In this paper we examined postpartum behavior in a population where fertility remains high, use of modern contraception remains low and where lactational amenorrhea, together with postnatal sexual abstinence, acts as the most important constraint on childbearing within marriage. Our main contribution to the extensive postpartum literature is its focus on the degree to which women in Cameroon are exposed to the risk of an unintended pregnancy in the 18 months following childbirth, considering the resumption of sex and menses as well as contraceptive use. The decision to restrict attention to the 18 months following childbirth was based partly on consideration that protection against pregnancy over this period of time safeguards against the shortest birth intervals that carry the most risk to child survival. At 18 months most women in Cameroon have weaned their child and have resumed menses. A longer period, for instance 24 months that has been used in some analyses of postpartum contraception, would have been inadvisable because of the selection bias against short intervals bias arising from the appreciable minority who had two births in the past two years. Even the cut-off at 18 months introduces a small bias. In the 2018 CDHS, 28 women out of 3,007 had two births in the past 18 months.
Our definition of risk is contentious, because it departs from the orthodoxy that, following childbirth, sexually active women should use a modern method of contraception if they wish to avoid pregnancy, unless they are observing the conditions of LAM. Specifically, we classified all amenorrheic women with an infant aged less than six months as protected against pregnancy risk, regardless of the extent of supplementary feeding. Further, amenorrheic women with a child aged six or more months were classified as partially at risk. The evidence to support these decisions is sparse and dated but nevertheless positive. It is also consistent with the behavior of Cameroonian women, many of whom await the return of menses before starting a method of contraception. At 12-17 months postpartum, for instance, only 14.6% of sexually active amenorrheic women reported use of a modern method compared with 27% of menstruating women. And as shown in the top panel of Table 3, the degree of protection from delayed return of sex and menses was the single biggest predictor of contraceptive use, even after adjustment for other co-variates.
In the definition of risk status, we made an explicit choice to focus on unintended pregnancy rather than all pregnancies. The implication was that women wanting another child soon or within the next 12 months were classified as not at risk. Overall, only 12% wanted another child soon though this proportion rose to 24% among those with a child aged 14 or more months (see Figure 1). As nearly 90% of our sample wished to delay pregnancy, it is most likely that an analysis based on all pregnancies would have yielded similar results to those reported in this paper. We were also faced with a choice of how to handle the small number of 92 women who reported that they were pregnant. As their exclusion would have introduced a bias, we decided to classify them on the basis of intendedness. Women who declared that the pregnancy was wanted at that time (i.e., intended) were classified, along with those wanting a child soon, as not at risk while others were classified as fully at risk.
Following this classification, we showed how exposure to risk evolved over the 18 postpartum months. In the first six months after birth, over 90% of women were classified as fully protected, predominantly because of delayed resumption of sex and menses. Reported use of LAM was trivially small at less than 1%. As time since birth lengthened, the protective effect of sexual abstinence and amenorrhea waned and was only partially compensated by a gradual increase in modern method use. The proportion fully at risk (i.e., sex and menses resumed, and no contraceptive use despite a desire to avoid pregnancy) rose from 8% at 0-5 months to 24% at 6-11 months postpartum and further to 30% among those with a child aged 12-17 months, as shown in Figure 2b. These estimates are much lower than estimates of unmet need that do not consider the protective effect of abstinence and amenorrhea, except for self-reported use of LAM [3, 9]. At 6-11 months, nearly one-third of women had resumed sex but remained amenorrheic and were not using a modern method and were thus defined as partially at risk.
There are several implications of this evolution for provision of postpartum contraception counselling and services. First, family planning programs should seek to inform women that reliance on lactational amenorrhea carries an unpredictable yet modest risk of conception, but a risk that rises with the duration since birth. Second, provision of short acting hormonal methods, such as pills and injectables, early after childbirth is ill-advised. High rates of discontinuation of these methods imply that many women will have discontinued use by the time the protective contribution of abstinence and amenorrhea has ceased. As shown in Figure 3, the method-mix in the first six months postpartum is dominated by short-acting methods with a tiny contribution from long-acting ones. Third, counselling about, and provision of, long-acting methods immediately after delivery or in the early postpartum months should be a key strategy. While less than half of married women were aware of IUDs in 2018 and use remains very low in Cameroon, awareness of implants among married women increased from 51–72% and use rose from 0.7–2.6% between 2011 and 2018. Bearing in mind that two-thirds of births take place in a health facility and that child immunization rates are high, it is probable that this trend could be accelerated by improved availability and counselling. One possible blueprint for action comes from a cluster-randomized trial in Burkina Faso, a country that has similarities with Cameroon. A set of low technology interventions that raised the profile of family planning across the spectrum of MCH services doubled the level of modern method use at 12 months postpartum with an equal share of long- and short-acting methods [22].
The results in Table 2 are instructive and justify our emphasis on a broad definition of risk. As expected, sharp gradients in current use of modern contraception by education and residence were observed, from 5% for women with no schooling to 28% for those with higher secondary or tertiary education and from 12% in the rural population to 27% in the two major cities. Based on these results, it is tempting to conclude that the need for postpartum contraception is most pressing among the rural, less-educated sectors of the population. However, when sexual abstinence and amenorrhea were considered, a very different picture emerged. The percent classified as fully at risk of an unintended pregnancy varied little by education or residence. Indeed, women with the highest educational attainment and those living in Douala or Yaoundé were most likely to be classified as fully at risk. The reason is clear. The protective effect of amenorrhea lasts much longer among rural, less educated women than among their urban, better educated counterparts. For instance, CDHS data show that the median duration of amenorrhea is 4.3 months in Yaoundé and Douala compared with 10.8 months in rural areas. Thus, more privileged couples in Cameroon are substituting contraception for lactational amenorrhea as their means of birth spacing. But greater resort to modern methods of contraception does not lead to longer median inter-birth intervals lengths, which are close to 30 months in all population strata. Furthermore, women with secondary schooling and those living in the two major cities were more likely than their counterparts to report the most recent birth as mistimed. Specifically, 24% of birth were reported as mistimed in Douala/ Yaoundé compared with 16% in rural areas. Only 7% of women with no education declared the birth as mistimed. This rose to 19%, 27.5% and 22% for women with primary, lower secondary and higher education, respectively (results not shown but available on request). The conclusion is counterintuitive but clear. The need for enhanced postpartum family planning is just as pressing, if not more so, among privileged strata as among the less privileged.
The correlates of modern method use were further explored by logistic regression in Table 3. One objective was to assess the associations between contraceptive practice and use of MCH services and receipt of family planning counselling at a health facility visit. As shown in Table 1, uptake of antenatal, maternity, and immunization services is high in Cameroon. Because of strong associations between use of these three services, only the number of child vaccinations was included in the regressions. After adjustment for risk status, education, residence, and other covariates, a statistically significant but modest association between number of vaccinations and modern method use was found, with odds of 1.64 for women whose child had received eight or more vaccinations, compared with those with less than three. This result is consistent with most studies elsewhere that found a positive association between uptake of MCH services and contraceptive use [23, 24, 10]. Causal interpretation from observational studies is problematic but the evidence from interventions that offered contraceptive advice and supplies at immunization clinics is encouraging, though meagre [25, 26].
Uptake of postpartum contraception may be enhanced by MCH service use indirectly insofar as women gain confidence and trust in modern medical services and staff and directly to the extent that family planning advice is actively offered when women attend facilities for health reasons. Despite frequent use of MCH services by the majority of mothers, only one-third reported discussion of contraception during a health facility visit in the past year. This result implies weak integration of family planning into routine health services and missed opportunities for encouraging contraceptive use. This defect is by no means confined to Cameroon. Similar evidence of missed opportunities has been documented in Senegal, Uganda, Malawi, Nigeria and Ethiopia [27–29]. In Cameroon, discussion was significantly related to contraceptive use but, after adjustment, the association was modest, perhaps a reflection that condom, which requires no contribution from a health worker, was the most commonly used method.
We identified a further type of missed opportunity. One-third of mothers reported a visit from a health worker in the past year, which suggests that outreach services in Cameroon have appreciable coverage. However, at only one-fourth of these visits was family planning discussed, with the consequence that only 8.3% of all women had discussed this topic at a home visit. Though contraceptive use was significantly higher among those with such a visit, it is clear that outreach services currently contribute little to family planning uptake.
Most of the other factors included in Table 3 had statistically significant associations with modern method use, after adjustment for risk status and other covariates. Unsurprisingly, the single most powerful correlate was women’s education. It is clear that, even after considering amenorrhea, better educated women have a higher propensity to adopt contraception than the less educated, though, as noted above, this has little effect on birth interval length or on the risk of mistimed pregnancies. The association between urban residence and contraception was less strong than that for education and it is of interest that the large unadjusted difference between rural and urban residents was greatly reduced after adjustment, probably because of the concentration of highly educated couples in the two major cities and other urban areas.
Regional differences are also of note. Unadjusted differences were very large but odds of use attenuated by about 50% after adjustment. Modern method use was lowest in the Northern zone (Adamawa, North, Far North), which is the poorest and least developed part of the country, with an ethnic and religious composition similar to Northern Nigeria where use is also very low. The large unadjusted difference between the Northern zone and the North-West/South-West zone disappeared after adjustment, suggesting that higher contraceptive use in the latter zone reflects other factors such as education and urbanisation. Contraceptive use in the North- and South-West may also have been depressed by civil unrest that has disrupted services. Conversely, the difference between the Northern zone and the Centre/East/South remained large, after adjustment.
There is much international concern that young women face particular problems in accessing contraceptive services and in persuading male partners to adopt contraception, with the consequence that their levels of contraceptive use are lower and unmet need higher than for older women. Such concern is unjustified in Cameroon, where young women were more likely to use a modern method than older women. Adjustment for education and other covariates made little difference to the age gradient in use. The reasons are unclear but two possible explanations can be identified. First, attitudes towards contraception may be more favorable among the younger than older couples. Second, fecundability among older couples will be lower than in the young and hence adequate inter-birth spacing can more easily be achieved without resort to contraception.
Slightly over one-tenth of our sample were single (i.e., never married), a reflection of an upward trend in premarital childbearing as age at marriage increased. Before adjustment, there was no difference in contraceptive use by marital status but, after adjustment, single women were significantly less likely to report current use than married women. Interpretation is difficult. Most single mothers are young and the adjustment for age is thus problematic. It is also likely that infrequent sex among women who are unmarried and not cohabiting may contribute to low use.