COST-UTILITY ANALYSIS OF A NOVEL PHARMACIST-GUIDED WARFARIN PHARMACOGENETIC SERVICE IN 1 TO 5 YEAR TIME HORIZONS
Incorporation of genotype information has been associated with improved warfarin dose prediction. Recently, improved clinical outcomes associated with a genotype-guided approach were demonstrated in a tertiary care setting where clinical pharmacists determined initial warfarin doses based on patient clinical characteristics and genotype (PGx). However, whether PGx testing is economically justifiable is controversial. The objective of this study was to determine the cost-effectiveness of using PGx versus usual care (UC) during a one to five-year time horizon from a healthcare provider perspective.
METHODS
A decision tree depicting warfarin-related complications over a 90-day period for patients newly initiated on warfarin for a diagnosis of either atrial fibrillation or venous thromboembolism was developed. For each chance node, the adjusted hazard ratios of warfarin related complications for PGx versus UC calculated from the cohort follow-up data were implemented. The decision tree was followed by a 90-day cycle Markov model projecting possible scenarios for the patients who have not experienced a fatal event during the decision tree follow-up. Transition probabilities between the Markov states were taken from literature and converted to 90-day estimates. Cost and health-related utility scores of each complication were extracted from publically available sources and adjusted to 2015 values. In order to see the impact of time horizon on the decision as well as the incremental cost-effectiveness ratio (ICER), the model ran for from one to five years with an annual discount rate of 3%. Uncertainty was tested through one-way sensitivity analyses within the range of ±25% of the all base case inputs, scenario analyses using clinical inputs identified from published studies in place of the cohort study results, and probabilistic Monte-Carlo simulations.
RESULTS
Over a 5-year time horizon, each unit of PGx service was associated with 0.0167 additional quality adjusted life-years (QALYs) and only $3.5 dollars in additional costs compared to UC. The corresponding ICER was $210/QALY. The ICER estimates were higher at shorter time horizons, with predicted ICERs of $4,729, $742 and $398 per QALY gained at years 1, 2 and 3, respectively. The ICER increased when we tested the alternative scenario, but differences from the base case scenario were small. Specifically, the ICERs were $1,512 at year 1 and $188 at year 5. Using probabilistic sensitivity analysis, the likelihood of the PGx being accepted as a cost-effective option at the willingness-to-pay (WTP) of $50,000/QALY were 70%, 87% and 91% over 1, 3 and 5-year time horizons, respectively.
CONCLUSION
PGx is associated with slightly increased costs of care and a modest gain in QALYs. The initial cost gap between the treatment arms diminished when a longer patient follow-up was implemented. Hence, at time horizons greater than one year, the PGx was likely to be cost-effective compared to UC at a WTP of $50,000/QALY making it an attractive option for health care systems considering longer time horizons.