Classification of stillbirth

BMJ 2005; 331 doi: http://dx.doi.org/10.1136/bmj.331.7527.1270 (Published 24 November 2005) Cite this as: BMJ 2005;331:1270

Author's reply

  1. Jason Gardosi On behalf of all of the authors., director (gardosi{at}perinatal.nhs.uk)
  1. Perinatal Institute, Birmingham B15 3QE

    EDITOR—We developed the new classification to get away from the limitations of seeking the cause of stillbirth, in favour of determining the conditions which could explain what had happened. This allowed a number of categories to be included which may have undetermined underlying causes, but which in themselves are relevant for an understanding of the events preceding and surrounding the demise.

    For instance, cord prolapse is a relevant category even though it may have a number of underlying, spontaneous, or iatrogenic causes. Fetal growth restriction is another example, being associated with a fivefold increase in risk for a fetus below the 10th customised centile, and an 11-fold risk for a fetus below the 2.5th centile.1 There are indeed many different causes for fetal growth restriction, although we are not aware of any evidence that some types might not be a risk for stillbirth, as Kirk suggests. Constitutional smallness has no increased risk, but we corrected for constitutional variation by using customised centiles. A low customised centile indicates that that fetus was failing to reach its growth potential.

    Kirk mentions the importance of establishing the risk of recurrence for purposes of counselling. This reasoning ought to be extended to fetal growth restriction, which has a high risk of recurrence, but which is currently not recognised in many classification systems. Sebire questions the relevance of a category for fetal growth restriction, as most fetuses with this condition do not die. This is not a convincing argument, as many other conditions used in ReCoDe and in conventional classification systems are not lethal either—for example, most fetuses do well in pregnancies complicated by maternal hypertensive disease.

    Postmortem examinations, when agreed to by the bereaved parents, are important for establishing causes. However, pathologists are often not rewarded for their efforts by an outdated classification with which they still end up with many “unexplained” reports. The proportion of stillbirths classified as unexplained under the old system has in fact been on the increase in relation to other causes (figure 3.2).2 In addition, overall stillbirth rates have increased,3 which means that more and more women are being told that they have had an unexplained stillbirth. Reducing the proportion of stillbirths which are classified as unexplained will, we believe, not only improve clinical care but allow a sharper focus for research into causes.

    Erwich et al recognise the difficulties with trying to establish a cause, especially in areas of overlapping pathophysiology. They question the need for hierarchy, but we believe that this is essential for reproducibility, an important criterion for a good classification system. Wherever possible, ReCoDe categories are mapped to ICD 10 codes, and the conditions are assigned and ordered by means of a computer algorithm. The secondary classification softens the hierarchy and increases descriptiveness by allowing another relevant condition to be coded. The example they quote (fetal growth restriction with few secondary codes for placental insufficiency) shows a strength of our system: growth restriction was recognised retrospectively, even though placental insufficiency may not have been established on histological grounds. Many placentas are also not sent for examination, a continuing problem which pathologists are seeking to address.

    Erwich et al point out that a separate classification for stillbirths might obscure related neonatal mortality due to iatrogenic preterm delivery. This is an argument for looking at stillbirth and neonatal death rates separately as well as in combination—that is, as perinatal mortality rates. However, the conditions which led to the death and their implications may be substantially different, and the quality of neonatal care may have contributed. Thus stillbirths need their own classification system.


    • Competing interests None declared.


    View Abstract