Dear Dr. Ram Kumar,
Thank you very much for your comments on our meta-analysis. After discussing with the co-authors and carefully re-checking the data. We found that we made a careless mistake. We are sorry for making the readers confused by this mistake. There is an error in showing the odds ratio of ‘the association between arterial umbilical cord pH and AIS’.
When we first submitted our manuscript to the journal, the reviewers’ comments including:
1. The authors should ‘perform some limited multivariable modelling to see if the univariable effects they are observing are confounded by baseline differences in patient characteristics’.
2. The authors should include more risk factors such as ‘pre-eclampsia, ventouse delivery and oxytocin induction’.
According to the reviewers’ comments, we re-performed the meta-analysis. Considering that ORs from univariable and multivariable models, we performed analysis using both fixed-effect and random-effect models to analyze the data. By fixed-effect model, we found that the odds ratio was 4.63 (95%CI, 2.14 to 9.98) for arterial umbilical cord ph. On the other hand, by random-effect model, the odds ratio was 4.80 (95% CI, 0.68 to 33.90). We therefore used the positive results and the figure detected by fixed-effect model in our manuscript. A few days later, after consulting with a statistical expert, we were told that it was better to use random-effect model to detect the odds ratio because of the significant heterogeneity (I² = 79.3%, p = 0.008). Therefore, it was better to sue the odds ratio 4.80 (95% CI, 0.68 to 33.90) to show the results, which means the data did not show significant change of the mean risk estimate (OR 4.80; 95% CI 0.68 to 33.90). However, we made a careless mistake. The negative data was just in the brain for discussing. The positive data and figure detected by the fixed-effect model were not be replaced by the negative data detected by the random-effect model when we submitted the revised version of our manuscript.
In clinic, cord arterial blood pH was supposed to be an important indicator of hypoxia and brain injury. Our meta-analysis did not show the risk estimate. We think these are associated with the significant heterogeneity and a few studies included in our analysis. Recently, we followed up the studies on the association between hypoxia and PAIS. We found there are two new studies related to hypoxia and PAIS. We would like to include the new studies and do a meta-analysis to update the new data if the editor agrees. And we would like Dr. Kumar to review our new submission.
Thank you very much for your time.

Sincerely,

Dezhi Mu