Main

Previous longitudinal studies of postnatal growth in infants born with IUGR or those SGA have shown that in the majority they effectuate their catch-up growth (recognizable by an increase of height velocity) during the first 2 y of life. Nevertheless, a significantly reduced mean final height has been reported in young adult populations born SGA(1, 2), and the incidence of final short stature in SGA subjects at 16-18 y is 7-fold higher than that for a non-SGA group(1).

Because several studies have shown that human GH administration improves growth velocity of children with short stature related to IUGR(3, 4), it is important to investigate the natural history of growth in this population. SGA subjects form a heterogenous group of patients. The mechanism(s), the type (symmetric or not), and the intensity of IUGR vary from one case to another. Moreover the definition of the criteria at birth defining the disease (either birth weight or birth length) may vary from one report to another, which makes comparison difficult. Because several perinatal factors, such as auxologic data at birth and length of gestation, are important for the future development of a child and subsequent final height, prospective studies are necessary to establish factors predictive of individual growth and final height.

In this study we have evaluated the effects of the duration of gestation(from 37 to 42 wk), birth weight, length, and ponderal index on the subsequent final height of each subject. In addition, the role of etiologic factors of IUGR and target height has also been evaluated. Finally, the impact of the pubertal growth spurt on final height was studied in a subgroup of subjects.

METHODS

In 1971 a longitudinal population-based registry was initiated in Haguenau(France). The registry aimed to record information on all pregnancies, deliveries, and perinatal events for all births in the given area. As previously described(2), all term singleton subjects born SGA between 1971 and 1978 (n = 452) were selected for the study. SGA subjects were defined as those showing a birth weight and/or a birth length below the 3rd percentile according to the local standards. Term singleton subjects born during the same period with a birth weight AGA (birth weight between the 25th and the 75th percentile) served as control subjects(n = 451). They were selected from the registry as the first AGA subject born immediately after each SGA index case. Of the subjects selected, 58% took part in the study. A detailed description of the cohort population followed until final height was reached (SGA, n = 236; AGA,n = 281) has been previously made(2).

At the time of the study, height and weight were measured, and the medical status was assessed by a questionnaire asking for details of any previous and present health problems or developmental disorders and pharmacologic treatments. Subjects with growth-related disorders (SGA, n = 14; AGA, n = 3) were excluded from the analysis. As previously shown(2), we had to consider that some of the younger men had not achieved their final height as the mean height of men was significantly lower from 16.6 to 18 y than at 18-24.5 y. Only those of the younger male subjects who were able to be remeasured 2 y later (n = 26) were included in the analysis. This was not true for the women, all of whom had completed puberty with menarche having occurred at least 18 mo before the study.

The clinical characteristics at birth of the 213 SGA and 272 AGA subjects included in the study and studied at a mean age of 20.8 ± 2.0 y are indicated in Table 1. Among them, 155 SGA and 188 AGA subjects were also studied at a mean age of 6.1 ± 0.7 y. No significant differences were found for gestational age at birth or sex between the SGA and AGA groups. Among the SGA subjects, birth weight alone was below the 3rd percentile for 53.5% (n = 114), birth length alone was below the 3rd percentile for 22.5% (n = 48), and both birth weight and length were below the 3rd percentile for 24% (n = 51) of subjects.

Table 1 Characteristics of the 213 SGA and the 272 AGA subjects at birth
Full size table

To evaluate the degree of IUGR at birth, weight and length were expressed as a SDS and corrected for gestational age and sex according to the growth standards of Haguenau, as previously described(2). The birth weight was from -2 to -3 SDS and below -3 SDS for 86 and 12 subjects, respectively. The birth length was from -2 to -3 SDS and below -3 SDS for 75 and 19 subjects, respectively. The birth head circumference was from -2 to -3 SDS and below -3 SDS for 25 and 7 subjects, respectively. To evaluate the intrauterine nutritional state, the ponderal index was calculated as the ratio of birth weight (g) to the cube of the length (cm3) × 100 corrected for gestational age according to Miller's standard(5). Two groups were constituted according to a ponderal index < or ≥ 3rd percentile at birth. Among the SGA subjects, 48 (23%) presented a low ponderal index (<3rd percentile) at birth. Gestational age was determined from the date of the last menstrual period of the mother and by physical examination, confirmed by ultrasound measurements when available.

During pregnancy, risk factors associated with IUGR included pregnancy-induced hypertension, n = 58 (27%); smoking, n = 66 (31%); and a history of SGA in offspring, n = 58 (27%), with some subjects showing more than one factor. No significant differences were found for clinical characteristics at birth among these groups.

Each subject's parents were asked for their current heights and weights, and 97% of them were measured in the same standardized way as their child. Target height was calculated from mid parental heights adjusted for the sex of the child(6).

As shown in Table 2, the mean target height was lower in the SGA than in the AGA group, reaching a significant difference only for women (p < 0.0001). In an attempt to investigate whether the final height was influenced by puberty in the SGA population, we analyzed the available height of 343 subjects during the prepubertal period (from 5 to 8 y and from 5 to 9 y for female and male subjects, respectively) as compared with final height in the SGA (n = 155) and AGA (n = 188) groups. Child growth data for subjects born between 1971 and 1974 was available in the Haguenau registry, because their height was recorded at 6 y of age(7) and for other subjects from the child's health record booklet, which is systematically given at birth to parents. This booklet contains information recorded by physicians throughout childhood, including weight and height measurements. No difference was found between the subjects studied for height before and after puberty (n = 343), and the whole population studied for final height (n = 485) concerning gestational age at birth, birth weight and length, risk factors associated with IUGR during pregnancy, target height, and sex distribution of subjects between SGA and AGA groups.

Table 2 Clinical characteristics at age 20.8 ± 2.0 y in a population of 213 subjects born SGA and 272 AGA subjects
Full size table

Each subject's height was measured twice by the same pediatrician using the same wall-mounted stadiometer to the nearest 0.1 cm and the average value was used in the analysis. Weight was measured using a portable scale to the nearest 0.1 kg, and weight for height was assessed as body mass index [body mass index = weight (kg)/height (m2) (SD for chronologic age](8).

Measurements were also expressed in SDS, with the total young adult AGA population of this study serving as the reference population because the average final height for each sex was above the growth reference value currently used in France and, as previously mentioned, to avoid bias caused by the documented upward secular drift in height in this population(2). To analyze the difference between the final height and height before puberty, the growth standard currently used in France was used(9), as no local standard could be generated due to the small number of children studied at each age in this study.

Statistical analysis. Statistical analysis was performed using the SAS software (SAS Institute, Cary, NC). Results concerning descriptive statistics are expressed as the mean ± SD with a 95% CI. The two-tailed t test and the Wilcoxon rank sum test at an α level of 0.05 were used to assess the significance of differences between groups. Theχ2 test or the Fisher exact test was used for nominal variables. To study the evolution within each group, the paired t test and the Wilcoxon signed rank test were used to assess the within-group evolution.

Multiple linear regression models were used to assess the differences between SGA and AGA adjusted by target height (or mother's and father's height separately when necessary). First-order interaction was tested between SGA/AGA groups and sex. If significant, separate analyses were carried out for female and male subjects. Logistic regression analysis was used to explore relative odds ratios and to construct CIs for dichotomic dependent variables.

The study protocol was reviewed and approved by the faculty ethics committee. It was explained to both subject and parent, each of whom signed a written consent.

RESULTS

As shown in Table 2, the mean final height, weight, and head circumference were significantly lower in the SGA than in the AGA group (p < 0.0001). Surprisingly, the mean final heights of SGA subjects were similar to (in the case of male subjects) or even higher than(p < 0.005) (in the case of female subjects) (at + 1.46 cm; 95% CI from 0.5 to 2.4) the target height. To interpret these data, it is important to take into account the secular trend in height, which is well demonstrated in our study. When looking at the AGA population, the mean final height was significantly higher (p < 0.0001) than target height in both sexes (for male subjects +4.3 cm with 95% CI from 3.3 to 5.3; for female subjects +4.1 cm with 95% CI from 3.2 to 5.0).

When adjusted for target height, a significant deficit in final height(p < 0.0001) was found in SGA as compared with AGA subjects for both male subjects (-3.99 cm with 95% CI from -5.6 to -2.4) and female subjects (-3.64 cm with 95% CI from -5.0 to -2.3).

As shown in Table 3, the prepubertal population was studied at a mean age of 6.3 ± 0.9 y for male subjects (n = 144) and 5.9 ± 0.6 y for female subjects (n = 199). At this age, the mean height was significantly lower in the SGA than in the AGA population (p < 0.0001). The increase in height during puberty was analyzed by the delta height SDS from the final height SDS minus the prepubertal height SDS for each case (n = 343). No significant difference was found in the mean delta height SDS between SGA and AGA groups in either sex (-0.20 ± 1 SDS for SGA and -0.04 ± 1 SDS for AGA subjects; p = 0.12).

Table 3 Mean height at 6.1 ± 0.7 y and at 20.5± 1.9 y of age and delta height SDS (final height minus prepubertal height) in a population of 155 subjects born SGA and 188 AGA subjects
Full size table

Moreover, it is interesting to note that a significant increase in the mean body mass index SDS (for male subjects, p < 0.01; for female subjects, p < 0.002) occurred in the SGA population (but not in the AGA population) between the prepubertal period compared with the same subjects studied at 20 y of age (delta body mass index (SDS) = +0.58 ± 1.7 SDS for SGA (n = 144) and +0.16 ± 1.4 SDS for AGA(n = 162) subjects; p (adjusted for sex) = 0.01).

Prediction factors in the determination of final height in the whole studied population (SGA and AGA). According to the multiple regression analysis, the target height and the group studied (SGA or AGA) were the most important variables in the prediction at birth of individual final height. Using these measurements, 35% (male) and 40% (female) of the variability of individual final height (cm) could be explained at birth, the models being as follows.

For male subjects (overall F ratio = 54.642,p < 0.0001): individual final height (cm) = 53.06 - 3.998 group(as dichotomous variable, value 1 for SGA or value 0 for AGA) +0.72 target height (cm), where 8.6% of the individual final height variability was explained by the group (SGA or AGA) (p < 0.0001) and 26.6% by target height (p < 0.0001).

For female subjects (overall F ratio = 89.057,p < 0.0001): individual final height (cm) = 61.03 - 3.64 group(as dichotomous variable, value 1 for SGA or value 0 for AGA) + 0.65 target height (cm), where 7% of the individual final height variability was explained by the group (SGA or AGA) (p < 0.0001) and 33% by target height(p < 0.0001).

Prediction factors in the determination of final height and catch up growth for SGA subjects. For SGA subjects and according to a multiple stepwise linear regression model, three variables were found to explain 31% of the individual final height SDS variability (r2 = 0.31, overall F ratio = 30.639, p < 0.0001). They were mother's height SDS (p < 0.0001, 18.7% of the total variability of final height was explained by this variable), father's height SDS(explaining 11% of the total variability, p < 0.0001) and birth length (SDS) (explaining 1.6% of the total variability of final height,p < 0.03) (final height SDS = -0.375 + 0.36 mother's height SDS+0.31 father's height SDS +0.13 birth length SDS).

Moreover, using these variables, 58% of the variability of individual height gain SDS (delta height SDS as expressed by final height SDS minus birth length SDS) could also be predicted at birth (overall F ratio = 91.798, p = 0.0001, r2 = 0.58), the model being: height gain SDS = -0.34-0.87 birth length SDS (p = 0.0001,r2 = 0.41) + 0.36 mother's height SDS (p = 0.0001, r2 = 0.11) + father's height SDS (p = 0.0001, r2 = 0.06).

The effects of sex, gestational age (37-42 wk), birth weight (SDS), ponderal index at birth, risk factors during pregnancy associated with IUGR such as pregnancy-induced hypertension, smoking, and history of SGA in offspring were evaluated by multiple regression analysis and were not found to have a significant influence on individual final height or height gain SDS in the SGA population.

Prediction of final short stature for SGA subjects. Of the SGA subjects, 29 (13.6%) presented final short stature (defined as height below -2 SD score as compared with the AGA population taken as control). Target height SDS was found to be the only variable which could predict final short stature in SGA subjects (p < 0.0001). In a logistic regression model, the relative risk of final short stature was found to be higher in the case of a short stature mother (but not in that of a short stature father) (as defined by height < -1.5 SDS) as compared with a normal height mother (height ≥ 1.5 SDS) (RR = 5.4; 95% CI = 1.9-15.2). Moreover, no RR of final short stature was found whether SGA was defined according to birth weight (RR = 2.1; 95% CI= 0.70-7.1) or birth length (RR = 1.5; 95% CI = 0.66-3.4).

DISCUSSION

Several studies have attempted to define the adult height deficit in SGA subjects(1, 10–13). In most of them(10–13) the SGA subject groups analyzed were too small to allow the definition. Only one previous report retrospectively described the natural postnatal growth pattern in full-term small for gestational age infants(1). In this study we have demonstrated a significant reduction in the mean final height of an SGA population (13.6% of subjects presenting short final stature) compared with subjects of normal birth weight.

Methodologic advantages in this population based registry were the case control cohort, the high rate of follow up from birth to final height and the known parental heights. The upward secular drift in height, clearly shown by the finding that parental heights were smaller than those of the control population caused no bias to the results, because the control population was of the same age as the SGA population. Subjects who did not reach their final height and those showing well defined causes of growth disorders were excluded from the study.

Data concerning puberty and the pubertal growth spurt in children born SGA is very limited(14–16). No data are available concerning the pattern of puberty. Two retrospective studies on small numbers of short adult subjects born SGA suggest that adult height SDS was less than expected at prepubertal ages(14, 15). The Swedish report on SGA children identified from a population cohort suggests that these children attained puberty at a normal or early age(16). We have already shown in our SGA subjects selected from a population-based registry that puberty in girls occurs at a normal age as their mean age of menarche was comparable with that of the AGA population(2). The present study demonstrates that the difference between the final height SDS and the prepubertal height SDS is similar for SGA and AGA groups suggesting that final height in the SGA population is not influenced by puberty and that the pubertal growth spurt is normal in this population.

Final height in individuals varies considerably. Target height and group(SGA or AGA) were seen to be predictive at birth for 35% in the case of male subjects and 40% in the case of female subjects of the variability of individual final height, where 8% was related to the group (SGA or AGA).

Parental height and birth length SDS (but not birth weight SDS) have previously been used to predict the individual variability of height gain SDS during the first 2 y of life(17) and the magnitude of the whole postnatal catch-up growth in subjects born SGA(1). In contrast in another study, birth weight SDS was found to be more sensitive than was birth length in predicting catch-up growth during the first 2 y of life in full-term SGA infants(18). In no retrospective studies were risk factors during pregnancy associated with IUGR taken into consideration. We recently demonstrated in a prospective study that the length of gestation, when premature SGA subjects were included, is also predictive of height gain during the first 2 y of life(17). In the present study, parental height and birth length SDS were seen to be predictive of individual final height and height gain in subjects born SGA. No other variables such as sex, gestational age (from 37 to 42 wk), birth weight SDS, ponderal index at birth, and risk factors during pregnancy associated with IUGR, such as pregnancy-induced hypertension, smoking, or a history of SGA in offspring, were seen to be predictive. Thus the major influence of parental height on final height was evident in this study. However, in the SGA population, the influence of maternal height was greater than that of paternal height. Moreover, there was a 5.4 times greater risk of being short for SGA individuals from short stature mothers than for SGA individuals from normal stature mothers. Thus short stature mothers constitute a high risk population for developing IUGR during gestation and short final stature among SGA children.

As in other reports(1), no relative risk of final short stature was found when SGA was defined in terms of birth weight or birth length, showing that short final stature among SGA individuals is independent of whether birth weight or birth length is used in defining SGA. An inverse relationship between birth weight and the risk of developing impaired glucose tolerance, insulin resistance, hypertriglyceridemia, and hypertension(syndrome X) in middle age has been reported in retrospective cohort studies(19–23). Recently we demonstrated that elevated insulin and proinsulin levels are present in this studied cohort of young adults at 20 y of age born SGA in comparison with AGA subjects(2). These findings could be considered to be markers of early changes in insulin sensitivity in subjects born SGA, but in contrast with what has been observed later in life, blood pressure and lipid concentrations were normal in these non-obese SGA subjects. Our present data show that a significant increase in body mass index occurs between the ages of 6 and 20 y in this SGA population as compared with AGA subjects. This is a potential factor for insulin resistance. Longitudinal studies in specific age groups are needed to better describe relationships between insulin sensitivity and body composition changes during the pediatric age in subjects born SGA.