1、Negative LifeSymptoms inL. LaMontagne,and Nancy Wellscognimenteveerabiativein a, ruminativetvONE INUnited Statedisorder that leadspoint during chilet al., 2010).adolescent deprested, and the prevaldisorders incr(Fergusson Lewinsohn,tudiadolesdepresoldmajoradulthood, anxiety disorders, nicotine or al
2、coholunderachieve-and sui-rs, Graaf,Horwood,Solomon,ies of de-that negativelnerabilities)symp-Available online at -N.doi:10.1016/j.apnu.2011.04.008Archivesrelated to depressive symptoms. Longi-nal investigations have demonstrated thatcents who reported subsyndromal levels ofsive symptoms (i.e., symp
3、toms below thresh-for diagnosis) were at an increased risk fordepressive episodes later in adolescence andUniversity Medical Center, Nashville, TN.Corresponding Author: Cara C. Young, PhD, RN, FNPBC,VanderbiltUniversitySchoolofNursing,Nashville,TEmail addresses: cara.youngvanderbilt.edu(C.C. Young),
4、 lynda.lamontagnevanderbilt.edu(L.L. LaMontagne), mary.dietrichvanderbilt.edu(M.S. Dietrich), nancy.wellsvanderbilt.edu (N. Wells) 2012 Elsevier Inc. All rights reserved.0883-9417/1801-0005$34.00/0required for an adolescent to experience adverseoutcomesAndrews, 1993; Reinherz et al., 1993; Weissmane
5、t al., 1999). The most devastating outcome,suicide, is the third leading cause of death foryoung people 1524 years of age (CDC, 2007). Adiagnosis of a depressive disorder, however, is nottoms when adolescents are faced with negative lifeFrom the Vanderbilt University School of Nursing,Nashville, TN;
6、 Vanderbilt University School of Medicine(Biostatistics, Psychiatry), Vanderbilt University Schoolof Nursing, Nashville, TN; Nursing Research, Vanderbiltof PsychiatrEVERY four to five youth in thes will be diagnosed with a mentalto severe impairment at somedhood or adolescence (MerikangasThe negativ
7、e sequelae related tosive disorders are well documen-ence ofdepressive symptoms andeases steadily across adolescenceWoodward, 2002; Glied CuijpeFergusson,Ridder, Lewinsohn,Seeley, Beck,1967; Hankin Clark Kovacs Brown, Hammen, Craske, Lewinsohn,Joiner, Abramson et al., 1989). For example,an adolescen
8、t girl who interprets a negative event(e.g., a breakup with her boyfriend) using anegative inferential style concludes that thebreakup was because she is not pretty enoughand infers she will never be pretty enough to haveanother boyfriend. She develops a negative self-evaluation that she is undesira
9、ble to the oppositegender leading to feelings of hopelessness andsadness. Strong empirical evidence supports therelationship between the cognitive vulnerability ofnegative inferential style and negative life eventsas predictors of the development of depressivesymptoms in both male and female adolesc
10、ents(Abela, 2001; Abela Hankin,Abramson, Nolen-Hoeksema, Wisco, Park, Goodyer, Schwartz Riso et al., 2003; Robinson Spasojevic Hankin Weissman, Orvaschell, Weissman Mezulis,Hyde, Nolen-Hoeksema i.e., education andoccupation). The FIQ was completed by a parentor guardian and returned with the signed
11、informedconsent. SES was determined using a combinationof Hollingsheads Two Factor Index of SocialPosition (1965) and Barratts (2006) SimplifiedMeasure of Social Status (BSMSS). The BSMSSwas developed based on Hollingsheads measureof social status. The BSMSS provides an updatedlist of occupations to
12、 improve relevance to present-day occupations. The updated occupations of theBSMSS and the highest level of educationalattainment were used to determine each partici-pants social status as defined by Hollingsheadsrating index (1965). Five classifications weresuggested by Hollingshead to determine SE
13、S.The highest class, Class I, includes major businessprofessionals (scores 1117); Class II includeslesser professionals (scores 1827); Class IIIincludes skilled craftsmen and clerical and salesworkers (scores 2843); Class IV includes semi-skilled workers (scores 4460); and Class Vincludes unskilled
14、laborers (scores 6177). Inthis study, education and occupation of bothparents/guardians were averaged and used tocalculate SES unless the adolescent lived withone parent only, in which case only that parentseducation and occupation were used. Lower scoresindicate higher SES.Negative Life EventsAdole
15、scent Life Events Questionnaire (Hankin(Weissman et al.,1980). The CES-DC is a modified version of theCES-D used extensively with adult populations.This modified version is a 20-item self-report scalethat assesses the frequency of depressive symptomsover the past week. Scoring of items range from0(n
16、ot at all)to3(a lot). Four items were reversescored. A total scale score was calculated bysumming item responses. Scores had a possiblerange from 0 to 60, with higher scores indicatinggreater frequency of depressive symptoms. Theinternal consistencies in adolescent samples havebeen reported to be go
17、od ( = .86) and testretestreliability moderate (r = .61; Faulstich, Carey,Ruggiero, Enyart, WeissmanWeich,Churchill, Weissman Lewinsohn, Gotlib, Just Nolen-Hoeksema, Parker, Nolen-Hoeksema therefore, for consistency of presentation,median and 25th75th interquartile ranges (repre-senting the middle 5
18、0% of values) are presented tosummarize all of the study measures. Skeweddistributions were rank-transformed to meet theparametric assumptions of the tests of association(Pearsons correlations and simple and multiplelinear regression) used in this study. With theexception of gender, all of the varia
19、bles included inthe correlational and regression analyses werecontinuous in nature; thus, Pearsons correlationswere used to assess univariate associations amongthe scales. The inclusion of gender in such ananalysis resulted in a point-biserial correlation.Tests of the differences between dependent c
20、orre-lations were conducted using z-statistics.Hierarchical linear regression analysis was usedto evaluate the unique (adjusted) associations ofexplicitly ordered sets of variables with depressivesymptoms. All variables were forced into themodel under the control of the investigator.Personal charact
21、eristics (i.e., gender and SES)were entered first in the analysis; negative lifeevents were entered in Step 2 to assess theirimportance after controlling for personal charac-teristics. Finally, because cognitive vulnerabilitiescomprise the variable of most interest in this study,those were entered i
22、n the final step (aftercontrolling for the associations of both personalcharacteristics and negative life events). Tolerancestatistics and Variance Inflation Factors (VIF) wereused to evaluate the possible existence of multi-collinearity in the final step of the hierarchicalmultiple-regression model
23、. Tolerance statistics foreach of the variables in the model were greaterthan .01, and all VIF were less than 10, indicatingthe model was not affected by multicollinearity. Amaximum alpha of .05 was used for tests ofstatistical significance.RESULTSThe sample with complete data consisted of 63females
24、 (57%) and 48 males (43%), ranging in agefrom 12 to 15 years. Most were Caucasian (93%),which was representative of the schools from whichthe potential participants were recruited (TennesseeDepartment of Education, 2010). In this sample,approximately 81% of the parents/legal guardiansreported a high
25、 school education/GED or higher.Using the updated occupations in the BSMSS(Barratt, 2006) and Hollingsheads Two FactorIndex of Social Status (1965) classification system,more than 75% of the participants social status fellinto Class IV (e.g., clerical and sales workers,technicians, and construction
26、laborers) and Class V(unskilled laborers). See Table 1 for a summary ofdemographic data.The median number of negative life eventsreported on the ALEQ was 21, with the middle50% ranging from 11 to 31. Ninety-six percent(n = 107 of 111) of participants reported 5 or moreessentially homogeneous for eth
27、nicity (93% Cau-casian); thus, no tests of association for ethnicitywere conducted.Statistically significant relationships were foundfor each of the three cognitive vulnerabilities withnegative life events (see Table 3). Participants whoreported higher numbers of negative life events(e.g., fights wi
28、th parents, breakup of a romanticrelationship, failing a test in school) reported moreTable 1. Summary of Participant Demographic DataVariable Total sample (N = 111), frequency (%)GenderMale 48 (43)Female 63 (57)Age (years)12 28 (25)14 YOUNG ET AL13 51 (46)14 29 (26)15 3 (3)Grade7th 46 (41)8th 65 (5
29、9)EthnicityCaucasian 103 (93)Hispanic 2 (2)African American 3 (3)Other 3 (3)Parental marital statusSingle 23 (21)Married/Livingwith partner 88 (80)Parental education levelSome high school or less 39 (19)High school grad/GED 66 (33)Some college 48 (24)College education 32 (16)Graduate degree 18 (9)SE
30、SI (highest) 0 (0)II 2 (2)III 25 (23)IV 50 (45)V (lowest) 34 (31)negative life events. Depressive symptom scores onthe CES-DC ranged from 1 to 53, with a median of14 (middle 50%, 8 to 23). Scores for negativeinferential style on the CCSQ ranged from 1.05 to3.75, with approximately 96% (n = 106) ofpa
31、rticipants scoring lower than 3. Descriptivestatistics for all study measures are presented inTable 2 (N = 111).Univariate Associations Among PersonalCharacteristics, Negative Life Events,Cognitive Vulnerabilities, and DepressiveSymptomsHigher numbers of negative life events werepositively correlate
32、d with higher levels of depres-sive symptoms (r = .61, P b .001). Higher levels ofall three cognitive vulnerabilities were statisticallysignificantly associated with increased reports ofdepressive symptoms (see Table 3), with thestrongest association found between the cognitivevulnerability of rumin
33、ative response style anddepressive symptoms (r = .88, P b .001). Asshown in Table 3, none of the personal character-istics of gender, age, or SES achieved a statisticallysignificant level of association with depressivesymptoms. The strongest pattern, however, was forgender (r = .18, P = .058). Altho
34、ugh not statisticallysignificant, females tended to report more depres-sive symptoms than males. One statisticallysignificant relationship was detected between SESand dysfunctional attitudes. Participants from lowerSES backgrounds reported more dysfunctionalattitudes (r = .24, P = .012). The sample
35、wasTable 2. Descriptive Statistics of Measures (N = 111)Scale (score range) Mdn IQR (25th, 75th) Min, MaxALEQ (070) 21 11, 31 0, 53CES-DC (060) 14 8, 23 1, 53DAS (040) 20 15, 24 10, 37CCSQ (15) 1.6 1.4, 2 1.05, 3.75RRS (188) 38 29, 49 22, 78NOTE. Mdn = Median IQR = 25th and 75th interquartile range.
36、dysfunctional attitudes (r = .49, P b .001), negativeinferential styles (r = .49, P b .001), and ruminativeresponse styles (r =.63,P b .001). Test ofdifferences were calculated among the dependentcorrelations of the three cognitive vulnerabilitieswith number of negative life events to determine ifth
37、ere were differences in the strength and directionof those relationships. As noted above, all theTable 3. Correlations Between KeyMeasure 23451. ALEQ .61 (b.001) .49 (b.001) .49 (b.001)2. CES-DC .41 (b.001) .59 (b.001)3. DAS .57 (b.001)4. CCSQ5. RRSNOTE. Values in each cell are r (P value). Gender:
38、female = 0, male = 1.Correlation is point-biserial. All other correlations are Pearsons correlati15ADOLESCENTS DEPRESSIVE SYMPTOMScorrelations were in the positive direction. Howev-er, one statistically significant difference in thestrength of those relationships was found. Therelationship of rumina
39、tive response style withnegative life events (r = .63) was statisticallysignificantly stronger than that of negative inferen-tial style with negative life events (r = .49, Z0=2.30, P = .021).Unique Contribution of Cognitive Vulnerabilitiesto Depressive SymptomsThe multivariate associations of the hy
40、pothe-sized explanatory variables with depressive symp-toms are summarized in Table 4.The initial step ofTable 4. Hierarchical Multiple Linear Regression Examining theContribution of Cognitive Vulnerabilities to DepressiveSymptoms (N = 111)Variables PRPR2change PStep 1 .20 .118 .04 .118Gender .18 .0
41、56SES .08 .834Step 2 .62 b.001 .34 b.001Gender .08 .309SES .02 .763ALEQ .60 b.001Step 3 .88 b.001 .39 b.001Gender .04 .354SES .01 .841ALEQ .10 .117DAS .03 .678CCSQ .03 .677RRS .80 b.001NOTE. Multiple R = .88, P b .001; R2= .773 (adjusted R2= .760).Gender: female = 0, male = 1.the analysis included t
42、he participants gender andSES. These personal characteristics accounted forapproximately 4% of the variability in depressivesymptoms, and the multivariate association was notstatistically significant (multiple R = .20, P = .118,adjusted R2= .02). To assess the unique contribu-tion of negative life e
43、vents to the variability indepressive symptoms after controlling for thepersonal characteristics, the negative life eventsscore was added to the model in the second step.With the inclusion of these scores, there was astatistically significant increase in the ability toexplaindepressive symptomsthat
44、went from 4%(inthe previous step) to a total of 38%. The resultingmultiple correlation was statistically significant(multiple R = .619, P b .001, adjusted R2= .366).After controlling for the contributions ofpersonal characteristics and negative life eventsscores in the first two steps, the set of co
45、gnitivevulnerability variables was forced into the model.This addition resulted in another statisticallysignificant increase (39%, P b .001) in the abilityto explain the variability in depressive symptoms(from the 38% shared variability of the previousstep to a final shared variability of 77%). Ther
46、esulting overall multiple correlation of the entireset of variables (gender, SES, negative life events,cognitive vulnerabilities) with depressive symp-toms was .88 and was statistically significant (P b.001). After controlling for the observed associ-ations of all of the other study variables in thi
47、sStudy Variables (N = 111)GenderAge SES.63 (b.001) .17 (.079) .01 (.903) .17 (.083).88 (b.001) .18 (.058) .03 (.782) .08 (.433).47 (b.001) .13 (.168) .16 (.097) .24 (.012).66 (b.001) .07 (.478) .02 (.842) .18 (.065).15 (.114) .01 (.935) .09 (.359)ons.model, ruminative response style remained theonly
48、 variable demonstrating a statistically signif-icant unique contribution to the number ofdepressive symptoms ( = .80, P b .001).DISCUSSIONThis study is one of the first to examine all threecognitive vulnerabilities (i.e., dysfunctional atti-tudes, negative inferential style, and ruminativeresponse s
49、tyle) with negative life events and16 YOUNG ET ALdepressive symptoms in a sample of youngadolescents. The high percentage of variance indepressive symptoms explained by the full regres-sion model and the unique contribution of rumina-tive response style provide strong support for therole that cognitive vulnerabilities have in thedevelopment of depressive symptoms when ado-lescents experience negative life events.The mean number and range of depressivesymptoms found in this sample were similar tothose previously presented
