Introduction
The age at onset of the menstrual cycle (menarcheal age) is a significant period in the life of a female adolescent. It is also an important indicator for certain diseases, with an early onset of menarche having been shown to be a risk factor for breast cancer (Kelsey, Reference Kelsey1993), pelvic inflammatory disease and spontaneous abortion (Helm et al., Reference Helm, Münster and Schmidt1996) and ischaemic heart disease (IHD) (Cooper et al., Reference Cooper, Ephross, Weinberg, Baird, Whelan and Sandler1999). One the other hand delayed menarche has been implicated as a high risk factor for irregular menstrual cycles and low peak bone mass (Anai et al., Reference Anai, Miyazaki, Tomiyasu and Matsuo2001). Young woman with type I diabetes mellitus have been shown to have moderately delayed age at menarche compared with the general population (Danielson et al., Reference Danielson, Palta, Allen and D'Alessio2005). The patterns of such diseases could be ameliorated with changes in the age at menarche over time.
Wyshak & Frisch (Reference Wyshak and Frisch1982) reported that secular changes in age at menarche have been described since the 18th century, and other researchers have reported changes in age at menarche worldwide (Cole, Reference Cole2000; Kac et al., Reference Kac, Auxiliadora de Santa Cruz and Velasquez-Melendez2000; Prebeg & Bralic, Reference Prebeg and Bralic2000; Okasha et al., Reference Okasha, McCarron, McEwen and Smith2001; Ayatollahi et al., Reference Ayatollahi, Dowlatabadi and Ayatollahi2002; Becker-Christensen, Reference Becker-Christensen2002; Hesketh et al., Reference Hesketh, Ding and Tomkins2002; Wang & Murphy, Reference Wang and Murphy2002; Hwang et al., Reference Hwang, Shin, Frongillo, Shin and Jo2003; Junqueira et al., Reference Junqueira Do Lago, Faerstein, De Souza Lopes and Werneck2003; Padez, Reference Padez2003; Padez & Rocha, Reference Padez and Rocha2003; Ersoy et al., Reference Ersoy, Balkan, Gunay, Onag and Egemen2004).
Okasha et al. (Reference Okasha, McCarron, McEwen and Smith2001) have suggested that the nature of the relationship between menarcheal age and adult anthropometric measures may be important in understanding the significance of the effects of menarcheal age on disease in later life. Numerous researchers have shown that age at menarche is associated with adult height, weight and body mass index (Shangold et al., Reference Shangold, Kelly, Berkeley, Freedman and Groshen1989; Georgiadis et al., Reference Georgiadis, Mantzoros, Evagelopoulou and Spentzos1997; Bharati & Bharati, Reference Bharati and Bharati1998; Laitinen et al., Reference Laitinen, Power and Järvelin2001; Ersoy et al., Reference Ersoy, Balkan, Gunay, Onag and Egemen2004). It has also been shown to be related to socioeconomic and demographic factors (see, for example: Padez, Reference Padez2003; Chavarro et al., Reference Chavarro, Villamor, Narváez and Hoyos2004; Wronka & Pawlińska-Chmara, Reference Wronka and Pawlińska-Chmara2005).
Specifically with respect to Bangladeshi populations, researchers have studied the relationship of age at menarche with nutritional status, post-menarcheal growth and marriage (Chowdhury et al., Reference Chowdhury, Huffman and Curlin1977; Ogata, Reference Ogata1979; Haq, Reference Haq1984; Riley et al., Reference Riley, Huffman and Chowdhury1989; Chowdhury et al., Reference Chowdhury, Shahabuddin, Seal, Talukder, Hassan, Tomkins, Costello and Talukder2000). Ogata (Reference Ogata1979) investigated age at menarche of 775 Bangladeshi housewives born between 1938 and 1957; he reported that mean age at menarche remained stable over birth-year cohorts.
The purpose of the present study was to test for the presence of any secular trends in age at menarche in university female students in Bangladesh in the birth-year cohorts from 1979 to 1986. In addition, the association between age at menarche and various adult anthropometric measures and socio-demographic factors was assessed.
Data and Methods
Data
The study sample consisted of 995 healthy Bangladeshi adult female students residing in student halls at the University of Rajshahi, Bangladesh, between July 2004 and May 2005. The university has four female halls of residence accommodating a total of 2900 students at any particular time. The University of Rajshahi is the second largest university in Bangladesh, with students coming from all over the country. The sample was selected using stratified random sampling with a proportional allocation technique.
A total of 1000 selected students were asked by a female co-author (Saima Islam), using a standard questionnaire, to report their age at menarche, and socio-demographic characteristics were recorded for each subject. Five students who could not remember their menarcheal age were excluded from the current analysis. Consequently, 995 female students were included in this study to evaluate the secular trend of age at menarche and association with anthropometric and socio-demographic factors. Body height was measured as the distance from the highest point of the top of the head in the mid-sagittal plane to the floor by anthropometer, and body weight was taken with thin clothing using a weighing scale. All measurements were done by a single researcher (Saima Islam). Body mass index (BMI), defined as the ratio of weight in kilograms to height squared in metres, was calculated.
Analysis
The sample was subdivided into eight classes according to birth-year from 1979 to 1986. Descriptive statistics were first used to calculate the mean and standard deviations for age at menarche by birth-year cohorts. Also, a probit method was used to calculate the median of age at menarche by birth-year cohorts (Table 1). The data were then subjected to further statistical analysis.
Age at menarche of Bangladeshi adult female students by birth-year cohorts
To examine the interclass variation of the response variable (age at menarche) the statistical linear model for one-way analysis of variance (ANOVA) was applied. The model corresponding to each variable is:
where Yij is the j th observation (response variable) for the i th birth-year cohort; µ is the general mean effect; α i=µi−µ (additional effect of i th birth-year cohorts); µi is the average effect of ith birth-year cohorts; εij is the random error term, which follows N(0, σ 2); p is the number of cohorts; and q is the number of observations for each cohort.
The ANOVA procedure tests the hypothesis H o:
or equivalently:
by means of a single F-test. If the hypothesis of the equality of cohort means is rejected, it may be concluded that there are differences among the cohort means. The randomness, normality and homogeneity of cohort variances were checked using the Kolmogorov–Smirnov non-parametric test, a normal probability plot, and the Levene test, respectively, for the validity of the ANOVA data.
Linear regression analysis was applied to detect the possible presence of trends in age at menarche among the birth-year cohorts from 1979 to 1986.
To examine the average relationship between the age at menarche and body measurements and socio-demographic factors, multiple regression analysis was utilized. The underlying multiple linear regression model corresponding to each variable is:
where Y is the response variable (age at menarche), Xi (i=1, 2, 3,…, k) are the predictor variables (independent variables), β0 is the intercept term, β1, β2,…, βk are the unknown regression coefficients, and ε is the error term with a N(0, σ 2) distribution. In multiple regression analysis, an important assumption is that the explanatory variables are independent of each other, i.e. there is no relationship between the explanatory variables to estimate the ordinary least squares (OLS). However, in some applications of regression, the explanatory variables are related to each other. This problem is called the multicollinearity problem (Chatterjee & Hadi, Reference Chatterjee and Hadi2006). In this study, a variance inflation factor (VIF) was used to check for the multicollinearity problem among the predictor variables. The variance inflation for independent variables Xj is:
where p is the number of predictor variables and R 2j is the square of the multiple correlation coefficient of the j th variable with the remaining (p−1) variables, where:
(1) if 0<VIF<5, there is no evidence of a multicollinearity problem;
(2) if 5≤VIF≤10, there is a moderate multicollinearity problem; and
(3) if VIF>10, there is a serious multicollinearity problem of variables.
Finally, Student's t-test was used to find the differences between the religions (Muslim and Hindu), residence (urban and rural), father's occupation (government or non-government service and self-employed) and mother's occupation (government or non-government service and housewife). All statistical analyses were performed using SPSS (version 15.0) and MINITAB.
Results
A total of 995 female students were interviewed and examined. The age at menarche of Bangladeshi adult female students varied from 10 to 15 years, with a mean menarcheal age of 13.12±1.16 and a median 13.17 years (Table 1).
Secular trends
Since the current study was subdivided into eight cohorts by birth-year, this facilitated a study of possible trends over time. Before utilizing the ANOVA, it was necessary to ensure that the standard assumptions underlying the ANOVA model were satisfied. Consequently, it was necessary to first test the data for randomness, normality and homogeneity. The Kolmogorov–Smirnov non-parametric test and the normal probability plot showed that there were no serious problems concerning the randomness and normality of the data. In addition, the Levene test demonstrated that the data were homogeneous. Thus, the data satisfied the standard assumptions of the ANOVA model.
The variations of mean age at menarche of Bangladeshi adult students by birth-year cohort from 1979 to 1986 were statistically significant (p<0.001) (Table 2).
Analysis of variance for age at menarche of Bangladeshi adult female students by birth-year cohorts
** 1% level of significance.
Linear regression
To examine the presence of secular trends a regression coefficient was computed. Mean age at menarche is depicted graphically in Fig. 1. This shows that there were yearly fluctuations in age at menarche, and this is a characteristic of such cohort studies. This fluctuation in age at menarche was further examined by linear regression analysis. The positive coefficient of linear regression analysis indicated that the age at menarche of university female students generally increased in value from the birth-year cohorts 1979 to 1986 (Table 3).
Mean age at menarche by birth-year cohorts.
Regression coefficient for the effect of year of birth on age at menarche of Bangladeshi adult female students
a Test for linear trend.
** 1% level of significance.
Multiple regressions
The multiple regression model used was:
where age at menarche (AAM) is the response variable and the other variables are predictors: Ht, body height; BMI, body mass index; MI, family monthly income; FEL, father's education level; MEL, mother's education level; NS, number of siblings; OB, order of birth.
The estimated model was:
The regression coefficients and the VIF of the independent variables are presented in Table 4. The VIF showed that there was no evidence of a multicollinearity problem among the predictor variables. The coefficients of the multiple regression analysis showed that there was a significant positive association between age at menarche and height (p<0.05), and a negative association between age at menarche and BMI (p<0.01) (Table 4). These results suggest that taller and slimmer females reached menarche later than shorter and heavier females. They also show that age at menarche tends to be higher with nutritional status using BMI as an indicator.
Multiple regression coefficients and the variance inflation factor (VIF) for body measurements and socio-demographic factors with age at menarche as the response variable
* 5% level of significance;
** 1% level of significance.
The coefficients of multiple regression analysis also show there was a significant negative association between a girl's age at menarche and her mother's education level (p<0.05), while social class (monthly family income), number of siblings and order of birth did not show association with age at menarche (Table 4). These results indicate that the daughters of better educated mothers reach menarche earlier.
Differences between age at menarche of Bangladeshi female students by socio-cultural status
* 5% level of significance;
** 1% level of significance.
The majority of students were Muslim (89.45%). The majority (64.49%) of student's fathers worked in government or non-government service, while only 11.66% of mothers worked in service (Table 5). To see if there was a significant difference in age at menarche between two groups, the t-test was applied. Students from rural locations had a later mean age at menarche than those who spent their adolescence in urban areas. Muslim students also tended to reach menarche earlier than Hindu students. Daughters of parents who worked in government or non-government service tended to reach menarche earlier than those whose parents were self-employed (Table 5).
Discussion
Secular trends in age at menarche
The study found that the mean menarcheal age of female students at Rajshahi University (data collected between July 2004 and May 2005) was 13.12±1.16 years, which was older than the value of 12.88 years found by Ogata in 1979 for Bangladeshi females (Ogata, Reference Ogata1979). A more recent study (roughly 20 years later) found that the mean menarcheal age of Bangladeshi females was 13.00±0.98 years (Chowdhury et al., Reference Chowdhury, Shahabuddin, Seal, Talukder, Hassan, Tomkins, Costello and Talukder2000). These results suggest that the secular trend toward an increase in age at menarche of Bangladeshi females is still continuing. The current study's investigation period, while limited to only eight years from 1979 to 1986, also displays a pattern of mean age at menarche that rises with cohort age (Fig. 1). Moreover, the regression coefficient (β=0.097) indicates that menarcheal age of Bangladeshi female students has increased over time (Table 3). Ogata (Reference Ogata1979), who reported on the secular trend in age at menarche of Bangladeshi females, indicated that the age at menarche remained stable among birth-year cohorts from 1938 to 1957. The results of the current study are in agreement with those of Cole (Reference Cole2000), who pointed out that the mean menarcheal age of most European countries had stabilized at approximately 13 years but may be subsequently rising, as was found in a Germany study by Gohlke & Woelfle (Reference Gohlke and Woelfle2009). The current results are also in agreement with those of Dann & Roberts (Reference Dann and Roberts1993), who reported that the mean age at menarche of Warwick University female students has also increased. This increasing tendency in age at menarche has also been found in the United States (Wyshak, Reference Wyshak1983), Finland (Rimpelä & Rimpelä, Reference Rimpelä and Rimpelä1993) and Croatia (Prebeg & Bralic, Reference Prebeg and Bralic2000).
Menarcheal age and anthropometric measures
The present study demonstrates that the age at menarche of Bangladeshi female students was negatively associated with adult BMI, but positively associated with adult height. These results are supported by the findings of Chowdhury et al. (Reference Chowdhury, Shahabuddin, Seal, Talukder, Hassan, Tomkins, Costello and Talukder2000), who found that the age at menarche of Bangladeshi females was related negatively with BMI and positively with height, and those of Ersoy et al. (Reference Ersoy, Balkan, Gunay, Onag and Egemen2004), who reported an inverse relationship between age at menarche and post-menarcheal weight and BMI of Turkish female students. The present results also corroborate the study of Okasha et al. (Reference Okasha, McCarron, McEwen and Smith2001), who found in female students at the University of Glasgow that age at menarche was positively associated with adult height and negatively associated with weight and BMI.
Menarcheal age and socio-demographic factors
The present study demonstrates that mother's educational level and occupation have a significant influence on their daughter's age at menarche. To the authors' knowledge, there have been no comparable studies on this in Bangladesh to date. However, a similar study conducted on female university students in Portugal (Padez, Reference Padez2003) found no association between a girl's age at menarche and her parent's educational level and occupation. The present study found that females from rural locations had a later age at menarche than those who spent their adolescence in urban areas. Identical results have been found in Portugal (Padez, Reference Padez2003) and Spain (Marrodán et al., Reference Marrodán, Mesa, Aréchiga and Pérez-Magdaleno2000). The current study showed that female Muslim students tended to reach menarche earlier than Hindu students. A similar result was found by Chowdhury et al. (Reference Chowdhury, Huffman and Curlin1977). Environment and social–cultural background, such as definitions of levels of education and classification of occupation types, vary significantly between different geographical areas. It may be difficult to make conclusions based on multiple studies from different parts of the world, and perhaps more local, regional studies should be conducted.
This study only investigated trends in age at menarche and its association with selected anthropometric measures and socio-demographic factors. It was not possible to look at other important factors directly related to age at menarche, such as birth weight (Silva et al., Reference Silva, De Stavola, Mann, Kuh, Hardy and Wadsworth2002; Terry et al., Reference Terry, Ferris, Tehranifar, Wei and Flom2009), childhood living conditions (Kac et al., Reference Kac, Auxiliadora de Santa Cruz and Velasquez-Melendez2000), food habits in childhood (Windham et al., Reference Windham, Bottomley, Birner and Fenster2004), physical activity, life-style factors and nutrition (Merzenich et al., Reference Merzenich, Boeing and Wahrendorf1993). Clearly, more research is required.
Conclusions
The mean age at menarche among Bangladeshi university students (data collected between July 2004 and May 2005) was found to be 13.12±1.16 years and the median was 13.17 years, showing an increase over time when compared with previous studies. Menarcheal age was negatively associated with adult BMI, but positively associated with height. Early menarche was noted among students from urban areas, those of Muslim religion and those from households with an educated mother.
