We read with interest the study by Dr Edwards and colleagues. We are not surprised that their investigation of the predictive power for severe sepsis of various modified obstetric early warning scoring systems (MOEWS) confirms the limitations of MOEWS in a ward setting. MOEWS are not designed to predict severe sepsis. They are intended to help detect the early signs of illness and trigger timely medical review with appropriate intervention. It could be that many of those in the Edwards et al cohort of women with chorioamnionitis who triggered the various MOEWS, but did not develop severe sepsis, avoided clinical deterioration because appropriate antimicrobial treatment and fluid resuscitation were instigated early in their clinical course. The absence of case-notes review to determine, for example, the interval from initial vital sign abnormalities to antibiotic administration, is a significant study design limitation that Dr Edwards and her co-authors could not address. It is apparent that the completeness of vital sign recordings for inpatients has often been suboptimal. Indeed, Dr Edwards and her co-authors have highlighted this by their exclusion of 549 of the 913 women in their database due to incomplete vital sign data.

In 2013, Ireland was the first country to implement nationally a standardized MOEWS for all pregnant and postnatal inpatients. A customized obstetric sepsis screening tool has been included since July 2014. The Irish Maternity Early Warning System was not among the 6 sample MOEWS selected for analysis from the global literature search of Edwards et al. Recent work has shown an improvement in the completeness of vital sign recordings since the Irish Maternity Early Warning System was introduced compared to the years before its development. Herein lies, in our view, the most significant potential benefit of MOEWS—improved recording of vital signs. MOEWS augment, but do not replace, clinical judgement and there will always be instances when, regardless of the validation or not of MOEWS, women with vital signs values defined as normal by the trigger thresholds require clinical assessment due to, for example, malodorous liquor or fetal tachycardia. We agree with Dr Edwards and her co-authors that further research should aim to validate MOEWS clinically and to optimize their sensitivity and specificity, thus enabling the efficient use of limited staff resources.