Methodologies used in cost-effectiveness models for evaluating treatments in major depressive disorder: a systematic review

BackgroundDecision makers in many jurisdictions use cost-effectiveness estimates as an aid for selecting interventions with an appropriate balance between health benefits and costs. This systematic literature review aims to provide an overview of published cost-effectiveness models in major depressive disorder (MDD) with a focus on the methods employed. Key components of the identified models are discussed and any challenges in developing models are highlighted.MethodsA systematic literature search was performed to identify all primary model-based economic evaluations of MDD interventions indexed in MEDLINE, the Cochrane Library, EMBASE, EconLit, and PsycINFO between January 2000 and May 2010.ResultsA total of 37 studies were included in the review. These studies predominantly evaluated antidepressant medications. The analyses were performed across a broad set of countries. The majority of models were decision-trees; eight were Markov models. Most models had a time horizon of less than 1 year. The majority of analyses took a payer perspective. Clinical input data were obtained from pooled placebo-controlled comparative trials, single head-to-head trials, or meta-analyses. The majority of studies (24 of 37) used treatment success or symptom-free days as main outcomes, 14 studies incorporated health state utilities, and 2 used disability-adjusted life-years. A few models (14 of 37) incorporated probabilities and costs associated with suicide and/or suicide attempts. Two models examined the cost-effectiveness of second-line treatment in patients who had failed to respond to initial therapy. Resource use data used in the models were obtained mostly from expert opinion. All studies, with the exception of one, explored parameter uncertainty.ConclusionsThe review identified several model input data gaps, including utility values in partial responders, efficacy of second-line treatments, and resource utilisation estimates obtained from relevant, high-quality studies. It highlighted the differences in outcome measures among the trials of MDD interventions, which can lead to difficulty in performing indirect comparisons, and the inconsistencies in definitions of health states used in the clinical trials and those used in utility studies. Clinical outcomes contributed to the uncertainty in cost-effectiveness estimates to a greater degree than costs or utility weights.