Secular trend in age at menarche for women born between 1920 and 1979 in Rio de Janeiro, Brazil

PRIMARY OBJECTIVE: This article presents data on the secular trend in age at menarche for 1955 women from 16 to 76 years of age born between 1920 and 1979 and studied under the Nutrition and Health Survey conducted in the municipality of Rio de Janeiro, Brazil, in 1996. METHODS AND PROCEDURES: Age at menarche was defined by the retrospective method. Women were grouped according to decade of birth, and the trend was estimated using simple linear regression between age at menarche and year of birth for the following specific periods: 1920-1940, 1920-1960, 1960-1979 and 1920-1979. MAIN OUTCOMES AND RESULTS: Mean age at menarche decreased from 13.07 to 12.40 years when comparing the group of women born in the 1920s with the 1970s birth cohort, corresponding to a mean rate of -0.0123 years per year (p < 0.001). The downward trend was -0.0120 years per year (p > 0.05) for the 1920s, 30s and 40s, -0.0093 years per year (p < 0.05) for the period from 1920 to 1960, and -0.0224 years per year (p < 0.01) for the 1960s/70s. CONCLUSIONS: The results suggest a secular trend in age at menarche. The literature points to such environmental variables as improved living conditions and expanded access to health services. Within this context, age at menarche could be used as a marker for social development.     

Menarcheal age in Turkey: Secular trend and socio-demographic correlates

Background: Menarche is an important indicator for assessing the developmental status of pubertal girls. Despite its importance, there is no nationwide information on menarcheal age in Turkey.

Aim: This paper is the first attempt to examine age at menarche for Turkey as a whole. The aim is to present the secular trend of menarcheal age and variations across different socio-demographic groups.

Methods: Data were employed from the Turkey Demographic and Health Survey, 2008. Mean menarcheal ages were estimated for birth cohorts and socio-demographic sub-groups. The pace of decline in menarcheal age has been estimated using multiple linear regression analysis, controlling for year of birth and other variables.

Results: Mean age at menarche was estimated as 13.30 (95% CI = 13.26–13.35). It was estimated as 13.17 years (95% CI 12.95–13.38) for the youngest birth cohort (1989–1993), as opposed to 13.44 (95% CI 13.37–13.52) years for the cohort born in 1959–1968.

Conclusion: Regression analysis indicated a decrease of 1.44 months per decade, providing evidence of a secular trend in menarcheal age in Turkey. Further results suggested childhood place of residence, education, welfare status and number of siblings to be significantly associated with menarcheal age.     

Time-related trends of age at menopause and reproductive period of women in a Chuvashian rural population

OBJECTIVE: The objective of this study was to examine in an observational, cross-sectional, community-based study the secular trend of age at menopause among women in a rural Chuvashian population and to identify factors associated with age at menopause. DESIGN: The sample included 316 postmenopausal women born between 1920 and 1950 with mean age at menopause of 48.47 +/- 4.63 (34-58) years. Statistical analyses included simple and multiple linear regression and "whiskers" plots. RESULTS: Significant association was found between year of birth and age at menopause (beta = 0.194, P < 0.001) and reproduction period (P = 0.193, P < 0.001). Mean values of age at menopause increased from 47.0 years (born during 1920-1925) to maximal values of 49.7 years (born during 1940-1945) and 49.3 years (born during 1945-1950). Mean values of their reproductive period increased from 30.7 (born during 1920-1925) to maximal values of 34.1 (born during 1940-1945) and to 33.7 (born during 1945-1950). Multiple linear regression analysis demonstrated that year of birth was the only statistically significant (P = 0.19, P < 0.01) predictor of age at menopause. Age at menarche can also be a possible predictor of age at menopause (beta = -0.12, P = 0.04). CONCLUSIONS: The study confirmed certain secular trends of age at menopause and reproductive periods in Chuvashian women. The authors also observed a negative association between age at menarche and age at menopause. The number of children and medical abortions as well as body mass index showed no association with age at menopause.     

Secular trend in age at menarche for South Korean women born between 1920 and 1986: the Ansan Study

BACKGROUND: There is strong evidence of a downward secular trend in age at menarche in Europe and the USA during the last century and in Japan and China during the past few decades. However, no study on this trend in age at menarche has been reported in South Korea. AIM: To measure the trend in age at menarche in South Korea during the past few decades and the association of height with this trend. SUBJECTS AND METHODS: A total of 1061 South Korean women born between 1920 and 1986 were randomly recruited from Ansan Cohort Study samples and separate school girl samples, and subjected to this analysis. The data on age at menarche were collected by the retrospective method. Height was measured at time studied and assumed to be relatively constant since age at menarche. Women were grouped with respect to decade of birth and mean age at menarche was determined. The secular trends in annual age at menarche and in height were analysed by the 3-year moving average. RESULTS: Mean menarcheal age decreased from 16.8 to 12.7 years during the past 67 years, corresponding to -0.64 years per decade. Height increased from 149.23 to 161.75 cm during the same period, showing an inverse relationship in the change of trend between height and mean age at menarche. CONCLUSION: Our data suggest that the downward secular trend in age at menarche may reflect the secular change in physical growth in South Korean women during the past 67 years.     

Age at menarche in a Chuvashian rural population

OBJECTIVES: The secular trend and familial influences on age at menarche among a rural Chuvashian population was examined in a cross-sectional community-based study. SUBJECTS AND METHODS: The cohort included 617 females aged 18-80 years, with age at menarche ranging from 10 to 24 years (mean 15.4 +/- 2.1). Statistical analyses included simple linear regression, maximum likelihood estimation and a whiskers plot. RESULTS: Women born during the second through the fourth decade of the 20th century showed increasing mean values of age at menarche from 15.4 (second decade) up to 16.5 (fourth decade). Women born after the fourth decade showed a decrease in mean values from 15.5 (fifth decade) to 13.0 (ninth decade). The highest peak of age at menarche was in women born in 1932. Significant familial correlations were observed between adjusted to year of birth age at menarche in mothers and daughters r = 0.195, p<0.01 and sisters and sisters r = 0.404, p<0.01. CONCLUSION: Our study demonstrated the certain secular trends of age at menarche in Chuvashian women who matured after World War II. Periods of socio-economic disasters such as war and famine disrupt the secular trend. Our findings also confirm the hypothesis of significant familial influences on menarcheal age.     

Evidence for a downward secular trend in age of menarche in a rural Gambian population

Abstract

Menarcheal age is a key indicator of female maturity and development. Studies in many countries have reported a downward secular trend in age of menarche over the past century. This study presents data gained using the ‘status quo’ method and interval regression to estimate median menarcheal age of girls in a rural Gambian community. Cross-sectional studies carried out in 1989, 2000 and 2008 revealed a median menarcheal age of 16.06 (95% CI 15.67–16.45), 15.03 (95% CI 14.76–15.30) and 14.90 (95% CI 14.52–15.28), respectively. The average rate of decline of median age of menarche was amongst the most rapid yet reported, at 0.65 years of age per decade (p < 0.00001). There was no evidence for a change in the rate of decline over the two decades studied. These results probably reflect ongoing socio-economic development within the region.     

Recent decline in age at menarche: The Fels Longitudinal Study

A number of recent reports suggest that the average age at menarche of US girls has declined over the past 20 years. Because the putative declines in the age at menarche are concurrent with increases in childhood body mass index (BMI), it has been suggested that these two trends may be causally linked. We examined differences in mean age of menarche in Fels Longitudinal Study girls who were born in six 10‐year birth cohorts (1930s, 1940s, 1950s, 1960s, 1970s, and 1980s) and simultaneous cohort changes in mean BMI measured cross‐sectionally at selected ages from 3–35 years (n = 371). Girls born in the 1980s had a mean age at menarche of 12.34 years, which was ～3–6 months earlier than that of girls born previously (P < 0.001). While the mean BMI values at ages 25 and 35 generally increased from the 1930s to the 1970s, the mean BMI during childhood and adolescence remained constant across the six birth cohorts. In summary, we found no evidence that the recent decline in the age at menarche in the Fels Longitudinal Study girls was reflected in concurrent increases in BMI at any point in childhood or adolescence. Conversely, girls born in the 1960s and 1970s have subsequently become heavier in young and mid‐adulthood than were girls from earlier birth cohorts, without any concurrent change in the mean age at menarche over that time period. These two findings suggest that population‐level shifts in BMI and the timing of menarche are largely independent, although sometimes coincident, processes. Am. J. Hum. Biol. 16:453–457, 2004. © 2004 Wiley‐Liss, Inc.     

Socio-demographic predictors of age at menarche in a group of Colombian university women

Background: Age at menarche is an important developmental milestone known to be modified by social and environmental factors. The determinants and potential change over time of age at menarche in Colombia are not well documented.

Aims: To obtain a current estimate of age at menarche in Colombia and to examine the associations between socio-demographic characteristics and age at menarche.

Subjects and methods: Between 1998 and 2001, we obtained recalled age at menarche, socio-demographic characteristics, and physical activity information in a group of 3206 students at the National University of Colombia who completed a self-administered questionnaire. We examined the cross-sectional associations between mean age at menarche and migratory pattern, parental education, socio-economic status indicators, practice of physical activity, and year of birth using linear regression.

Results: Mean age at menarche (±?SD) was 12.68 ± 1.31. In a multivariate adjusted regression model, age at menarche was positively associated with family size and the practice of at least 2 daily hours of physical activity, and was inversely related to urbanization level, socio-economic status, and year of birth.

Conclusions: Socio-economic status is a strong predictor of menarche in this population. There appears to be a trend towards earlier menarche by year of birth.     

Variability of reproductive history across the Swiss SAPALDIA cohort – Patterns and main determinants

Background: Reproductive characteristics play an aetiological role for many diseases, including reproductive cancers. They have been shown to vary internationally and nationally, but have not yet been described for the whole Swiss population.

Aim: The study investigated the variability of reproductive characteristics, their patterns, and main determinants across a population-based female study population in Switzerland.

Methods: Reproductive characteristics obtained from 3119 women (28–72 years) participating in the SAPALDIA cohort survey in 2001–2003 are described across birth cohorts, study areas, language regions, and levels of urbanization. Determinants of age at menopause were analysed by Cox regression.

Results: Reported median age at menarche was 13 years and median age at natural menopause was 52 years. The prevalence of nulliparity was 27%, and the fertility rate 1.6. Across birth cohorts there was a decline of menarcheal age and fertility rates, and an increase of nulliparity and age at last pregnancy. All characteristics varied across study areas, language regions, and levels of urbanization. Smoking, parity, and physical exercise were stronger predictors of earlier (<52 years) than older (>52 years) menopausal age.

Conclusion: Reproductive events show secular and geographic variation within Switzerland. Smoking, parity and physical activity significantly predict age at natural menopause, particularly before age 52. The secular trend of earlier menarche and increased nulliparity may result in a higher risk of reproductive cancers in younger generations.     

Nutritional status and age at menarche in a rural area of Bangladesh

The age at menarche and its association with nutritional status in a rural area of Bangladesh was determined. A cross-sectional study was conducted in four villages of Rupganj Thana of Narayanganj district. Data was collected through October to December 1996 using a pre-tested structured questionnaire interview schedule, and nutritional status was measured by weight, height, body mass index (BMI) and physical examination. Data were obtained on 436 adolescent girls aged 10-17 years. Among them, 165 (37.8%) girls had commenced menarche. The mean age at menarche as determined by retrospective recall was 13 years SD 0.89 (n=165). The median age at menarche determined by the status quo method was 13.0. Among the adolescents 60.1% were thin (BMI&lt; 5th centile WHO recommended reference) and 48.2% were stunted (&lt; 3rd centile NCHS/WHO). The mean weight and BMI were significantly higher among the menstruating girls of 13, 14 and 15 years (p &lt; 0.01) than non-menstruating girls. The mean height was found to be significantly higher at 11-14 years among the menstruating girls (p &lt; 0.05). A lower prevalence of angular stomatitis was found among the menstruating adolescent girls compared with the non-menstruating girls, 36.4% versus 46.5%, although this was statistically non-significant (odds ratio=0.66, 95% CI 0.43-1.00). For glossitis, no significant difference was found. Among the menstruating girls 12.1% were suffering from menorrhagia and 31.5% from dysmenorrhoea. We conclude that the age of menarche among this rural Bangladeshi community is not as delayed as expected. Not surprisingly, menarche is associated with better nutritional status. The surveyed population had extremely high rates of undernutrition which suggests that adolescents in this and similar situations require specific intervention programmes to improve their nutritional status.     

Book review

Background: Due to the secular trend in length and height, growth references need to be updated regularly. Reference charts that were until recently used in Belgium are based on samples collected more than 30 years ago, and references for body mass index (BMI) and pubertal development have not been established before.

Aims: To establish contemporary cross-sectional reference charts for height, weight, BMI, head circumference, and pubertal development from birth to 21 years of age, based on a representative sample of children from Flanders, Belgium.

Subjects and methods: 15 989 healthy subjects of Belgian origin, 0–25 years of age, were measured in 2002–2004. Growth curves were fitted with the LMS method, and percentiles for the pubertal development were estimated with generalized additive models on status quo data from 8690 subjects aged 6–22 years of age.

Results: A positive secular trend in height and weight is observed in children above 5 years of age. Adult median height has increased by 1.2 cm/decade in boys and 0.8 cm/decade in girls; median weight by 0.9 kg/decade in boys, and 1.0 kg/decade in girls, and the weight distribution became more skewed. The BMI curve is comparable to that of other populations, except for higher percentiles. This reflects the increasing prevalence of overweight and obesity. Median age at menarche (13.0 years) has not advanced any more over the past 50 years. Median ages at menarche and B2 in girls and G2 or T4 in boys are comparable to other West European estimates, but approximately 10% enter G2/T4 before 9 years of age.

Conclusions: The ongoing secular trend in height and weight makes growth charts previously used in Belgium obsolete. New representative charts for growth and pubertal development are introduced. For weight monitoring, it is advised that the now-available BMI growth charts are used.     

Social background and age at menarche in Portuguese university students: A note on the secular changes in Portugal

Menarcheal age of a sample of Portuguese university students (n = 3,366), born between 1972 and 1983, age 18–23 years, was analyzed. The influence of parents' educational level and occupation, family size, birth order, and degree of urbanization of girl's locality of residence during childhood and adolescence were analyzed as well as secular trend in the Portuguese population. Mean age at menarche for girls born in 1983 was 12.32 years. Parents' educational level and occupation did not show any significant influence on mean age at menarche. Place of residence during childhood years and adolescence showed a significant effect on mean age at menarche, with girls from rural places with a later age at menarche than those who spent their childhood or adolescence in urban areas (P ≤ 0.01 and P ≤ 0.05 for childhood and adolescence, respectively). Family size and birth order showed the highest effect. Girls born in small families, with one child, matured earlier (12.32 years) than those born in large families with four or more children (12.67 years), (P ≤ 0.01). Also, girls that were first‐born had an earlier menarche (12.41 years) than those who were third or later (12.58 years, P ≤ 0.01). Regression analysis selected family size and place of residence in childhood as the most important determinants of mean age at menarche in our university students. In this sample, from 1972 to 1983, mean age at menarche remained stable. When we considered published data from all the Portuguese population we found a decrease in mean age at menarche from 15 years for girls born in 1880 to 12.44 for those born in the 1980s. This decrease was the result of great improvements in the social and economic living conditions that occurred in Portugal especially after the 1970s concerning nutrition and health care, among many other environmental factors. Am. J. Hum. Biol. 15:415–427, 2003. ©2003 Wiley‐Liss, Inc.     

Secular trends in height,weight and body mass index of 6-year-old children in Bremerhaven

Secular trends in growth processes of children can be important indicators of changes in public health. Common to studies on secular trends in children is that evaluation is based on comparison of data collected at two (or more) distinct points on a time scale. The quantitative characteristic of the secular trend is estimated by linear interpolation between the two end points of the underlying time interval, which in studies of children are usually at least 10 years apart. The purpose of the present paper is to analyse secular trends in height, weight and body mass index (BMI) of 6-year-old children from Bremerhaven over the period 1968-1987 (the year refers to the birth cohort). The results are based on data drawn from health records of the City Health Centre, where all 6-year-old children are routinely measured in a school entrance examination. Thus the data represent complete birth cohorts of children entering school in Bremerhaven and not selected samples. The data reported here refer only to children of German origin. The sample sizes vary from n = 313 (girls born in 1982) to n = 737 (boys born in 1968), and total sample size is n = 7601. Regression of the arithmetic means of height on year of birth showed that the trend in stature for children born between 1968 and 1987 was 0.67cm/decade for boys and 0.49cm/decade in girls. Both trends are statistically significant (p &lt; 0.05). Although there was an increasing tendency for weight as well, which was more marked for the 95th percentile than for the median, neither of the trends in both sexes was statistically significant. While the BMI in both sexes showed no trend at all for the median and the 5th percentile, there was a significant linear increase of the 95th percentile. Furthermore, the results for height show that an evaluation of secular trends under qualitative and quantitative perspective critically depends on the selection of points on the time scale.     

Secular change in age at menarche in rural Oaxaca, southern Mexico: 1968--2000

OBJECTIVE: To evaluate secular change in the age at menarche between 1978 and 2000 in residents of a rural Zapotec-speaking community in Oaxaca, southern Mexico, using status quo and retrospective methods. MATERIALS AND METHODS: Status quo menarcheal status of girls 9-18 years of age in 1978 (n= 101) and 2000 (n=238) and retrospective ages at menarche of adult women 19+ years of age in 1978 (n = 228) and 2000 (n = 246) were obtained via interview. Probit analysis was used to estimate median ages at menarche and 95% confidence intervals (CI) for the status quo data. Analysis of variance and linear regression was used to compare the retrospective data across surveys. RESULTS: The median ages at menarche of adolescents are 14.8+/-1.2 years (0.24 year, 95% CI 14.2-15.4 years) in 1978 and 13.0+/-1.0 years (0.10 year, 95% CI 12.7-13.3 years) in 2000. Age at menarche has declined by 1.8 years over about 23 years, 0.78 year/decade (95% CI = 0.65-0.91 year/decade). Recalled ages at menarche do not differ by age group in the 1978 survey, but differ significantly by age group in the 2000 survey (p <0.001). Within the 2000 survey, the two youngest age groups (< 29, 30-39 years) do not differ, but attain menarche earlier than women in the four older age groups (p < 0.05), who do not differ from each other in age at menarche. The estimated rates of secular decline in age at menarche in adult women vary between 0.38 and 0.42 years/decade (0.26-0.56 year/decade). CONCLUSIONS: There is a major secular decline in the age at menarche of adolescent girls and young adult women between 1978 and 2000. The estimated rate of decline in adult women is about one-half of that in adolescent girls. The secular decline in age at menarche is consistent with corresponding secular gains height, sitting height and estimated leg length of children and adolescents in the community; corresponding secular gains are smaller in young adults 19-29 years.     

Menarcheal age for Norwegian women born 1830–1960

Data from birth records from the maternity hospitals in the three main cities in Norway have been used to study the trend in menarcheal age for women born from about 1830 to about 1960. The investigation is based on a sample of 200–300 records around every 10th year from each of the three clinics in partly overlapping time periods relating to a total of 9152 women. The recollected age at menarche fell from just above 16 years for women born around 1830 to just above 13 years for those born around 1960, the decrease being not totally linear. These results correspond closely with a previously published investigation from Oslo from about the same period of time (Brudevoll, Liestøl and Walløe, 1979), but our results, which cover more of Norway, show a more linearily shaped curve than the results covering only Oslo. We have also analysed the relationships of several independent variables to menarcheal age, using multivariate linear regression methods. Besides the woman's year of birth, which was the most important variable throughout the whole period of time, various geographical variables were found to be of moderate importance. Being born in the countryside and in towns other than Oslo and Bergen led to a slightly higher age at menarche. No significant difference between Oslo, Bergen and Trondheim was detected except for the period up to about 1880 where the Bergen women had about 2·6 months earlier menarche than the Oslo women. Married women amongst the sample had experienced menarche a little earlier than the unmarried, and among married women there was an association between occupation and menarcheal age, women from the lowest social classes having the latest ages at menarche. The importance of these socially related parameters declined with time, and for women born after 1945 the difference seemed to have disappeared. The age at menarche was found to be positively related to a woman's age at first birth; the further back in time the stronger the relation. In addition, delayed age at menarche was also found to be associated with irregularities in the menstrual cycles in later life.     

Changes of recollected menarcheal age and month among women in Tokyo over a period of 90 years

Data on the recollected age at menarche of 47,881 women born between 1881 and 1970 were examined. The mean menarcheal age had changed from 15.1 years in those born up to 1900 to 12.5 years in those born during the 1960s. The age at menarche differed according to the month of birth, and the pattern of average age distribution by month of birth was not the same when the year of birth was different. Among women born before 1955 the menarcheal age was earlier in those born in the summer. However, among women born after 1955, menarche occurred later when they were born in the summer. The monthly distribution of menarche had also changed during this 90-year period. Two peaks in April and August were prominent among those born up to 1960. A third peak in January became also apparent in those born after 1960.     

Menarcheal age for Norwegian women born 1830-1960

Data from birth records from the maternity hospitals in the three main cities in Norway have been used to study the trend in menarcheal age for women born from about 1830 to about 1960. The investigation is based on a sample of 200-300 records around every 10th year from each of the three clinics in partly overlapping time periods relating to a total of 9152 women. The recollected age at menarche fell from just above 16 years for women born around 1830 to just above 13 years for those born around 1960, the decrease being not totally linear. These results correspond closely with a previously published investigation from Oslo from about the same period of time (Brudevoll, Liest?l and Wall?e, 1979), but our results, which cover more of Norway, show a more linearily shaped curve than the results covering only Oslo. We have also analysed the relationships of several independent variables to menarcheal age, using multivariate linear regression methods. Besides the woman's year of birth, which was the most important variable throughout the whole period of time, various geographical variables were found to be of moderate importance. Being born in the countryside and in towns other than Oslo and Bergen led to a slightly higher age at menarche. No significant difference between Oslo, Bergen and Trondheim was detected except for the period up to about 1880 where the Bergen women had about 2.6 months earlier menarche than the Oslo women. Married women amongst the sample had experienced menarche a little earlier than the unmarried, and among married women there was an association between occupation and menarcheal age, women from the lowest social classes having the latest ages at menarche. The importance of these socially related parameters declined with time, and for women born after 1945 the difference seemed to have disappeared. The age at menarche was found to be positively related to a woman's age at first birth; the further back in time the stronger the relation. In addition, delayed age at menarche was also found to be associated with irregularities in the menstrual cycles in later life.     

Secular trends of growth and pubertal maturation of school children in Southern Thailand

Background: In Thailand, studies of growth date back to 1975, but there have been no studies examining any trends in increasing/decreasing growth.

Aim: To determine if there have been any secular trends of increasing/decreasing growth and/or ages at puberty in Thai children.

Methods: In 2012, a cross-sectional study of growth was conducted in 3460 children. The median heights and weights and the ages of pubertal maturation were compared with previous studies. Correlations between the secular trends and the health statistics indicators were calculated.

Results: From 1975–2012, the median final height of boys and girls had increased by 1.32 and 0.86?cm/decade and weight by 2.49 and 1.76?kg/decade, respectively. In girls, the age at thelarche and menarche had declined by 0.39 and 0.12 years/decade, respectively. In boys, the age at testicular enlargement Tanner II had declined by 0.15 years/decade. Increased physical growth was positively correlated to life expectancy, per capita income and prevalence of overweight/obesity was negatively correlated to prevalence of malnutrition and under-five mortality rate.

Conclusion: The positive secular trend towards an increase in growth and a decline in the age at onset of puberty of Thai children is correlated with improvements in overall living conditions in Thailand.     

Stature and stature distribution in Portuguese male adults 1904–1998: The role of environmental factors

The records of height of 841,457 18‐year‐old Portuguese males were analyzed by area of residence (districts). The sample included all Portuguese 18‐year‐old males born between 1966 and 1979 and examined between 1985 and 1998, in the north, center, and south of Portugal. They represent all social strata. There were statistically significant differences (P ≤ 0.001) among districts: males from Lisbon (172.8 cm) and Setúbal (172.7 cm), the most developed districts, were tallest, and those from Madeira (169.7 cm) and Coimbra (171.6 cm) were shortest. Compared with published data for 1904, there was a positive secular trend in height. The average increase was 8.93 cm and the estimated rate was 0.99 cm per decade. The changes that occurred were mainly the result of the reduction of the shortest classes of stature, those <150 cm to 170 cm, and an increase in the frequency of the highest classes, ≥170 cm. This positive trend and the changes in stature distribution must be related to the general improvement in standard living conditions that occurred in Portugal primarily after the 1960s and 1970s, especially in terms of nutrition and the health system. Taking into account the socioeconomic differences that still exist between districts, the results suggest that the secular trend in height should continue for the Portuguese population in future decades. Am. J. Hum. Biol. 14:39–49, 2002. © 2002 Wiley‐Liss, Inc.