Longitudinal growth of stature in boys according to age and puberty: Prediction of adult stature from the age of 13 years

The purpose of this study was to cross‐validate and demonstrate how adult stature can be predicted in 13‐year‐old teenager's boys by using a new reference specific growth curve obtained from chronological age and maturity.


| INTRODUCTION
Pubertal growth consists of a phase of acceleration followed by a phase of deceleration and then the cessation of growth (Ferrandez et al. 2009;Prader, 1992). Longitudinal studies tend to present data as a single group or in three maturity groups corresponding to early, intermediate and late puberty (Hagg & Taranger, 1991;Largo & Prader, 1983;Tanner et al., 1976). Other authors have established percentile curves of stature and weight according to chronological age and growth peak (Cole, 2012;Rosario et al., 2011). Maturity stages have been defined from the age of peak velocity or also with sexual maturation (Ferrandez et al., 2009). Most longitudinal studies focus on stature and weight (Borghi et al., 2006;Buckler, 2012;Himes, 2006;Lee et al., 2008;Sorkin et al., 1999).
Several techniques that provide prediction of adult stature have been developed. Most of them require skeletal age as a predictor in the regression equations (Roche et al., 1975;Tanner et al., 1983). Other non-invasive techniques to predict adult stature without the use of skeletal maturity have been developed By Beunen and Malina method. The prediction of adult stature can be obtains with chronological age, current stature, sitting height, subscapular skinfold as predictors. (Beunen et al., 1997). Different predictive methods have been performed based on chronological age and anthropometric dimensions. The peak stature velocity (PHV) which is an indicator of biological maturation can be obtained from the velocity of leg length and sitting height (Mirwald et al., 2002;Sherar et al., 2005).
Different studies have focused on the prediction of adult stature. Recently, the Beunen-Malina Freitas method was developed in girls aged 12-15 years (Beunen et al., 2011). This method which does not involve radiations is regularly used on European populations. Biological maturation is an essential criterion to better describe the growth kinetics in longitudinal and not a transverse studies.
The purpose of the present study is to accurately predict the adult stature of adolescent boys from the age of 13 using a non-invasive method.
Secondary sexual characteristics associated with chronological age represent the variables that most influence the timing of the puberty. Secondary sexual characteristics are the qualitative criteria that allow us to describe the different pubertal stages. In our study, the maturity stage was based on the age of peak height velocity (APHV) which was the determining criterion to develop the new standards based on three groups each made up of individuals with advanced, standard and delayed puberty.

| METHODS
The main objective of this study is to obtain greater accuracy in the monitoring of individual stature growth. To achieve this, we developed new stature growth curves using chronological age and biological maturation from longitudinal data. These curves were also used to accurately predict the adult stature of teenagers from the age of 13 and a half.

| Sampling
The prediction of adult stature of young boys was established from the results of a longitudinal study within the framework of a National Research Agency program. In this program, we developed new growth curves using age and maturation from longitudinal data collected from 125 sedentary boys aged from 12 to 17.5 years. The teenagers in our study are mostly sedentary because they spend more than 2 h of screen time and less than an hour of physical activity per day. Less than 10% of adolescents engage in physical activity for 4 h per week outside of school time.
The essential longitudinal study to build these growth curves took place over a period five consecutive years. This survey was carried out during the school year in three secondary schools in the region of the town of Soissons, in the department of Aisne. Longitudinal data were collected twice every school year. So this reference sample consists of 1234 measures collected for a period of 5 years.

| Ethics
To ensure the sustainability of the data collection for a study of more than 5 years, all participants and parents gave their oral and written informed consent to participate. The permission of the medical council and the regional education authority of Aisne was also procured. Written informed consent was given in accordance with institutional Human investigation committee guidelines in accordance with the Declaration of Helsinki amended the October 2013, after informion about the procedures used in the experiments.
The parents of the boys of our sample arise from various social and occupational environment: employees (office and hospital workers, 30%); skilled workers (corporate administrative and commercial employees, 30%); and middle managers (technical and commercial executives, 40%). Furthermore, only 2% are immigrants coming from North Africa. So, our sample can be considered as being very homogeneous with a large majority of Caucasian boys. Respecting the inclusion criteria of individuals in good health and in a particular age group led to the undertaking of this study in schools in the Soissons area. The exclusion criterion concerned boys who had a stature for age in Z score, lower or higher than three standard deviations. Longitudinal data were collected twice every school year.
The data collection focused on stature, chronological age and biological maturation. Stature was taken to the nearest millimeter using procedures describes by Claessens from a GPM anthropological instruments-DSSH/, Zurich, Switzerland. Two measurements were taken and a third measurement was required if the two results differed by more than 4 mm. All measurements were performed by the same trained technician throughout the study.

| Pubertal stages
The maturity stage was based on the age of peak stature velocity (APHV). Knowing the APHV was the determining criterion to develop new standards based on three groups each made up of individuals with advanced, standard and delayed puberty. Moreover, an evaluation of secondary pubertal stages was also conducted using a simplification of the Tanner stages. These puberty stages were assessed based on facial and axillary hair pilosity as well as voice changes (Pineau, 2020) (Table 1). The degree of maturation was thus determined using 4 different stages. The stage below the age of puberty (ST1) (Stage 0 of Tanner); The pre-pubescent stage (ST2) (Stages 1 and 2 of Tanner); The para -pubescent stage (ST3) (Stage 3 of Tanner) and the pubescent stage (ST4) (Stages 4 and 5 of Tanner). When there was concordance between axillary and face pilosity, we retained the identified stage. In case of conflict between stages, we selected the higher level. In practice, the majority of stages concerned no more than two consecutive stages.
Three average longitudinal stature curves were drawn for each group of boys with advanced, standard and delayed puberty. We carried out a prediction of the adult stature of 67 teenagers among the 125 adolescents for which we know, the real adult stature. The procedure to predict adult stature was carried in two steps: First the chronological age between 160 and 164 months associated with pubertal stage informed us if the sedentary boy is in advanced, standard or delayed puberty. Then we calculated the Z-score value of the teenager's and made a projection of this value to have an estimation of adult stature.

| Example of adult stature prediction
For example, one boy measured 168 cm at 163 months with a pubertal stage of ST2. He thus had a standard puberty and his growth kinetics are represented by the curve (Figure 1): y = À0.0053 t 2 + 2.30 t -73.0. On this curve the average stature at 163 months is 161.1 cm. Thus Zscore = (166-161.1)/6 = 1.31 where 6 is the standard deviation.
The projection of his stature to 18 years of age (216 months) is: 177 + 6 x 1.31 = 184.9 cm.

| Statistical analysis
Results are expressed as mean ± SD with the range (maxmin) value. We have plotted the linear relationship between the predicted and the actual adult stature. The correlation coefficient as well the standard error of prediction was specified. Statistical analyses were carried out using Statistica software (Statsoft, Tulsa, Okla, USA). The 95% confidence limits for the predicted stature have been specified graphically. P < .05 was considered to be the significance level. Table 2 gives mean, standard deviation and the range of variables of the total sample. We observed a wide variability of stature and biological maturity between the ages of 12 and 17.5 years. Table 3 shows the repartition of ages at peak stature velocity (APHV) and the puberty stages between individuals with advanced, standard and delayed puberty. Significant differences in APHV (P < .01) were observed between the three maturity groups. These APHV differences allow us to classify the teenagers into three groups according to their maturity according to 24 adolescents (19%) with advanced puberty, 24 with delayed puberty (19%) puberty and 77 with standard puberty (62%).

| RESULTS
In our study, all subjects with delayed puberty between the ages of 160 and 164 months are at stage 1, all subjects with standard puberty are pre-pubescent (stage 2) and all subjects with advanced puberty are pubescent (stage 3). Therefore, these results show that there is a connection between the descriptive secondary pubertal stage and the APHV. Figure 1 shows the new stature growth curves based on three groups made up of individuals with advanced, standard and delayed puberty. At the same age, we observe significant differences in average stature between three growth curves from 12 to 16 years. At the end of puberty, the average values converge at about 177 cm.
Among the 67 adolescents 24% had delayed puberty, 51% standard puberty and 25% advanced puberty. The mean and SD values for stature, age and pubertal stages are shown in Table 4.
There is a linear relationship between adult stature estimates and actual stature with a strong correlation and low SEE (r = 0.97; SEE = 1.68) (Figure 2). Figure 3 represents the differences between the estimated and actual adult stature at the age of 18 years (Figure 3). No significant bias was observed between the differences and the actual stature (r = À0.11; P = .35).

| DISCUSSION
This study demonstrates that, knowing the biological maturation, determined from the age of Peak Height Velocity (APHV), we can differentiate the subjects with delayed, standard and advanced puberty. Consequently, new stature growth curves based on these three groups enable us to monitor individual growth more accurately.
Mean standard puberty curve are similar to the United States growth charts (Kuczmarscki et al. 2000). There were significant differences in the APHV between delayed, standard and advanced maturers. For example, the mean value of our standard maturers APHV (13.9 years) is similar to the Swedish population APHV (Liu et al., 2000).
At the same age, adolescents with advanced puberty are taller than those with delayed puberty (Vizmanos et al., 2001). Mean Stature for each maturity group are similar when the subjects are compared at the time of their PHV. Moreover, the curves show that final stature T A B L E 2 Mean and SD values in the total sample (n = 125)

Variables Mean ± SD Range
Age ( (18 years old) are more or less equal regardless of the maturity group. Age at takeoff correlates highly with pubertal stage, but correlates negatively with duration of puberty (Abassi, 1998). Adolescents with delayed puberty are taller before puberty but equal in final stature to adolescents with delayed puberty (Karlberg et al., 1987). In our study, the timing of puberty is also based on secondary sexual characteristics for which four different stages were defined. The collection of data to describe the secondary sexual maturation stages is not invasive and does not require any medical assistance. We found a close relationship between the descriptive secondary pubertal stage and the APHV. Between 160 and 164 months, these results allow us to classify subjects with delayed, standard or advanced puberty without waiting for the onset of the PHV. Today's baseline growth curves were elaborated in the 1970's and updated in 2006 by the World Health F I G U R E 2 Linear correlation between estimated and actual adult stature.
F I G U R E 3 Difference between estimated and actual adult stature.
Organization (WHO, 2006). This method gives us the values of percentiles and Z scores for 5 to 19-year-old subjects. In addition to the WHO standards, other growth curves are commonly used globally (Juliusson et al., 2009;Roelants et al., 2009;Rosario et al., 2011). However, the individual growth of an adolescent may be significantly vary around his canal according to his maturity.
The mean growth curves established from the reference population were used to predict the adult stature of 67 individuals with a good accuracy. The mean differences between predicted and real adult stature of the sedentary boys is À0.11 cm with 95% limits of agreement of [À3.2; +2.8 cm].
Different models have been proposed to predict adult stature. Sperlich et al. (1995) performed a stature prediction for boys with untreated constitutional stunting using bone age methods. 32.6% of boys tested have a prediction that deviates from the final stature of more than 5 cm. Ostojic (2013) has established a model for predicting the stature of young Caucasian boys playing basketball, football, volleyball and swimming from the Tanner Whitehouse method based on hand x-rays. This method results in a prediction of adult stature between À5.8 cm and + 4.5 cm in 95% of cases. Ali and Ohtsuki (2001) developed a model for predicting stature from curvilinear regression without exposing individuals to X-rays. Their method includes age at peak growth, therefore it is applicable only beyond 12 years for girls and 14 years for boys. They obtained a prediction with a 90% confidence interval between À3.3 and +5 cm. Lee et al. (2008) made a prediction of adult stature with good precision from a multiple regression that includes biological parameters. However, they introduced into the prediction equation the age immediately preceding the age at the peak of growth on the growth curve. For this reason, this method is only applicable at the beginning of the pubertal period. Cumulative growth velocity curves specific to maturity were developed for advanced, standard and delayed maturity. Area under these curves was used to establish reference values to predict adult stature. APHV was used to differentiate subjects with delayed, standard or advanced puberty (Sherar et al., 2005). This method gives a rather inaccurate prediction of the stature as adults at ±5.3 cm in 95% of cases observed on 224 boys.
The non-invasive method of predicting the stature, obtained from skeletal age, suggested by Beunen et al. (2010), makes it possible to differentiate the maturation of individuals. The mean deviation of the predicted stature is 2.3 cm with a standard error of estimation of 4.7 cm at the age of 13 years. More recently, Beunen et al. (2011) proposed a model for predicting the adult stature of girls without using skeletal maturation. Adult stature is estimated based on stature at age 13 or 14, leg length, sitting height, and age at the menarche. In 95% of cases, the prediction is in the range ±5.4 cm at 13 years and ±4.6 cm at 14 years. Ultimately, most adult stature prediction methods have an accuracy greater than 3 cm. In addition, these methods are sometimes restrictive because they use X-rays of the hand and wrist.
In conclusion, the new average growth curves of stature developed from chronological age and pubertal maturation allow us to monitor individual growth more accurately.

ACKNOWLEDGMENTS
This project was sponsored by the National Agency of the Research. Data from this study are available.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.