercept would remain unchanged, the incremental cost-effectiveness ratio would also be reduced by 29%. CONCLUSIONS: By incorporating a link between HAQ and resource use into an existing economic model, it has been established that a substantial proportion of a treatment’s incremental cost may be offset by reduced use of resources in other areas.