Pyne JM (REAP - Little Rock), Martin BC
(University of Arkansas for Medical Sciences), Bodhani A
(University of Arkansas for Medical Sciences)
Objectives:
Genetic testing to determine metabolic genotypes and personalize pharmacotherapy for depression is in the early stages of implementation however it is unknown when these evolving technologies are cost-saving or cost-effective. The objective of this presentation is to estimate the cost-effectiveness of metabolic genetic testing compared to no testing.
Methods:
A deterministic decision analysis model which was inspired by a model developed by Sullivan et al, 2004 was created to estimate the cost effectiveness of initiating antidepressant therapy. The model was built using Tree Age Pro and validated with the original Sullivan model inputs and adapted to model three treatment strategies: initiate treatment with paroxetine; initiate treatment with citalopram; and conduct a genetic test to test CYP2D6 polymorphisms, where fast metabolizers are prescribed paroxetine and slow metabolizers are prescribed citalopram. The model used a 6 month time horizon to incorporate all costs and utilities. Model inputs were obtained from published sources and clinical assumptions. Since polymorphisms of the CYP2D6 vary across race, race specific cost effectiveness estimates were estimated. One way sensitivity analyses were conducted to explore the robustness of the findings.
Results:
The preliminary base case results show that initiating therapy with citalopram was expected to cost $3,790 and produce 0.378 QALYs over the 6-month time horizon, and this option dominated the other two strategies. When the cost of citalopram exceeded $120/month, gene testing became the most cost-effective strategy using an ICER threshold of $50,000/QALY. When the clinical decision was simplified to two treatment strategies, initiate paroxetine or the gene testing strategy, the gene testing strategy dominated the no-gene testing strategy up until the cost of genetic testing exceeded $100/patient.
Implications:
Because citalopram has a lower adverse event rate than paroxetine with similar monthly costs, it is the preferred agent across a range of sensitivity analyses. Metabolic genetic testing may be a cost-effective strategy in some clinical situations, particularly when adverse event rates vary with genetic polymorphisms, and where there may be substantial cost differences between drugs.
Impacts:
Cost-effectiveness modeling studies will be an important tool to identify situations when genetic testing can be routinely implemented in clinical practice.