Paper discussed: Dodd JM, Crowther CA, Haslam RR, Robinson JS, Group TToBT. Elective birth at 37 weeks of gestation versus standard care for women with an uncomplicated twin pregnancy at term: the Twins Timing of Birth Randomised Trial. BJOG 2012;119(8):964–73.
Date of exchange: 10 September 2012
Number of participants: 7
Number of tweets: 70 (35 storified)
|Participants||All women with a twin pregnancy at gestational age of 36+6 weeks or more|
|Intervention||Elective induction of labour or elective caesarean sections|
|Outcome||A composite mortality and morbidity index, including perinatal mortality and serious neonatal morbidity|
|Study design||Multinational multi-centred randomised trial|
Research literacy: randomisation and allocation concealment
The current study used computer-generated numbers to determine the allocation sequence, which was an appropriate method. However, there was a lack of description regarding the concealment of allocation, a factor critical in ensuring the comparability of groups at baseline to minimise the risk of selection bias. Table 1 reassured us that mothers’ risk factors in the two groups were similar. The achievement of an ideal trial methodology is not always practically feasible, but the potential impact of any deficiencies can be explored through careful examination of the relevant data reported.
Feedback on the study
Are the results different in monochorionic and dichorionic twins?
One criticism of the trial was the mixing of monochorionic and dichorionic twins in the analysis. Clinical management differs between the two. Some anxiety was expressed at the randomisation of monochorionic twins, because they are often delivered before 37 weeks (RCOG 2008 Green-top guideline 51; Visintin C, Mugglestone MA, James D, Kilby MD. BMJ 2011;343:d5714). However, the numbers of monochorionic twins were small and equally distributed in the two groups (19 in intervention and 21 in comparison). The smallness of the numbers can limit the power to obtain a reliable answer to this question.
How could we interpret the composite primary outcome in this study?
In this study, perinatal mortality and serious neonatal morbidity were combined and used as a composite primary outcome. The suitability of composite outcomes is always debated. When various elements combined in the composite influenced by the intervention in the same direction (reduction in rate through the effect achieved) are pre-specified, there is greater confidence in the clinical significance of the results. The clinical significance of the average difference in birthweight of 90 g was unclear; however, the rate of small for gestational age was lower with intervention. Maternal outcomes were also considered to be clinically important for decision making.
Would a twin-specific growth chart be more appropriate?
There are twin-specific growth charts (Joseph KS, Fahey J, Platt RW, Liston RM, Lee SK, Sauve R, Liu S, Allen AC, Kramer MS. Am J Epidemiol 2009;169(5):616–24.) which could have been used in this trial. For the comparison of groups, these would be unlikely to change the inferences made.
What are the implications of under-recruitment?
BJOG has a guide on whether we can trust the results of trials that are stopped early (Khan KS, Hills R. BJOG 2006;113(7):766–8). The concern in this situation is that there may be a lack of statistical power to detect a difference when one exists. This study failed to recruit the planned sample size because of the lack of ongoing funding, but, as the difference detected for the main outcome measure was statistically significant, stopping early did not lead to an unreliable conclusion.
Take home messages
Allocation concealment in randomised controlled trials is a critical feature for study validity. The interpretation of a composite primary outcome requires caution. Just because a study stops early, it does not automatically mean that it is unreliable. In women with an uncomplicated twin pregnancy delivered at 37 weeks of gestation, the composite outcome was favourable compared with control.