Study sites

The present report is part of a longitudinal, prospective study using data from two Brazilian cohorts of singleton liveborns from Ribeirão Preto, SP (1994) and São Luís, MA (1997/1998) [26]. The study sites have different socioeconomic characteristics.

Ribeirão Preto (RP), located in the Southeast of the country, a wealthier and more industrialized region with a population of 457,653 inhabitants in 1994 and 551,312 in 2005, had a Municipal-Human Development Index (M-HDI) of 0.855, occupying 22^nd place in the national ranking in 2000. São Luís (SL) is the capital of the State of Maranhão, located in the Northeast of Brazil, one of the poorest regions in the country with a population of 801,895 inhabitants in 1997 and 978,822 in 2005. Its M-HDI was 0.778 in 2000, corresponding to 1112nd place in the national ranking [27].

Sampling

In RP, data were collected for all liveborns delivered during four months [28]. Hospital deliveries represented more than 99% of all births. Twin births and infants whose mothers did not reside in the city were excluded. Thus, the final sample consisted of 2,846 births. Losses represented fewer than 5% of the births.

In SL, the study was conducted from March 1997 to February 1998. A systematic sampling was performed, with proportional stratification according to the number of births at each of the 10 public and private maternity hospitals. A total of 2,541 hospital births were studied, including infants born to women residing in the city, liveborns, stillborns and single or multiple deliveries [29]. The sample was representative of the births in the city since hospital deliveries represent 96.3% of them. After multiple deliveries and stillborns were excluded, the final sample consisted of 2,443 births. Losses due to refusal or impossibility of locating the mother amounted to 5.8%.

For the follow-up at school age sampling was stratified by birth weight. Five groups were constituted: very low birth weight (VLBW, <1500 g), low birth weight (LBW, 1500 to 2499 g), insufficient birth weight (IBW, 2500 to 2999 g), and normal birth weight (NBW, 3000 to 4249 g). Children whose birth weights were at least two standard deviations above average were classified as high birth weight (HBW, ≥4250 g). All parents or persons responsible for children in the VLBW, LBW and HBW groups and one in every three in the IBW and NBW groups were invited to participate in the follow-up study. Thus, VLBW, LBW and HBW children were oversampled to increase the study power [27].

In RP, after exclusion of multiple births, stillborns and deaths during the first year of life, 1150 children were eligible for follow-up. A total of 790 children from the original birth cohort were evaluated (24 VLBW, 145 LBW, 174 IBW, 419 NBW, and 28 HBW) at age 9 to 11 years, in 2004/05, with a follow-up rate of 68.7%. In SL, after exclusion of multiple births, stillborns and deaths during the first year of life, 926 children were eligible for follow-up, with 673 children of the original cohort being followed up (5 VLBW, 76 LBW, 134 IBW, 439 NBW, and 19 HBW) at age 7 to 9 years, in 2005/06, representing a follow-up rate of 72.7%. Children whose birth length was missing were excluded. Thus, 1,402 schoolchildren remained in the analysis, 748 from RP and 654 from SL [27].

Sample size

A sample of about 700 children in each city has an 80% power to detect a 7% difference in height catch-up or catch-down growth (estimated at about 30%) between exposed and non-exposed groups, with a 5% probability of type I error.

Variables

The response variable was the difference between z-score for height at school age, based on the NCHS reference [30], and z-score for birth length, based on the Swedish reference [31]. It was classified using the definition proposed by Ong et al. (2000), who considered catch-up as a change in z-score≥0.67 and catch-down growth as a change in z-score≤−0.67.

The variable “adequate weight for gestational age/preterm birth” was created by combining preterm birth (gestational age <37 weeks, based on the date of the last normal menstrual period) and IUGR, defined on the basis of the birth weight ratio, which is the ratio between the newborn's weight and the mean weight for gestational age of the Williams et al. reference [32]; a birth weight ratio <0.85 was defined as IUGR [33]. This variable was categorized as follows: term newborns without IUGR (normal); preterm newborns without IUGR (preterm without IUGR); term newborns with IUGR (term with IUGR), and preterm newborns with IUGR (preterm with IUGR).

The remaining variables were: newborn's sex, maternal age (<20, 20 to 34 and ≥35 years), maternal schooling (0 to 4, 5 to 8, 9 to 11, and ≥12 years), parity (1, 2 to 4 and ≥5), maternal marital status (married, cohabiting, and no companion), family income in minimum wages (≤5, >5), and occupation of family head (non-manual, skilled and semi-skilled manual, and unskilled or unemployed). Neonatal ponderal index (PI) was calculated according to Rohrer's formula, PI = weight/length³×100. A child was considered “thin” when PI<10th percentile of the Lubchenco et al. reference [34], and “normal” otherwise.

Age groups were established as decimal ages according to the standardization recommended by Ross and Marfell-Jones [35], being 7 years for ages between 6.50 and 7.49 years, 8 years for ages between 7.50 and 8.49, and so on. A “missing” category was included for family income, because more than 10% of the families did not report income.

Statistical analyses

Mean z-scores for birth length and schoolchild's height were calculated. Associations between preterm birth/IUGR status and z-scores for birth length and height at school age were determined by analysis of variance (ANOVA). The chi-square (χ²) test and, when necessary, the Fisher exact test were used to assess differences in proportions.

The risk ratios of height catch-up and catch-down growth were obtained by multinomial logistic regression in separate models for each city. Since the follow-up sampling of schoolchildren was not equiprobabilistic, because LBW, VLBW and HBW children were oversampled estimates were corrected by weighting. The variables used for weighting were birth weight and preterm birth. Sampling stratification according to birth weight was also taken into account. In the tables and in the text, absolute frequencies are presented without weighting and percentages are weighted.

Ethics

The project was approved by the Research Ethics Committee of the Clinics Hospital, Faculty of Medicine of Ribeirão Preto, University of São Paulo (protocols 28/2004 and 10073/2009) and by the Research Ethics Committee of the University Hospital, Federal University of Maranhão (protocol 3104-476/2005). All parents or persons responsible for the children gave written informed consent to participate in the study.

Height growth in a middle-income country from birth to 7–11 years

Children from both cities had negative mean birth length z-scores. Thus, their mean values for birth length were below the mean values of the NCHS reference, indicating that in two different settings of a middle-income country, one more and the other less affluent, children face constraints to their growth. Preterm children without IUGR were the only exception. They showed positive mean values for birth length z-scores.

From birth to 7–11 years, although some groups caught-up with their peers, they remained showing negative mean height z-scores, indicating that constraints to their growth were alleviated but persisted through infancy and childhood. Children in the more affluent setting were able to present higher catch-up growth than those in the less affluent setting. As a result, children from the more developed setting of a middle-income country, although still presenting negative mean height z-scores at school age were closer to the mean height values of the NCHS reference than children from the less developed setting.

Risk factors for height catch-up growth

IUGR is a heterogeneous condition affecting both term and preterm newborns. However, causes, complications and prognosis of IUGR differ between term and preterm children [4], [5]. Thus it is important to look at height catch-up separately for term and preterm children with IUGR.

Itabashi et al. [36], in a cohort study of SGA (small for gestational age) term and preterm newborns, showed that approximately 90% of term infants had obtained height catch-up at 5 years. In our study, in both cities, term children with IUGR presented higher height catch-up growth than normal children. In addition, children from the more affluent setting seemed to present higher height catch-up growth compared to those in the less affluent city. Thus, our results were consistent in the two cities, indicating that in a middle-income country, term children with IUGR were able to present compensatory height growth and catch-up partially with normal children. However, their height mean values were still below those from their peers from developed countries at the same age.

Results were not so consistent for preterm children with IUGR. They had higher probability of height catch-up compared to normal children in RP only, the more affluent setting. With a similar definition of height catch-up, Darendeliler et al. [5], in Turkey, showed that SGA preterm children were smaller and lighter than AGA (adequate for gestational age) PT children, but had significant weight and height catch-up growth, reaching the same weight and height of AGA PT children at about 5 years of age. However, Brandt et al. [4], in Germany, showed that 54% of SGA preterm children did not show height catch-up. In addition, Itabashi et al. [36], in Japan, showed that only 74% of SGA preterm newborns with a gestational age of less than 32 weeks had obtained height catch-up at 5 years. The authors concluded that SGA children with less than 32 weeks of gestation are at higher risk not to achieve catch-up growth compared to SGA children with a gestational age of more than 32 weeks. In our study, PT children with IUGR were smaller at birth than PT children without IUGR when taking into account mean z-score for birth length. In RP, the more affluent city, although PT children with IUGR presented increased catch-up growth, this compensatory growth was not enough to allow them to catch up completely with their peers. In SL, the less affluent city, preterm children with IUGR did not show height catch-up growth. As a result, at school age in both cities PT children with IUGR remained smaller than normal children. Their height mean values were approximately −1 SD below those from their counterparts in developed countries at 7–11 years.

Other findings of prospective studies conducted in developed countries also indicate that, despite partial catch-up during the first years of life [3], [7], [9], children with IUGR continue to have a slightly smaller body size compared to their peers with no IUGR during infancy [23], [24]. According to Darendeliler et al. [5], AGA PT children with stable intrauterine growth react in a different manner after delivery and may be more sensitive to environmental factors than SGA PT children who, once freed from the intrauterine constraints to their growth, react differently, presenting an “exaggerated” catch-up, as long as they receive good care after birth. This mechanism may explain the increase in mean height z-score between birth and school age in preterm children with IUGR compared to preterm children without IUGR in RP, the more developed city. Our results suggest that term children with IUGR tend to experience height catch-up but preterm children with IUGR experience more difficulty in reaching their potential and perhaps more so in less developed settings.

Regarding parity, there is no remarkable evidence that the number of children does interfere with height catch-up or catch-down. Our data show that in RP children of primiparous mothers had higher risk of height catch-up, whereas in SL parity was not associated with catch-up growth. Guimarães et al. [37], observed that children whose mothers had given birth to four or more children had a 3.5 times higher chance of having short stature than children whose mothers had given birth to only one child. In the UK, infants of primiparous pregnancies were thin at birth but showed dramatic catch-up growth, and were taller than infants of non-primiparous pregnancies from 12 mo onwards. It has been shown that availability of postnatal nutrition may be affected by family size but perhaps other unknown factors may be involved to explain why first-born children presented higher catch-up growth than their counterparts [38].

Risk factors for height catch-down growth

In the present study, preterm children without IUGR presented higher risk of catch-down in both cities. This risk was slightly higher in SL, the less developed city. Darendelier et al. [5] showed that some AGA PT children presented height catch-down, with a decrease in height z-score midway through childhood in relation to birth length. The authors speculated that these AGA PT children were possibly already in a process of catch-down before birth and their postnatal growth may have been the continuation of their insufficient intrauterine growth. Alternately, as already mentioned, AGA PT children with stable intrauterine growth may be more sensitive to restrictive extrauterine environmental factors than SGA PT children. The authors concluded that AGA PT children may be at risk for impaired growth and recommended that they should be monitored as carefully as SGA children.

Children who were thin at birth showed higher risk of height catch-down at school age only in RP. Thinner children usually show catch up growth in weight, especially when they are growth-restricted [2], but in our analysis PI was adjusted for preterm birth and IUGR. So, one of the reasons why these children in the more developed city may show catch-down in height at school age is possibly because they are genetically short.

Among children younger than five years, influence of environmental factors on growth is much more important than that of genetic factors. The younger a child, the more dependent and vulnerable it is regarding the environment [39]. The observation that in SL, the less developed city, there was a slightly higher risk of height catch-down for PT children without IUGR suggests that better socioeconomic conditions facilitate height catch-up and reduce the risk of height catch-down. Although in the present study z-score distributions for birth weight and, more markedly, for height at school age were shifted towards lower values in both cities, in SL this left shift was even more intense at school age. Children from families of different socioeconomic levels have a different size, on average, at all ages, and high-income groups are always ahead along the course of maturity [39], [40]. This may explain why children born from low-income families presented higher risk of catch-down only in SL. Poverty, an inadequate diet and infections during childhood underlie unfavorable socioeconomic conditions [41]. In RP, these conditions may have been partially overcome even by low-income families due to a more equitable distribution of goods and services, including health services, in this city compared to SL.

In our study, sex interfered with growth only in SL, since girls had a higher risk of height catch-down than boys. However, no association between catch-down and sex was detected in RP. Regarding catch-up, Knops et al. [42], when studying factors associated with weight catch-up between 5 and 10 years of age, did not find difference between boys and girls.

Strengths and limitations

The present study permitted evaluation of height catch-up and catch-down growth, the latter being less studied, in two cities of contrasting socioeconomic conditions in a middle-income developing country. Studies of growth in developing countries are important because factors that influence growth failure during early ages seem to differ from those observed in developed countries [43]. The present investigation had the advantage of being a longitudinal, prospective population-based cohort study. The oversampling of groups more likely to present growth faltering permitted an increase in the power of detecting differences between groups. Selective losses occurred in the sample. In SL there was a selective loss of families of more privileged socioeconomic level, in contrast to what was observed in RP, where follow-up rates were higher for the better off.

Because mortality rate of preterm infants with IUGR was higher in SL than in RP [44] fewer children in this group were alive at school age to participate in the follow-up study in SL. In addition, oversampling of preterm children in SL was not as successful as in RP.

The difficulty in comparing the present results to those reported in the literature is that most studies have investigated weight catch-up and not height catch-up. In addition, different criteria have been used to assess catch-up. While some authors consider catch-up as an increase in weight or height higher than average, others consider catch-up as an increase in z-score higher than a defined cut-off point. The more traditional criterion considers catch-up as an increase in z-score for weight and height higher than 2SD [45]. The criterion proposed by Ong et al. [2], which was used in the present study, considers catch-up as an increase in z-score for weight or height higher than 0.67 SD.

Another difficulty is that few studies have looked at the association between PT/IUGR and height catch-down, since most studies focused on height catch-up.

We conclude that in both cities term children with IUGR presented higher height catch-up growth than normal children. Only in RP, the more developed city, children from primiparous mothers, thin children and preterm children with IUGR had a higher risk of presenting height catch-up.

In both cities, PT children without IUGR were at a higher risk of height catch-down. Only in SL, the less developed city, children from low-income families and females were at higher risk of height catch-down.