This week our paper, a systematic review of both the non-experimental and randomised controlled trial evidence comparing intermittent auscultation with CTG use for women considered to be at high risk during labour, has been published in an issue of Women and Birth. You can access a free copy here for a short time. Today’s post explains the backstory about why writing this paper was important for me.
One of the goals I set myself during the completion of the PhD was to read, and understand, pretty much everything that has been published on CTG monitoring. I started collecting and reading papers about two years before I enrolled. I began by obtaining copies of the original papers cited in the Cochrane review. That was eye-opening. The research was sound, and clearly showed no benefit from CTG use.
As I have spoken with obstetricians about the evidence, the majority are quite sure that I am wrong, and are very sure that CTG use improves outcomes for women with risk factors. This surety about the evidence is curious given that the Cochrane review (Alfirevic et al., 2017) has always included information about the use of CTG monitoring in high-risk populations concluding that it didn’t improve outcomes, even back to when it was the book called Effective Care in Pregnancy and Childbirth. It seemed to me that this information was hidden in plain sight.
That was my first motivation for wanting to write this paper. I wanted to pull the information about high-risk populations out from where it was buried in Cochrane review and make it more visible. And I wanted to go further than this. The RANZCOG guidelines suggested that evidence from cohort studies supports CTG use (Royal Australian and New Zealand College of Obstetricians and Gynaecologists, 2019). Somewhat tellingly, the evidence they gave for this statement was not a cohort study, but the Dublin randomised controlled trial (MacDonald et al., 1985), and another paper about the use of Bayesian approaches to data analysis (Hornbuckle et al., 2000). I wanted to find out for myself whether the non-experimental research been compellingly in favour of better outcomes from CTG use.
My co-authors were on board with the plan for the paper, so I (with the help of our excellent librarians here at Griffith University) set about finding all the original research and making quality assessments. We did a re-analysis of the same evidence included in the Cochrane review. Not surprisingly given the skills of the Cochrane team, we found the same that they did: no improvement in mortality and an increase in cerebral palsy rates when CTG monitoring is used rather than intermittent auscultation.
We moved to looking at the non-experimental research, excluding research which didn’t include women considered to be at high risk. We kept all of what we found regardless of quality, and in table 1 listed the non-experimental research according to a quality assessment using the ROBINS-I tool. We didn’t attempt to summarise the findings of all the non-experimental research with a single relative risk calculation (as we had for the randomised controlled trials), as there was far too much variability within the research for this to be meaningful. None of this research reported on cerebral palsy as an outcome.
There wasn’t much space in the paper to discuss the findings from non-experimental research, so let me share a few thoughts that didn’t make it into the article. (I haven’t reproduced table 1 here for copyright reasons, but you can find it in the paper.) We found a total of 26 non-experimental research projects, some published over more than one paper. Of these, 20 were at critical risk of bias according to the ROBINS-I tool. That is, the possibility that the findings were due to something other than the form of heart rate monitoring that was used was very high.
Of the papers in the critical risk category, 15 compared outcomes from a period prior to the introduction of CTG monitoring with a period after CTG monitoring was introduced. These studies were published between 1972 and 1982. Three compared the use of intermittent auscultation in a low risk population with CTG use in a high-risk population, and the remaining two papers compared intermittent auscultation with CTG use in a high-risk population. Only four of the 26 found a statistically significant reduction in stillbirth, four found a significant reduction in neonatal mortality, and five found a significant reduction in the overall perinatal mortality rate. One study found a statistically significant increase in neonatal death when CTGs were introduced. The risk ratios for perinatal mortality ranged from 0.10 (a large reduction in deaths) to 1.56 (a 50% increase in deaths).
Five studies were considered to be at moderate risk of bias according to ROBINS-I. This evidence is considered by the makers of the ROBINS-I tool to be sufficiently robust to guide practice. These were conducted between 1978 and 2018, with the largest including 1.2 million women. Two studies compared the use of intermittent auscultation with CTG use in a high-risk population, neither finding significant benefits. The remaining three studies examined large population data sets comparing the use of intermittent auscultation with CTGs, having corrected for known confounders. One found a statistically significant increase in neonatal deaths with CTG use, with a risk ratio of 1.46. The others found no statistically significant differences.
Having examined the non-experimental research, what emerges is the absence of compelling evidence that CTG use improves perinatal outcomes for women considered to be a high risk. How long might it be before RANZCOG and other guideline writing authorities acknowledge the reality of the evidence and produce guidelines that are evidence-aligned?
Alfirevic, Z., Devane, D., Gyte, G. M. L., & Cuthbert, A. (2017, Feb 03). Continuous cardiotocography (CTG) as a form of electronic fetal monitoring (EFM) for fetal assessment during labour. Cochrane Database of Systematic Reviews, 2(CD006066), 1-137.
Hornbuckle, J., Vail, A., Abrams, K. R., & Thornton, J. G. (2000, Jan). Bayesian interpretation of trials: the example of intrapartum electronic fetal heart rate monitoring. BJOG: An International Journal of Obstetrics and Gynaecology, 107(1), 3-10.
MacDonald, D., Grant, A., Sheridan-Pereira, M., Boylan, P. C., & Chalmers, I. (1985, Jul 01). The Dublin randomized controlled trial of intrapartum fetal heart rate monitoring. American Journal of Obstetrics and Gynecology, 152(5), 524-539.
Royal Australian and New Zealand College of Obstetricians and Gynaecologists. (2019). Intrapartum fetal surveillance clinical guideline. 4th Edn.
Small, K. A., Sidebotham, M., Fenwick, J., & Gamble, J. (2020, Sept). Intrapartum cardiotocograph monitoring and perinatal outcomes for women at risk: Literature review. Women and Birth, 33(5), 411-418.