Access the full text.
Sign up today, get DeepDyve free for 14 days.
Loading next page...
/lp/sage/influence-of-maternal-education-on-second-childbirth-interval-among-60Wi9XYp9F
- Publisher
- SAGE
- Copyright
- Copyright © 2022 by SAGE Publications Inc, unless otherwise noted. Manuscript content on this site is licensed under Creative Commons Licenses.
- ISSN
- 2158-2440
- eISSN
- 2158-2440
- DOI
- 10.1177/21582440221079920
- Publisher site
- See Article on Publisher Site
Abstract
Inter-birth interval lengthening is a key component of fertility decline. Although South Africa fertility rate remains the lowest in sub-Saharan Africa, information on the effect of education on the interval between first- and second-birth across residential contexts is rarely documented. The study investigated the relationship between maternal education and second- birth interval (SbI) by residence among South African women. The study analyzed the 2016 South Africa Demographic and Health Survey data on 6,039 women aged 15 to 49 years who had reported at least one childbirth at the time of survey. Survival analysis methods were applied at 5% significance level. The SbI was significantly longer (p < 0.001) among urban (76 months) relative to rural (66 months) women. About a fifth of rural women and about a tenth of urban women had at most a primary education. Women who had a secondary education (aHR = 0.86; 95% CI [0.76, 0.96]) were 14% times more likely to delay second-birth compared to those who had at most a primary education in rural setting. Other determinants of SbI included region in rural; age at-first-birth and household wealth in urban; ethnicity, marital status at-first-birth and employment in both residential settings. The length of SbI remains long in both residential contexts, but longer in urban. Findings demonstrated rural-urban differentials in the relationship between maternal education and second birth interval, suggesting contextual impact. Fertility strategies targeted at strengthening health education for improved maternal and child health should be residential-context specific. Keywords survival analysis, maternal education, second birth interval, extended Cox-regression, South Africa, urban and rural differential. Amongst African countries, South Africa is the sixth most Introduction populous country with 2.6 children per woman and a median Fertility decline in most sub-Saharan African countries is birth interval of 55.3 months (NDoH et al., 2019; Stats SA, remarkable over the last two to three decades. Compared to 2018; United Nations, 2017). Rapid urban growth is a key 1998, the current total fertility rate (TFR) amongst women feature of South Africa population. Urban population stood declined from 6.3 to 4.2 in Malawi, 5.6 to 4.4 in Ethiopia, 4.6 at about 68% which is marginally above global average and to 3.9 in Ghana, 4.4 to 3.8 in Gabon, and 2.9 to 2.4 in South is estimated to be about 70% by 2030 (Stats SA, 2018; Africa (Population Reference Bureau, 2018; Tabutin & Atkinson, 2014). Previous studies have established Schoumaker, 2004). Although South Africa fertility rate that exploring fertility dynamics by residential contexts is remains the lowest in sub-Saharan Africa, her observed fer- key to unraveling demographic changes (Lerch, 2018; tility decline seems rather stalled, particularly in urban areas. Within the last two decades, TFR declined from 3.9 to 3.1 University of Ibadan, Ibadan, Nigeria amongst women in rural but marginally increased from 2.3 to North-West University, Mmabatho, South Africa 2.4 amongst women in urban (NDoH, et al., 2019). Although Corresponding Author: literature suggested different patterns of sub-national fertility Rotimi Felix Afolabi, Department of Epidemiology and Medical Statistics, changes among sub-Saharan Africa countries (Lerch, 2017; Faculty of Public Health, College of Medicine, University of Ibadan, Queen White et al., 2008), urban pattern may have a strong impact Elizabeth Road, UCH Campus, Ibadan, Oyo 23402, Nigeria. on South Africa national fertility. Email: rotimifelix@yahoo.com Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License (https://creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage). 2 SAGE Open Shapiro & Tambashe, 2000; Towriss & Timaeus, 2018), is yet to be attained. The knowledge of the relationship most especially as it relates to urbanization in developing between SbI and maternal education could scale-up interven- countries. Urban population, therefore, remains a platform tions that may further lower fertility and, consequently, through which the fertility changes could be investigated reduce the likelihood of negative maternal and child health (Towriss & Timaeus, 2018). outcomes. Unarguably, there are needs for the examination of fertil- Empirical evidence has indicated multifarious factors ity transitions characterized by lengthening inter-birth inter- associated with inter-birth intervals including SbI. These vals at sub-population levels. Studies have linked the have been identified as proximate (biological and behavioral postponement of all higher-order births, including the second factors) and distal (socio-demographic and -cultural factors) birth as a driver of sub-Saharan fertility decline (Lerch, determinants (Bongaarts, 2015; Erfani & McQuillan, 2009; 2017; Moultrie, et al., 2012). Even though sub-optimal inter- Finlay et al., 2018; Palamuleni, et al., 2007; Ramarao, et al., birth intervals often have adverse consequences for maternal 2006; Yohannes, et al., 2017). According to Bongaarts and child health outcomes (Ball, et al., 2014; Chen, et al., (1978), four proximate factors: marriage, postpartum in- 2014; Mahfouz, et al., 2018; Molitoris, et al., 2019; Stevens, fecundability, contraception, and abortion were identified to et al., 2018; Zhang, et al., 2017), the risk of second childbirth define direct determinants through which all socio-economic (and other higher-order births) can be curtailed compared to and cultural factors operate to influence human fertility. the risk of first birth. In contemporary South Africa, the tran- Although Bongaarts (1978) model continues to be relevant, sition from first to second birth is a critical segment in fertil- fertility behavior has changed substantially in the contempo- ity change. With fertility preference of two children (Stats rary population (Bongaarts, 2015). For instance, exposure to SA, 2015), having second childbirth is an event that most sexual activity and childbearing take place not only within South African women look forward to. The present study but also outside marriages in South Africa. Additionally, the therefore focused on second birth interval (SbI), defined as usage of contraception has diminished the effects of conven- time elapsed between first and second childbirth. tional prolonged breastfeeding and postpartum abstinence on Although with a varying direction of association between inter-birth intervals owing to increasing urbanization, cul- educational attainment and the tendency of having second tural changes, and government policies (Baschieri, 2004; childbirth, literature had long recognized education as a Kim, 2003). Some researchers, meanwhile, have claimed major socio-economic driver of fertility transition (Bartus, et that socio-economic and -cultural factors could directly al., 2013; Bongaarts, 2020; Klesment, et al., 2014). influence birth interval dynamics (Baschieri, 2004; Rindfuss, Education, particularly women’s education, has a consider- et al., 1987). Intuitively, these factors could suggest either to able impact on fertility transition in contemporary demogra- delay time to second birth or otherwise. phy due to many pathways between education and family For instance, women may experience a longer birth interval behavior. Such pathways include better access to and use of after the birth of a son compared to the birth of a daughter health information, higher decision-making autonomy, better (Baschieri, 2004; Mace & Sear, 1997; Rossi & Rouanet, contraception and reproductive system knowledge, greater 2015). In addition, literature has documented a significant potential for earnings, and increasing opportunity cost of influence of wealth status, religion, or survival of the first birth childbearing (Bongaarts, 2020; Impicciatore & Tomatis, on the length of SbI (Bao, et al., 2017; Ramarao et al., 2006; 2020). In lieu of the aforementioned and the growth of edu- Singh et al., 2012). Similarly, studies have reported the asso- cation especially among women, exploring the influence of ciation between birth interval lengthening, including SbI and education on SbI while controlling for other correlates may contraceptive use: the more improved access to contraception, clarify the role of education in fertility transitions in South the longer the birth interval (Towriss & Timaeus, 2018). Africa. Women’s age, work status, partner education, and marital sta- Like any other birth interval, SbI could impact on the tus have also been documented as determinants of SbI maternal and child health (Ball et al., 2014; Chen et al., 2014; (Ahammed, et al., 2019; Fiori, et al., 2014; Singh et al., 2012). Stevens et al., 2018; Zhang et al., 2017). Mahfouz et al. Women education, particularly, is considered as the most (2018) submitted that: on the one hand, inter pregnancy important determinant of fertility change (Bongaarts, 2020). interval less than 24 months is attributed to a high risk of Studies have suggested a strong association between preterm birth and low birth weight; on the other hand, an maternal education and birth interval (Bongaarts, 2020; interval higher than 59 months is linked to high likelihood of Klesment et al., 2014; Van Bavel, 2010); an increased educa- having a stillbirth and pregnancy-induced hypertension. tional attainment often suggests a reduced fertility. Women Likewise, other researchers (Grundy & Kravdal, 2014; who had no formal or a lower education have been cited to Nisha, et al., 2019) have documented that a long birth inter- have higher odds for having second childbirth compared val (>59 months) is attributed to poor maternal and child with those who had a higher education (Ahammed et al., health outcomes. Meanwhile, literature had estimated a 2019; Fagbamigbe et al., 2020; Newman & McCulloch, median birth interval of 72 months for the late 1990s in South 1984). Aside from this negative association, a positive or Africa (Moultrie et al., 2012; Towriss & Timaeus, 2018). non-negative relationship between maternal education and Nearly two decades thereafter, fertility at replacement level the tendency of having second birth has been documented. Afolabi and Palamuleni 3 For instance, women who attained a higher education have selected for the survey. A detailed sampling design and pro- second births earlier than those with a lower education attain- cedures has been reported (NDoH et al., 2019). ment in a study conducted among some European countries (Bartus et al., 2013; Klesment et al., 2014). We hypothesized Study Population and Variables that maternal education would significantly influence the SbI lengthening in South Africa. The knowledge of the South A total of 6,039 (2,666—rural; 3,373—urban) women who African fertility transition may be relevant and useful for had ever given birth were included in the study, having other sub-Saharan Africa countries, especially with relatively excluded women who had a multiple first births. high fertility rates. This study investigated the relationship between maternal Outcome variable. The dependent variable of interest was education and time interval from first to second birth by resi- time interval between first and second childbirth amongst dence among women of reproductive age in South Africa. women in South Africa. Women who did not have second The goal was to identify both similarities and differences of childbirth as at the time of the survey were right-censored the inter-relationship between women’s educational attain- and were coded 0; otherwise, 1 in the analysis. ment and SbI for the two residential contexts. This could improve the knowledge of national fertility dynamics includ- Independent variables. The main independent variables were ing the timing and extent of demographic dividends which is maternal education (at-most primary = 0, secondary = 1; critical for national development policy. Consequently, the higher = 2) and place of residence (urban = 1 or rural = 2). study addressed two questions. (i) For how long would a Also, an interaction between maternal education and place of woman residing in rural or urban delay the second childbirth residence variable was generated (at-most primary/rural = 0, after her first birth? (ii) To what extent is the discrepancy in at-most primary/urban = 1, secondary/rural = 2, secondary/ SbI among women who live in the same residential area due urban = 3, higher/rural = 4, and higher/urban = 5). Other inde- to their level of educational attainment? pendent variables considered for the study were region/prov- ince, ethnicity, wealth index, age at first-birth, marital status at first-birth, employment status, contraceptive use, first- Method birth sex, first-birth survival, and desire for more children. Study Design and Setting The detailed description of these variables is presented in Table 1 and included demographic, socio-economic, cultural, A cross-sectional and nationally representative 2016 South and bio-behavioral factors. Africa demographic and health survey (SADHS) data was used for this study (NDoH et al., 2019). The data contain self-reported information on sexual and reproductive health Data Analysis history of the sampled women in South Africa. South Africa As the main goal of the study is to investigate the influence is made up of nine provinces and each of these provinces was of maternal education on time interval of first to second categorized as urban, traditional and farm areas, to enhance childbirth, survival analysis methods were used for the anal- survey precision. ysis. The “true survival time” for women who have had sec- The multistage cluster sampling technique was used, and ond-birth was the SbI. The “censored time” for the women the master sampling frame containing enumeration areas yet to have second birth was time since the first-birth and (EAs) was adopted from the 2011 census frame of the interview date. The Kaplan-Meier survival method was used Republic of South Africa. At the first stage, a stratification- to describe the women’s time interval to second-birth while based probability proportional-to-size sampling method was the log-rank test was employed to examine the association used to sample EAs (known as clusters) as the primary sam- between SbI and the explanatory variables. Also, the Mann- pling unit. A systematic sampling thereafter was used to Whitney test was applied to examine the rural-urban differ- select residential dwelling units referred to as households at ences in average survival time. Semi parametric-based the second stage. Of 750 clusters sampled for the survey, extended Cox hazard regression was thereafter used to 468 were from urban, 224 traditional, and 58 farm areas. For explore maternal education effect on SbI, having confirmed the purpose of this study, settlement/residence type is cate- violation of proportional hazard assumption. The model is gorized as urban and rural (traditional and farm) areas. A expressed as follows (Akinyemi, et al., 2020): total of 15,000 households were sampled from 750 clusters p p 1 2 {( bx + γ xg t )} ∑∑ ll i jjij [] bx ++ bx ++ [( γγ xg tx ) + g () t ] 11ip pi 11ip 1 pi p l == 1 j 1 11 22 2 2 ht (, xt ())( = ht) = ht () i 0 0 p p 1 2 ht (, xt ()) imlying i → = ln bx + γ xg () t llij ji j ∑ ∑ ht () l=1 j=1 4 SAGE Open Table 1. Definitions of Independent Variables. Characteristics Descriptions Place of residence Type of place of residence (urban, rural) Region Region of residence is categorized into nine provinces: Western Cape, Eastern Cape, Northern Cape, Free state, Kwazulu-Natal, North west, Gauteng, Mpumalanga, and Limpopo Ethnicity Women ethnicity (Black/African, others). White, Colored, Indian/Asian ethnic groups were grouped as “other” due to their respective lower percentage compared to Blacks Wealth status The wealth index variable, derived from the generated weighted factor score using principal component analysis as contained in the women recode file, was categorized into tertiles: low, middle, and high. It is a proxy measure of household socio-economic status due to non-existence of information on household income. The household scores were generated based on ownership of household assets (e.g., television, radio, or bicycle/car) and housing characteristics (e.g., flooring, wall, or roofing materials) Education Women highest level of educational attainment categorized as: at most primary, secondary, and higher Partner education Partner/husband’s highest level of educational attainment categorized as: at most primary, secondary, higher, no partner/husband or don’t know Age at FB Women age at first birth (in years) categorized as: <20, 20–24, and ≥25 years Marital status Women marital status at first birth: Never married/cohabiting, cohabiting before the first birth, and cohabiting after the first birth Desire more children Fertility desire, derived from the questions asking women to indicate whether they desired more children was re-categorized as “desired, desired no-more, or undecided” Employment Women current employment status: working, not working FB sex Sex of the first birth (male or female) FB survival The first childbirth survival variable categorized as: not alive, alive Contraceptive Contraceptive use was derived from the questions asking women to indicate “whether they ever used anything or tried to delay or avoid getting pregnant” and their “current contraceptive use by method type”; this was categorized as “never, former, or current” where b is the jth coefficient of the predictor variable X ; model approach while controlling for other predictors as well j j h (t) is the baseline hazard function and h(t)/h (t)—hazard as accounting for multicollinearity in each of the settings. In 0 0 ratio (HR); γ is the jth overall effect of X (t) = X × g (t) (time model 1, maternal education and variables to define other j j j j dependent predictor(s)) such that its positive value indicates socio-demographic characteristics were included; model 2 is the HR increases with increasing time; otherwise, it the full model with the exclusion of partner’s education due to decreases; p —number of predictors, independent of time; collinearity with marital status at first birth. However, only p —number of predictor(s) which interact with time; and the full model tagged model 3 was considered for the pooled g(t) = t (in which estimated HRs increase or decrease expo- data. The adjusted HRs (aHRs) including their 95% confi- nentially as t increases). A detailed description of Cox pro- dence intervals (CIs) and/or p-values were reported. portional hazard model including the extended Cox hazard Importantly, provisions were made for intracluster correlation model’s time function, g (t) is presented in Supplemental and the data was weighted to adjust for differences in popula- Appendix 1. tion sizes of each province in South Africa. All analyses were The coefficient b indicates the changes in the expected conducted at 5% level of significance using Stata 14. duration of SbI for every unit change in the jth predictor. The exponentials of the coefficients suggest the likelihood Ethical Approval of having a second-birth; thus, HR > 1 indicates higher like- lihood, HR < 1 lower likelihood, and HR = 1 equal The South African Medical Research Council (SAMRC) likelihood. Ethics Committee and the ICF Institutional Review Board The crude cox proportional hazard model was used to reviewed and approved the survey protocol, instruments, and explore the association between maternal education and SbI material prior to the collection of data. Refer to SADHS in both rural and urban settings, including the interaction 2019 report for the details of the ethical approvals for the between the two key variables. All significant variables parent study (NDoH et al., 2019). The Demographic and (p < .15) premised on the log-rank test were thereafter Health Surveys Program, ICT International, USA also per- included in the extended Cox regression model using a-two mitted the usage of the dataset for the present analysis. Afolabi and Palamuleni 5 rural-urban differences in SbI were evident. For instance, Results the rural-urban differences in MtSb were observed among Participants’ Characteristics women who never married (rural—95% CI 74 [69, 80] vs. urban—95% CI 92 [86, 98] months, p ≤ .001), who were The average (±SD) ages of the rural and urban women were from low-wealthier households (rural—95% CI 63 [61, 65] 33.1 (±8.6) and 34.0 (±8.5) years, respectively. While vs. urban—95% CI 69 [64, 77] months, p = .007) or lived in women aged <20 years at first birth constituted the highest Eastern Cape province (rural—95% CI 54 [48, 60] vs. percentage of rural (51.1%) and urban (44.4%) women stud- urban—95% CI 78 [68, 88] months, p < 0.001). Meanwhile, ied, those of aged ≥25 years constituted the least (Table 2). nearly all the considered characteristics had significant About three-quarters of the women had secondary educa- (p < 0.15) differences in their respective survival curves cat- tion, while about one-fifth never used contraceptives in both egory except for first-birth sex and survival of the first child rural and urban settings. Nearly all the women’s first child- in both settings (Table 2). birth survived in both the rural (96.1%) and urban (96.6%) areas. Of 6,039 women analyzed, 4,015 (66.5%) had had second Influence of Maternal Education on Time to births—1,847 (69.3%) in rural and 2,168 (64.3%) in urban— Second Childbirth by Rural-Urban Residence prior to the date of the interview. Black/African was the most Tables 3a and 3b respectively present unadjusted and adjusted common race of both the rural (97.7%) and the urban (78.7%) hazard ratios of the influence of maternal education on sec- women; however, women of other (White, Colored, and ond birth interval in both rural and urban contexts. Indian/Asian) race had the majority of second births in both Rural women who had a secondary (HR = 0.79; 95% CI the rural (77.0%) and urban (70.4%) settings. The percentage [0.71, 0.88]) and a higher (HR = 0.67; 95% CI [0.55, 0.82]) of women who had second birth decreased as the women’s education respectively were 21% and 33% more likely to age at first birth and their educational attainment increased, lengthen SbI relative to their rural counterparts who had at in both settings. While 85.6% of rural and 82.9% of urban most a primary education. Similarly, women who had a post women who had at most a primary education had second- primary level of education (secondary—HR = 0.78; 95% CI birth, only 60.4% of rural and 56.5% of urban women who [0.69, 0.88]; higher—HR = 0.74; 95% CI [0.63, 0.88]) were had a higher education had second-birth. Most women who about 22% to 25% times more likely to delay second birth neither decided nor desired another child had had second- compared to those who had at most a primary education in birth. One-third of the rural women reported that they were urban setting. Relative to rural women who had at most a currently working compared to 43.0% of urban women. primary education, women who had at most a primary and Compared to 44.9% of urban women, more than half of rural lived in urban including those who had a post primary edu- women were never married. The median survival time to sec- cation irrespective of their residential abode were about ond birth (MtSb) was 66 (95% CI [64.0, 69.0]) and 76 (95% 20% to 40% significantly more likely to delay second-birth CI [74.0, 78.0]) months amongst rural and urban women, (Table 3a). respectively (Table 2). This finding is also shown in Figure 1, These associations remained relatively alike when other a probability plot showing the risk and cumulative survival socio-demographic factors were controlled for in rural or curve of having a second birth. urban context. Relative to women who had at most a primary education, the risk of having second-birth was significantly Urban-Rural Differences in the Pattern of Second lower among rural residents who had a post primary educa- Birth Interval tion (secondary—aHR = 0. 68, 95% CI [0.70, 0.82]; higher— The findings show a shorter MtSb among women who aHR = 0. 62, 95% CI [0.62, 0.93]). A similar but reduced-lower attained at most a primary education compared to those who likelihood of having second-birth was observed among urban had a secondary or higher education, irrespective of their women who had a post primary education (secondary— residential abode. Urban women who had at most primary aHR = 0. 79, 95% CI [0.69, 0.90]; higher—aHR = 0. 80, 95% (urban—63 [57, 71] vs rural—55 [51, 59] months, p = .025) CI [0.66, 0.98]; model 1). For model 2, when all other sig- or secondary (urban—77 [75, 80] vs rural—69 [66, 71] nificant variables were controlled for, women who had a sec- months, p < .001) education had a significant longer MtSb ondary education were 14% more likely to delay second-birth compared to their rural counterparts (Table 2). Figure 2 cor- in rural setting; however, maternal education was not signifi- roborates this finding; it shows the probabilities of having cantly associated with SbI among urban women. For model second birth by maternal education and its interplay with 3, compared to rural residents who had at most a primary place of residence. education, women who had a secondary education (rural— By and large, findings showed a significant longer aver- aHR = 0.87; 95% CI [0.77, 0.97]; urban—aHR = 0.79; 95% age length of SbI among urban women relative to their rural CI [0.69, 0.89]) or a higher education (urban—aHR = 0.82; counterparts (p < .001). Even when disaggregated by the 95% CI [0.86, 0.98]) were more likely to delay second-birth categories of other characteristics considered, significant (Table 3b). 6 SAGE Open Table 2. Distribution of Women’s Characteristics and Their Association With the Median Time to Second Birth According to Rural and Urban Settings. Rural Urban Percentage Percentage Characteristic n at risk (%) had SB MtSb (95% CI) p-Value^ n at risk (%) had SB MtSb (95% CI) p-Value^ p-Value# Maternal education <.001* <.001* At most primary 486 (18.2) 85.6 55 (51–59) 369 (11.0) 82.9 63 (57–71) .025* Secondary 1973 (74.0) 66.2 69 (66–71) 2574 (76.3) 62.9 77 (75–80) <.001* Higher 207 (7.8) 60.4 79 (65–87) 430 (12.7) 56.5 78 (71–87) .596 Region .002* .095 Western Cape 21 (0.8) 85.7 66 (49–70) 430 (12.7) 66.5 74 (67–80) .256 Eastern Cape 380 (14.3) 73.2 54 (48–60) 369 (10.9) 60.7 78 (68–88) <.001* Northern Cape 148 (5.6) 69.6 78 (67–86) 369 (10.9) 66.7 78 (71–86) .633 Free State 69 (2.6) 60.9 83 (59–99) 520 (15.4) 62.9 80 (75–93) .735 Kwazulu-Natal 496 (18.6) 66.5 61 (56–65) 410 (12.2) 62.4 73 (68–79) .007* North West 346 (13.0) 72.3 67 (58–71) 311 (9.2) 67.8 71 (59–76) .955 Gauteng 60 (2.3) 66.7 78 (62–104) 565 (16.8) 65.0 79 (74–84) .715 Mpumalanga 483 (18.1) 66.3 69 (63–78) 288 (8.5) 63.5 68 (60–77) .916 Limpopo 663 (24.9) 70.3 70 (65–77) 111 (3.3) 61.3 81 (63–103) .247 Ethnicity .034* .004* Black/African 2605 (97.7) 69.1 66 (64–69) 2654 (78.7) 62.6 78 (75–80) <.001* Others 61 (2.3) 77.0 60 (44–70) 719 (21.3) 70.4 69 (63–75) .021* Wealth Index .001* <.001* Low 1558 (58.4) 70.5 63 (61–65) 610 (18.1) 65.7 69 (64–77) .007* Middle 980 (36.8) 67.0 71 (66–78) 1159 (34.4) 66.2 72 (68–76) .922 High 128 (4.8) 71.1 73 (68–89) 1604 (47.6) 62.3 80 (77–86) .234 Partner’s education <.001* <.001* At most primary 296 (11.0) 86.5 59 (52–65) 235 (7.0) 83.8 65 (57–77) .118 Secondary 586 (22.0) 82.1 62 (59–65) 1061 (31.5) 76.4 67 (63–71) .104 Higher 79 (3.0) 79.7 65 (48–72) 235 (7.0) 73.2 68 (61–76) .453 No partner/don’t 1705 (64.0) 61.4 70 (67–75) 1842 (54.6) 53.6 87 (81–93) <.001* know Age at FB .111 .007* <20 1361 (51.1) 73.2 65 (62–69) 1496 (44.4) 69.9 76 (73–78) <.001* 20–24 1008 (37.8) 69.2 66 (62–69) 1285 (38.1) 63.3 75 (71–80) .002* ≥25 297 (11.1) 51.5 72 (63–83) 592 (17.6) 52.2 77 (72–91) .002* Marital status at FB <.001* <.001* Never married 1441 (54.1) 56.4 74 (69–80) 1515 (44.9) 48.4 92 (86–98) <.001* FB before cohabiting 652 (24.5) 85.6 62 (58–66) 1030 (30.5) 79.3 76 (72–80) <.001* FB after cohabiting 573 (21.5) 83.1 61 (56–63) 828 (24.5) 74.5 59 (57–63) .357 Desire more children <.001* <.001* Desired 820 (30.8) 44.9 92 (82–105) 1008 (29.9) 35.6 122 (104–139) .140 Desired no more 1552 (58.2) 82.2 60 (57–62) 1926 (57.1) 76.6 69 (66–71) <.001* Undecided 294 (11.0) 66.9 67 (59–75) 439 (13.0) 76.1 71 (65–78) .004* Employment status .022* <.001* Not working 1778 (66.7) 64.8 64 (61–67) 1924 (57.0) 62.1 71 (68–75) <.001* Working 888 (33.3) 78.3 69 (65–74) 1449 (43.0) 67.2 81 (78–86) .010* FB sex .982 .899 Male 1399 (52.5) 69.3 67 (63–70) 1739 (51.6) 64.2 75 (71–77) <.001* Female 1267 (47.5) 69.3 65 (62–69) 1634 (48.4) 64.4 77 (74–81) <.001* FB survival 0.873 .355 Dead 104 (3.9) 74.0 53 (48–73) 115 (3.4) 65.2 78 (60–104) .251 Alive 2562 (96.1) 69.1 66 (64–69) 3258 (96.6) 64.2 76 (74–78) <.001* Contraceptive <.001* <.001* Never 589 (22.1) 66.7 70 (65–77) 761 (22.6) 62.8 86 (78–95) <.001* Formal 581 (21.8) 69.9 67 (63–74) 714 (21.2) 64.3 75 (71–82) .058 Current 1496 (56.1) 70.1 64 (61–67) 1898 (55.2) 64.9 74 (70–77) <.001* Total 2666 69.3 66 (64–69) 3373 64.3 76 (74–78) <.001* *p < .05; ^ based on log-rank test; # differences in SbI between urban and rural areas based on Mann-Whitney test; CI = 95% Confidence interval; n = number of women; FB = First-Birth; SB = second birth; MtSb = Median survival time to second childbirth. Afolabi and Palamuleni 7 Figure 1. Overall survival and hazard function of second birth interval by residence. The probability plot showing the cumulative survival curve and the hazard of having a second childbirth. Other significant predictors of SbI were ethnicity, employ- likelihood of having a second-birth at baseline, the likeli- ment status, marital status at first-birth and desire for more hood significantly increased by 0.5% for every unit increase children for both urban and rural residents. While factors in survival time for both settings (Table 3b). such as wealth index and age at first-birth had significant relationships with SbI among women residing in urban areas, Discussion ethnicity was a peculiar factor associated with SbI among rural residents. Of these, age at first-birth of urban and wom- The present study investigated the influence of maternal edu- en’s region of rural residents had time-varying effects on SbI. cation on first to second childbirth interval by residence The chance of having second-birth increased with increasing amongst women in South Africa using 2016 SADHS nation- age at first-birth. Urban women aged ≥25 years at first birth ally representative data. The proportion of the women who were 29% times more likely to shorten SbI at baseline; how- had a second-birth was lower in urban compared to rural. ever, the likelihood declined by 1% at every unit increase in Even though a lengthened SbI compared to the optimal inter- survival time relative to teenage mothers at first-birth. birth interval of 3 to 5 years (World Health Organization, Whereas among rural residents, Eastern Cape and Kwazulu- 2007) was observed in both settings, it was significantly lon- Natal women respectively had about two times the likeli- ger in urban. The SbI is longer among the women studied hood of having second-birth at baseline, but declined by 1% compared to the median birth interval of 55 and 56 months for every increase in unit of time compared with their (though confined to non-first births in the 5 years prior to the Limpopo counterparts. survey year) respectively among rural and urban sub-popula- Black/African women (aHR = 0.87; 95% CI [0.77, 0.99]) tions in South Africa (NDoH et al., 2019). had a 13% lower tendency to shorten SbI compared to These results suggest a considerable difference in SbI women of other race living in urban setting. However, Black between South Africa women residing in rural and urban women had a 31% lower risk of welcoming a second-birth areas. Urban women, essentially, have a higher tendency to among rural residents. Women who had their first births after increase SbI than those in rural. Similar result has been cohabiting had about 44% higher risks of having a shortened reported in other studies (Ahammed et al., 2019; Yohannes, SbI among rural residents, but 79% among urban residents, et al., 2011) that women from rural areas were more likely to than those who never married. Relatively, women who were have a second birth relative to their urban counterparts. The working had about 15% lower likelihood of having a second- observed differential may be attributed to the heightened birth irrespective of their place of residence (rural— costs of childbearing in urban relative to rural settings (Lerch, aHR = 0.87; 95% CI [0.79, 0.96]; urban—aHR = 0.83; 95% 2018). Besides, the difference could be due to disparities in CI [0.76, 0.91]). Urban women who were from wealthier the women’s residence-specific socio-economic and demo- households had a 18% reduced risk of having a second-birth, graphic characteristics. For instance, a higher percentage of but not for rural residents. Although women who formerly or rural women were unemployed, from low-wealthier house- currently used contraceptives had a lower nonsignificant holds, lived in Eastern Cape and Kwazulu-Natal provinces or 8 SAGE Open Figure 2. Hazard function of time to second birth by maternal education (left-panel, unadjusted; right-panel, adjusted) in rural and urban residence. Afolabi and Palamuleni 9 Table 3a. Unadjusted Hazard Ratios of Second Birth Interval by Maternal Education Among Rural and Urban Women. Rural Urban Pooled data Characteristics HR (95% CI) p-Value HR (95% CI) p-Value HR (95% CI) p-Value Education ≤Primary (Ref) 1 1 Secondary 0.79 (0.71, 0.88) <.001 0.78 (0.69, 0.88) <.001 Higher 0.67 (0.55, 0.82) <.001 0.74 (0.63, 0.88) <.001 Education/residence ≤Primary and rural (Ref) 1 ≤Primary and urban 0.81 (0.70, 0.94) .006 Secondary and rural 0.79 (0.71, 0.88) <.001 Secondary and urban 0.63 (0.56, 0.70) <.001 Higher and rural 0.67 (0.55, 0.82) <.001 Higher and urban 0.60 (0.51, 0.71) <.001 LL 12,823.3 15,761.0 31,349.9 AIC 25,650.6 31,525.9 62,709.8 HR = hazard ratio; 95% CI = Confidence interval; Ref = reference category; LL = log-likelihood; AIC = Akaike information criterion. Table 3b. Adjusted Hazard Ratio of the Relationship Between Maternal Education and SbI According to Rural-Urban Residence and Their Pooled Sample. aHR (95% CI)—Rural aHR (95% CI)—Urban aHR (95% CI)—pooled Characteristics Model 1 Model 2 Model 1 Model 2 Model 3 Education ≤Primary (Ref) 1 1 1 1 a c a Secondary 0.68 (0.55, 0.83) 0.86 (0.76, 0.96) 0.79 (0.69, 0.90) 0.93 (0.82, 1.06) c c Higher 0.62 (0.42, 0.93) 0.84 (0.67, 1.04) 0.80 (0.66, 0.98) 0.96 (0.80, 1.16) Education/residence ≤Primary and rural (Ref) 1 ≤Primary and urban 0.85 (0.73, 1.00) Secondary and rural 0.87 (0.77, 0.97) Secondary and urban 0.79 (0.69, 0.89) Higher and rural 0.87 (0.70, 1.07) Higher and urban 0.82 (0.68, 0.98) Age at FB <20 (Ref) 1 1 1 1 1 20–24 1.05 (0.95, 1.16) 1.07 (0.97, 1.18) 1.08 (0.91, 1.29) 1.05 (0.88, 1.25) 1.03 (0.96, 1.10) c c c ≥25 0.89 (0.75, 1.06) 0.97 (0.81, 1.16) 1.37 (1.06, 1.75) 1.29 (1.01, 1.66) 0.89 (0.80, 0.99) Region Western Cape 0.81 (0.29, 2.27) 0.69 (0.25, 1.94) 0.98 (0.74, 1.29) 0.89 (0.67, 1.17) 1.23 (0.93, 1.63) a a a Eastern Cape 2.02 (1.52, 2.68) 1.89 (1.42, 2.51) 1.04 (0.79, 1.36) 0.91 (0.69, 1.20) 1.51 (1.19, 1.91) Northern Cape 0.73 (0.46, 1.14) 0.70 (0.45, 1.11) 0.98 (0.74, 1.30) 0.87 (0.66, 1.15) 0.82 (0.62, 1.07) Free State 1.01 (0.54, 1.91) 0.96 (0.52, 1.77) 1.05 (0.81, 1.37) 0.87 (0.67, 1.13) 1.07 (0.83, 1.39) a a a Kwazulu-Natal 1.83 (1.39, 2.40) 1.83 (1.39, 2.40) 1.23 (0.94, 1.60) 1.16 (0.89, 1.52) 1.70 (1.35, 2.13) North West 1.29 (0.96, 1.73) 1.20 (0.89, 1.63) 1.31 (0.99, 1.72) 1.20 (0.91, 1.58) 1.29 (1.01, 1.64) Gauteng 0.79 (0.41, 1.53) 0.75 (0.39, 1.44) 1.06 (0.82, 1.37) 0.98 (0.76, 1.28) 1.13 (0.88, 1.46) Mpumalanga 1.20 (0.91, 1.58) 1.17 (0.89, 1.55) 1.17 (0.88, 1.54) 1.06 (0.80, 1.40) 1.28 (1.01, 1.62) Limpopo (Ref) 1 1 1 1 1 Ethnicity c a c b Black/African 0.74 (0.52, 1.06) 0.69 (0.48, 0.99) 0.80 (0.70, 0.90) 0.87 (0.77, 0.99) 0.84 (0.75, 0.94) Other (Ref) 1 1 1 1 1 Wealth status Low (Ref) 1 1 1 1 1 Middle 0.86 (0.78, 0.96) 0.91 (0.82, 1.01) 1.01 (0.89, 1.14) 1.02 (0.90, 1.15) 0.94 (0.87, 1.02) a b a High 0.85 (0.67, 1.07) 0.80 (0.63, 1.02) 0.79 (0.70, 0.90) 0.82 (0.72, 0.92) 0.77 (0.70, 0.85) (continued) 10 SAGE Open Table 3b. (continued) aHR (95% CI)—Rural aHR (95% CI)—Urban aHR (95% CI)—pooled Characteristics Model 1 Model 2 Model 1 Model 2 Model 3 Employment Status Not working (Ref) 1 1 1 1 1 b a a a Working 0.93 (0.85, 1.03) 0.87 (0.79, 0.96) 0.74 (0.63, 0.88) 0.83 (0.76, 0.91) 0.85 (0.80, 0.91) Partner’s education At most primary 1.15 (1.00, 1.33) 1.27 (1.08, 1.50) a a Secondary 1.32 (1.18, 1.47) 1.52 (1.38, 1.68) a a Higher 1.61 (1.23, 2.11) 1.67 (1.39, 1.99) No partner (Ref) 1 1 Marital status at FB Never married (Ref) 1 1 1 a a a FB before cohabiting 1.30 (1.16, 1.45) 1.42 (1.28, 1.57) 1.35 (1.26, 1.46) a a a FB after cohabiting 1.44 (1.28, 1.63) 1.79 (1.59, 2.01) 1.62 (1.49, 1.76) Desire for children Desired (Ref) 1 1 1 a a a No more desired 1.86 (1.65, 2.09) 2.11 (1.88, 2.38) 1.96 (1.81, 2.13) a a a Undecided 1.51 (1.27, 1.80) 1.89 (1.62, 2.20) 1.71 (1.52, 1.91) Contraceptive Never (Ref) 1 1 1 Former 0.91 (0.70, 1.20) 1.22 (0.96, 1.55) 1.04 (0.87, 1.25) Current 0.96 (0.77, 1.21) 1.09 (0.89, 1.33) 0.99 (0.85, 1.15) Education ≤Primary (Ref) 1 Secondary × t 1.00 (1.00, 1.01) Higher × t 1.00 (1.00, 1.01) Age at FB <20 (Ref) 1 1 20–24 × t 1.00 (1.00, 1.00) 1.00 (1.00, 1.00) a a ≥25 × t 0.99 (0.99, 1.00) 0.99 (0.99, 1.00) Region Western Cape × t 1.01 (0.99, 1.02) 1.01 (0.99, 1.02) 0.99 (0.99, 1.00) a a a Eastern Cape × t 0.99 (0.99, 1.00) 0.99 (0.99, 1.00) 0.99 (0.99, 1.00) Northern Cape × t 1.00 (1.00, 1.01) 1.00 (1.00, 1.01) 1.00(1.00, 1.00) Free State × t 1.00 (0.99, 1.01) 1.00(0.99, 1.01) 1.00(0.99, 1.00) a a a Kwazulu-Natal × t 0.99(0.99, 1.00) 0.99(0.99, 1.00) 0.99(0.99, 1.00) North West × t 1.00(0.99, 1.00) 1.00(0.99, 1.00) 1.00(0.99, 1.00) Gauteng × t 1.00(0.99, 1.01) 1.00(0.99, 1.01) 1.00(0.99, 1.00) Mpumalanga × t 1.00(0.99, 1.00) 1.00(0.99, 1.00) 1.00(0.99, 1.00) Limpopo (Ref) 1 1 1 Employment status Not working (Ref) 1 Working × t 1.00(1.00, 1.00) Contraceptive Never (Ref) 1 1 1 Former × t 1.00(1.00, 1.01) 1.00(1.00, 1.00) 1.00(1.00, 1.00) b c a Current × t 1.00(1.00, 1.01) 1.00(1.00, 1.01) 1.00(1.00, 1.01) -LL 12,767.5 12,682.3 15,672.5 15,553.6 31,025.4 AIC 25,593 25,428.6 31,389 31,159.2 62,120.8 aHR = adjusted hazard ratio; CI = 95% confidence interval; Ref = reference category; LL = log-likelihood; AIC = Akaike information criterion; n = number of observations; partners education variable was excluded in models 2 and 3 due to collinearity with marital status at first birth. a b c p < .001. p < .01. p < .05. had at most a primary education compared to their urban 2019). Saliently, the study suggests that rural women who counterparts in this study. This comparative differential of had a secondary education were more likely than those who the rural and urban SbI’s lengthening demonstrates the had at most a primary education to have a longer SbI. This impact of urbanization on fertility transition. finding remained rather similar after controlling for other The role of non-optimal inter-birth intervals as a critical correlates among rural residents; this may suggest that mater- risk factor of poor maternal and child health outcomes is nal education influence on SbI is relatively independent of a negatively related to maternal education (Molitoris et al., collective effect of other correlates among rural women. On Afolabi and Palamuleni 11 the other hand, the effect of maternal education on SbI sub- at addressing non-use of contraceptives, contraceptive dis- stantially changed while controlling for other variables. This continuation and failure, and by extension, the unwanted indicates that maternal education alone may not have con- births are therefore required. tributed to the relative lengthened SbI among urban women. Furthermore, marital status had a significant impact on This differential aligns with earlier literature (Ahammed et SbI in that women in a relationship (married or cohabiting) al., 2019; Kim, 2003; Klesment et al., 2014). Educational were more likely to have a second birth relative to never- attainment affords women the opportunity of non-childbear- married women in both settings. This finding aligns with ear- ing activities access. Unlike in urban, the rural’s SbI pattern lier literature (Fagbamigbe et al., 2020). The plausible is not surprising for educated women are more likely to post- explanation could be that married women are at a greater risk pone marriage, use contraceptives and desire no more chil- of exposure to sexual activity and pregnancy that could lead dren (Olatoregun, et al., 2014). This could partly be explained to accelerated second childbirth. More so, about half of the from the point of view that an increased in educational attain- South African women who had at least a child were never ment promotes women’s social and economic status. married; this could lead to lengthened SbI as having a second Addressing poor maternal and childbirth outcomes due to child might be delayed until marriage. non-optimal birth interval would therefore require women However, the results of a few predictors of SbI that are empowerment through education, especially in rural setting. peculiar to each of the residences imply some contextual After adjusting for the effect of other variables, ethnicity, impacts. Among urban women, age at first-birth significantly employment status, marital status at first-birth and desire for influenced SbI as older women were more likely to welcome more children significantly impacted on SbI in both residen- a second childbirth. The reason could partly be due to the tial contexts. Generally, a poor level of education could non-attainment of desired parity (De Jonge et al., 2014; induce unemployment, poor wealth status and consequently Mcguire & Stephenson, 2015). The implication of this may impact on SbI. Being employed is a significant factor of SbI be associated with the fact that women who have not attained in that employed women were less likely to have a second their preferred family size are more likely to long for a sec- birth in both the rural and urban settings. This is in line with ond childbirth (Yohannes et al., 2011). This corroborates the existing theory that wage employment or career generally findings from other sub-Saharan countries like Ethiopia, has a delaying impact on women’s fertility transition (Benzies Uganda, and Zimbabwe (Hailu & Gulte, 2016; Mcguire & et al., 2006). The result, however, disagrees with a previous Stephenson, 2015) which claimed older women were more study (Levin et al., 2016) which opined that employed likely to welcome second childbirth. women were more likely to have a second childbirth. Similarly, increasing household wealth lengthened SbI in Black/African women were less prone to having a second urban areas. This suggests that time elapsed between first childbirth either in rural or urban relative to other ethnicities. and second-birth is significantly influenced by the availabil- Of note, rural and urban Black/Africans respectively were ity of financial/material resources. This finding corroborates about 31% and 13% times more likely to delay a second- the claim of other empirical studies (Hailu & Gulte, 2016; birth. This suggests urban Black/Africans proceed sooner to Yohannes et al., 2011). A plausible explanation could be have a second childbirth relative to their rural counterparts. attributed to the accessibility to health care services and This could partly be explained by ample apartheid govern- information relating to the implication of shorter inter-birth ment policies that favored Whites -most of whom lived in intervals on women and child health, which may motivate urban-fertility in South Africa (Palamuleni et al., 2007). the women to delay a second-birth. This buttresses the A high percentage of women who had a second-birth may importance of women empowerment through education and suggest that most women had already attained their desired employment as a means of improving economic status which fertility, for nearly 60% of the women desired no additional may affect SbI. child(ren). Hence, the early second-birth and subsequently, In rural area, on the other hand, women who lived in the the shortened SbI among women who neither decided nor provinces of Eastern Cape or Kwazulu-Natal had a signifi- desired to have more children as observed in both settings. cantly shorter SbI compared with their Limpopo counter- Other likely reason may be ascribed to the increased in pro- parts, though the effect subsequently declined over time. The portion of women with an unmet need for contraception in finding of significant fertility changes in the two provinces is the last two decades, as documented in the 2019 SADHS corroborated by an earlier study (Chersich et al., 2017). report (NDoH et al., 2019). Another plausible explanation Contraception used to delay childbirth has been reported could be that a high proportion of teenage mothers who may in literature as a positive predictor of lengthened inter-birth be unwilling to raise another child, due to economic and psy- intervals (Towriss & Timaeus, 2018). However, this study chological effect of having the first child before marriage, demonstrated a nonsignificant opposing association between may rescind the decision afterwards. Perhaps, this contracts contraception and SbI in the two settings. While the use of the belief that a culture of early sexual intercourse has a contraceptive is protective against shortened SbI among higher tendency of producing many children linked with rural women, contraceptive use among urban women por- shortened birth intervals (Fitaw, et al., 2003). Policies aimed tend to shorten SbI. The nonsignificant result may be linked 12 SAGE Open to the fact that nearly 80% of the women used contraceptive Declaration of Conflicting Interests in both settings. Although, such contraception may be dis- The author(s) declared no potential conflicts of interest with respect continued. According to NDoH et al. (2019), the rate of con- to the research, authorship, and/or publication of this article. traceptive use discontinuation is 29% within 1 year after commencement. Funding Meanwhile, this study is not without limitations. First, The author(s) received no financial support for the research, author- the study design is cross-sectional. The analyzed variables ship, and/or publication of this article. can only provide evidence of a statistical relationship but not a causal relationship between the variables and time to ORCID iD second birth. Second, there may be a possibility of recall Rotimi Felix Afolabi https://orcid.org/0000-0002-0744-1787 bias as the survey entailed self-reported data without any means of verification. Also, the usage of secondary data Supplemental Material limited our potential to sufficiently assess the influence of Supplemental material for this article is available online. some characteristics like breastfeeding practices, abortion rate, and first birth interval length as drivers of second birth interval. Nonetheless, the study has been strengthened References using the most recent large nationally representative data- Ahammed, B., Kabir, M. R., Abedin, M. M., Ali, M., & Islam, set. Besides, the strength of the work includes provision of M. A. (2019). Determinants of different birth intervals of information on how maternal education impacted on SbI in ever married women: Evidence from Bangladesh. Clinical both rural and urban residential settings, which is rarely Epidemiology and Global Health, 7(3), 450–456. https://doi. documented. org/10.1016/j.cegh.2019.01.011 Akinyemi, J. O., Afolabi, R. F., & Awolude, O. A. (2020, October). Semi-parametric model for timing of first childbirth after HIV Conclusion diagnosis among women of childbearing age in Ibadan, Nigeria. PLoS ONE, 15(10), e0240247. https://doi.org/10.1371/journal. Knowledge of the SbI is not only critical at directing the sub- pone.0240247 sequent childbearing experiences but also at improving Atkinson, D. (2014). Rural-Urban linkages: South Africa case maternal and child health. The information is valuable and study. RIMISP. www.rimisp.org important to policy and decision makers in making informed Ball, S. J., Pereira, G., Jacoby, P., Klerk, N., & Stanley, F. J. (2014, decisions on the direction of future fertility patterns at sub- July). Re-evaluation of link between interpregnancy interval national population levels. A longer SbI was observed among and adverse birth outcomes: Retrospective cohort study match- urban women relative to their rural counterparts, though a ing two intervals per mother. BMJ, 349, g4333. https://doi. longer duration of SbI compared to the optimal birth interval org/10.1136/bmj.g4333 Bao, L., Chen, F., & Zheng, Z. (2017). Transition in second birth length is evident in both settings. The findings demonstrate intention in a low fertility context: The case of Jiangsu, China. residence context-specific differential in maternal education Asian Population Studies, 13(2), 198–222. https://doi.org/10.1 effect on SbI among South African women. Increasing 080/17441730.2017.1291125 maternal education is negatively associated with SbI in rural; Bartus, T., Murinkó, L., Szalma, I., & Szél, B. (2013). The effect of however, education may not have distinctively influenced education on second births in hungary: A test of the time-squeeze, the relative lengthened SbI among urban women. Other con- selfselection, and partner-effect hypotheses. Demographic siderable predictors—ethnicity, marital status at first-birth, Research, 28(1), 1–32. https://doi.org/10.4054/DemRes.2013.28.1 desire for more children, and employment status—have been Baschieri, A. (2004). The second birth interval in Egypt: The role identified to impact on SbI in both settings. While being of contraception. In S3RI applications working paper A04/05. older at first-birth could shorten the SbI among urban Southampton United Kingdom. women; rural women living in Eastern Cape or Kwazulu- Benzies, K., Tough, S., Tofflemire, K., Frick, C., Faber, A., & Newburn-Cook, C. (2006). Factors influencing women’s Natal province may welcome a second childbirth early. The decisions about timing of motherhood. Journal of Obstetric, SbI’s differential between the residential contexts should be Gynecologic & Neonatal Nursing, 35(5), 625–633. https://doi. addressed by strengthening the health education systems. org/10.1111/J.1552-6909.2006.00079.X These may have a far-reaching impact on women’s socioeco- Bongaarts, J. (1978). A Framework for Analyzing the Proximate nomic empowerment and adherence to the optimal inter- Determinants of Fertility. Population Council. birth interval, and consequently an improved maternal and Bongaarts, J. (2015). Modeling the fertility impact of the proximate child health. determinants: Time for a tune-up. Demographic Research, 33(19), 535–560. https://doi.org/10.4054/DemRes.2015.33.19 Acknowledgments Bongaarts, J. (2020). Trends in fertility and fertility preferences in We acknowledge the National Demographic and Health Survey’s sub-Saharan Africa: The roles of education and family plan- supervisory bodies in South Africa for granting us access to the data ning programs. Genus, 76(1), 32. https://doi.org/10.1186/ used for our study. s41118-020-00098-z Afolabi and Palamuleni 13 Chen, I., Jhangri, G. S., & Chandra, S. (2014). Relationship between fertility patterns (Working Paper No. 7643). World Bank interpregnancy interval and congenital anomalies. American Policy Research. http://econ.worldbank.org. Journal of Obstetrics and Gynecology, 210(6), 564.e1–564.e8. Mace, R., & Sear, R. (1997). Birth interval and the sex of children https://doi.org/10.1016/j.ajog.2014.02.002 in a traditional African population : An evolutionary analysis. Chersich, M. F., Wabiri, N., Risher, K., Shisana, O., Celentano, Journal of Biosocial Science, 29(4), 499–507. http://eprints. D., Rehle, T., & Rees, H. (2017). Contraception coverage lse.ac.uk and methods used among women in South Africa: A national Mahfouz, E. M., El-Sherbiny, N. A., Wahed, W. Y. A., & Hamed, household survey. South African Medical Journal, 107(4), N. S. (2018). Effect of inter-pregnancy interval on pregnancy 307–314. https://doi.org/10.7196/SAMJ.2017.v107i4.12141 outcome: A prospective study at Fayoum, Egypt. International De Jonge, H. C., Azad, K., Seward, N., Kuddus, A., Shaha, S., Journal of Medicine in Developing Countries, 2(2), 38–44. Beard, J., & Fottrell, E. (2014). Determinants and conse- https://doi.org/10.24911/IJMDC.51-1520268317 quences of short birth interval in rural Bangladesh: A cross- Mcguire, C., & Stephenson, R. (2015). Community factors influ- sectional study. BMC Pregnancy and Childbirth, 14(1), 427. encing birth spacing among married women in Uganda and https://doi.org/10.1186/s12884-014-0427-6 Zimbabwe. African Journal of Reproductive Health, 19(1), 14. Erfani, A., & McQuillan, K. (2009). Rapid fertility decline and the www.measuredhs.com. changing timing of births in Iran. Annual meeting of the popu- Molitoris, J., Barclay, K., & Kolk, M. (2019). When and where birth lation association of America. Detroit, Michigan. spacing matters for child survival: An international comparison Fagbamigbe, A., Ojo, A., Onyeike, N., Okafo, I., Olabuyi, R., & using the DHS. Demography, 56(4), 1349–1370. https://doi. Afololabi, R. (2020). Survival analysis of time interval between org/10.1007/s13524-019-00798-y first and second childbirth among women in Nigeria | African Moultrie, T. A., Sayi, T. S., & Timaeus, I. M. (2012). Birth intervals, Journal of Medicine and Medical Sciences. African Journal of postponement, and fertility decline in Africa: A new type of Medicine and Medical Sciences, 49(1), 241–252. http://www. transition? Birth intervals, postponement, and fertility decline ojshostng.com/index.php/ajmms/article/view/679/358 in Africa: A new type of transition? Population Studies, 66(3), Finlay, J. E., Mejía-Guevara, I., & Akachi, Y. (2018). Inequality 241–258. https://doi.org/10.1080/00324728.2012.701660 in total fertility rates and the proximate determinants of fer- National Department of Health (NDoH), Statistics South Africa tility in 21 sub-Saharan African countries. PLoS One, 13(9), (Stats SA), South African Medical Research Council (SAMRC), e0203344.https://doi.org/10.1371/journal.pone.0203344 & ICF. (2019). South Africa Demographic and Health Survey Fiori, F., Graham, E., & Feng, Z. (2014). Geographical variations 2016. NDoH, NDoH, Stats SA, SAMRC, and ICF in fertility and transition to second and third birth in Britain §. Newman, J. L., & McCulloch, C.. (1984). A hazard rate approach to Advances in Life Course Research, 21, 149–167. https://doi. the timing of births. Econometrica, 52, 939–962. org/10.1016/j.alcr.2013.11.004 Nisha, M. K., Alam, A., Islam, M. T., Huda, T., & Raynes-Greenow, Fitaw, Y., Berhane, Y., & Worku, A. (2003). Differentials of fertil- C. (2019). Risk of adverse pregnancy outcomes associated with ity in rural Butajira. Ethiopian Journal of Health Development, short and long birth intervals in Bangladesh: Evidence from 17(1), 17–25. https://doi.org/10.4314/ejhd.v17i1.9778 six Bangladesh demographic and health surveys, 1996–2014. Grundy, E., & Kravdal, Ø. (2014). Do short birth intervals have BMJ Open, 9(2), e024392. https://doi.org/10.1136/bmjo- long-term implications for parental health? Results from pen-2018-024392 analyses of complete cohort Norwegian register data. Journal Olatoregun, O., Fagbamigbe, A. F., Akinyemi, O. J., Yusuf, O. B., of Epidemiology and Community Health, 68(10), 958–964. & Afolabi, B. E. (2014). A comparative analysis of fertility https://doi.org/10.1136/jech-2014-204191 differentials in Ghana and Nigeria. Afr J Reprod Health, 18(3), Hailu, D., & Gulte, T. (2016). Determinants of short interbirth inter- 36–47. val among reproductive age mothers in Arba Minch District, Palamuleni, M., Kalule-Sabiti, I., & Makiwane, M. (2007). Fertility Ethiopia. International Journal of Reproductive Medicine, and childbearing in South Africa. In Y. Amoateng & T. B. 2016, 6072437. https://doi.org/10.1155/2016/6072437 Heaton (Eds.), Families and households in post-apartheid Impicciatore, R., & Tomatis, F. (2020). The nexus between educa- South Africa: Socio-demographic perspectives (pp. 113–133). tion and fertility in six European countries. Genus, 76(1), 1–20. HSRC Press. www.hsrcpress.ac.za https://doi.org/10.1186/s41118-020-00104-4 Population Reference Bureau. (2018). 2018 World population data Kim, J. (2003). Women’s education in the fertility transition: An sheet. https://www.prb.org/wp-content/uploads/2018/08/2018_ analysis of the second birth interval in Indonesia. Brown. WPDS.pdf Klesment, M., Puur, A., & Rahnu, L. (2014). Varying association Ramarao, S., Townsend, J., & Askew, I. (2006). Correlates of inter- between education and second births in Europe: Comparative birth intervals: Implications of optimal birth spacing strategies analysis based on the EU-SILC data. Demographic Research, in Mozambique. http://www.rhcatalyst.org/ 31(27), 813–859. https://doi.org/10.4054/DemRes.2014.31.27 Rindfuss, R. R., Palmore, J. A., & Bumpass, L. L. (1987). Analyzing Lerch, M. (2017). Urban and rural fertility transitions in the devel- birth intervals: Implications for demographic theory and data oping world: a cohort perspective (MPIDR Working Paper No. collection. Sociological Forum, 2(4), 811–828. https://doi. WP-2017-011). Max Planck Institute for Demographic Research. org/10.1007/BF01124385 http://www.demogr.mpg.de/papers/working/wp-2017-011.pdf Rossi, P., & Rouanet, L. (2015). Gender Preferences in Africa: A Lerch, M. (2018). Fertility decline in urban and rural areas of devel- comparative analysis of fertility choices. World Development, oping countries. Population and Development Review, 45(2), 72, 326–345. https://doi.org/10.1016/j.worlddev.2015.03.010 301–320. https://doi.org/10.1111/padr.12220 Shapiro, D., & Tambashe, B. O. (2000). Fertility transition in Levin, V., Besedina, E., & Aritomi, T. (2016). Going beyond urban and rural areas of Sub-Saharan Africa (Working Paper). the first child analysis of Russian mothers’ desired and actual Department of Economics Pennsylvania State University. 14 SAGE Open Singh, R., Tripathi, V., Kalaivani, M., Singh, K., Dwivedi, S. N., & Van Bavel, J. (2010). Second birth rates across europe: Interactions Dwivedi, S. N. (2012). Determinants of birth intervals in Tamil between women’s level of education and child care enrolment. Nadu in india: Developing cox hazard models with validations Vienna Yearbook of Population Research, 8(1), 107–138. and predictions. Rev Colomb Estad, 35(2), 289–307. http:// https://doi.org/10.1553/populationyearbook2010s107 www.scielo.org.co/pdf/rce/v35nspe2/v35nspe2a07.pdf White, M. J., Muhidin, S., Andrzejewski, C., Tagoe, E., Knight, R., Statistics South Africa (Stats SA). (2015). Census 2011: Fertility in & Reed, H. (2008). Urbanization and fertility: An event-his- South Africa. www.statssa.gov.za tory analysis of coastal Ghana. Demography, 45(4), 803–816. Statistics South Africa (Stats SA). (2018). Men, women and chil- https://doi.org/10.1353/dem.0.0035 dren: Findings of the living conditions survey 2014/15 / statis- World Health Organization. (2007). Report of a WHO technical tics South Africa. www.statssa.gov.za consultation on birth spacing: Geneva, Switzerland 13-15 Stevens, J., Lutz, R., & Osuagwu, N. (2018, February 1). Brief June 2005. World Health Organization. https://apps.who.int/ interpregnancy interval: Are 75% of adolescent mothers iris/bitstream/handle/10665/69855/WHO_RHR_07.1_eng. unaware of the prematurity risk? American Journal of Public pdf?sequence=1&isAllowed=y Health, 108(Suppl 1), S11–S12. https://doi.org/10.2105/ Yohannes, S., Wondafrash, M., Abera, M., & Girma, E. (2011). AJPH.2017.304129 Duration and determinants of birth interval among women of Tabutin, D., & Schoumaker, B. (2004). The demography of sub- child bearing age in Southern Ethiopia. BMC Pregnancy and Saharan Africa from the 1950s to the 2000s: A survey of Childbirth, 11(38), 1–8. https://doi.org/10.1186/1471-2393-11-38 changes and a statistical assesment. Population, 59(3–4), 457– Yohannes, T., Laelago, T., Ayele, M., & Tamrat, T. (2017). 556. https://doi.org/10.2307/3654914 Mortality and morbidity trends and predictors of mortality in Towriss, C., & Timaeus, I. M. (2018). Contraceptive use and under-five children with severe acute malnutrition in Hadiya lengthening birth intervals in rural and urban Eastern Africa zone, South Ethiopia: a four-year retrospective review of Catriona A. Towriss. Demographic Research, 38(64), 2027– hospital-based records (2012–2015). BMC Nutrition, 3(1), 18. 2052. https://doi.org/10.4054/DemRes.2018.38.64 https://doi.org/10.1186/s40795-017-0135-5 United Nations Department of Economic and Social Affairs Zhang, Y., Quist, A., & Enquobahrie, D. (2017). Short birth-to- Population Division. (2017). World Population Prospects: pregnancy intervals among African-born black women in The 2017 revision, key findings and advance tables. United Washington State. The Journal of Maternal-Fetal & Neonatal Nations. Medicine, 2017(1), 1–5.
Journal
SAGE Open – SAGE
Published: Mar 9, 2022
Keywords: survival analysis; maternal education; second birth interval; extended Cox-regression; South Africa; urban and rural differential.