Many factors have been investigated for their potential role in the prevention of mother-to-child transmission of HIV. Cesarean section has been associated with a reduced transmission rate in some prospective cohort studies [1,2], but not in others . Since 1985, some centres in Switzerland have observed a policy of routinely offering HIV-positive pregnant women an elective Cesarean section by 38 weeks of pregnancy, whereas others have not. The Swiss prospective cohort study of children born to HIV-infected mothers has previously reported a marked reduction in transmission rate after elective Cesarean section, defined as an abdominal delivery performed before onset of labour and with intact membranes .
In 1994, the AIDS Clinical Trials Group (ACTG) 076 trial  demonstrated that the combined prenatal, intrapartum and postnatal use of zidovudine (ZDV) reduced the rate of mother-to-child transmission of HIV from 25.5 to 8.3%. This led to the recommendation of ZDV prophylaxis in HIV-infected pregnant women in many countries, including Switzerland. The present study reports on the effect of ZDV use in the presence or absence of elective Cesarean section in the Swiss prospective cohort study of children born to HIV-infected mothers.
Since 1986, children of mothers known to be HIV-infected at delivery were enrolled in a prospective study encompassing the whole of Switzerland . For the present study, follow-up data on all children born before 1 July 1996 were analysed. Follow-up until determination of infection status, but at least for 6 months, was required. In twins, the data of only one member of a pair, randomly chosen, were used.
Before the beginning of the study some obstetric centres had already chosen a policy of routinely offering HIV-positive pregnant women an elective Cesarean section by 38 weeks of pregnancy, whereas others did not. Both categories of centres participated equally in the national study. In mid-1994 a national recommendation for ZDV use for the prevention of vertical HIV transmission was issued and accepted by all study participants.
Children were classified as infected if they had detectable HIV antibodies beyond 24 months of life or positive virus detection tests in at least two blood samples by at least two different methods (antigen after heat-mediated disruption of immune complexes [7,8], polymerase chain reaction, virus culture). For classification as uninfected, the absence of HIV-related symptoms and either no detectable HIV antibodies beyond 15 months or negative virus detection tests in at least two samples after the age of 1 month by at least two different methods were required. All other children were classified as indeterminate.
Potential risk factors for vertical transmission ascertained at birth were as follows: maternal age, parity, gravidity (not distinguishing between spontaneous and induced abortions), injecting drug use, health status (HIV-associated systemic symptoms corresponding to class IV according to the 1986 Centers for Disease Control and Prevention classification system) , use of ZDV during pregnancy and intrapartum, sex of infant, gestational age, mode of delivery (elective Cesarean section before onset of labour with intact membranes, other Cesarean section, spontaneous vaginal birth, operative vaginal delivery), use of invasive fetal monitoring (scalp electrode or scalp blood sampling for pH), breastfeeding, and postnatal ZDV prophylaxis.
Approximate confidence intervals (CI) were calculated for odds ratios (OR) and proportions. Proportions were compared by the χ2 or Fisher's exact test. Logistic regression was performed using STATISTICA for Macintosh (StatSoft, Inc, Tulsa, Oklahoma, USA).
A total of 494 children were born during the study period (excluding six randomly chosen members of twin pairs), and 414 (84%) of them had their infection status determined. All twins were concordant (five were not infected, one was indeterminate). Infection status was indeterminate in nine children because of early non-HIV-related death, in 48 because of loss to follow-up, and in 23 because of parental dissent to venipuncture although clinical follow-up was possible. Eighteen of these latter children were younger than 12 months old. Except for ZDV use, distribution of risk factors was similar among children with known and indeterminate infection status. ZDV use was over-represented amongst those with indeterminate infection status because of the younger age of these children [21 indeterminate out of 76 (28%) after ZDV exposure versus 59 (14%) out of 417 without ZDV]. Sixty-seven out of 414 children were found to be HIV-infected, giving a transmission rate of 16.2% (95% CI, 13.0–18.5).
After elective Cesarean section, which was performed in 117 cases, transmission rate was 6% (seven out of 117; 95% CI, 2–12), compared with 20% (59 out of 295; 95% CI, 16–23) for other delivery modes (OR, 0.25; 95% CI, 0.11–0.58; P < 0.001). Twenty per cent of children were born in centres offering elective Cesarean section routinely, and 76% of those were in fact delivered by Cesarean section. Frequency of Cesarean section in the other centres was 16%. Stratified analysis according to type of centre by the Mantel-Haenszel method yielded no evidence for a modification of the effect of elective Cesarean section by the type of centre, and gave a combined OR of 0.33 (95% CI, 0.13–0.88; P < 0.025). The influence of obstetric factors on transmission rate is summarized in Table 1. Taking together all children who had an invasive procedure including vacuum or forceps delivery or invasive fetal monitoring, the transmission rate for this group was 27% (14 out of 52; 95% CI, 16–41). The intermediate group, born by spontaneous vaginal delivery or non-elective Cesarean section and with documented absence of invasive fetal monitoring, had a transmission rate of 18% (39 out of 218; 95% CI, 13–23). Finally, the children with the least invasive mode of delivery, namely elective Cesarean section, had, as stated above, a transmission rate of 6%. Thus, a direct relationship between transmission rate and invasiveness of the delivery process could be demonstrated (χ2 test for trend, P < 0.001).
ZDV use during pregnancy was reported for 57 women: in three women as a maternal therapy (in one case in combination with lamivudine), and in 54 women for prevention of mother-to-child transmission. Median duration of ZDV use was 17 weeks (range, 1–33 weeks). Intrapartum intravenous ZDV was used in 66 cases and postnatal ZDV prophylaxis in 68 children. Altogether, 76 mother-child pairs had any ZDV exposure, and 45 of these had a complete prophylactic regimen defined as antenatal ZDV starting before 34 weeks of pregnancy, intravenous intrapartum application and postnatal prophylaxis for 6 weeks. Of the 89 children born after publication of the recommendations for general prophylaxis, 69 (78%) had some form of ZDV exposure, and 44 (49%) received a complete regimen. Transmission rate was 7% (four out of 55) in children with any ZDV exposure and known infection status compared with 17% in the absence of ZDV (OR, 0.37; 95% CI, 0.13–1.07; not significant). Two (7%) out of 30 children were infected after a complete ZDV regimen, one of them by a ZDV-resistant strain of HIV . Because transmission rates were identical after complete or incomplete prophylaxis, the variable ‘any ZDV exposure’ was chosen for multivariate analysis.
Other factors found to be associated with transmission rate (Table 2) were maternal HIV-related symptoms and a history of one or more previous miscarriages or induced abortions. The number of these events in a particular women had no influence on the size of this effect. Maternal age, parity, drug use, sex of infant, gestational age and breastfeeding [transmission rate for children ever breastfed, 16% (five out of 31), versus 16% (59 out of 378) in those never breastfed] had no association with transmission rate. Many of the risk factors analysed showed considerable association amongst themselves. Mothers with elective Cesarean section had more ZDV use (20 versus 5%; P < 0.001), more HIV-related symptoms (13 versus 7%; P = 0.004), were more often primiparous (78 versus 61%; P = 0.002) and less likely to breastfeed (1 versus 10%; P = 0.001) than their counterparts, but did not differ by age, history of drug use, and previous miscarriage or induced abortion. Mother-child pairs who, despite delivery after publication of recommendations for prophylaxis, did not receive ZDV had less elective Cesarean section (26 versus 57%; P < 0.05), fewer HIV-related symptoms (6 versus 16%), and were breastfed more often (21 versus 0%; P < 0.001) than their counterparts, but did not differ by age, parity, history of drug use, and previous miscarriage or induced abortion. Adjustment for all these factors by logistic regression (Table 2) strengthened the association with maternal HIV-related symptoms, confirmed an independent, statistically highly significant effect of elective Cesarean section, and did not alter the effect of ZDV, which did not reach statistical significance.
The combined effect of ZDV use and elective Cesarean section resulted in the absence of mother-to-child transmission in 31 consecutive children (Table 3). There appeared to be an independent and additive effect on transmission rate. Risk differences for both ZDV use and elective Cesarean section were not significantly different between strata, so no effect modification could be demonstrated. The pooled risk difference for ZDV use was 7.6% (95% CI, 0–16) and for elective Cesarean section 13.0% (95% CI, 5–22).
The most striking finding of the present study was that the combination of elective Cesarean section with ZDV prophylaxis seemed to eliminate vertical transmission altogether, with 31 consecutively born infants shown to be not infected. Although numbers after combined prophylaxis are still small (upper limit of the 95% CI is 11%, a rate observed after ZDV prophylaxis alone) and the potential biases of an observational study cannot be ruled out, an additive protective effect of the two interventions is biologically plausible.
Soon after mother-to-child transmission of HIV was recognized, some obstetricians suggested a possible preventive effect of elective Cesarean section. Despite a lack of scientific data, as early as in 1985 some obstetric centres in Switzerland started to routinely offer delivery by elective Cesarean section by 38 weeks of pregnancy to all their HIV-infected pregnant women, whereas other centres chose to leave their obstetric practice unaffected by the presence of HIV infection. During the following decade it became increasingly clear that a substantial proportion of vertical infections do indeed take place at the time around delivery, and that the potential pathways of intrapartum virus transmission, namely transplacental maternofetal haemorrhage during labour, swallowing of cervicovaginal secretions or blood during birth, or direct contact of infectious fluid with mucous membranes or traumatic skin lesions in the birth canal, could theoretically be circumvented by elective Cesarean section . In the present analysis of the Swiss cohort, a very clear reduction in transmission rate after elective Cesarean section was found, contrasting with an increase associated with the use of invasive procedures. Other maternal factors found to influence transmission rate were HIV-related symptoms, as already reported in many other studies , and a history of previous miscarriage or induced abortion. This latter finding was unexpected and cannot readily be explained. Unfortunately, data on maternal CD4+ cell counts and viral load were too incomplete to allow an adjustment for these known predictors.
A reduction of transmission rate after Cesarean section has been observed in the much larger European Collaborative Study , but not in the equally large French Multicentre Study . Most other studies yielded inconclusive results, except for a small cohort from South Africa showing a reduction of marginal statistical significance . The results of a meta-analysis were mainly influenced by the two very large European studies and indicated a small protective effect of Cesarean section . The power of this analysis was limited by the impossibility of distinguishing between elective and emergency Cesarean sections and by methodological differences among studies, including indications for Cesarean sections as well as the recording of practical details such as the timing of the intervention in relation to labour and rupture of membranes. Our study is unique in that a large proportion of women had Cesarean section for the only reason of being HIV-positive and the intervention was prospectively documented as having occurred before labour and with intact membranes.
The recommendation for ZDV prophylaxis of mother-to-child transmission of HIV had a very good uptake among physicians in Switzerland. However, a substantial minority of pregnant women were reluctant to take the drug before the immediate intrapartum period or refused it altogether. Most of them were in relatively good health and seemed to have a critical attitude towards modern medicine, as reflected by their surprisingly high rate (21%) of breastfeeding against medical advice. Complete or partial ZDV prophylaxis resulted in a reduction of transmission rate from 17 to 7%.
Transmission despite ZDV prophylaxis can occur in the presence of resistant HIV [10,14] or be associated with a lack of antenatal decrease in maternal viral load . This latter finding of a small cohort study was, however, not completely confirmed by the viral load analysis of the ACTG 076 study , suggesting a post-exposure preventive effect of ZDV in the fetus or new-born as an additional mechanism. In this case, the size of the viral inoculum, given by the concentration of infective virus in the transmitting fluid times the volume inoculated, may be crucial in peripartum transmission. If elective Cesarean section effectively reduces exposure to infectious fluids, then an additive protective effect to ZDV prophylaxis appears likely.
A definite answer to the question of whether and to what extent elective Cesarean section protects from perinatal HIV transmission and how such a protective effect would be affected by the presence or absence of ZDV prophylaxis can only be given by a randomized controlled trial. In addition, only an intervention trial can evaluate the potential risks of routine Cesarean section to maternal health. An international randomized mode of delivery trial enrolling women accepting or refusing antiretroviral prophylaxis is now ongoing in Europe. Depending on the results of this trial, elective Cesarean section might prove to be a valuable alternative or adjunct in the prevention of mother-to-child transmission of HIV, especially for women fearing potential toxic effects of ZDV or combinations of antiretroviral agents towards the fetus and newborn.
The authors thank J. Micallef for editing assistance.
1. The European Collaborative Study: Vertical transmission of HIV-1: maternal immune status and obstetric factors
2. Kuhn L, Bobat R, Coutsoudis A, et al.
: Cesarean deliveries and maternal–infant HIV transmission: results from a prospective study in South Africa
. J Acquir Immune Defic Syndr Hum Retrovirol
3. Mandelbrot L, Mayaux MJ, Bongain A, et al.
: Obstetric factors and mother-to-child transmission of human immunodeficiency virus type 1: the French perinatal cohorts
. Am J Obstet Gynecol
4. Kind C, Paediatric AIDS Group of Switzerland: Mother-to-child transmission of human immunodeficiency virus type 1: influence of parity and mode of delivery
. Eur J Pediatr
5. Connor EM, Sperling RS, Gelber R, et al.
: Reduction of maternal-infant transmission of human immunodeficiency virus type 1 with zidovudine treatment
. N Engl J Med
6. Kind C, Brändle B, Wyler C-A, et al.
: Epidemiology of vertically transmitted HIV-1 infection in Switzerland: results of a nation-wide prospective study
. Eur J Pediatr
7. Schüpbach J, Böni J, Tomasik Z, et al.
: Sensitive detection and early prognostic significance of p24 antigen in heat-denatured plasma of human immunodeficiency virus type 1-infected infants
. J Infect Dis
8. Schüpbach J, Flepp M, Pontelli D, Tomasik Z, Lüthy R, Böni J: Heat-mediated immune complex dissociation and enzyme-linked immunosorbent assay signal amplification render p24 antigen detection in plasma as sensitive as HIV-1 RNA detection by polymerase chain reaction
9. Centers for Disease Control: Classification system for human T-lymphotropic virus type III/lymphadenopathy-associated virus infection
10. Siegrist C-A, Yerly S, Kaiser L, Wyler C-A, Perrin L: Mother to child transmission of zidovudine resistant HIV-1
11. Scarlatti G: Paediatric HIV infection
12. The European Collaborative Study: Caesarean section and risk of vertical transmission of HIV-1 infection
13. Dunn D, Newell ML, Mayaux MJ, et al.
: Mode of delivery and vertical transmission of HIV-1: a review of prospective studies
. J Acquir Immune Defic Syndr
14. Frenkel LM, Wagner LE, Demeter LM, et al.
: Effects of zidovudine use during pregnancy on resistance and vertical transmission of human immunodeficiency virus type 1
. Clin Infect Dis
15. Dickover RE, Garratty EM, Herman SA, et al.
: Identification of levels of maternal HIV-1 RNA associated with risk of perinatal transmission. Effect of maternal zidovudine treatment on viral load
16. Sperling RS, Shapiro DE, Coombs RW, et al.
: Maternal viral load, zidovudine treatment, and the risk of transmission of human immunodeficiency virus type 1 from mother to infant
. N Engl J Med
Principal collaborators of the Swiss Neonatal HIV Study Group
J.J. Cheseaux, B. Vaudaux, C-L. Fawer (University Children's Hospital, Lausanne); P. Birrer (University Children's Hospital, Bern); H.E. Gnehm (Children's Hospital, Aarau); G. Schubiger (Children's Hospital, Luzern); J. Klingler (Children's Hospital, Biel); U. Hunziker (Division of Paediatrics, Kantonsspital, Winterthur); H.F. Kuchler (Division of Paediatrics, Hôpital Cantonal, Sion); M.P. Gianinazzi (Division of Paediatrics, Ospedale Civico, Lugano); U. Bühlmann (Division of Paediatrics, Stadtspital Triemli, Zürich); S. Suter (University Children's Hospital, Genève); U.B. Schaad (University Children's Hospital, Basel); G. Spoletini (University Women's Hospital, Lausanne, now Regional Hospital La Chaux-de-Fonds); K. Rohling (University Women's Hospital, Zürich); J. Böni, J. Jendis, Z. Tomasik (National Centre for Retroviruses, University of Zürich); L. Perrin (Laboratory for Viral Serology, University Hospital Genève); P. Erb (Institute for Medical Microbiology, University of Basel); M. Künzel, I. Felber, M. Holzherr, G. Schneidewind (Study administration, Kantonsspital, St Gallen).