C. Data analyses
Results were analysed according to the method of IOL received. Mean costs (and standard deviation) are expressed in Australian dollars (AUD) at 2019. The cost-effectiveness was estimated for the two competing methods of IOL received, and the expected values of direct costs and health utilities were calculated for each model pathway (see Supplementary material, Figure S2). Total costs were calculated as a sum of all applicable cost components per model pathway, including mean priming cost, cost of CS and mean length of stay (in hours) in each clinical unit (antenatal ward, birth suite, postnatal ward, special nursery) multiplied by the corresponding mean hourly cost of stay. Deterministic model outcomes are presented as cost per quality-adjusted life year (QALY) and as incremental cost-effectiveness ratio (ICER), where the difference in cost (∆C) is divided by the difference in effectiveness (∆E). We also calculated the Net Monetary Benefit (NMB) including the willingness-to-pay (λ) per additional outcome where NMB = (λ*∆E)-∆C (19). The strategy with the highest NMB is the recommended strategy. Results were not discounted due to the short time frame of the evaluation. TreeAge Pro 2020, R1.1 software was used for all cost-effectiveness, sensitivity and subgroup analyses (20).
Probabilistic sensitivity analysis was performed as Monte Carlo simulations by adding model distributions for input parameters at the same time and running the model 10,000 times. Beta distributions were fitted around utility parameters and gamma distributions around cost and length of stay parameters using the mean values and standard deviations for each pathway. The results are presented graphically on the cost-effectiveness acceptability curve and on the cost-effectiveness plane. We assumed a willingness-to-pay (WTP) of the standard $50,000 per QALY gained (21). In one-way sensitivity analyses only a single value from a range of possible values is tested at a time whilst all other parameters remain the same. We performed one-way sensitivity analyses for the cost of CS, hourly cost of stay in antenatal ward, birth suite, postnatal ward, intensive care nursery and special care nursery. Subgroup analyses were performed for parity (nulliparous vs parous) and cervical favourability at start of IOL (modified Bishops score (MBS) ≤3 vs >3) and the impact on model outcomes was recorded.