The authors used two decision-analytic models to synthesise data from a range of sources and model costs and outcomes of spontaneous preterm delivery and induced preterm deliveries or therapeutic abortions. The time horizon of the analysis was lifetime. The authors stated that the study perspective taken was societal for costs and maternal for utilities.
Effectiveness data were taken from known relevant studies based on observational databases collected in USA and expert opinion. The authors used their judgement to select the most appropriate estimate from the available evidence found in the literature. The main clinical effectiveness estimates were proportions of preterm infants resuscitated under a selective resuscitation regime (all infants were assumed to be resuscitated under a universal regime) and the outcome of resuscitation. Outcomes were death, moderate disability, severe impairment and survival intact or with mild sequelae. Average life expectancy of parturients was assumed to be 50 years.
Monetary benefit and utility valuations:
Maternal Utility weights were taken from a published study that assessed maternal quality of life for prenatal diagnosis of Down syndrome which estimated the preferences of the mother. For the termination of pregnancy model utilities were assumed as none were available from the literature.
Measure of benefit:
Quality-adjusted life-years (QALYs) were used to produce the ratio of cost per QALY. The numbers of infants surviving was reported. Future QALYs were discounted at a rate of 3% per annum.
Cost categories included in the analysis were costs of maternal care, foetal and neonatal care costs and the cost of long-term care for surviving infants. The sources of resource use and prices were based on estimates from the published literature. The price year for the analysis was 2010. Costs were presented in US Dollars ($). The authors used the medical component of the US Consumer Price Index to inflate costs and discounted future costs at a rate of 3% per annum.
Analysis of uncertainty:
The authors conducted univariate, bivariate and probabilistic sensitivity analysis. The results of the uncertainty analysis were presented using confidence ellipses.