Baby-OSCAR: Outcome after Selective early treatment for Closure of patent ductus ARteriosus in preterm babies-a statistical analysis plan for short-term outcomes.
Bell JL., Gupta S., Juszczak E., Hardy P., Linsell L.
BACKGROUND: The Baby-OSCAR trial is a multi-centre, randomised, placebo-controlled parallel group trial of early treatment of large patent ductus arteriosus (PDA) with ibuprofen in extremely preterm infants. This paper describes the statistical analysis plan for the short-term health outcomes of the Baby-OSCAR trial. METHODS AND DESIGN: This is a randomised controlled trial to determine if early-targeted treatment of a large PDA with parenteral ibuprofen in extremely preterm babies improves short and long-term health and economic outcomes. Infants born between 23+0 and 28+6 weeks of gestation, confirmed by echocardiography as having a large PDA (with a diameter of at least 1.5 mm and unrestricted pulsatile PDA flow pattern), with parental informed consent, were randomly allocated to receive either ibuprofen or placebo within 72 h of birth. The primary outcome is a composite of death by 36 weeks' postmenstrual age or moderate or severe bronchopulmonary dysplasia (BPD) at 36 weeks' postmenstrual age. RESULTS: Baseline demographic and clinical characteristics will be described by randomised group. The primary analysis will be on the modified intention to treat (ITT) population. Counts and percentages will be presented for binary and categorical variables, and mean and standard deviation or median and interquartile range will be presented for continuous variables. For binary outcomes, risk ratios and confidence intervals will be calculated using log binomial or Poisson regression with a robust variance estimator. Continuous outcomes will be analysed using linear regression models, or quantile regression models if skewed. Analyses will be adjusted for all minimisation factors where technically possible, and correlation between siblings from multiple births will be accounted for by nesting the clusters as a random effect. Both crude and adjusted effect estimates will be presented, with the primary inference based on the adjusted estimates. Ninety-five per cent confidence intervals will be used for all pre-specified outcome comparisons. CONCLUSION: This paper describes the statistical analysis plan for short-term health outcomes of the trial, including the analysis principles, definitions of important outcomes, methods for primary analysis, pre-specified subgroup analysis, and secondary analysis. The plan was finalised prior to completion of short-term follow-up. TRIAL REGISTRATION: ISRCTN registry ISRCTN84264977 . Registered on 15 September 2010.