Incidence and predictors of severe obstetric morbidity: case-control studyCommentary: Obstetric morbidity data and the need to evaluate thromboembolic disease

The study by Waterstone et al1, which reports the incidence and
predictors of severe obstetric morbidity, contributes significantly to our
understanding of maternal health in the U.K. and is welcomed. We would,
however, urge caution in the interpretation of some of the reported
results, particularly with regard to potential pathways of effects.

For instance, all these examples of serious obstetric morbidity have
multiple causes, and these causes may occur at different points before or
during pregnancy, and during or after labour. What happens at a later
stage may be conditional on what has happened earlier. There is no
indication that this temporal sequence was considered in the model.

Many factors may contribute independently to the outcomes. For
example, a plausible explanation for the relatively low odds ratio (1.36;
95% confidence interval 0.99 to 1.88) for post term induction of labour is
that the association is masked by the inclusion of emergency caesarean
section in the statistical model. Induction of labour is independently
associated with emergency caesarean section, which in turn is associated
with haemorrhage and infection. Its inclusion in the model would be
justified if emergency caesarean section behaved as a confounder in the
relationship between induction of labour and maternal morbidity. We would
suggest that emergency caesarean section is on the causal pathway and its
inclusion in the model makes it impossible to determine the true effect of
induction of labour as a predictor for maternal morbidity.

The odds ratio for emergency caesarean section suggests that women
who underwent the procedure were four times more likely to suffer
morbidity, when compared to women who did not. The commentary that
follows this paper by D.Murphy2, states that because the odds ratio
reported is adjusted, it reflects the additional risk of the procedure
itself. However, the estimated effect of emergency caesarean section
reported in the study actually represents the combined effect of emergency
caesarean section and those confounding variables that had not been
accounted for. Drawing a causal interpretation from this four-fold odds
ratio is incorrect and we regard the statement in the Abstract that
‘Caesarean section quadruples the risk of morbidity’ as misleading,
particularly since no mention is made of the fact that this statement
really only applies to emergency caesarean sections.

There is an upward trend in published caesarean section rates; this
increase may in part reflect changes in clinical beliefs about the
relative safety of caesarean section and vaginal birth. There is a paucity
of data available to help inform decision-making in this area and there is
an urgent need for further research to evaluate maternal morbidity for
caesarean section in comparison to vaginal delivery.

References:

1. Waterstone M., Bewley S. and Wolfe C. (2001) ‘Incidence and predictors
of severe obstetric morbidity: case-control study’ BMJ 322 1089-1093.

2. Murphy DJ. Commentary: Obstetric morbidity data and the need to
evaluate thromboembolic disease. BMJ 322

Authors:

Shantini Paranjothy

Clinical Research Fellow, CESU

Royal College of Obstetricians and Gynaecologists

Christ Frost

Senior Lecturer in Medical Statistics

Medical Statistics Unit,
London School of Hygiene and Tropical Medicine

Diana Elbourne

Reader in Health Care Evaluation

Medical Statistics Unit,
London School of Hygiene and Tropical Medicine

Jane Thomas

Director, CESU

Royal College of Obstetricians and Gynaecologists