In their recent publication on ST analysis as an adjunct to standard electric fetal monitoring, Schuit et al based their analysis on a case-by case–based data analysis of 4 randomized controlled trials using multiple imputation for missing data. They claim this approach to be superior to the more standard aggregated data analysis.

There are 3 main issues to be discussed in connection with the study design:

As their primary outcome parameter, they have used BD _ecf with no defined rules of data validation. In our recent reanalysis of the Swedish randomized controlled trial data, such rules have been identified and should have been commented on because they limit the need for imputation techniques and include composite neonatal markers as required. Furthermore, despite our previous scrutiny of the Swedish randomized controlled trial data, the current data analysis adds 4 more cases of metabolic acidosis and a total of 13 in the whole material (Table 3 in their article contains an error in the addition of metabolic acidosis cases). Those cases should be presented and discussed with the use of new alternatives for data verification, such as percentiles for arterial pCO ₂ for a specific arterial pH.

Exclusion of the Plymouth data because the ST protocol was substantially different. Although absolute T/QRS values were used, they clearly were associated with an ST rise as stated in the protocol and negative ST (today termed biphasic ST) were included as an indication for intervention (further information can be found in Yli et al ). I can understand why Plymouth patient data could not be used for a case-by-case analysis because they do not exist in a standardized database; however, to claim that the study should be excluded on the grounds that are presented should be considered to be bias.

Why did the authors not separate the studies into their first and second half because this may have added to our understanding of the implementation of a new method for fetal surveillance?

The authors state that metabolic acidosis is the best surrogate marker of intrapartum asphyxia at the same time that they argue the need for composite markers to reduce the size of the trials. I would recommend composite markers to be used with caution and primarily to include or exclude a case of metabolic acidosis as has been done previously. Our ability to document pathophysiologic condition should still be the aim, and statistics should be only a tool to assist.