Age at menarche association with metabolic, reproductive, and sociodemographic factors: a population-based cross-sectional analysis using an outcome-wide framework in Geneva, Switzerland.

Menarche, the onset of menstruation, usually occurs between ages 10 and 16 [ 1 ]. The average age at menarche has declined since the 1930s, with variations across countries [ 2 ]. This decline has mainly been attributed to improved nutritional status [ 3 ], particularly higher childhood body mass index (BMI) [ 4 ], possibly via leptin-mediated effects on puberty [ 5 ], and high intakes of animal protein [ 6 ], which have been increasingly consumed over the past 20 years [ 7 ]. Other influences include exposure to endocrine-disrupting chemicals (e.g. bisphenol A (BPA) and phthalates [ 8 ]), socioeconomic status, psychosocial stressors (including adverse childhood experiences and unstable family environments [ 9 ]), urban lifestyle [ 10 ] and cultural norms [ 11 ]. The reproductive lifespan, from menarche to menopause, globally averaging 35.8 years [ 12 ], is a critical period influencing long-term health outcomes [ 13 ]. Prolonged hormonal exposure may benefit cardiovascular [ 14 ] and bone health [ 15 ], but can raise risks of hormone-sensitive cancers [ 16 ]. Classification of menarche (early, normal, and late) has been commonly used to facilitate risk factor association [ 1 ]. A recent umbrella review reported strong evidence linking early menarche (<12 years) to metabolic syndrome, endometrial cancer, and higher adult BMI [ 17 ]. Other studies have demonstrated that women with early menarche are also at greater risks of various types of cancers [ 18 ], cardiovascular diseases [ 19 ], metabolic diseases [ 20 ], and premature mortality of all causes [ 19 ]. Late menarche, although less studied, has been linked to higher odds of infertility (not statistically significant) [ 21 ], and emerging evidence suggests association with higher risks of dementia [ 22 ], osteoporosis [ 23 ], or adult-onset asthma (with a U-shaped relationship since early menarche is also associated) [ 24 ]. Conversely, it appears protective against hormone-sensitive cancers such as breast [ 25 ] and endometrial cancer [ 26 ]. Despite growing interest, most research has examined menarche timing in relation to single outcomes, such as cancer [ 25 ] or metabolic risk [ 27 ], limiting understanding of its broader, interconnected effects on women’s health. Menstrual health and age at menarche are a fundamental part of preventative and primary care and should be integrated in routine primary and pediatric care [ 28 ]. However, studies show that primary care physicians do not routinely provide guidance on menstruation [ 29 ]. To inform preventive care using age at menarche as a clinical indicator, we used a population-based, cross-sectional study from Geneva, Switzerland, a unique high-income and international setting, with two precise aims. First, to assess 50-year trends in age at menarche, and second, to examine associations with sociodemographic characteristics, metabolic profiles, and reproductive health outcomes using an outcome-wide framework (particularly suited to assess the association strength across multiple outcomes).

Data originated from the Bus Santé study, a cross-sectional population-based survey of Geneva residents since 1993. Each year, ∼1000 participants are recruited using sex- and age-stratified random sampling from the official registry, as described previously [ 30 ]. Recruitment was suspended between 2020 and 2023 due to the COVID-19 pandemic. Average participation rate was 49% [ 30 ]. In this study, we selected women aged 34–75 years, recruited between January 2005 and May 2025, as the question regarding menarche timing was only introduced in 2005. Participants reporting menarches before age 4 or after age 20 were excluded to reduce the influence of data entry errors or rare endocrine pathologies. The study protocol is compliant with the Declaration of Helsinki and was approved by the Geneva Ethics Committee (IRB00003116, PB2016-00363 and CCER2022-01544). All participants gave written informed consent. Participants completed questionnaires covering socioeconomic status, lifestyle, reproductive history, and general health. All information are collected at the time of the survey; older participants may report events that occurred further in the past. Trained nurses conducted a clinical examination including weight, height, waist, and hip circumference. Retrospectively, self-reported during the survey, age at menarche (years) was treated either as continuous or categorized as early (14). This classification is widely used in the literature [ 1 ] and supported by distribution and percentile analyses of the Geneva population ( Supplementary Fig. S1 ). Education (primary/secondary vs tertiary), income (gross monthly household income: low < 7000CHF vs high [ 30 ], with categories unaffected by inflation in the past 20 years [ 31 ]), relationship status (single vs in a relationship), Swiss-born (Swiss vs foreign-born), number of siblings (only available for 2005–2019). Self-reported lifetime diagnosis of diabetes, hypercholesterolemia, hypertension; nursemeasured BMI, height and waist-to-hip ratio. Self-reported variables included mean menstrual cycle length and period duration (averages of reported minimum and maximum values), age at first pregnancy, number of pregnancies (including zero), and reproductive years (menopause age minus menarche age). Binary outcomes recorded as ever experienced included breast cancer, oophorectomy, hysterectomy, nulliparity (no pregnancy), age at first pregnancy, and pregnancy outcomes (live birth, stillbirth, ectopic pregnancy, miscarriage, abortion). Descriptive statistics were calculated for all variables. Group comparisons across menarche categories used Kruskal–Wallis tests ( post hoc Dunn’s) for continuous variables and chi-square tests for categorical variables. Menarche trend by birthdate was assessed via linear regression, including participants born 1936–1989 (ages 34–75 at survey, excluding years with <15 participants). Multivariable linear regression was used with age at menarche as the outcome variable and age at survey, Swiss birth status, and number of siblings as independent variables. We applied VanderWeele’s outcome-wide epidemiology framework [ 32 ]. This approach is particularly suited to settings in which multiple outcomes are examined in relation to a single exposure (in this case, age at menarche) and allows for a systematic comparison of the magnitude and robustness of associations across outcomes. Associations between age at menarche and outcomes in sociodemographic, metabolic, and reproductive domains were examined using linear regression for continuous variables [(adjusted) a β , 95% confidence interval (CI)], and logistic regression for binary outcomes (adjusted odds ratios aOR, 95% CI). We considered metabolic outcomes as primary given stronger prior evidence, while reproductive and sociodemographic domains were secondary/exploratory. In line with an outcome-wide framework, all models were consistently adjusted for age, education level, and survey year. Further details on confounder selection and the underlying conceptual causal structure are provided in Supplementary Fig. S2 . To ensure comparison across outcomes we calculated standardized effect size, linear models using standardized β (Cohen’s d equivalent), and logistic models by converting ORs to d using d  = log(OR) × (√3/π) as proposed by Chinn [ 33 ]. To assess the robustness of observed associations to potential unmeasured confounding factors, we calculated E -values [ 34 ] by converting standardized effect sizes to RR where needed (RR ≈ exp(0.91 ×  d )). For instance, an E -value of 2 means that a risk ratio of 2 is the minimum strength of association an unmeasured confounder must have with both exposure and outcome to nullify the observed effect. Statistical significance was set at α  = 0.05; FDR-adjusted P- value (Benjamini–Hochberg) were used for multiple comparisons. Findings were categorized into five evidence tiers to guide interpretation, without implying definitive causal strength. Robust: required FDR < 0.05, a meaningful | d  | ≥0.20 [ 35 ], and E -value ≥ 2.0 (threshold introduced by VanderWeele and Ding [ 34 ] as a benchmark for meaningful confounding robustness). Probable: met FDR < 0.10, | d | ≥ 0.15, and E -value ≥ 1.75, acknowledging suggestive but less certain evidence. Exploratory: are hypothesis-generating with FDR < 0.20, | d | ≥ 0.10 (can represents the lower bound of a meaningful effect in large observational studies [ 36 ]) and E -value ≥ 1.5. Preliminary: had nominal P  < 0.05 but did not survive FDR correction, E -value < 1.5, warranting caution. Null: ( P  ≥ 0.05) are reported descriptively only. Analyses were performed in R v4.4.2 and RStudio v2025.05.0 using dplyr , tidyr , gtsummary , FSA , stats , and ggplot2 . OpenAI-ChatGPT (version GPT-4-turbo) was used to improve readability and language of the text with oversight of the authors.

The study population included 6462 participants, with a mean age of 51.8 years (SD = 11), and a mean age at menarche of 13.1 years (SD = 1.6). Almost half (46.2%) completed tertiary education, and less than half (46.1%) were born in Switzerland ( Table 1 ). Characteristics of women participants aged 35–74 in the Bus Santé study between 2005 and 2025 n (%); mean (SD). Pearson’s Chi-squared test; Kruskal–Wallis rank sum test. All P- value remained significant ( q  < 0.05) after FDR (BH) correction. Early vs late is significant ( P  < 0.05) after post hoc Dunn’s test, when Kruskal–Wallis rank sum test is significant. Late vs normal is significant ( P  < 0.05) after post hoc Dunn’s test, when Kruskal–Wallis rank sum test is significant. Early vs normal is significant ( P  < 0.05) after post hoc Dunn’s test, when Kruskal–Wallis rank sum test is significant. The N for individual variable is specified only if more than 2.5% of the information was missing. Information about the number of siblings was not collected after 2019. A modest declining trend in mean age at menarche was observed from 14.06 years old (SD = 2.07), for women born in 1936, to 12.89 years old (SD = 1.74) for those born in 1989, associated with overall limited variance explained ( β  = −0.008; R 2  = 0.003; P  < 0.001) ( Fig. 1 ). Trend in the mean age at menarche for participants of the Bus Santé study born between 1936 and 1989. Red regression line is fitted using a linear model. 95% confidence intervals are shown around each line and annotation above the plot provides key statistics from the linear regression. In multivariable linear regression, Swiss-born women had a significantly later age at menarche compared to foreign-born women (adjusted β  = 0.177; 95% CI: 0.089–0.266; P  < 0.001). Number of siblings was positively associated with age at menarche, with each additional sibling corresponding to a 0.062-year increase (95% CI: 0.043–0.081; P  < 0.001). The model explained 1.1% of the variance in age at menarche ( R 2  = 0.011) ( Supplementary Table S1 ). Late menarche was associated with lower education levels (aOR = 0.81; 95% CI: 0.71–0.93), and early menarche with lower income (aOR = 0.79; 95% CI: 0.67–0.93), compared to normal menarche. Age-specific analyses were performed to account for the slight overrepresentation of younger women in the early menarche group and possible career-stage effects. The association with lower income persisted mainly in the older age group, indicating a stable pattern ( Supplementary Table S3 ). No significant associations were found between menarche categories and relationship status ( Fig. 2 ). Association between early/late menarche timing and socioeconomic and metabolic outcomes. Forest plot showing standardized effect sizes with 95% confidence intervals for the associations between timing of menarche (early or late vs normal) and various metabolic outcomes in adulthood. The reference group is indicated on the y -axis. A table with the corresponding statistical analyses is presented on the left side of the plot. Results are highlighted in bold if significant after FDR correction (adjusted P  < .05) and if the E -value is reasonably robust to unmeasured confounding ( E -value < 2). The Robustness column indicates the strength of evidence for each association, as described in the “Methods” section. Women with early menarche had a higher prevalence of diabetes (7.9%), hypercholesterolemia (28.4%), and hypertension (26.4%) compared to those with normal (5.8%, 24.5%, 21.7%) or late menarche (5.4%, 26.1%, 23.3%) ( Supplementary Table S2 ). After adjusting for age, education, and survey year, early menarche was associated with increased odds of diabetes (aOR: 1.40, 95% CI: 1.05–1.83), hypercholesterolemia (aOR: 1.27, 95% CI: 1.07–1.49), and hypertension (aOR: 1.41, 95% CI: 1.18–1.67). It was also linked to shorter stature (−2.12 cm, 95% CI: −2.58 to −1.65), higher BMI (a β : 1.20, 95% CI: 0.84–1.55), and a slightly higher waist-to-hip ratio (standardized effect size: 0.08, 95% CI: 0.01–0.15, adjusted P  = .022). Conversely, late menarche was associated with taller stature (a β : 1.03, 95% CI: 0.61–1.45) and lower BMI (a β : −0.79, 95% CI: −1.11 to −0.48) ( Fig. 2 ). Women with late menarche demonstrated shorter reproductive lifespans compared to their counterparts (mean: 33.1 years vs 36.2 years for normal menarche and 37.6 years for early menarche; P  < .001) ( Supplementary Table S2 ). This association had the largest effect size in the outcome-wide models (standardized effect size: −0.57; 95% CI: −0.65 to −0.48; E -value: 2.74). Early menarche was associated with a longer reproductive lifespan (standardized effect size: 0.29; 95% CI: 0.19–0.39; E -value: 1.93), higher odds of hysterectomy (aOR: 1.69; 95% CI: 1.33–2.13) and similar trend for oophorectomy, although not significant after FDR correction. Age at breast cancer onset was, on average, 2.49 years earlier in women with early menarche (although not significant) but did not affect overall breast cancer rates ( Fig. 3 ). Association between early/late menarche timing and adult reproductive outcomes. Forest plot showing standardized effect sizes with 95% confidence intervals for the associations between timing of menarche (early or late vs normal) and various reproductive outcomes in adulthood. The reference group is indicated on the y -axis. A table with the corresponding statistical analyses is presented on the left side of the plot. Significant results after FDR correction (adjusted P  < .05) are in bold and a * indicates nominal significance only ( P  < .05). E -value reasonably robust to unmeasured confounding ( E -value < 2) are also highlighted in bold. The Robustness column indicates the strength of evidence for each association, as described in the Methods section. Late menarche was associated with distinct menstrual cycle patterns, including longer cycle length (a β  = 0.41 days; 95% CI: 0.10–0.69) and shorter bleeding duration (a β  = −0.20 days; 95% CI: −0.32 to −0.06) relative to normal menarche. Early menarche showed opposite trends, though not significant. Women with early menarche tended to have a younger age at first pregnancy (a β  = −0.56 years vs the normal menarche group), though this did not remain after FDR correction. There were no differences in the number of pregnancies or nulliparity. Late menarche was associated with stillbirth and ectopic pregnancy (aOR: 1.80; 95% CI: 1.12–2.84 and aOR: 1.45; 95% CI: 1.04–2.00, respectively), however not significant after FDR correction. No differences in miscarriage or abortion rates were observed ( Fig. 3 ). Overall, potential cultural, nutritional, and socioeconomic differences across the extended data collection period did not modify the association between age at menarche and the outcomes of interest, as shown by estimates consistency across models adjusted for age alone or for both age and survey year ( Supplementary Table S4 ).

This is the first population-based analysis of age at menarche in Geneva, confirming the global trend toward earlier onset, albeit less pronounced in this high-income setting. Using an outcome-wide analytical approach, we identified key associations between menarche timing and later-life socioeconomic, metabolic, and reproductive outcomes. The strongest association was with the reproductive timing, which may partly contribute to the observed health consequences across the life course. Earlier menarche was associated with prolonged hormonal exposure which in turn showed association with increased odds of hysterectomy, oophorectomy, and metabolic disorders. In contrast, late menarche was associated with a shorter reproductive window, possible ovulatory differences and was linked to higher odds of pregnancy complications such as stillbirth and ectopic pregnancy. These findings underscore the importance of menarche timing as a marker of long-term health. Routine assessment and documentation of the age at menarche in primary care may help improve risk stratification and guide preventive care throughout the life course.

This study examined trends in age at menarche over the past 20 years in a population-based study in Geneva, and assessed its associations with sociodemographic, metabolic, and reproductive health outcomes. Our findings reflect a highly educated and international sample, with a mean age at menarche of 13.14 years. A gradual decline in age at menarche was observed in the Geneva population over the study period. This pattern is consistent with global trends, although the pace and extent of decline vary considerably between low- and high-income countries [ 2 , 37 , 38 ]. Globally, this shift is commonly attributed to improvements in socioeconomic conditions and nutritional status [ 2 , 39 , 40 ]. Interestingly, a European multicenter study reported a steeper decline ( β  = −0.12; SD = 0.002) [41], compared to the more modest trend observed in Geneva ( β  = −0.008), possibly reflecting earlier stabilization of higher socioeconomic conditions. In addition, other factors such as increased urbanization [42] or changes in the genetic background of immigrant populations [43] may also contribute to the observed trends. Various factors may explain the downward trends in age at menarche. Among them improved nutrition and socioeconomic status have been frequently associated, but other factors such as birth and infancy weight [44], genetic predisposition [43], environmental factors (e.g. endocrine-disrupting chemicals) [ 8 ], and urban living [42] were associated with menarche onset. Our results show that Swiss-born women and those with larger families experienced late menarche. These findings align with existing literature documenting population-specific variations in pubertal timing [45, 46], with European populations generally experiencing later onset compared to populations of African or Hispanic descent [47]. The association with family size may reflect the “resource competition hypothesis,” where limited parental resources are distributed among children, affecting individual growth trajectories and nutritional status during critical developmental periods [48, 49]. Yet, the associated R 2 of 1.1% for these predictors is notably low, indicating that most variation in menarcheal timing likely stems from other unmeasured factors, underscoring the complexity of the determinants of reproductive timing. Associations between age at menarche and sociodemographic outcomes remain inconsistent in the literature [50, 51]. In our study, late menarche was associated with lower odds of tertiary education, although this relationship appeared susceptible to unmeasured confounding, consistent with previous observations that association strength depends on unaccounted covariates. Early menarche, conversely, was associated with lower income, independent of education. These findings indicate that menarche timing may have associations with socioeconomic trajectories in ways not entirely captured by formal education. Further analyses controlling for childhood and early-life circumstances, parental socioeconomic conditions, and psychosocial stressors are needed to strengthen these findings. Accordingly, these associations should be interpreted cautiously as indicators of social patterning rather than evidence of causal effects. Nevertheless, they provide preliminary insight into the potential long-term social impact of menarche timing and underscore the need for dedicated studies. Our findings confirm previously established associations between early menarche and increased risk of cardiometabolic diseases in adulthood [52, 53]. Specifically, our observed 41% increased odds of hypertension compare favorably to the 25% higher odds reported across 17 studies [52], while our 40% increased odds of diabetes are consistent with a 39% higher risk of type 2 diabetes identified across 28 studies [53]. We additionally observed 27% increased odds of hypercholesterolemia, though direct comparisons are limited by the absence of analogous estimates in the existing literature. Interestingly, age-specific analysis showed that the increased odds of diabetes and hypertension are most pronounced in the younger group, suggesting a possible link between early menarche and earlier onset of these conditions. Recent analyses have identified a U-shaped association between age at menarche and cardiovascular risk, with a moderately increased risk observed among women with late menarche [ 27 ]. This pattern was not observed in our study population, and late menarche was overall not associated with cardiometabolic risks. Anthropometric analyses revealed significant associations between menarche timing and adult body composition. Early menarche was associated with higher BMI (+1.20 kg/m 2 ) and shorter adult height (−2.12 cm) compared to the normal menarche group. These findings align with longitudinal studies indicating that while early and late maturing girls tend to reach menarche at similar weights, early maturers often experience earlier growth cessation [54]. Moreover, a Mendelian randomization study confirmed that early menarche has a causal effect on increased adult BMI, independent of childhood BMI [55]. The effect sizes for BMI and height associations were substantially larger than those observed for waist-to-hip ratio (effect size: 0.08 vs 0.24 for BMI), suggesting that the apparent association between early menarche and elevated adult BMI may be partially confounded by the concurrent effect on adult stature. This finding indicates that the relationship between early menarche and obesity risk may be more complex than previously characterized, with height-related factors serving as important mediating variables. Conversely, late menarche demonstrated significant associations with both increased height and reduced BMI (effect sizes: 0.15 and −0.16, respectively), indicating that both extremes of menarche timing exert substantial effects on adult anthropometric characteristics. We found no significant differences in number of pregnancies or nulliparity between the groups, suggesting no strong association between menarche timing and overall fertility. However, late menarche was associated with specific adverse outcomes, including higher odds of stillbirth and ectopic pregnancy. Although these associations were not significant after FDR correction, they demonstrated moderate to strong E -values suggesting potential clinical significance. Biologically, late menarche is associated with prolonged prepubertal endocrine maturation, resulting in suboptimal endometrial development and placental implantation processes that could predispose to specific pregnancy complications, including stillbirth and ectopic implantation [56–58]. Although the evidence remains preliminary, our results indicate that women with late menarche exhibit longer cycle duration and shorter bleeding duration compared to normal menarche timing, consistent with suboptimal endometrial development and ovulatory irregularity [59]. Our findings suggest these developmental differences may persist into adulthood, manifesting as specific reproductive vulnerabilities rather than generalized infertility. However, given the exploratory nature of the results, dedicated prospective studies are needed before firm conclusions can be drawn. Early menarche, was associated with 69% higher odds of reporting hysterectomy and 47% higher odds of reporting oophorectomy (although not reaching statistical significance after FDR correction), potentially due to associations between early menarche and increased incidence of ovarian and uterine cancers, as well as endometriosis [60–62]. Although early menarche is often linked to higher breast cancer incidence [63, 64], we did not observe this association. Instead, we found a non-significant trend toward breast cancer occurring 2.49 years earlier (limited sample size). Given the relatively older age of our population (mean age 51.8 years), this may reflect a shift toward earlier onset rather than increased overall incidence. The prevailing hypothesis attributes these risks to prolonged exposure to endogenous estrogen and progesterone. Early menarche extends hormonal cycling duration, promoting cellular proliferation in hormone-sensitive tissues, including breast, endometrial, and ovarian epithelia. Metabolic factors associated with early menarche such as insulin resistance, chronic inflammation, and altered adipokine profiles may further elevate cancer risk [ 17 , 18 ]. Women with early menarche experienced a 1.59-year longer reproductive lifespan, while those with late menarche had 3.12 fewer reproductive years compared with normal menarche. These results showed strong effect sizes and robust E -values, supporting the idea that early or late menarche critically shapes cumulative hormonal exposure and potential long-term reproductive and health associated outcomes. Recognizing the health implications of menarche timing is essential and should be more effectively integrated into primary care routine consultations. Several authoritative guidelines [65–68] emphasize systematically recording menstrual health data however this lacks integration in routine clinical practice. A national survey in the USA, showed that up to one-third of pediatricians do not routinely address adolescent menstruation [69]. Our results support documenting age at menarche during routine primary care consultations, improving understanding of menstrual health and risk stratification for metabolic and reproductive health outcomes. Enhanced dialogue around menstrual health and its monitoring in primary care can reinforce preventive health strategies throughout women’s life course. Several limitations warrant consideration. First, the reliance on self-reported age at menarche introduces potential recall bias, particularly in older participants, as previously described [70]. We attempted to mitigate this bias by adjusting all associations for age at survey. Second, given the population-based design of the study and the voluntary nature of participation, we acknowledge that women with severe conditions may be underrepresented. Also, although we aimed for a comprehensive analysis, some outcomes known to be associated with age at menarche, such as depression, dementia, and osteoporosis were not included. Additionally, early-life factors that may influence both menarche timing and later health outcomes (e.g. birth weight, childhood BMI, parental socioeconomic status, genetic traits) were unavailable. Nevertheless, the calculation of E -values helped identify associations with the highest susceptibility to unmeasured confounding, highlighting variables that warrant further investigation. Also, despite a reasonable overall sample size of 6462 women, some analyses require independent confirmation due to the low incidence of certain events. That said, we believe that some associations can be considered confirmatory in nature, as they are supported by previously reported findings in the literature, such as the association between early menarche and increased odds of hypertension, hypercholesterolemia, diabetes, as well as greater height and BMI. In contrast, other findings, such as the association of late menarche with lower educational attainment, the lower odds of higher income associated with early menarche, and the association between late menarche and adverse reproductive outcomes (e.g. stillbirth and ectopic pregnancy), should be considered exploratory and therefore interpreted with greater caution pending replication.