Childbearing desires and behaviour: a prospective assessment in Nairobi slums

For decades, the concept of unmet need for contraception has provided the single strongest argument for investment in family planning in low- and middle-income countries. Though its precise measurement is complex, the central idea is straightforward: unmet need refers to women who state that they want no more children or wish to delay the next child for at least two years but are using no method of contraception. It has long been acknowledged that these simple questions on future childbearing preferences cannot capture nuance, ambivalence, emotional orientation and uncertainty of sentiments [1, 2].

It has been argued that the concepts of fertility preferences and planning may be particularly problematic in sub-Saharan Africa. Preferences may be ambivalent. For instance, among women in Burkina Faso, Ghana and Kenya who did not want a child at all, appreciable minorities of women reported that they would experience no problem if they became pregnant in a few weeks and about 10% of women wanting to delay the next childbearing said that they would be happy with an early pregnancy [3]. Preferences may also be unstable. A high degree of fluctuation in reproductive desires was reported among young women in longitudinal studies in Kenya, Ghana and Malawi [4‐7]. Johnson-Hanks and Sennott and Yeatman argued that fertility preferences of young women in Cameroon and southern Malawi may be tentative and unstable as an adaptation to unpredictable environments [5, 8].

Unpredictability of life is likely to especially be high in slums. Slum dwellers often face pervasive poverty, congestion, inadequate services, personal violence and breakdown of relationships. In our study setting, Korogocho and Viwandani, two Nairobi slums, the majority of adults is engaged in casual work; in 2009 only 14% were engaged in salaried employment and one-quarter were economically inactive. In contrast to rural areas where many families rely on farming and grow much of their own food, the urban slum population relies on the market economy and needs to buy food in addition to paying rent and other costs. For example, a study by Macharia et al. found that only 20% of households are food secure in Korogocho and Viwandani [9]. Additionally, not only household incomes but also conjugal relationship are unstable in these settings and divorce and separations are more common in slums than elsewhere in Kenya [10].

In African cities, women experience a constant dilemma between concerns for the costs of an additional child and the desire for a larger family [11‐13]. Having an additional child in these highly uncertain circumstances may put extra pressure not only on their household economic circumstances but also the spousal relationship [11]. Women who want to postpone childbearing for a long time or do not want any more children may also have conflicted preferences. They may revise their preferences if their partner or partner’s family want more children or if financial circumstances improve. These considerations imply that the link between desired future childbearing and subsequent behaviour may be weak among slum dwellers.

In spite of the aforementioned concerns, examination of the utility of fertility preference measures and their link to reproductive behaviour is still rare in Africa. While an increasing but small number of prospective studies assessed to what extent fertility preferences predict future reproductive outcomes among women in sub-Saharan Africa [14‐17], none has been conducted in urban informal settlement settings.

Sub-Saharan Africa will account for one-half of world population increase between 2015 and 2050 [18]. The population growth will occur almost entirely in urban areas. It is projected that over the next three decades (2020–2050), Africa will be the fastest urbanising region in the world [19]. While many new arrivals in the outlying informal neighbourhoods of cities are migrants from rural areas, a large part of this growth will be due to the larger number of births than deaths among urban residents. Inequality in cities is higher than in rural areas, and most cities are falling behind in housing provision and improving vital infrastructure compared with the rapid pace of population growth. It is reported that a majority of the urban population (62%) in sub-Saharan Africa live in slums or in slum-like conditions [20]. While the urban fertility rate has fallen in recent years, urban unmet need for modern contraception and unintended pregnancy are still high. The rapid growth in the number of women of reproductive age living in cities with limited access to vital services such as family planning creates enormous need for targeted sexual and reproductive health services.

Advancing an understanding of childbearing desires is an important precursor to achievement of the policy goal of reducing unintended pregnancies. The aim of this paper, therefore, is to assess the extent to which baseline childbearing preferences predict subsequent reproductive outcomes among women aged 15–49 years in two slums in Nairobi, Kenya, using longitudinal data from the Nairobi Urban Health Demographic Surveillance System (NUHDSS). To our knowledge, this is the first prospective analysis to assess predictive power of fertility preferences in urban slums in sub-Saharan Africa.

A total of 4140 NUHDSS female residents aged 15–49 at baseline and their 4577 children born between September 2006 and October 2010 were identified as eligible for enrolment. Among 4140 women, 3719 women had one index child, contributing a single mother-child dyad each, 405 women formed two mother-child dyads and 16 women contributed three dyads. On average, women contributed 19 person-months of follow-up. As shown in Figs. 1, 874 subsequent pregnancies were identified during the follow-up period. Fifty-eight percent (509 pregnancies) were self-reported pregnancies by the women at follow-up interviews and 42% were identified from the NUHDSS’s vital registration data. Among the self-reported pregnancies, 437 were later reported as live births, 52 were loss to follow-up due to out-migration and 20 pregnancies were not reported as live births in the NUHDSS, indicating that they probably ended in miscarriage, abortion or stillbirth.

A summary of selected baseline background characteristics of the women is presented in Table 1. Over one-half of the women lived in the fourth largest slum in Nairobi, Korogocho, and 47% were members of the youthful and mobile population in Viwandani. About 54% of the women were between the ages of 15 and 24 years, and most women reported being currently married or living with partner (85%). Almost one-half (48%) had completed primary school and 23% had attended secondary school or higher education. Concerning ethnicity, Kikuyu was the dominant group, followed by Kamba, Luo and Luhya. One-third of the women had one living child, one-quarter had two and 35% had more than two children. In two-thirds of cases, the index child was aged less than six months at the time of the baseline interview.

Few studies have used a prospective design to improve our understanding of fertility desires among women in sub-Saharan Africa and none has been conducted in a slum population in this region. In this study, we were specifically interested in the extent to which baseline childbearing desires predict subsequent reproductive outcomes. This study makes a significant contribution to assessing the validity of responses to simple questions on future childbearing wishes.

The single most important finding is the strong link between baseline fertility desires and subsequent reproduction. A large difference in pregnancy risk was observed between those who wanted no more children and those who wanted another child. This is in line with previous longitudinal studies conducted in rural Africa: three prospective studies conducted in Nigeria, Ghana and northern Malawi found that preferences, particularly desire to limit childbearing, among married women have strong power to predict childbearing over two to three years period [14, 16, 17]. Women who wanted no more children had one-half or lower odds of getting pregnant in two to three years compared with women wanting more children. Similarly, a study in selected cities in Kenya, Nigeria and Senegal found very large differences in pregnancy risk between women wishing to limit family size and those wanting more children [15]. The single partial exception is a study in rural Mozambique where 56% of women stating at baseline that they wanted no more children gave birth in the three year follow-up compared with 69% of those wanting more [28].

Prior studies in Taiwan, US, South Korea and Sri Lanka between the 1970s and 1990s also showed strong predictive validity of the prospective measure on reproductive outcomes over three to five years [29‐33]. More recent similar results have been reported in Egypt and Pakistan [34, 35]. An important conclusion can be reached: the link between woman’s stated desire to stop childbearing and subsequent childbearing is just as strong in sub-Saharan Africa as in other regions of the world.

As well as consolidating the African evidence on the predictive strength of the preference for stopping childbearing, this study advances understanding of the relative effects of desired limitation and spacing or postponement. Nearly all other studies have employed a simple dichotomy of wanting no more children versus wanting more or divided the latter group into those wanting the next child within two years or after two years. Our more detailed classification revealed a pronounced gradient in pregnancy risk according to preferred spacing length. Indeed, the difference in pregnancy risk over three years between women wanting no more children and those wishing to wait for five or more years was marginal: 28% versus 32%.

The importance of this result is underscored by the large proportion of women in the study expressing a desire to wait many years before the next child. This desire was also mentioned by the majority of respondents in a qualitative study in the same slums [11]. In this study, women explained that they would not want to have two children in secondary school at the same time. It may also be true that women want older children to have a degree of independence before another child is born because of the greater challenges in arranging child-care and bringing up children in slums than it is in the rural areas. This result also supports other evidence that the fertility decline in Africa is being driven to an appreciable extent by prolonged spacing or postponement, in contrast to Asia and Latin America where the transition to lower fertility initially took the form of limiting family sizes [36].

Yet implementation of baseline preferences is far from perfect. Among women stating a desire to have no more children, 16% became pregnant within two years, rising to 28% at three years. The corresponding estimates for women wanting to wait for five or more years were 21 and 32%. This finding is consistent with studies in Egypt, Malawi, and Pakistan where between 29 and 34% of women who wanted no more children became pregnant within three years [34, 35, 37]. One possible reason, already mentioned, is that women did genuinely change their minds. However, it is also likely that many women who stated this desire to stop childbearing faced barriers to contraceptive adoption and sustained use. Indeed, another analysis using the same source of data showed a high discontinuation of contraception among postpartum women [38]. It is estimated that unmet need for contraception is about 24% among women in Nairobi’s slums [10]. It is important to note that failure to implement preferences seemed to work in two opposite directions. Many women who at baseline indicated that they wanted a child in less than two years had not gotten pregnant even after three years. Failure to conceive as quickly as desired may be partially responsible; at recruitment 35% of the women were still amenorrheic and for some of these postpartum insusceptibility may have been prolonged. But this result also suggests that prospective preferences change overtime. Our study collected relevant data only at recruitment and cannot assess stability of fertility desire. Additionally, life in the Nairobi slums is unpredictable, incomes fluctuate and relationships are often fragile. It is thus to be expected that changes in circumstances give rise to change in reproductive desires.

The lack of intensity in the desire to protect against future unwanted pregnancy among our study population in the slums is worth highlighting. For example, among postpartum women who wanted to delay child bearing for two to four years about 17% said having a child in the next few weeks would not be a problem, while another 10% said it would only be a small problem. This indicates a tolerance concerning the timing of future childbearing which is consistent with results from other countries including studies among Kenyan women [15]. In a study of three African countries by Speizer [3], Kenyan women reported the highest levels of ambivalence toward childbearing, though the level of contraceptive use was the highest. As suggested by several studies, women with ambivalence are less likely to use contraception or are more likely to discontinue, especially if they face any method-related dissatisfaction [3].

There are very large differences in implementation of preferences by education, locality and ethnicity. We attempted to account for these by examining strength of fertility desires (i.e. big problem or not) but no variation was found. It is not so surprising that educated women had much lower hazard of getting pregnant than the less educated because educated women probably have better knowledge and more agency than less educated women. Other studies among the same slum population showed that women with more education were significantly more likely than others to have adopted a modern contraceptive method, even during the 12 months postpartum [7, 38]. It is possible that, for women with more education, the potential loss of income which will be incurred by time away from the labour force to raise an additional child may be the impetus pushing them to realize their fertility desires [39]. This is supported by the fact that financial reasons were cited as a main reason for limiting by women who wanted no more children. It is, however, worth noting that our results are inconsistent with findings from northern Malawi, where, though contraceptive use was higher among the more educated [40], education did not modify the relationship between reproductive desire and future childbearing [14].

Kikuyu women who wanted to limit family size or postpone the next child for five or more years were significantly more likely to avoid pregnancy than corresponding women from other ethnic backgrounds. This result is consistent with large differentials in contraceptive prevalence between the Kikuyu dominated Central province and the Luo/Luhya dominated Western region but it nevertheless intriguing that similar differences persist when locality and presumably service access is controlled. This suggests cultural factors not yet understood are playing a significant role. Finally, the Viwandani-Korogocho difference is equally intriguing. Even when education, age and ethnicity are controlled for, a big difference persists between the two sites, with women in Viwandani more likely to realise their fertility desires than women in Korogocho. This difference may be partly explained by the lower level of under-five mortality in Viwandani than in Korogocho [22]. Further analysis is therefore needed to understand the specific role community factors play in women (and couples) realising their fertility desires.

This study should be interpreted in light of several limitations. First, the study suffered high loss to follow-up and it is possible that mobile women differ from those who remained in the same slum locations over the observation period. Second, the average follow-up time period was 19 person-months. Although we used survival analysis, the high attrition and short follow-up period prevented the estimation of long-term impact of preferences on reproductive outcomes. Third, we were unable to assess the influence of partners’ preferences on reproductive decisions and outcomes because the interviews were conducted among women. Other studies show that partners’ approval influences woman’s decision to use contraceptive methods and contributes to achieving their fertility desire [41‐43]. However, there is also evidence that the influence of a woman and her partner was equal, and in some studies the wife’s preference was found to exert a greater influence than that of her partner [44, 45]. For example, the cross-sectional analysis from 18 national surveys reported few differences in family planning use between couples in whom the wife, but not the husband, wanted more children and vice versa [46]. Fourth, we did not attempt to assess the stability of desires nor the role of contraceptive adoption, failure and discontinuation or abortion in mediating the effect of preferences on reproductive outcomes. Fifth, the eligibility criterion of this study was having a child recently. Such women may be particularly likely to report not wanting to have another child or wishing to wait longer before the next child. Women tend to revise their fertility preferences after they have a new pregnancy [4, 5]. Lastly, the HIV status at the interview was not collected in this nested study. In 2008/2009 HIV prevalence was 12% in the adult population in the two slums [47]. HIV status may have impacted women’s fertility desires and intentions and the implementation of their desire [48, 49].

Our study adds to the small body of evidence that the stated wishes of women to limit or postpone future childbearing in sub-Saharan Africa have the same meaning and similar predictive validity as preferences in other regions. Concerns that these wishes need not be taken seriously because they are superficial and highly volatile are misplaced. It follows that estimates of unmet need for contraception in Africa are broadly valid and that the increase in the proportion of women in many East African countries who state a desire to have no more children is a trend of considerable importance. In addition, the findings revealed a pronounced gradient in pregnancy risk according to preferred spacing length, which support other evidence on the important contribution of long-term spacing or postponement to fertility decline in sub-Saharan Africa.