Randomised controlled trial: effect of delayed versus early umbilical cord clamping on neonatal outcomes and iron status at 4 months
- By: Candice L Downey, Susan Bewley
“Randomised controlled trial: effect of delayed versus early umbilical cord clamping on neonatal outcomes and iron status at 4 months” by Ola Andersson and colleagues (BMJ 2011;343:d7157, doi:10.1136/bmj.d7157)
Objective—To investigate the effects of delayed umbilical cord clamping, compared with early clamping, on infant iron status at 4 months of age in a European setting.
Design—Randomised controlled trial.
Setting—Swedish county hospital.
Participants—400 full term infants born after a low risk pregnancy.
Intervention—Infants were randomised to delayed umbilical cord clamping (≥180 seconds after delivery) or early clamping (≤10 seconds after delivery).
Main outcome measures—Haemoglobin and iron status at 4 months of age with the power estimate based on serum ferritin levels. Secondary outcomes included neonatal anaemia, early respiratory symptoms, polycythaemia, and need for phototherapy.
Results—At 4 months of age, infants showed no significant differences in haemoglobin concentration between the groups, but infants subjected to delayed cord clamping had 45% (95% confidence interval 23% to 71%) higher mean ferritin concentration (117 μg/L v 81 μg/L, P<0.001) and a lower prevalence of iron deficiency (1 (0.6%) v 10 (5.7%), P=0.01, relative risk reduction 0.90; number needed to treat=20 (17 to 67)). As for secondary outcomes, the delayed cord clamping group had lower prevalence of neonatal anaemia at 2 days of age (2 (1.2%) v 10 (6.3%), P=0.02, relative risk reduction 0.80, number needed to treat 20 (15 to 111)). There were no significant differences between groups in postnatal respiratory symptoms, polycythaemia, or hyperbilirubinaemia requiring phototherapy.
Conclusions—Delayed cord clamping, compared with early clamping, resulted in improved iron status and reduced prevalence of iron deficiency at 4 months of age, and reduced prevalence of neonatal anaemia, without demonstrable adverse effects. As iron deficiency in infants even without anaemia has been associated with impaired development, delayed cord clamping seems to benefit full term infants even in regions with a relatively low prevalence of iron deficiency anaemia.
Why do the study?
The optimal timing of umbilical cord clamping after birth is debated. In this study the authors define delayed cord clamping as more than three minutes after delivery and early clamping as within 10 seconds of birth. Delayed clamping has been linked with a decreased risk of iron deficiency anaemia because it allows physiological transfusion of placental blood to the neonate. This is important because childhood anaemia might be associated with poor neurodevelopment. Uncommon adverse effects of deferring clamping can include polycythaemia, hyperbilirubinaemia, and respiratory problems. Most of the available cord clamping data, however, come from studies done in low and middle income countries. Much of the evidence is low quality. The aim of this trial was to provide robust evidence for delayed versus early umbilical cord clamping in infants born in a high income country, Sweden.
What study design was used?
A randomised controlled trial is the most robust study design when comparing interventions. It provides the strongest evidence for cause and effect because this type of study design eliminates as much bias as possible. Examples of such bias include confounding bias, selection bias, and measurement bias.
This randomised controlled trial had two groups: those receiving the new intervention (delayed clamping) and the control group (early clamping). A control group allows researchers to compare the effect of one intervention with another. Ideally, the only difference between the intervention and control groups of a trial would be their exposure to the intervention (here, delayed clamping compared with the standard procedure of early clamping). Unfortunately, human studies will always have some degree of random variation in terms of individual characteristics within the sample population. This can lead to confounding—a type of bias that occurs when this variation exposes a factor that will affect the outcome, independent of the intervention.
The authors used eligibility criteria to eliminate confounding factors they could predict, such as babies with known congenital malformations. Randomisation minimises the effects of confounding by ensuring that each of the study participants, and therefore each confounding factor, has an equal chance of being in the intervention or control groups (see table). Confounding factors should therefore affect each group’s results equally, cancelling out the overall effect. The randomisation was done by a computer program, eliminating selection bias. Another way to combat confounding is to increase the sample size. This dilutes the effect of the confounding factor but does not eliminate it.
|Cord clamping||P value of difference|
|Early (n=189)||Delayed (n=193)|
|Age (years)||31.6 (4.2)||30.9 (4.7)||NS|
|Parity (including study child)||1.76 (0.85)||1.74 (0.73)||NS|
|Weight at first antenatal visit (kg)||66.6 (12.1)||67.4 (12.0)||NS|
|Body mass index||23.8 (4.0)||24.2 (3.7)||NS|
|Haemoglobin at first antenatal visit (g/L)||128 (9)||128 (11)||NS|
|No (%) with Rh negative blood group||27 (14)||40 (21)||NS|
|No (%) of vaginal deliveries:|
|Non-instrumental||176 (93)||174 (90)||NS|
|Vacuum extraction||11 (6)||16 (8)||NS|
|Forceps||1 (0.5)||1 (0.5)||NS|
|No (%) of caesarean sections||1 (0.5)||2 (1)||NS|
|Gestational age (weeks)||40.1 (1.1)||40.0 (1.1)||NS|
|No (%) of males||93 (49)||86 (45)||NS|
|No (%) with Apgar score 7–10 at 1 minute||181 (96)||188 (97)||NS|
|Birth weight (g)||3533 (486)||3629 (460)||0.05*|
|Birth length (cm)||50.8 (1.9)||50.9 (1.9)||NS|
|Head circumference (cm)||34.7 (1.4)||34.9 (1.4)||NS|
|Umbilical cord haemoglobin (g/L)||163 (16)||159 (18)||0.01†|
|Umbilical cord packed cell volume||0.49 (0.04)||0.47 (0.05)||0.01‡|
An important element of randomised controlled trial design is blinding. This is the concealment of the patient’s allocated group from the patient (single blinding) or from the patient and the person administering the intervention (double blinding). Blinding was not achieved in this study as it would be impossible to hide the timing of cord clamping from either midwives or mothers. Blinding is more important in trials where the outcome is subjective, such as symptom control. The main outcome of this study was based on laboratory blood tests of iron status, and so there was little chance of biased assessment.
What did the authors do?
Having first informed pregnant women of the study at the antenatal clinic, the authors recruited eligible women on the delivery ward. A total of 400 participants were recruited (200 in each arm) from more than 1500 approached. This number was predetermined by a pre-study power analysis. Assuming they would find similar results to a previous study, the researchers calculated the number of participants required to achieve statistical significance with those results. Smaller samples decrease the importance of a result because it is more likely that the result is due to chance.
The intervention group received delayed cord clamping (more than three minutes after delivery) whereas the control group received early cord clamping (within 10 seconds). Blood tests were taken soon after delivery, two to three days after birth, and at 4 months. The primary outcome of the study was iron status, determined from these blood results. Other (secondary) outcomes included neonatal anaemia, early respiratory symptoms, polycythaemia, and need for phototherapy.
What did the study find?
Iron status was determined by a number of indicators including serum ferritin and haemoglobin concentration. No significant difference in haemoglobin concentration was found between the early and delayed cord clamping groups (both averaged 113 g/L). Mean ferritin concentration was 117 μg/L for the delayed clamping group and 81 μg/L for the early clamping group; therefore mean ferritin concentration was 45% higher in the delayed clamping group. This had a 95% confidence interval range of 23 to 71, meaning that if the study was repeated, 95 times out of 100 the difference in ferritin concentrations between the groups would be between 23% and 71%.
The P value indicates the probability of getting these results if the null hypothesis is true. The null hypothesis is that early versus delayed cord clamping has no effect on iron status at 4 months. In other words, with P <0.001 it is very unlikely (less than 1 in 1000 chance) that the null hypothesis is correct.
Although the prevalence of anaemia was not different between the groups, iron deficiency was significantly more prevalent in the early cord clamping babies. The number needed to treat was 20 (95% confidence interval 17 to 67)—that is, the number of babies who need delayed cord clamping for one baby to avoid iron deficiency is 20.
There were no significant differences between the early and delayed cord clamping groups in terms of baseline characteristics, except for a difference in birth weight explained by placental transfusion. There was no significant difference in secondary outcomes.
What are the strengths and limitations of the study?
The strengths lie in the study design. A large, well conducted randomised controlled trial is the most robust test of an intervention. A number of systems can be used to appraise the quality of a randomised controlled trial; the BMJ recommends the Critical Appraisal Skills Programme toolkit, and most evaluate issues such as randomisation, blinding, follow-up, and study size. Despite the size and randomisation in this study, confounding factors may still be present, such as genetic variation, birth weight, and growth restriction. However, it is positive that the results mimic previous similar trials and make biological sense.
Of the 400 neonates recruited, only 350 were evaluated at 4 months, largely as a result of participants withdrawing or violating the inclusion criteria of normal births—for instance, by unexpectedly having an assisted delivery. The exclusion of 50 babies may have caused biased results, especially if the criteria were linked to the intervention.
Not every baby received the treatment they had been allocated, mostly because of complications at birth. However, it would not be fair to exclude these babies from the results as we would no longer be comparing like with like, despite the numbers being similar between the two arms (26 and 30 babies did not receive the allocated intervention in the early and delayed cord clamping groups, respectively). Instead, the authors avoided analysis bias by employing an intention to treat analysis. This means that babies were included in the analysis by allocation group, regardless of the treatment they actually received. This allows us to evaluate confidently the effects of allocating the timing of cord clamping, and not just the effects on babies who adhered to protocol.
An important limitation is the use of indirect estimates of iron status, such as ferritin concentration. Ferritin is an acute phase protein and may be incorrectly interpreted in inflammatory conditions. The authors tried to minimise the effects of using indirect indicators by measuring a range of variables such as transferrin saturation and mean cell volume, all of which showed significant differences in iron status between cord clamping groups, providing stronger evidence for their findings.
What does the study mean?
Deferring cord clamping for three minutes in healthy, term babies improves their iron status at 4 months. Although statistically significant, this finding is not necessarily clinically significant. The results cannot be generalised to all births as the sample population did not include complicated pregnancies or births. Iron status cannot be used as a surrogate outcome for neurodevelopmental status as there is no direct link between the two, merely an association. The long term effects are not known, although this paper is the first in an anticipated series of reports from this trial, which will include longer term outcomes. Nevertheless, healthy term babies might benefit in the short term from delayed cord clamping. Implications for practice might include noting the timing of cord clamping, as an indicator of placental transfusion.
In this study the intervention was delayed cord clamping because early clamping was standard at the Swedish hospital. Early cord clamping is an intervention, despite its routine clinical use. Observe the practices of midwives and doctors on your obstetrics rotation and reflect on the influence of routine.Candice L Downey, foundation doctor1, Susan Bewley, professor of complex obstetrics2
1Leeds Teaching Hospitals Trust, Leeds, UK, 2Division of Women’s Health KCL, Women’s Health Academic Centre, King’s Health Partners, St Thomas’ Hospital, London, UK
Correspondence to: firstname.lastname@example.org
Competing interests: SB has shared registered intellectual property rights in the BASICS trolley (Bedside Assessment, Stabilisation and Immediate Cardiorespiratory Support), for which all profits will be given to charity.
Provenance and peer review: Commissioned; not externally peer reviewed.
- Chaparro CM, Neufeld LM, Tena Alavez G, Eguia-Liz Cedillo R, Dewey KG. Effect of timing of umbilical cord clamping on iron status in Mexican infants: a randomised controlled trial. Lancet 2006;367:1997-2004.
- Milton Keynes Primary Care Trust. Critical Appraisal Toolkit for Randomised Controlled Trials. Milton Keynes, 2002. www.hello.nhs.uk/.../CAT6-Randomised_Controlled_Trial.pdf.
Cite this as: BMJ 2012;20:e2761
- Published: 08 May 2012
- DOI: 10.1136/sbmj.e2761