Sir, we read with interest the article by Groutz et al., 1 who state that a first vaginal delivery after the age of 37 years is associated with a higher risk of stress urinary incontinence (SUI) than a first delivery before the age of 30 (hypothesis A) and that caesarean delivery has a protective effect in elderly primiparae (hypothesis B). We think, however, that their conclusions must be interpreted with extreme prudence.
Testing these two hypotheses requires taking four factors into account: parity, mode of delivery, age at delivery, and age at questionnaire. Their respective interactions must also be examined. The difference observed by Groutz between the group of elderly and young primiparae who delivered vaginally (38% vs. 10% postpartum SUI) may be attributable simply to the age difference of the women at the moment of the questionnaire. Rortveit et al. 2 showed that SUI is three times (from 2.0 to 4.4 times) more frequent in nulliparae aged 40-49 years than in those aged 20-29. Although the authors might argue that the exclusion of preexisting SUI is likely to correct for the effect of age at questionnaire, they do not demonstrate this point, and it leads to the selection bias discussed below. To determine the possibility of an effect of age at first delivery, it is essential to assess the interaction between parity (nulliparae vs. primiparae) and age (young vs. old) and therefore to include nulliparae and primiparae of different ages. The absence of nulliparae in Groutz's study prevents the testing of hypothesis A. Other studies do not support the existence of an interaction between age and delivery on SUI risk: Kuh et al. 3 questioned a sample of 1,333 women aged 48 years and found no association between SUI and age at first delivery. Questioning 2,625 women aged 49-61 years, we found no excess risk (OR ¼ 1.0; 95% CI 0.7-1.3) for women who gave birth for the first time after the age of 27. 4 Confirmation of hypothesis (B) requires an assessment of the interaction between mode of delivery and age. It can only be studied with groups of women differing for age (young vs. old) and mode of delivery (vaginal vs. elective caesarean). This study does not have a group of young primiparae with caesarean deliveries.
We must also express doubts about the validity of the constitution of the three groups that are included. We regret that Groutz et al. do not indicate how they avoided the possible recall bias and partial response rate we encountered, although they seem to have used a method similar to ours. 5 It appears to us (but is not explicitly stated) that the authors constituted their groups of women retrospectively. We would like to know the response rate and the factors associated with