Main

It is known that some phases of growth are more important than others in their effect on final height. For instance, a child who is born short has a 7-fold greater chance of being a short adult (1, 2); the timing and duration of puberty is also crucial in determining final height (36). Another critical period is that between 6 and 18 mo of age in undeveloped countries. During this period, stunting, i.e. low growth velocity, is a common feature. As a result, >50% of children living in poor circumstances are short or stunted by 2 y of age (710). This explains much of the observed difference in final height between developed and developing communities (11), as well as the long-term trend in pubertal maturation (12, 13).

The ICP model divides the human growth process into three additive, partly superimposed phases: the infancy, childhood, and puberty phases (4, 1416). The infancy phase of the ICP model begins in the middle of gestation and tails off at about 3–4 y of life. This phase constantly decelerates with age. It has been claimed to represent the postnatal continuation of fetal growth; and can be represented by an exponential function. The second phase, the childhood phase, starts during the second half of the first postnatal year, slowly decelerates, and continues into puberty until growth ceases. This phase can be represented by a simple second-degree polynomial function. The puberty phase represents the additional growth, in addition to the childhood phase, induced by puberty and accelerates up to the age at PHV, then decelerates until growth ceases. This phase can effectively be expressed by a logistic function. The onset of the childhood phase of growth usually takes place between 6 and 12 mo of age in normal children. On the average, girls have an earlier onset of childhood phase than do boys. The mean and standard deviation of the age at childhood phase onset have been calculated to be 8.86 and 1.96 mo for boys, and 8.13 and 1.92 mo for girls, respectively (4). A delayed onset of the childhood phase of growth is assumed as the causal factor of stunting in early life (10, 16, 17). In two previous studies (10, 17), we have shown that stunting at 6 to 18 mo of age is highly related to a delay in the childhood phase of growth.

The extent to which stunting or delayed childhood onset of growth is prevalent in a healthy, normal population of a developed community and the long-term consequences for final height have not been outlined thus far. Those were also the objectives of this study.

METHODS

Subjects.

The data were obtained from a community-based, prospective, observational growth study, and the details have been described elsewhere (1). The study was conducted in all the schools in the city of Göteborg, Sweden, from April to November 1992. In total, 4487 children formed the study cohort; of these, 55.8% were born in Göteborg and 28.5%, in the outskirts of the city. Nearly 77% were born in 1974, 16.7% in 1973, 3% before 1973, and 3.5% in 1975. Boys and girls were equally represented.

In total, 1341 children were excluded from the present analysis; some children did not have complete birth records (n = 529), such as birth weight, birth length, or gestation age. Such data were often absent when children were born outside of Sweden (n = 242). Other excluding factors included: multiple births, gestational age beyond the range of 37–41 wk, and the presence of growth-related disorders (Table 1).

Table 1 Reasons for subjects to be excluded from the present study (one or more reasons might be present in some subjects)

To estimate the age at onset of the childhood phase and the age at PHV, at least three to four repeat length measurements are needed at approximately 1 y of age and at puberty, respectively (1, 4, 15). Because 714 subjects did not have these measurements taken at these ages, they were deleted from the analysis. As a result, 2432 children remained in the analysis, 1214 boys and 1218 girls.

Protocol.

All visits and examinations were made at the various schools. Health records for each child from birth to the last grade at school were retrieved. At each examination, height and weight were measured for each school child, and a questionnaire was completed that looked at history of diseases and any specific pharmacologic treatment. Meanwhile, the parents of each child were informed of the study. Data pertaining to birth and the perinatal period, such as size at birth, length of gestation, and any health problems, were obtained from the Swedish Birth Register at the Swedish National Board for Health and Welfare. Body measurements were taken in a standardized manner (1).

The age at onset of the childhood phase of growth was determined by visual inspection of the measured attained length or height (noninterpolated) and the change in growth velocity displayed on a computer-generated ICP-based growth chart (Fig. 1a) (1, 15). The determination of PHV was also identified from the computer-generated individual growth chart (noninterpolated) based on the growth velocity curve smoothed by a spline function. The age at the peak (maximum) of this curve during the adolescent period was identified by visual inspection and defined as the age at PHV for an individual child (Fig. 1b) (1, 4).

Figure 1
figure 1

a, Computer-generated ICP growth chart from birth to 3 y of age. Onset of the childhood phase of growth was identified by visual inspection of an abrupt increase in length growth velocity and its corresponding change in attained length growth pattern. b, Computer-generated ICP growth chart from 3 to 21 y of age. The age associated with the peak of the velocity curve during the adolescent period was identified as the age at PHV.

Because the children were not examined at exactly the same ages except for at birth, the individual length or height values were interpolated to the exact age at 0.5, 1, 1.5, 2, 5, 8, 12, 15, and 18 y. For each child, a higher-degree polynomial function was fitted to the length or height values between birth and 3 y of age and between 3 and 18 y of age.

The data from the interpolated length or height values were used to compute the corresponding SDS values. The software used to compute SDS was Epi Info, version 604b (Centers for Disease Control and Prevention, Atlanta, GA) using the NCHS reference values from birth to 18 y. We used this international reference, as recommended by the World Health Organization (18), to make the results comparable with other population study results.

Ethical Clearance.

The study was approved by the ethics committee of the Medical Faculty, University of Göteborg, Sweden, and by the Swedish Data Authorities. Written informed consent was obtained from the parents or those children who were >16 y of age.

Statistical Analysis.

The statistical analysis included two-sample t test, Wilcoxon rank test, Spearman correlation, and multiple linear regression. A p < 0.05 was regarded as statistically significant. Only two-tailed tests were used in the analysis. All statistical analyses and graphics were performed using SAS software for Windows, version 6.08 (SAS Institute Inc., Cary, NC).

RESULTS

There was no significant difference (p > 0.05) in the size at birth, i.e. birth length and birth weight, in both boys and girls between the infants included (n = 2432) and those excluded (n = 714) when adjusted for length of gestation. No significant difference (p > 0.05) was found in the mean final height in both sexes between included and excluded children. From birth to 18 y of age, the mean values for length or height in our series were significantly (p < 0.0001) higher than the NCHS mean reference values. The mean final height SDS was 0.44 and 0.61 for boys and girls, respectively, which corresponds to 2.9 cm and 3.7 cm above the NCHS reference mean.

The central tendency for both the age at onset of the childhood phase of growth and the age at PHV showed a significant difference between the two sexes (two-sample Wilcoxon rank test, p < 0.0001 in both cases). Girls had on average an earlier onset of the childhood phase of growth and experienced PHV at an earlier age also, with a median age at 9 mo and 11.5 y, respectively, compared with 10 mo and 13.5 y in boys (Fig. 2). Eighty-six percent of boys and 92.3% of girls had a childhood onset of growth at or before 12 mo of age.

Figure 2
figure 2

a, Cumulative frequency distribution of the age at onset of the childhood phase of growth for boys and girls. The median age was 9 mo for girls and 10 mo boys. b, Cumulative frequency distribution of the age at PHV for boys and girls. The median age was 11.5 y for girls and 13.5 y for boys.

There were no significant differences in the central tendency in the age at PHV between different childhood onset groups for both boys and girls (Kruskal-Wallis k-simple test, p > 0.1). The bivariate correlation coefficient between the age at onset of the childhood phase and the age at PHV was −0.01 (p > 0.7) for boys and 0.05 (p > 0.1) for girls.

Table 2 gives the gain in length or height SDS from 6 mo of age to final height for the various childhood onset groups for both sexes, both separately and combined. A linear regression was fitted to the data, and the 95% confidence interval of the mean is also included in Figure 3; for each month of a late childhood onset, final height values decreased by 0.09 SDS or 0.5 cm.

Table 2 Mean length/height SDS gain from 6 mo of age to 18 y for different childhood onset groups given by sex and for the two sexes pooled. The 95% confidence interval of the pooled mean is also included *p < 0.0001, Kruskal-Wallis k-sample test over all childhood onset groups. † CI, confidence interval.
Figure 3
figure 3

Simple linear regression between delta height SDS from 0.5 to 18 y of age and the age at onset of the childhood phase of growth. The 95% confidence interval for the mean is also included. Data for both sexes were pooled.

A multiple regression model with length at 6 mo, the age at onset of the childhood phase of growth, and the age at PHV as independent variables was applied to the series (Table 3). Both length at 6 mo of age and the age at onset of the childhood phase of growth were significantly (p < 0.0001) related to height at the various ages with fairly constant t values from about 2 y of age on; the total R2 was 0.33. The importance of the age at onset of the childhood phase of growth for final height was similar for both sexes, i.e. a 1-mo delay in childhood onset produced a final height reduction of about 0.5 cm. To adjust for any possible influences of the timing of puberty on the results given in Table 3, the age at PHV was also included in the multiple linear regression model. The age at PHV contributed significantly (p < 0.05) to the model from 5 y of age on. However, the slopes and t values for both length at 6 mo of age and for the age at childhood onset of growth were virtually the same as those found using the regression model without age at PHV (results not shown).

Table 3 Impact of age at onset of childhood phase of growth on linear growth at various ages analyzed by multiple linear regression model with adjustment for height (cm) at 6 mo of age* * Function: Height = α + β1 × length at 6 mo + β2 × childhood onset.p>

DISCUSSION

Linear growth retardation (stunting) is a commonly seen feature in infants in developing communities and explains much of the difference in mean body size between developed and developing countries (9, 1920). The difference in mean final height can be as much as 14–15 cm, and approximately 65–70% of this difference can be related to stunting occurring at 6–18 mo of life. However, this study shows for the first time that stunting is also prevalent in healthy children living in a developed community, i.e. Sweden. For the first time, we can also estimate the long-term consequences of stunting in early life on final height.

Adult or final height is dependent on a number of intrinsic and extrinsic factors. It is known that size at birth and at the time of puberty are significant predictors of final height (16). In this work, we have defined the importance of the age at onset of the childhood phase of growth for linear growth in childhood and for the final height achievement in a normal population. In addition, we have quantitatively isolated this effect after adjusting for the onset of puberty. For each month that the age of childhood onset is delayed, there is a 0.5-cm decrease in final height.

To our knowledge, this is the largest, longitudinal growth study ever conducted on individual data ranging from birth to final height. Analysis was based on full-term healthy children (n = 2432) for whom a longitudinal series of growth data during the first year of life and puberty was available. We could not document any statistical differences between the included and excluded children in terms of mean birth size or final height, which implies that the subjects we analyzed represent the population we intended to study.

The methodology for determining the age at onset of the childhood phase of growth has been evaluated and discussed in detail elsewhere (10, 15) and has been used in many studies during the past decade, such as ones investigating Turner's syndrome, cystic fibrosis, coeliac disease, congenital hip dislocation, Sotos syndrome (16), and infants living in poor environments (10, 17). The methodology is thus well established and easy to learn. In this study, one of the authors (Y.X.L.) estimated the age at childhood onset of growth. In a previous study of 425 children, however, two of the authors (Y.X.L., J.K.) made an interobserver validation with a high level of agreement. For these reasons, we believe that the age at childhood onset of growth has been accurately determined in the present study.

The previous normal range of childhood onset of growth (6–12 mo of age) has to be reconsidered, inasmuch as only 86% of the boys and 92.3% of the girls reached the onset before or at 12 mo of age (15). The number of children with an onset later than 12 mo of age was higher than the expected figure in this normal and healthy population, but not as many had delayed childhood onset as occurs in children with growth-related disorders, for instance, 36% of Turner's syndrome patients, 56% of children with coeliac disease, and 74% of children in an urban area of Lahore, Pakistan (16). The age at childhood onset of growth has been taken to represent the age at which growth hormone starts to influence linear growth significantly (4, 10, 1416).

Surprisingly, we could not document any significant correlation between the age at childhood onset of growth and the age at PHV; the Spearman correlation coefficient was −0.01 for boys and 0.05 for girls. The timing of puberty is known to be correlated to attained height before puberty; short children go into puberty later than tall children (4). Because the age at childhood onset of growth obviously has an impact on height in childhood, we could expect that late childhood onset of growth will lead to a delayed puberty. Possibly, subgroup analysis will show results in this direction, but this has not yet been performed. Our results show that a delayed childhood onset of growth, say, to 15 mo of age, is not compensated for by a delayed puberty. This is one reason why the impact of the age at childhood onset of growth on subsequent height remains into maturity.

We have previously shown that when there is a high incidence of stunting in early life, height at 2 y of age is strongly negatively correlated with the age at which childhood onset of growth occurs (10, 17). Unfortunately, as we did not have final height data available to compare with height in childhood or adulthood, we can only speculate that a delayed childhood onset of growth has a lifelong effect on height. However, the main finding of this study is that there is a clear association between the age at onset of the childhood phase of growth and subsequent height in a normal population as well. It is well recognized that the timing of puberty is an important determinant of final height, and in this work, we have shown another, as yet unrecognized, milestone of linear growth that is at least equally important, i.e. the age at childhood onset of growth.

Previously, it was known that early linear growth retardation is a common problem in developing countries, but we have shown that it is also present in a developed country, inasmuch as approximately 10% of the Swedish children showed a growth pattern in line with early linear growth retardation, i.e. a delayed childhood onset of growth (>12 mo of age). This is the first time we have been able to demonstrate this, as well as the lifelong consequences of early linear growth retardation. For every month of delay in the childhood onset of growth, final height is reduced by 0.5 cm; the delay can be by 12 mo or more in some children, thus producing a final height reduction of 6 cm or more (Fig. 3). The impact of the delay is linear and has an effect already at 6 mo of age, not just from 12 mo of age on, as we previously believed (15).

We conclude that early linear growth retardation is not only a problem in a developing community, but apparently is also present in developed communities. The long-term impact of linear growth retardation in early life is also clearly demonstrated here, with the potential for a reduction in final height by 6 cm or one SDS less than that expected in a healthy population. A more systematic follow-up of infants during the first 2 y of life in any community is needed to be able to prevent such a growth default.