| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Cancer Research UK Health Behaviour Unit, Department of Epidemiology and Public Health, University College London, London, UK.
Address correspondence and reprint requests to Cornelia van Jaarsveld, Cancer Research UK Health Behaviour Unit, Department of Epidemiology and Public Health, University College London, Gower Street, London WC1E 6BT, UK. E-mail: c.jaarsveld{at}ucl.ac.uk
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Methods: The Health and Behaviour in Teenagers Study (HABITS) followed a cohort of 5863 adolescents from ages 11 to 12 years (UK year 7; US grade 6) for 5 years. Puberty was assessed with the Pubertal Development Scale. Three pubertal timing groups were created by identifying adolescents who reached midpuberty relatively early, average, or late, compared with their peers. Longitudinal trends in health behaviors and stress were compared between the three groups.
Results: Smoking rates were higher throughout adolescence among early-maturing students, with no evidence that late-maturers "caught up" when they reached puberty, although group differences narrowed over time. Early-maturing students had higher rates of sedentary behaviors but also reported higher rates of vigorous activity than their "on-time" developing counterparts. Patterns in dietary behaviors and stress showed lower rates of daily breakfast and higher stress among early-maturing girls, but not boys. Overall, the effects were largest in early adolescence (ages 11–13 years) and became smaller at older ages (ages 14–16 years).
Conclusion: Early-maturing adolescents are at increased risk for unhealthy behaviors, especially smoking, and although differences attenuate during adolescence, they remain significant at age 16 years. This suggests that early maturation may be a cause of, or is at least a marker for, differences in lifestyle.
Key Words: adolescence • pubertal timing • health-related behaviors • smoking • diet • activity • stress
Abbreviations: HABITS = Health and Behaviour in Teenagers Study; OR = odds ratio; CI = confidence interval.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Many studies of substance abuse or problem behaviors (e.g., smoking, alcohol, drug use, delinquency) have identified early-maturing boys and girls as at risk (5–12). Other studies, particularly those focusing on boys, have identified both early- and late-maturers as exhibiting higher rates of problem behavior than "on-time" maturers (13–15). A common interpretation has been that early-maturers (and possibly also late-maturers) are disadvantaged by being deviant from the normal developmental patterns; this leads to lower self-esteem and increased emotional stress, which in turn influences initiation and use of substances (7,16). In addition, early-maturing boys and girls are hypothesized to feel the need to bond with others of a similar maturational age (older adolescents or adults), and in attempting to achieve this, they adopt some accessible adult behaviors (e.g., smoking, drinking) in order to fit in (4,16).
Given that on average, girls mature earlier than boys, associations between pubertal timing and health behaviors need to take into account gender differences. Gender differences in trends in health behaviors as a function of pubertal timing have also been reported. For example, higher smoking rates have been found in prepubertal girls compared with prepubertal boys, although no gender differences in smoking rates were observed in postpubertal children (17). Sedentary activities were shown to increase with advancing puberty in girls, but not in boys (18).
Evidence for higher rates of health risk behaviors such as smoking and alcohol consumption among early-maturing adolescents is reasonably convincing given the number of different studies that have reported it, although conclusions are limited by many studies being cross-sectional (8,9,13,15), having only two assessment points (10,11), or focusing only on one gender (5,7,12–14). However, very few studies have looked at pubertal timing effects on a wider range of health behaviors such as food choice (19) or activity levels (18,20). A reduction in energy intake and a parallel decrease in energy expenditure with advancing pubertal maturation has been reported (19,20). The lack of longitudinal studies exploring the continuing impact of pubertal timing beyond midadolescence makes it difficult to know whether puberty per se, or the early timing relative to peers, is the critical factor. It is also unclear whether early maturation brings forward the onset of certain behaviors, with late-maturers "catching up" once they pass puberty, or actually confers risk—or is a marker for higher risk—and, therefore, the differences persist after adolescence.
Cross-sectional analyses using baseline data from the Health and Behaviour in Teenagers Study (HABITS) showed that pubertal stage at age 11 years was related to health behaviors and that patterning of results varied both by behavior and gender (21). The present study examines the developmental trends in health behaviors, as well as stress levels, using longitudinal data from the same large, mixed gender sample. Patterns in smoking uptake, food choices, activity levels, and stress are compared between adolescents who reach midpuberty relatively early, at the average time, or later than others in their peer group. Our objectives were a) to replicate the established association between early puberty and smoking uptake; b) to see whether differences in smoking rates between early, average, and late-maturers narrowed between ages 11 and 16 years and differed by gender; and c) to determine whether other health behaviors (food choice, physical activity, sedentary behaviors) and stress followed the same pattern, by plotting their longitudinal trajectories from 11 to 16 in adolescents with early, average, or late timing of puberty.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
1 year of the study, but the present analyses included 5229 (89%) participants, excluding those with missing data on pubertal timing (n = 553) or ethnicity (n = 81). The final sample consisted of 2982 boys and 2247 girls. Just over half the sample were White (58%), 26% were Black, 11% were Asian, and 5% reported other ethnic backgrounds (Table 1).
|
Measures
Pubertal Timing
Puberty was assessed with the Pubertal Development Scale (24), which is a five-item, gender-specific, self-report scale, developed for settings where measures such as direct observation or explicit photographs (e.g., Sexual Maturation Scale) (25) are less appropriate. Students rated themselves on growth spurt, pubic hair, skin changes, along with menarche and breast development for girls and voice change and facial hair for boys. Total scores at each school year are recoded into five stages: prepubertal, beginning pubertal, midpuberty, advanced pubertal and postpubertal (24).
Three pubertal timing groups were created by identifying adolescents who reach midpuberty relatively "early," "average," or "late," compared with their peers. Definitions of pubertal timing differed by gender, because girls reach midpuberty on average 6 to 12 months earlier than boys (26). For boys, pubertal timing was defined "early" for those reaching midpuberty in Year 7 to 8 (ages 11–13 years); "average" when midpuberty was reached in Year 9 (ages 13–14 years), and "late" when midpuberty was reached in Year 10 (ages 14–15 years) or later. For girls, "early" was defined as reaching midpuberty in Year 7 (ages 11–12 years), "average" as reaching midpuberty in Year 8 (ages 12–13 years), and "late" as reaching midpuberty in Year 9 (ages 13–14 years) or later. This categorical variable was used rather than the continuous variable age at midpuberty because the exact age when midpuberty was reached was unknown.
Health Behaviors
Data on health behaviors were collected at each year. Smoking status was assessed using questions from the UK National Smoking Surveys (27). Students were asked which of the following statements best described them: I have never smoked; I have only ever tried smoking once; I used to smoke sometimes but I never smoke cigarettes now; I sometimes smoke cigarettes now but I don't smoke as many as one a week; I usually smoke between one and six cigarettes a week; and I usually smoke more than six cigarettes a week. For the purpose of the following analyses, responses were split in two ways to indicate those participants who had "ever tried smoking" versus those who had never smoked, and into "current smokers" (smoking sometimes or more often) versus noncurrent smokers (never smokers, one time triers, and ex-smokers). Salivary cotinine, a metabolite of nicotine, was used to give biochemical verification of recent smoking. Students provided saliva samples on dental rolls which were sent for assay by gas chromatography. Where students reported noncurrent smoking but cotinine levels exceeded 15 ng/ml (indicating recent smoking behavior), self-reported smoking status was adjusted accordingly (28). This resulted in 10 to 21 adjustments per study year (<1% of data). The two measures of smoking (i.e., ever tried smoking and current smoking rates) were included because the number of students who currently smoked was too low at earlier years to examine pubertal timing effects. It has been shown that "one time triers" are at increased risk for smoking uptake in later years (29). It should be noted that the ever tried smoking variable is cumulative across years.
Intake of high-fat foods was based on frequency of eating eight food items that provide proportionally high levels of fat from the "foods that you eat scale" (30). They included crisps (chips), sweets/chocolate, cakes, other puddings/desserts, biscuits, chips (fries), sausages/burgers, and processed meat. Frequency of consumption of each item was dichotomized as infrequent (no more than twice a week) and frequent (3 days a week). As previously reported for Year 7, a high-fat diet was defined as at least three high-fat foods consumed frequently (23). Fruit and vegetable intake was assessed using two questions, originally based on items from the Dietary Instrument for Nutrition and Education (31) and used in a similar, modified format in other studies (32), which asked: "About how many servings of fruit [vegetables] do you usually eat in a day?" Responses were classified according to national guidelines as less than five versus at least five servings a day. Daily breakfast routine was assessed by asking participants how often they usually ate breakfast, and responses were recoded as daily versus less than daily.
Activity was assessed with separate questions for vigorous activity and sedentary behaviors. Vigorous activity was assessed using a question from the Youth Risk Behavior Surveillance System Questionnaire (33), which asks how many days in the past week the person did exercise that made them "sweat and breathe heavily." Vigorous activity was coded as at least 3 days a week versus <3 days a week. Sedentary behavior was assessed by asking students how much time they usually spent watching television or videos, playing video games on the computer on school days, and were coded as 2 hours or 
or >2 hours (34).
Stress
Stress was assessed with the four-item version of the Perceived Stress Scale (PSS) (35). The scale ranges from 0 to 16; higher scores indicate more stress. The PSS has good concurrent and predictive validity and was correlated in the expected direction with a wide range of outcomes (35). This shortened version had a relatively low internal reliability in the present sample at baseline (
= 0.50), possibly because of the young age of the respondents. The internal reliability increased at follow-up assessments ( = 0.64 at Year 11, ages 15–16 years). Its advantage is that it is quick to complete and therefore enables repeated measures of perceived stress in large samples.
Anthropometric and Demographic Characteristics
Students were individually weighed in indoor clothes and without shoes (scales, Tanita Corporation of America, Arlington Heights, Illinois). Height was measured (Leicester height measure, Medical Scales and Measuring Systems seca ltd., Birmingham, UK). Students were asked their exact date of birth, and age was calculated in decimal years. Ethnicity was reported in predefined categories and combined into White, Black or mixed Black, Asian or mixed Asian, and "other." Socioeconomic deprivation of each student's area of residence was indexed using Townsend scores, which are based on census data on car ownership, unemployment, and overcrowded living conditions by postcode (36). The Townsend scores in the HABITS sample were divided into quintiles from least to most deprived.
Analyses
The present analysis included 5229 (89%) of the 5863 participating students. Typical of longitudinal studies, full data on all five assessments are available for <50% of subjects. However, our analyses are not restricted to students with complete data because the mixed models analysis fits a linear trend line for each student and does not need data from every visit, thereby maximizing the use of available data. On average, the 5229 students contributed 3.7 observations to the analyses (range 1–5). Trends in health behaviors and stress were analyzed separately for boys and girls according to pubertal timing (i.e., reaching midpuberty early, average, or late). We also examined interactions between gender and pubertal timing effects in the total sample, to check whether gender-specific analyses were indicated. There were significant interactions between gender and pubertal timing for ever tried smoking, fat intake, fruit and vegetable intake, breakfast routine, and vigorous exercise, and therefore gender-specific analyses are presented.
Multilevel analyses included clustering by school, and all analyses were adjusted for age, deprivation, and ethnicity. Multivariate multilevel logistic regression analyses used "average" pubertal timing as the reference category in the first series of analyses, and additionally the "early" pubertal timing group is compared with "late." The interaction between pubertal timing and school year was assessed to determine whether the effect of pubertal timing changed over time. When the interaction was significant, odds ratios were calculated for each year separately; only a single estimate is presented when the interaction with time was nonsignificant. When significant, a quadratic time parameter (year2) was added to models for nonlinear effects. Analyses were carried out with SPSS 13 and MLwiN (Centre for Multilevel Modeling, University of Bristol, UK), and p < .05 was considered significant. Effects are illustrated by presenting results from the three pubertal timing groups in the same graph. Using the example of smoking: if pubertal status is the key factor, then as each group reaches midpuberty, smoking rates should increase. If early puberty has a maintained adverse effect, then the early pubertal group would be expected to show maintained higher smoking rates than the other two groups. If being "off-time" versus "on-time" is important, both early and late groups would have higher smoking rates than the average pubertal timing group. The same principles apply to the other health behaviors.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Sample Characteristics of the Three Pubertal Timing Groups
Among boys, 30% were classified as early maturing, 30% as average, and 40% as late maturing relative to peers. Among girls, the percentages were 27% as early maturing, 45% as average, and 29% as late maturing. Characteristics of the three groups in Year 7 (the first year of the study) are shown in Table 1. In Year 7, the female early puberty group had already reached midpuberty (n = 605), but relatively few of the boys had reached midpuberty in Year 7 (n = 341). As would be expected, the early puberty group was slightly older for the year group. Those with early puberty were also more likely to be Black and more socioeconomically deprived (boys only). In Year 7, early puberty was significantly related to a number of unhealthy behaviors (Table 1). Higher stress was also observed in both early and late compared with average-maturing girls, but differences were not significant in boys. Subsequent analyses were adjusted for age, deprivation, and ethnicity.
Trends in Health Behaviors During Adolescence
Table 2 shows trends in health behaviors, separately by gender. The most striking observation is the rapid development of unhealthier lifestyles across the domains of smoking and activity (Figures 1 and 2). Effects on food choices were mixed, showing both increases and decreases of different unhealthy food choices. The percentage of students who had tried smoking cigarettes increased from around 20% to 60% and current smoking increased from around 3% in Year 7 to 30% in Year 11. Trends in dietary behaviors are also shown in Table 2. The percentage of students with high-fat diets and daily breakfast routines decreased over time, whereas fruit and vegetable intake peaked at Year 8 and decreased during later years. The percentage of students meeting the recommendations for vigorous physical activity decreased over time, whereas the percentage of students with high sedentary behavior increased.
|
|
|
Pubertal Timing Effects on Smoking
Table 3 presents the association between pubertal timing and smoking in boys and girls. A combined odds ratio (OR) across all years of follow-up is presented in the main table, and separate ORs for each year are presented in the footnote under the table when the effect of pubertal timing changes during follow-up (assessed by a significant interaction between puberty group and year). Early-maturing boys and girls had significantly higher rates of smoking compared with average and late-developing peers. The interaction of puberty group and year was significant for both smoking variables in boys and "ever tried smoking" in girls, indicating that the difference in smoking rates between puberty groups significantly decreased over time. As shown in the footnote under Table 3, the odds for having ever tried smoking in the early compared with average puberty group decreased from 1.71 (95% CI = 1.38; 2.12) in Year 7 to 1.28 (1.03; 1.59) in Year 11 for boys, and from 1.68 (1.34; 2.11) in Year 7 to a nonsignificant 1.13 (0.92; 1.40) in Year 10 for girls. Even though the association between early puberty and trying smoking decreased over time, it was still significant at Year 11 in boys (Figure 1). The odds for current smoking in the early puberty boys group decreased from 3.52 (2.41; 5.16) in Year 7 to a nonsignificant 1.02 (0.81; 1.29) in Year 11, compared with average-developing boys, but differences between the early and late puberty groups remained significant at all years. Early-maturing girls also reported higher rates of current smoking and this effect was stable over time (OR = 1.33 (1.07; 1.64) compared with the average puberty group, and OR = 1.62 (1.28; 2.05) compared with late-developing girls). Late-maturing boys had significant lower rates of current smoking (OR = 0.77, 95% CI = 0.65; 0.93) and late-maturing girls had significant lower rates of ever tried smoking (OR = 0.67, 95% CI = 0.56; 0.81) compared with average-developing peers, and these lower smoking rates in late-maturing students were stable over time (Table 3).
|
Pubertal Timing Effects on Dietary Behavior and Activity Levels
Differences in dietary behaviors between pubertal timing groups in boys were nonsignificant (Table 4). Among girls, the early-developing girls reported lower-fat intake compared with late-maturing girls (OR = 0.83, 95% CI = 0.70; 0.98) and in comparison with average-developing girls, late-maturing girls also had significantly higher rates of high-fat intake, an effect that increased over time and was significant from Year 10 (OR = 1.22, 95% CI = 1.04; 1.44) and increased to 1.33 (1.08; 1.64) in Year 11. There were no differences regarding fruit and vegetable intake. Early-maturing girls reported significantly lower rates of a daily breakfast routine compared with average-developing girls, but these effects decreased over time and became nonsignificant by Year 9.
|
Trends in vigorous physical activity indicated healthier behaviors in early-maturing boys and girls (Table 4 and Figure 2). Early-maturing boys had significantly higher rates of vigorous activity compared with average- and late-developing boys, and this effect was stable over time. However, early-maturing boys also reported more sedentary behavior compared with late-developing boys (OR = 1.23, 95% CI = 1.08; 1.41). Early-maturing girls reported significant higher rates of vigorous physical activity compared with average-developing girls, but these effects decreased over time and became nonsignificant by Year 9. Trends in sedentary behavior in girls were more complicated. Early-maturing girls reported significant higher rates of sedentary behavior in Year 7, and significant lower rates in Year 11, compared with average-developing girls.
Pubertal Timing Effects on Perceived Stress Levels
Differences in perceived stress levels between puberty groups were nonsignificant in boys (Table 4). In girls, the pattern was for higher stress in both early- and late-maturing girls compared with average-maturing peers. Multivariate analyses showed that the early-maturing girls reported significantly higher stress than average-maturing peers (OR = 1.49, 95% CI = 1.19; 1.88).
The influence of perceived stress on the puberty group effects on health behaviors was examined by repeating all analyses of the pubertal timing effects on health behaviors adjusting for perceived stress levels. Although stress was significantly related to most health behaviors, including it in the model did not change the reported pubertal timing effects (data not tabulated).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
As in previous studies, we found that smoking rates were highest among early-maturing students, intermediate in average-maturing, and lowest in late-maturing boys and girls. Trend analyses indicated that differences between puberty groups decreased during follow-up, but significant differences in smoking rates remained at ages 15 to 16 years. Longer follow-up is needed to see whether differences wash out by adulthood. These patterns were less obvious for other health behaviors, but some associations between pubertal timing and food choices and activity levels were observed. Earlier-maturing adolescents reported higher rates of sedentary behavior and not having a daily breakfast routine (girls only). In contrast, early development was also related to some healthier behaviors, including higher rates of vigorous physical activity in both early-maturing boys and girls, and lower- (reported) fat intake in early-maturing girls. The higher rates of both vigorous activity and sedentary behavior among early-maturers may seem contradictory, but there is growing evidence that activity and inactivity behaviors are distinct behaviors and not two extremes of one (34). Associations of pubertal timing with vigorous activity and sedentary behavior were stable during the 5 years of follow-up in boys, whereas associations with breakfast routine in girls decreased over follow-up. In contrast, differences in fat intake in girls increased over time, resulting in significantly higher fat intakes in late-maturing girls at ages 14 to 16 years compared with average-maturing girls. Whether this reflects true differences related to body size, or, as may be more likely, more concern about and underreporting of fat intake in the earlier-maturing—and fatter—girls cannot be determined from these results.
Even though some effects of early pubertal timing decreased over time, our longitudinal analyses showed a continuing impact of pubertal timing beyond midadolescence, indicating that early timing relative to peers is the critical factor rather than puberty per se. Early maturation does not simply bring forward the onset of certain behaviors, it actually confers risk—or is a marker for higher risk. Few other studies have explored the continuing effect of pubertal timing throughout adolescence into young adulthood. Dick et al. (37) studied substance use in girls ages 16 to 18 years and concluded that there was some catch-up effect for average- and late-maturing girls, with the influence of early maturation dampening out as the majority of adolescents begin to experiment with substances. However, later-maturing girls were still demonstrating higher levels of abstinence at the age of 18 years, suggesting a persisting influence of pubertal timing into early adulthood (37). Graber et al. also reported continued excess of psychopathology at the age of 24 years in early-maturing girls although not in boys (38).
The pattern of the pubertal timing effect was similar for smoking and other health behaviors, indicating the early-maturing group at one extreme, the average-maturing group as the intermediate group, and late-maturing students at the other extreme. However, perceived stress was different, with both early- and late-maturing girls ("off-time") having higher stress during adolescence. These different patterns argue against stress mediating the relationship between pubertal timing and health behaviors in adolescents and consistent with this, our results were unchanged after adjusting the analyses for perceived stress levels. Others have also found little evidence for the mediating effect of stress in the relationship between pubertal timing and health behaviors (7,21,39).
A limitation of previous studies on pubertal timing effects was the focus on one gender. Our study showed that, for some behaviors (smoking, vigorous physical activity, and sedentary behaviors), the effects of pubertal timing are similar for boys and girls. The gender difference became obvious in dietary behavior, showing clearer effects of pubertal timing in girls than boys. Another strength of the present study was the inclusion of a range of health risk behaviors, which is important because the effects of pubertal timing are not the same across behaviors. General explanations for a negative impact of early maturation on lifestyle and behavior might not be applicable to all behaviors. Correlations between smoking, dietary habits, physical activity, and sedentary behavior in our sample were generally low (r < .20) indicating that adolescents' decisions and actions regarding smoking may be triggered differently from other health risk behaviors. From a problem-behavior perspective, smoking as a normative transgression could be motivated by goals such as rejecting the norms of conventional society, affirming membership in a peer group, asserting independence from parents, or being seen as more mature (40). Such functions are not necessarily implicated by other health behaviors, such as food choices and physical activity.
One possible explanation for the negative impact of early puberty on health behaviors is a mismatch between intraindividual attributes and physical maturity. Although physically developed, the adolescents may still be psychologically immature, lacking the cognitive skills to resist social pressure from peers (12). It has also been shown that early-maturing boys and girls tend to have older friends (41). This explanation might be particularly relevant for behaviors that are highly socially related, such as smoking. A qualitative study concluded that adolescent smoking is profoundly social and motivated by attempts to achieve the status of "cool" and "grown-up," or to conform to the norm of certain groups and gain group membership and identity (42). Yet, another explanation is that puberty brings about a transformation in parent-child relations, and thus early pubertal timing instills in adolescents a desire to break family ties by engaging in "deviant" behavior. Trends observed in daily breakfast routine might be explained by this theory.
Studies on the association between pubertal timing and behavior are potentially difficult because of confounding effects of age and change in school environment. The longitudinal design of the present study and inclusion of both genders is a major advantage, and we were also able to adjust for age. Moreover, the study focused on continuous years in the same secondary school, ruling out the influence of change in school environment. Loss to follow-up was a limitation, in common with most longitudinal studies, but our analyses are not restricted to students with complete data because the mixed models analysis does not need data from every time point and maximizes the use of available data. A strength of the study was that the measure of pubertal timing was based on assessments at several follow-up times, which increases the validity. Pubertal development scores at each assessment were checked for inconsistencies (i.e., decreases in pubertal development over time). Reliance on self-reported behavioral data was a short-coming, which might result in socially biased answers and underreporting of unhealthy behaviors. However, smoking data were validated with a biochemical measure of smoking exposure (saliva cotinine). Other health behaviors were measured with questionnaires that demonstrated to be valid in other studies. Although underreporting of unhealthy behaviors cannot be ruled out, the major area where it is likely to have affected our analyses relate to dietary fat. Earlier puberty is related to higher body weight, which is in turn usually related to greater body dissatisfaction and dieting. Dieters may be motivated to underreport fat intake as part of impression management.
In conclusion, the rapid development of unhealthier lifestyles across the domains of smoking, physical activity, sedentary behavior, and breakfast routine during adolescence in both genders warrants concern. Early-maturing adolescents are at increased risk for unhealthy behaviors, especially smoking, and although differences attenuate during adolescence, they are still significant at age 16 years. This suggests that early maturation could be as a cause of, or at least a marker for, differences in lifestyle.
We gratefully acknowledge the participation of the 36 schools and 5863 students and the work of all those involved in collecting the data.
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Received for publication September 26, 2006; revision received June 20, 2007.
DOI:10.1097/PSY.0b013e3181576106
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
