You are on page 1of 9

Mendez Rojas et al.

BMC Public Health (2016) 16:991
DOI 10.1186/s12889-016-3666-9


Community social capital on the timing of
sexual debut and teen birth in Nicaragua: a
multilevel approach
Bomar Mendez Rojas1, Idrissa Beogo2, Patrick Opiyo Owili1, Oluwafunmilade Adesanya1 and Chuan-Yu Chen3,4*

Background: Community attributes have been gradually recognized as critical determinants shaping sexual
behaviors in young population; nevertheless, most of the published studies were conducted in high income
countries. The study aims to examine the association between community social capital with the time to sexual
onset and to first birth in Central America.
Methods: Building upon the 2011/12 Demographic and Health Survey conducted in Nicaragua, we identified a
sample of 2766 community-dwelling female adolescents aged 15 to 19 years. Multilevel survival analyses were
performed to estimate the risks linked with three domains of community social capital (i.e., norms, resource and
social network).
Results: Higher prevalence of female sexual debut (norms) and higher proportion of secondary school or higher
education (resource) in the community are associated with an earlier age of sexual debut by 47 % (p < 0.05) and
16 %, respectively (p < 0.001). Living in a community with a high proportion of females having a child increases the
hazard of teen birth (p < 0.001) and resource is negatively associated with teen childbearing (p < 0.05). Residential
stability and community religious composition (social network) were not linked with teen-onset sex and birth.
Conclusions: The norm and resource aspects of social capital appeared differentially associated with adolescent
sexual and reproductive behaviors. Interventions aiming to tackle unfavorable sexual and reproductive outcomes in
young people should be devised and implemented with integration of social process.
Keywords: Teen birth, Sexual debut, Multilevel, Social capital

Background before 15 years is 13 % [7], consequently heightening the
The connection of teen births with an array of adverse incidence of sexually transmitted diseases and unintended
outcomes, such as early neonatal death, low birth weight, pregnancy [8]. This high fertility occurs in a socio-cultural
anemia, postpartum hemorrhage, puerperal endometritis environment that applauds men with high number of
and high caesarean section rate, has been consistently doc- sexual partners and where abortion is illegal [9]. Although
umented [1–4]. In Nicaragua, 20 % of the total population adolescent sexual and reproductive behavior are context-
are adolescents [5]. The fertility rate was estimated 89 per sensitive [10, 11], until now most of existing research in
1000 girls aged 15 to 19 years in 2014—the second highest this field has mainly provided evidence concerning the
rate in Latin America [5] and teenage girls account for effect of individual-level characteristics, as highlighted in
approximately 25 % of births and about half of those recent systematic reviews [12–14].
births are unintended [6]. The prevalence of sexual debut Findings from prior nationwide representative studies
supporting the existence of contextual effect on ado-
* Correspondence: lescent reproductive outcomes in Nicaragua are mani-
fested in urban –rural differentials in sexual debut
Institute of Public Health, National Yang-Ming University, Medical Building,
Rm 210, 155, Sec. 2, Linong Street, Taipei 112, Taiwan
Center of Neuropsychiatric Research, National Health Research Institutes, 35
among adolescents [9, 15]. Aside from residence area,
Keyan Road, Zhunan, Miaoli County 350, Taiwan smaller geographical units such as neighborhood are of
Full list of author information is available at the end of the article

© 2016 The Author(s). Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (, which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
( applies to the data made available in this article, unless otherwise stated.

Social network. 29]. 33]. especially in high income countries [16]. women was 91 % [34].K. reflects live birth was recorded from the birth certificate or vac- the attachment and social interaction with community as cination card. the ecological study approach [19. 32]. mea- the respondents to be interviewed by a same-sex mem- sures of organizational life. Date of first dential stability and religious affiliation [28. Given its potential malleability. For teens with no sexual debut at survey time. the gested that social capital (measured by group membership analytic sample includes 2 766 females Nicaraguan who and social trust) may serve as a mediator for income in- were aged 15 to 19 years at survey time. 20. interview women aged 15 to 49 years and men aged 15 social capital is of relevance to policy makers [22–24]. As a proxy for community we used the point is that studies of ecological design are often limited census tract (n = 356). In this paper we adopted the multilevel ap.e. with cross-sectional design select- the provision of resources and the formation of norms. probably face-to-face interview. sexual debut) hood’s ethnic composition when social capital is measured and provides enough statistical power to detect differ- through collective efficacy (a composite scale based on ences in the distribution of the outcome across main social cohesion and social control) [25]. and membership in community organizations [30]. asked the month and year when it occurred. This series of surveys are carried out every Nevertheless. resource. BMC Public Health (2016) 16:991 Page 2 of 9 interest for policy decision since they represent a proxim. control) and ever having sexual intercourse [18]. some scholars have urban area [37]. the available results pertaining to the con- 5 years and aim to generate information about maternal nection between social capital and sexual and reproductive and child health.Mendez Rojas et al. Social ate environment where adolescents commonly spend spare norms. tion and material resources for collective efforts [27. A noteworthy predictors [36]. social connections and social Nicaragua (years 2011/2012). One study reported DHS’ data collection procedure strongly recommended a negative correlation between social capital (e. To address this limitation. defined as the period between sexual debut capital is associated with sexual onset and teen birth in and the day. which define the acceptability of behaviors and time out of their households and create new relationships can be viewed as the basis to build and maintain trust [16]. A study from the U. involvement in public affairs ber who was also responsible for filling the information and social trust) and teen pregnancy [19].S. In order to facilitate responses Several studies in the United States had adopted the to sensitive issues and decrease social desirability. turned to the multilevel approach. . This sample equality on teen births [20] and the link between social size exceeds the minimum sample size of 1 365 adoles- capital and teen births may be moderated by neighbor- cents required to achieve 65 events (i. social variable concerning sexual debut. lected first live birth rather than date of first pregnancy proach to tackle potential effects of three perspectives of because the latter is not available in the DHS. indexed by resi. were measured by percentage of childbearing social network. we used the percentage of resi- mechanism of neighborhood attributes on teenage repro. which usually corresponds to a in disentangling the contextual effect from compositional village in the rural area or to a neighborhood in the effect [26].e. We se- Nicaragua. the respondents were norms. re. ing households through a two-stage sampling strategy to values. Outcome variable ports an insignificant association between neighborhood The first outcome variable is the age at sexual onset (in social capital (defined as social cohesion/trust and social years). and norms. Study design and data The theory based on social capital posits that the inter- This research used data from the DHS conducted in play of social processes (e. and beliefs [21]. and some sug- given during the survey [35]..g. The DHS is a nationally interaction) will have impact on individual health through representative survey. the current study used the latest Demographic analysis).. Among the neighborhood-level factors linked with popula- tion health outcomes. month and year of first live birth. Recognizing the neighborhood-based homogeneity in [31. Study variables ated with teen birth [17]. censor under the terminology of survival literature. and the response rate of eligible due to variation in analytical approach and measure. dents with higher education to index the access to informa- ductive health.g. The second outcome variable is the time to and Health Surveys (DHS) to examine whether social first birth. For the community-level social capital: social network. For teens with no birth (censor) the date well as the degree of participation in community affairs of survey interview was used as the ending time. social capital is one of the commonly Methods reported neighborhood characteristics [16–20]. an in. the age at interview was used to include them in the In order to address the abovementioned gaps in the analysis (i. creasing attention has been paid to the potential role and For resource accessibility. relationship. The information was collected by health amongst young people were rather mixed. women with sex behaviors or already having a child. whereas a study from U. to 59 years. found that social capital (defined as social cohesion and social control) at neighborhood level is negatively associ.. and resource [27]. 25]. In the present study.

marital porates the notion of access and utilization of common status (at survey time). Among the entire surveyed sample.. we focused on this religious munity level. women between 15 to 49 years in each census tract the median survival time is prominently longer in the fol- [43]. Regarding the debut and having a child) based on the responses of subsample of female youth who ever had sex (n = 1 247). This variable was created to take the utilization of standard Cox regression would lead to im- into account the responses from men aged 15 to precise estimations. norms and networks” [38]. Model 1 displays the community”.g. A simi. we implemented survival regres- be less socially integrated which may hinder the access to sion models to estimate the hazard rate of the timing of emotional and material resources that operate in favor of sexual debut and first birth. the efficiency of society by facilitating coordinated ac- tions. although .e. we hy. to on the existence of trust to others community members describe the population characteristics and estimate the and the strength of connections among them [39]. woman’s religion. People from the lowest wealth index hence we estimated community social norms indicators (21. 27]. The percentages of Catholics in the 12. two-level survival model. below 20 years. A parent or guardian must pro- in each census tract. 13]. unmarried (30 months). In model 2 each contextual variable was ad- mean proportion by census tract was estimated. we used the complex pothesized that those social processes are more fragile for survey analysis to account for a multi-stage sampling people who have lived in the same community for less design and correct for unequal probabilities of selection than 10 years (the same cutoff point of 10 years has been [44].g. to 49 years and men aged 15 to 59 years in each census Table 2 shows the distribution of the community vari- tract. The allows us to capture the heterogeneous composition of DHS guarantee respondent’s anonymity. the majority (54. At individual-level.0 using frailtypack. and the responses for those who have unadjusted hazard ratios from community. To facilitate the interpretation. We anticipated this approach vide consent prior to participation of adolescents [35]. First. 41]. and prior history of abortion cerns “features of social organization that can improve into account by statistical adjustment [12. proportion of childbearing age women with sexual est (16 years) median time to sexual debut. central tendency measures used in prior research using DHS) [37. we employed a 59 years and women aged 15 to 49 years: each respond. ables. we also took woman’s years of educa- ital on Bordieu and Putnam’s theories. which leads to a violation of the as- proportion of people who have lived in the community sumption of independent distribution of error terms [26]. they may were estimated. being in the with secondary school or higher education was also es. tion. for community variables. 40. Because 50 % of Nicaraguans are For all models we incorporated probability weights at com- affiliated with Catholicism. such as trust. Descriptive analyses were conducted in Stata denomination [42]. for less than 10 years. The justed for individual characteristics. all community. for less than 10 years in the community is 17. timated by averaging the responses of women aged 15 or residing in urban setting (30 months).and community-level variables in model 3. women from her families or neighborhood). female age at sexual debut.4 %) have the short- (i. The mean percentage of inhabitants that have lived level variables were standardized. To overcome this issue.. BMC Public Health (2016) 16:991 Page 3 of 9 Main predictors Covariates This study based its definition of community social cap. and the latter con. Therefore. Ethics lar procedure was followed to estimate the percentage All DHS protocols were approved by the ICF’s Institutional of community inhabitants with no religious affiliation Review Board (IRB) [37]. residence area. Given Statistical analysis that the formulation of community social capital depends The analyses were broken down into two parts.0 %) and those with no religion (15.75.Mendez Rojas et al. richer wealth index (34 months). women’s household wealth index quintiles. Second. The former incor. Table 1 portrays the baseline characteristics. Since adolescents are nested health [28]. Finally. our first main predictor is the within communities. religious institutions) and therefore obtain more accurate estimations of the Results extent to which the community-level of trust and par. the proportion of community members lowing variables: high education (30 months).7 %. 7 years of education and has the lowest median time to first ment (e. median survival time to first birth. community were derived using the responses from se- lected women and men regardless of their age. community social networks (e. and finally we entered second main predictor is related to religious commu.and individual- lived for less than 10 years were aggregated and the level variables. As regards social norms. nity composition.6 %) has less than reproductive behaviors in their surrounding environ. sexual experience.2. with teenage girls as the first level ent was asked “How long they have been living in the and communities as the second level. all individual. the mean year of edu- enced not only by their peers but also by adult women’s cation attainment is 7..0 and regression analysis in R 3. This study included a total of 2 766 young females aged ticipation influence adolescent’s reproductive outcomes. resources for mutual benefit [21. a teenage girls may be influ.

3 and the inter.7.6 33 Middle 26.5 63.6–64.7 18.0 72 64.3 30 Wealth index quintiles Poorest 21.6.5 17 37.5 Percentage of childbearing aged females with onset sexual debut (age 15 to 49 years) 85. IQR =26.1 24 0.1 27.3 28 Residence area Urban 54.4).60 31 Marital status Unmarried 44.4 25 Have you ever had abortions? Yes 14.9 11. The mean percent.9 30 0.9 44.5 8.5–64.0 17 51.4 27 Residential stability: stables.9 10.7 13. the adolescent have lived less than 10 years in the community Respondents who belong to other religion were excluded from the analysis a %wt: Weighted percentages between-community variation was observed as denoted proportion of having experienced sexual debut among by the standard deviation (SD) of 13.9 % Table 3 exhibits the results of three models of survival (SD = 10.90 (0.7 Percentage of community inhabitants with no religion 14.50 Catholics 44.0 16 <0.3). 21.7).5 27 0.5 30 0.001 26.3 to 23.3 19 42.2 31 Richer 18. 85.91 No 85.0 24 Age of sexual debut (years) 15 or under 62.3 19 9.1.9 18 <0.08) 6.7–80 . IQR = 80. and the average proportion of 49 years in the community have had at least a child.9.8 9.3 7. BMC Public Health (2016) 16:991 Page 4 of 9 Table 1 Descriptive statistics of female participants aged 15 to 19 years Variables Female youth N = 2 766 Female youth who ever had sex N = 1 247 Mean (SE) %wta Median time to first Log rank test Mean (SE) %wta Median time to Log rank test sexual encounter (years) (p-value) first birth (months) (p-value) Women’s education (years) 7.6 13.6–91.0.6 17 20.5 34 Richest 17.4 23.002 Rural 45.4 26.05.6 25 0.4 16 <0.1–21.001 57.75 (0.4 36.1 Percentage of community inhabitants that are Catholic 46.6–64.02 High (8 to 16 years) 45.5 25 0. As indicated Table 2 Distribution of community variables (N = 356) Community-level characteristics Mean SD Median Interquartile range Percentage of community inhabitants that have lived in the same community 10 or less years 17.Mendez Rojas et al. 80) of women aged 15 to 23.1.001 48.7 26 Protestants 37.0001 Married 55.3 Percentage of childbearing age women currently having a child (age 15 to 49 years) 71. the adolescent have lived 10 or more years in the community. IQR = 7.6.3 14.001 Poorer 16. women with childbearing ages at the time of survey is quartile range (IQR) of 9.05 85.5 18 24.9 % (SD = 11.7 18 40.3–23.80 16–19 37. respondents with no religion in the community is 14. IQR = 64.7 80. 91.4 25 Woman’s religion No religion 15.11) Low (0 to 7 years) 54.4 % (SD = 71.5 27 0.6 16 <0.5 % (SD = 8. unstable.0 40. and on average age of Catholics at community level is 46.4 Percentage of community inhabitants with secondary school or higher 43.4 17 19.001 20. The community average analysis predicting transition to sexual onset.

89)** Residence area Urban 1 1 Rural 1.aHR: Adjusted hazard ratio were estimated with community – and individual-level variables entered simultaneously All hazard ratios were estimated taking into account household sampling weight in model 1.70) and justed (model 2).00.47 (1.81 (0.07) % of community inhabitants with no religion 1.88 (0.71–1.16 (1. our results demonstrated that social capital. and hazard to have sexual onset.83)*** 0.89–1.04–1.54–1.31–0.16) having a child % of community inhabitants with secondary school or higher 0.e.14) Middle 0.14 (1.92)*** 0.07 (0.95–1.85–1. However.52–2.56)*** (aged 15 to 49 years) % of females in the community aged 15 to 49 years currently 1.22) ** Significant at 5 % level *** Significant at 1 % level Model 1.01 (0.47 (1..051).02–1.43–1. all community capital variables appeared to be percentage of female-onset sexual debut and of resi- statistically significant except for community inhabitants dents with higher education may reduce the hazard to with no religion. community-onset sexual debut.96 (0.43)*** 1.94–1. CI 95 % = 1. On the basis of population-based survey in Central characteristics predicting the transition from sexual debut America.66 (0.72)*** 0.13 (1. only social norms variables remained sig.62–0. showed that one social norms hazard by 76 % (95 % CI = 1.05).24)*** 1.59 (0.52 (1. p = 0.45–0.56)*** 1.62)*** 0.99).22)** 1.56) and social resource 1.67) greater hazard of birth occurrence while this (aHR = 1.07 (0. respectively.91)*** Woman’s religion No religion 1 1 Catholics 0.75 (0. may play an important role in shaping teenage .49–0. with all individual variables ad.10) % of females in the community with onset sexual debut 1.06–1. Discussion The association estimates for community.82)*** 0.04 (0.98 (0. manifested in the social norm and resource perspec- With all variables adjusted. the hazard is reduced by 17 % per 1 year of education CI = 0.33 (1.99–1. (95 % CI = 0.06 (0. 30 % for richest wealth index (p < 0.70 (0.aHR: Adjusted hazard ratio were estimated entering each community variable separately while adjusting for individual-level variables Model 3.63–0.81–0.78–0.39–1.02).25–1.39–1.33) may increase the girls’ hazard was decreased by 5 % per year of education.39–1.74)*** 0.06) Richest 0.53–0. was 27 % lower for the poorer wealth index group (95 % ables.12–1.98 (0.12) 1.02–1.93 (0.84 (0.85)*** Wealth index quintiles Poorest 1 1 Poorer 0.90)** Protestants 0.15) 1.54–0. we found that a higher tives.39 (0.54–0.09) same neighborhood less than 10 years % of community inhabitants that are Catholics 0.75 (0.52–0.75 (0.79–0.23) Richer 0.52 (1.81)*** 0.61)*** 1.uHR: Unadjusted hazard ratio Model 2. At individual-level (i. As for individual-level vari.81–0.97–1.75)*** 0. CI 95 % = 1. to the first birth occurrence were summarized in Table 4.70–0. and 25 % for Catholics and Protestants (all p < 0.85).66–1. Finally.99–1. including both community.Mendez Rojas et al.53 (0.and individual.11) 1. whereas a nificant. adjusted Hazard Ratio variables.14) 1.33–1.63–0.85–1.16.92 (0.64 (0. model 3.93 (0. 11 % (95 % CI = 0.47. women who were married have 36 % (95 % CI = [aHR] = 1.001).33)** Individual characteristics Women’s education (years) 0.76 (0.76–1.83 (0.81–0.48)*** 0. BMC Public Health (2016) 16:991 Page 5 of 9 Table 3 Individual and community characteristics predicting the transition to sexual onset among females aged 15 to 19 years Variables Model 1 Model 2 Model 3 uHR (95 % CI) aHR (95 % CI) aHR (95 % CI) N = 2 766 N = 2 766 N = 2 766 Community social capital (z-score) % of community inhabitants that have lived in the 1.and higher percentage of females having a child may elevate the individual-level variables.99–1.74–1.00) 0. have the first birth by 38 % (95 % CI = 0.95)** 0.02 (0.

aHR: Adjusted hazard ratio were estimated with community – and individual-level variables entered simultaneously All hazard ratios were estimated taking into account household sampling weight girls’ timing of sexual debut and first birth.51 (1.04–1.96)*** 0.98)** 0.91–1. respectively.22)*** 1.aHR: Adjusted hazard ratio were estimated entering each community variable separately while adjusting for individual-level variables Model 3.05 (0.46 (1.Mendez Rojas et al.36)*** 1.99)** Age of sexual onset < 15 years 1 16 to 19 years 0.94 (0.03 (0.00) Middle 0.06 (0. BMC Public Health (2016) 16:991 Page 6 of 9 Table 4 Individual and community characteristics linking transition from sexual debut to first birth among females aged 15 to 19 years Variables Model 1 Model 2 Model 3 uHR (95 % CI) aHR (95 % CI) aHR (95 % CI) n = 1 247 n = 1 247 n = 1 247 Community social capital (z-score) % of community inhabitants that have lived in the 1.70)*** with onset sexual debut % of females in the community aged 15 to 49 years 1.11 (0.87)** 0.74–1.02 (0.12–1.93–1.02 (0.52) Wealth index quintiles Poorest 1 1 Poorer 0.15) 1.15) Protestants 0. The lack of significant association for social network vailing social norms of female-onset sexual debut and suggested that social interactions and social cohesiveness resources may increase the hazard of sexual onset by generated within community organizations may have 47 and 16 %.76 (1.83–1.05 (0.74 (0.74)*** 1.94)** 1.99)** Individual characteristics Women’s education (years) 0.45–0.52–0.94–1.85–1.12) % of females in the community aged 15 to 49 years 0.80 (0.89 (0.02) 0.97–1.54–1.73 (0.49) Marital status Unmarried/no cohabitation 1 1 Married 1.16) Richest 0.36 (1.95 (0.uHR: Unadjusted hazard ratio Model 2.66–1.91–1. As compared with other . 45].10) Richer 0.52–2.22–1.10) 1. of very young women [29.71 (0.23 (1.09) Residence area Urban 1 1 Rural 1.06) 0.03 (0.89 (0.92 (0.84 (0.91–0.92 (0.54) Woman’s religion No religion 1 1 Catholics 0.01 (0.20)** 1.89–1.34) 1.53–0.01–1.56–1. yet reduce the hazard of little influence on both sexual debut or early childbearing having the first birth by 38 and 11 %.91–1.93 (0.02)*** currently having a child % of community inhabitants with secondary school or higher 1.68–1.68 (0.99 (0.12) 1.86)** 0.81–0.84)*** 1.00 (0.92–0.05) 0.89 (0.35) ** Significant at 5 % level *** Significant at 1 % level Model 1.84 (0.62 (0.94 (0.95–1.10 (1.24)*** 0.10) 0.73–1.25–1.82–1.82–1.53–0.20 (0.14) same neighborhood less than 10 years % of community inhabitants that are Catholics 1.78–0.65 (0.50–1.16) 1.67)** Have you ever had abortions No 1 1 Yes 1.89)** 0.65–1.11) 1. The pre.93–1.68–1.14 (1.50–0.62 (0.16) % of community inhabitants with no religion 1.06–1.04 (0.81–1.12–1.13 (1.16) 0.

55]. which may undermine its influence at the geo. through individual religious affiliation. Even though no separ. Further. unit of high. Prior research has established that more sexual debut) and supervision of quality connections to developed areas are less likely to follow conservative and others outside the family may help to mitigate the ero- traditional lifestyle. our results from the subsample of adolescents influence could be exerted by inculcating teachings that who have ever had sexual encounters should be inter- encourage female adolescents to remain a virgin until preted with caution since the younger respondents (e. counties instead of a proxy sexual debut was collected. ute to deter adolescent childbearing [13. Finally. and statistical techniques were used sectional design of our study.. communities with higher pro- and being able to cope with physical demanding activ. This obser- vices). since the DHS with each other and eventually creating bonds that allow offers a limited set of social capital related-variables.. sporadic.g. middle income countries.. the current place where the respondent live and not ne- olics) on sexual intercourse may be operated indirectly cessarily the characteristics before the outcome occurred. echoing the individual-level findings that higher graphical level [46].e. Our community education variable was shown to influ. the sampling proce- and not part of a plan to build a family [52].. Second. of the 19-year old. employment. both individually affairs [47]. nity percentage of female. Further. our multilevel ana- sequent abandonment and stigma (for losing virginity) [9] lytical strategy allows us to vividly comprehend the effect and may make it difficult for girls to build future family or each level exerts on sexual debut and teen birth. indicating the possibility community social capital reflect the characteristics of that religion-related social network (particularly the Cath. the Catholicism may offer adolescents more opportunities such as easier access be less active in neighborhoods or villages (for example: to schooling.g. Nevertheless. this paper is one of the influence on sexual debut and transition to adolescent first studies examining the role of community social motherhood. national representativeness. the access to sexual education (oriented to higher education are expected to have higher levels of social increase awareness of the risks associated with early development [54]. the unmarried level studies (individual-level or ecological). Communities with and therefore may require long-term interventions to be high percentage of inhabitants with secondary school and modified.Mendez Rojas et al. for example. Future studies using a richer set of departure from traditional norms (e. 48–50]. Living in a community with high percent. are known to be more willing to participate in community highlighted the importance of education. Philippines. help fill gaps in current literature mainly based on single- ation occurred after first sexual intercourse. 15-years) are more vulnerable to be right censored even if Prevailing social norms at community-level predict they had the first sexual debut at the same age with those sexual onset and transition to the first birth: high commu. Conclusions matized [53]. It is possible that the first sexual encounter capital on adolescent sexual behavior in the context of was a consequence of male counterpart pressure. were unable to use more comprehensive social capital our study may not be directly comparable with some prior measures used in prior research [19]. portion of inhabitants with secondary school or higher ities associated with motherhood [31. marriage [51]. girls (as 36. community members are actively participating. although multilevel studies conducted in Kenya. we them to cope with adverse situations [29]. For example. or healthcare that may contrib- having fewer churches and offering fewer worship ser. no temporal sequence [37. (a form of human capital) and collectively. adoles- transition to first birth which may be an expression of cents residing in areas with more access to information social pressure to start motherhood while still young and material resources (i.S.g.5 % of our analytical sample) may opt for avoiding having a child out of wedlock and being stig. 37]. due to the cross- of neighborhood). Finally. Given that religious-affiliated people education may delay the time to having the first birth. our measurements of delay the timing of sexual debut. More developed communities may also findings. it is also possible that our measure of religion.. Third. delayed sexual social capital measures may be needed to validate these debut) [9.sexual onset exerts an opposite In spite of these limitations. 32]. We conclude that social norms are significantly associated age of females currently having a child predicts faster with rapid sexual onset and teen birth. which successfully transit to motherhood. thus the observed positive association sion of traditional values observed in more developed of community education with sexual onset may reflect communities [16]. Relationships dures of the DHS would ensure the interpretation with with those characteristics likely end up in disarray and sub. First. and the quality of data collection from DHS is known to be U. . individual-level religion was found to can be established. this Finally. BMC Public Health (2016) 16:991 Page 7 of 9 religious groups such as Protestants. in young popu- related social network did not capture the extent to which lation in order to have healthy development [56]. interacting This paper has several limitations. While social norms are embedded in the cultural context ence both sexual onset and first birth. vation. social desirability or recall bias may arise when date of geographical aggregation (e. where different indicators of social network. education) are also at high risk of early sexual debut.

BJOG-Int J Obstet Gy. manuscript: BMR. Gold R. Managua: Ministerio de Calverton: Measure DHS. BJOG-Int J Obstet Gy. Raneri LG. Accessed 24 Nov 2015. 33. ICF International Rockville. Rm 210. Received: 16 April 2016 Accepted: 12 September 2016 26. 2006. Ouagadougou. IB. BMC Public Health. Sec. Solar O. 7. Wulf D.36:26. Agampodi SB. Taiwan. Canada. Acquisition of data: BMR and CYC. Lammers C. et al. et al. Taiwan. J Adolesc. Hystad PW. Taiwan. Instituto Nacional de Información y Desarrollo. National Yang-Ming University. National Yang-Ming University. CYC is faculty in behavior: an ecological network approach. 2015. et al.4:350–66. We address many thanks to 10. Wu C-C. Moore JL. 2013.121:40–8. p. Althabe F. Vyncke V. Jackson AL. Psychiatry. Research. Social indicators and 32. 5. 2008. OA and CYC. Way S. IRB number: YM103093E. Maryland. 14. Soc Sci Med. income inequality. municipalities. IB was in charge of the department of Nursing of of health. 2011. Risk and protective factors that affect adolescent reproductive Analysis and interpretation of data: BMR. Fisher J.12:1. In: Marmot M. 2012. 23.8:77–83. Salud. Soller B. et al. Browning CR. Newborn Health Registry study. Poulin F. 2014. 2005. Washington. 35 Keyan Road. Blandón L. CEPALSTAT Databases and Statistical Publications. varied opportunities given to complete the study. Holtgrave DR. Taipei. Irwin A. 2012. Heagerty P. Koenig H. 2nd ed. 19. Sabherwal S. A brief conceptual tutorial on multilevel analysis in social epidemiology: investigating contextual phenomena in different groups of people. Neighborhoods and adolescent health-risk Program. Social capital in its place: using social theory to understand Not applicable. 2003. Flap H. Carpiano RM. IB and CYC. 2010. Teen‐aged mothers in contemporary Britain. Conde-Agudelo A. Income inequality and pregnancy ICF International that provided the datasets used in this study. Medical Building. Study concept and design: BMR and CYC. outcomes in adolescent pregnancies: the Global Network’s Maternal 30. Crosby RA. 2 École Nationale de Santé Publique.192:342–9. et al. Trends in adolescent reproductive and 34. Connell F. 2014. Mohnen SM. Glob Public Health. Carson VB. Connell FA. 2005. J Child Psychol presently a postdoctoral fellow at Nursing Department. social capital and inequalities in health. Morisaki N.asp? Relig. spacing. “Sense of community belonging” in health 2. Taiwan. BMC Public Health.42:17–30. Prata N. and adolescents? A literature review. Ganchimeg T. Theorizing religious effects among American adolescents. et al. Stewart C. National Yang-Ming University. Ota E. Early childbearing in Nicaragua: a continuing behind the relationship between neighborhood social capital and individual challenge. Individual. Lancet. health.379:1641–52. Head S. 28.62:148–54. Gibbons L. 2006. Social Determinants of Health discussion Paper 2 (Policy and Practice). . BMC Public Health (2016) 16:991 Page 8 of 9 Acknowledgments 8. the École Nationale de Santé Publique. A conceptual framework for action on the social determinants of Public Health. National Yang Ming 16.128:95–104. De Clercq B.53:562–72. et al. Reprod Health. Adolescence and the social determinants University. Am J Obstet Gynecol. Handbook of religion 3. Taiwan. Factors associated with pregnancy among adolescents in low-income and lower middle-income countries: a systematic Authors’ contributions review. 2012. BMR is PhD candidate in International Health Program. Does neighbourhood social capital Author details aid in levelling the social gradient in the health and well-being of children 1 International Health Program. among adolescent mothers: a World Health Organization multicountry 31. Income inequality is associated with Funding adolescent fertility in Brazil: a longitudinal multilevel analysis of 5. National Yang-Ming University. Stevens V. social capital: developing an understanding of the causal pathway. Social ecological predictors of repeat adolescent study. 2002. 15.59:1117–26. The effects of social structure and Ethics approval and consent to participate social capital on changes in smoking status from 8th to 9th grade: results of This study was also approved by the Institutional Review Board (IRB) of the Child and Adolescent Behaviors in Long-term Evolution (CABLE) study. Stephens C. Chaix B.59:729–36. 11.69:918–24. Adolescent sexual behavior and reproductive outcomes in Central America: trends over the past two decades. health-a multilevel analysis. Health-related behavior as a mechanism 6. Encuesta nicaraguense de sexual behaviors: a comparative analysis. Int Authors’ information Perspect Sex Reprod Health. Competing interests 20. Taipei 112. influence of collective efficacy on adolescent childbearing. demografia y salud 2011/2012: informe preliminar. Health Place. Völker B. J Epidemiol Community Health. 2009.120:1622–30. 2nd ed. Viner RM. Soc Sci Med. 3Institute of 24. (ICDF) for scholarship provided for some of the authors. pregnancy.cepal. Institute of Public Health. 2006.34:289–300. Chen C-Y. Burkina Faso and is 17. 2002. New York: Oxford University Press. Perspect Sex Reprod Health. 2010. 148–71. Arch Pediatr Adolesc Med.39:39–47. familial. 22.43:727–42. J Epidemiol Community Health.565 This work was supported by National Health Research Institutes. et al. Merlo J. Chang H-Y. Ganchimeg T. Mmari K. friends-related and contextual predictors of We are grateful to Taiwan International Cooperation and Development Fund early sexual intercourse. J Sci Study statistics. J Adolesc Health. nulliparous adolescents in low‐and middle‐income countries: a multi‐country 29. References 27. p. London: Oxford Scholarship 4. Kawachi I. 2004. National Yang-Ming University. Maternal and perinatal outcomes among surveys: what social capital is it measuring? Health Place. Stansfeld SA. Miaoli 25. 2013. Ota E. 155. Kothari M.13:1. Blum R. Understanding the religion-physical health study. Study supervision: CYC. Zhunan. Hispanic concentration and the conditional County 350. Wilkinson R. Burkina Faso. Pradhan R. PO and OA are PhD candidates in International Health 18. Soc Sci Med. 2006. BMC Public Health.15:103.125:163–72. Lion KC. DC: Alan Guttmacher Institute. et al. Taiwan. Teen births. 2015. Social support and social cohesion. Available at: http://interwp. et al. et al. review. Gold R. Drafting of the health in developing countries: a structured literature review. Measurement of social capital 1. Taipei. Linong Street. Yang M. Chiavegatto Filho AD. 2015. Québec. Issues Brief. 4Center of Neuropsychiatric Geneva: World Health Organization.160:925–30. National Health Research Institutes. 2007. Pregnancy and childbirth outcomes Online. Wang S. Speizer IS. Samandari G. Finch BK. Cohen D. We are also indebted 9. PO. 2015. Glozier N. sectional study.35:91– All authors read and approved the final manuscript. 2013. 2. Denny S. 12. Carson VB. Prev Med. King D. DHS comparative reports 29. Ozer EM. et al. 13. Taipei. editors.17:606–17.38:556–9. Mmari K. Wynter K. In: Koenig H. Kennedy B. Adverse maternal and perinatal and health. J Adolesc Data supporting this study can be found upon request and permission of Health. 2010. Carballo PL. King D. Maternal-perinatal morbidity and in relation to health in low and middle income countries (LMIC): a systematic mortality associated with adolescent pregnancy in Latin America: cross. 2011. 2012. Agampodi TC.Mendez Rojas et al. Belizán JM. IdAplicacion=1&idTema=1&idIndicador=31&idioma=i. et al. Ouagadougou. Université Laval. 2009.66:1174–84. Moffitt TE. The protective value of social capital against teen pregnancy: a state-level analysis.12:1. Smith C. Adolescent childbearing in Nicaragua: a quantitative to National Yang Ming University and its International Health Program (IHP) for assessment of associated factors. 579–99. and The authors declare that they have no competing interests. relationship. 2015. Wiemann CM. IB and CYC. editors. A review of risk and protective factors for adolescent Availability of data and materials sexual and reproductive health in developing countries: an update. Mori R. Soc Sci Med. Social determinants of health. Consent for publication 21. Int Perspect Sex Reprod Health. Statistical analysis: BMR.

Gregson S. J Prim Prev. Adolescent outcomes and opportunities in a Canadian province: looking at siblings and neighbors. Making democracy work: Civic traditions in modern Italy. Princeton: Princeton University Press. Berkman LF. 2004. Submit your next manuscript to BioMed Central and we will help you at every step: • We accept pre-submission inquiries • Our selector tool helps you to find the most relevant journal • We provide round the clock customer support • Convenient online submission • Thorough peer review • Inclusion in PubMed and all major indexing services • Maximum visibility for your research Submit your manuscript at www. Community group participation: can it help young women to avoid HIV? An exploratory study of social capital and school education in rural Zimbabwe. Ksobiech K.14:1. Marín FM. 52. 44. Gender and social differences in adolescent sexuality and reproduction in Nicaragua. social. Wiemann CM. 2008. Residential mobility in childhood and health outcomes: a systematic review. Soc Sci Med. 42. Religion and volunteering in context disentangling the contextual effects of religion on voluntary behavior. Garrard-Burnett V. 2010. Adamczyk A. 43. 2000. Bose S. The effects of religious contextual norms. 2008. Chen K-H. Multilevel influences of school and family on alcohol-purchasing behaviors in school-aged children. Religious differentials in the sexual and reproductive behaviors of young women in the United States. J Marriage Fam. social networks. García J. Roos LL. Adolescent fertility and religion in Rio de Janeiro. Jones RK. 2011. Zelaya E. Community effects on pregnancy intention among cohabiting women in the Philippines: implications for maternal and child health. Soc Work Health Care. Darroch JE. MacGregor CA. 2008. 46. 54. Available at: http://linus.pdf. Community context and the prevalence of premarital sex among young women in Kenya and the Philippines: trends and differences from 1993 to 2003. Preventing adolescent pregnancy: biological. Walld R.16:512–22. Witt J. Brazil in the year 2000: the role of Protestantism. Putnam RD. et al. Jelleyman T. 2005. 2004. cultural. 49.71:616–25. Training field staff for DHS surveys.biomedcentral. BMC Public Health.66:721–38. 53. Soc Sci Res. 37. Ogali I.62:289–303. 2013. Cooperman A. 36. et al. Mushati P. Liu C-Y. Potter JE. e357. Lim C.31:3–14. Chiao C. Drug Alcohol Depend. Van Malderen C. Pires R.html. 55. Int J Equity Health. Khasakhala A. 45. Tenkorang EY. Soc Sci Med. Matern Child Health J. Pop Stud-J Demog. 2014.37:657–72. 2010. Crowder K. Haynie DL. et al. Social support.12:1. et al. J Adolesc Health. Are pregnant adolescents stigmatized by pregnancy? J Adolesc Health. 38. 39. 41. Spencer N. Accessed 20 Jul 2016.36:279–88. DHS. com/pubs/pdf/DHSM3/Training_Field_Staff_for_DHS_Surveys_Oct2009. 48. Pereira J. Chen C-Y. J Adolesc Health. Singh S. 2012.36:e1-352. et al.58:2119–32.21:39–46. Bell J. Health Place. Washington. 47. Residential mobility and the onset of adolescent sexual activity.77:747–79. 1997. structural constraints. 2005. Chiao C.nih. 40. Berenson AB. Decomposing Kenyan socio-economic inequalities in skilled birth attendance and measles immunization. 50.Mendez Rojas et al.62:584–92.nci. and personal religiosity on abortion decisions. et al. BMC Public Health (2016) 16:991 Page 9 of 9 35. Religion in Latin America widespread change in a historically Catholic region.16:1293–303. Am Sociol Rev. Do residential conditions explain the relationship between living arrangements and adolescent behavior? J Marriage Fam. A multi-level model of condom use among male and female upper primary school students in Nyanza. Teachman J. 1994. Araújo-Pedrosa A. 2005.67:499–514. 2014. D. NIH. South SJ. 2012. Available at: https://dhsprogram.35:239– . Terceira N. 51.C: Pew Research Center. Leonardi R. McKinnon S. Sahgal survival. Yi C-C. social cohesion and health.114:127–33. Accessed 20 Jul 2016. 2014. Nanetti RY. Stratified Design with Prospective Analysis Plan and Time-to-Event Endpoint. Maticka-Tyndale E. J Epidemiol Community Health. Rickert VI. Kenya. and political influences on age at first sexual intercourse. 56.