In this population-based cohort study of children from the Brazilian Amazon, socioeconomic background was positively related to linear growth during the school-aged years, whereas maternal height and child birth weight and length were associated with height up until age 10 y.
Previous prospective studies in developed and developing countries have shown associations between socioeconomic variables and attained height. The 1958 British birth cohort found that manual social class, family size, and household crowding were inversely related to considerable differences in height at ages 7 y, 11 y, 16 y, and 33 y [21
]. Although in past decades these characteristics affected linear growth throughout childhood, results from a more recent generation of British children have shown that socioeconomic disparities now have a major impact on birth length – an indication that, in a high-income setting, socioeconomic position expresses its effects on height mostly through mechanisms before birth rather than during childhood [22
Among low- to middle-income settings, nationally representative studies performed in India and Thailand have concluded that linear growth retardation is disproportionately concentrated among children from poor households [23
], but few studies have been conducted in cohorts followed for several years. Our findings indicate that the socioeconomic background is an important predictor of linear growth in this population of pre-school and school-aged children from the Brazilian Amazon. Interventions to ameliorate poverty could have positive effects on linear growth. A conditional cash transfer program in Mexico was found to enhance linear growth among infants by approximately 0.40 Z after 2 year’s implementation [25
]. It is unknown whether this effect can be sustained through school age. Four Brazilian cross-sectional national household surveys performed over a 33-year period showed a steep decline in the overall prevalence of stunting among children aged <5 y due to economic growth coupled with equity-oriented public policies and improvements in the population’s purchasing power, maternal education, sanitation, and access to health care [26
]. These surveys, however, did not include older children, nor were they representative of Amazonian populations.
Our results are consistent with the literature regarding the constant and positive association of maternal height with child’s linear growth. An analysis of 109 cross-sectional demographic and health surveys in 54 low- to middle-income countries confirmed that maternal stature is inversely associated with the likelihood of stunting of offspring up until age 5 y [27
]. Among these Brazilian Amazon children, and as reported from Great Britain [21
] and Pelotas [8
] birth cohorts, maternal nutrition may represent the combined effects of genetics and early-life environmental factors, reflecting the intergenerational transfer of both socioeconomic conditions and biologic mechanisms that have consequences for child health. For example, shorter mothers might provide an inadequate supply of nutrients to their fetuses and have narrower pelvises, thereby increasing risk for deliveries with complications [28
Concerning perinatal exposures, child’s birth weight and length were strong and positive determinants of HAZ throughout school age, in agreement with previous studies [6
]. Although birth weight is in some ways conditioned to maternal height [21
], it is noteworthy that the association of birth weight with HAZ was virtually unchanged after controlling for socioeconomic and maternal characteristics, suggesting that the influence of birth weight is independent of maternal stature. Consistent with this notion, a cohort of Belgian monozygotic twin girls (allowing control for genetic and maternal factors) found that the twin who was at least 5% heavier at birth was also taller as an adult [30
In our study, early introduction of cow’s milk was not significantly associated with HAZ. Only a few previous longitudinal investigations have reported long-term associations between infant feeding practices and anthropometric outcomes [8
]. Evidence linking untimely introduction of cow’s milk with diseases such as type 1 diabetes [32
] suggests that complementary feeding should provide appropriate foods in addition to breast milk at around age 6 mo [33
] to ensure satisfactory nutritional status during infancy.
The present findings should be considered in light of the limitations and strengths of our study. Although this was a population-based study, sample size was relatively small. Because of the high mobility of Acrelândia’s residents, mostly driven by job offers, the follow-up rate was 55% (256 of 468 children who participated at baseline). However, children included in the analyses were not statistically different from those who were not included with respect to sex, age, length/height, and the socioeconomic, maternal, and child characteristics at baseline. Another limitation is that we lacked information on child’s father. Birth weight and length were obtained from child health cards rather than through direct measurement by the research team. Nonetheless, there is evidence that these birth weight records have high validity in Brazil [35
]. There are several strengths to the study, including its longitudinal design, the extended follow-up period, the large number of determinants examined, and the fact that a dropout analysis involving baseline determinants showed no significant differences between children lost to follow-up and those who stayed in the cohort. Furthermore, our results were based on direct and standardized length/height measurements for both children and their mothers.