Re: Association between day of delivery and obstetric outcomes: observational study
The impact of different models of labour ward staffing on perinatal outcomes is part of the important current debate about how the safety of NHS maternity care for women and babies can be assured. It is therefore disappointing that the study by Palmer et al.1 contains a number of clinical and methodological flaws, which ultimately lead to an inaccurate presentation of the results, unjustified extrapolations of what these results mean in terms of avoidable harm, and misleading evidence guiding policy. The publication of this paper is even more regrettable as it immediately follows the recent controversy that arose after another BMJ paper which purported that “patients admitted on Saturday and Sunday…have an increased likelihood of death within 30 days even when severity of illness is taken into account,”2 a result “prone to be misrepresented”.3
We have the following concerns with the paper by Palmer et al.:
1. The authors state that for in-hospital perinatal mortality, there is a ‘highly statistically significant increase’ observed at the weekend. However, the mortality indicator selected includes antepartum stillbirths that occur before the onset of labour, which account for 86% of stillbirths.4 This undermines the interpretation that weekend deliveries are more "dangerous" as these antepartum deaths will in most cases occur some days prior to the delivery of the baby. Therefore for a relevant and interpretable analysis of the association between day of delivery and perinatal mortality, deliveries of babies who have died before the onset of labour should have been excluded from the analysis.
2. The day-of-the-week results are presented using Tuesday as the reference day. Using this reference day, in-hospital perinatal mortality (defined by the authors as including antenatal stillbirth; see above) is statistically significantly increased not only on Saturday and Sunday but also on Wednesday, Thursday and Friday (according to results presented in the paper’s Figure). In our view, this demonstrates that the authors’ statement “that babies born at the weekend had an increased risk of being stillborn or dying in hospital” is at best an example of selective reporting and at worst an example of a misleading conclusion that can misguide the discussion on adequate staffing levels on weekdays and weekends.
3. Another issue is that since 2012 over half of maternity units have moved to staffing levels that provide full consultant cover during the daytime at weekends. This is a further argument against the interpretation that the safety of deliveries at the weekend is compromised. Furthermore, a visual inspection of the pattern of the other six outcomes (apart from in-hospital perinatal mortality) represented in the Figure included in this paper could be used as another indication that current staffing patterns provide consistent safety during the entire week (including the weekend). When Palmer et al. studied the relationship between consultant staffing levels and their outcomes of interest, they found no consistent evidence for a relationship. Only one out of seven outcomes was marginally statistically significant.
4. The authors suggest “770 perinatal deaths and 470 maternal infections per year above what would have been expected if performance was consistent across the week”. This statement is based on all days having the same rate as Tuesdays. A large proportion of this number is therefore not related to delivery at the weekend and is explained by the higher rates observed during other weekdays. The authors’ statement can be easily misinterpreted, as it has been in a number of headlines that appeared in the national press immediately following the publication of this paper.5,6 It is misleading because it suggests that the difference corresponds to avoidable events, whilst ignoring the influence of the play of chance (i.e. random variation) and other factors that are not amenable to quality improvement. We should not underestimate that impact that such headlines will have on public confidence in maternity services. It is regrettable that the authors provide this estimate in their paper because it has previously been pointed out in this journal that this type of interpretation is “enough to make a statistician sob.” 7
In our view, this paper presents misleading evidence which will unnecessarily undermine the public’s confidence in maternity services and potentially misguide policies aiming to increase out-of-hours safety levels. This could have been avoided if the authors had sought essential clinical input in the design, analysis and interpretation of the study. There are a number of ongoing national programmes that are examining the quality of care in maternity services which aim to provide evidence that is clinically relevant and methodologically robust so that it can adequately inform quality improvement initiatives to further improve the outcomes for pregnant women and their babies. 4,8,9,10
4 Manktelow BM, et al. on behalf of the MBRRACE-UK collaboration. Perinatal Mortality Surveillance Report UK Perinatal Deaths for births from January to December 2013. Leicester: The Infant Mortality and Morbidity Group, Department of Health Sciences, University of Leicester. 2015.
5 Gregory A. Betrayal of our babies: 770 die a year because they’re born at the weekend. The Daily Mirror. 2015 Nov 25.
6 Spencer M. Risk of having a weekend baby: Major study reveals greater threat of stillbirth or death. The Daily Mail. 2015 Nov 25.
8 Knight HE, Cromwell D, van der Meulen J, et al. Patterns of maternity care in English NHS hospitals 2011/12. Royal College of Obstetricians and Gynaecologists, 2013.
Competing interests: No competing interests