| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the School of Nursing, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina.
Address correspondence and reprint requests to Hyekyun Rhee, Department of Family, Community and Mental Health Systems, School of Nursing, University of Virginia, McLeod Hall, PO Box 800782, Charlottesville, VA 22908. E-mail: hr3k{at}virgnia.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Methods: This secondary data analysis used longitudinal data from a nationally representative sample of adolescents in grades 7 through 12 (n = 9,141) who participated in the National Longitudinal Study of Adolescent Health during 1994 to 1996. Ten selected physical symptoms (i.e., headache, stomachaches, fatigue) were used to construct clusters. Each cluster was compared in regard to demographic factors, psychological adjustment and interpersonal relationships.
Results: K-means in combination with Ward method clustered the sample into 4 groups according to the overall patterns of the 10 symptoms: nonsymptom (41%), moderate symptom (38%), high symptom (19%) and extreme symptom (2%). Adolescents in higher symptom clusters were more likely to be girls, nonwhites, or from families on welfare and reported high depressive symptoms, low self-esteem, and poor perceptions of parental affection and friendship quality. About 16% in clusters with lower symptom patterns develop somatizing patterns in Wave II; new onset was predicted by gender, younger age, and depressive symptoms.
Conclusion: Symptom patterns characterized by overall high frequencies of multiple symptoms may indicate somatization. This study also suggests that adolescents with a somatizing tendency are more likely to experience psychological and interpersonal difficulties, and girls and younger adolescents are more vulnerable. Targeted prevention programs are needed for these vulnerable individuals by addressing their psychosocial functioning.
Key Words: multiple symptoms • clusters • psychological adjustment • interpersonal relationships • stability
Abbreviations: Add Health = the National Longitudinal Study of Adolescent Health; CES-D = the Center for Epidemiological Studies-Depression Scale; SES = socioeconomic status; ANOVA = analysis of variance; OR = odds ratio; 95CI = 95% confidence interval; NS = nonsymptom; MS = moderate symptom; HS = high symptom; ES = extreme symptom.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Studies have shown relationships between physical symptoms and various psychosocial factors in children and adolescents (8). Children's and adolescents' perceptions of themselves and their emotional status have consistently been found to be associated with their physical symptom experience (4,5,9,10), and poor relationships with parents or peers have been linked to a high incidence of physical symptoms. Because of the complicated relational web surrounding the psychological and interpersonal relationship factors in the context of adolescent physical symptoms, failure to address these factors simultaneously may result in misleading or incomplete conclusions. Thus, it is desirable to take into consideration those multiple factors concurrently when examining physical symptoms.
The purpose of this study was to reveal patterns of physical symptoms using a clustering approach and to examine relationships between the identified patterns and psychosocial factors. The current study addressed the following research questions: (1) Can adolescents be clustered in a meaningful way according to patterns of 10 physical symptoms? (2) What are the psychosocial factors (psychological adjustment, interpersonal relationships, and demographic factors) associated with each symptom pattern? (3) Will the identified symptom patterns be stable over time? (4) What is the extent to which individuals maintain their symptom patterns?
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
The present study used 9140 adolescents from the core sample who participated in both Waves I and II. The demographic characteristics of the study sample are summarized in Table 1. Age of adolescents in this sample ranged from 11 to 21 years, and the majority (52%) were between 15 and 18.
|
Variables and Measurements
Physical Symptoms
Items for 10 physical symptoms were extracted from a general health survey questionnaire consisting of 60 questions total. Four symptoms—headache, stomachache, musculoskeletal pain, and fatigue—were selected because of common occurrence. In addition, 6 symptoms—feeling hot, cold sweats, sore throat, urinary problems, dizziness, and chest pain—were selected to represent physical discomforts. In individual in-home interview questions, adolescents rated the frequency of having experienced each of the 10 physical symptom during the past 12 months, using a 5-point scale: 0 = never, 1 = just a few times, 2 = about once a week, 3 = almost every day, and 4 = every day.
Psychological Adjustment
Self-esteem questions included 8 items selected and modified from the Self-Esteem Inventory (12). Each item was measured on a 4-point rating scale, ranging from 1 (most of the time or all of the time) to 4 (absence). All items were inversely coded for conceptual consistency, with a higher score indicating better self-esteem. Each total summed score was averaged by the number of items answered. Mean scores ranged from 1 to 4. Cronbach 
in this study sample was 0.86.
Depressive symptoms were assessed using the Center for Epidemiological Studies-Depression Scale (CES-D) (13). The Add Health study dropped 2 items on sleep problems and crying spells and added 1 item on a feeling of worthlessness. Each of the 19 items was scored from 0 (complete absence of the symptom) to 3 (most of the time or all of the time during the past week). Individuals' total summed scores were averaged by the numbers of items answered; the higher the score, the greater the depressive symptoms. The content, criterion, and construct validity of the CES-D scale were demonstrated in numerous studies, and the scale has shown excellent reliability in adolescent populations (14). Cronbach in the present study was 0.87.
Interpersonal Relationships
Three aspects of parent-child relationships were examined based on adolescents' reports: parental affection, parental involvement and parental control. Parental affection was evaluated using 5 items measuring the level of closeness, warmth, care, and communication with parent and overall satisfaction with the relationship with parent. The values of each item ranged from 1 (strongly agree) to 5 (strongly disagree). Summed scores were averaged; higher scores indicate higher perceived parental affection. Cronbach were 0.88 for both mothers and fathers in the present sample. A single variable was constructed by taking the maximum level of parental affection from either parent.
Parental involvement was measured by 9 activities (e.g., going shopping, playing a sport, participating in cultural events) that the parent and adolescent did together in the last 4 weeks. Maternal and paternal involvements were separately measured on a dichotomous rating scale, 0 (no) and 1 (yes). The total number of shared activities was counted, with possible scores ranging from 0 to 9; higher scores indicate higher parental involvement. The higher of the values between the maternal and paternal scores were used to construct a single variable.
Parental control was constructed from a series of 7 questions; whether the adolescent was allowed to make his or her own decisions about weekend curfew, friends, clothes, how much TV to watch, which TV program(s) to watch, what time to go to bed on weeknights, and what to eat. A dichotomized response set, 0 (no) and 1 (yes), was reversely coded so that a score of 1 indicated parental control on each particular matter. A single score was obtained by summing up each answer, ranging from 0 to 7; higher scores indicate high overall parental control.
Peer relationships focused on friendships. Two aspects of friendship were used: friendship quantity and perceived quality of friendship. Peer friendship quantity took into account the number of best friends (maximum 5 from each gender) and shared time and activities (e.g., going to the friend's house, hanging out after school, talking) with them during the past 7 days. Each item had a dichotomized response set, no (0) and yes (1). The total friendship quantity score ranged from 0 to 50. Individuals reporting many best friends scored higher than those with fewer best friends. A single item was used to capture the perceived quality of peer friendship: "How much do you feel that your friends care about you?" Respondents rated this item on a 4-point scale, ranging from 0 (not at all) to 4 (very much).
Demographic Factors
As a proxy measure of socioeconomic status (SES), parental welfare status was classified into 2 groups, "welfare" versus "no welfare." If 1 or both parents were on welfare, then "welfare," and if neither of parents was on welfare, then "no welfare." The adequacy of using the self-report indicator of welfare status has been supported in a study using adolescent respondents (15). Regarding family structure, adolescents were categorized into either "living with both parents" or "not living with both parents" based on their reports. In addition, gender, race/ethnicity, and age were also measured.
Human Subject Protection
A signed consent from a parent or guardian to interview the adolescent was obtained, along with the adolescent's signed assent before participation. To ensure confidentiality, no paper questionnaires were used; all answers were recorded on laptop computers. All Add Health procedures and questionnaires were reviewed and approved by the institutional review board for the Protection of Human Subjects. Extensive care was taken to ensure the confidentiality of data throughout analysis.
Data Analysis
Cluster analyses were conducted using 10 physical symptoms, including headache, stomachache, musculoskeletal pain, fatigue, feeling hot, cold sweats, sore throat, urinary problems, dizziness, and chest pain. As each symptom was measured on the same 5-point scale, no standardization was necessary. Given the large sample size, a nonhierarchical clustering method, "k-means" was used (16,17). This method was complemented by a hierarchical technique, "Ward method," to establish the number of clusters and to profile cluster centers. Canonical discriminant function analysis technique was used to obtain variables useful in describing differences among clusters. This analysis also yielded information about the degree of variance accounted for by these factors. Procedures were implemented using SAS (17). To validate the obtained cluster solution through replication (16), the present study randomly split the study sample into 2 groups, and analyzed each group using identical clustering procedures (18,19). Each cluster was compared using 
2, multiple univariate analyses of variance (ANOVAs) or logistic regressions, depending on the levels of measurement of the psychosocial variables.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
|
Cluster Analysis and Discriminant Function Analysis (Research Question 1)
A total sample was randomly divided into 2 subgroups; "the main group" and "the replication group." No significant differences between the 2 groups in demographic characteristics confirmed a successful random split. Using the main group, Ward's minimum variance approach was initially applied using the 10 physical symptoms. Because Ward method is an exploratory analysis identifying the natural groups of elements and is less sensitive to outliers than other cluster methods (20), outliers were not removed in analysis. Decisions on the appropriate number of clusters were made based on the between sum of squares (BSS), root mean square SD (RMSSTD), R-square (RSQ), and semipartial R-square (SPRSQ). Evaluation of the values of each measure indicated four as the appropriate number of clusters. The Ward method provided four seeds to act as the initial centroids for the k-means approach that refines the nearest centroid sorting. In the k-means method, observations are repeatedly reassigned to the cluster with nearest seeds that are subsequently updated. Maximum iteration was fixed to 20 in this study as suggested by Khattree and Naik (17). Table 3 summarizes the profiles of means for the 10 symptoms in each cluster.
|
Cluster 1, as the largest group (40.7%, n = 1869), was characterized by a profile in which the mean of each symptom was below average and thus was labeled "nonsymptom (NS)." The profile of cluster 2 (38%, n = 1732) almost overlapped with the average profile, as such labeled "moderate symptoms (MS)." Cluster 3 (19%, n = 861) had an overall profile higher than average with a peak in fatigue and was labeled "high symptoms (HS)." Last, cluster 4 (2.3%, n = 107), exhibited an overall profile highest among the 4 clusters and was thus labeled "extreme symptoms (ES)." Cluster 4 was distinguished from cluster 3 by its strikingly high mean in urinary symptoms. Standardized canonical discriminant function coefficients confirmed that 94.3% of originally grouped individuals were correctly classified in each cluster. An identical number of clusters with identical configurations was successfully reproduced on the replication group, providing evidence of the validity of the cluster solution. The successful replication indicates that outliers were unlikely to be a significant factor determining the cluster solution.
Clusters and Psychosocial Factors (Research Question 2)
2 and logistic regressions were used to assess significant differences among the 4 clusters in terms of categorical demographic variables (Table 4). Significant gender and race differences were noted. The odds of girls being classified into HS and ES (versus either MS or NS) were significantly higher than that of boys (odds ratio, OR = 1.8; 95% confidence interval, 95CI, 1.65–2.02, p < .001; OR = 1.7, 95CI, 1.54–1.92, p < .01, respectively), whereas boys were more likely to be in the NS than the other clusters (OR = 1.4, 95CI, 1.22–1.82, p < .001). Over a fourth of the girls were classified into the 2 higher-symptom groups.
|
The likelihood of nonwhites being classified in either NS and ES were 1.5 to 2 times that of whites (p < .001) who were more likely to be in MS (versus other clusters) (OR = 1.6; 95CI, 1.25–1.78; p < .001). The rate of respondents being grouped into ES was tripled when parents were welfare recipients in comparison with respondents of nonwelfare status (OR = 2.9; 95CI, 2.76–4.02; p < .001). Over 30% of welfare status adolescents were grouped into either HS or ES. In contrast, the likelihood that respondents of nonwelfare status would be in MS (versus the other clusters) was significantly higher (OR = 1.7; 95CI, 1.25–1.97; p < .001). Respondents who did not live with both parents were more likely to be in either HS or ES (versus the other 2 clusters) than those living with both parents (OR = 1.3; 95CI, 1.19–1.59; p < .01; OR = 1.5; 95CI, 1.43–1.75; p < .05, respectively).
Multiple univariate ANOVAs were performed to examine the extent to which each cluster differed for psychosocial variables. Due to the skewed data distribution, a log-transformation was performed for all the variables. Pairwise comparisons revealed that the most frequent significant differences were found in the NS cluster who exhibited favorable psychological adjustment, low depressive symptom scores, and high self-esteem, and a high level of parental affection. HS and ES showed substantially greater depressive symptoms and significantly lower self-esteem than either NS or MS. A low level of parental affection and a high level of parental involvement were also evident in HS. The ES group appeared most vulnerable in most variables included.
Such differences were reaffirmed by constructing a profile of those variables, after being standardized with a Z-score, for each cluster, which illustrates the relative strength of each factor compared with the others (Figure 1). The profiles of NS and MS were quite similar, with most scores within ±0.5 SD range. An approximately inverse pattern was exhibited in HS; in particular, depressive symptoms were higher and self- esteem and parental affection were lower. Despite differences in patterns, the size of score variations was small, confined within ±0.5 SD. ES exhibited an atypical profile in which the Z-score of depressive symptoms elevated close to 1.5 SD, and self-esteem was below –0.5 SD.
|
Stability of Symptom Pattern Configurations (Research Question 3)
As in Wave I, Ward approach and k-means method suggested 4 clusters as an ideal solution for Wave II. Similarly, these 4 clusters differed in the level of overall symptoms, although identical patterns to those from Wave I data were not reproduced. The same labels as Wave I were applied: cluster 1 "NS cluster" (56%, n = 2574); cluster 2 "MS cluster" (20%, n = 906); cluster 3 "HS cluster" (12%, n = 555); cluster 4 "ES cluster" (12%, n = 532).
Stability of Cluster Membership (Research Question 4)
A 4 x 4 contingency table was constructed, using the cluster memberships in Wave I and Wave II (Table 5). Stability in cluster memberships was demonstrated by those in NS, in which 78% stayed in their cluster over time. Over 50% of adolescents in MS in Wave I showed an improved symptom pattern in Wave II by changing their membership to NS in Wave II, whereas 33% in HS changed their membership to ES in Wave II. Individuals in ES did not show any significant trend of membership change. Overall, individuals' cluster memberships in lower levels of symptoms patterns (NS and MS) tended to be stable as the majority of adolescents remained in these clusters rather than progressing to either HS or ES. Instability in cluster membership was more evidenced among individuals in the HS and ES clusters who were more likely to change their cluster membership during the course of a year.
|
Additional analyses were conducted for the subgroup of individuals in the 2 lower level clusters, NS and MS, in Wave I. The majority (n = 3031, 84%) maintained their memberships in either NS or MS in Wave II, and only a small proportion (n = 567, 16%) reported changing to higher level clusters. The stable subgroup and the unstable subgroup were compared through logistic regressions, in terms of psychosocial factors, and found to differ in multiple ways. The unstable subgroup included more girls (OR = 1.63; 95CI, 1.36–1.95; p = .0001), adolescents who did not live both parents (OR = 1.22; 95CI, 1.02–1.46; p = .03), and adolescents from families receiving welfare (OR = 1.35; 95CI, 1.02–1.79; p = .04). Most significantly, the odds of being in the unstable subgroup increased substantially with an increase in depressive symptoms (OR = 2.93; 95CI, 2.30–3.73; p < .0001) and a decrease in self-esteem at Wave I (OR = 0.60; 95CI, 0.51–0.71; p < .0001). The likelihood of being in the unstable subgroup also decreased with an increase in parental affection (OR = 0.74; 95CI, 0.63–0.87; p < .001). A multiple logistic model was computed using the score of each psychosocial factor in Wave I to examine unique contribution of each factor to membership changes by addressing multicollinearity. Three factors emerged as significant predictors for being in the unstable subgroup: being a girl (OR = 1.45; 95CI, 1.19–1.77; p < .001), being younger (OR = 0.93; 95CI, 0.87–0.99; p < .05), and having more depressive symptoms (OR = 2.54; 95CI, 1.89–3.42; p < .001).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
The ES cluster was differentiated from the HS cluster in Wave 1 by an unusually high occurrence of urinary symptoms. Urinary symptoms have rarely received attention in studies of adolescent somatic symptoms. A few studies have suggested that childhood urinary symptoms, particularly painful and frequent urination, are psychosomatic rather than organic in nature (22). The involvement of underlying organic diseases, such as urinary tract infection, cystitis, or sexually transmitted diseases (23,24), cannot be completely excluded. Nonetheless, when viewed in the context of a pattern in which reports of other unrelated symptoms were also elevated simultaneously, it seems plausible to understand urinary symptoms as a manifestation of somatization rather than organic dysfunction.
Symptom Clusters and Psychosocial Factors
In agreement with earlier studies reporting girls' high propensity toward somatization (8), we found that girls were more likely to be in the 2 clusters associated with somatizing. Nonwhites tended to be in either the NS or the ES cluster. This finding may reconcile conflicting reports that have identified nonwhites either as symptom prone (25,26) or symptom resistant (27). The most distinctive contrast related to SES was observed between the MS and the ES clusters. The former cluster tended to be composed of well-to-do adolescents, whereas the latter consisted primarily of families on welfare. The SES difference in the 2 clusters could simply reflect a covarying relationship between SES and race in the data. The MS cluster included more whites and those from higher SES, whereas the ES cluster consisted of more nonwhites and those with lower SES.
The youth with higher levels of symptom patterns had lower self-esteem and higher depressive symptoms. A striking cormorbidity between psychological maladjustment and physical symptoms has been documented (28,29). A negative association between self-esteem and physical symptoms has been reported in other studies of adolescents (29,30). This study confirmed that such relationships hold true even when patterns of physical symptoms are considered.
A recent study showed that children with recurrent stomach pains had poorer perceptions of their relationships with parents than their counterparts without such pains (30). Similarly, we found that adolescents with a somatizing tendency reported lower perception of parental affection than those with low levels of symptoms. As in a prior study (31), we found positive associations between the levels of symptom patterns and parental involvement. Based on clinical observations, Minuchin et al. (32) identified family "enmeshment" as a factor related to the development and maintenance of psychosomatic symptoms in children. Similarly, Roy (33) found marked degree of closeness or togetherness between family members as a risk factor for chronic pain. Further research is needed to delineate the specific mechanisms by which particular attributes of parent-child relationships influence adolescents' somatizing tendencies.
Consistent with previous findings (3,30,34), we found associations between peer relationships and symptom patterns. Adolescents in the clusters associated with a somatizing tendency reported high numbers of friends but low friendship quality (feelings of being cared about), whereas those in the NS cluster reported the fewest number of friends but high friendship quality. Thus, supportive and caring peer relationships, rather than the numbers of friends, may protect youngsters from physical symptoms. Future research is needed to better understand the influence of the quantity and quality of friendship on adolescent development and health.
Stability of Physical Symptom Patterns
Literature addressing the issue of stability of multiple physical symptoms is limited. Some studies have examined the stability in single or only a few specific symptoms (19,29,35). Using high school students, Poikolainen et al. (29) found a striking persistence of 1 or more recurrent physical symptoms in 73% of the total sample over a 5-year course. Wangby (6) examined the stability of symptom patterns involving stomachache and tiredness and concluded that patterns in midadolescent girls (15–16 years) were fairly stable over a year. However, when 3 or more symptoms were considered, stability over 1-year period was not evident in a majority of adolescents.
Overall, our study did not support the stability of symptom patterns as the analyses failed to generate identical sets of symptom patterns for both Waves. This instability may be explained in three ways. First, recollection bias might affect physical symptom reporting (36). In this study, adolescents were required to recall the frequencies of 10 symptoms over the past year. Recollection errors may have been greater for symptoms occurring rarely, such as urinary symptoms, chest pain, and cold sweat. Second, the manifestation of somatic symptoms may change situation by situation, a possibility supported by other population-based studies (37,38). Finally, instability in symptom patterns may be a reflection of unstable developmental status during adolescence. Although the manifestation of the symptom patterns at Wave I and Wave II were not exact duplicates of each other, data from Waves I and II generated the same numbers of symptom patterns. The long-term stability of individuals' general propensity to multiple somatic symptoms, rather than any single complaint, may be an indication of somatic syndrome rather than of an organic base (39).
The majority of those in the low symptom clusters (84%) remained free of somatization over a 1-year period. A small proportion (16%) developed patterns associated with somatizing. Depressive symptoms emerged as the most powerful contributing factor to the maintenance and development of these patterns. Simon and Gureje (36), using an adult sample, suggested that the presence of depression sustained somatic symptoms over a 1-year period. Many studies have debated causal links between depression and somatic symptoms (10,37,38,40). The current study renders support for depression as an antecedent of somatic symptoms by demonstrating that the new onset of somatic patterns is predicted by depression in the early time point.
We also found that the younger the participants, the more likely they were to develop patterns associated with somatizing. A study showed that 2-year stability coefficients of somatic symptoms obtained from 11 to 12 years were lower than those in 15- to 16-year-old adolescents (41). In another study, somatic symptom scores were less stable in younger ages (10 to 13 years) and became more stable as youngsters aged (15 to 16 years) (6). These findings suggest that the tendency for somatic symptoms may not be established until youngsters reach midadolescence.
This study applied clustering methods to multiple symptoms. Also, this study was more comprehensive in scope, capturing multidimensional human functioning germane to adolescence. A longitudinal approach enabled us to address an important developmental concept, namely, continuity versus discontinuity, in the realm of physical symptoms. Moreover, the use of a nationally representative large sample of adolescents strengthened generalizability of the findings. Nonetheless, caution should be exercised in determining the scope of applicability of current findings for several reasons. First, the selection of physical symptoms was not based on existing standardized instruments such as Children's Somatization Inventory (4) or Somatic Symptom Checklist (1); thus, replication of this study could be hampered if different definitions of symptoms were used. Also, analyses were solely based on the frequency of symptoms; results might not have been the same if intensity or severity of symptoms had been taken into consideration. In addition, a 1-year time interval may have been too long to guard against recollection errors or too brief to observe meaningful changes in symptom patterns or psychosocial factors. Importantly, because this study was still correlational in nature, implications for causal relationships between the cluster solutions and the demographic characteristics and psychosocial factors should not be made. Thus, the findings should be interpreted with caution. Further research is warranted to address the limitations and to test the replicability of the findings.
The classification of individuals based on similar patterns of multiple symptoms made it straightforward to obtain a rich description pertaining to each group's psychosocial properties. Such information allowed us to recognize areas needing attention or interventions for a group of individuals sharing the similar symptom pattern. In addition, the inconsistent nature of symptom patterns and adolescents' tendency to physical symptoms guard against labeling young people "somatizers" yet underscore the importance of continuous monitoring of symptom complaints while being vigilant about potential risk for somatization. Clinicians need to be attentive to youngsters' psychosocial status, such as emotional conditions, family dynamics, and peer relationships, in evaluating symptom complaints. The study also highlighted the importance of early detection of risk factors and interventions for younger adolescents before these individuals develop somatizing tendencies as they age. Particularly, screening for and treatment of depression is a necessary step in addressing somatizing complaints in adolescents.
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
This research uses data from Add Health, a program project designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris and funded by a grant P01-HD31921 from the National Institute of Child Health and Human Development, with cooperative funding from 17 other agencies. Special acknowledgment is due Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design. Persons interested in obtaining data files from Add Health should contact Add Health, Carolina Population Center, 123 W. Franklin Street, Chapel Hill, NC 27516-2524 (www.cpc.unc.edu/addhealth/contract.html).
DOI:10.1097/01.psy.0000188404.02876.8b
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  G. S. Goldfield, R. Mallory, T. Parker, T. Cunningham, C. Legg, A. Lumb, K. Parker, D. Prud'homme, and K. B. Adamo Effects of Modifying Physical Activity and Sedentary Behavior on Psychosocial Adjustment in Overweight/Obese Children J. Pediatr. Psychol., August 1, 2007; 32(7): 783 - 793. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
