This second part of the appraisal of RCTs comparing CTG+ST to CTG focus on the five meta-analyses (MAs) published in 2012 and 2013. As discussed in the first part, there are considerable differences in several variables in the five RCTs that make a comparison between the trials, and hence performing adequate meta-analyses, challenging. The type of meta-analyses to use, fixed- or random-effect MA, depends on the heterogeneity of the trials. A random-effect MA as a rule gives a more conservative 95% confidence interval. Both types have been used adequately in the five MAs.
None of the MAs included complete and relevant data from all five RCTs and several calculation errors were found that changed the result in the final analyses. There were for example mix up of two different ways to calculate base deficit (in blood and extracellular fluid), mistakes of not including the complete revised data from the Swedish and Dutch trials, and calculation errors using the wrong denominators or summary discrepancies.
Two of the MAs decided to exclude the Plymouth RCT because of non-computerized ST analysis. The authors find this unfortunate since this trial contributes considerable to the analyses of metabolic acidosis and argue that using manual interpretation of ST data presents a greater challenge to the ability to improve perinatal outcomes. On the other hand, all MAs include the French RCT despite the fact that the study design of this trial does not conform to the intended use (and Stan guidelines) of the methodology (see first part of this article series). Hence, the authors recommend including the Plymouth RCT but excluding the French RCT in future MAs.
To resolve the errors found in the present MAs the authors have renewed the calculations of several of the outcomes. Results below show the reduction in the CTG+ST group, where * denotes a significant reduction.
Data for neonatal encephalopathy stage ≥2 (Sarnat & Sarnat) are only available in the Swedish and Dutch trials, and show no significant difference between the groups (RR 0.75, 95% CI 0.04-15.69). For other variables presented in the MAs (admission to NICU, neonatal intubation, perinatal mortality) adequate data were not available to make correct comparisons.
The authors conclude that the new meta-analyses shows significant reductions in total operative delivery rate and in neonatal metabolic acidosis rate in addition to the significant reduction in FBS usage that was previously shown in the MAs.
Abstract – Olofsson et al. 2014
