Unlike our previous study of a California cohort from similar birth years, this study shows a slight delay of almost four months among daughters of heavy prenatal smokers compared to those of non-smokers, after adjustment for other factors. These results also do not confirm other small or more recent studies that reported a lower age at menarche for daughters of smoking mothers, one of which was also conducted in California with an overlapping, but broader, birth cohort. 39–41
These three studies did not ascertain in utero
smoke exposure, but at least some smokers are likely to have smoked during their pregnancies, particularly before the late 1960s or in Poland. Similar to some other studies, in our current study, maternal smoking was crudely associated with smaller child height at ages four and seven, and to some degree lower weight, , but not BMI.33,45
This did not, however, appear to entirely explain the association of smoking with later age at menarche as we adjusted for size, as well as preterm birth. On the other hand, some more recent studies have found that maternal prenatal smoking is associated with obesity or higher ponderal index during childhood. 35,36,46
In an analysis using the entire CPP cohort, maternal smoking during pregnancy was associated with having children more likely to be short and overweight, 34
which might lead to an expectation of earlier, not later, puberty.
We found a number of predictive factors that should be considered in studies of exogenous exposures that may affect puberty. An important predictor is mother’s age at menarche, as shown previously, 7
thereby supporting the validity of this data set. Consistent with the literature, larger size at either age four or seven measured in several ways was related to an earlier age at menarche in this study. Whether body size or amount of adipose tissue is the trigger for menarche has not been clear, although the suggestion that a minimal amount of body fat is required for onset of menses and the role of leptin has led to the hypothesis that the presence of adipose tissue is important for at least one pathway to puberty. 11,12,47
With both weight and height in a model, only weight remained significantly associated with AAM in our study, whereas with BMI and height, both remained significantly associated, indicating body “fatness” is important. Consistent with our results, other studies have shown a relationship of earlier pubertal development with greater weight and height at young ages,38,48,49
and in one, the association with weight appeared independent of height. 49
These studies were primarily of White girls, whereas our association of adiposity at a young age with AAM was more apparent among Blacks than Whites. As birthweight is considered predictive of later weight, a finding of later menarche with low birthweight might be expected. However, a few studies have found the opposite; e.g., an earlier onset of menarche,14,50,51
but not all, especially after controlling for growth during childhood.15,16
We did not find an association with low birthweight in this or our previous study. Rather, girls born prematurely tended to have a later mean age at menarche, somewhat similar to a Swedish study,14
but this was weakened after adjustment.
Our findings with SES were equivocal, although some measures were consistent with the hypothesis of trends towards earlier pubertal onset being related to higher SES.7,8
An interesting finding was an earlier age at menarche among girls who were early in the birth order (low maternal parity) or had few other siblings later in childhood. In a model together these were likely over-adjusted as they are correlated. These factors may reflect higher SES or available resources, but the association remained after adjustment, pointing to potential family structure issues that have been noted previously. 38,52–54
Alternatively, parity may be related to hormonal factors, with higher estrogen levels observed in primiparous women or first versus subsequent pregnancies.55,56
An earlier age at menarche has been observed with greater psychosocial stress, such as parental divorce, but not all studies find effects in the same direction, 57–59
including ours. Alternative hypotheses suggest different physiologic reactions to the presence of biologic versus step parents and siblings to integrate some of these findings.52
The population we studied previously was very different from this primarily low income, Northeastern, Black population. 38
Other differences in this study sample from the previous one include much younger mothers, many more of whom were working during pregnancy. We could control for most of these variables as needed, so there may be related factors that were not measured or measurable to explain the differences in results, including diet, other exposures, and several childhood factors. The predictors of daughter’s age at menarche varied somewhat between this and the prior study, with maternal age, education, and employment status more related in the prior than in this study. The lack of a strong association of age at menarche with race in this cohort is unusual, but was found in only one site. The relationship of early puberty with race seen in the U.S. is unlikely to be entirely genetic, given the heterogeneous population and later ages seen in African and other Black populations, and may more likely reflect lifestyle factors. In the Pathways cohort, White mothers tended to be younger and less educated than Black mothers, but more likely to be married, so various SES factors may be at play. Smoke exposure may interact in some way with either genetic or other factors such as stress that differ between the populations studied.
Limitations of this analysis include lack of data on other prenatal exposures (such as alcohol and caffeine consumption, or lead level) and retrospective reporting of age at menarche. Age at menarche is generally considered to be fairly well-recalled,60
but this study only recorded it to the year of age rather than year and months. In addition, there was no data on paternal smoking or childhood ETS in this study. Results from this cohort of primarily Black girls born in the 1960’s may not be entirely generalizable to current cohorts. Strengths are similar to our previous study, including the prospective collection of data on smoking during pregnancy, births from the 1960s when smoking rates were higher, and longitudinal follow-up of children’s growth.
Although there is inherent biologic variation in onset of menarche, shifts towards earlier or later puberty may signify an insult to the hypothalamic or endocrine systems that results in later reproductive problems. Furthermore, other debilitating chronic diseases in women are hormonally-related, such as cancer and even heart disease, so an alteration in age at reproductive maturation caused by an exposure could potentially have long lasting effects. Therefore it is important to identify factors related to puberty, but the logistical problems of long-term follow-up have made this difficult. More definitive answers require further studies conducted among new cohorts, ideally including longitudinal follow-up from birth to maturation, where other indicators of puberty can be measured, as well as including boys. The role of early exposure to tobacco smoke in influencing age at menarche has not been clarified, but evidence presented here and elsewhere suggests this and other potentially important environmental determinants of age at menarche that should continue to be examined.