OBJECTIVES: To perform cost-effectiveness analysis (CEA) for routine use of a transparent dressing integrating a chlorhexidine gluconate (CHG)-containing gel pad versus standard transparent dressings, with a classical decision tree model and a Non-Homogeneous Markov Model (NHMM) previously developed.  METHODS: Clinical efficacy data was extracted from a multicentre randomized controlled trial (RCT) with 1,879 patients and economical data obtained from micro and macro-costing published studies. The baseline method is a NHMM previously developed in Microsoft Excel® with VBA using the same data sources. The decision tree was built with the TreeAge Pro® software 2013. One-way deterministic (DSA) and probabilistic sensitivity (PSA) analyses were conducted on key clinical and economic parameters.  RESULTS: Based on the decision-tree model, the CHG-dressing is a dominant strategy compared to standard dressings. The intervention prevents 13.5 infections per 1,000 patients and saves €157 per patient. These results are robust across a range of values for several parameters in DSA. The PSA with the NHMM resulted in 11.8 infections avoided per 1,000 patients (95%CI: [3.85; 19.64)]) and a mean extra cost of €141 per patient (95%CI: [€-975; €1,258]) when using antimicrobial dressing. Effectiveness as calculated by both models is similar while cost estimations diverge.  CONCLUSIONS: Decision-tree and the NHMM are structurally different and even though their outcomes cannot be directly compared, they were coherent. The decision-tree model indicates that CHG-dressings are cost-saving and a dominant preventative strategy for CRBSIs. The Markov model supports cost-effectiveness compared to standard dressing. The main disadvantages of the decision-tree are the inability to integrate changes among health states during the ICU stay and to simulate possible observable trajectories in the patient history. The structure of the non-homogeneous Markov model does not allow DSA for the incidence of the disease.