Incidence of pregnancies among African HIV-infected women, Abidjan, 1995–2000

A study was conducted in 1999–2000 in Abidjan within the ANRS 049 Ditrame project, to describe the adherence of African HIV-infected women to counselling on protected sexual intercourse and contraceptive use after the announcement of their HIV test result during pregnancy. The incidence of pregnancy was 16.5 per 100 women-years at risk during the first 24 months postpartum. The factor most influencing the decision to start a new pregnancy was a high level of education, which tripled this risk. In Abidjan, Côte d'Ivoire, the ANRS 049 Ditrame programme consisted first of a double blinded placebo-controlled randomized trial, which evaluated the tolerance and efficacy of a short zidovudine regimen up to February 1998 [1], and then of an open cohort in which HIV-infected women systematically received the zidovudine prophylactic regimen until September 1998. After written consent, each woman was followed during the last month of pregnancy and at least 18 months postpartum with her child. Women received regular counselling on the protection of sexual intercourse and contraceptive use, with free delivery of condoms and contraceptives to reduce the risk of sexual transmission to their partner, and to avoid unexpected pregnancies. The aim of the study was to describe the reproductive behaviour of these women who had recently delivered, received zidovudine prophylaxis and a comprehensive counselling package [2]. Between April 1999 and March 2000, a supplementary questionnaire was administered by social workers to each HIV-infected woman who came for a follow-up appointment; 149 of a total of 427 who had been enrolled in the Ditrame project, without any refusal. The sociodemographic characteristics of this sample were no different from those of the rest of the women included. The questionnaire focused on the occurrence of pregnancies since inclusion in the programme, the resumption of sexual activity after childbirth, the history of contraception and condom use, and the desire for a new child expressed by the woman and her partner. The woman was also asked if she had notified her partner about her seropositive status. We calculated the incidence of pregnancy in relation to the period of observation (number of pregnancies per 100 women-years at risk) in order to take into account the variability of the duration of follow-up (interquartile range 12–24 months). Finally, the influence of sociodemographic factors as well as proximate determinants of fertility (breastfeeding, postpartum abstinence and contraceptive use) were studied in relation to the probability of the occurrence of a new pregnancy using a Cox regression model, with time-dependent covariates: cessation of breastfeeding, cessation of postpartum abstinence, and the death of the child whose gestation had been followed in the programme (Ditrame child). The survival of the Ditrame child and the duration of breastfeeding, which were too closely related covariates, could not be tested in the same model. Out of the 149 women surveyed, 37 had at least one pregnancy during the postpartum follow-up, a pregnancy incidence of 16.5 per 100 women-years at risk for the first 24 months postpartum [95% confidence interval (CI) 11.2–21.9]. Of these 37 pregnancies, 46% (17) were desired by the women and 51% (19) were not (one woman did not answer the question). Out of the unwanted pregnancies, 68% (13/19) were terminated voluntarily by induced abortion. The modelling of the incidence of a new pregnancy using Cox regression shows that, given equal sociodemographic characteristics, the probability that a woman whose Ditrame child was stillborn or died in the first year started a new pregnancy was greater than for a woman whose baby was still alive [hazard ratio (HR) 6.6, CI 2.5–17.0, model A, Table 1). We did not take into account whether the child was HIV infected or not, because the mother did not know the serostatus of her child at the time of the survey. Cessation of breastfeeding of the Ditrame baby increased the probability of the occurrence of a new pregnancy (HR 9.0, CI 1.9–42.4, model B, Table 1). The only sociodemographic variable that influenced the occurrence of a new pregnancy was the level of education: having a secondary level of education tripled the risk of the occurrence of a new pregnancy (model A: HR 3.2, CI 1.0–9.5; model B: HR 2.8, CI 0.9–8.7, Table 1). The incidence of pregnancy in the first 24 months postpartum was thus 14.9 (5.2–24.7) per 100 woman-years for women with no schooling, 15.2 (7.1–23.3) for women with primary education, and 26.4 (12.1–40.7) for those with secondary education.Table 1: Determinants of the risk of the initiation of a new pregnancy during postpartum follow-upa. Time-dependent Cox regression. Project ANRS 049 Ditrame, Abidjan, Côte d'Ivoire, 1995–2000. When postpartum abstinence was included in the models (models C and D), the same tendencies were observed, but the effects were reduced and did not reach statistical significance. The length of postpartum abstinence thus became the only predictor of the occurrence of a new pregnancy. Neither the use of contraception nor notification of the partner significantly affected the occurrence of a new pregnancy in any of the models (Table 1). Counselling on protection of sexual intercourse was followed poorly: only one out of five of the women who had notified her partner of their serostatus claimed to use condoms systematically and half of them never did. We observed the same on counselling on contraceptive use: only 39% of women interviewed used a modern method of contraception at the time of the survey. This low adherence to family planning counselling was independent of the woman's desire for a child and was not therefore a voluntary strategy. It was not influenced by the partner either. The relatively high rate of postpartum pregnancy we documented is close to what was observed in Rwanda [3]. In the SEROCO French cohort, the overall incidence observed for a 3 year postpartum follow-up was 7.9 pregnancies per 100 woman-years [4]. The low use of contraceptives is explained by the general context of Côte d'Ivoire, in which a population growth control policy [5] was started only at the end of the 1990s, with limited family planning services available. Furthermore, contraceptives are still perceived as a health hazard, and abortion, although illegal, seems to be used as a means of birth control [6]. On the other hand, an increase in the duration of postpartum abstinence was observed in our study (12 months) compared with a survey conducted in 1997 (8.4 months) [7]. This increase in the length of postpartum abstinence, which was greatest among women who had not informed their partner of their seropositivity, can be considered as one of the strategies adopted by these women aware of their seropositivity, in order to avoid sexual transmission, anticipating the difficulty in proposing condoms to their partner. However, the level of education appears to be an impediment to this strategy: in order to explain the extended length of abstinence to the husband without revealing their seropositivity, women apparently hide behind the tradition of protracted postpartum abstinence, a tradition that is more difficult to practice among educated women than among uneducated women. In conclusion, in this study, HIV-infected women hardly followed the counselling given on protection of sexual intercourse and contraceptive use. Some women developed their own strategy, such as protracted postpartum abstinence. Paradoxically, the most educated women were the most vulnerable to the risk of early pregnancy, because they could not use this strategy, but then did not adopt reliable contraception early enough. The context of Abidjan, where contraceptive use is still low [6] clearly accounts for this pattern. This must be taken into account in future public health projects targeted to HIV-infected women in this city or comparable settings. Acknowledgements The authors extend special thanks to Katia Castetbon and Valeriane Leroy for reviewing the manuscript. Annabel Desgrées-du-Loûa,b Philippe Msellatia,c Ida Vihoc Angèle Yaoc Delphine Yapic Pierrette Kassic Christiane Welffens-Ekrac,d Laurent Mandelbrote François Dabisc,f for the ANRS 049 Ditrame Study Group