Content
Data description
AaM
) registered in years at the first interview, was the response variable of interest. The average was 13.18 with a minimum of 8 and a maximum of 24. We could raise the question about the existence of bias in the reported ages, but many studies have demonstrated that this reported age is quite accurate, most likely due to the emotional significance of menarche for a young girl [28]. Figure 1 shows the histogram for the unconditional distribution of this variable with a modal class at 13 years old. Summarized in Table 1 are its descriptive percentiles by decade from 1920 to 1973, and Table 2 depicts the mean and the median by municipality and decade. Year of birth (Byear
) and the demographic information given by the municipality of residence (Muni
) were the independent variables.
AaM (Percentiles) | ||||||||
---|---|---|---|---|---|---|---|---|
Decade | 5 | 10 | 25 | 50 | 75 | 90 | 95 | women |
20’s | 11 | 12 | 12 | 14 | 15 | 17 | 18 | 19844 |
30’s | 11 | 11 | 12 | 14 | 15 | 17 | 17 | 72206 |
40’s | 11 | 11 | 12 | 13 | 14 | 16 | 17 | 109335 |
50’s | 11 | 11 | 12 | 13 | 14 | 15 | 16 | 122045 |
60’s | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 109266 |
70’s | 10 | 11 | 12 | 12 | 14 | 15 | 15 | 19652 |
Decade | Decade | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
20’s | 30’s | 40’s | 50’s | 60’s | 70’s | 20’s | 30’s | 40’s | 50’s | 60’s | 70’s | ||
Municipality |
\(\bar{x}; \;\tilde{x}\)
|
\(\bar{x}; \;\tilde{x}\)
|
\(\bar{x}; \;\tilde{x}\)
|
\(\bar{x}; \;\tilde{x}\)
|
\(\bar{x}; \;\tilde{x}\)
|
\(\bar{x}; \;\tilde{x}\)
| Municipality |
\(\bar{x}; \;\tilde{x}\)
|
\(\bar{x}; \;\tilde{x}\)
|
\(\bar{x}; \;\tilde{x}\)
|
\(\bar{x}; \;\tilde{x}\)
|
\(\bar{x}; \;\tilde{x}\)
|
\(\bar{x}; \;\tilde{x}\)
|
ÁGUEDA | 14.12; 14 | 13.77; 14 | 13.48; 13 | 13.04; 13 | 12.79; 13 | 12.67; 12 | MONTEMOR-O-VELHO | 14.01; 14 | 13.71; 14 | 13.28; 13 | 12.85; 13 | 12.58; 13 | 12.63; 13 |
AGUIAR DA BEIRA | 13.51; 13 | 13.59; 13 | 13.25; 13 | 12.79; 13 | 12.44; 12 | 12.54; 12 | MORTÁGUA | 14.71; 16 | 14.67; 14 | 14.18; 14 | 13.6; 13 | 13.1; 13 | 12.64; 13 |
ALB.-A-VELHA | 14.61; 15 | 14.34; 14 | 13.88; 14 | 13.42; 13 | 13.02; 13 | 12.55; 13 | MURTOSA | 14.13; 14 | 13.73; 13 | 13.47; 13 | 13.21; 13 | 12.83; 13 | 12.49; 12 |
ALCOBAÇA | 14.26; 14 | 14.02; 14 | 13.67; 14 | 13.21; 13 | 12.94; 13 | –;– | NAZARÉ | 14.01; 14 | 13.65; 13 | 13.49; 13 | 12.93; 13 | 12.74; 13 | 12.76; 13 |
ALMEIDA | 14.28; 14 | 13.72; 14 | 13.57; 14 | 13.24; 13 | 12.71; 13 | 12.68; 13 | NELAS | 14.11; 14 | 13.5; 13 | 13.25; 13 | 12.74; 13 | 12.52; 12 | 12.41; 12 |
ALVAIÁZERE | 13.95; 14 | 14.01; 14 | 13.5; 13 | 13.15; 13 | 12.8; 13 | 12.68; 13 | OLEIROS | 13.56; 13 | 13.49; 13 | 13.04; 13 | 12.76; 13 | 12.43; 12 | 12.39; 12 |
ANADIA | 14.1; 13.5 | 14.07; 14 | 13.61; 13 | 13.18; 13 | 12.95; 13 | 12.82; 13 | OLIVEIRA DE FRADES | 13.85; 14 | 13.58; 14 | 13.18; 13 | 12.92; 13 | 12.6; 12.5 | 12.7; 13 |
ANSIÃO | –;– | 13.93; 14 | 13.54; 13 | 13.14; 13 | 12.67; 13 | 12.04; 12 | OLIV. DO BAIRRO | 14.04; 14 | 13.79; 14 | 13.36; 13 | 12.89; 13 | 12.62; 12 | 12.6; 12.5 |
ARGANIL | 13.92; 14 | 13.58; 13 | 13.17; 13 | 12.74; 13 | 12.51; 12 | 12.55; 12 | OLIV. DO HOSPITAL | 14.52; 14 | 14.16; 14 | 13.76; 14 | 13.32; 13 | 13.19; 13 | –;– |
AVEIRO | 14.01; 14 | 13.72; 14 | 13.35; 13 | 12.95; 13 | 12.67; 13 | –;– | OURÉM | 13.83; 14 | 13.75; 14 | 13.33; 13 | 12.98; 13 | 12.67; 13 | 12.62; 12 |
BATALHA | –;– | 13.64; 14 | 13.27; 13 | 12.93; 13 | 12.58; 12 | 12.8; 12.5 | OVAR | 12.69; 12 | 13.54; 13 | 13.21; 13 | 12.83; 13 | 12.66; 13 | 12.61; 12 |
BELMONTE | 13.87; 14 | 14.1; 14 | 13.67; 13 | 13.22; 13 | 12.85; 13 | 12.65; 12 | PAMPI. DA SERRA | 14.58; 14 | 14.2; 14 | 13.57; 13 | 13.11; 13 | 12.85; 13 | 12.8; 13 |
CANTANHEDE | 13.6; 13 | 13.36; 13 | 13; 13 | 12.77; 13 | 12.36; 12 | 12.49; 12 | PEDROGÃO GRANDE | 14.1; 14 | 13.96; 14 | 13.55; 14 | 13.08; 13 | 12.94; 13 | 12.81; 13 |
CARREGAL DO SAL | 14.47; 14 | 14.26; 14 | 13.74; 14 | 13.32; 13 | 12.94; 13 | 12.7; 12.5 | PENACOVA | 13.56; 13 | 13.45; 13 | 13.34; 13 | 12.86; 13 | 12.61; 12 | 12.43; 12 |
CAST. DE PÊRA | 14.33; 14 | 14.11; 14 | 13.71; 14 | 13.22; 13 | 12.66; 13 | 12.5; 12.5 | PENALVA DO CASTELO | 13.64; 14 | 13.54; 13 | 13.13; 13 | 12.79; 13 | 12.58; 12 | –;– |
CASTELO BRANCO | 13.76; 14 | 13.7; 14 | 13.33; 13 | 13; 13 | 12.66; 13 | 12.7; 13 | PENAMACOR | 13.37; 13 | 13.41; 13 | 13; 13 | 12.63; 13 | 12.43; 12 | 12.36; 12 |
CASTRO DAIRE | 14.34; 14 | 14.28; 14 | 13.74; 14 | 13.27; 13 | 12.88; 13 | 12.7; 13 | PENEDONO | 14.18; 14 | 14.08; 14 | 13.58; 13 | 13.15; 13 | 12.91; 13 | 12.81; 13 |
CELORICO DA BEIRA | 14.26; 14 | 14.06; 14 | 13.62; 13 | 13.14; 13 | 12.8; 13 | 12.77; 13 | PENELA | –;– | 13.7; 14 | 13.41; 13 | 12.93; 13 | 12.94; 13 | –;– |
COIMBRA | –;– | 13.84; 14 | 13.51; 13 | 12.9; 13 | 12.48; 12 | 12.65; 13 | PINHEL | 14.19; 14 | 13.87; 14 | 13.41; 13 | 12.92; 13 | 12.75; 13 | 12.75; 13 |
CONDEIXA-A-NOVA | 13.88; 14 | 13.77; 14 | 13.37; 13 | 12.8; 13 | 12.54; 12 | 12.25; 12 | POMBAL | 14.62; 15 | 14.08; 14 | 13.76; 14 | 13.28; 13 | 12.96; 13 | 12.82; 13 |
COVILHÃ | –;– | 13.79; 14 | 13.38; 13 | 12.89; 13 | 12.53; 12 | 13.3; 13 | PORTO DE MÓS | 14.27; 14 | 14.05; 14 | 13.54; 13 | 13.15; 13 | 12.96; 13 | 12.79; 13 |
ESTARREJA | 14.01; 14 | 13.56; 13 | 13.23; 13 | 12.85; 13 | 12.7; 13 | 13.06; 13 | PROENÇA-A-NOVA | 13.88; 14 | 13.66; 13 | 13.27; 13 | 12.74; 13 | 12.5; 12 | 12.77; 12 |
FERREIRA DO ZÊZERE | 14.27; 14 | 13.98; 14 | 13.45; 13 | 13; 13 | 12.83; 13 | 12.58; 12 | RESENDE | 14.35; 14 | 14.02; 14 | 13.61; 13 | 13.18; 13 | 12.62; 12 | 12.62; 13 |
FIGUEIRA DA FOZ | 13.95; 14 | 13.57; 13 | 13.14; 13 | 12.71; 13 | 12.55; 12 | 12.62; 12 | SABUGAL | 13.94; 14 | 14; 14 | 13.42; 13 | 13.15; 13 | 12.63; 12 | 12.49; 12 |
FIG. CAST. RODRIGO | 13.39; 13 | 13.61; 14 | 13.21; 13 | 12.83; 13 | 12.53; 12 | 12.27; 12 | SANTA COMBA DÃO | 14.4; 14 | 14.32; 14 | 13.87; 14 | 13.32; 13 | 12.82; 13 | 12.55; 12 |
FIGUEIRÓ DOS VINHOS | 13.8; 14 | 13.88; 14 | 13.55; 13 | 13.3; 13 | 12.81; 13 | 12.66; 12 | SÃO PEDRO DO SUL | 14.27; 14 | 14.17; 14 | 13.97; 14 | 13.55; 13 | 12.92; 13 | 12.6; 12.5 |
FORNOS DE ALGODRES | 14.54; 15 | 14.22; 14 | 13.57; 13 | 13.26; 13 | 12.95; 13 | –;– | SÁTÃO | 13.77; 14 | 13.45; 13 | 13.15; 13 | 12.71; 13 | 12.55; 12 | 12.65; 12 |
FUNDÃO | 13.62; 13 | 13.35; 13 | 12.96; 13 | 12.7; 13 | 12.47; 12 | 12.43; 12 | SEIA | 14.38; 14 | 14.43; 14 | 13.99; 14 | 13.53; 13 | 13.14; 13 | 12.77; 13 |
GÓIS | 13.93; 14 | 13.48; 13 | 13.24; 13 | 12.8; 13 | 12.48; 12 | 12.21; 12 | SERNANCELHE | 14.54; 14 | 14.51; 14 | 13.99; 14 | 13.48; 13 | 13.03; 13 | 13.24; 13 |
GOUVEIA | 13.58; 14 | 13.64; 13 | 13.4; 13 | 13.01; 13 | 12.67; 13 | 12.63; 13 | SERTÃ | 13.8; 14 | 13.55; 13 | 13.16; 13 | 12.83; 13 | 12.55; 12 | 12.41; 12 |
GUARDA | 14.22; 14 | 13.97; 14 | 13.61; 13 | 13.25; 13 | 12.82; 13 | 12.33; 12 | SEVER DO VOUGA | 14.6; 15 | 14.19; 14 | 13.74; 14 | 13.18; 13 | 12.84; 13 | 12.84; 13 |
IDANHA-A-NOVA | 14.33; 14 | 14.2; 14 | 13.96; 14 | 13.34; 13 | 13.01; 13 | 12.89; 13 | SOURE | 13.81; 13 | 13.54; 13 | 13.32; 13 | 12.97; 13 | 12.63; 13 | 12.53; 12 |
ÍLHAVO | 13.94; 14 | 13.88; 14 | 13.48; 13 | 12.89; 13 | 12.76; 13 | 12.49; 12 | TÁBUA | 13.31; 13 | 13.89; 14 | 13.36; 13 | 13.16; 13 | 12.72; 13 | 12.4; 12 |
LAMEGO | 13.92; 14 | 13.73; 14 | 13.4; 13 | 13; 13 | 12.67; 13 | 12.48; 12 | TOMAR | 14.05; 14 | 13.97; 14 | 13.51; 13 | 13.18; 13 | 12.83; 13 | 12.91; 13 |
LEIRIA | 13.61; 13 | 13.43; 13 | 13.02; 13 | 12.9; 13 | 12.46; 12 | 12.47; 13 | TONDELA | 13.59; 13 | 13.57; 13 | 13.11; 13 | 12.7; 13 | 12.46; 12 | 12.22; 12 |
LOUSÃ | 13.73; 14 | 13.51; 13 | 13.14; 13 | 12.78; 13 | 12.42; 12 | 12.51; 12 | TRANCOSO | 14.19; 14 | 14; 14 | 13.55; 13 | 13.01; 13 | 12.72; 13 | 12.5; 12 |
MAÇÃO | 14.44; 14 | 14.34; 14 | 14.06; 14 | 13.68; 13 | –;– | –;– | VAGOS | 13.4; 13 | 13.21; 13 | 12.89; 13 | 12.82; 13 | 12.54; 12 | 12.21; 12 |
MANGUALDE | 14.73; 15 | 14.6; 14.5 | 13.83; 14 | 13.48; 13 | 12.99; 13 | –;– | V. DE REI | 14.29; 14 | 14.09; 14 | 13.54; 13 | 13.13; 13 | 12.73; 13 | 12.81; 13 |
MANTEIGAS | 15; 14 | 13.77; 14 | 13.45; 13 | 13.12; 13 | 12.78; 13 | 12.93; 13 | V. NOVA DE FOZ CÔA | 13.76; 14 | 13.66; 13 | 13.29; 13 | 12.85; 13 | 12.63; 13 | 12.53; 12 |
MARINHA GRANDE | 14.04; 14 | 14.01; 14 | 13.55; 13 | 13.03; 13 | 13.11; 13 | –;– | V. NOVA DE PAIVA | 13.98; 14 | 13.23; 13 | 13.03; 13 | 12.59; 12 | 12.53; 12 | 11.8; 11 |
MEALHADA | 14.03; 14 | 13.79; 14 | 13.37; 13 | 12.93; 13 | 12.6; 13 | 12.39; 12 | V. NOVA DE POIARES | 13.86; 14 | 13.72; 14 | 13.35; 13 | 12.86; 13 | 12.53; 12 | 12.69; 13 |
MÊDA | 13.88; 14 | 13.59; 14 | 13.25; 13 | 12.81; 13 | 12.54; 13 | –;– | V. VELHA DE RODÃO | 14.04; 14 | 13.95; 14 | 13.53; 13 | 13.15; 13 | 13.04; 13 | –;– |
MIRA | 14.12; 14 | 13.48; 13 | 13.2; 13 | 12.84; 13 | 12.6; 12 | 12.6; 13 | VISEU | 13.97; 14 | 13.88; 14 | 13.48; 13 | 13.06; 13 | 12.68; 13 | 12.64; 13 |
MIRANDA DO CORVO | 13.39; 13 | 13.1; 13 | 13.02; 13 | 12.64; 13 | 12.5; 12 | 12.5; 12 | VOUZELA | 14.48; 14 | 14.46; 14 | 13.82; 14 | 13.38; 13 | 13.01; 13 | 12.54; 13 |
MOIMENTA DA BEIRA | 14.29; 14 | 14.02; 14 | 13.64; 14 | 13.11; 13 | 12.84; 13 | 12.72; 13 |
Methods
Statistical models
AaM
, given the covariates Byear
and Muni
.AaM
is right skewed (vide Fig. 1), and a non-linear assumption for the relation between the AaM
and Byear
is better supported by the data. Taking this into consideration, we opted to analyse the data with a model within the aforementioned class – GAMLSS. This method offers a highly flexible approach in that constraints to the traditional distributional assumptions, such as normality, are removed. The method enables the use of skewed distributions without having to transform the data, allowing us to work on the scale of the data, which is a very important feature. Additionally, as already said, all the parameters of the response probability distribution can be modelled by explanatory variables and not only the location. For instance, it is possible to model the variance and the skewness (for some distributions it is also possible to model the kurtosis). All the explanatory variables are introduced into the model parameters through predictors, which can be linear functions of the explanatory variables or can take the form of structured additive predictors with non-linear or smoothing functions of explanatory variables.Byear
and Muni
. It is a multiplicative model resulting from the log-link, which in turn ensures positive values for the parameter. For this distribution the scale parameter is approximately the coefficient of variation. The skewness parameter, \(\nu\), is modelled only with an intercept term. It was found that adding covariates (e.g. age) to its linear predictor did not improve the adjustment because those models produced greater GAIC values.Results
R
software (v 4.1.0), namely using the main package gamlss
(version 5.1-4) and two additional packages that provide a set of functions to fit models with spatial variables (gamlss.spatial
and gamlss.add
).R
syntax based on the gamlss package:
Age at menarche – temporal trends
Byear
for exploring the temporal trends in the central region of Portugal. It should be noted that the percentiles curves displayed in Fig. 2 are not linear. This behaviour could not be captured within a typical linear regression analysis and is facilitated by the non-linear approach permitted by the generalized additive models. Additionally, the percentiles 25%, 50%, 75%, 90% and 95% are steadily decreasing at a greater rate since 1920 than the percentiles 5% and 10%, which could mean that there is not much more space for reductions in the menarche age for the girls with the earliest menarches. The stagnation trend reported by other studies also appears to be emerging, mainly in the 1960s. From 1970 onwards the downward trend reappears, which is in line with the results published in [18].
Age at menarche – spatio-temporal trends
Byear
and Muni
) on the AaM location parameter, i.e. its median. The interpretation is relatively straightforward. A positive time effect was found until approximately 1950 and thereafter a negative effect. This means that a woman born before 1950 had an AaM above the median for the overall population and accounting all the years, i.e. 13.05, and a woman born after 1950 will tend to have an AaM below 13.05. For the spatial effects, a municipality with a negative effect means that women residing there have ages at menarche below the median when comparing to the overall population. On the other side, areas with a positive spatial effect will tend to have ages at menarche above the median of the overall population. Looking to Figs. 3 and 4 and comparing them to Fig. 5, it is clear that larger values of the median AaM are associated with a positive spatial effect and vice-versa.
getPEF
function (Partial Effect function) which allows us to calculate the slope of the curve at each year (vide Table 3). For example, for the year 1960, in gamlss
syntax this can be done by and the result is the slope of the curve’s tangent for the year 1960, which is not very different from the other years, at least until 1960. The result is approximately \(-0.04\), meaning that for that year the median AaM was reduced at a rate of \(0.04\times 365\approx 14.6\) days, or 4.87 months per decade. This value is in accordance with the findings of other studies, e.g. [35] which examines the evolution of the menarche in the period 1830–1960 in Europe and reported a decreasing of about 3 months per decade. For the year 1970, we have seen from Fig. 2 a curve with a horizontal aspect, meaning a stabilization. Calculating the slope at this time, the value of \(-0.013\) appears, meaning that the median AaM is reducing in this year at a rate of \(0.013\times 365=4.745\) days, or 1.58 months per decade, pointing toa slowing down in the reduction rate.
Birth Year | Decreasing Rate |
---|---|
1920 | 8.91 |
1925 | 9.1 |
1930 | 10.47 |
1935 | 13.04 |
1940 | 14.43 |
1945 | 14.71 |
1950 | 15.23 |
1955 | 13.72 |
1960 | 10.48 |
1965 | 6.26 |
1970 | 4.75 |