Outcomes
Only clinical outcomes were used as endpoints in the analysis, as shown in Figure 1. The reasons why certain secondary outcomes were not explicitly included in the decision tree are shown in Appendix S2. While these outcomes were not assigned individual nodes, the use of NHS guidelines (Appendix S3) allowed for the cost of these outcomes to be grouped and used in the cost of treatment complications included in the decision tree.
The data on outcomes was collected in two ways in the ANODE trial. Firstly, the number of people with one of the primary outcomes of suspected maternal infection was recorded 6 weeks after delivery through telephone interviews and hospital records. The secondary outcomes were collected based on a questionnaire also completed at the same time and included their Mean EQ-5D-5L values1.
The EQ-5D-5L score ranges from 0 to 1, with a higher score indicating greater quality of life. The ANODE trial reported a mean EQ-5D-5L score of 0.935 and 0.927 for the co-amoxiclav and placebo groups respectively1. However, within these groups there was variation that needed to be considered such as those patients that experienced sepsis or no infection at all. To account for this, an EQ-5D-5L score for sepsis of 0.848 was acquired from the literature16. All other endpoints of the decision tree involved forms of localised infection or systemic infection not resulting in sepsis and were accordingly attributed the mean EQ-5D-5L value reported in the ANODE trial1. Due to the short 6-week time horizon, no discounting of benefit data was required17.