Timing of Voice Breaking in Males Associated with Growth and Weight Gain Across the Life Course

Abstract

Background:

In contrast to the many studies in females, there are few data in males on the relationships between childhood growth and weight gain and the timing of pubertal maturation and its relevance to adult body mass index (BMI) and body composition.

Methods:

A total of 2008 males in the 1946 British Birth Cohort Study had assessment of pubertal status including voice-breaking status (no change, starting, or complete) at age 14 yr. These responses were related to growth measurements at birth (weight only) and at 2, 4, 6, 7, 11, 14, 20, 26, 36, 43, 53, and 60–64 yr. Body composition was assessed by dual-energy x-ray absorptiometry at 60–64 yr.

Results:

Males with more advanced voice-breaking status at age 14 yr had similar birth weights compared with other males; they showed faster weight gain from 0–2 yr and had higher mean weight and BMI at age 2 yr. Subsequently, they continued to accelerate in weight and BMI, and also in height, and maximum differences in body size were seen at age 14 yr. Adult height did not differ between groups, but males with advanced voice breaking had higher adult BMI and greater whole-body lean mass and greater android fat mass at 60–64 yr.

Conclusion:

Similar to females with earlier menarche, the trajectory to earlier sexual maturation in males is manifested by faster early postnatal growth and weight gain and leads to higher adult BMI. Timing of pubertal maturation has potential relevance to adult disease risks in males. We also describe conditional height difference in sd score as a proxy marker of pubertal timing in males.

There is a wide variation in the timing of puberty in boys and girls. In girls, age at menarche, the onset of the first menstrual bleed during late pubertal development, can be self-recalled in women many years later and has been widely used in epidemiological studies despite only moderate accuracy (1–3). Earlier menarche has been related to low birth weight and faster postnatal weight gain, which is apparent even within the first few weeks of life (4–6). In adult life, earlier menarche is associated with shorter height, higher body mass index (BMI), increased risks for type 2 diabetes and cardiovascular disease (CVD), higher all-cause mortality, and higher mortality from CVD and cancer (7, 8). In boys, more advanced pubertal maturation has been associated with taller childhood height, larger childhood and adult BMI, and higher adult systolic and diastolic blood pressure in one study (9). The lack of additional evidence in boys is largely due to the lack of easily recordable and validated measures of pubertal timing.

Male pubertal onset and progression are manifested by the gradual enlargement of genital size and spread of pubic and axillary hair, as opposed to any distinct changes in appearance (10). The five Tanner stages of sexual maturation in boys were originally described in 1969 (11) and are universally regarded as the convention. However, these stages are difficult to assign accurately without direct observation and palpation of testicular size by trained observers, which has become increasingly unacceptable by many participants in research studies. Self-reported pubertal stage in boys is prone to large errors. Other physical measures have been proposed, such as the timing of the adolescent growth acceleration, which occurs in boys in late puberty; however, this requires frequent growth measures.

Voice breaking in boys usually occurs as a distinct event during late puberty, due to the increased length of the vocal cords that follows the growth spurt of the larynx and represents a further noninvasive measure of pubertal timing, akin to age at menarche in girls (12). In Danish choir boys, data on weekly voice assessments supplemented by fiberoptic laryngoscopy demonstrated a secular trend to earlier puberty associated with increasing BMI (13). In the National Survey of Health and Development (NSHD), voice-breaking status was more simply recorded on one occasion at age 14 yr (14). It was used in a composite measure of pubertal maturation related to growth and blood pressure (9) and was more strongly associated than other parameters of male puberty with the genetic locus for age at menarche at LIN28B (15).

Wehkalampi et al. (16) recently described height difference (HD) in sd scores (SDS) (between ages 11.5–17.5 yr in girls and 14.0–17.5 yr in boys) as a proxy measure of pubertal timing, because it showed strong correlations with age at peak height velocity in boys (r = 0.84) and girls (r = 0.78) (16). However, simple differences between any two repeated measurements may be prone to regression to the mean bias, where more extreme initial values tend to be less extreme at follow-up simply by chance, and vice versa. Cole (17) described conditional change in SDS to correct for regression to the mean in infant weight monitoring.

We therefore assessed voice-breaking status at age 14 yr and also a newly derived parameter, conditional HD-SDS, as noninvasive markers of timing of pubertal maturation in boys and related these and other available measures of pubertal timing in the NSHD to childhood growth and weight gain and also to adult BMI and body composition at age 60–64 yr.

Materials and Methods

Study sample

The NSHD is a socially stratified birth cohort of 2547 men and 2815 women of white European descent born during 1 wk in 1946 who have been followed with repeated data collections since then (14). Between 2006 and 2010, at 60–64 yr, contact was made with 2661 (84%) of 3164 eligible study members still alive and living in England, Scotland, or Wales. No contact was attempted for those who had died (n = 717), were living abroad (n = 567), had previously withdrawn from the study (n = 595), or had been lost to follow-up for more than 10 yr (n = 319). Of the 2661 who provided information, 1690 (63.5%) attended one of six regional clinical research facilities; the remainder were visited at home or completed a postal questionnaire only (18). Inclusion criteria for the current analysis was men with data on birth weight and voice breaking at age 14 yr (n = 2001). The study received Multi-Centre Research Ethics Committee approval, and informed consent was given by cohort participants.

Anthropometry

Birth weight to the nearest quarter pound (113 g) was extracted from medical records and converted to kilograms. Heights and weights were measured using standard protocols at ages 2, 4, 6, 7, 11, 15, 36, 43, 53, and 60–64 yr and were self-reported at ages 20 and 26 yr.

Growth and pubertal assessment

Pubertal characteristics were described by the child's school doctor (medically qualified practitioners who regularly attended schools as part of routine primary care), at a special clinic examination in 1961 when the boys were 14 yr old (mean 14.5, range 14.3–15.2 yr). In boys, the school doctor was asked to rate appearance of the genitals as infantile, early, or advanced. Pubic hair was rated as none, yes sparse, or yes profuse. Axillary hair was rated as no or yes. Voice-breaking status was rated as no, starting to break, or completely broken.

Body composition

During clinic visits at 60–64 yr, measures of body composition were obtained in the supine position using a QDR 4500 Discovery scanner (Hologic Inc., Bedford, MA); the same software (Hologic APEX version 3.1), hardware, and procedures were used in each clinic, and all measures were calibrated internally and across clinics using a European spine phantom. From these scans, whole-body fat and lean mass, android (abdominal) region fat mass and gynoid region fat mass, and appendicular lean mass were obtained and converted into kilograms. The ratio of android to gynoid fat mass was derived, with higher values indicating greater fat distribution in the abdomen than hips. Lean mass was defined as mass excluding fat and bone mass, and in all measures, data from the head were excluded due to the high proportion of bone mass known to affect the accuracy of soft-tissue measures (19). Routine anthropometric measures were taken during the clinic visit using standardized protocols by trained nurses (18). In total, 746 male participants had data available for all body composition outcomes, with missing data mostly due to artifacts such as pacemakers obstructing the x-rays.

Calculations

BMI at each measurement and maternal BMI were calculated as weight in kilograms divided by height in meters squared. Measures of BMI, weight, and height were converted into SDs using internally generated growth charts, which is preferable for historical cohorts, and were constructed using the LMS method (where L is skewness, M is median, and S is coefficient) (20).

HD-SDS was originally described as the difference between adolescent height SDS (at age 14 yr in males) and adult height SDS (16), and we calculated this parameter as height SDS at 36 yr minus height SDS at 14 yr because 36 yr was the first time adult height was measured in the NSHD. HD-SDS was then categorized into three groups for display purposes. Lower HD-SDS values indicate those males with less growth occurring after age 14 yr and who therefore had earlier pubertal growth acceleration and more advanced pubertal maturation at age 14 yr than those with higher HD-SDS values (16).

Conditional HD-SDS was calculated as adolescent height SDS − (r × adult height SDS)/(√1 − r²), where r is the Pearson correlation coefficient for the association between adolescent height SDS and adult height SDS. This equation was as previously described for conditional weight gain (17), but we arranged the variables in the equation to design a parameter that is independent of adult height. Note that higher conditional HD-SDS values indicate males with more advanced puberty at age 14 yr.

Analyses

Linear regression was initially performed to test the cross-sectional linear trends in body size across groups of voice breaking at age 14 yr, adjusted for precise age at pubertal assessment.

Random intercept mixed models were used to test the longitudinal associations between voice-breaking status at age 14 yr and changes in weight, BMI, and height SDS with age, using the xtmixed command in Stata. This approach takes correlations between repeated measures on the same individual into account and allows for missing measurement data assuming that data are missing at random. The linear change in outcome with age was allowed to vary by voice-breaking status by adding a voice-breaking group by age interaction term. We used Wald test statistics to compare models that included linear, quadratic, and cubic functions of age to test whether changes in body size by voice-breaking status were nonlinear between birth to 60–64 yr. To estimate linear coefficients between growth and age during specific developmental periods, we fitted linear piecewise models with knots at 2 (weight only), 7, 14, and 20 yr to allow different linear coefficients for 0–2 yr (infancy; data available for weight only), 2–7 yr (prepubertal childhood), 7–14 yr (early adolescence), 14–20 yr (late adolescence), and 20 to 60–64 yr (adult). All analyses were performed using Stata version 11.0.

Results

At age 14 yr, 537 (26.8%) of 2001 males had no voice breaking, 743 (37.1%) were starting, and 721 (36.0%) had complete voice breaking. Age at the time of pubertal assessment was slightly later in those with complete (14.57 yr) vs. starting (14.54) or no voice breaking (14.50; P trend < 0.0001), and this was adjusted for in all subsequent analyses. Males with different voice-breaking status at age 14 yr showed divergent patterns of growth in weight, height, and BMI during childhood and adulthood (Fig. 1); longitudinal models confirmed nonlinear associations with age (quadratic age terms in models for weight, height, and BMI SDS were all P < 0.001, not shown). Separate analyses were subsequently performed in the prespecified age groups.

Fig. 1.

Mean height, weight, and BMI SDS from birth to age 60–64 yr by voice-breaking status at age 14 yr.

Infancy weight 0–2 yr

There was no difference in birth weight between groups of voice-breaking status (P trend = 0.7; Table 1); thereafter, the groups diverged in weight (Fig. 1). More advanced voice breaking at age 14 yr was associated with heavier weight and BMI at age 2 yr and with a trend to taller height (Table 1). In longitudinal analyses, compared with males with no voice breaking, those with starting to break (P = 0.04) and complete voice breaking at age 14 yr (P < 0.0005) showed faster weight gain between birth and 2 yr (Table 2).

Table 1.

Childhood and adult weight, height, and BMI in males by voice-breaking status at age 14 yr

	Voice broken at age 14 yr			Total	P trend
	No	Starting	Complete
n	537	743	721	2001
Weight (kg) at age (yr):
0	3.47, 3.42–3.51	3.49, 3.45–3.52	3.48, 3.44–3.52	2001	0.6
2	13.0, 12.9–13.1	13.2, 13.1–13.3	13.4, 13.3–13.5	1696	<0.0001
4	17.0, 16.8–17.1	17.4, 17.2–17.6	17.9, 17.8–18.1	1863	<0.0001
6	20.1, 19.9–20.3	20.7, 20.6–20.9	21.5, 21.3–21.7	1814	<0.0001
7	22.1, 21.9–22.4	22.9, 22.7–23.1	23.9, 23.7–24.2	1789	<0.0001
11	32.3, 31.8–32.8	34.1, 33.7–34.5	36.0, 35.6–36.5	1845	<0.0001
14	45.8, 45.0–46.6	51.3, 50.6–51.9	56.7, 56.1–57.3	1883	<0.0001
20	68.7, 67.9–69.5	71.1, 70.3–71.8	72.1, 71.4–72.8	1619	<0.0001
26	72.0, 71.0–72.9	73.6, 72.8–74.4	75.1, 74.3–75.9	1541	<0.0001
36	74.9, 73.7–76.1	76.3, 75.3–77.3	78.3, 77.4–79.3	1364	<0.0001
43	77.5, 76.2–78.7	78.8, 77.8–79.9	81.4, 80.3–82.4	1338	<0.0001
53	82.3, 80.9–83.8	83.8, 82.5–85.0	85.6, 84.4–86.9	1193	0.0008
60–64	84.1, 82.4–85.8	85.8, 84.3–87.3	87.0, 85.6–88.5	886	0.01
Height (cm) at age (yr)
2	85.6, 85.1–86.0	86.1, 85.7–86.5	86.1, 85.6–86.5	1667	0.1
4	102.9, 102.4–103.3	103.4, 103.0–103.7	104.4, 104.0–104.8	1819	<0.0001
6	113.1, 112.6–113.6	114.5, 114.1–114.8	115.7, 115.3–116.1	1799	<0.0001
7	118.7, 118.2–119.1	120.3, 119.9–120.7	122.0, 121.6–122.4	1854	<0.0001
11	138.2, 137.6–138.8	140.8, 140.4–141.3	142.7, 142.2–143.2	1862	<0.0001
14	155.8, 155.1–156.5	161.9, 161.3–162.4	166.8, 166.2–167.3	1899	<0.0001
20	176.5, 175.9–177.2	177.2, 176.7–177.7	177.0, 176.5–177.5	1629	0.7
26	177.1, 176.5–177.8	177.5, 177.0–178.0	177.1, 176.6–177.7	1543	0.8
36	175.1, 174.4–175.8	175.5, 175.0–176.1	175.5, 174.9–176.1	1360	0.6
43	175.0, 174.2–175.7	175.5, 174.9–176.1	175.1, 174.6–175.7	1337	1.0
53	174.6, 173.9–175.3	175.0, 174.4–175.6	174.3, 173.7–174.9	1193	0.3
60–64	174.2, 173.4–175.1	175.1, 174.4–175.8	174.7, 174.0–175.5	886	0.4
BMI (kg/m2) at age (yr)
2	17.9, 17.6–18.1	17.9, 17.7–18.1	18.2, 18.0–18.4	1600	0.03
4	16.1, 15.9–16.2	16.3, 16.1–16.4	16.5, 16.3–16.6	1785	<0.0001
6	15.7, 15.6–15.8	15.8, 15.7–15.9	16.1, 16.0–16.2	1728	<0.0001
7	15.7, 15.6–15.8	15.8, 15.7–15.9	16.1, 16.0–16.2	1780	<0.0001
11	16.8, 16.7–17.0	17.1, 17.0–17.3	17.6, 17.5–17.8	1831	<0.0001
14	18.7, 18.5–19.0	19.5, 19.3–19.7	20.3, 20.2–20.5	1867	<0.0001
20	22.0, 21.8–22.2	22.6, 22.4–22.8	23.0, 22.8–23.2	1591	<0.0001
26	22.9, 22.6–23.2	23.3, 23.1–23.6	23.9, 23.7–24.1	1541	<0.0001
36	24.4, 24.1–24.7	24.8, 24.5–25.0	25.4, 25.1–25.7	1357	<0.0001
43	25.3, 24.9–25.6	25.6, 25.3–25.9	26.5, 26.2–26.8	1337	<0.0001
53	27.0, 26.5–27.4	27.4, 27.0–27.7	28.2, 27.8–28.5	1193	<0.0001
60–64	27.7, 27.2–28.2	28.0, 27.5–28.4	28.5, 28.0–28.9	885	0.02

Results are shown as means, 95% confidence interval. P values for linear trend were adjusted for precise age at pubertal assessment.

Table 2.

Longitudinal analyses of voice-breaking status at 14 yr and growth and weight gain

Age period (yr)	Starting to break			Complete voice breaking
	Weight	BMI	Height	Weight	BMI	Height
0–2	0.054 ± 0.026	NA	NA	0.150 ± 0.026	NA	NA
2–7	0.028 ± 0.010	0.009 ± 0.012	0.035 ± 0.008	0.053 ± 0.010	0.016 ± 0.012	0.080 ± 0.008
7–14	0.048 ± 0.008	0.037 ± 0.009	0.058 ± 0.006	0.081 ± 0.008	0.067 ± 0.009	0.095 ± 0.006
14–20	−0.063 ± 0.009	−0.026 ± 0.010	−0.095 ± 0.007	−0.123 ± 0.009	−0.052 ± 0.010	−0.186 ± 0.007
20 to 60–64	−0.003 ± 0.001	−0.003 ± 0.002	−0.001 ± 0.001	−0.003 ± 0.001	−0.003 ± 0.002	−0.002 ± 0.001

Regression coefficients ± se, estimated from random intercept mixed models with piecewise specification of age periods, represent changes in SDS per year during the specific age periods for males in the starting to break or complete voice breaking groups compared with those with no voice breaking as the reference group. NA, Not available.

Coefficients with P < 0.05 are shown in bold.

Childhood

Males with complete or starting voice breaking at 14 yr showed faster gains in weight and also in height during childhood compared with those with no voice breaking, during both prepubertal childhood (2–7 yr) and early adolescence (7–14 yr) (Table 2 and Fig. 1). Differences in BMI were maintained during the prepubertal years (2–7 yr; Table 1) and widened further during early adolescence (7–14 yr; Table 2). Differences in weight, height, and BMI were all maximal at age 14 yr and reduced during late adolescence (14–20 yr; Tables 1 and 2 and Fig. 1).

Adult size and body composition

Differences in weight and BMI, but not height, by voice-breaking status at age 14 yr were apparent at all time points between 20 and 60–64 yr, with only modest attenuation in the association with weight with increasing age (Table 2). At age 60–64 yr, when body composition was assessed by dual-energy x-ray absorptiometry (DXA), more advanced voice breaking at age 14 yr was associated with greater whole-body lean body mass (P = 0.002) and a trend toward greater appendicular lean mass (P = 0.07; Table 3). Males with more advanced voice breaking at age 14 yr also showed a trend toward greater whole-body fat mass (P = 0.1), specifically greater android fat mass (P = 0.02), and a greater android to gynoid fat mass ratio (P = 0.02); they also had greater waist circumferences at age 53 yr (P = 0.002; Table 3).

Table 3.

Body composition outcomes at 60–64 yr by voice-breaking status at 14 yr

	Voice broken at age 14 yr?			P trend
	No	Starting	Complete
DXA at 60–64 yr (n)
n	167	221	236
Whole-body fat mass (kg)	23.5, 22.4–24.6	23.8, 22.8–24.8	24.7, 23.8–25.6	0.1
Whole-body lean mass (kg)	52.7, 51.6–53.9	53.4, 52.5–54.3	55.0, 54.1–55.9	0.002
Appendicular fat mass (kg)	10.1, 9.6–10.6	10.2, 9.7–10.6	10.3, 9.9–10.7	0.7
Appendicular lean mass (kg)	24.4, 23.8–25.0	24.5, 24.1–24.9	25.1, 24.6–25.5	0.07
Android fat mass (kg)	2.4, 2.3–2.6	2.5, 2.3–2.6	2.7, 2.5–2.8	0.02
Gynoid fat mass (kg)	3.7, 3.6–3.9	3.8, 3.6–3.9	3.9, 3.7–4.0	0.2
Android to gynoid fat mass ratio	0.65, 0.63–0.67	0.65, 0.63–0.67	0.68, 0.66–0.70	0.02
Fat mass (%)	29.5, 28.8–30.3	29.4, 28.7–30.1	29.7, 29.0–30.4	0.6
Waist circumference at 53 (cm)	96.9, 95.7–98.1	97.7, 96.7–98.7	99.3, 98.3–100.2	0.002
n	321	439	436

Results are shown as means, 95% confidence interval; n = 624 for all variables. P values for linear trend were adjusted for age at pubertal assessment and height at 60–64 yr.

Comparison with other markers of pubertal maturation at age 14

There were moderate intercorrelations between voice-breaking status and other markers of pubertal status at age 14 yr, with r² values for linear associations ranging from 22.1–32.7% (Table 4). All markers of pubertal maturation at age 14 yr showed consistent associations with childhood weight and growth, with the exception of HD-SDS, which was the only marker of pubertal maturation that was associated with birth weight and adult height (Table 5). Lower HD-SDS values, which indicate earlier pubertal maturation, were associated with lower birth weight and shorter adult height, and HD-SDS also showed different patterns of association with childhood size compared with the other markers (Table 5).

Table 4.

Intercorrelations between voice breaking and other markers of pubertal maturation at age 14 yr

	Genital development (%)	Pubic hair (%)	Axillary hair (%)	HD-SDS (%)
Voice breaking (%)	28.1	32.7	22.4	28.2
Genital development (%)		46.5	21.3	22.4
Pubic hair (%)			23.2	19.2
Axillary hair (%)				15.0

Values are r² from linear associations between each pair of variables. HD-SDS is difference in height SDS between age 14 yr and adult (continuous variable).

Table 5.

Comparison of associations between different markers of pubertal maturation at age 14 yr and childhood and adult size

Marker	Child-like	Intermediate	Advanced	Total	P trend
Voice breaking	No change	Starting	Complete
n	537	743	721
Birth weight SDS	0.00, −0.09–0.08	0.02, −0.06–0.09	0.01, −0.07–0.08	2001	0.9
Weight SDS at 2 yr	−0.15, −0.24 to −0.06	−0.03, −0.11–0.05	0.14, 0.07–0.22	1696	<0.0001
Weight SDS at 36 yr	−0.17, −0.28 to −0.06	−0.04, −0.13–0.05	0.16, 0.07–0.24	1364	<0.0001
Height SDS at 7 yr	−0.34, −0.42 to −0.25	−0.03, −0.10–0.04	0.27, 0.19–0.35	1854	<0.0001
Height SDS at 36 yr	−0.05, −0.16–0.06	0.02, −0.07–0.10	0.01, −0.07–0.10	1360	0.6
Genital development	Infantile	Early	Advanced
n	87	804	1085
Birth weight SDS	−0.09, −0.28–0.11	0.01, −0.06–0.07	0.00, −0.05–0.06	1976	0.7
Weight SDS at 2 yr	−0.15, −0.39–0.10	−0.10, −0.17 to −0.03	0.10, 0.03–0.16	1676	<0.0001
Weight SDS at 36 yr	−0.26, −0.57–0.05	−0.10, −0.19 to −0.01	0.09, 0.02–0.17	1345	<0.0001
Height SDS at 7 yr	−0.43, −0.67 to −0.19	−0.20, −0.27 to −0.13	0.18, 0.12–0.24	1833	<0.0001
Height SDS at 36 yr	−0.14, −0.44–0.16	−0.02, −0.11–0.06	0.03, −0.04–0.10	1341	0.2
Pubic hair	No	Yes sparse	Yes profuse
n	245	796	947
Birth weight SDS	−0.03, −0.15–0.10	0.01, −0.05–0.08	−0.01, −0.07–0.05	1988	0.8
Weight SDS at 2 yr	−0.27, −0.42 to −0.12	0.00, −0.08–0.07	0.08, 0.01–0.14	1684	<0.0001
Weight SDS at 36 yr	−0.23, −0.39 to −0.06	−0.02, −0.11–0.07	0.06, −0.01–0.14	1353	<0.0001
Height SDS at 7 yr	−0.49, −0.62 to −0.37	−0.11, −0.18 to −0.04	0.21, 0.14–0.28	1842	<0.0001
Height SDS at 36 yr	−0.13, −0.27–0.02	0.03, −0.05–0.12	0.00, −0.07–0.08	1349	0.2
Axillary hair	No	Yes
n	821	1154
Birth weight SDS	0.01, −0.05–0.08	−0.02, −0.07–0.04		1975	0.5
Weight SDS at 2 yr	−0.12, −0.19 to −0.05	0.09, 0.03–0.15		1672	<0.0001
Weight SDS at 36 yr	−0.15, −0.24 to −0.07	0.11, 0.05–0.18		1346	<0.0001
Height SDS at 7 yr	−0.25, −0.32 to −0.18	0.17, 0.11–0.23		1830	<0.0001
Height SDS at 36 yr	−0.01, −0.09–0.08	0.00, −0.06–0.07		1342	0.9
HD–SDS	>0.67	−0.67–0.67	<−0.67
n	430	432	430
Birth weight SDS	0.07, −0.02–0.16	−0.01, −0.10–0.08	−0.11, −0.20 to −0.03	1292	0.008
Weight SDS at 2 yr	0.00, −0.10–0.11	0.00, −0.10–0.10	0.04, −0.06–0.14	1105	0.8
Weight SDS at 36 yr	0.12, −0.01–0.24	−0.01, −0.09–0.06	−0.06, −0.17–0.05	1289	0.04
Height SDS at 7 yr	−0.06, −0.15–0.04	−0.03, −0.13–0.07	0.08, −0.02–0.17	1201	0.2
Height SDS at 36 yr	0.43, 0.33–0.52	−0.05, −0.13–0.04	−0.38, −0.46 to −0.30	1292	<0.0001
Conditional HD–SDS	<−0.67	−0.67–0.67	>0.67
n	355	577	360
Birth weight SDS	−0.02, −0.12–0.08	−0.02, −0.10–0.06	−0.01, −0.11–0.08	1292	0.9
Weight SDS at 2 yr	−0.15, −0.27–0.04	−0.01, −0.10–0.08	0.22, 0.12–0.32	1105	<0.0001
Weight SDS at 36 yr	−0.18, −0.29–0.07	0.02, −0.07–0.10	0.16, 0.06–0.26	1289	<0.0001
Height SDS at 7 yr	−0.40, −0.51–0.30	−0.06, −0.14–0.02	0.48, 0.38–0.58	1201	<0.0001
Height SDS at 36 yr	0.03, −0.08–0.14	−0.05, −0.13–0.03	0.05, −0.04–0.14	1292	0.8

HD-SDS is the difference in height SDS between age 14 yr and adult (lower HD-SDS values indicate more advanced pubertal maturation at age 14 yr). Conditional HD-SDS is the conditional difference in height SDS between adult and age 14 yr (higher HD-SDS values indicate more advanced pubertal maturation at age 14 yr).

Our new parameter, conditional HD-SDS, designed to correct the bias inherent in (unconditional) HD-SDS (17), showed a pattern of associations with childhood and adult weight and height that was more consistent with the other physical markers of pubertal maturation at age 14 yr (Table 5).

Discussion

Similar to females with earlier menarche, the trajectory to earlier sexual maturation in males, as indicated by earlier voice breaking, is manifested by faster weight gain and growth in infancy and early childhood and leads to persistently higher adult BMI. Unlike findings in females, timing of sexual maturation in males was not associated with birth weight or with adult height (9).

It is well established that earlier pubertal maturation is linked to a faster tempo of growth, that is to say a faster growth trajectory that approaches the adult height target at an earlier age. Such differences in tempo are particularly marked during adolescence and are consequent to the pubertal rise in sex steroids (21). Our current findings support the concept that the trajectory to earlier sexual maturation in males, as well as in females, is manifested as a faster tempo of growth throughout infancy and childhood, rather than being confined to the pubertal years. Such early life processes could potentially include the transient pituitary-gonadal activation that has been described in both male (22) and female (23) infants. Alternatively, it could reflect mechanisms independent of sex hormones, such LIN28 and LIN28B, regulators of microRNA preprocessing that have been linked to childhood growth and pubertal timing in human genetics studies (15, 24) and in mouse models (25). Males with earlier voice breaking had higher BMI than other males, and this was already apparent from ages 2 yr onward. Higher childhood body weight and BMI could also directly promote earlier pubertal onset and progression through leptin, insulin, and/or other hormonal mechanisms (26).

In adult life, males with earlier voice breaking continued to have higher BMI than other males throughout the duration of follow-up with no detectable attenuation up to age 60–64 yr. The differences in BMI at age 60–64 yr by timing of voice breaking were accompanied by differences in whole-body lean mass and in regional fat mass, such that males with earlier voice breaking had greater total lean mass and greater abdominal fat mass. In females, puberty is associated with a rise in percent body fat, relative to body weight, from around 23–29% between early to late puberty (27), and females with earlier timing of menarche have greater fat mass in late adolescence (27) and as adults (28). In contrast, puberty in males is characterized by greater gains in fat-free mass than in fat mass (29) and by greater gains in central fat relative to total body fat (30); we now show that these pubertal changes appear to be exacerbated in those males with earlier timing of sexual maturation.

Our findings are consistent with data from the few other studies in males, which invariably estimated the timing of puberty by analysis of adolescent height growth. Earlier age at peak height velocity has been associated with greater central but not peripheral fat mass by DXA at age 18 yr (31); greater central to peripheral skinfold ratio at age 30 yr (32); greater BMI, waist circumference, and percent body fat at age 27–35 yr (33); and higher BMI at age 63 yr (34). Only one other study used a non-growth-based measure of pubertal timing; in the 1958 British birth cohort, more advanced axillary hair stage at age 16 yr was associated with greater BMI at 23 and 33 yr old (35).

Elevated lifetime BMI and adult abdominal fat in males with earlier voice breaking may have important implications for increased CVD and type 2 diabetes risks, although conversely, their greater lean body mass may be protective. In the NSHD, we previously reported that male early developers, defined by a composite of pubertal measures at age 14 yr, had substantially higher blood pressure (BP) at age 53 yr; mean systolic BP was 6.4 mm Hg higher in the earliest puberty group compared with the latest, and for diastolic BP, the difference was 4.6 mm Hg (9). Other studies have associated earlier age at peak height velocity to higher fasting plasma triglyceride and insulin levels (33). Earlier pubertal maturation might also have adverse consequences for insulin sensitivity and glucose metabolism through mechanisms other than BMI and body composition, such as the regulation of microRNA (36). Additional studies are needed to confirm these differences, to understand their mechanisms and to identify their consequences for disease and mortality as have been reported in women with early menarche (7).

A key limitation in characterizing such associations in males is the difficulty in accurately assessing the timing of puberty in cohort studies. Although our findings show that the timing of voice breaking appears to be a noninvasive marker, additional studies should assess the risks of bias in observed, self-reported, or even recalled timing of voice breaking. Alternatively, where archival data are available, the difference between height at age 14 yr and adult height may provide a reasonable proxy (16). However, we found that when using HD-SDS, the early maturers showed lower birth weights and lower adult heights, which was at odds with our direct markers of puberty (Table 5). Our earlier analysis in NSHD also reported no association between pubertal timing and adult height in men, although earlier puberty was associated with greater trunk to leg length ratio (9). It is therefore highly likely that the positive association between HD-SDS and adult height is spurious due to regression to the mean inherent in the formula (HD-SDS = adult height SDS − adolescent height SDS) and that this bias led to a spurious association with birth weight and masked the true differences in infancy and early childhood size.

We propose a simple modification of HD-SDS to make this value independent of adult height. Conditional change in SDS was originally described for weight gain from birth where it was designed to make that value independent of birth weight (17). We confirmed that conditional HD-SDS was indeed independent of adult height and shows a similar pattern of associations with weight and height with the direct markers of pubertal timing (Table 5). However, in females, in whom earlier timing of puberty is associated with modest reductions in adult height (7), neither HD-SDS (which is strongly related to adult height) nor conditional HD-SDS (which is unrelated to adult height) may be appropriate alternatives for pubertal timing, and the best noninvasive measure remains age at menarche.

Strengths of our study include the direct assessment of pubertal timing during adolescence by medically qualified school doctors, combined with unique very-long-term follow-up over the life course and adult body composition assessed by whole-body DXA. However, our study has several limitations. Pubertal maturation was assessed by a large number of school doctors working in routine practice throughout England, Scotland, and Wales in 1961, before publication of the current Tanner stage classification (11), and the categories used provided relatively limited discrimination between groups. However, these categories still provided sufficient power to detect highly significant associations with childhood and adult growth and weight gain, with strong consistency in these associations between the various measures of pubertal maturation. Adult height was not measured until age 36 yr, and attrition in follow-up meant that available numbers were smaller for HD-SDS than for other markers of pubertal timing; however, highly significant associations were still observed. Unfortunately, body composition was not assessed before age 60–64 yr of age, and it would be interesting to monitor the effects of pubertal timing on adult changes in body composition with aging. Finally, although the NSHD cohort remains broadly representative of the national population of a similar age (37), in both males and females, there has been a secular trend to earlier puberty in more recent birth cohorts (38).

In conclusion, similar to females with earlier menarche, the trajectory to earlier sexual maturation in males is manifested by faster weight gain from birth and faster growth from early childhood and leads to higher adult BMI, greater lean mass, and greater abdominal fat mass, with potential relevance for adult health. In studies where direct assessments of pubertal development are not available, conditional HD-SDS is an appropriate proxy marker for pubertal timing in males.