The association of neighbourhood and individual social capital with consistent self-rated health: a longitudinal study in Brazilian pregnant and postpartum women

Social capital are features of social structure, such as interpersonal trust, networks and norms of mutual aid and reciprocity, which act as resources for individuals and facilitate cooperation and collective action [1]. In general, social capital can be conceptualized at both individual and contextual levels. Individual social capital is defined in terms of resources and support that are embedded within individual’s social network [2]. Contextual social capital emphasizes the resources that can be drawn upon by individuals to pursue collective aims by being interconnected. Contextual social capital is also referred to as collective social capital, and has been measured as neighbourhoods at country, state and local community levels [2]. Neighbourhood social capital is related to the relationships between social groups and their neighbourhoods and is largely based on day-to-day interaction between neighbours [1]. Neighbourhoods include social and environmental structures that not only include networks themselves, but also shared norms and mutual trust that facilitate cooperation for mutual benefit [3].

Cultural context and social rules influence reciprocity and mutual cooperation, and therefore affect the level of social capital. A report of the Brazilian Commission on Social Determinants of Health found that social trust is extremely poor among Brazilians [4]. Citizens feel that politicians in Brazil are not very trustworthy [5]. Trust in others is important to sustain collective feelings [6]. However, the low sense of effectiveness of politicians is reflected in the low level of social participation of younger Brazilians [6]. In Brazil, neighborhoods with low levels of social trust and low social control have high violence rates [7]. The increase in homicide rates has a strong social impact regarding neighbourhood security. Fear and insecurity are associated with feelings of impunity and indifference in the urban population as a result of violent crimes [7]. In addition, there is pessimism about safety and violence [8].

Social capital and its contextual dimensions, social trust, reciprocity, neighbourhood safety, neighbourhood support, social control, empowerment and political efficacy, has been associated with a large number of health outcomes; poor mental health [9, 10], infant well-being [11], mortality [12, 13], oral conditions [14‐17], respiratory diseases [10], coronary diseases [18] and teenage pregnancy [19]. Social capital may play an important role in self-rated health (SRH) [20‐24] as SRH is highly correlated with objective health measures [25, 26]. Multilevel studies have shown an association between contextual social capital and SRH [27, 28], although there is some inconsistency in the measurement of social capital. The studies reviewed by Engstöm et al.[28] reported that lack of contextual social capital (civic trust; political trust; civic participation; and/or political participation), was related to poor SRH in adults and young adolescents. Eriksson et al.[29], who investigated the associations between collective social capital and SRH in men and women from Sweden, showed that women living in very high social capital neighbourhoods were more likely to rate their health as good or fair compared to women living in neighbourhoods with very low social capital [29]. Different aspects of the neighbourhood environment are associated with people's SRH [20, 21, 23, 30‐33]. Neighbourhood safety and political participation, dimensions of neighbourhood social capital, are related to better SRH in women [31]. Furthermore, women living in states with low levels of women’s political participation were more likely to rate their health as poor [32].

The social environment related to neighbourhoods is particularly important for women's health [31, 34, 35], and has also been considered as a determinant of health during the gestational period [36, 37]. Place of residence [38], neighbourhood poverty and deprivation [39‐42], community participation [37] and neighbourhood crime rates [43] accounted for different rates of preterm birth, low birth weight, inappropriate weight gain [44], gestational hypertension [45] and miscarriage/perinatal death [39]. Moreover, social capital was associated with health-related behaviours during pregnancy [45] and the use of prenatal care [46]. However, studies on the influence of neighbourhood environment on pregnant and postpartum women’s health, including those focusing on social capital, concentrated on clinical outcomes of pregnancy [37, 44, 47]. No study has explored the association between neighbourhood-level social capital and subjective measures of health, such as self-rated health, in pregnant women despite of its relevance to both maternal and child health [28, 48]. SRH has been used reliably in studies involving proximal causes of diseases in pregnant women and up to one year postpartum [28, 48‐50].

Although environmental characteristics embedded in the neighbourhood structure may be a potential determinant of subjective measures of health, social support and social networks, considered in this study as individual-level social capital [2, 3, 51, 52], have been cited as risk factors for maternal well-being during pregnancy and postpartum [53‐55]. Some observational and interventional studies showed that social support is linked to outcomes of pregnancy and health in the postpartum period [56]. Social support has been associated with low levels of stress after childbirth [57], and lack of social support was associated with poor SRH in the postpartum period [58, 59]. Lack of social support and ongoing physical and emotional problems were more important for SRH than socio-demographic background [60]. Social support and social networks have been also associated with reproductive outcomes. Inadequate social networks and weak social support affected intrauterine growth, especially among socially deprived women [61].

Changes in women’s SRH during and after pregnancy are expected because of body changes, physical recovery, adaptation to motherhood and adverse factors related to the infant [60]. Existing medical conditions such as gestational hypertension and gestational diabetes may get worse during pregnancy. The most common adverse condition during pregnancy is urinary tract infections [62]. It causes more symptoms than other conditions and may lead to maternal and neonatal morbidity and mortality [63]. Diseases of the lower urinary tract occurring during pregnancy lead to more symptoms than other conditions. Pregnant women with the abovementioned diseases usually experience malaise, indisposition, discomfort, polyuria and lower abdominal pain [64], which may influences their SRH.

Despite the changes during pregnancy, SRH remained 'very good' or 'good' in the majority of pregnant women during one year postpartum [49]. However, the factors contributing to the consistency in SRH are unknown. Among the factors suggested, individual social capital, such as social networks that reflect the way women receive social support, is considered an important proximal determinant of their health [65]. It has been argued that different forms of individual social capital are protective resources that exert a buffering effect against life stresses and can improve emotional well-being of mothers [66], and therefore, may have a positive effect on their SRH.

Despite the evidence showing a link between the place where women live and their health during pregnancy, the effect of neighbourhood-level social capital on SRH during pregnancy and for some time after childbirth is unknown and needs investigation. A multilevel longitudinal study was therefore conducted to gain a better understanding of what affects SRH in women during pregnancy and six months after delivery. It is important to distinguish the independent effects of contextual determinants (neighbourhood social capital) and individual determinants (social support and social networks) on SRH levels during pregnancy and during the postpartum period [49]. To address those research questions a theoretical framework encompassing a number of interlinked neighbourhood and individual-level characteristics was developed and employed in the present research (Figure 1). The objective of the study was to investigate the association of neighbourhood and individual social capital with self-rated health in women between the first trimester of pregnancy and six months postpartum.

The comparisons of demographic, socioeconomic and housing conditions characteristics between SRH groups are presented in Table 1. The study women were predominantly head of family (66.2%), without paid work (59.4%) and living in adequate housing conditions. Of the sample, 41.3% had schooling for more than 9 years, 85.7% had a family income more than 1 minimal wage and 70.8% were married. Most participants have 2 or 3 children, were Brown ethnicity and aged between 20 and 30 years (Table 1). Regarding socioeconomic covariates, the proportion of women who reported as head of family, living in poor housing conditions and with low schooling were statistically higher in the Poor SRH group (p ≤ 0.01). Bivariate analysis also showed the association between number of children in family after pregnancy and Brown and Black ethnicity with poor SRH (p ≤ 0.01).

The comparison of health-related behaviours and self-reported diseases between SRH groups is presented in Table 2. Overall, the participants in both SRH groups reported that they consumed lower levels of alcohol and smoked less than before pregnancy. Whereas the prevalence of diabetes and hypertension before pregnancy was low (1.2% and 6.7%, respectively), urinary infection during pregnancy was common (45.5%). Self-reported diseases were higher in women with poor SRH (p ≤ 0.001).

There were statistically significant differences of individual social capital variables between the SRH groups. The mean scores of all social support dimensions were significantly higher in the Good SRH group compared to the Poor SRH group (p < 0.001). Women with poor SRH had less friendship social networks than those with good SRH. There was a borderline association between family social networks and Good SRH group (p = 0.063). Women in the Good SRH group reported significantly more participation in associations or social groups (p ≤ 0.05) (Table 3).

In the multivariate analysis four statistical models were developed to test the association of neighbourhood and individual social capital with SRH adjusted for covariates (Table 4). In Model 1, neighbourhood social capital and social class were adjusted for each other. Low neighbourhood social capital was statistically associated with poor SRH [OR 1.94 (95% CI: 1.02-3.88)]. Model 2 included individual social capital variables. Low neighbourhood social capital remained statistically associated with poor SRH [OR 2.10 (95% CI: 1.05-4.24)]. In addition, both individual social capital variables, social support (positive social interaction) [OR 0.78 (95% CI: 0.73-0.90)] and social networks (one or more friends) [OR 0.51 (95% CI: 0.32-0.80)] were inversely associated with poor SRH. In Model 3, neighbourhood-level variables were adjusted for individual socioeconomic and demographic covariates. Low neighbourhood social capital did not remain significantly associated with SRH. Apart from that, the statistical association between low social support and low social networks with poor SRH persisted. High social support was associated with 19% lower odds of poor SRH [OR 0.81 (95% CI: 0.73-0.90)] and social networks was associated with 40% lower odds of poor SRH [OR 0.60 (95% CI: 0.35-0.98)]. Of the socio-demographic covariates, only family income and age were not statistically associated with poor SRH.

In the final adjusted model (Model 4), neighbourhood-level variables were adjusted for all individual variables, including self-reported urinary infection. Neighbourhood-level variables were not associated with SRH and the association between low social support and low social networks with poor SRH persisted in this final model. Women with higher levels of positive social interaction and those with one or more friends had 18% [OR 0.82 (95% CI: 0.73-0.90)] and 39% [OR 0.61 (95% CI: 0.37-0.99)] lower chances of reporting poor SRH. Among socio-demographic variables, water supply outside the house [OR 1.85 (95% CI: 1.05-3.28)] and low schooling [OR 2.06 (95% CI: 1.15-3.71)] were also associated with poor SRH. Women from Black and Brown ethnicity were 2.11 and 2.02 times more likely to report poor SRH compared to White women. Women with 2 or 3 children and those with 4 or more children were 3.23 and 3.39 times more likely to report poor SRH compared to those with no or one other child. Women reporting urinary infection had a higher likelihood of reporting poor SRH [OR 2.11 (95% CI: 1.28-3.49)].

In additional analysis that included the 222 women who changed SRH status, individual social capital was the major differential characteristic among the four groups. The mean score of affectionate and emotional supports (dimensions of social support) was higher among women with good SRH, and reduced significantly among those who changed from good SRH to poor SRH. Conversely, women who had poor SRH at baseline and changed to good SRH showed higher social support than those who remained with poor SRH at both assessments. In addition, in comparison to those with good SRH at baseline and poor SRH at follow-up, those with good SRH on both occasions had more relatives. However, no statistical differences were found concerning individual social capital between groups who changed their SRH status from good to poor, or from poor to good SRH. Data are presented in Additional file 3.

The interaction term, composed of neighbourhood social capital and individual social capital, was not statistically significant when added to the Model 1, and was therefore removed.

Despite the claim that neighbourhood social capital may be a contextual influence on the health of residents [88], and that there was a stronger associations between contextual social capital and health in women compared to men [89, 90], this study found that neighbourhood social capital was more weakly associated with women’s SRH during pregnancy and up to 6 months after delivery than individual social capital and other socio-demographic characteristics. After adjustment for individual variables, no association was found between neighbourhood social capital and SRH. Therefore, findings of the present study do not support the hypothesis that neighbourhood social capital was related to degree of SRH during pregnancy and up to six months after childbirth.

Even though there is evidence of a moderate association between neighbourhood environment and health [10], some studies did not find an association between neighbourhood social capital and SRH [28, 90]. Others found a substantial reduction in the association after adjusting for individual variables [31, 91]. Findings from the Stockholm Public Health Cohort study revealed no or moderate associations between contextual social capital and SRH [28]; the presence or absence of such an association depended on how individual social capital was statistically analyzed. Similarly, the Caerphilly Health and Social Needs Survey found that the association between neighbourhood deprivation and SRH was substantially reduced after adjusting for individual socioeconomic status. In addition, the association was attenuated by perceptions of the neighbourhood and by housing problems [31]. In a study of 30 districts of Saskatchewan, Canada, the political and civic participation, proxies measures of social capital, were not related to SRH [92].

Our findings differ from other similar studies in Brazil that found an association between neighbourhood characteristics and SRH. In a cohort study involving public university workers, neighbourhood characteristics were associated with SRH after controlling for individual factors such as age, ethnicity, income, education, social class, home, and health conditions [24, 93]. However, a large proportion of SRH variability was at the individual level and only 4% was explained by neighbourhood characteristics, which was reduced to 1% when individual variables were included in the model. Cremonese et al.[94] also reported that SRH depends more on individual characteristics than the socio-demographic context of neighbourhoods. According to Borges et al.[22], social capital may play an important role on SRH although the pattern of association differs according to the specific dimensions of social capital employed. They showed that social trust and civic participation as well as bridging social capital and social support remained associated with SRH in a group of adolescents after adjustment for other social capital indicators and confounders.

A positive association between social capital and SRH has been frequently reported [20, 21, 23, 32, 33]. In most studies, cross-sectional and ecological study designs were employed and demographic characteristics of the samples varied considerably. Variations involved demographic groups - adolescents, adults, elderly and pregnant women, and differences related to the levels of analysis - nations, states/regions, cities, neighbourhoods, and local institutions [33]. In addition, the interpretation of evidence is affected by the heterogeneous ways in which social capital was conceived and measured. Most studies used different composite indexes to characterize contextual social inequalities such as Index of Relative Socioeconomic Disadvantage (IRSD), neighbourhood socioeconomic status, Human Development Index (HDI), census-based composite indexes and Gini Index. Moreover, neighbourhood environment has been measured using the number and quality of local services and amenities [10], or by subjective evaluations obtained directly from local residents about their concerns and perceptions of the neighbourhood (about people and place), such as proxy measures or dimensions of neighbourhood social capital - social control, social trust, norms of reciprocity, feelings of safety (neighbourhood security), political efficacy and empowerment [15, 16, 44]. Therefore, valid comparisons between our findings and those from previous studies are difficult.

Although there was no association between neighbourhood social capital and poor SRH in the present study, there was a positive association between social support and social network and SRH. Therefore, the latter hypothesis was confirmed. The lower the social support and social networks, the higher the likelihood of maintaining poor SRH during pregnancy and 6 months postpartum. That finding accords with previous findings [56, 60, 95].

The period between pregnancy and 6 months after childbirth is a unique one in most women’s life. Individual social capital appears to be important for women, especially during pregnancy and after delivery. The lack of social support constitutes an important risk factor for maternal well-being during pregnancy [53]. Having low social support during pregnancy may reflect insufficient emotional support from the partner, family and/or friends that increase maternal stress, anxiety and depression, which in turn are related to well-being and SRH of pregnant women [50]. Psychosocial factors are also considered risk factors for undesirable pregnancy outcomes such as low birth weight, prematurity and intrauterine growth retardation [96, 97]. Dissatisfaction with partner’s support increased the risk of poor SRH [48] and insufficient social support increased the risk of poor SRH in multiparas [60]. In the postpartum period, the lack of social support was the most consistent predictor of poor health outcomes [58]. In addition, social networks influenced health behaviours and lifestyle habits during pregnancy, including dietary habits and smoking [28, 54, 55, 98]. Even though smoking during pregnancy was not assessed in the present study, previous studies showed that current smoking did not affect the association between social capital and SRH [31].

The strong influence of social support and social networks on women’s health [31, 35] may be due to effective psychosocial resources, particularly social stability and social participation, that provide emotional and instrumental support [66]. During pregnancy and the months thereafter, the association between individual social capital and women’s SRH can be explained by the quality of personal and social resources [99]. As personal resources can be very limited compared to those from social ties, especially during pregnancy, social resources are necessary and should be accessible through strong social connections, which represents a sense of attachment or sharing one’s sentiments [99]. Moreover, high social participation was associated with lower odds of poor SRH in women after controlling confounders [100], probably because women have more ties with their neighbours than do men [101, 102] and are more able to create and maintain local social networks that connect families and communities [103]. They are more likely to spend more time in the neighbourhood where they live, carrying out domestic tasks, going to supermarkets or grocery shops nearby or taking care of children and their elderly relatives [31], as well as participating in local volunteer work and/or religious groups [104].

Neighbourhood social capital may be affected by aspects of individual social capital that are associated with health [28]. Strong ties, such as social support and social networks, seem to mediate the association between neighbourhood social capital and consistent SRH [26]. However, in this study the interaction term was not statistically significant, suggesting that for pregnant women, the combined effect of neighbourhood social capital and individual social capital was not associated with poor SRH. Moreover, neighbourhood social capital was less important for those with consistent SRH during pregnancy and six months postpartum than individual social capital.

Women whose SRH status changed during the period of the study were not included in the multilevel analysis. By excluding them from the analysis, the study population may have been conditioned based on an effect of the exposure. However, the possibility of a ‘collider-stratification bias’ [105], a form of selection bias, was rejected by running a model without using postpartum SRH alone as the outcome and including all women regardless of SRH during pregnancy (data not shown). The results were very similar, suggesting that selection bias did not occur.

Additional analysis showed that individual social capital differed between groups whose SRH status changed. In addition, the possible influence of individual social capital in women not changing from good to poor levels of SRH over a period of great stress and transition was not confirmed. Conversely, individual social capital was associated with consistent SRH. The findings indicate that it is unlikely that including women who changed their SRH status over time in the overall analysis would affect the results. The influence of individual and neighbourhood social capital on SRH in women whose pregnancies were aborted or miscarried should be considered in future studies.

The findings of this study have several important implications, but are not without limitations. Recruiting pregnant women from public antenatal care units might affect the generalizability of our findings resulting in a homogeneous sample concerning some socioeconomic characteristics. This may explain the lack of association between social class and SRH. The number of women per neighbourhood was slightly below the ideal and might have affected the power of the study [72]. Our study is limited to pregnant and postpartum women, and considering the specificity of the sample employed in the present study, our findings must be viewed with caution and must not be generalized.

A point to be taken into account when considering our findings is that urinary infection was prevalent in our sample (45.5%), and was statistically higher in women with poor SRH. The prevalence of urinary infection was slightly higher than in other studies [62, 106]. The presence of urinary infection was based on self-reports and subject to misreporting. The reported associations in the final model did not change significantly when the variable urinary infection was removed. Future studies should consider collecting data about complications such as ‘morning sickness’ in early pregnancy or backache, perineal problems during pregnancy and postpartum, and complications of caesarean section postnatally, to validate subjective measures of health.

Positive features of the study were the use of social support and social networks questionnaires with adequate psychometric properties for the Brazilian population. The neighbourhood social capital questionnaire had been used in Brazilian populations and had good reliability. The data collection was standardized and performed by trained interviewers reducing information bias. In addition, the response rate was high and losses to follow-up were low.

There are some points related to the social capital assessment that should to be considered in future studies. One is the measure per si. There is no consensus on how to measure social capital. Contextual social features have been routinely assessed using aggregated measures and indexes, like percentage of poverty or median of income, mainly those available in census databases [10]. In this study, neighbourhood social capital was assessed individually and then aggregated at area level. However, the characterization of places and neighbourhoods by interviewers was not done. Social epidemiologists need to assess specific characteristics of places and neighbourhoods through primary data collection (compositional characteristics), paying attention to resources embedded in social networks [26].

The present study enhances knowledge on social capital and health in several ways. Our findings indicate a relationship between individual social capital and SRH. In addition, multilevel analysis is an adequate statistical approach to evaluate simultaneously the role of neighbourhood-level and individual factors on health. Above all, more attention was given to low socioeconomic status women, where undesirable pregnancy outcomes rates remain relatively high in other low and middle income countries. Knowledge about how proximal and distal social determinants affect women’s health during pregnancy and the initial period of motherhood has important policy implications on the health and social areas. The identification on what level of social capital was more important to women´s health is relevant in developing interventions. Although social inequalities are viewed as a broad and unspecific determinants of health, our findings suggest that there are specific social risk factors during and after pregnancy that should be taken into account to improve both maternal and child health.

Low individual social capital during pregnancy, such as social support and social networks, independently influenced women who had consistently poor SRH whereas neighbourhood social capital did not affect their SRH during pregnancy and the postpartum period. From pregnancy and up to six months after delivery the effect of individual social capital better explained consistent SRH in women than did neighbourhood social capital.