Effects of sanitation on child growth in Serbian Roma communities

1 Introduction

Height is a highly inheritable trait, frequently used as an indicator of growth, nutrition, and social environment in earlier life [1]. A child’s growth potential is mainly determined by genetics; however, the level of its attainment is dependent on environmental conditions during the growth period, especially in resource poor settings, where inadequate nutrition and environmental hazards may influence growth more than genetic factors [2]. Improving infant and child nutrition may benefit childhood growth, but adequate dietary intake alone does not eliminate growth deficits altogether, suggesting necessary additional implementation of measures to reduce pathogen exposure, such as through WASH (safe water, improved sanitation, and hygiene) [3]. Exposure to poor sanitation-related enteric pathogens have been associated with poor growth outcomes in children, i.e., lower height-for-age Z score (HAZ) and greater stunting (HAZ less than two standard deviations [HAZ−2 SD]) [4,5,6]. HAZ is used to evaluate the nutritional status of children at a particular age, relative to the growth standards, allowing for comparisons across and within populations. HAZ measurement points to important information about the general health of children: decreased height for age is an important risk factor for illness during childhood and indicates long-term, cumulative insufficiencies of health or nutrition (stunting) [7]. Additionally, HAZ is positively correlated with survival, cognitive, communication, and motor development among young children throughout the HAZ range, without a distinct threshold effect or any other cut-point [8]. In turn, childhood impaired growth has been directly linked with adverse outcomes in later life [9].

Unimproved toilet facility, in particular, open defecation, and subsequent contamination of sources of drinking water, is a leading risk factor for infectious diseases and can lead to persistent diarrhea and enteric parasite infection in children under 5 years [10,11]. It is estimated that over one billion people worldwide, mostly in rural areas, defecate in the open (i.e., without using any toilet or latrine); in European low- and middle-income countries, 92% of the population use basic sanitation services; however, more than 300,000 people still practice open defecation, mainly in the countryside [11,12].

It is generally assumed that children who live without adequate sanitation do not grow and develop as well as children with access to adequate sanitation, with poor sanitation being a risk factor for stunting, iodine deficiency, and inadequate development. Empirical evidence on the linkages between household sanitation infrastructure/open defecation and child nutrition, however, has been mixed. Thus, studies have found negative effects of open defecation on child’s growth [4,10,13] but also reduced the burden of diarrheal diseases and less undernutrition after moving from open defecation to a fixed latrine [14]. In contrast, randomized clinical trials in low-and middle-income countries found no association between sanitation interventions and improved linear growth in children younger than 5 years [15,16]. Additionally, no effect of open defecation on child’s height and other anthropometric outcomes has been found in a number of studies [17,18]. Inconsistent results may arise due to methodological issues and/or unaccounted confounding factors, such as the phenotypic variation in parental condition (height), community/neighbors’ sanitation facilities, or because family size confounds the potentially differential effects that siblings can have on young children’s growth.

Therefore, the primary aim of this study is to estimate the association between household sanitation infrastructure (open defecation vs any kind of toilet) and child’s height (HAZ and stunting) among under 5 years Serbia Roma children.

Roma, a diverse population of South Asian origin, is the largest ethnic group in Europe with low levels of integration. Despite some positive changes with the launch of the Decade of Roma Inclusion 2005–2015 in Central and Southeastern Europe, the Roma remains a marginalized population in many countries, experiencing poverty, unemployment, poor education, poor quality housing, and poorer health than the majority of population [19]. In Serbia, the Roma officially makes up around 2.1% of the population; however, due to ethnic mimicry, estimates run between 400,000 and 500,000 – or between 5.8 and 6.5% of the total population. The demographic distribution for those that declare as Roma shows a very young population, with a large proportion being poorly educated albeit with high fertility [20]. At the same time, child mortality rates remain notably higher for the Roma compared with the national average [21]. In contrast to non-Roma, Roma children grow up in large families and often lag behind in development and nutrition, negatively influencing life outcomes and leading to an intergenerational cycle of poverty [22]. Furthermore, lower respiratory tract infections, intestinal infectious diseases, and viral diseases were found to be more common among Roma children than in the majority [23]. A limited number of studies have addressed the relationship between Serbian Roma anthropometrics and health and child outcomes. Roma children’s stunting has been found to be associated with maternal (low) care and (un)registration at birth but unrelated to socioeconomic status [24,25]. Still, some Roma children may face additional risks, as suggested by the height differentials [20].

Sanitation coverage has improved in Serbian Roma communities in recent decades, although a recent survey found that over 65% of Roma settlements lack access to the sewerage system [26]. In fact, a lack of sewage infrastructure is common in the Republic of Serbia as a whole: there are numerous problems pertaining to the administration and management of wastewater. Thus, around 30% of the cities do not have a sewerage system, while only 60% of the population is connected to the sewerage network [27]. Even the largest cities lack wastewater treatment plant facilities, such that industrial and municipal wastewater is being discharged directly into the rivers or the receiving water body, endangering not only aquatic life but potentially human health as well. This results in the degradation of not only surface water but also underground sources of drinking water.

Although data are available on the differences in basic physical infrastructure, including water, sanitation, and hygiene between Roma and non-Roma [28], no population-level study has examined whether the variation in height among Roma children can be explained by differences in sanitation infrastructure-type of toilet usage. This study used large surveys of data sets from UNICEF 2014 and 2019 Multiple Indicator Cluster Surveys (MICS 5 and 6) for Serbian Roma settlements, to examine whether the type of toilet was associated with HAZ and stunting across children aged 0–59 months.

2 Data and methodology

3 Results

This section presents descriptive statistics of the sample, differences in Roma children HAZ scores by the type of toilet use, regressions assessing the relationship between the type of toilet facility used and Roma children HAZ scores, and the relationship between the type of toilet and Roma children stunting, respectively, for children aged 0–24 and 25–59 months.

Roma children were on average 31 months old (SD = 17.28); children aged 0–24 months were on average 12 months old (SD = 7.02); and children aged 25–59 months were on average 43 months old (SD = 17.28).

More than 16% (499) of children lived in a household without any kind of toilet (no service – open defecation). There were more boys (52.3%) than girls in the sample; for children aged 0–24 months, most children were breastfed at the time of survey (92%), the birthweight was on average 3.17 kg (SD = 1.04), while 11% were born with a low birth weight. For the total sample, 23% of children were stunted (614 out of 2,683), while HAZ was −0.93 (SD = 1.68). For children aged 0–24 months, HAZ was −1.02 (1.73), and for children aged 25–59 months, it was −1.59 (1.74). Roma mothers were on average 25 (SD = 5.83) years old, more than one third were illiterate (37%), and the majority (67%) lived in urban areas. The number of siblings was 3.04 (SD = 1.67), and birth spacing was 27 months on average (SD = 1.36).

Table 1 summarizes the differences in demographics and children’s nutritional status reflected in HAZ, by the type of toilet use (any kind of toilet vs no facility/open defecation) for Roma children and their mothers.

Children living in households with any kind of toilet facility were on average taller (M = −0.90, SD = 1.67) than children in households without service (open defecation) (M = −1.59, SD = 1.74), and the difference was statistically significant t(2,672) = 4.13, p = 0.00, and the effect size was small (η ² = 0.01). Furthermore, maternal literacy was higher in households with any kind of sanitation/toilet facility (χ ²(1, n = 2,755) = 53.16, p = 0.00, and small size effect φ = 0.14), while open defecation was more common in rural than in urban areas ( χ ²(1, n = 3,174) = 44.98, p = 0.00, small size effect, φ = 0.15). The number of siblings was higher in households practicing open defecation (M = 3.94, SD = 2.26) than in households with any kind of toilet facility (M = 2.85, SD = 1.57), the difference being significant (t(119.49) = −5.16, p = 0.00, and small size effect η ² = 0.01).

Differences in birth weight, birth spacing, maternal age, breastfeeding practice, stunting, and gender in regard to the type of toilet facility used were not statistically significant.

Table 2 summarizes the results of multiple regressions assessing the relationship between the type of toilet facility used and Roma children individual-level HAZ, separately for children aged 0–24 and 25–59 months.

For children aged 0–24 months, the type of toilet facility was not statistically significant in predicting Roma children HAZ (p > 0.05). Instead, birthweight was shown to have a significant effect on Roma children’s HAZ. Children born with weight above 2.5 kg were, on average, taller for 0.19 SD compared with children born with weight below 2.5 kg (β = 0.19, 95% CI = 0.23–0.46, p = 0.00). In addition, girls were taller than boys (β = 0.02, 95% CI = −0.03 to 0.01, p = 0.00). Also, birth spacing (β = 0.06, 95% CI = 0.00–0.46, p = 0.05) and maternal age (β = 0.09, 95% CI = 0.10–0.53, p = 0.00) were positively associated with children’s height, while child’s age (β = −0.13, 95% CI = −0.048 to −0.017, p = 0.00) showed a negative association with HAZ.

For children aged 25–59 months, the type of toilet facility, birth spacing, area, and the number of siblings residing in the same household were significantly associated with height. Children living in households with any kind of toilet facility were 0.13 SD taller than children living in homes without access to a toilet (β = 0.13, 95% CI = 0.01–0.03, p = 0.00). Children living in urban areas were, on average, taller for 0.08 SD compared with their rural counterparts (β = 0.08, 95% CI = 0.01–0.52, p = 0.04). Birth spacing was positively associated with children’s height (β = 0.14, 95% CI = 0.22–0.71, p = 0.00), while the number of siblings was negatively associated with children’s height (β = −0.18, 95% CI = −0.65 to −0.13, p = 0.00).

Bootstrap sensitivity analysis confirmed the significance and stability of predictors for children aged 0–24 months: gender (B = 0.33, bias = 0.00, SE = 0.10, p = 0.001), child’s age (B = −0.04, bias = 0.00, SE = 0.01, p = 0.001), weight at birth (B = 0.85, bias = 0.004, SE = 0.12, p = 0.001), and birth spacing (B = 0.37, bias = 0.002, SE = 0.18, p = 0.003). The confidence interval and low bias confirmed the reliability of the effect of these variables on the child’s height.

For older children, bootstrap sensitivity analysis confirmed the significance and stability of predictors: type of toilet (B = 0.35, bias = 0.002, SE = 0.13, p = 0.01), area (B = 0.03, bias = 0.00, SE = 0.01, p = 0.001), birth spacing (B = −0.02, bias = 0.000, SE = 0.01, p = 0.02), and number of siblings (B = −0.16, bias = 0.001, SE = 0.04, p = 0.001). The confidence interval and low bias confirmed the reliability of the effect of these variables on the child’s height.

Table 3 presents a summary of the logistic regressions accounting for the relationship between the type of toilet and Roma children stunting, separately for children aged 0–24 and 25–59 months.

In children aged 0–24 months, those born with normal birthweight were less likely to be stunted than their counterparts (OR = 1.50; 95% CI = 1.06–2.13; p = 0.02). Also, children who were more widely spaced were less likely to be stunted than children who were more narrowly spaced (OR = 1.03; 95% CI = 1.01–1.06; p = 0.01). Bootstrap sensitivity analysis confirmed the significance and stability of predictors for children aged 0–24 months: birthweight (B = 0.038, bias = 0.001, SE = 0.012, p = 0.002) and birth spacing (B = 0.430, bias = 0.010, SE = 0.186, p = 0.019). The confidence interval and low bias confirmed the reliability of the effect of these variables on the child’s height.

For children aged 25–59 months, those living in households with access to any kind of toilet facility had lower chance of being stunted than children living in households without toilet (OR = 1.58; 95% CI = 1.07–2.34; p = 0.02). Children living in households with more siblings were more likely to be stunted than those with fewer siblings (OR = 0.51; 95% CI = 0.29–0.91; p = 0.02). For these children, bootstrap sensitivity analysis confirmed the significance and stability of predictors: type of toilet (B = 0.502, bias = 0.006, SE = 0.221, p = 0.026) and number of siblings (B = −0.429, bias = −0.004, SE = 0.154, p = 0.008). The confidence interval and low bias confirmed the reliability of the effect of these variables on the child’s height.

4 Discussion

The present study assessed the relationship between the type of toilet used and Roma children’s height, and whether, after controlling several confounding factors, the variation in height among Roma children can be explained by differences in the sanitation infrastructure – type of toilet usage. While previous studies yielded inconsistent results regarding the relationships between the type of toilet and children’s height [17,18], in this study, after adjusting for potential confounding factors, the results indicate that Roma children aged 25–59 months, living in households with no toilet service, faced a significant deficit in terms of height. These children were more likely to be stunted and had a significantly shorter stature than their counterparts in households with access to any kind of toilet, and the effect was aggravated if the children lived in rural areas. The results indicate that one-sixth of Roma children grew up in households with no toilet service and almost one-fourth were stunted, with a mean HAZ less than 0 (−0.93). However, in contrast to their older counterparts, younger Roma children, up to 24 months of age, were not affected by the type of toilet.

As height serves as an indicator of health in early life, for older Roma children, growing up in households with no toilet service and in rural areas can be associated with worse health outcomes. This pattern was also observed after comparing children in households with access to a toilet and those with no service, as the former were on average taller than the latter, and the difference in height was statistically significant. These findings are in consistent with other studies where the lack of proper facilities to deal with waste contributed to impaired growth [6,34,35,36]. Even with good personal sanitation habits, exposure to fecal pathogens may cause serious health effects [37]. That is, child health outcomes are affected not just by an individual household’s sanitation facilities but also by the rates of open defecation within one’s community [11]. Studies have found that in dense, impoverished areas with poor infrastructure, the risk of enteric infection may be the greatest [36]. Repeated exposure to fecal pathogens, particularly common in areas where open defecation is practiced, can cause nutrient loss through diarrhea and lead to undernutrition and growth stunting. In settlements where many of the households defecate in the open, children are 10% more likely to be stunted than children living in open defecation-free settlements [13]. MICS does not account for the community-level sanitation; however, open defecation was more common among rural Roma than in urban areas, suggesting that children in rural areas, especially those residing in households with no service, were likely exposed to adverse environmental risk factors with negative implications for health and growth. These findings are reflected in height differentials among the older Roma children: those living in households with no toilet service and in rural areas were more likely to be stunted and shorter than their counterparts. In other studies too, stunting was more prevalent among children in rural settings compared with children in urban settings [38].

By contrast, younger Roma children (0–24 months) bore no negative consequences of residing in households with no service: the type of toilet had no effect on their height. In other studies, the relationship between sanitation and growth outcomes was observed more often in children older than 2 years of age [3]. MICS does not account for infant and young children care/hygienic habits (i.e., wearing diapers, household’s disposal of child stools, etc.), but given their young age, it is likely that the very young children had less direct exposure to fecal pathogens than their older counterparts. Another possible explanation may lie in the protective effects of breastfeeding. Breastfeeding status (whether the child is still being breastfed) was not statistically significant in the present study; nevertheless, as the majority of younger Roma children were still breastfed at the time of the survey, breastfeeding may still have played its part as a protective factor in regard to children’s health and thus height. A large body of evidence points to the protective effects of breastfeeding against diarrhea incidence, prevalence, hospitalizations, diarrhea mortality, and all-cause mortality [39]. Roma mothers tend to invest heavily in their children through breastfeeding: the duration of breastfeeding is typically more than 12 months on average and the practice of breastfeeding remains an integral part of the Eastern European Roma mothers’ cultural identity [25].

Maternal literacy skills per se, as a proxy for socioeconomic status (SES), had no influence on Roma children’s height outcomes. However, poor sanitation/open defecation is closely associated with poor socioeconomic position [10]: mothers of Roma children residing in households with no service had poorer literacy skills than their counterparts, implying differences in SES among Roma children. Numerous studies have found a strong association between maternal education and child health, associated with differences in parenting behaviors in nutrition and health promotion, with poverty strongly connected with undernutrition and its risk factors [38]. No statistical significant difference was found for stunting between children in households without service and their counterparts; however, HAZ was greater among children residing in households with any kind of toilet, and the difference in height was statistically significant.

Other factors associated with Roma children’s growth outcomes include different maternal and child characteristics. Thus, maternal investment (weight at birth), reproductive behavior (maternal parity, i.e., number of siblings and birth spacing), maternal age, and child’s age and sex, additionally accounted for children’s growth.

Of the controlled variables for the younger Roma children, weight at birth emerged as the strongest predictor of HAZ and stunting. Accordingly, in this study, as in many other studies, weight at birth appeared as one of the main specific correlates of child growth in regard to height [19,40]. Birthweight is dependent on the maternal stature, reproductive behavior, and environmental constraints and may be the largest contributory factor for later growth during early life [33].

HAZ of the younger Roma children was further influenced by the child’s sex and age and maternal age. In many settings, due to the combination of both biological and social mechanisms, growth faltering in children under 5 years is more likely to affect boys than girls [41]. Thus, Roma girls were more likely to be taller than Roma boys, suggesting that boys might be more susceptible to nutritional inequalities than girls of the same age. This pattern is consistent with previous findings, possibly as a consequence of parents favoring specific children due to greater fitness payoffs [19,42].

Both the maternal and child age influenced younger children’s HAZ, albeit in opposite ways. Increasing maternal age was associated with a taller stature of their children. There is no clear explanation for the observed better growth outcomes with increasing maternal age: not only epigenetic changes but also pre- and post-natal maternal age-related behaviors may play a role in the association [43]. Variations in the child-rearing environment also affect the childhood growth. Thus, generally, older mothers tend to invest more in their children, as they are less likely to have additional children such that the investment is focused on the children they already have, resulting in better child outcomes [44]. In contrast to maternal age, child’s age was inversely associated with HAZ: older Roma children lagged behind in height in regard to their age reference group compared with younger children. Additionally, other findings suggest a decline in the mean HAZ at early stages of life. In many low-and-middle income countries, falloffs in mean HAZ with age may be due to a downward shift of the entire HAZ distribution, suggesting that children across the HAZ range may experience slower growth when compared to international standards, due to nutritional and infectious conditions [45].

In all the children surveyed, birth spacing was positively associated with HAZ, while in the younger group those more widely spaced were less likely to be stunted than the more narrowly spaced children. The results are comparable to other studies and suggest an increased height and lower chances of stunting among the more widely spaced children relative to children who were more narrowly spaced [46]. Studies have found that short (<18 months) and long (>59 months) birth intervals are associated with increased risk of poor child outcomes: children born either at short or long birth intervals are more likely to be preterm, small for gestational age, and have low birthweight accompanied by a greater risk for stunting [32,47]. These effects appear to be stronger in low- and middle-income countries than those at higher levels of development. Roma women’s reproductive and parental behaviors are shaped by the traditional encouragement of early marriage as well as high fertility: women start reproducing at an optimum age and continue having relatively short-spaced children in their most fertile years, thus minimizing potential reproductive losses [48]. The poor outcomes of short birth spacing have often been explained by the physiological effects of maternal depletion, i.e., maternal inability to fully recover from one pregnancy before supporting the next one [47]. In addition, short birth spacing may increase the competition between siblings for parental financial resources and/or time and also higher disease transmission among closely spaced siblings [47]. These are likely reflected in the negative association between the sibship size and children’s HAZ and stunting for the older group of children: for Roma and other disadvantaged children living under resource-scarce settings, having numerous siblings may be particularly costly.

5 Conclusions

The relationship between the type of toilet and growth outcomes among Roma children varied with age. Having no toilet and living in rural area were associated with compromised child growth for Roma children aged 25–59 months, as they were shorter and more likely to be stunted than their counterparts in urban households and in households with any kind of toilet. There was no association between household sanitation infrastructure and height in children under 24 months of age; nevertheless, this finding should not undermine the benefits of sanitation overall. In recent decades, the Serbian government has introduced numerous affirmative measures aimed at improving the Roma’s situation in several areas, including housing and sanitation, education, healthcare, and facilitated access to financial social assistance. At the same time, significant progress is being made by water management instances in Serbia with construction of the sewage network and wastewater treatment plants in selected municipalities [27]. Despite these efforts, however, the prevalence of childhood stunting among the Roma remains high, with possible adverse health outcomes for the affected children: increased susceptibility to morbidity, mortality, and infectious diseases. To reduce the risks, access to and use of improved sanitation facilities with hygienic separation of human excreta from human contact is essential [11].

Additionally, the findings of this study point toward other contributing factors to growth outcomes in all Roma children: in the present sample of Roma children, child growth was shown to be affected by maternal age, child age and sex, maternal investment, and reproductive behaviors.

The present study had a number of limitations. The survey data were self-reported and thus susceptible to potential biases: self-reported data rely on the memory and recall of the respondents, which may be influenced by various factors, such as time, context, emotions, motivation, and social norms. Furthermore, being a secondary analysis of cross-sectional MICS datasets, temporal trends, directionality of associations, or inferences of causality between the measured variables could not be assessed. Thus, because the outcome and exposure variables were measured at the same time, it is not possible to establish causal relationships.

Other potential confounding factors, such as parental height and health and birthweight for the older children, were not collected. Maternal height, for instance, was found to influence offspring linear growth and other child outcomes such as mortality and anthropometric failure [49]. These and other influences should be examined in future research. In particular, future research should also focus on Roma cultural features, such as age-increasing women’s autonomy that may also show associations with child health and growth outcomes [50].

Despite these limitations, the present study is the first to examine the association between the type of toilet used and child growth outcomes using the Serbian Roma national dataset, thus contributing to the literature on child outcomes in poor ethnic minority populations and with findings that may have broader applicability beyond the Roma to other ethnic and social minorities.