Search

ObjectiveTo determine whether early low-dose hydrocortisone treatment in extremely preterm infants is associated with brain damage assessed by MRI at term equivalent of age (TEA).Patients and outcomesThis is a predefined secondary analysis of brain abnormalities, observed by MRI at TEA, of patients randomly assigned to receive either placebo or hydrocortisone in the PREMILOC trial. Outcomes were based on brain abnormalities graded according to Kidokoro scores.ResultsAmong 412 survivors at TEA, 300 MRIs were performed and 295 were suitable for analysis. Kidokoro scoring was completed for 119/148 and 110/147 MRIs in the hydrocortisone and placebo groups, respectively. The distribution of the Kidokoro white matter (WM) subscore and other subscores was not significantly different between the two groups. There was, however, a significant association between a higher overall Kidokoro score and hydrocortisone treatment (5.84 (SD 3.51) for hydrocortisone and 4.98 (SD 2.52) for placebo; mean difference, 0.86; 95% CI 0.06 to 1.66; p=0.04). However, hydrocortisone was not statistically associated with moderate-to-severe brain lesions (Kidokoro overall score ≥6) in a multivariate logistic regression model accounting for potential confounding variables (adjusted OR (95% CI) 1.27 (0.75 to 2.14), p=0.38). Bronchopulmonary dysplasia at 36 weeks postmenstrual age significantly predicted both WM damage (adjusted OR (95% CI) 2.70 (1.03 to 7.14), p=0.04) and global brain damage (adjusted OR (95% CI) 2.18 (1.19 to 3.99), p=0.01).ConclusionsEarly hydrocortisone exposure in extremely preterm infants is not statistically associated with either WM brain damage or overall moderate-to-severe brain lesions when adjusted for other neonatal variables.Trial registration numberEudraCT number 2007-002041-20, NCT00623740

Additional file 1. Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) Checklist.

Additional file 1. Data collected by patients in the diaries and also in the CRFs.

International audience; A Bayesian methodology is proposed for constructing a parametric prior on two treatment effect parameters, based on graphical information elicited from a group of expert physicians. The motivating application is a 70-patient randomized trial to compare two treatments for idiopathic nephrotic syndrome in children. The methodology relies on histograms of the treatment parameters constructed manually by each physician, applying the method of Johnson et al. (2010). For each physician, a marginal prior for each treatment parameter characterized by location and precision hyperparameters is fit to the elicited histogram. A bivariate prior is obtained by averaging the marginals over a latent physician effect distribution. An overall prior is constructed as a mixture of the individual physicians' priors. A simulation study evaluating several versions of the methodology is presented. A framework is given for performing a sensitivity analysis of posterior inferences to prior location and precision and illustrated based on the idiopathic nephrotic syndrome trial.

Table 1: List of the participating centers. (DOCX 36 kb)

Characteristics of included published studies (1â 142) and protocols of future ongoing trials registered on ClinicalTrials.gov (143 to 170). (DOCX 67 kb)

PubMed search, ClinicalTrials.gov and Cochrane library search strategies. (DOCX 26 kb)

Plots of posterior class a and class b misclassifications according to the stopping thresholds for each of the 13 pairs of priors. The plots analogous to Fig. 3 for the 3 other events: (a) Death, (b) Necrotizing enterocolitis, (c) Retinopathy. (PDF 2969 kb)

Distribution of the successive conclusions and misclassifications, obtained by applying the decision rule to the 5000 simulated trials at each interim analysis and in overall. The plots analogous to Fig. 4 for the 3 other events: (a) Death, (b) Necrotizing enterocolitis, (c) Retinopathy. (PDF 2498 kb)

Supplemental information’s on methods. Additional information on methods : (i) Method for comparison of the 3 different estimation ways used to fit the mixtures of Beta distributions; (ii) Definition of the non-informative and informative priors compared in the sensitivity analysis; (iii) Definition of the “good decision” for each scenario of the simulation study. (PDF 246 kb)

Final overall rates of misclassifications obtained with the decision rule, according to the function used to define the thresholds at each successive interim analysis. The final overall rates of misclassifications obtained with the decision rule summarized in the Table 5, provided for the 4 functions used to define the thresholds at each successive interim analysis. (PDF 107 kb)

Parameters of the mixtures retained to fit the acceptable differences between arms Dj. A table summarizing the mixtures retained for the acceptable differences. (PDF 108 kb)

Expected distribution of gestational age and events in the control arm (FD arm) of the BETADOSE trial, imputed from the prevalence observed in the ePIPAGE-2 cohort study. A table providing the expected event rates and gestational age in the motivated trial (based on the EPIPAGE-2 cohort study [17]). (PDF 78 kb)

Full Patients’ Verbatim Comments. (DOCX 19 kb)

Histogram of the acceptable differences in events, and mixtures of beta distributions fitted from experts’ elicitation, through 3 different methods, with their criteria for goodness of fit. The plots analogous to Fig. 2 for the 3 other events: (a) Death, (b) Necrotizing enterocolitis, (c) Retinopathy. (PDF 992 kb)

Comparison of the use of a mixture distribution versus the use of an average value of elicitated margins from experts. In this additional work, two approaches were proposed to compute the acceptable difference from expertsâ elicitation: the method described in the main manuscript and the use of average values. (PDF 368 kb)