The Royal College of Obstetricians and Gynaecologists (RCOG, based in the UK) recently released an update on the progress of their “Each Baby Counts” program. Back in 2015 the program set an ambitious target to reduce perinatal mortality and severe brain injury in term babies by 50% by the year 2020. This most recent report covered outcomes for the year 2017, and found that nothing has changed since the program started.
Deaths and brain injury occurred at a rate of 1.52 per 1000 term births. 12% of the babies included in this outcome had an intrapartum stillbirth, 13% experienced early neonatal death (first seven days after birth), and 75% had severe brain injury. In 2016, this combined outcome occurred in 1.61 per 1000 term births, and in 2015 it was 1.57 per 1000 term births. To be considered a success, the rate would need to be 0.79 per 1000 by 2020. It is clear that significant progress is not being made.
In 72% of cases, at least one reviewer had the opinion that the outcome might have been prevented had care been different. There was also no change in this rate, having been 71% in 2016 and 76% in 2015. It is fair to say then that not only has there been no change to the outcome of interest, there also appears to have been little improvement in the rate of poor care provision. Why might this be the case?
The Each Baby Counts program is effectively a body of research. Good researchers understand that they need to examine closely the assumptions they make and consider the impact they have on their findings. The designers of the Each Baby Counts program appear to have not adequately addressed some key assumptions, which might contribute to the lack of benefit from the program.
Each Baby Counts seeks to generate new knowledge by reviewing data from health services about poor outcomes. RCOG then assumes that sharing this knowledge will lead to a change in practice and an improvement in outcomes. I’m going to call this assumption the quality improvement assumption. This is a common assumption in healthcare services, and is a feature of a managerial approach that has been called New Public Management (Siltala, 2013). There is actually little hard evidence to show that quality improvement processes robustly achieve improvements in quality or safety.
The other assumption that has been made is that the definitions accurately represents a concrete diagnosis which is stable over time and permits comparisons to be made from one year to the next. While this is true of the mortality figures, this is not the case for the outcome which dominates the overall outcome, namely severe brain injury. Severe brain injury was defined as a diagnosis of grade III hypoxic-ischaemic encephalopathy; or a combination of decreased central tone, coma, and seizures; or the use of therapeutic cooling. Establishing the grading of hypoxic-ischaemic encephalopathy requires clinical judgement to apply criteria such as those set out by Sarnat and Sarnat in 1976. Different clinicians might arrive at a different grading even when using the same criteria, and if there is variation in the criteria used over time then this alone can explain differences. The use of therapeutic cooling is a particular risk for definitional accuracy as the indications for the use of this vary from unit to unit, clinician to clinician, and over time. Therapeutic cooling is a decision, not a diagnosis. The rate of use of therapeutic cooling is liable to change for reasons other than because of a change in the health of newborn infants.
The other important assumptions that shape this body of knowledge are the framework through which data were examined and the tools used to generate knowledge. The Each Baby Counts report is clear that the reviewers decided to focus on “clinical escalation in maternity care” (p. x). The chosen framework is strongly aligned with an obstetric world view. The assumptions built in to the framework are that women’s bodies are a site of risk for the fetus, it is the role of the clinician (particularly the midwife) to detect this risk and to initiate additional monitoring, and that midwives must detect abnormalities in this monitoring (generally based on obstetrical standardised understandings of how women’s bodies function). Having detected the problem, the midwife is expected to communicate this with an obstetric staff member, who is meant to take a course of action designed to prevent the anticipated risk. The decision to use these particular frameworks excludes other possible ways to frame up the data. In making this choice, the focus is predominantly on the role played by individual clinicians within the healthcare system, ignoring the birthing woman or person as an active agent, the role of the healthcare system, and the impact of the social world outside healthcare.
A range of different methodological tools were used to collect and make sense of data at the local health service level. These were set out on page 6 and 7 of the report, and include Root Cause Analysis, Datix, the Perinatal Mortality Review Tool, and others. Little effort has been put into validating these tools as fit for purpose. Again, these tools use a particular frame through which to view care provision, and exclude other ways of knowing.
There is the distinct possibility then that findings generated by the Each Baby Counts review don’t adequately explain the reasons for the poor outcomes. Given that the recommendations then stem from these findings, there is the possibility that they might, at best, have no impact on perinatal outcomes. At worst, there may be perverse consequences where other harms occur but go relatively unmeasured within the metrics used by Each Baby Counts (maternal traumatic stress for example).
As an interesting aside, Figure 10 on page 11 of the report appears to indicate that the use of intermittent auscultation was associated with fewer instances of errors or interpretation, equipment problems, or failure to act on abnormal findings than the use of CTG monitoring. This is worthy of further examination, particularly in light of the evidence that CTG monitoring doesn’t improve perinatal outcomes, while being associated with higher rates of surgical birth (Small, et al, 2020).
Work to improve maternity care processes and outcomes is important. It is also important that it is done well. We should perhaps question whether the RCOG and other obstetric led organisations are best placed to do this given the cognitive biases that the obstetric knowledge paradigm brings to maternity care. The solutions to improving perinatal outcomes are likely to be found in social change and health system reform which are principally aimed at improving the social welfare of reproductive age women.
Sarnat, H.B., & Sarnat, M.S. (1976). Neonatal encephalopathy following fetal distress. A clinical and electroencephalographic study. Archives of Neurology, 33(10):696-705.
Siltala, J. (2013, May 22). New Public Management: The Evidence-Based Worst Practice? Administration & Society, 45(4), 468-493. https://doi.org/10.1177/0095399713483385
Small, K. A., Sidebotham, M., Fenwick, J., & Gamble, J. (2020). Intrapartum cardiotocograph monitoring and perinatal outcomes for women at risk: Literature review. Women and Birth, 33(5), 411-418. https://doi.org/10.1016/j.wombi.2019.10.002