delivery at first birth could be an indica- tor of low UI risk without being really protective. In the current state of knowl- edge, this alternative explanation is not.
BJOG Exchange
Quebec, which probably explains why the rate of PTB is the same as in Canada. In conclusion, the Quebec programme focuses on the wrong risk factors. Moreover, this costly public health intervention probably does harm by increasing harmful lifestyle habits (smoking, drinking, and physical inactivity) and by reducing the birth interval. The programme should only offer early work leave to the very small group of women who are exposed to dangerous working conditions, including self-employed women who are presently excluded. Decision makers should use an evidence-based approach to promote healthy pregnancies. Finally, there is an urgent need to test my hypothesis that this programme does harm. &
References 1 Takser L. Pregnant pause: the need for an evidence-based approach for work leave in the prevention of preterm birth and low birthweight. BJOG 2013;120:517–20. 2 Palmer KT, Bonzini M, Harris EC, Linaker C, Bonde JP. Work activities and risk of prematurity, low birth weight and pre-eclampsia: an updated review with meta-analysis. Occup Environ Med 2013;70:213–22. 3 Palmer KT, Bonzini M, Bonde JP. Multidisciplinary Guideline Development Group; Health and Work Development Unit; Royal College of Physicians; Faculty of Occupational Medicine. Pregnancy: occupational aspects of management: concise guidance. Clin Med 2013;13:75–9. 4 Fall A, Goulet L, V ezina M. Comparative study of major depressive symptoms among pregnant women by employment status. Springerplus 2013;2:201. 5 Croteau A, Marcoux S, Brisson C. Work activity in pregnancy, preventive measures, and the risk of preterm delivery. Am J Epidemiol 2007;166:951–65. 6 England LJ, Grauman A, Qian C, Wilkins DG, Schisterman EF, Yu KF, et al. Misclassification of maternal smoking status and its effects on an epidemiologic study of pregnancy outcomes. Nicotine Tob Res 2007;9:1005–13.
L Takser Departement de Pediatrie, Universite de Sherbrooke, Sherbrooke, QC, Canada Accepted 27 July 2013. DOI: 10.1111/1471-0528.12440
1578
Does the SWEPOP (Swedish Pregnancy, Obesity and Pelvic Floor) study suffer from a selection bias?
Sir, Gyhagen et al.1 have reported a prevalence of urinary incontinence (UI) 20 years after the first birth that is higher in cases of vaginal delivery (40%) than in caesarean delivery (29%). They conclude that it would be necessary to achieve eight or nine caesarean sections to avoid one case of UI, thereby suggesting that caesarean section has a protective effect. This conclusion seems imprudent because a number of preconditions have not been met. A prophylactic intervention should ideally occur before disease onset. UI is particularly common in late pregnancy, when it affects 30–50% of nulliparous women.2 In reports from the SWEPOP study the UI starting date was not specified, probably because it is not known.1,3 We therefore have no precise idea of the antenatal prevalence of UI in the group of women with infants delivered vaginally compared with the caesarean section group, or whether prevalence was similar in both groups. The apparent protective effect of caesarean section in this survey could be explained by a selection bias. Two studies have shown that nulliparous women whose pregnancy will end with a caesarean section have prenatal characteristics putting them at a reduced risk of UI such as lower urethral mobility or levator hiatus.4,5 The mode of delivery could quite possibly be a consequence of antenatal pelvic floor characteristics. Urethral mobility measured during pregnancy is predictive of postnatal UI.2 Caesarean delivery at first birth could be an indicator of low UI risk without being really protective. In the current state of knowledge, this alternative explanation is not refutable. Only a randomised trial could answer the question. The available randomised trial focusing on breech presentation at term is negative with respect to stress UI prevention.
The absence of a pathophysiological mechanism to explain a causal link between vaginal delivery and UI is also to be considered. The main mechanism of stress UI seems to be intrinsic sphincter deficiency and not urethral mobility.6 Postnatal values of maximum urethral pressure or sphincter volume are similar in cases of vaginal and of caesarean delivery.2,4 Six months after vaginal birth, the values of urethral mobility are similar to the prenatal values.3 In the latest analysis by Gyhagen et al.,3 it is disturbing to note that the distributions of UI type, severity and bothersomeness are similar in incontinent women regardless of delivery mode. A protective effect of caesarean section would have to be accompanied by a different distribution, with more stress UI, more severe UI and more bothersome UI in the vaginal delivery group. Instead the authors observed 12.4% with severe stress UI and 18.1% with troublesome stress UI in the vaginal delivery group versus 13.7 and 17.7%, respectively, in the caesarean section group. In conclusion, a selection bias cannot be excluded and we need to be careful not to promote a prophylactic caesarean section for which the expected benefits remain unclear. &
References 1 Gyhagen M, Bullarbo M, Nielsen TF, Milsom I. Prevalence and risk factors for pelvic organ prolapse 20 years after childbirth: a national cohort study in singleton primiparae after vaginal or caesarean delivery. BJOG 2013; 120:152–60. 2 Fritel X, Ringa V, Quiboeuf E, Fauconnier A. Female urinary incontinence, from pregnancy to menopause, a review of epidemiologic and pathophysiologic findings. Acta Obstet Gynecol Scand 2012;91:901–10. 3 Gyhagen M, Bullarbo M, Nielsen TF, Milsom I. A comparison of the long-term consequences of vaginal delivery versus caesarean section on the prevalence, severity and bothersomeness of urinary incontinence subtypes: a national cohort study in primiparous women. BJOG 2013;120:1548–55.
ª 2013 Royal College of Obstetricians and Gynaecologists
BJOG Exchange
4 Toozs-Hobson P, Balmforth J, Cardozo L, Khullar V, Athanasiou S. The effect of mode of delivery on pelvic floor functional anatomy. Int Urogynecol J Pelvic Floor Dysfunct 2008; 19:407–16. 5 Dietz HP, Moore KH, Steensma AB. Antenatal pelvic organ mobility is associated with delivery mode. Aust N Z J Obstet Gynaecol 2003;43:70–4. 6 DeLancey JOL. Why do women have stress urinary incontinence? Neurourol Urodyn 2010;29:S13–7.
X Fritela & A Fauconnierb a
Poitiers University Hospital, INSERM CIC802, Poitiers, France bPoissy– Saint-Germain Hospital, Versailles-Saint-Quentin University, Poissy, France
Accepted 12 February 2013. DOI: 10.1111/1471-0528.12417
Does the SWEPOP (Swedish Pregnancy, Obesity and Pelvic Floor) study suffer from a selection bias?
Authors’ reply Sir, We thank Professors Xavier Fritel and Arnoud Fauconnier1 for their contribution to the discussion about the aetiology of pelvic floor disorders such as urinary incontinence (UI) in women after childbirth. Fritel and Fauconnier argue that caesarean delivery cannot be prophylactic for UI because it does not occur before disease onset and it is known that UI is common in late pregnancy. Urinary incontinence is multifactorial and pregnancy and childbirth are only two of many independent risk factors. It is well known that UI exists in nulliparous women and even in men and children.2 We are of the opinion that the difference in prevalence of UI between vaginal delivery and caesarean section in our study3,4 represents the excess risk resulting from pelvic floor injury at vaginal birth compared with caesarean section. We assume that UI symptoms before and during pregnancy are similarly distributed between the
two groups. Fritel and Fauconnier challenge our assumption and introduce the hypothesis that antenatal maternal characteristics cause a selection bias, which allocates women—with tissue characteristics favouring continence but hindering vaginal delivery—to acute caesarean section, explaining the lower prevalence of UI in this group. Observations however contradict what must follow from this allocation bias hypothesis. If this allocation hypothesis is correct, it must follow that UI prevalence should differ between women delivered by acute as opposed to elective caesarean section. This was not observed in our study. Elective caesarean section is performed mainly on medical indications, irrespective of pelvic floor status and thus UI prevalence following elective caesarean section ought to differ from acute caesarean section and be more similar to that after VD if the hypothesis of Fritel and Fauconnier is correct. However, UI prevalence was similar after acute and elective caesarean section and for both groups was significantly lower than after vaginal delivery, which contradicts the hypothesis of Fritel and Fauconnier. The allocation bias hypothesis can also be tested against the findings in the study of Nielsen et al.5 Women with parity of one (n = 160) who had been delivered by caesarean section due to abnormally slow progress of labour (dystocia) were referred to trial of vaginal delivery (pelvic disproportion excluded by X-ray pelvimetry). They were compared with a matched control group of women who had not been delivered by caesarean section earlier. The rate of vaginal birth was 94% in both groups, indicating that the slow progress of labour during the first delivery (in the experimental group) could not be attributed to particular tissue or muscle characteristics of these women.5 The proposition that stress UI (as opposed to mixed UI and urge UI) is the subtype of UI especially linked to vaginal delivery—shown by a higher
ª 2013 Royal College of Obstetricians and Gynaecologists
prevalence when compared with caesarean section—has not been demonstrated in earlier reports. With sufficient power, as in our study, it is convincingly shown that all three subtypes are more common after vaginal delivery in comparison with caesarean section (odds ratios 1.42 for stress UI, 1.66 for urge UI, and 1.46 for mixed UI, table 1 in Ref. 3), which means that the association in fact was numerically weakest for stress UI and strongest for urge UI. The common view that urge UI is especially linked to caesarean section can also be questioned by these results. Regardless of the higher prevalence of UI and UI-subtypes after vaginal delivery, the distribution of subtypes among those with incontinence is strikingly similar between the two modes of delivery and does not contradict the fact that they were more prevalent after vaginal delivery. The explanation for this striking similarity in distribution is, however, a matter for future research. The cause of Fritel’s and Fauconnier’s criticism for inconsistency in our results is based on the confusion of prevalence and proportionality. We also find it astounding to read that there is an ‘absence of pathophysiological mechanism to explain a causal link between vaginal delivery and urinary incontinence’. Although all the details are not yet at hand there is an abundance of evidence from magnetic resonance imaging, multidimensional ultrasound, neurophysiological studies and data simulation studies demonstrating that it is during the second stage of delivery, when the head of the fetus passes through the pelvic canal, that injuries to the muscles, nerves and connective tissues of the pelvic floor occur due to compression, stretch, laceration and ischaemia. These injuries were not observed after caesarean section. Finally, the concept of number needed to treat is not a recommendation to promote caesarean section but a measure of the protective effect of an intervention based on prevalence of the outcome of interest, and is a simple, practical and
1579
