Introduction

Chronic diseases have emerged as the leading cause of death in recent decades [1]. Among chronic diseases, hypertension is one of the ten main isolated causes of death worldwide [2]. The large global disease burden of hypertension is due directly to its high prevalence and considerable mortality risk and indirectly through its influence on other associated cardiovascular diseases (e.g., chronic kidney disease, ischemic heart disease, and cerebrovascular diseases) [3]. In Brazil, an estimated 265 thousand people died of hypertension in 2015 [3].

The etiology of hypertension is complex. Multilevel factors, such as race, chronological age, stress, physical activity level, and dietary patterns, have influence on the onset and development of hypertension [4,5,6]. However, one of the strongest proximal risk factors of hypertension is overweight/obesity [4, 5]. Thus, increased body adiposity (especially centrally located obesity) represents a higher risk for the development of hypertension.

In addition to the traditional determinants, early biological maturation has been related to obesity and high blood pressure (BP) in adolescents [7]. However, it is not clear whether changes in biological processes in the early years can have impact on cardiovascular and hemodynamic outcomes across the life span [8,9,10,11,12]. As obesity is directly associated with biological maturation, it has been speculated that the relationship between biological maturation and cardiometabolic risk factors can be mediated by obesity [9, 12, 13]. Although speculative, the mediating role of adiposity on the effect of early maturation on BP in adulthood has not been tested. However, this is especially important given that the onset of biological maturation, especially age at menarche, now occurs at a much earlier age than in previous decades [14]. Moreover, understanding the mediating effect of adiposity on the relationship between early maturation and BP in adulthood could help identify women subgroups that are at risk (based on adiposity status) and those that should be targeted for early intervention. Thus, our aim was to investigate  the associations between biological maturation (based on age at menarche) and adiposity and BP in adulthood and to examine whether adiposity mediates the associations between age at menarche and BP in adulthood.

Materials and methods

Sample

The Brazilian Health Survey (PNS in Portuguese) is a nationally representative cross-sectional epidemiological survey that was conducted in 2013. The survey employed a complex sampling design. Census tracts were first randomly selected from a national plan, households were then randomly selected from a national registry of addresses, and finally, within each household, one randomly selected adult 18 years or older was invited to participate in the study. Interviews were eventually conducted in 64,348 households, which represented an overall response rate of 78%. Person-level survey weights were applied to account for the probability of selection as well as non-response rates [15]. For the present analyses, only data from female participants were explored. Overall, 34,282 women participated in the PNS, but 567 women were excluded due to missing data. Therefore, our final sample size comprised 33,715 women between 18 and 100 years of age. The Brazilian national council of ethics in research approved all procedures, which were in accordance with the Helsinki Declaration.

Age at menarche

Age at menarche was used as an indicator of sexual maturation, as it marks the final stage of sexual maturation [16]. The assessment was conducted through self-report with participants recalling their age at menarche retrospectively. For the descriptive analysis, age at menarche was categorized as ≤ 11, 12, 13, 14, or ≥ 15 years, and continuous data for actual age at menarche were used in the mediation models.

Obesity indicators

Waist circumference (WC) and body mass index (BMI) were adopted as obesity (adiposity) indicators. For measurements, a portable stadiometer, a portable electronic weighing machine and an anthropometric tape were used. All body adiposity measurements were conducted by trained staff. Obesity indicators were categorized according to standard cut-points: WC > 88 cm and BMI ≥ 30 kg/m2 [17].

Blood pressure

The measurement of BP was carried out by trained staff using calibrated digital equipment (TECH® model MA 100). Initially, participants rested for 10 min and then three measurements were taken. Before these measurements, all the participants were instructed to empty their bladders and not to smoke or drink for at least 30 min before the measurements. The measurements were taken with the individual in a seated position and with the right arm supported by a table. The average of the second and third measurements was used for the present study. The adopted cut-point for high systolic BP was ≥ 140 mm Hg and for high diastolic BP was ≥ 90 mm Hg. More detailed information was previously published [18].

Covariates

Participants reported their own chronological age in years and details of their highest education level were used to create three education categories (no academic qualification, high school, and more than high school). Race was self-reported and the information was dichotomized (white and non-white). Leisure-time physical activity was assessed by asking participants if they had (no/yes) performed any sports or exercise in the preceding 3 months and how many days per week, and hours and minutes per day they usually performed sports or physical exercise. Participants who engaged in ≥ 150 min of physical activity per week (weekly frequency multiplied by daily time) were classified as active [19]. Participants further reported how many days per week they usually smoked tobacco; participants were classified as smokers if they engaged in the behavior ≥ 1 day weekly. With regards to sodium consumption, a five-point Likert scale was used to assess whether participants perceived their diet was characterized by “a little” to “a lot” of salt (diets containing “quite a lot” and “a lot” were considered to be indicative of high salt intake). In addition, participants reported how many days per week they usually consumed alcohol and their responses were categorized as follows: none, moderate (1–3 days per week), and heavy (4 or more days per week). Daily TV viewing was self-reported by participants in response to the question “How many hours a day do you usually spend watching TV?” One of the following responses was permitted: (a) Less than 1 h; (b) More than 1 h but less than 2 h; (c) More than 2 h but less than 3 h; (d) More than 3 h but less than 4 h; (e) More than 4 h but less than 5 h; (f) More than 5 h but less than 6 h; (g) More than 6 h; or (h) I do not watch TV. The data were used to create three categories of daily TV viewing time (less than 2 h; between 2 h and 4 h; and more than 4 h). The participants also reported if they had already entered menopause and if they used any antihypertensive medication. Responses to these questions were treated as dichotomous variables (no or yes). All covariates were treated as controlled variables in the main analyses.

Statistical analysis

Means and 95% confidence intervals were used to describe the sample characteristics. Values with non-overlapping confidence intervals were considered significantly different [20, 21]. Pearson’s correlations and adjusted linear regression analyses were used to assess the relationships between age at menarche and adiposity indicators (BMI and WC) and BP (systolic and diastolic BP). In the regression analysis, model 1 tested the association of age at menarche with BP, adjusted for chronological age, race, menopause status, antihypertensive medication use, education status, smoking status, leisure physical activity, TV viewing, and alcohol consumption. Model 2 tested the association of WC with BP, including all the covariates that were adjusted for in model 1, whereas model 3 tested the association between BMI and BP, including all the covariates that were adjusted for in models 1 and 2. As women usually gain weight after menopause, the correlation and regression analyses were conducted overall and stratified by menopause status (i.e., before menopause and after menopause).

The mediation analysis was performed according to the principles of Baron and Kenny [22]. In the first instance, the mediator variable (WC and BMI) was regressed onto the independent variable (age at menarche). In the second equation, the dependent variable (systolic and diastolic BP) was regressed onto the independent variable. Finally, in the third equation, the dependent variable was regressed onto the independent variable, adjusted for the mediator. Mediating effects were identified if the following criteria were met: (a) in the first equation, the independent variable was a significant predictor of the mediator (if age at menarche was significantly associated with WC/BMI); (b) in the second equation, the independent variable was a significant predictor of the dependent variable (if age at menarche was significantly associated with systolic/diastolic BP); (c) in the third equation, the mediator was a significant predictor of the dependent variable and the association between the dependent and independent variables (from equation two) was either partially or fully removed (if WC/BMI was significantly associated with systolic/diastolic BP and if this association accounted for part or all of the significance of the association between age at menarche and systolic/diastolic BP). All covariates were adjusted for in the mediation models.

Analyses including the indirect and direct effects with 95% confidence intervals were calculated using structural equation modeling building in STATA 14 using the maximum likelihood method, with a 5% significance level. Analyses were conducted during 2017.

Results

The final sample consisted of 33,715 women with an average age of 43.2 years (95% confidence interval from 42.9 to 43.6). The characteristics of the sample according to age at menarche are presented in Table 1. In general, chronological age was greater among the groups with a greater age at menarche than in those with a younger age at menarche. Moreover, the group of women who were ≤ 11 years old at menarche presented a greater average WC and BMI than their older counterparts. The group of women who were ≥ 15 y of age at menarche presented the greatest prevalence of women who had already entered menopause. Descriptive relationships between age at menarche and WC, BMI, and BP are presented in Fig. 1. Both adiposity indicators decreased with a higher age at menarche, whereas a weak association was observed between age at menarche and BP. Using categorical data, a greater prevalence of obesity was found among women with an earlier age at menarche, especially those who were ≤ 11 years of age at menarche (Fig. 2).

Table 1 Characteristics of the sample according to age at menarche
Full size table
Fig. 1
figure 1

Waist circumference, body mass index, and blood pressure according to age at menarche

Full size image
Fig. 2
figure 2

Prevalence of outcomes (systolic blood pressure and diastolic blood pressure) and mediators (body mass index and waist circumference) according to age at menarche

Full size image

Correlations between independent, mediators, and dependent variables, according to menopause status, are presented in Table 2. All correlations were small but significant, with the exception of the correlation between age at menarche and diastolic BP after menopause. Moreover, the correlations between age at menarche with mediators and outcomes were weaker after menopause. In this sense, age at menarche was inversely associated with obesity indicators as well as BP (p < 0.001), whereas obesity indicators were positively associated with BP. After this analysis, the mediating effect of obesity indicators on the relationship between age at menarche and BP was tested (Fig. 3). In general, WC and BMI partly mediated the relationship between age at menarche and systolic BP (accounting for 41% and 44% of the effects, respectively), given that the mediators (WC and BMI) accounted for part of the effects and that age at menarche remained significant in equation 3. On the other hand, the relationship between age at menarche and diastolic BP was fully mediated by WC and BMI, given that age at menarche was no longer significant in equation 3 (with the addition of the obesity indicator).

Table 2 Partial correlations between dependent, independent and mediator variables among whole sample and according to menopause status
Full size table
Fig. 3
figure 3

Mediation models of waist circumference and body mass index on the relationship between age at menarche and blood pressure. Note Adjusted by chronological age, race, menopause status, hypertension drug use, educational status, tobacco smoking, leisure physical activity, TV viewing, sodium consumption and alcohol consumption

Full size image

The results of the linear regression analyses stratified by menopause status using collapsed age at menarche ( ≤ 11, 12, 13, 14, ≥ 15 years) are presented in Table 3. The negative associations between age at menarche and systolic (β = − 0.620, P = 0.008) and diastolic (β = − 0.477, P = 0.005) BP were significant only among pre-menopausal women. In addition, WC significantly contributed to predicting systolic (R2 = 0.170, P = 0.046) and diastolic (R2 = 0.160, P = 0.025) BP only among pre-menopausal women.

Table 3 Regression model for prediction of systolic and diastolic blood pressure by age at menarche
Full size table

Moreover, we tested the interaction between race and age at menarche, as well as the interaction between antihypertensive medication use and age at menarche in predicting systolic and diastolic BP. The interaction term of race*age at menarche was not significant for predicting either systolic BP (WC adjusted p = 0.951, BMI adjusted p = 0.903) or diastolic BP (WC adjusted p = 0.879, BMI adjusted p = 0.749). In addition, the interaction term of antihypertensive medication*age at menarche was not significant for predicting either systolic BP (WC adjusted p = 0.773, BMI adjusted p = 0.823) or diastolic BP (WC adjusted p = 0.516, BMI adjusted p = 0.484).

Discussion

This study investigated the association between age at menarche and body adiposity and BP in adulthood and examined the mediating effects of body adiposity on the relationship between age at menarche and BP. The main findings were that early maturation was related to obesity and higher BP in adulthood and that greater WC and BMI partly or fully mediated the relationship between early maturation and BP in adulthood. In general, this finding is consistent with the results of previous studies that suggest that earlier biological maturation is a risk factor for several negative health outcomes in late adolescence and adulthood [8, 23, 24]. Many previous studies support the conclusion that early biological maturation or puberty is related to obesity [8,9,10] and other metabolic risk factors [8, 11, 13]. On a somewhat similar note, previous studies have also suggested that the relationship between maturation and cardiovascular risk factors seems to be mediated by obesity-related variables, especially lipid variables [9, 13, 24, 25].

Our findings suggest the likelihood of a dose–response relationship between age at menarche and obesity indicators (WC and BMI). The age of 12 years appears to be the point at which there is a decline in obesity risk, as obesity indicators among the participants dropped from a high point at 8 years at menarche to a lower point at approximately 12 years of menarche, where they remained stable. This finding seems to be consistent with the literature that classifies menarche at less than 12 years old as early menarche and that found early menarche to be associated with obesity in adulthood [9, 26, 27]. This association could be explained by the fact that early maturing girls already have greater adiposity before puberty, given that adipose tissue, through leptin release, has a role in the initiation of the biological maturation process [28, 29]. Considering the potential tracking of adiposity from adolescence to adulthood [30], early maturing girls could maintain their higher adiposity in adulthood [31]. Another mechanism could be related to the time of the maturation process. There is evidence that early maturing adolescents experience a more accelerated maturation process than late maturing adolescents [32]. Thus, greater interference with biological and behavioral pathways in adolescents with faster maturation can lead to increased adiposity [32, 33].

We also found a significant inverse relationship between age at menarche and BP in adulthood, although this association was weak. This finding corroborates previous findings regarding other cardiovascular risk factors [8, 11] and is especially relevant given that the measure of BP was, on average, 30 years after the age at menarche. The weaker linear association between age at menarche and BP in adulthood could, in part, be explained by the fact that early maturers were younger than women with on time/late menarche. This finding may be a reflection of the fact that BP is an outcome with a higher latency period than adiposity.

The main novelty of our study is the finding that obesity indicators partly mediated the relationship between age at menarche and systolic BP, and fully mediated the relationship between age at menarche and diastolic BP in adulthood. Obesity could more strongly mediate the association between age at menarche and diastolic BP than systolic BP, because BMI is strongly associated with diastolic function [34, 35]. Nevertheless, our findings confirm the assumption and previous speculation that obesity could be a mediator of the association between age at menarche and onset of chronic diseases [9, 12, 13]. Considering that it is already clear that subjects with higher levels of adiposity present with greater BP [4], and that age at menarche is a risk factor for high adiposity in adulthood, it is biologically plausible that early maturation could influence BP in adulthood through greater body adiposity. However, greater adiposity can be associated with higher BP through different paths, such as increased peripheral resistance, increased sympathetic nervous system activity, renin–angiotensin–aldosterone system, and stimulation of negative hormones for BP [36].

Interestingly, previous studies that adjusted for important covariates and confounders of high BP, including lifestyle, socio-demographic factors and chronological age, suggest that, at least for systolic BP, there are other potential unobserved/unmeasured variables beyond adiposity that can mediate the effect of age at menarche[5, 6]. However, our findings suggest that obesity, due to its inverse relationship with age at menarche, is an important mediator of both systolic and diastolic BP in adulthood. In the present study, the association between age at menarche and diastolic BP was weaker; thus, obesity indicators could exert a greater influence. Moreover, among women who already entered menopause, the effects of menarche can be even weaker, as our results suggest, and obesity can present a greater effect on BP. Therefore, obesity in childhood/adolescence [31] or even behavioral factors through life (the relationship between physical activity and biological maturation) [37] are potential mediators to target when designing future and current interventions for BP control and management.

Our results have clear practical implications and applications. Given the adverse relationship between early age at menarche and body adiposity and BP, early interventions that aim to reduce body adiposity, even in childhood, should be promoted [38]. Moreover, keeping in mind that obesity indicators during childhood mediate the relationship between age at menarche and BP, interventions focusing on obese girls and women should be implemented [39, 40]. Furthermore, given that greater BMI maintenance and weight gain are associated with a greater likelihood of high BP [41], interventions aiming to avoid obesity throughout the life span should also be incorporated as part of public health strategies to control hypertension in women.

The main limitations of the current study are related to the cross-sectional nature of the study and the recall method used for the assessment of age at menarche. However, the results provided baseline data that could be explored further in future prospective studies. Moreover, the recall assessment method has good validity with a low error and bias of less than 1 year [42, 43]. A major strength of the study is the analysis of objectively measured obesity indicators and BP in a large sample of more than 34,000 Brazilian women. Another strength is the analytical method that controlled for potential covariates and confounders, such as chronological age, race, educational status, physical activity level, TV viewing, tobacco smoking, alcohol consumption, sodium ingestion and antihypertension medication use [5, 6].

In conclusion, age at menarche was inversely related with adiposity and BP in adulthood. Early maturation could predict obesity in adulthood, and adiposity indicators mediate the relationship between age at menarche and BP in adulthood. Obesity is an important risk factor to target to reduce the negative influence of early maturation on high BP in women.

Informed consent

Informed consent was obtained from all participants included in the study.

Data sharing statement

All data are open published as microdata at http://www.ibge.gov.br/home/.