Objective: This study updated a 2001 decision economic model that used indirect data and confirmed its findings by developing a new cost-effectiveness model by using now available head-to-head data. The models compared olanzapine with ziprasidone in the treatment of schizophrenia in the United States. Methods: A decision analytic modeling approach was used to estimate annual health-care costs and health outcomes, incorporating events such as response, relapse, and suicide. Patients without response to first-line treatment switched to the other comparator. Decision tree probabilities were extracted from head-to-head studies and other published clinical literature. Direct health-care costs and quality-adjusted life-years (QALYs) were estimated on the basis of resource use and utility weights for initial and relapse episodes, maintenance therapy, and extended episodes of illness. Disutilities associated with treatment-emergent adverse events were included. Results: Consistent with the 2001 model, this model found that first-line treatment with olanzapine is associated with fewer hospital days, fewer days with extrapyramidal symptoms, and higher QALYs than is first-line treatment with ziprasidone. Total costs were lower for the olanzapine pathway ($70,232-$72,776 vs. $73,086-$73,310 in the Positive and Negative Syndrome Scale analysis) due to the cost savings associated with reduced health-care resource use. The incremental cost per QALY gained indicated that the olanzapine pathway dominated the ziprasidone pathway. Conclusions: Decision analytic models should be continuously assessed against new data. This case study shows that incorporating new data confirmed results of a previously published model in which olanzapine was associated with better expected health outcomes and lower total health-care costs than was ziprasidone.
