SEARCH

SEARCH BY CITATION

Keywords:

  • epidemiology;
  • puberty;
  • risk assessment

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Author contribution
  8. Acknowledgement
  9. Conflict of interest
  10. References

Epidemiological studies have raised concern about the reproductive consequences of prenatal cigarette smoking exposure, possibly affecting semen quality and onset of pubertal development of the offspring. The aim of this study was to further investigate pubertal development in young men exposed to cigarette smoking in foetal life. In a Danish pregnancy cohort, information on maternal smoking during pregnancy was available from questionnaires administered in 1984–1987, and information on pubertal development, assessed by age at first nocturnal emission, acne, voice break and regular shaving, was obtained from a follow-up questionnaire administered in 2005 to the young men (age: 18–21). We found no significant association between prenatal cigarette smoking exposure and earlier onset of puberty, but we did observe a tendency towards earlier age of first nocturnal emission, acne and voice break, indicating an accelerated age of pubertal development. Men exposed to ≥15 cigarettes/day had 3.1 months (95% CI: −6.4; 0.2) earlier age at acne and 2.2 months (95% CI: −7.3; 3.0) earlier age at first nocturnal emission, 1.2 months (95% CI: −4.6; 2.2) earlier age at voice break, however, 1.3 months (95% CI: −1.6; 4.3) later age at regular shaving, compared with unexposed men. Prenatal cigarette smoking exposure may induce an earlier age at onset of puberty in young men, but larger studies with prospectively collected data on pubertal development are needed to explore this hypothesis further.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Author contribution
  8. Acknowledgement
  9. Conflict of interest
  10. References

A secular trend towards an earlier onset of puberty in girls has been proposed in several publications (Zacharias & Wurtman, 1969; Wyshak & Frisch, 1982; Lee et al., 2001; Mul et al., 2001; Euling et al., 2008; Aksglaede et al., 2009). Reports on boys are sparse and the results are less clear than for girls (Lindgren, 1996, Lee et al., 2001; Reiter & Lee, 2001; Tomova et al., 2011), but a possible advancement in the timing of male puberty have also been suggested (Lee et al., 2001; Tomova et al., 2011). Altered timing of puberty may have health implications later in life and have been associated with, for instance, a higher risk of metabolic syndrome, polycystic ovarian syndrome (Ibanez et al., 2000), as well as testicular cancer (Moss et al., 1986; Forman et al., 1994; Moller & Skakkebaek, 1996).

Among healthy children, there is a wide physiologic variation in the age at onset of puberty. The variability is generated by genetic factors, ethnicity, as well as environmental factors such as socioeconomic background and nutritional conditions (Parent et al., 2003), but other environmental factors may also play a role. Recently, the possible role of maternal cigarette smoking during pregnancy has been considered a risk factor that may affect the onset of puberty. Cigarette smoke consists of toxic components, such as polycyclic aromatic hydrocarbons, nicotine and its metabolite cotinine, that cross the placenta and appear to be potential endocrine disrupters (MacKenzie & Angevine, 1981; Jauniaux et al., 1999). Maternal cigarette smoking has been associated with subfecundity, measured as longer time to pregnancy (Olsen et al., 1983; Weinberg et al., 1989; Jensen et al., 1998, 2006) and impaired semen quality (Storgaard et al., 2003; Jensen et al., 2004, 2005; Ramlau-Hansen et al., 2007; Paasch et al., 2008) in the male offspring. During recent years, studies have focused on whether prenatal exposure to cigarette smoking affects sexual maturation, and, although conflicting results have been published (Fried et al., 2001; Windham et al., 2008), age of menarche may be accelerated in girls exposed to cigarette smoking in foetal life (Windham et al., 2004; Rubin et al., 2009; Shrestha et al., 2011). To the best of our knowledge, only two studies have addressed boys, and their findings indicate that prenatal exposure to cigarette smoke may also lead to an earlier male pubertal development (Fried et al., 2001; Ravnborg et al., 2011).

In this follow-up study with prospectively collected data on maternal cigarette smoking, we evaluate whether exposure to cigarette smoking during foetal life induces an earlier age at onset of puberty among male offspring.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Author contribution
  8. Acknowledgement
  9. Conflict of interest
  10. References

Study population

The study was based on data from the ‘Healthy Habits for Two’ cohort (Olsen et al., 1989) which was established as a health campaign aimed at all pregnant women in two Danish municipalities (Aalborg and Odense) between April 1984 and April 1987. A total of 11 980 pregnant woman (87% of all invited) participated and filled out self-administered questionnaires handed out by midwifes around the 36th gestational week, on parents’ sociodemographic and lifestyle factors before and during pregnancy, including cigarette smoking. Of the 11 980 participants, 11 144 delivered live born singletons of which 5716 were of male gender.

In 2005, the young men (between 18 and 21 years of age), who were alive and still living in Denmark, were identified in the Danish Civil Registration System and asked to complete a follow-up questionnaire. The questionnaire was administered through the Internet to a total of 5142 young men and 2810 (55%) responded

Assessment of prenatal cigarette smoking

Information on maternal lifestyle and demographics were available from the questionnaires filled in by the pregnant women around 36th week of gestation. The main information on maternal cigarette smoking stem from the question: ‘Do you smoke now?’, and those who reported current smoking were asked ‘How much do you currently smoke on average’? with the following response options: ‘do not smoke now’, ‘1–4 cigarettes/1–2 cheroots daily’, ‘5–9 cigarettes/3–5 cheroots daily’, ‘10–14 cigarettes/6–7 cheroots daily’, ‘15–19 cigarettes/8–10 cheroots daily’, or ‘≥20 cigarettes/≥11 cheroots daily’. Furthermore, the women were asked ‘How many cigarettes did you smoke the year before you got pregnant?’ with the following response categories: ‘non-smoker’, ‘1–4 cigarettes daily’, ‘5–9 cigarettes daily’, ‘10–14 cigarettes daily’, ‘15–19 cigarettes daily’, or ‘20 or more cigarettes daily’ as well as ‘Have you smoked since you got pregnant (yes or no)?’. From these answers, five groups were defined: non-smokers (included women who neither smoked the year before or during the pregnancy), past smokers (included women who smoked before pregnancy or stopped sometime before 36th gestational week), 1–9 cigarettes/day during pregnancy, 10–14 cigarettes/day during pregnancy and ≥15 cigarettes/day during pregnancy. These five groups are used as the main exposure variable in the statistical analyses.

Assessment of indicators of pubertal development

In the follow-up questionnaire administered in 2005, the young men were asked about four different indicators of pubertal development: ‘Have you had acne?’, ‘Has your voice broken?’, ‘Have you started to shave regularly?’ and ‘Have you had your first nocturnal emission?’ The following response categories were given: ‘no’, ‘yes’, ‘prefer not to answer’ and ‘do not remember’. Those who answered ‘yes’ were also asked to provide the age (in years and months) at which the event first occurred. To create continuous outcome variables for each of the four events, we converted the month into the fraction of a year before adding years and months.

Covariates

A priori, we identified potential confounders and included the following covariates in the statistical analyses: maternal age at delivery in years (continuous), maternal pre-pregnancy body mass index (BMI) (<18.5, 18.5–24.9, >24.9 kg/m2), maternal alcohol consumption during pregnancy (0 drink/week, 0.5–1.5 drinks/week, ≥2.0 drinks/week), chronic diseases of the mother (diabetes mellitus, epilepsy, arthrosis, heart disease, cancer, psychiatric disorders, allergy or other chronic diseases, combined into one variable, present or not present), municipality of residence at the time of delivery (Odense and Aalborg, other places), family's socioeconomic status based on the highest ranking of job description or academic background between the mother and the father at the time of pregnancy (white-collar workers, blue-collar workers, unemployed, students) and cohabitation status of the parents (yes, no).

Statistical analyses

Missing information

The number (%) of young men who provided an age at acne, voice break, regular shaving and first nocturnal emission were 1804 (64%), 1696 (60%), 2128 (76%) and 924 (33%), respectively. About three quarter of these only provided age in years (and not months), and consequently there were a relatively large amount of partly missing information in the data set. The level of missing values according to maternal cigarette smoking during pregnancy is presented in Table 1. As complete case analysis may lead to biased estimates of the association studied (Greenland & Finkle, 1995; Sterne et al., 2009), we addressed the missing data problem by using multiple imputations, which yield unbiased and more precise estimates if data are missing at random (Rubin, 1987; Sterne et al., 2009). Briefly, multiple imputation is an approach that creates several (m > 1) different imputed data sets, based on other known subject characteristics from the whole data set and thus incorporates an appropriate variability. The m complete data sets are then analysed and the results are combined by use of the so-called Rubin's rule, thus producing a single set of inference that include the variability associated with the missing data.

Table 1. Number (%) of missing values according to maternal cigarette smoking during pregnancy
 Maternal cigarette smoking during pregnancy
 Non-smoker (= 1544)Past smoker (n = 305)1–9 cigarettes/day (n = 473)10–14 cigarettes/day (n = 322)≥15 cigarettes/day (n = 161)
Indicators of pubertal development
First nocturnal emission
Age in years, no. (%)1034 (66.9)202 (66.2)315 (66.6)207 (64.3)124 (77.0)
Age in months, no. (%)1401 (90.7)278 (91.1)439 (92.8)288 (89.4)152 (94.4)
Acne
Age in years, no. (%)562 (33.4)108 (35.4)162 (34.3)116 (36.0)58 (36.0)
Age in months, no. (%)1333 (86.3)275 (93.4)405 (85.6)269 (83.5)141 (87.6)
Voice break
Age in years, no. (%)620 (40.2)118 (38.7)181 (38.3)128 (39.8)65 (40.4)
Age in months, no. (%)1302 (84.3)275 (90.2)400 (84.6)258 (80.1)142 (88.2)
Regular shaving
Age in years, no. (%)374 (24.2)86 (28.0)114 (24.1)68 (21.1)40 (24.8)
Age in months, no. (%)1248 (80.8)252 (82.6)370 (78.2)255 (79.2)133 (82.6)
Maternal and parental characteristics
Maternal age, no. (%)2 (0.1)0 (0)1 (0.2)0 (0)0 (0)
Maternal pre-pregnancy BMI, no. (%)100 (6)12 (4)39 (8)28 (9)13 (8)
Maternal alcohol consumption, no. (%)0 (0)0 (0)0 (0)0 (0)0 (0)
Mother with chronic diseases, no. (%)23 (2)5 (2)8 (2)8 (3)4 (3)
Municipality of residence, no. (%)3 (0.2)0 (0)0 (0)0 (0)0 (0)
Family's socioeconomic status, no. (%)0 (0)0 (0)0 (0)0 (0)0 (0)
Cohabitation status, no. (%)70 (5)18 (6)32 (7)20 (6)14 (9)

We restricted the sample to subjects who had at least provided an age in years for one of the four events. Those who failed to meet these inclusion criteria were excluded from the analyses. Only one of the men had missing information on age at all of the four markers of pubertal development, but in total, 288 participants were excluded and 2522 participants constituted the final study population. We generated 100 complete data sets with imputed data. The following variables were included in the main imputation model: age at first nocturnal emission, age at acne, age at voice break, age at regular shaving, maternal cigarette smoking, maternal pre-pregnancy BMI, maternal age at delivery, alcohol consumption during pregnancy, chronic diseases of the mother, municipality of residence at the time of delivery, family's socioeconomic status and cohabitation of the parents. We present the results from the data analyses based on the imputed data sets. We performed different imputation models and compared the results, to check for consistency, that is, the sensitivity of the results to the choice of model used for imputations.

Data analyses

Data for the timing of the four indicators of pubertal development were approximately normally distributed. We calculated the mean (SD) age for each event. We performed multiple linear regression analyses, using the five strata of maternal cigarette exposure as the explanatory variable, with non-smoker as the referent and adjusted for the covariates described above. Furthermore we tested for trend. As 2.1% of the women contributed with two children to the cohort, we applied robust standard errors in the adjusted analyses to account for ‘clustering’. We assessed the appropriateness of the model by examining residual plots using the complete case analyses, and found symmetrically distributed residuals.

We performed subanalyses to check the consistency of our results. First, we repeated the analyses using different models to impute the missing data from. Secondly, we performed restricted analyses, with those who reported at least age in years at all event. Finally, we repeated the analyses including the men's own smoking habits and alcohol consumption as covariates. Further, we tested the correlations between the four indicators of pubertal development adjusted for the covariates described above, both for the imputed dataset and the complete cases.

The statistical analyses were performed using Stata 12 software (Stata Corporation, College Station, TX, USA). A two-tailed p value of <0.05 was considered statistically significant. All estimates are accompanied by 95% confidence intervals (CIs) calculated from robust standard error.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Author contribution
  8. Acknowledgement
  9. Conflict of interest
  10. References

The initial cohort included a total of 2810 young men. After exclusion of 288 men prior to the multiple imputation methods, data were available for 2522 men, who constituted the study population.

In Table 1, the pattern of missing values in the final study population according to the level of maternal cigarette smoking is presented. Maternal cigarette smoking was associated with a higher level of missing data on first nocturnal emission, whereas the missing data for the other outcomes as well as covariates were approximately similarly distributed in the cigarette smoking categories.

Table 2 shows the characteristics of the study population stratified according to the number of indicators of pubertal development with provided age. Compared with the men who provided information on age of pubertal development (n = 2522), the non-responding men (n = 288) were slightly older and were more often smokers. Also, they had more often mothers who smoked ≥15 cigarettes/day, were overweight and their parents were less often white-collar workers. Finally, there were a lower proportion of the mothers who reported drinking ≥2.0 drinks/week and mothers with chronic diseases among these men.

Table 2. Characteristics according to number of pubertal development events with provided age
 Number of pubertal development events with provided age
4 events (= 522)1–3 events (n = 2000)0 events (n = 288)
Maternal and parental characteristics
Age, years [mean (SD)]27.9 (4.4)27.8 (4.6)27.8 (4.8)
Maternal cigarette smoking
Non-smoker, no. (%)289 (56)1,093 (55)163 (57)
Past smoker, no. (%)53 (10)207 (11)36 (12)
1–9 cigarettes/day, no. (%)90 (17)342 (17)41 (14)
10–14 cigarettes/day, no. (%)67 (13)227 (11)28 (10)
≥15 cigarettes/day, no. (%)22 (4)119 (6)20 (7)
Pre-pregnancy BMI
<18.5 kg/m2, no. (%)41 (8)153 (8)20 (7)
18.5–24.9 kg/m2, no. (%)395 (81)1494 (80)209 (78)
>24.9 kg/m2, no. (%)52 (11)212 (12)41 (15)

Alcohol consumption during

pregnancy

0 drink/week, no. (%)85 (16)333 (17)46 (16)
0.5–1.5 drinks/week, no. (%)293 (56)1118 (56)178 (62)
≥2.0 drinks/week, no. (%)144 (28)549 (27)64 (22)
Mother with chronic diseases
Yes, no. (%)96 (19)330 (17)45 (16)
No, no. (%)422 (81)1632 (83)237 (84)

Municipality of residence at the

time of delivery

Aalborg and Odense, no. (%)323 (62)1302 (65)190 (66)
Other places, no. (%)198 (38)696 (35)98 (34)
Family's socioeconomic status
White-collar workers, no. (%)405 (77)1556 (78)199 (69)
Blue-collar workers, no. (%)99 (19)354 (18)66 (23)
Unemployed, no. (%)15 (3)69 (3)20 (7)
Students, no. (%)3 (1)21 (1)3 (1)
Cohabitation status

Mother living with the father

of the child, no. (%)

481 (98)1857 (98)258 (97)

Mother not living with the father

of the child, no. (%)

9 (2)42 (2)9 (3)
Person-related characteristics of sons
Birth weight, g [mean (SD)]3534 (507)3552 (500)3562 (525)
Cigarette smoking
Yes, no. (%)93 (22)392 (23)56 (24)
No, no. (%)336 (78)1299 (77)176 (76)

Age when responding to questionnaire, years

[mean (SD)]

19.4 (0.8)19.5 (0.8)19.6 (0.9)

About 40% of the mothers reported smoking sometime during pregnancy, and 34% reported smoking in late pregnancy (around 36th week of gestation). We observed that maternal cigarette smoking was positively associated with maternal alcohol consumption during pregnancy and family's socioeconomic status and inversely associated with pre-pregnancy BMI. A higher proportion of heavy smoking mothers (≥15 cigarettes/day) reported chronic diseases and reported not living with the father of the child, compared with non-smoking mothers. Compared with the unexposed, young men prenatally exposed to tobacco smoking had a lower birth weight (data not shown).

The crude mean (95% CI) age at the four indicators of pubertal development among all participants were as follows: acne: 14.6 (14.5; 14.7) years, voice break: 14.5 (14.5; 14.6) years, regular shaving: 17.2 (17.2; 17.3) years and first nocturnal emission: 14.8 (14.7; 14.9) years. Table 3 shows the correlations among the four outcomes adjusted for maternal cigarette smoking during pregnancy and other potential confounding factors.

Table 3. Correlation matrices
 EmissionAcneVoice breakRegular shaving
  1. * Results are adjusted for maternal smoking during pregnancy, maternal alcohol consumption during pregnancy, maternal pre-pregnancy BMI, maternal age at delivery, chronic diseases of the mother, municipality at the time of delivery, socio-economic status of the family and cohabitation status of the parents.

Correlation matrix for imputed dataset, adjusted for potential confounders*
Emission1
Acne0.331
Voice break0.390.601
Regular shaving0.300.300.381
Correlation matrix for complete cases, adjusted for potential confounders*
Emission1
Acne0.331
Voice break0.370.631
Regular shaving0.290.270.401

In Table 4, results from the multiple linear regression analyses after mutiple imputation, are presented with the crude mean (SD) age at appearance of the indicators of pubertal development across exposure levels as well as adjusted differences between the exposure groups. We observed tendencies towards lower age at first nocturnal emission, acne and voice break with higher levels of maternal cigarette smoking exposure, although, none of the trends were statistically significant. Compared with unexposed men, men exposed to ≥15 cigarettes/day had 0.27 (95% CI: −0.54; 0.01) years earlier age at acne, corresponding to 3.2 months (95% CI in months: −6.5; 0.1). Furthermore, men exposed to ≥15 cigarettes/day during pregnancy had 2.3 months (95% CI: −7.4; 3.0) earlier age at first nocturnal emission and 1.7 months (95% CI: −4.8; 1.4) earlier age at voice break. In contrast, age at regular shaving among high-exposed men was observed to be 1.6 months (95% CI: −1.4; 4.6) later than among unexposed men.

Table 4. Age at indicators of pubertal development according to level of maternal cigarette smoking during pregnancy, imputed dataset (N = 2522)
Maternal cigarette smokingDistribution (%)First nocturnal emissionAcneVoice breakRegular shaving
Age mean (SD)Adjusteda difference (95% CI)Age mean (SD)Adjusteda difference (95% CI)Age mean (SD)Adjusteda difference (95% CI)Age mean (SD)Adjusteda difference (95% CI)
  1. Results from test for trend: emission: p for trend = 0.42, acne: p for trend = 0.09, voice break: p for trend = 0.29, regular shaving: p for trend = 0.34. aAdjusted for maternal age at delivery (continuous), pre-pregnancy BMI (<18.5, 18.5–24.9, >24.9 kg/m2), maternal alcohol drinking during pregnancy (0 drink/week, 0.5–1.5 drinks/week, ≥2.0 drinks/week), chronic diseases of the mother [diabetes mellitus, epilepsy, arthrosis, heart disease, cancer, psychiatric diseases or other chronic diseases, combined into one variable (present or not present)], municipality of residence at the time of delivery (Odense and Aalborg, other places), family's socioeconomic status based on the highest ranking of job description or academic background between the mother and the father at the time of delivery (skilled workers, unskilled or semiskilled workers, unemployed, students), and cohabitation status of the parents (Mother not living with the father of the child). The following characteristics were chosen as the referent in the models: maternal age = 25, pre-pregnancy BMI between 18.5–24.9, no maternal alcohol consumption, no chronic diseases of the mother, mothers living in Odense or Aalborg at the time of delivery, family's socioeconomic status = skilled workers, cohabitation status of the parents = Mother living with the father of the child.

Non-smoker5514.8 (1.7)Reference14.6 (1.4)Reference14.5 (1.3)Reference17.2 (1.4)Reference
Past smoker1114.8 (1.7)0.03 (−0.26; 0.32)14.7 (1.5)0.05 (−0.17; 0.27)14.5 (1.3)−0.01 (−0.20; 0.18)17.3 (1.4)0.09 (−0.11; 0.28)
1–9 cig/day1714.9 (1.8)0.05 (−0.21; 0.30)14.7 (1.5)0.04 (−0.15; 0.22)14.6 (1.4)0.04 (−0.13; 0.20)17.2 (1.5)0.07 (−0.11; 0.24)
10–14 cig/day1114.7 (1.7)−0.15 (−0.46; 0.16)14.5 (1.4)−0.15 (−0.36; 0.06)14.4 (1.4)−0.11 (−0.31; 0.09)17.2 (1.3)0.03 (−0.15; 0.21)
≥15 cig/day614.6 (1.7)−0.19 (−0.62; 0.25)14.4 (1.4)−0.27 (−0.54; 0.01)14.4 (1.3)−0.14 (−0.40; 0.12)17.3 (1.3)0.13 (−0.12; 0.38)

We performed secondary analyses. First, analyses based on the other imputation models showed the same results as those presented in Table 4 (data not shown). In the analyses restricted to individuals with complete information, we found similar associations as above, however, the mean differences between exposed and unexposed men were larger (Table 5).

Table 5. Age at indicators of pubertal development according to level of maternal cigarette smoking during pregnancy, restricted analyses (complete cases)
Maternal cigarette smokingFirst nocturnal emission (n =789)Acne (n = 1552)Voice break (n = 1476)Regular shaving (n = 1848)
Age mean (SD)Adjusteda difference (95% CI)Age mean (SD)Adjusteda difference (95% CI)Age mean (SD)Adjusteda difference (95% CI)Age mean (SD)Adjusteda difference (95% CI)
  1. Results from test for trend: emission: p for trend = 0.37, acne: p for trend = 0.06, voice break: p for trend = 0.36, regular shaving: p for trend = 0.15. aAdjusted for maternal age at delivery (continuous), pre-pregnancy BMI (<18.5, 18.5–24.9, >24.9 kg/m2), maternal alcohol drinking during pregnancy (0 drink/week, 0.5–1.5 drinks/week, ≥2.0 drinks/week), chronic diseases of the mother [diabetes mellitus, epilepsy, arthrosis, heart disease, cancer, psychiatric diseases or other chronic diseases, combined into one variable (present or not present)], municipality of residence at the time of delivery (Odense and Aalborg, other places), family's socioeconomic status based on the highest ranking of job description or academic background between the mother and the father at the time of delivery (skilled workers, unskilled or semiskilled workers, unemployed, students), and cohabitation status of the parents (Mother living with the father of the child, Mother not living with the father of the child). The following characteristics were chosen as the referent in the models: maternal age = 25, pre-pregnancy BMI between 18.5–24.9, no maternal alcohol consumption, no chronic diseases of the mother, mothers living in Odense or Aalborg at the time of delivery, family's socioeconomic status = skilled workers, cohabitation status of the parents = Mother living with the father of the child.

Non-smoker14.8 (1.8)Reference14.6 (1.5)Reference14.5 (1.3)Reference17.2 (1.4)Reference
Past smoker14.9 (1.8)0.09 (−0.32; 0.50)14.7 (1.5)0.03 (−0.21; 0.27)14.5 (1.2)−0.05 (−0.27; 0.18)17.3 (1.4)0.08 (−0.14; 0.29)
1–9 cig/day14.9 (2.0)0.12 (−0.25; 0.49)14.7 (1.5)0.14 (−0.07; 0.35)14.6 (1.4)0.11 (−0.09; 0.30)17.2 (1.5)0.09 (−0.09; 0.27)
10–14 cig/day14.7 (1.8)−0.07 (−0.49; 0.35)14.4 (1.4)−0.27 (−0.52; −0.03)14.4 (1.4)−0.18 (−0.41; 0.05)17.2 (1.3)0.06 (−0.15; 0.26)
≥15 cig/day14.4 (1.8)−0.62 (−1.28; 0.04)14.4 (1.3)−0.34 (−0.67; 0.00)14.4 (1.4)−0.13 (−0.44; 0.19)17.3 (1.2)0.21 (−0.08; 0.50)

Finally, repeating the analyses including the men's own smoking and alcohol consumption (potential mediators), showed results in the same direction (data not shown).

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Author contribution
  8. Acknowledgement
  9. Conflict of interest
  10. References

To the best of our knowledge, this is the first large-scale follow-up study based on prospectively collected data on maternal cigarette smoking, investigating the influence of this prenatal exposure on pubertal development in male offspring.

We found no statistically significant association between prenatal cigarette smoking exposure and earlier onset of puberty, but we observed tendencies towards earlier age of first nocturnal emission, acne and voice break, indicating an accelerated age of pubertal development among the men exposed to ≥15 cigarettes/day. However, results on regular shaving were in the opposite direction and caution is needed when interpreting the results, which include considering a substantial bias towards the null based on an expected random misclassification of puberty dates.

Despite the relatively large amount of literature on the association between prenatal cigarette smoking and age at pubertal onset in girls, similar studies in boys have been lacking. The first study by Fried et al. (2001) found that voice break occurred at an earlier age among male offspring exposed to cigarette smoke in utero, although the study was limited by a very small sample size (83 men). More recently, a large cross-sectional study of 3486 men by Ravnborg et al. (2011) also indicated that prenatal exposure to tobacco may lead to earlier voice break, growth of penis and pubic hair development, but their study was limited by the use of retrospectively collected data on prenatal cigarette exposure (binomial, yes/no) through the sons and limited confounder adjustment.

This study addresses some of the limitations of the previous published articles (Fried et al., 2001; Ravnborg et al., 2011). We used prospectively collected data on maternal cigarette smoking with information on the amount of cigarettes smoked per day and a large fraction of heavy smokers, thus giving us the opportunity to study dose-response effects. Moreover, we had the ability to adjust for potential confounders. However, residual confounding or confounding by other unknown factors cannot be ruled out.

A noteworthy limitation of this study is the retrospectively collected, self-reported indicators of pubertal development with a recall time varying from 1 to 12 years. The majority of previous studies on age at onset of puberty among boys have found that the first ejaculation occurred from the age of 13.0–13.4 years (Carlier & Steeno, 1985; Nielsen et al., 1986; Guizar-Vazquez et al., 1992; Veldre & Jurimae, 2004; Campbell et al., 2005), earlier than reported by the men in our study. A study by Schaefer et al. (1990) found that spermache occurred approximately at 14 years. Thus, there are differences in the reported age at spermache. Although age at first nocturnal emission is considered to be a reliable way of measuring puberty in boys compared with menarche in girls (Laron et al., 1980; Carlier & Steeno, 1985), recall is expected to be poor at longer time periods. The mean age of 14.5 years at voice break in our study is also slightly higher than 14.0 years reported by Juul et al. (2007). There is a risk that the discrepancy is owing to the recall time and that all our outcome measures are subject to some misclassification, but because of the sequence of reporting and the novelty of the hypothesis, misclassification will probably be of random nature and bias results in the null direction.

Self-reports of maternal smoking during pregnancy may be prone to misclassification, most likely underestimation. However, in the 1980s, smoking was widely accepted in Denmark, even during pregnancy, thus, the level of misclassification is most likely lower than what is expected now with increasing public health concern of smoking in pregnancy. Moreover, we found the expected effect of smoking on birth weight (Kramer, 1987; Wilcox, 1993): 89 g (95% CI: 75; 103) lower crude mean birth weight per increase in smoking level, which support the validity of the exposure data.

The participation rate in the birth cohort was high (87%), but the response rate at follow-up of the offspring in 2005 was relatively low (55%), and among the responders, there were missing outcome data. There is a risk of selection bias if participation depended on both prenatal cigarette smoking exposure and pubertal development. To cause bias away from the null highly exposed men with earlier onset of puberty must be oversampled and we find this implausible. We have no information on the pubertal development of those who did not respond to the questionnaire in 2005, however, we have data on birth weight of the young men as well as information on maternal smoking during pregnancy. We observed no difference in the distribution of maternal smoking during pregnancy or birth weight between responders and non-responders (data not shown).

Our findings were based on combining multiple regression models adjusted for potential confounders with multiple imputation methods to compensate for missing data. Multiple imputations allow for analyses of the whole study population, and thus it may minimize the risk of selection bias, strengthen the statistical power and provide more precise estimates if the data are missing at random (Rubin, 1987; Sterne et al., 2009). To investigate the credibility of this assumption, we compared results derived from several imputation models, and found that the results from the different approaches were approximately of the same magnitude and in the same direction. Moreover, we compared the results from the restricted analyses (Table 5) with the main analyses (Table 4) and the results were in the same direction, however, attenuated in the main analyses.

The previous studies on this subject (Fried et al., 2001; Ravnborg et al., 2011) also considered the possible effect of current smoking in adulthood on the association studied. In our data, maternal cigarette smoking was associated with the young men's own smoking habits; the high-exposed men were more often smokers than the non-exposed men (32% vs. 18%). However, puberty may well have occurred before smoking initiation and, in our opinion, it can hardly be considered a confounding factor. Instead, it should be considered a possible intermediate factor in the causal path we study, in which case, it would be incorrect to include smoking in adulthood as a covariate. We do not have information on age when the boys began to smoke, however, we performed subanalyses, including the men's smoking and alcohol consumption, and this did not change the results.

Puberty is initiated by an increase in gonadotropin secretion, thus activating the hypothalamo-pituitary-gonadal hormonal axis, resulting in gonadal growth and function. However, the mechanisms behind the increase in gonadotropin secretion remain unknown. We observed a tendency towards earlier onset of pubertal development in highly exposed male offspring, but the cause of this association remains unclear. Ravnborg et al. (2011) put forward the hypothesis that cigarette smoking during pregnancy induced an increase in free testosterone which may result in earlier pubertal development. A previous study by Ramlau-Hansen et al. (2008) on a subsample of 347 of the men from this birth cohort found higher free testosterone with higher levels of prenatal cigarette smoking exposure, and the free testosterone/free estradiol ratio was positively associated with maternal cigarette smoking during pregnancy. If these results apply to the larger study population, and if the levels of testosterone were elevated at the time of puberty initiation, it could possibly explain the accelerated pubertal development.

From a clinical point of view, early age of pubertal development among girls is a risk factor for metabolic syndrome, polycystic ovarian syndrome and insulin resistance in adulthood (Ibanez et al. 2000). Epidemiological studies have reported that an accelerated puberty in boys may be a risk factor for testicular cancer (Moss et al., 1986; Forman et al., 1994; Moller & Skakkebaek, 1996) and prostate cancer (Habel et al., 2000; Giles et al., 2003), however, others failed to find such associations. The implications of an altered age at puberty on male fertility remain unknown.

In summary, we observed a tendency towards an accelerated age of pubertal development among young men exposed to high levels of cigarette smoking during foetal life. Due to the limitations of the self-reported data and the risk of recall bias, the results must be interpreted cautiously. Future studies with more detailed prospectively collected data on pubertal development are needed.

Author contribution

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Author contribution
  8. Acknowledgement
  9. Conflict of interest
  10. References

Contributions to conception and design, data interpretation, manuscript revision and approval of the submitted version: L.B.H., J.O., H.S., A.M.T., A.E., J.L.Z., A.S., C.H.R.H.

Acquisition of data: J.O., A.M.T., C.H.R.H.

Data analysis: L.B.H., J.O., H.S., C.H.R.H.

Drafting the article: L.B.H.

Acknowledgement

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Author contribution
  8. Acknowledgement
  9. Conflict of interest
  10. References

The study was supported by a grant from the Danish Medical Research Council (271-05-0115).

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Author contribution
  8. Acknowledgement
  9. Conflict of interest
  10. References