|Home | About | Journals | Submit | Contact Us | Français|
Editor—In their study on the prospective risk of unexplained stillbirth in singleton pregnancies at term Cotzias et al performed a secondary analysis1 of data that we published.2 The proposed rationale for this mathematical exercise was that no published data provide accurate gestation-specific risks of stillbirth at term. In fact, our publication evaluates risks of stillbirth and neonatal and infant mortality throughout pregnancy.2
The original data for the North East Thames region, with the addition of births for 1992, were used to inform the confidential inquiry into stillbirths and deaths in infancy (CESDI) concerning antepartum stillbirths (L Hilder and N Datta, unpublished data). When both fetal and neonatal causes cited on stillbirth registrations were used the proportion of stillbirths that are unexplained increased from 0.1392 at 37 weeks to 0.5000 at 43 weeks. Even if stillbirth is explainable it is not necessarily preventable and is inevitably unexpected. We acknowledge that when doctors deal with parents who have had a recent stillbirth, information on aetiology is invaluable. We continue to believe, however, that when prospective risks are being estimated for clinical purposes all stillbirths should be included.
The results presented by the authors are critically flawed by the reliance on cumulative prospective risk of stillbirth. The authors total the number of stillbirths in the remaining weeks of pregnancy in order to estimate the prospective risk of stillbirth at a specific week of gestation. This methodology produces clinically implausible results, explaining the authors' paradoxical conclusion that the risk of stillbirth at 38 weeks is greater than that at 42 weeks. If this was taken to absurdity their prospective risk of stillbirth at 24 weeks would be 1 in 330 while that at 43 weeks would be 1 in 633.
We analysed data from 158945 singleton pregnancies in the North East Thames region in which the fetuses were congenitally normal (table). Our data give clinically relevant estimates of prospective weekly risk of stillbirth, showing a sharp increase in risk after 40 weeks. Current strategies of elective induction of labour after 41 weeks seek to avert fetal death without increasing rates of obstetric intervention. Until large enough trials are performed to justify this course of management the correct interpretation of observational data from large population based analysis is vital.
Editor—Cotzias et al report the risk of unexpected stillbirth as a function of increasing gestational age.1-1 We would like to comment on the methodology of this risk estimation.
Fetal mortality can be expressed by two measures. It can be expressed as a risk (or cumulative incidence), defined as the number of stillbirths diagnosed at or beyond a specific week divided by the number of ongoing pregnancies at the beginning of that week. Alternatively it can be expressed as a rate, defined as the number of stillbirths diagnosed at a specific week divided by the number of ongoing pregnancies at mid-week (an approximation of person time at risk).
Cotzias et al point out that the risk of stillbirth is similar across gestational weeks at term. In an earlier paper based on the same database, results were expressed as rates, increasing with gestational age.1-2 Although the risk of death is similar across gestational age, the period during which the fetus is at risk is clearly much longer at 35 weeks (about five weeks) or 38 weeks (about two weeks) than at 42 weeks (only a few days). The figure shows Cotzias et al's data presented as risks and as rates.
In their comments in their electronic response about Cotzias et al's paper 1-3 [see next letter] Yudkin and Redman refer to their own study, where the risk during the next two weeks increased with gestational age.1-4 This shows that, in the interpretation of risk estimates, the period for which the risk is present should not be ignored. Reporting fetal mortality as cumulative incidence is misleading if readers are not told the length of the period at risk. The clinician caring for a pregnant woman, and the woman herself, balance the risk of stillbirth with the probability of spontaneous onset of labour during a specific period (for example, until the next visit). The clinically relevant measure is either the rate or the risk of stillbirth during a limited period.
Cotzias et al state that at 38 weeks the risks of stillbirth near term exceed those at 42 weeks. Obstetricians may be tempted to induce labour at 38 weeks on the basis of these conclusions. This policy is not based on scientific evidence. In contrast, results of randomised trials suggest a reduction in perinatal mortality when labour is induced at 41 weeks, a time when the stillbirth rate increases.1-5 Until randomised trials of earlier induction of labour or of specific tests show a reduction in perinatal mortality, we will continue to believe that gestation should continue until spontaneous onset of labour up to 41 weeks.
Editor—Cotzias et al estimated that the prospective risk of stillbirth at 38 weeks (1 in 529 ongoing pregnancies) was greater than at 42 weeks (1 in 565).2-1 On the basis of this finding, they question whether routine delivery should be considered at 38 weeks, rather than at 42 weeks as is usually recommended.
This approach makes sense only if the costs of delivery are equal at each gestational age. But, as the authors imply, this is not the case. Not only would one expect a higher incidence of caesarean section and iatrogenic prematurity at earlier gestational ages, but important non-clinical outcomes, such as maternal satisfaction, would differ too. Routine induction as early as 38 weeks would be unacceptable to most women. Although the risk of prospective stillbirth at a given gestation represents the greatest cost of allowing the pregnancy to continue, a knowledge of all the costs involved, including those associated with delivery at each gestation, is needed to inform a policy of routine early delivery.
The alternative form of management is to improve fetal monitoring so that impending fetal death can be identified and pre-empted. In our study of births to Oxfordshire residents at the John Radcliffe Hospital in 1978-85 we calculated the risk not of prospective stillbirth at any future gestation but of impending stillbirth—that is, stillbirth occurring in the next two weeks.2-2
The risk of unexplained impending stillbirth was very low until 40 weeks (0.2-0.4/1000 ongoing pregnancies) and increased to 1.2/1000 at 41 weeks and beyond. These results set in context the difficulty of identifying the rare impending fetal death and also (subject to cost considerations) support a policy of increased surveillance at and beyond 41 weeks. A recent update of this analysis, based on 44450 singleton births in 1991-8, shows a similar pattern, although the peak level of risk, of 1.3/1000, is delayed until 42 weeks.
Our updated analysis suggests, contrary to the assumption of Cotzias et al, that the proportion of total stillbirths that are unexplained stillbirths is not constant between 35 and 42 weeks but rising, being highest at and after 39 weeks. Applying these changing proportions to the data of Cotzias et al would support the view that, contrary to their claim, the prospective risk of unexplained stillbirth is greatest in post-term pregnancies.
Editor—We are surprised by Hilder et al's comments, as we invited collaboration and discussed initially the study design and later the results with their corresponding author. We are unable to understand why they view our analysis as critically flawed: of course the prospective risk of stillbirth is higher at 24 weeks than at any other subsequent gestation, because there is a longer period during which stillbirth can occur. We expressed the risk of stillbirth for the remainder of the pregnancy, whereas Hilder et al have expressed the risk as a rate over the next week3-1 and Yudkin et al over the next fortnight.3-2
The fact that our figures suggest broadly comparable risks per pregnancy at 38 weeks and at 42 weeks is entirely consistent with the observation that the risk per day or week at 42 weeks is many times higher than that at 38 weeks. This issue of risk versus rate per week or fortnight is illustrated in Boulvain et al's graph. In our view what matters to women at term is the overall chance of their baby dying, not the chance in the next day, week, or fortnight. The point of our analysis was to inform women and obstetricians what the chance of a stillbirth was once fetal maturity was achieved.
Yudkin and Redman observe that the percentage of unexplained stillbirths increases with gestation. As we argued in the short report, however, the more relevant risk is that of unexpected, rather than unexplained, stillbirth, which in our view includes all stillbirths after 38 weeks. We agree that routine delivery at term would have cost implications. Obviating these risks, however, should be seen in the context of changing obstetric risk:benefit ratios and the wider debate on optimal timing and mode of delivery.
Yudkin and Redman have suggested the obvious alternative of improving fetal monitoring techniques. There are currently no data to support routine monitoring from 38 weeks, just as after 41 weeks meta-analysis clearly shows expectant management with fetal monitoring is associated with a higher stillbirth rate than is elective delivery.3-3