We thank Dr Rosén for his interest in our article on the effectiveness of ST analysis and electronic fetal monitoring (EFM) in the reduction of neonatal metabolic acidosis. We would like to respond to his comments on (1) the validation of BD _ecf values (2) the exclusion of the Plymouth data, and (3) the investigation of the learning effect of ST analysis.

In our individual participant data (IPD) metaanalysis, we analyzed the data as they were shared by the principal investigators. Based on these, BD _ecf was recalculated according to the Siggaard-Andersen method. All analyses were performed according to the intention-to-treat principle; therefore, we did not exclude any measurements of pH and pCO ₂ that were found to be invalid according to the checklist previously described by Kro et al. Application of the checklist revealed that only 61% of all measurements fulfilled all 5 criteria and that only 79% fulfilled the first 4 criteria, which indicated that a large proportion of arterial pCO ₂ measurements was discordant to the measured pH. This was also the case for 2 of the new cases of metabolic acidosis, which were introduced through multiple imputation techniques. This percentage (2 of 13 cases of metabolic acidosis with a discrepancy between pH and pCO ₂ ) is comparable with the percentage in the total IPD. Therefore we are confident that the additional cases because of imputation techniques are a reliable representation of the original data.

Initially we did contact the principle investigator of the Plymouth trial; however, it was not possible to find anyone with access to the IPD. Given that the ST-analysis method that was used was slightly different from the methods that had been used in more recent trials, we decided to exclude the trial of Westgate et al. Dr Rosèn is correct that biphasic ST changes were incorporated in their guideline. However, although investigating T/QRS changes, their threshold for performing an intervention was based on the absolute T/QRS ratio rather than a change in T/QRS ratio. Another important difference was that they identified ST changes by visual analysis rather than by automatic assessment. Therefore, it was decided to leave the Westgate trial out of our metaanalysis. After the addition of the results of this trial to the results of the metaanalysis with a 2-step approach, we found a significant reduction of metabolic acidosis in the EFM + ST-analysis arm compared with EFM alone (relative risk [RR], 0.70; 95% confidence interval [CI], 0.50–0.995).

Indeed, the favorable effect of ST analysis in the trial by Amer-Wåhlin et al was obtained mainly in the second part of the trial, which indicated a certain learning curve of ST analysis. Because we aimed to investigate the effectiveness of ST analysis based on all available data, followed an intention-to-treat principle, and did not prespecify such analysis, we choose not to differentiate between the first and second half of the trials. However, analysis of the second part of the IPD indicated that EFM + ST analysis statistically significant reduced metabolic acidosis _BDecf (RR, 0.50; 95% CI, 0.29–0.85), arterial pH <7.05 (RR, 0.67; 95% CI, 0.49–0.92), neonatal intensive care admission (RR, 0.76; 95% CI, 0.59–0.97), and composite adverse neonatal outcome (RR, 0.60; 95% CI, 0.40–0.89), compared with EFM alone. A remarkable finding was that the number of adverse outcomes in the EFM arm increased in the second part of the IPD compared with the first part, while they decreased in the EFM + ST-analysis arm. A possible explanation for these findings could be that, with time, the experience with and confidence in ST analysis increased, which consequently decreased the confidence in the conventional monitoring techniques and resulted in more doubtful and uncertain decisions.

Further studies into long-term neonatal outcomes, follow-up evaluation after introduction of ST analysis, and cost-effectiveness should give further evidence about the usefulness of additional ST analysis in EFM. Although the overall IPD data did not show a significant reduction in metabolic acidosis and neonatal intensive care admissions, such an improved outcome was found for the second part of the trials, which indicated that a learning curve has to be taken into account.