OBJECTIVES: Coffee (Coffea arabica) contains over 1,000 distinct molecular compounds and is one of the most widely consumed beverages worldwide. Epidemiologic studies have shown an inverse relationship between coffee consumption and all-cause mortality. This analysis aims to assess the cost-effectiveness of coffee from the perspective of the US consumer and payer.

METHODS: A cohort life-table analysis was developed to model life-years (LYs) of US coffee consumers vs non-consumers over a lifetime horizon. Age- and gender-specific mortality rates were used to model survival outcomes. Relative risks of death by average coffee intake (cups/day) were obtained from a recent large, prospective US cohort study. Costs per cup were estimated for home preparation and obtained from a national sample of low- and high-cost vendors. Incremental analyses were conducted by cost, sex, and level of daily coffee consumption. Deterministic (DSA) and probabilistic (PSA) sensitivity analyses were conducted. The model was validated by comparing life expectancy results to a model utilizing relative risks of cancer and chronic diseases based on coffee consumption. RESULTS: Coffee increased undiscounted LYs in 1, 2-3, 4-5, and 6+ cup/day male (0.72, 1.23, 1.48, and 1.22) and female (0.53, 1.42, 1.77, 1.65) consumers, respectively, versus non-consumers. ICERs per discounted LY gained were $5,460, $11,326, $24,254 for males and $5,885, $8,910, $20,145 for females, respectively, for 1, 2-3, and 4-5 cups/day consumption of home-prepared coffee; 6+ cups/day was strictly dominated. Consumption of 4-5 cups per day purchased from high-cost vendors was not cost-effective, with ICERs >$50,000/LY-gained.  DSA showed that coffee effectiveness in preventing death and coffee acquisition cost had the largest impact on ICERs. CONCLUSIONS: In this analysis, coffee consumption was associated with increased LYs and was shown to be potentially cost-effective, especially if home-prepared or purchased from low-cost vendors. Given the observational nature of the study data, further research is warranted to validate these findings.