On 2017 Sep 21, Hendrik S. Fischer commented:

In response to Dr. Fleming’s comment: Odds ratios (OR) must not be misinterpreted as risk ratios (RR). If outcomes are rare, OR approximate RR. If outcomes are common, OR and RR differ, but both OR and RR still have specific advantages and disadvantages (Cummings P. The relative merits of risk ratios and odds ratios. Arch Pediatr Adolesc Med. 2009;163:438–45). Importantly, OR cannot exaggerate differences across groups given as RR. They are just two different methods to express these differences. In the present meta-analysis, the odds ratio (95% confidence interval) of death or BPD is 0.83 (0.71–0.96) and the risk ratio is 0.90 (0.83–0.98). Admittedly, both OR and RR may be difficult to interpret from a clinical point of view. We therefore recommend that meta-analyses should calculate a number needed to treat (NNT) as a meaningful measure of effect for the clinician. In our meta-analysis, the NTT was 35, indicating the small but beneficial effect of avoiding mechanical ventilation, which we have discussed in the abstract and in our paper.

On 2017 Jul 12, Robert Fleming commented:

In the meta-analysis presented in this manuscript, the results are presented as odds ratio (OR) rather than relative risk (RR). When an outcome (BPD in this setting) is common (> 10% in the unexposed group, which is the case for these studies), using OR can exaggerate differences across the groups. The original report from the SUPPORT trial reports a RR of 0.9. In this meta-analysis the OR from the same study is reported as 0.81. Likewise, converting the OR from this meta-analysis (0.83) to RR results in a value of 0.9. This point is made not to imply that mechanical ventilation should not be avoided; however the effect of avoiding mechanical ventilation on the incidence of BPD calculates to be less than concluded in this meta-analysis.

On 2017 Jul 12, Robert Fleming commented:

In the meta-analysis presented in this manuscript, the results are presented as odds ratio (OR) rather than relative risk (RR). When an outcome (BPD in this setting) is common (> 10% in the unexposed group, which is the case for these studies), using OR can exaggerate differences across the groups. The original report from the SUPPORT trial reports a RR of 0.9. In this meta-analysis the OR from the same study is reported as 0.81. Likewise, converting the OR from this meta-analysis (0.83) to RR results in a value of 0.9. This point is made not to imply that mechanical ventilation should not be avoided; however the effect of avoiding mechanical ventilation on the incidence of BPD calculates to be less than concluded in this meta-analysis.