Reply: We thank the authors of the recent letter for their careful reading of the article and useful comments. Four issues were raised and we will respond to these in turn. The design effect that is relevant to cluster-randomized controlled trial was overlooked in calculating pooled 95% confidence interval (CI)1. All the 3 cluster-randomized controlled trials 2–4 reported cluster-adjusted risk ratios (RRs) and 95% CIs. Our reported confidence interval for the risk of neonatal mortality of chlorhexidine versus dry cord care (RR 0.83, 95% CI: 0.74–0.94)5 was narrower than it should have been, and the recalculated cluster-adjusted 95% CI is slightly wider as the correspondents point out (RR 0.83, 95% CI: 0.72–0.96). The Intention to Treat (ITT) analysis reported in the primary studies did have some deficiencies. If we take account of this then the reported 17% reduction in overall neonatal mortality may be an over estimation of effect. When we try to account for this then the estimate of effect is somewhat reduced to 13% RR 0.87 (95% CI: 0.79–0.96) while remaining a significant effect. This fact was referred to by Goldenberg et al6 in their commentary. Again the correspondents point out that further data may be included by assuming no interaction in the trial by Soofi et al4 reporting a factorial design. We have explored this after including these data the direction of effect is not altered and the point estimate for effect is slightly larger (RR 0.82, 95% CI: 0.73–0.92). Finally, the variation in clinical definitions for omphalitis does make pooling results more challenging. Due to differences in the description of omphalitis by the original study authors,2–4 we felt that combining the studies in a meta-analysis was not appropriate and thus we did not give an overall effect on omphalitis as an outcome as a main finding. We hoped that Figure 2 in our original article5 demonstrating significant heterogeneity (I2 of 92%) would illustrate this. Overall, the correspondents have provided useful input to help adjust and explore further the estimates of effect. There remains evidence of effectiveness for 4% chlorhexidine for neonatal cord care, but we would agree that these findings should be used as one part of a broader decision-making process on whether or not to introduce this intervention into routine settings. As further trials in Africa7,8 are under way then additional evidence to inform policy will be welcome. Jamlick Karumbi, B.Pharm Division of Paediatrics Ministry of Medical Services Afya House, Nairobi, Kenya SIRCLE Collaboration KEMRI-Wellcome Trust Research Programme Nairobi, Kenya School of Pharmacy University of Nairobi Nairobi, Kenya Mercy Mulaku, B.Pharm SIRCLE Collaboration KEMRI-Wellcome Trust Research Programme Nairobi, Kenya School of Pharmacy University of Nairobi Nairobi, Kenya Jalemba Aluvaala, MMed Division of Paediatrics Ministry of Medical Services Afya House, Nairobi, Kenya SIRCLE Collaboration KEMRI-Wellcome Trust Research Programme Nairobi, Kenya Department of Paediatrics University of Nairobi Nairobi, Kenya Mike English, PhD SIRCLE Collaboration KEMRI-Wellcome Trust Research Programme Nairobi, Kenya Nuffield Department of Medicine University of Oxford John Radcliffe Hospital Headington, Oxford, United Kingdom Newton Opiyo, MSc SIRCLE Collaboration KEMRI-Wellcome Trust Research Programme Nairobi, Kenya
Read full abstract