OBJECTIVES: To compare the cost-effectiveness of prolonged sedation with dexmedetomidine compared with propofol and midazolam for mechanically ventilated adult patients in Colombia.

METHODS: A Markov model was designed to capture the duration of mechanical ventilation, intensive care unit (ICU) length of stay and delirium episodes of dexmedetomidine, propofol and midazolam. The time horizon of the model was 376 hours. The hourly probabilities to move from one state to another were estimates using Bayesian mixed treatment comparison and the data was obtained from previously published prospective, randomized, double-blind trials. Effectiveness measure was the patients ICU length of stay at the end of the follow-up period. The ICU costs were determined daily and the medication costs were calculated as the number of hours each patient remained intubated. The used exchanged rate was COP$ 2.957 per USD$ 1. Probabilistic sensitivity analysis was performed.

RESULTS: Our base-case analysis demonstrated that the median time to extubation with dexmedetomidine was shorter, 92.55 hours (CrI95% 86.83, 98.31), compared to midazolam 129.4 hours (CrI95%121.2, 137.4) and 116.6 hours (CrI95% 90.89, 142) with propofol. In terms of median ICU length of stay with dexmedetomidine was 7.7 days (CrI95% 6.5, 8.8) with midazolam was 8.8 days (CrI95% 6.7, 11.05) and 8.6 days (CrI95% 4.1, 12.91) with propofol. The model indicated that midazolam is a dominated by dexmedetomidine and propofol. The incremental cost-effective ratio (ICER) for dexmedetomidine was USD$ 5.60 with propofol. The probabilistic analysis demonstrated that dexmedetomidine dominated midazolam in all simulations and with a willingness to pay above USD$ 100 dexmedetomidine have a higher probability to be cost effective compared to propofol (52.9%).

CONCLUSIONS: The model suggests that dexmedetomidine is cost effective due to their ability to reduce ICU length of stay, duration on mechanical ventilation and occurrence of delirium; this finding is robust with sensitivity analyses conducted.