The utility of clinical tests of eligibility for a trial of labour following a caesarean section: A decision analysis

Objective There is evidence to suggest that women who have a failed trial of labour following caesarean section are at the highest risk of major morbid events, compared with those who have a successful trial of labour or elective repeat caesarean section. Attempts are being made to predict, either by radiological or ultrasonic measurements or clinical factors, who should attempt a trial of labour after a prior caesarean section and who should not, in the hopes of reducing the number of failed trials of labour with their associated morbidity. The goal of this study was to determine the necessary sensitivity and specificity for such a test to be clinically useful in this setting. Methods A decision analytic approach was used which compared two strategies for treating women with a prior caesarean section: Trial of labour for all or application of a hypothetical test, the results of which would determine whether a woman would be offered a trial of labour. It was assumed that the goal of the strategies was to minimise the number of major maternal morbid events (hysterectomy, uterine rupture, operative injury) with an acceptable caesarean section rate. Probability estimates for these outcomes were obtained from the existing literature. We evaluated several combinations of sensitivity and specificity to determine the impact of these characteristics on the number of major complications and the number of caesarean sections, compared with the strategy of trial of labour for all women. The results are reported per 1000 women with a previous caesarean section. Results Under baseline assumptions, a clinical test for determining eligibility for a trial of labour must have both a sensitivity and specificity in excess of 75% to obtain a reasonable trade-off between reduction in morbidity and the total rate of caesarean sections. In women with a low empiric likelihood of a successful trial of labour, tests with lesser degrees of both sensitivity and specificity are reasonable. In women with a high empiric likelihood of a successful trial of labour, much higher sensitivities and specificities are required. Conclusions In developing tests to determine to whom to offer a trial of labour, investigators and clinicians must realise that in most cases, a highly sensitive and specific test is needed, although this is dependent on the empiric likelihood of the success of a trial of labour.