We read with great interest the article by Drassinower et al, who investigated the impact of prolonged latency after preterm premature rupture of membranes (PPROM) on neonatal sepsis. The main finding highlights that, for a given gestational age at PPROM, prolonged latency does not increase the risk of neonatal sepsis, except for latencies >4 weeks associated with reduced risk of sepsis. As rightly underlined by the authors, this result makes sense as the most stable cases of PPROM with the longest latency durations are probably those with the lowest risks of chorioamnionitis and neonatal sepsis.

However, we would like to raise some key points. First, the primary outcome included both early- and late-onset sepsis. This definition seems inappropriate as late-onset sepsis mostly results from pathogens horizontally transmitted after birth, thus do not reflect intrauterine inflammation. Second, cases of stillbirths and neonatal deaths were not taken into consideration while potentially linked to severe prenatal or postnatal infection. This can lead to underestimating adverse effects of prolonged latency, especially for ruptures at 22-24 weeks of gestation that were included in the trial. Third, in the multivariable model, adjustment for both latency and time-dependent covariates (eg, multiple courses of steroids) may have introduced overadjustment bias. We would also have appreciated being able to interpret global P values for each covariate included in the multivariable model.

Such results are likely to greatly impact daily practice, obstetric management, and counseling of women with PPROM. Further analyses addressing the previous issues are therefore needed.