Age at menarche and its association with blood pressure in adult women of developing countries: a systematic review and meta-analysis

Abstract

Context

Evidence about the effect of age at menarche (AAM) on blood pressure (BP) has largely been drawn from studies in developed countries. Studies in developing countries are expanding recently but have not been summarised.

Objective

To systematically explore the association between AAM and BP and the potential modifiers in developing countries.

Methods

We searched PubMed, Embase, and Web of Science for publications until March 2022. A random-effects model was used to calculate the pooled relative risk (RR) with 95% confidence interval (CI).

Results

Twenty studies were eligible. In studies with participants’ mean age at BP assessment <55 years, women in the oldest group as compared with the middle or the youngest group of AAM had a higher risk of hypertension in those studies without adjustment for confounders (RR 1.79, 95% CI 1.41–2.28, I²=97.0%), those with adjustment for confounders excluding adiposity (1.25,1.04–1.51, I²=84.8%), and those with adjustment for confounders including adiposity (1.38,1.03–1.86, I²=91.8%). In studies with participants’ mean age at BP assessment ≥55 years, no significant differences were found for studies without adjustment for confounders (RR 1.07, 95% CI 0.78–1.47, I²=90.3%), studies with adjustment for confounders excluding adiposity (0.85, 0.78–0.92, I²=12.3%), or studies with adjustment for confounders including adiposity (0.95, 0.80–1.11, I²=45.5%). A similar association was observed between AAM and baseline systolic BP and diabolic BP.

Conclusion

Late menarche was associated with a higher risk of BP and this association was modified by age and adiposity in developing countries.

Keywords:

• Age at menarche
• blood pressure
• heterogeneity
• meta-analysis

Introduction

Hypertension is regarded as a public health problem in the world and is generally defined as having systolic blood pressure (SBP) of 140 mmHg or greater or diastolic blood pressure (DBP) of 90 mmHg or greater, or taking medication for hypertension (NCD Risk Factor Collaboration (NCD-RisC) 2021). It was reported that ageing, dietary factors (high sodium intake, low potassium intake, and unhealthy diet), obesity, and lack of physical activity were common risk factors for hypertension (Mills et al. 2020). There are several factors unique to women that may be related to hypertension including menarche (Bubach et al. 2018), parity (Moazzeni et al. 2021), menopause (Song et al. 2018), and oral contraceptive use (Afshari et al. 2021).

Menarche signals the start of reproductive capacity which is influenced by various factors. Several studies have demonstrated that approximately half of the normal variation observed in the timing of the onset of menses in girls is attributable to genetic factors (Towne et al. 2005; Morris et al. 2011). In addition to genetics, extrinsic factors such as living conditions, nutrition, overall health, physical activity, and socioeconomic status are also important and modifiable aspects in the timing of menarche (Yermachenko and Dvornyk 2014). One study including 40,803 female adolescents found that adolescents with excess weight had earlier menarche than their peers (Barros et al. 2019). Another study observed that undernutrition (stunting or thinness) was associated with relatively older ages for puberty stages (Campisi et al. 2021). Besides, an analysis of 24,380 Mexican women indicated that high socioeconomic status (SES, defined as the income level by an equivalent adult in each household) women reached menarche at the earliest ages and low SES women at the latest (Marván et al. 2020). Compared with developed countries, people in developing countries have a lower average standard of living and poor healthcare and nutrition. A low mean age at menarche has already been observed in developed countries such as France (12.5 years), and Hong Kong (11.67 years) (Tang et al. 2003; Lalys and Pineau 2014) and much higher mean age at menarche can be seen in developing countries such as China (14.8 years) and Bangladesh (15.69 years) (Bosch et al. 2008; Chen et al. 2021). Early age at menarche is associated with increased body mass index (Wei et al. 2019) and early exposure to a high level of oestrogen, while late menarche is associated with late exposure to a low level of oestrogen which is demonstrated to be related to blood pressure (Liu et al. 2018; Chen et al. 2021). Several studies have also shown a link between age at menarche and blood pressure and the results were very inconsistent (Hardy et al. 2006; Canoy et al. 2015; Yang et al. 2018; Chen et al. 2021; Liu D et al. 2021).

Bubach et al. conducted a systematic review and meta-analysis dated March 2016 and demonstrated that early menarche was associated with a higher risk of hypertension (Bubach et al. 2018). We noted that the vast majority of studies reporting data were drawn from populations living in developed countries (USA, UK, Spain, Germany, Portugal, Finland, Japan, Singapore, Korea), with relatively fewer studies publishing information from developing countries (China, Bangladesh, Iran). From 2016 to 2021, some studies were conducted in developing countries and some of them observed that not early menarche, but late menarche, was associated with a higher risk of hypertension (Liu et al. 2018; Zhou et al. 2019; Chen et al. 2021). Besides, evidence from one study including only postmenopausal women indicated there was no statistical connection between age at menarche and hypertension after adjusting for a series of confounders (Yang et al. 2018). A U-shaped association between age at menarche and hypertension was found in one study that leveraged 60,135 healthy young Chinese women (Guo et al. 2018).

Considering this, we thought it was necessary to explore the association between age at menarche and hypertension in developing countries. Therefore, we conducted this systematic review and meta-analysis aiming to evaluate the association between age at menarche and hypertension in developing countries and further explore the potential modifiers of the association.

Methods

Search strategy

We systematically searched three online databases (PubMed, Embase, and Web of Science) up to January 2022. We used the following terms “(menarche or menstruation) combined with (“hypertension”, “blood pressure”, “systolic pressure” or “diastolic pressure”)” to search for pertinent studies. We also reviewed studies included in the previous systematic review and the reference lists of our included studies to identify additional pertinent studies (Prentice and Viner 2013; Bubach et al. 2018). We researched articles in March, 2022, and no additional articles met the inclusion criteria. The present study was not registered. No restriction was given to the language.

Study inclusion and exclusion criteria

Studies were included if they met the following inclusion criteria: (1) participants were adult women aged ≥18 years old and living in developing countries – the Human Development Index (HDI) is a summary measure of average achievement in key dimensions of human development: a long and healthy life, being knowledgeable and having a decent standard of living - a country was defined as a developing country if its HDI is less than 0.800 (Bainbridge et al. 2012); (2) the exposure variable was the age at menarche; (3) the outcome was hypertension with or without baseline values of SBP/DBP; (4) peer-reviewed original cohort, case-control, or cross-sectional studies; and (5) studies presented prevalence ratios (PRs) or odds ratios (ORs) or hazard ratios (HRs) with 95% confidence intervals (CIs) or sufficient data for calculating them. For simplicity, we referred to relative risk (RR) for these three types of measures of association and they were considered to be similar estimates of RR. Studies were excluded if they had one or more of the following characteristics: (1) participants were younger than 18 years old or they had specific diseases; (2) studies were reviews, meta-analyses, book chapters, editorials, or conference abstracts; (3) studies didn’t provide RR (95% CI) or enough data to calculate effect estimates.

Data extraction

Two authors (W.G. and S.W.) independently extracted the data from eligible studies. We solved any disagreements by conducting an open discussion. The following information was extracted: the name of the first author, publication year, study name, study design, study region, year of enrolment, the number of cases and total participants in each menarcheal age group, the timing/methods/ages/testers of BP assessment, definitions of hypertension, baseline values of SBP and DBP, the mean or median age of participants at baseline, mean or median age of menarche, relative risk (RR) and confidence interval (CI), and confounders adjusted in models. We extracted RR (95% CI) for studies: (1) in models without adjustment for confounders; (2) in models with adjustment for the largest number of confounders excluding adiposity [Body mass index (BMI)/Waist circumference (WC)]; (3) in models with adjustment for the largest number of confounders including adiposity. If a study didn’t provide the crude RR (95% CI) but provided the number of cases and total participants in each menarcheal age group, we calculated the crude RR (95% CI) based on the number of cases and total participants in each group.

Statistical analysis

When we analysed the association between age at menarche and blood pressure, women in the oldest group of AAM were compared with the women in the youngest or middle group. If one study reported RR (95% CI) by comparing the youngest with the oldest group (reference), we calculated the reciprocals to make the youngest group the reference. When we calculated the crude RR (95% CI) based on the number of cases and total participants in each group, if this study didn’t specify the reference group, we defaulted to the youngest group as the reference group. Then we calculated the summary RR (95% CI) using the Der Simonian-Laird random-effects model which takes into account both within-study and between-study variations (DerSimonian and Laird 1986). The strength of heterogeneity was divided into mild (<50%), moderate (50–70%), and high (75%) – three groups according to the inconsistency index (I²) among studies. To test the robustness of our results, we conducted a sensitivity analysis by analysing the rest of the studies after removing one study. Funnel plots, Egger’s, and Begg’s regression tests were used to assess whether publication bias existed or not. We performed all statistical analyses using the “meta” package in Stata, Version 16 (Stata Corp LLC 4905 Lakeway Drive College Station, TX 77845 USA). A two-tailed p-value <0.05 was considered statistically significant.

Results

Study selection

A total of 5285 studies were identified from PubMed (N = 1146), Embase (N = 2746), and Web of Science databases (N = 1393). After removing 1765 duplicate studies, we continued to exclude 3456 irrelevant studies and screened out 64 studies for full-text reading according to the titles and abstracts. We also identified another six studies by reviewing previous systematic reviews and the reference lists of our included studies. Of the 70 studies for full-text reading, 20 eligible studies containing 728,127 participants were included in this meta-analysis (Heys et al. 2007; Akter et al. 2012; Qiu et al. 2013; Cao et al. 2016; Chang et al. 2016; Liu et al. 2016; Zheng et al. 2016; Guo et al. 2018; Liu et al. 2018; Yang et al. 2018; Farahmand et al. 2019; Jansen et al. 2019; Petersohn et al. 2019; Shen et al. 2019; Wei et al. 2019; Zhou et al. 2019; Zhang et al. 2020; Chen et al. 2021; Hu et al. 2021; Liu D et al. 2021). The detailed reasons for excluding 50 studies are shown in Figure 1.

Figure 1. Flow chart of study screening.

Study characteristics

Table 1 shows the characteristics of eligible studies. Of the 20 eligible studies, sixteen studies were conducted in China, two in Mexico, one in Iran, and one in Bangladesh. Eighteen studies reported participants’ mean AAM. The mean AAM was between 14 years and 16.5 years in thirteen studies. In five studies the mean AAM of participants was between 12.5 years and 14 years. Late menarche was defined as AAM ≥ 16 years in 16 studies and as AAM >13 years in 4 studies. Six studies defined early menarche as AAM ≤12 years, eight studies defined early menarche as AAM ≤14 years, three studies defined early menarche as AAM ≤15 years, two studies defined early menarche as AAM ≤16 years, and one study defined early menarche as AAM ≤17 years. The reference category of AAM was in the middle category in 8 studies and the youngest category in 12 studies. Fourteen studies reported the crude RR (95% CI) or provided the number of hypertension cases and total participants in each menarchal age group. Eleven studies reported the RR (95% CI) in models with adjustment for a series of potential confounders but not adjusting for adiposity. Eight studies presented the RR (95% CI) in models with adjustment for adiposity and other potential confounders. For studies with adjustment for confounders, age of enrolment, educational level, physical activities, smoking, and alcohol drinking were common confounding factors for adjustment. The mean age at BP assessment was less than 55 years in eleven studies. In nine studies the mean age of participants at BP assessment was more than 55 years.

Table 1. Summary characteristics of eligible studies for age at menarche and blood pressure.

First author, year	Country	Number		Average menarche (years)^a	Year at enrolment (years)	Menarche versus		Methods and testers of blood pressure (BP) assessment	Definitions of hypertension	Age at blood pressure assessment (years)^a
		Cases	Subjects			Late	reference
Liu et al. 2021	China	3187	7518	16.00^b	2013–2014	≥18	≤12	BP was taken as the average of the 3 measurements, using an electronic sphygmomanometer (HEM-770A Fuzzy; Omron, Kyoto, Japan) by well-trained staff.	SBp ≥ 140 mmHg or DBp ≥ 90 mmHg or current use of antihypertension medication.	56.00^b
Hu et al. 2021	China	86207	121431	15.38 ± 2.00	2011–2012	>17	14–17	BP was taken as the average of the 3 measurements, using an automated electronic device (OMRON Model HEM-752 FUZZY) by well-trained staff.	Bp ≥ 120/80 mmHg /treated.	56.96 ± 9.14
Chen et al. 2021	China	59852	234867	14.80± (−)	2012–2016	≥17	15	BP was taken as the average of the 3 measurements, using an OMRON HBP-1300 Professional Portable Blood Pressure Monitor (OMRON, Kyoto, Japan) by well-trained staff.	SBp ≥ 140 mmHg, and/or DBp ≥ 90 mmHg, and/or use of antihypertensive medication within 2 weeks.	46.1 ± 18.90
Zhang et al. 2020	China	6429	15361	16.10 ± 2.20	2015–2017	≥18	16	BP was taken as the average of the 3 measurements, using an electronic sphygmomanometer (HEM-770AFuzzy; Omron, Kyoto, Japan) by well-trained staff.	SBp ≥ 140 mmHg and/or DBp ≥ 90 mmHg, or a previously diagnosed hypertension and use of antihypertensive medication in the previous 2 weeks.	61.5 ± 7.70
Zhou et al. 2019	China	1495	5102	15.50 ± 2.10	2013–2014	≥18	≤15	BP was taken as the average of the 3 measurements, using an electronic BP monitor (Omron HBP-1300; OMRON, Kyoto, Japan) by well-trained staff.	Previously diagnosed hypertension with or without current treatment or the average SBp ≥ 140 mmHg and/or DBp ≥ 90 mmHg at physical examinations.	52.6 ± 17.70
Wei et al. 2019	China	11147	30895	15.64 ± 1.92	2004–2008	≥17	≤12	BP was taken as the average of the 2 measurements, UA-779 digital BP monitor by well-trained staff.	—	51.31 ± 10.36
Shen et al. 2019	China	3268	7893	16.00 ± 2.00	2011–2012	18.6	13.3	BP was taken as the average of the 3 measurements, using an Automatic Upper Arm Blood Pressure Monitor (Omron HEM-7200) by well-trained staff.	Currently taking antihypertensive medicine, SBp ≥ 140 mmHg or DBp ≥ 90 mmHg.	58.00 ± 10.00
Petersohn et al. 2019	Mexico	5349	30626	13.00± (−)	1999–2000	>14	≤12	BP was measured using a mercury column sphygmomanometer, model TXJ-10 by well-trained staff.	SBp > 140 mmHg or DBp > 90 mmHg.	<55
Jansen et al. 2019	Mexico	8116	54921	12.50 ± 1.50	2008–2010	≥17	13	Self-reported.	Self-reported if they have ever been diagnosed by a physician.	42.10 ± 7.60
Farahmand et al. 2019	Iran	—	5344	13.27 ± 1.50	1999–2000	17–19	13–14	BP was taken as the average of the 2 measurements, using a mercury sphygmomanometer and the Korotkoff sound technique	SBp ≥ 130 mmHg, DBp ≥ 85 mmHg, or antihypertensive drug treatment.	13.00–50.00^c
Yang et al. 2018	China	3496	6252	16.15 ± 3.46	2013–2015	≥18	≤13	BP was taken as the average of the 3 measurements, using a mercury sphygmomanometer (Yutu Brand, Yutu, Shanghai Medical Instruments Co., Ltd., Shanghai, China) and oscillometric BP monitors (Omron HBP-1300, Kyoto, Japan) by well- trained staff.	SBp ⩾140 mmHg and/or DBp ≥ 90 mmHg and/or a diagnosis of hypertension.	63.75 ± 9.10
Liu et al. 2018	China	1633	7119	15.40 ± 2.00	2012-2014	≥18	14-15	BP was taken as the average of the 3 measurements, using an electronic sphygmomanometer (Omron HEM-7052, Dalian, China) by well-trained staff.	Previously diagnosed hypertension with or without current treatment, or the average SBP ≥140 mmHg and/or DBP ≥90 mmHg at physical examination.	44.70 ± 19.70
Guo et al. 2018	China	1798	60135	14.00^b	2014–2015	≥16	14	BP was taken as the average of the 2 measurements, using an automated device (HBP-9021 J, Omron, Japan) by well-trained staff.	History of diagnosed hypertension, use of blood pressure medication or SBp ≥ 140 mmHg and/or DBp ≥ 90 mmHg.	29.00 ± 4.12
Zheng et al. 2016	China	1401	12654	14.70 ± 1.90	2009–2013	≥17	≤12	Self-reported	Self-reported physician diagnosis of hypertension or self-reported medication use at baseline.	47.68 ± 8.90
Liu et al. 2016	China	1021	1251	15.35 ± 1.96	2009–2010	≥17	≤14	BP was taken as the average of the 2 measurements, using a standard mercury sphygmomanometer by well-trained staff.	SBp ≥ 130 mmHg and/or DBp ≥ 85 mmHg, antihypertensive drug treatment in a patient with a history of hypertension.	70.39 ± 6.21
Chang et al. 2016	China	1606	3292	13.90 ± 1.69	2004–2008	>14	<12	BP was measured using an electronic device (COLIN, VP-1000, Japan) by well-trained staff.	SBp ≥ 140 mmHg, and/or DBp ≥ 90 mmHg, and/or hypertension history and on ant-hypertensive drug treatment.	<55
Cao et al. 2016	China	—	1625	—	2011–2014	16–20	11–13	BP was measured with a mercury manometer in the seated position after a 5-min rest by well-trained staff.	SBp ≥ 130/DBp ≥ 85 mmHg or current use of antihypertensive medications.	60.45 ± 8.19
Qiu et al. 2013	China	937	3304	16.00^b	2011–2012	≥18	16	BP was taken as the average of the 3 measurements, using an automatic-sphygmomanometers by well-trained staff.	History of hypertension.	>55
Akter et al. 2012	Bangladesh	—	1423	—	2009–2010	>13	<12	BP was taken as the average of the 2 measurements, using the standard mercury manometer by well-trained staff.	SBp ≥ 130/DBp ≥ 85 mmHg or subjects diagnosed with hypertension.	<55
Heys et al. 2007	China	—	7349	15.40 ± 2.10	2003–2004	≥14.5	<12.5	BP was taken as the average of the last 2 of 3 measurements, using the Omron 705CP sphygmomanometer (Omron Corporation, Kyoto, Japan) by well- trained staff.	SBp ≥ 130 or DBp ≥ 85 mmHg or treatment of previously diagnosed hypertension.	64.00 ± 6.00

^aMean ± SD apart from some other indication; ^bmedian of age; ^c range of age.

Abbreviations: Blood pressure (BP); systolic blood pressure (SBP); diastolic blood pressure (DBP)

Age at menarche and hypertension

The summary RR (95% CI) of the oldest groups versus earlier groups of 20 studies according to participants’ mean age at BP assessment is shown in Figure 2. In studies with participants’ mean age at BP assessment <55 years, women in the oldest group as compared with the middle or the youngest group of AAM had a higher risk of hypertension for those studies without adjustment for confounders (RR 1.79, 95% CI 1.41–2.28, I²=97.0%), studies with adjustment for confounders excluding adiposity (1.25, 1.04–1.51, I²=84.8%), and studies with adjustment for confounders including adiposity (1.38, 1.03–1.86, I²=91.8%). In studies with participants’ mean age at BP assessment ≥55 years, no significant differences were found for the risk of hypertension when women in the oldest category of AAM were compared with the middle or the youngest category of studies without adjustment for confounders (RR 1.07, 95% CI 0.78–1.47, I²=90.3%), studies with adjustment for confounders excluding adiposity (0.85, 0.78–0.92, I²=12.3%), and studies with adjustment for confounders including adiposity (0.95, 0.80–1.11, I²=45.5%).

Figure 2. Forest plots of the association between late menarche and hypertension stratified by participants’ mean age at blood pressure assessment. (A) Studies without adjustment for confounders. (B) Studies with adjustment for confounders excluding adiposity. (C) Studies with adjustment confounders including adiposity. Squares represent the relative risk compared with the middle or early menarche group (reference). The size of each square represents the weight (%) assigned to each study. Weights are from the random-effects analysis. RR: relative risk; CI: confidence interval.

Age at menarche and SBP/DBP

For baseline SBP and DBP, we also observed that women in the oldest group as compared with the middle or the youngest group of AAM were associated with 6.82 mmHg higher SBP (95% CI 2.46–11.18, I²=99.7%) and 2.49 mmHg higher DBP (95% CI 1.43–3.56, I²=97.2%) in studies with participants’ mean age at BP assessment <55 years. However, there were no significant differences for SBP (MD-SBP =1.38 mmHg 95% CI: −2.06 to 4.82, I²=98.0%) and DBP (MD-SBP = −0.77 mmHg 95% CI: −2.01 to 0.47, I²=95.5%) when women in the oldest category of AAM were compared with the middle or the youngest category of studies with participants’ mean age at BP assessment ≥55 years (Figure 3).

Figure 3. Forest plots of the association between late menarche and SBP/DBP stratified by participants’ mean age at blood pressure assessment. (A) Late menarche and SBP. (B) Late menarche and DBP. Squares represent the mean difference of SBP compared with the middle or early menarche group (reference). The size of each square represents the weight (%) assigned to each study. Weights are from the random-effects analysis. MD: mean difference; CI: confidence interval; SBP: systolic blood pressure; DBP: diastolic blood pressure.

Sensitivity analysis and publication bias

We further conducted sensitivity analysis among seven studies adjusting for confounders excluding adiposity (I²=84.8%) and four studies adjusting for confounders including adiposity (I²=91.8%) with participants’ mean age at BP assessment <55 years. After excluding one study by Zhou et al. (2019), the high heterogeneity disappeared. The pooled RR (95% CI) for six studies adjusting for confounders excluding adiposity and three studies adjusting for confounders including adiposity with participants’ mean age at BP assessment < 55 years were 1.16 (1.04–1.30, I²=44.2%) and 1.13 (1.10–1.17, I²=0.0%), respectively (Supplementary Figure 1). A series of test results indicated there was a low probability of publication bias. The funnel plot shown in Supplementary Figure 2. was symmetrical. P values for Begg’s and Egger’s text were more than 0.05 (14 studies without adjustment for confounders: Begg’s: p = 0.511; Egger’s: p = 0.213; 11 studies with adjustment for confounders excluding adiposity: Begg’s: p = 1.000; Egger’s: p = 0.527; 8 studies with adjustment for confounders including adiposity: Begg’s: p = 1.000; Egger’s: p = 0.923).

Discussion

In this meta-analysis, we observed that women in the oldest group of AAM as compared with the middle or the youngest group had a significantly higher risk of BP among studies with participants’ mean age at BP assessment <55 years. This association was weaker but still evident among studies with adjustment for adiposity and other confounders.

Our subgroup analysis according to the studies’ participants’ mean age at BP assessment indicated that age was a modifier of the association between AAM and BP, although the exact mechanisms of age on AAM and BP were unclear, given that age itself is a risk factor for blood pressure (Vokonas et al. 1988). A reasonable explanation might be that the traditional risk factors for hypertension might mask the effects of AAM on BP (Abu Bakar et al. 2021).

One previous review in which the populations of fourteen studies, from a total of 17 studies, came from developed countries reported that early menarche was associated with a higher risk of hypertension, which is inconsistent with our study in developing countries. Though the exact mechanisms for earlier menarche being associated with higher hypertension risk in developed countries and later menarche being associated with higher hypertension risk in developing countries are unclear, the disparity of the AAM-BP relationship between developing countries and developed countries may be explained by a series of physiological changes caused by the difference of AAM. As for early age at menarche, it has been suggested that it may result in a longer period of exposure to oestrogen and higher oestrogen levels (Vihko and Apter 1984). Higher cumulative exposure to oestrogen can decrease serum sex hormone-binding globulin levels and low sex hormone-binding globulin levels are associated with an increased risk of BP (Daka et al. 2013; Hu et al. 2021). On the other hand, an earlier age at menarche can promote the development of obesity later in life and then increase the risk of higher blood pressure through adiposity (Wei et al. 2019; Zhang et al. 2020; Liu D et al. 2021). In contrast to early menarche, late menarche is associated with a lower cumulative endogenous exposure to oestrogens and progestogens during the life course (Chen et al. 2021). The absence of protective effects of oestrogens and long-term imbalance of oestrogen and androgen levels may be contributing factors to higher blood pressure (Dubey et al. 2002). Besides, women with late menarche were more likely to smoke and drink, which is demonstrated to increase the risk of hypertension (Liu et al. 2018; Chen et al. 2021). It is necessary to further unravel the related mechanisms for future research.

Compared with earlier menarche, the mean difference between SBP and DBP in the oldest group was significantly higher, especially for SBP. However, we only used the baseline values for SBP and DBP. Whether this relationship was affected by potential confounders, requires further exploration. Up to the present, several studies have focused on the relationship between age at menarche and SBP and DBP by adjusting a series of confounders. Shen, using the baseline data of 7893 women from the China Health and Retirement Longitudinal Study, found that the early onset of menarche was significantly associated with increased SBP and DBP (Shen et al. 2019). Another study conducted in an undeveloped region in southern China found that AAM was independently and positively associated with SBP and DBP (Zhou et al. 2019). A cross-sectional analysis based on a small population of 7119 females found that age at menarche was positively associated with SBP but not with DBP after adjusting for confounders (Liu et al. 2018). However, not all studies observed the relationship between age at menarche and DBP after adjustment for a series of confounders. One study has observed that adiposity indicators mediated the relationship between age at menarche, and systolic (partly) and diastolic (fully) blood pressure. The results indicated that obesity indicators could exert a greater influence on DBP (Werneck et al. 2018). Apart from obesity indicators, other important variates and confounders may also affect the relationship between age at menarche and SBP.

The heterogeneity among seven studies adjusting for confounders excluding adiposity (I²=84.8%) and four studies adjusting for confounders including adiposity (I²=91.8%), with participants’ mean age at BP assessment <55 years, was very high. After excluding one study written by Zhou (Zhou et al. 2019), the high heterogeneity disappeared. A very strong positive association between late age at menarche and hypertension was observed in this study. We found that it was the only one that adjusted for basal metabolism rate, waist circulation, body fat percentage, visceral adiposity index, heart rate, and sleeping duration and the association became stronger (RR = 2.08, 95% CI =1.71 − 2.53). The over-adjustment for confounders might be a reason for the stronger association than in other studies.

There are several strengths to our study. First, this is the first comprehensive and quantitative meta-analysis to explore the relationship between age at menarche and blood pressure in developing countries. Second, we further explore the potential modifiers of this association. Our study also has several limitations. First, all eligible studies in our meta-analysis were cross-sectional analyses and collected age at menarche was self-reported, which might cause recall bias. However, self-reporting is still considered an effective method for measuring age at menarche among middle-aged and older women. One study examined the validity of age at menarche and found a high correlation (r = 0.79, p < 0.001) between recalled and actual age at menarche even after 30 years. (Must et al. 2002). Meanwhile, another study found no systematic difference in self-reported age at menarche in Tromsø 3 and Tromsø 4 (7 years later) and the association seems to be at least as strong in older women (aged 55–73) as in younger women (aged 25–34)(Lundblad and Jacobsen 2017). Second, we could not directly quantify the attenuation in the association between age at menarche and blood pressure among studies adjusting for a series of confounders, because not all studies presented the unadjusted and adjusted estimates at the same time. Third, the majority of studies in our study were conducted in China, research on the association between age at menarche and blood pressure is still scarce in other developing countries. In the future, related research in other developing countries needs to be included to further test the robustness of our results.

Conclusions

Late menarche was associated with a higher risk of blood pressure in developing countries. This association was modified by age and adiposity. It is important to monitor blood pressure among women with late menarche.

Acknowledgments

The researchers sincerely thank all partners who contributed to this study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.