- Email: [email protected]
Nature of Life Events and Difficulties in Depressed Adolescents
Nature of Life Events and Difficulties in Depressed Adolescents DO UGLAS E. WILLIAMSON, B.A., BORIS BIRMAHER, M.D., ELLEN FRANK, PH.D., BARBARA P. ANDERSON, M.S., MARY K. MATTY, B.S., AND DAVID]. KUPFER, M.D.
ABSTRACT Objective: To examine the significance of acute life events and ongoing difficulties in adolescents with a recent major depressive disorder. Method: Adolescents (aged 13-18 years) with a recent episode of major depress ive disorder based on DSM-If/·R (n
=26) and normal controls free of any Axis I lifetime psychiatric disorder (n =15) were assessed using the
investigator·based Life Events and Difficulties Schedule (LEOS). Results: Traditionally defined severe events were more likely to occur in the year prior to onset among depressed adolescents (46%) than in a comparable period among normal controls (20%), but these differences did not reach statistical significance . Expanding the definition of severe events to include those events focused on others impo rtant to the adolescent resulted in a significantly higher percentage of depressed adolescents having one or more refined "severe" events in the year prior to onset (62%) compared with normal controls (27%) (p s .02). It is interesting that one half of the depressed adolescents had two or more refined severe events occur during the year prior to onset compared with none of the normal controls (p :s; .01). Further analyses showed that depressed adolescents were significantly more likely to have a major difficulty precede the onset of the ir depression (27%) compared with normal controls (0%) (p :s; .04). Conclusions: The results suggest that depressed adolescents are exposed to high levels of stress prior to becoming depressed. Future investigations might benefit from using the LEOS with adolescents to assess acute and ongoing stressors. J. Am . Acad. Child Ado/esc. Psychiatry. 1998, 37( 10):1049-1 057 . Key Words: life events, adolescence, major depressive disorder.
Several studies of children (Banez and Compas, 1990; Mullins et al., 1985) and adolescents (Johnson and McCutcheon, 1980; Swearingen and Cohen, 1985) have examined the relation between life events and symptoms of depression. Overall, these studies have found life events to be positively correlated with symptoms of depression (average r ~ 0.30) , suggesting that life events may playa causal role in the development of depressive symptoms. However, the fact that life events have been found to be positively correlated with depressive symptoms does not definitively establish a causal relation between the two. Increased life events could very well be due to the presAcapttd M ay 5, 1998. From ~sttm Psychiatric lnstiture and Clinic, Piwburgh. This rtudy u/as fill/ded by a gram fr om the Research and Druelopmrnt Network On Psychopathology and Develop ment under the auspices of the John D. and Catherine T Ma c/irthur Foundation (to Dr. Birmaher). Correspondencr to M r. WilliamJOn, Department ofPsychiat ry. University of Pittsburgh School ofMedici ne. ~sttm Psychiatric Institu te and Clinic. 3811 O 'Hara St reet, Pittsburgh, PA 15213. 0890·85 67/98 /3 71O- 1049/$03 .00/0 © 1998 by the American Academy of Child and Adolescent Psychiatry.
J.
AM . ACA D . C H ILD ADOLESC. PSYCHI ATR Y. 37: 10. O CTO BE R 1998
en ce of depressive symptoms as much as the onset of depressive symptoms could be precipitated by the occurrence of life events. Similarl y, studies of clinic ally depressed youth have found life events to be elevated in depress ed children (Kashani et al., 1990) and adolescents (Daniels and Moos, 1990; Williamson er al., 1995a), with a few studies specifically showing life events to be significantly elevated prior to the onset of a depressive episode (Goodyer er al., 1985, 1988). Extending the findings from the co rrelatio nal studies, these studies suggest that life events play an important and possibly etiological role in the onset of depression in children and adolescents. However, because of methodological limitations, the role that life events play in the onset of depression among youth is not clear. One of the most significant limitations in life events studies of depression in children and adolescents reported to date has been their reliance on self-report life events checklists. Several limitations of life events checklists include the following: First, the checklist or respondentbased method of assessing life events does not ascertain
1049
WILLIAMSON ET AL.
the specific date that each life event occurred, making it difficult to assess the temporal relation between event occurrence and illness onset. Second, life event checklists have generally been found to have low test-retest reliability, with fall-off in reporting approaching 5% per month (Cohen and Wills, 1985) that has been attributed to respondents not being specifically prompted to remember events (Jenkins et al., 1979). Third, life events checklists do not assess whether an event is dependent on or independent of the individual's behavior. In the absence of the assessment of event dependence, events likely to be associated with the onset of depression cannot be specifically examined. Fourth, life events checklists do not distinguish between different severity levelswithin a particular stressor. For example, events such as birth of a sibling are considered a life event; however, in the absence of relevant contextual information (i.e., the meaning of the event for the individual given a particular biographical milieu), the severity of the event cannot be determined. Finally, acute stressors are often not differentiated from ongoing chronic stressors. Due in part to these limitations, the nature of the causal relation between life events and depression in children and adolescents is not well understood. The state-of-the-art method for assessing life events in adults is the investigator-based Bedford College Life Event and Difficulty Schedule (LEOS) (Brown and Harris, 1978. 1989). This life event interview collects detailed information about the event itself, the timing of its occurrence, and relevant contextual information for each event. Based on the contextual information, the threat for each event is rated by a panel of raters using standardized rating procedures. The LEOS provides a number of methodological advances over life events checklists that include (1) careful dating of event occurrence, (2) a scaling of the degree of stress ("threat") for an event, (3) the use of "event dictionaries" and consensus judgments to ensure reliability for including life occurrences as events and for accurately demarcating threat level, (4) ratings of eventbehavior and event-illness independence, (5) scoring of event dimensions (e.g., loss, danger), and (6) definition and inclusion of chronic stressors or difficulties. Several studies have supported the reliability (e.g., interrater) and validity (e.g., multiple informant) of the LEOS with adults exhibiting a variety of psychiatric symptoms (see Brown and Harris, 1978, 1989). The Bedford College LEOS methodology was originally adapted for use with adolescents by Monck and
1050
J.
Dobbs (1985). They found that the adult version of the LEOS was easily adaptable for use with adolescents and that adolescents were able to report reliably on the occurrence of severe and nonsevere events. In a study of 67 mother/daughter pairs from the community, adolescents aged 15 to 20 years were able to report on 93% of all severe events and 88% of all nonsevere events occurring during the prior year, suggesting that adolescents are quite reliable informants. On the basis of their work, Monck and Dobbs (1985) developed a LEOS manual with accompanying event dictionaries to be used for assessing stressors in adolescents. To date, the LEOS as developed for adolescents by Monck and Dobbs (1985) has not been used to assess stressors in an adolescent population in the United States. Furthermore, the LEOS has not been applied to a sample of depressed adolescents to determine whether the nature of life events and difficulties is the same for depressed adolescents as has been found in depressed adults. In this article we report on the application of the Bedford College LEOS to a sample of clinically depressed and normal control adolescents. To our knowledge, this report represents the first full application of the LEOS to a sample of depressed adolescents in the United States. Our main hypothesis, based on the studies of adults, was that depressed adolescents would have significantly more severe events and/or major difficulties in the year period prior to onset of their depression compared to a comparable period in normal control adolescents. Specific issues relevant to assessing life events and difficulties in adolescents living in the United States are discussed. METHOD Subjects Depressed adolescents aged 13 to 18 years were recruited through the outpatient Child and Adolescent Depression Program at the Western Psychiatric Institute and Clinic (WPIC), University of Pittsburgh, where they presented for treatment. Adolescents were evaluated using the Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present Episode version (K-SADSP) (Chambers et al., 1985), with both parenrts) and adolescent serving as informants about the adolescent's symptoms. For all cases, a child psychiatrist interviewed the adolescent, confirmed the presence of each of the criterion symptoms for depressive disorder with the adolescent and parenns), and reviewed all related diagnostic information. Adolescents were classified as depressed if they currently or recently (i.e., were remitted during the past month) had met DSM-I1I-R (American Psychiatric Association, 1987) criteria for a major depressive disorder (MOD). Depressed adolescents with any lifetime episodes of conduct disorder, attention deficit disorder, bipolar dis-
AM. ACAD. CHILD ADOLESC. PSYCHIATRY. 37:10, OCTOBER 1998
LIFE EV ENTS O F D EPR ESSED AD OLESC EN T S
order, su b sta n ce abuse/dependence, panic di sorder , ob sessivecompulsive disorder, o r psychosis were excluded fro m the study. C o nversely, depressed ad olescent s were not excluded if th ey had any lifetime ep isodes of an xiety di sorder not exclud ed ab ove. In addition, depressed adole scents were required to have had the ons et of their cu rrent o r most recent depressive ep isode in th e past 12 months, and at the time of interview, to ha ve had tw o co nsec utive weekly Beck Depression Inventory (Beck et al., 1961 ) sco res :s; 10. Depressed ado lescents wh ose depressive epi sode had per sisted for more th an 1 year were excl ude d. All de p ressed adolescen ts received o ngoi ng o u tpatient tre atment for th eir depression. Normal cont rol adolescents aged 13 to 18 years were recruited from existing research protocols be ing conducted at WPIC (Da h l et al., 1996; Ryan et al., 1994). Control adolescents were assessed using th e Sched u le for Affect ive D isorders and Sch izophrenia for Schoo l-Age C h ild ren-E pidem iologic version (K-SAD S-E) (O rvaschel et aI., 1982), with both parenti s) and adolescent serving as in fo rma nt s about th e ad o lesce nt 's sym pto ms. For th e cu rrent study, ad olescents were classified as no rm al controls o n ly if they had never m et DSM -III-R criteria for any psych iatric dis order. The demographic characteristics of the MOD and normal control ad olescents are pres ented in Table 1. The two groups were comparable with regard to age, gender, and race . Depressed ado lescents tended to belong to th e lower st rata of soci oeconomic stat us as measu red by Hollingshead 's Four Fac to r Ind ex o f Soc ial Posit ion (H o llingshea d, 1960) , although thi s finding was not stati stically significant. For adolescents w ith MOD , the average du rati on o f th eir depressive episode was 33.3 ± 18 .0 weeks (range 15-51 weeks). The severity of their depressive episode, as measured by a Hamilton scor e extracted from the K-SADS-P (Williamson et al., 1992) , was 19.0 ± 4 .8 during the worst per iod of th e depressive ep isod e.
Life Events Assessment The LEO S (Brow n an d H arris, 1978) was used to assess acut e and ongoing srresso rs in adolescents via a dire ct interview with the ad olescent. T he LEOS , designed for use with ad u lts, wa s o riginally adapted for use w ith ado lescents in a study of British fem ale adolescents (Mon ck an d Dobbs, 1985 ). More recently, as pan o f the co m prehensive work bein g conducted with the assessme nt of stressors in ad ults (Ellen Frank, Ph.D., prin cip al investigator), we have adap ted the LEO S fo r use in ado lescents living in the Un ited States (Boris Birmaher, M .D., pr in cipal inv est igator ) and have devel op ed anc ho ring exam ples for th e rat ing of events and d ifficulties described below based on the life experiences of U .S. adolescents. The LEOS is an in vestigator -based se m ist ru ctured int erview sched ule designed to elicit relevant cont extual in for m ation fo r events an d difficulties coverin g 10 domains: educat ion , work, rep roduction , money/possessio ns , hou sing , crime / legal, health , romant ic relatio nsh ips, other relatio nships , and m iscellaneous events (includ ing deaths). The LEOS int erviews were co nd ucted by interviewers with bachelor 's or ma ster 's degrees w ho had also received 3 m onths of LEO S tr aining prior to interviewing. The rating of events and diffi culties began with a ta pe-recorded int erview of an adolescent. After th e interview, th e inte rviewer listene d to the reco rding of the interview an d rated th e occur rence of events and d ifficul ties o n th e bas is of th e co n textua l information pro vided by the ado lescent . T he degree of stress for each event was rated on a 4- poi nt scale (I -m arked , 2-mode rate, 3so me, 4 -little or none) for both sho rt-term (pea k threat o f the event in th e 48 ho u rs followi ng th e event) and long-term threat (pea k threat in th e 10- to 14-day period after th e event oc curred). Diffi culties, lasting a min imum of 1 m onth (i.e ., 28 days), we re rated on
J.
AM . A CA D . C H ILD ADOLES C. PSYCHI AT RY. .~ 7 :
IO.
TABLE 1 Demograph ic C ha racte ristics of Depressed and Normal Cont rol Adolescents
Age G ender Fem ale/male Race White/nonwhite Soci oeconomic sta tu s 1 (low) 2 3 (m idd le) 4 5 (high) Durati on of d epression (wk) Extracted Hamilton sco re h Ep isodes Recurrent/first
Depressed (n = 26)
Normal Co nt rol (n = 15)
15.4 ± 1.4
15.8 ± 1.4
2 1/ 5
p
Sta tistic
= 0.95
.3 5
9/6
FET
. 14
22/4
15/0
FET
. 15
2 4 6 9 5
0 0 2
X2 = 2.56
t 39
d
.08
11 2
33. 3 ± 18.0 19.0 ± 4. 8 9/ 17
Note: Valu es represent m ean
± SO or number. FET = Fisher exact
test. a
Xl test of lin ear trend (Cochran, 1954 ).
H amilton sco re extracted from the Schedule for Affective Disorders and Schi zophren ia for School-Age C h ild ren- Present Ep isode (W illia mso n cr al.. 1992). b
a 7-po int scale ( I -high marked , 2-low marked, 3-high moderate, 4-low mode rat e, 5- m ild, 6- ver y m ild , 7-n ot/n o longer a d ifficul ty). The short- and long-te rm threat ratings for events and the threat ratings for diffi culties were made by com paring the contextual information for each event w ith a diction ary contain ing contextual examples based o n interviews w ith ad olescents and ad ults. For each event, the focu s of the event was determ ined to be specifically on the subject, on th e subject jo intl y w ith ano ther person, on ano ther per son, or involving a po ssession or a pet . After the int erviewe r rated each occ u rre nce (i.e., a chronological co m posi te o f events and incidents follow ed by ch ro n ic diffi culties), each LEOS chan was presented to a panel of con sen sus raters consisting of two or more raters blind to the ad olescent's di agnostic status. During the consen sus meeting, the interviewer p resented an overview of the ado lesce nt 's biographical circum sta nces that include d personal and fam ily demographics as wel l as wh o the adolescent's close relatives and co n fid ant s were (since events occurring to them ma y be eligible for inclusion depending o n th e nature of the even t). The agreem ent between all raters and fin al co nsensus ratings was high . Intraclass correl at ion coeffic ients ranged from 0.8 4 to 0.95 for all events. For all events, raters agreed exactl y more th an 90% of th e time (range 84 - 9 1%) and were within o ne level 98 % o f the time (range 98-100% ). For the m ost pan, we found the LEOS as origin ally adapted for use wi th adolescents by M on ck and Dobbs (l985 ) was sufficient for assessing srresso rs in our sam p le of adolescents. The sections o f the LEOS th at need ed to be refin ed we re those fo r wh ich th e cu ltu ral expe rience of U .S. adolescen ts differed from that of adolescents living in the United Kingdom. In general , education is d ifferent between U .S. and British ad olescents . For example, one difference was found
OCTO B ER 19 9 8
1051
WILLIAM SON ET AL.
in U.S. adolescents taking the Scholastic Aptitude Tests (SAr s). Taking the SATs is eligible to be included as an event . and the sho rtand long-term threat ratings will depend on the adolescent 's plan s and commitments for going to college and how poorly or well he/ she may have scored on previous SATs. While the rules for including school tests as events and for assessing their objective th reat are clearly outlined in the LEDS diction aries, the content pert ain ing to taking SAT tests is unique to U.S. adolescents. In add ition . we found it necessary to refine the interview schedule to better reflect the probes necessary for determining wh eth er events and diffi culties met the criteria outlined in the rating manuals. For example. in the romantic relationship domain. the original version of the interview schedule did not probe for boyfriend/girlfriend relationships. In the revised version of the interview schedule. detailed information about boyfriends/ girlfriends is probed for. including specific information about the start of a relationship. which includes (1) whether they go out together. (2) whether the adolescent is considered by the other person as his/ her girlfriend/boyfriend. and (3) whether others see them as a couple. T hese revisions have produced an int erview schedule that is more tightl y linked with the manual and the diction aries of events and difficulties, mak ing the determ ination of what should be included as an event or difficulty much more straightforward. Toward th is end. we have comprehensively adapted the LEDS for use in U.S. adolescents by providing broader-based prim ary probes for specific context and have added U.S. examples of events and difficulties whose context is specific to U.S. adolescents. In general. we were quite impressed with the conceptual flexibility of the LEDS methods and the relative ease with which we were able ro use the LEDS with our sample of adolescents living in the Un ited States. Events and difficult ies were categorized on the basis of Brown and Harris' work examin ing the relation berween life events and dep ression in adult s (Brown and Harris. 1978 , 1989). Event s were classified as severe events if they were (I) rated a l-rnarked on long-term th reat or (2) rated a 2-moderate on long-term threat and focused on the adolescent him/herself or jointly with someone else. Difficulti es were classifiedas major difficulties if they were (1) rated on the top 3 points of severity (l-high marked . 2-low marked, 3-high moderate); (2) had lasted at least 2 years; and (3) did not involve purely health probl ems.
survival analysis based on the Kaplan-Me ier product limit estimate was used to examine the temporal pattern of severe events occurring nearest in time to th e onset of depression or the linked period. Accordingl y. the on set of depre ssion or the start of the linked period was set at 0 and the occu rrence of events prior to these periods was coded in absolute times. Thus, an event occurring 1 month prior to onset would cause the subject to be coded as "dead" and his survival time to be coded as 1. For those adolescents who did not have a severe event in the 12-m onth period . they were coded as "alive" and their survival time was coded as 12. All significance tests were two -tailed with U2 ; 0.05 .
RESULTS
Before comparing acute and ongoing stressors in depressed and normal control adolescents, we examined the relation between age, sex, race, and socioeconomic status and stressors. None of the demographic characteristics were significantly related (Q any of the measures of stress. Presence of Severe Events and Major Difficulties Prior to Onset
Severe events have been shown to be more likely to be present in the year prior to onset in depressed adults (Brown and Harris, 1978 , 1989). To exam ine whether the same is true for depressed adolescents, we compared depressed and normal control adolescents on the presence of at least one traditionally defined severe event in the year prior to onset. More than twice as many depressed adolescents reported having had a traditionally defined severe event occur in the year prior to onset (46%) compared with normal control adolescents (20%), although this difference failed to reach statistical significance (X2 = 2.22, P < .14). Survival analyses similarly showed a trend for greater events in the depre ssed adolescents (generalized Wilcoxon (X 2 = 2.47 , P < .12). More so than adults, adolescents are int imately dependent on and attached (Q others in their environment including parents, grandparents, other close relatives, and friends. To determine whether events focused solely on these significant others in the adolescent's life (as opposed to jointly on the adolescent and the significant other) were associated with depression, we expanded the traditional definition of severe events to also include those events that were focused on others and rated as 2-moderate on long-term threat. These refined "severe" event s were significantly more prevalent among the depressed adolescents prior to onset (62%) than among normal control adolescents (27%) (X 2 = 5.23, P < .02). Survival analyses
Statistical Methods Demographic variables were compared using X2 with Yates continu ity cor rection or t tests as applicable. The relations hip between demograph ic variables and life events measures was assessed using eith er Spearman correlations or Mann -Whitney rank sum tests. For all analyses. the occurrence of life events was examined in the 12month period prior to the onset of the depressiveepisode in depressed adolescents and dur ing a comparable "linked" period in normal control adolescents . The linked period in the normal control adolescents represented the 12-month period co rrespondi ng to th e I2-month per iod that prec eded th e average on set of the dep ressive episode among the depressed adolescents. Because the average onset of illness was approximately 8 months (see Table 1), the linked period for each normal control adolescent represent ed a period covering 8 to 20 months prior to the date of interview.The advantage of using a linked period in normal cont rol adolescents is that the average length of recall for events and d ifficulties was the same for depressed and norm al control adolescents. Therefore. it is unl ikely that the rates of events and difficulties were artificially inflated in the normal control group. Analyses of life events and difficult ies proceeded in rwo steps. First. the presence of severe events and major difficulties was comp ared in MDD and normal control adolescents using X2 analyses. Second.
1052
J. AM .
AC A D. C H I L D ADO LES C. PSYC H I AT RY, 3 7: 10, OCTO B E R 19 9 8
LIF E EVENTS O F D EPR ESSED AD O LES CE N TS
similarly found the depressed adolescents to have significantly more refined severe events (generalized Wilcoxon X2 = 5.67, P < .02). The backward survival plot depi cting the survival fun ction for depres sed and normal control adolescent s is presented in Figure 1. Depressed adolescents had significantly more major d ifficulties compared with normal control adolescents (27% versus 0%, Fisher exact test, p < .04). Considering the refined severe events and major difficulties together, significantly more depressed adolescents had at least one in the year prior to ons et (69 % versus 27%, Yates X2 = 5.32, P < .02). It is interesting that two depressed adolescents (7% ) had a major difficulty but did not have a refined severe event prior to on set. Total Number of Refined Severe and Nonsevere Events
Compared with normal control adolescents , depressed adolescent s had a significantl y highe r average number of refined severe events (2.0 ± 2.1 versus 0.3 ± 0.5 , p < .0 1) and a trend for more nonsevere events (8.0 ± 5.7 versus 4.6 ± 3.3, p < .07) in th e year prior to onset. It is interesting that compared with normal control adolescen ts, a significantly higher percentage of depres sed adolescent s had two or more refined severe events (50% versus 0% , Fisher exact test , p < .00l) and three or more refined
o.
severe events (35 % versus 0% , Fisher exact test , p < .02) in the year prior to on set. Timing of Refined Severe and Nonsevere Events During the Year Prior to Onset
Original stud ies of life event s and difficulties in depressed adults showed that only severe events di stinguished between depressed and community control adults and did so for the entire year period prior to onset (Brown et al., 1973). Whether adolescents similarly experience higher rates of severe events for the entire year prior to onset or have higher levels of events across all threat levels remains an empirical question that has yet to be exam ined. Therefore, we examined the frequen cy of refin ed severe and nonsevere even ts among depressed and normal control adolescent s in the twelve I -rnonth intervals prior to on set. As shown in Figure 2, refined severe events were elevated in the depres sed adol escents across the twelve l -rnonth periods prior to on set compared with normal cont rols, although only the 7- and 10month comparisons reached statistical significance (p < .0 5). Similarly, nonsevere events were also elevat ed throughout the twelve l-monrh periods pr ior to on set; however, only the 7- and 1O-month comparisons reached statistical significance.
»,
--
.....
"E 0.8
~"S
.
.
....
0.7
~ 0.6
.€8
8. 0.5
~ o,
0.4
Q)
~
"5
0.3
(J
0.2
E ::)
---.. .
Depressed (n=26) Normals(n=15)
0.1
o
2
3
62% of the Depressedhad a ' Severe" Event 27% of the Normal Controlshad a "Severe"Event Generalized Wilcoxon= 5.67, df=1, P < .02
456 789 Months Priorto Onset or "Linked"Year
10
11
12
Fig. 1 Backward survival analysis for refined "severe" events during the year prior to o nset in th e depressed and the corre spondi ng "linked" year in the normal contro l adol escents.
J.
AM . ACA D. C H ILD ADOL ESC . PSYC H IATR Y• .n :lo . O CTOB ER 199 8
1053
WILLIAMSON ET AL.
-
MDOsI'l=26
CJ
NCs11=15
12
11
10 9 8 7 6 5 4 3 Month Prior to Onset or "Linked" Year
Fig. 2 Frequency of severe events during the 12 months prior disorder; NC s = normal control subjects.
1
onset . MDDs = subjects with major depressive
lar immune function, EEG sleep, nocturnal growth hormone secretion, and exp ression of depression in depressed children and adolescents (Birmaher et al., 1994; Williamson et al., 1995a,b, 1996). In all of these studies, we relied on a life events checklist, primarily Coddington's Life Event Record (Coddington, 1972), to assess stressors. As noted above, there are a number of limitations to assessing stressors using life events checklists, and these compromised our ability to investigate fully the relationship between stress and various neu roendocrine systems in depressed children and adolescents. For example, in our EEG sleep study (Williamson et al., 1995b), the finding that stressors decreased rapid eye movement sleep latency in normal controls but not depressed adolescents might very well have been a result of differences between the two groups in when the stressors occurred in relation to the sleep study. Because the Life Event Record was used, events were indicated only as occurring in the year prior to the EEG sleep studies and were not specificallydated. To the best of our knowledge, the current report represents the first full application of the Bedford College LEOS method for assessing acute and chronic stressors among depressed adolescents. As hypothesized, depressed adolescentsexperiencedsignificantlymore stressors during
Comparison of First Episode and Recurrent Depression
Of the 26 depressed adolescents , 9 had at least one prior episode of depression. Among these adolescents with recurrent depression, none had a prior episode occur during the I-year period prior to the onset of their most recent episode of depression. Depressed adolescents who had their first episode of depression reported experiencing a similar rate of refined severe events prior to onset (65%) compared with adolescents with recurrent depression (56%) (generalized Wilcoxon X2 = 0.62, P < .43). In contrast , significantly more depressed adolescents with a recurrent depressive episode reported a major difficulty (67% versus 6%, Fisher exact test, p ~ .0 1). Considering refined severe events and major difficulties together, adolescents with recurrent depression had similar rates of at least one acute or chronic stressor prior to the onset of their most recent depressive episode (78% versus 65%, Fisher exact test, not significant). DISCUSSION
Our interest in using the LEOS, an investigator-based interview schedule, to assess stressors in adolescents evolved out of our previous studies in which we examined the relations between stressful life events and cellu-
1054
to
2
J.
AM . AC AD. C HI LD A DOL ESC . PSYCHIATRY, 37 : 10 , OCTO BE R 19 9 8
LIFE EVENTS OF DEPRESSED ADOLESCENTS
the year prior to onset when compared with a comparable 12-month period in normal controls. A significantly greater number of depressed adolescents experienced a traditionally defined severe event (46% versus 20%, not significant), those rated I-marked or 2-moderate longterm threat and focused on the adolescent or jointly with another person, in the year prior to onset. However, probably because of the small sample size, this increase did not reach statistical significance (the power to detect a between-group difference was 0.41). Although severe events occurred across each of the 10 domains, typically they were centered around the adolescent's romantic relationships. Following is an example of one such severe event that occurred in the year prior to onset in a depressed adolescent female:
A I6-year-old female had had a confiding and sexual relationship with her boyfriendfor 72 months. Their relationship had a history ofongoing physical violence. While at a party, the subject's boyfriend hit her and she immediatelybroke offthe relationship. During the next 2 weeks he telephoned herseveral times, but she did not return hiscalls. Additional severe events in this sample of adolescents included the following: close friend moving away, first sexual intercourse, becoming pregnant, having an abortion, friend being raped, sister returning home to live, being sexually harassed by a stranger, argument and breakdown in relationship with mother, and a fight at school. Expanding severe events to include those events rated as 2-moderate on long-term threat and focused on others (parents, siblings, grandparents, friends, etc.) resulted in a significantly higher percentage of depressed adolescents having a refined severe event prior to onset (62% versus 27%, p < .02). Below is an example of an "otherfocused" event that was rated 2-moderate on long-term threat and reclassified by us as a refined severe event:
A I4-year-old malereported havinga very close relationship with his70-year-oldgrandfather; they livedon thesame street and saweach other at least every other day. Thegrandfather died ofkidneyfailure after having been hospitalized for 2 weeks; hehad been scheduled to undergo kidney dialysis. In this sample of adolescents, additional events that were other-focused and classified as refined severe events included the following: mother having surgery for breast cancer, father being involved in a car accident and taken by ambulance to the hospital, death of an uncle, exboyfriend shooting another person, and classmate being killed in an automobile accident.
J.
The finding that events rated 2-moderate on long-term threat and focused on others appear related to depression onset in adolescents differs from Brown's findings in depressed adults. In his original Camberwell sample of women (Brown and Harris, 1978), only events rated as I-rnarked or 2-moderate and focused on the woman herself or jointly with someone else were related to the presence of depression. From findings of the current study it appears that the type of events likely to precede the onset of a depressive episode during adolescence differs from the type of events found to precede the onset of depression in adults. In adolescents, stressful events occurring to primary attachment figures such as parents and grandparents seem to be centrally involved in the onset of depression in this age group. Future studies will need to confirm the importance of these other-focused events in the development of depression in adolescents. On the basis of their study of depression among women living in Camberwell, Brown and Harris (1978) defined provoking agents for depression as either a severe event or a major difficulty. Major difficulties are ongoing stressors lasting a minimum of 2 years, rated on the top 3 points of severity, and not purely involving health problems. Studies of depressed adults have shown that severe events account for most of the variance; however, major difficulties significantly add to the variance for depression (Brown and Harris, 1978). In the current study, 27% of the depressed adolescents had a major difficulty compared with none of the normal control adolescents. Following is an example of an ongoing stressor that was classified as a major difficulty:
A I5-year-old white female reported that she and her father did not get along. The subject and herfather fought constantly about hergrades, herfriends, and her behavior. She said that he did not showany interest in her life he rarely went to teacher conferences). He told her that he wouldnot besurprised ifshe were a tramp. The[ather often threatened topunch her and had attackedher in the past. He had slapped her in the face and about the head, had thrown a remote control device and typewriter at her, and had punched her in her back. These physical altercations resulted in the subject's sustaining bodily bruises and welts.
AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 37:10, OCTOBER 1998
««.
Additional major difficulties in this sample of adolescents included mother's health, brother's health, tension in relationship between parents, and relationship with parents. Sixty-nine percent of the depressed adolescents in the current study had a severe acute or chronic stressor, with
1055
W[LLIAMSON ET AL.
62% having a refined severe event and 7% having only a major difficulry. The finding that refined severe events account for most of the between-group variance is in agreement with studies of depressed adults (Brown and Harris, 1978) . In their study of Camberwell women, Brown and Harris found that 61 % of depressed patients had a provoking agent, with 49 % having a severe event and 12% having only a major difficulry. The results from this study suggest that the onset of depression in adolescents is similarly preceded by acute stressful events, although ongoing stressorsappear to contribute to the onset of depression in at least some adolescents. In the future, it will be interesting to examine the relation between refined severe events and major difficulties and their combined importance for the development of depression in adolescents. In this investigation, depressed adolescents were significantly more likely to have had two or more refined severe events prior to becoming depressed (50 versus 0%) compared with normal controls. Considered together with the high rate of nonsevere srressors experienced by both depressed and normal control adolescents, the se data suggest that depressed adolescents are experiencing higher levelsof stressors compared with depre ssed adults . Future investigations might fruitfully explore the source of these refined severe events to determine whether they evolve from similar or separate domains. Clinical Implications
The main clinical implication of this study is that adolescents who experience high levels of severe stressful events might be at increased risk for the development of depression. As reponed herein, 50% of the depressed adolescents reponed experiencing two or more severe events prior to onset compared with none of the normal controls! Therefore, clinical efforts focused on providing clinical intervention soon after the occurrence of the first severeevent might help to prevent the onset of depression in adolescents. Future studies should include prospective assessment of life events in nondepressed adolescents, some of whom will subsequently become depressed . In summary, this study represents one of the first applications of the LEOS to a sample of adolescents in the United States. The finding that LEOS-defined stressors are elevated in depressed adolescents prior to onset parallels the work of Brown and Harris (1978, 1989) in depressed adults . Future studies of adolescents are needed
1056
which use an investigator-based measure of acute and chronic stress, such as the LEOS , that also include other potential risk factors such as parental psychopathology and specifically examine the role of stressorsin the onset, maintenance, and recurrence of depression in this age group.
REFERENCES American Psychiatric Association ( 1987). Diagnostic and Statistical Manual of Mental Disorders, 3rd edition-reuised (DSM -III-R). Washington. DC: American Psychiatric Association Banez GA. Compas BE (1990 ), Children's and parent's daily stressful events and psychological symptoms. ] Abnorm Child Psycho! 18:591- 605 Beck AT. Ward C H. Mendel son M (196 1). An invento ry for measur ing depression . Arch Gm Psychiatry 4:53- 63 Birmaher B. Rabin BS. Garcia MR er al. (1994). Cellular im munity in depressed. cond uct disord er. and norm al adolescents: role of adverse life events. ] Am Acad Child Adolesc Psychiatry 33:671- 678 Brown GW. Harris TO. eds (1978), Socia! Origins of Depression: A Study of Psychiatric Disorder in Womrn. Lond on : Tavisrock Brown GW, H arr is TO. eds (1989). Lift Events and Illness. New York: G uilfo rd Brown GW. Sklair F. Harris T O . Birley JLT (1973) . Life-events and psychiatric disorders part 1: some methodological issues. Psycho! Med 3:74- 87 Cha mbers WJ. Puig-Ant ich J. Hirsch M et al. (1985) . T he assessment of affective d isorders in child ren and adolescents by semi-str uctu red interview: test-retest reliability of the Sched ule for Affective Disord ers and Schizophren ia for School-Age Children. Present Episode version . Arch
Gen Psychiatry 42:696- 702 Coc hran WG (1954). Some methods for strengthening the common X' test. Biometrics 10:417-44 I Co ddington RD (1972), The significance of life events as etiologic factors in diseasesof children. I: a survey of professionals.] Psychosom Res [6:205-2 13 Co hen S. W ills TA (1985), Stress. social suppo rt and the buffering hypothesis. Psy cho! BuI/9 8:31O-357 Dahl RE. Ryan N D . Matty MK et al. (1996 ). Sleep onset abnormalities in depressed ado lescents. Biol Psychiatry 39:400-410 Daniels D. Moos RH (1990), Assessinglife stressorsand social resources among adolescents: applications to depressed youth.] Adolesc Res 5:268-289 Goo dyer I. Kolvin I, Garzani s S ( 1985). Recent unde sirable life events and psychiatric disorder in childhood and adolescence. Br ] Psychiatry 147: 5 17- 523 Goo dyer [M . Wright C, Altharn PME ( 1988). Maternal adversity and recent stressful life events in anxious and depr essed children. ] Child Psycho! Psychiatry 29:6 5 1-667 Hollingshead AB (1960). Four-Factor Index ofSocia!Position. New Haven. CT: Yale Un iversity Dep artment of Sociology Jenki ns CD. Hu rst MW. Rose RM (1979). Life chang es: do peopl e really remember? Arch Gen Psychiatry 36:379-384 John son JH . McCutcheon SM (1980 ). Assessing life stress in older child ren and adolescent s: prelim inary finding s with the Life Events C hecklist. In : Stress and Anxiety. Sarason [G . Spielberger CD. eds. Washington. DC: Hemisphere Kashan i JH . Vaidya AF.Soltys SM. Dandoy AC, Reid JC (1990). Life events and major depression in a sample of inpatient childr en. ComprPsychiatry 31:266 -274 Monck E. Dobb s R (1985 ). Measuring life events in an adolescent popul ation : method ological issues and related finding s. Psycho! Med 15:841-850 Mullins LL. Siegel LJ. Hodges K (198 5). Co gnitive problem -solving and life event correlates of depre ssive symptoms in ch ildren. ] Abnorm Child Psycho! 13:30 5-314 O rvaschel H . Puig-Am ich J. C hambers WJ. Tabrizi MA. Johns on R ( 1982). Retrospect ive assessment of child psychopathology with the K-SADS-E. ] Am Acad Child Psychiatry 2 I :392-397
J. AM . ACA D . C H I1.D AD01. ESC. " SYCHI ATRY .
3 7 : 10 . OC T O BE R 199 8
LIFE EVENTS OF DEPRESSED ADOLESCENTS
Ryan NO, Dahl RE, Birrnaher B er al. (1994), Stimulatory tests of growth hormone secretion in prepubertal major depression: depressed versus normal children. JAm Acad ChildAdolesc Psychiatry 33:824-833 Swearingen EM, Cohen LH (1985), Life events and psychologicaldistress: a prospective study of young adolescents. DeuPsychol21: 1045-1054 Williamson DE, Birmaher B, Anderson BP, Al-Shabbour M, Ryan NO (1995a), Stressfullife events in depressed adolescents: the role of dependent events during the depressiveepisode. JAm AcadChildAdolesc Psychi-
atry 34:591-598
Williamson DE, Birmaher B, Dahl RE, Al-Shabbour M, Ryan NO (1996), Stressful life events influence nocturnal growth hormone secretion in depressed children. BioiPsychiatry 40: 1176-1180 Williamson DE, Dahl RE, Birmaher B, Goetz RR, Ryan NO (1995bl, Stressfullife events and EEG sleep in depressed and normal control adolescents. BioiPsychiatry 37:859-865 Williamson DE, Ryan NO, Dahl RE, Jeanette L (1992), Hamilton depression scorescan be extracted from the K-SADS-P in adolescents.J ChildAdolesc
Psychopharmacol2: 175-181
The Changing Pattern of Substance Abuse in Urban Adolescents. Cynthia Brasseux, MA, Lawrence J. D'Angelo, MD, MPH, Mark Guagliardo, PhD, Jocelyn Hicks, PhD
Objectives: To determine the prevalence of specific drug use in adolescents attending an adolescent health clinic and to compare current rates with a similar previous study. Design: Blinded and anonymous urine samples obtained from patients presenting for routine health care were tested for the presence of cannabinoids, phencyclidine (PCPl, amphetamines, opiates, and cocaine. Setting: Adolescent medicine outpatient clinic. Patients: Patients were between 12 and 21 years of age. Specimens from 1313 patients in 1995 to 1996 and 1312 patients in 1989 to 1990 were tested. Main Outcome Measures: Current drug use rates were compared with a similar screening of patients conducted in 1989 to 1990. Comparisons between studies were made on the basis of specific drug, age, and SeX. Results: For the most recent patient group, 14% Were positive for 1 or more drugs and 13% were positive for cannabinoids. Males were significantly more likely to test positive for drug use than females. The oldest adolescents Were more likely to test positive for drug use than younger adolescents. Comparing the 2 study year cohorts, patients tested recently were significantly more likely to have urine tests positive for at least 1 drug and cannabinoids in particular and less likely to have urine tests positive for cocaine. Conclusions: There has been an increase in positive urine tests in patients seen in our ambulatory clinic, with a strong shift toward cannabinoids and a shift away from cocaine. Practitioners need to be aware that drug use patterns in adolescents can shift relatively abruptly and counseling should be targeted to current drug USe patterns. Arch Pediatr Adolesc Med 1998;152:234-237 Abstracts selected by Michael]. Maloney, M.D., Assistant Editor.
j. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 37:10, OCTOBER 1998
1057
ABSTRACT Objective: To examine the significance of acute life events and ongoing difficulties in adolescents with a recent major depressive disorder. Method: Adolescents (aged 13-18 years) with a recent episode of major depress ive disorder based on DSM-If/·R (n
=26) and normal controls free of any Axis I lifetime psychiatric disorder (n =15) were assessed using the
investigator·based Life Events and Difficulties Schedule (LEOS). Results: Traditionally defined severe events were more likely to occur in the year prior to onset among depressed adolescents (46%) than in a comparable period among normal controls (20%), but these differences did not reach statistical significance . Expanding the definition of severe events to include those events focused on others impo rtant to the adolescent resulted in a significantly higher percentage of depressed adolescents having one or more refined "severe" events in the year prior to onset (62%) compared with normal controls (27%) (p s .02). It is interesting that one half of the depressed adolescents had two or more refined severe events occur during the year prior to onset compared with none of the normal controls (p :s; .01). Further analyses showed that depressed adolescents were significantly more likely to have a major difficulty precede the onset of the ir depression (27%) compared with normal controls (0%) (p :s; .04). Conclusions: The results suggest that depressed adolescents are exposed to high levels of stress prior to becoming depressed. Future investigations might benefit from using the LEOS with adolescents to assess acute and ongoing stressors. J. Am . Acad. Child Ado/esc. Psychiatry. 1998, 37( 10):1049-1 057 . Key Words: life events, adolescence, major depressive disorder.
Several studies of children (Banez and Compas, 1990; Mullins et al., 1985) and adolescents (Johnson and McCutcheon, 1980; Swearingen and Cohen, 1985) have examined the relation between life events and symptoms of depression. Overall, these studies have found life events to be positively correlated with symptoms of depression (average r ~ 0.30) , suggesting that life events may playa causal role in the development of depressive symptoms. However, the fact that life events have been found to be positively correlated with depressive symptoms does not definitively establish a causal relation between the two. Increased life events could very well be due to the presAcapttd M ay 5, 1998. From ~sttm Psychiatric lnstiture and Clinic, Piwburgh. This rtudy u/as fill/ded by a gram fr om the Research and Druelopmrnt Network On Psychopathology and Develop ment under the auspices of the John D. and Catherine T Ma c/irthur Foundation (to Dr. Birmaher). Correspondencr to M r. WilliamJOn, Department ofPsychiat ry. University of Pittsburgh School ofMedici ne. ~sttm Psychiatric Institu te and Clinic. 3811 O 'Hara St reet, Pittsburgh, PA 15213. 0890·85 67/98 /3 71O- 1049/$03 .00/0 © 1998 by the American Academy of Child and Adolescent Psychiatry.
J.
AM . ACA D . C H ILD ADOLESC. PSYCHI ATR Y. 37: 10. O CTO BE R 1998
en ce of depressive symptoms as much as the onset of depressive symptoms could be precipitated by the occurrence of life events. Similarl y, studies of clinic ally depressed youth have found life events to be elevated in depress ed children (Kashani et al., 1990) and adolescents (Daniels and Moos, 1990; Williamson er al., 1995a), with a few studies specifically showing life events to be significantly elevated prior to the onset of a depressive episode (Goodyer er al., 1985, 1988). Extending the findings from the co rrelatio nal studies, these studies suggest that life events play an important and possibly etiological role in the onset of depression in children and adolescents. However, because of methodological limitations, the role that life events play in the onset of depression among youth is not clear. One of the most significant limitations in life events studies of depression in children and adolescents reported to date has been their reliance on self-report life events checklists. Several limitations of life events checklists include the following: First, the checklist or respondentbased method of assessing life events does not ascertain
1049
WILLIAMSON ET AL.
the specific date that each life event occurred, making it difficult to assess the temporal relation between event occurrence and illness onset. Second, life event checklists have generally been found to have low test-retest reliability, with fall-off in reporting approaching 5% per month (Cohen and Wills, 1985) that has been attributed to respondents not being specifically prompted to remember events (Jenkins et al., 1979). Third, life events checklists do not assess whether an event is dependent on or independent of the individual's behavior. In the absence of the assessment of event dependence, events likely to be associated with the onset of depression cannot be specifically examined. Fourth, life events checklists do not distinguish between different severity levelswithin a particular stressor. For example, events such as birth of a sibling are considered a life event; however, in the absence of relevant contextual information (i.e., the meaning of the event for the individual given a particular biographical milieu), the severity of the event cannot be determined. Finally, acute stressors are often not differentiated from ongoing chronic stressors. Due in part to these limitations, the nature of the causal relation between life events and depression in children and adolescents is not well understood. The state-of-the-art method for assessing life events in adults is the investigator-based Bedford College Life Event and Difficulty Schedule (LEOS) (Brown and Harris, 1978. 1989). This life event interview collects detailed information about the event itself, the timing of its occurrence, and relevant contextual information for each event. Based on the contextual information, the threat for each event is rated by a panel of raters using standardized rating procedures. The LEOS provides a number of methodological advances over life events checklists that include (1) careful dating of event occurrence, (2) a scaling of the degree of stress ("threat") for an event, (3) the use of "event dictionaries" and consensus judgments to ensure reliability for including life occurrences as events and for accurately demarcating threat level, (4) ratings of eventbehavior and event-illness independence, (5) scoring of event dimensions (e.g., loss, danger), and (6) definition and inclusion of chronic stressors or difficulties. Several studies have supported the reliability (e.g., interrater) and validity (e.g., multiple informant) of the LEOS with adults exhibiting a variety of psychiatric symptoms (see Brown and Harris, 1978, 1989). The Bedford College LEOS methodology was originally adapted for use with adolescents by Monck and
1050
J.
Dobbs (1985). They found that the adult version of the LEOS was easily adaptable for use with adolescents and that adolescents were able to report reliably on the occurrence of severe and nonsevere events. In a study of 67 mother/daughter pairs from the community, adolescents aged 15 to 20 years were able to report on 93% of all severe events and 88% of all nonsevere events occurring during the prior year, suggesting that adolescents are quite reliable informants. On the basis of their work, Monck and Dobbs (1985) developed a LEOS manual with accompanying event dictionaries to be used for assessing stressors in adolescents. To date, the LEOS as developed for adolescents by Monck and Dobbs (1985) has not been used to assess stressors in an adolescent population in the United States. Furthermore, the LEOS has not been applied to a sample of depressed adolescents to determine whether the nature of life events and difficulties is the same for depressed adolescents as has been found in depressed adults. In this article we report on the application of the Bedford College LEOS to a sample of clinically depressed and normal control adolescents. To our knowledge, this report represents the first full application of the LEOS to a sample of depressed adolescents in the United States. Our main hypothesis, based on the studies of adults, was that depressed adolescents would have significantly more severe events and/or major difficulties in the year period prior to onset of their depression compared to a comparable period in normal control adolescents. Specific issues relevant to assessing life events and difficulties in adolescents living in the United States are discussed. METHOD Subjects Depressed adolescents aged 13 to 18 years were recruited through the outpatient Child and Adolescent Depression Program at the Western Psychiatric Institute and Clinic (WPIC), University of Pittsburgh, where they presented for treatment. Adolescents were evaluated using the Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present Episode version (K-SADSP) (Chambers et al., 1985), with both parenrts) and adolescent serving as informants about the adolescent's symptoms. For all cases, a child psychiatrist interviewed the adolescent, confirmed the presence of each of the criterion symptoms for depressive disorder with the adolescent and parenns), and reviewed all related diagnostic information. Adolescents were classified as depressed if they currently or recently (i.e., were remitted during the past month) had met DSM-I1I-R (American Psychiatric Association, 1987) criteria for a major depressive disorder (MOD). Depressed adolescents with any lifetime episodes of conduct disorder, attention deficit disorder, bipolar dis-
AM. ACAD. CHILD ADOLESC. PSYCHIATRY. 37:10, OCTOBER 1998
LIFE EV ENTS O F D EPR ESSED AD OLESC EN T S
order, su b sta n ce abuse/dependence, panic di sorder , ob sessivecompulsive disorder, o r psychosis were excluded fro m the study. C o nversely, depressed ad olescent s were not excluded if th ey had any lifetime ep isodes of an xiety di sorder not exclud ed ab ove. In addition, depressed adole scents were required to have had the ons et of their cu rrent o r most recent depressive ep isode in th e past 12 months, and at the time of interview, to ha ve had tw o co nsec utive weekly Beck Depression Inventory (Beck et al., 1961 ) sco res :s; 10. Depressed ado lescents wh ose depressive epi sode had per sisted for more th an 1 year were excl ude d. All de p ressed adolescen ts received o ngoi ng o u tpatient tre atment for th eir depression. Normal cont rol adolescents aged 13 to 18 years were recruited from existing research protocols be ing conducted at WPIC (Da h l et al., 1996; Ryan et al., 1994). Control adolescents were assessed using th e Sched u le for Affect ive D isorders and Sch izophrenia for Schoo l-Age C h ild ren-E pidem iologic version (K-SAD S-E) (O rvaschel et aI., 1982), with both parenti s) and adolescent serving as in fo rma nt s about th e ad o lesce nt 's sym pto ms. For th e cu rrent study, ad olescents were classified as no rm al controls o n ly if they had never m et DSM -III-R criteria for any psych iatric dis order. The demographic characteristics of the MOD and normal control ad olescents are pres ented in Table 1. The two groups were comparable with regard to age, gender, and race . Depressed ado lescents tended to belong to th e lower st rata of soci oeconomic stat us as measu red by Hollingshead 's Four Fac to r Ind ex o f Soc ial Posit ion (H o llingshea d, 1960) , although thi s finding was not stati stically significant. For adolescents w ith MOD , the average du rati on o f th eir depressive episode was 33.3 ± 18 .0 weeks (range 15-51 weeks). The severity of their depressive episode, as measured by a Hamilton scor e extracted from the K-SADS-P (Williamson et al., 1992) , was 19.0 ± 4 .8 during the worst per iod of th e depressive ep isod e.
Life Events Assessment The LEO S (Brow n an d H arris, 1978) was used to assess acut e and ongoing srresso rs in adolescents via a dire ct interview with the ad olescent. T he LEOS , designed for use with ad u lts, wa s o riginally adapted for use w ith ado lescents in a study of British fem ale adolescents (Mon ck an d Dobbs, 1985 ). More recently, as pan o f the co m prehensive work bein g conducted with the assessme nt of stressors in ad ults (Ellen Frank, Ph.D., prin cip al investigator), we have adap ted the LEO S fo r use in ado lescents living in the Un ited States (Boris Birmaher, M .D., pr in cipal inv est igator ) and have devel op ed anc ho ring exam ples for th e rat ing of events and d ifficulties described below based on the life experiences of U .S. adolescents. The LEOS is an in vestigator -based se m ist ru ctured int erview sched ule designed to elicit relevant cont extual in for m ation fo r events an d difficulties coverin g 10 domains: educat ion , work, rep roduction , money/possessio ns , hou sing , crime / legal, health , romant ic relatio nsh ips, other relatio nships , and m iscellaneous events (includ ing deaths). The LEOS int erviews were co nd ucted by interviewers with bachelor 's or ma ster 's degrees w ho had also received 3 m onths of LEO S tr aining prior to interviewing. The rating of events and diffi culties began with a ta pe-recorded int erview of an adolescent. After th e interview, th e inte rviewer listene d to the reco rding of the interview an d rated th e occur rence of events and d ifficul ties o n th e bas is of th e co n textua l information pro vided by the ado lescent . T he degree of stress for each event was rated on a 4- poi nt scale (I -m arked , 2-mode rate, 3so me, 4 -little or none) for both sho rt-term (pea k threat o f the event in th e 48 ho u rs followi ng th e event) and long-term threat (pea k threat in th e 10- to 14-day period after th e event oc curred). Diffi culties, lasting a min imum of 1 m onth (i.e ., 28 days), we re rated on
J.
AM . A CA D . C H ILD ADOLES C. PSYCHI AT RY. .~ 7 :
IO.
TABLE 1 Demograph ic C ha racte ristics of Depressed and Normal Cont rol Adolescents
Age G ender Fem ale/male Race White/nonwhite Soci oeconomic sta tu s 1 (low) 2 3 (m idd le) 4 5 (high) Durati on of d epression (wk) Extracted Hamilton sco re h Ep isodes Recurrent/first
Depressed (n = 26)
Normal Co nt rol (n = 15)
15.4 ± 1.4
15.8 ± 1.4
2 1/ 5
p
Sta tistic
= 0.95
.3 5
9/6
FET
. 14
22/4
15/0
FET
. 15
2 4 6 9 5
0 0 2
X2 = 2.56
t 39
d
.08
11 2
33. 3 ± 18.0 19.0 ± 4. 8 9/ 17
Note: Valu es represent m ean
± SO or number. FET = Fisher exact
test. a
Xl test of lin ear trend (Cochran, 1954 ).
H amilton sco re extracted from the Schedule for Affective Disorders and Schi zophren ia for School-Age C h ild ren- Present Ep isode (W illia mso n cr al.. 1992). b
a 7-po int scale ( I -high marked , 2-low marked, 3-high moderate, 4-low mode rat e, 5- m ild, 6- ver y m ild , 7-n ot/n o longer a d ifficul ty). The short- and long-te rm threat ratings for events and the threat ratings for diffi culties were made by com paring the contextual information for each event w ith a diction ary contain ing contextual examples based o n interviews w ith ad olescents and ad ults. For each event, the focu s of the event was determ ined to be specifically on the subject, on th e subject jo intl y w ith ano ther person, on ano ther per son, or involving a po ssession or a pet . After the int erviewe r rated each occ u rre nce (i.e., a chronological co m posi te o f events and incidents follow ed by ch ro n ic diffi culties), each LEOS chan was presented to a panel of con sen sus raters consisting of two or more raters blind to the ad olescent's di agnostic status. During the consen sus meeting, the interviewer p resented an overview of the ado lesce nt 's biographical circum sta nces that include d personal and fam ily demographics as wel l as wh o the adolescent's close relatives and co n fid ant s were (since events occurring to them ma y be eligible for inclusion depending o n th e nature of the even t). The agreem ent between all raters and fin al co nsensus ratings was high . Intraclass correl at ion coeffic ients ranged from 0.8 4 to 0.95 for all events. For all events, raters agreed exactl y more th an 90% of th e time (range 84 - 9 1%) and were within o ne level 98 % o f the time (range 98-100% ). For the m ost pan, we found the LEOS as origin ally adapted for use wi th adolescents by M on ck and Dobbs (l985 ) was sufficient for assessing srresso rs in our sam p le of adolescents. The sections o f the LEOS th at need ed to be refin ed we re those fo r wh ich th e cu ltu ral expe rience of U .S. adolescen ts differed from that of adolescents living in the United Kingdom. In general , education is d ifferent between U .S. and British ad olescents . For example, one difference was found
OCTO B ER 19 9 8
1051
WILLIAM SON ET AL.
in U.S. adolescents taking the Scholastic Aptitude Tests (SAr s). Taking the SATs is eligible to be included as an event . and the sho rtand long-term threat ratings will depend on the adolescent 's plan s and commitments for going to college and how poorly or well he/ she may have scored on previous SATs. While the rules for including school tests as events and for assessing their objective th reat are clearly outlined in the LEDS diction aries, the content pert ain ing to taking SAT tests is unique to U.S. adolescents. In add ition . we found it necessary to refine the interview schedule to better reflect the probes necessary for determining wh eth er events and diffi culties met the criteria outlined in the rating manuals. For example. in the romantic relationship domain. the original version of the interview schedule did not probe for boyfriend/girlfriend relationships. In the revised version of the interview schedule. detailed information about boyfriends/ girlfriends is probed for. including specific information about the start of a relationship. which includes (1) whether they go out together. (2) whether the adolescent is considered by the other person as his/ her girlfriend/boyfriend. and (3) whether others see them as a couple. T hese revisions have produced an int erview schedule that is more tightl y linked with the manual and the diction aries of events and difficulties, mak ing the determ ination of what should be included as an event or difficulty much more straightforward. Toward th is end. we have comprehensively adapted the LEDS for use in U.S. adolescents by providing broader-based prim ary probes for specific context and have added U.S. examples of events and difficulties whose context is specific to U.S. adolescents. In general. we were quite impressed with the conceptual flexibility of the LEDS methods and the relative ease with which we were able ro use the LEDS with our sample of adolescents living in the Un ited States. Events and difficult ies were categorized on the basis of Brown and Harris' work examin ing the relation berween life events and dep ression in adult s (Brown and Harris. 1978 , 1989). Event s were classified as severe events if they were (I) rated a l-rnarked on long-term th reat or (2) rated a 2-moderate on long-term threat and focused on the adolescent him/herself or jointly with someone else. Difficulti es were classifiedas major difficulties if they were (1) rated on the top 3 points of severity (l-high marked . 2-low marked, 3-high moderate); (2) had lasted at least 2 years; and (3) did not involve purely health probl ems.
survival analysis based on the Kaplan-Me ier product limit estimate was used to examine the temporal pattern of severe events occurring nearest in time to th e onset of depression or the linked period. Accordingl y. the on set of depre ssion or the start of the linked period was set at 0 and the occu rrence of events prior to these periods was coded in absolute times. Thus, an event occurring 1 month prior to onset would cause the subject to be coded as "dead" and his survival time to be coded as 1. For those adolescents who did not have a severe event in the 12-m onth period . they were coded as "alive" and their survival time was coded as 12. All significance tests were two -tailed with U2 ; 0.05 .
RESULTS
Before comparing acute and ongoing stressors in depressed and normal control adolescents, we examined the relation between age, sex, race, and socioeconomic status and stressors. None of the demographic characteristics were significantly related (Q any of the measures of stress. Presence of Severe Events and Major Difficulties Prior to Onset
Severe events have been shown to be more likely to be present in the year prior to onset in depressed adults (Brown and Harris, 1978 , 1989). To exam ine whether the same is true for depressed adolescents, we compared depressed and normal control adolescents on the presence of at least one traditionally defined severe event in the year prior to onset. More than twice as many depressed adolescents reported having had a traditionally defined severe event occur in the year prior to onset (46%) compared with normal control adolescents (20%), although this difference failed to reach statistical significance (X2 = 2.22, P < .14). Survival analyses similarly showed a trend for greater events in the depre ssed adolescents (generalized Wilcoxon (X 2 = 2.47 , P < .12). More so than adults, adolescents are int imately dependent on and attached (Q others in their environment including parents, grandparents, other close relatives, and friends. To determine whether events focused solely on these significant others in the adolescent's life (as opposed to jointly on the adolescent and the significant other) were associated with depression, we expanded the traditional definition of severe events to also include those events that were focused on others and rated as 2-moderate on long-term threat. These refined "severe" event s were significantly more prevalent among the depressed adolescents prior to onset (62%) than among normal control adolescents (27%) (X 2 = 5.23, P < .02). Survival analyses
Statistical Methods Demographic variables were compared using X2 with Yates continu ity cor rection or t tests as applicable. The relations hip between demograph ic variables and life events measures was assessed using eith er Spearman correlations or Mann -Whitney rank sum tests. For all analyses. the occurrence of life events was examined in the 12month period prior to the onset of the depressiveepisode in depressed adolescents and dur ing a comparable "linked" period in normal control adolescents . The linked period in the normal control adolescents represented the 12-month period co rrespondi ng to th e I2-month per iod that prec eded th e average on set of the dep ressive episode among the depressed adolescents. Because the average onset of illness was approximately 8 months (see Table 1), the linked period for each normal control adolescent represent ed a period covering 8 to 20 months prior to the date of interview.The advantage of using a linked period in normal cont rol adolescents is that the average length of recall for events and d ifficulties was the same for depressed and norm al control adolescents. Therefore. it is unl ikely that the rates of events and difficulties were artificially inflated in the normal control group. Analyses of life events and difficult ies proceeded in rwo steps. First. the presence of severe events and major difficulties was comp ared in MDD and normal control adolescents using X2 analyses. Second.
1052
J. AM .
AC A D. C H I L D ADO LES C. PSYC H I AT RY, 3 7: 10, OCTO B E R 19 9 8
LIF E EVENTS O F D EPR ESSED AD O LES CE N TS
similarly found the depressed adolescents to have significantly more refined severe events (generalized Wilcoxon X2 = 5.67, P < .02). The backward survival plot depi cting the survival fun ction for depres sed and normal control adolescent s is presented in Figure 1. Depressed adolescents had significantly more major d ifficulties compared with normal control adolescents (27% versus 0%, Fisher exact test, p < .04). Considering the refined severe events and major difficulties together, significantly more depressed adolescents had at least one in the year prior to ons et (69 % versus 27%, Yates X2 = 5.32, P < .02). It is interesting that two depressed adolescents (7% ) had a major difficulty but did not have a refined severe event prior to on set. Total Number of Refined Severe and Nonsevere Events
Compared with normal control adolescents , depressed adolescent s had a significantl y highe r average number of refined severe events (2.0 ± 2.1 versus 0.3 ± 0.5 , p < .0 1) and a trend for more nonsevere events (8.0 ± 5.7 versus 4.6 ± 3.3, p < .07) in th e year prior to onset. It is interesting that compared with normal control adolescen ts, a significantly higher percentage of depres sed adolescent s had two or more refined severe events (50% versus 0% , Fisher exact test , p < .00l) and three or more refined
o.
severe events (35 % versus 0% , Fisher exact test , p < .02) in the year prior to on set. Timing of Refined Severe and Nonsevere Events During the Year Prior to Onset
Original stud ies of life event s and difficulties in depressed adults showed that only severe events di stinguished between depressed and community control adults and did so for the entire year period prior to onset (Brown et al., 1973). Whether adolescents similarly experience higher rates of severe events for the entire year prior to onset or have higher levels of events across all threat levels remains an empirical question that has yet to be exam ined. Therefore, we examined the frequen cy of refin ed severe and nonsevere even ts among depressed and normal control adolescent s in the twelve I -rnonth intervals prior to on set. As shown in Figure 2, refined severe events were elevated in the depres sed adol escents across the twelve l -rnonth periods prior to on set compared with normal cont rols, although only the 7- and 10month comparisons reached statistical significance (p < .0 5). Similarly, nonsevere events were also elevat ed throughout the twelve l-monrh periods pr ior to on set; however, only the 7- and 1O-month comparisons reached statistical significance.
»,
--
.....
"E 0.8
~"S
.
.
....
0.7
~ 0.6
.€8
8. 0.5
~ o,
0.4
Q)
~
"5
0.3
(J
0.2
E ::)
---.. .
Depressed (n=26) Normals(n=15)
0.1
o
2
3
62% of the Depressedhad a ' Severe" Event 27% of the Normal Controlshad a "Severe"Event Generalized Wilcoxon= 5.67, df=1, P < .02
456 789 Months Priorto Onset or "Linked"Year
10
11
12
Fig. 1 Backward survival analysis for refined "severe" events during the year prior to o nset in th e depressed and the corre spondi ng "linked" year in the normal contro l adol escents.
J.
AM . ACA D. C H ILD ADOL ESC . PSYC H IATR Y• .n :lo . O CTOB ER 199 8
1053
WILLIAMSON ET AL.
-
MDOsI'l=26
CJ
NCs11=15
12
11
10 9 8 7 6 5 4 3 Month Prior to Onset or "Linked" Year
Fig. 2 Frequency of severe events during the 12 months prior disorder; NC s = normal control subjects.
1
onset . MDDs = subjects with major depressive
lar immune function, EEG sleep, nocturnal growth hormone secretion, and exp ression of depression in depressed children and adolescents (Birmaher et al., 1994; Williamson et al., 1995a,b, 1996). In all of these studies, we relied on a life events checklist, primarily Coddington's Life Event Record (Coddington, 1972), to assess stressors. As noted above, there are a number of limitations to assessing stressors using life events checklists, and these compromised our ability to investigate fully the relationship between stress and various neu roendocrine systems in depressed children and adolescents. For example, in our EEG sleep study (Williamson et al., 1995b), the finding that stressors decreased rapid eye movement sleep latency in normal controls but not depressed adolescents might very well have been a result of differences between the two groups in when the stressors occurred in relation to the sleep study. Because the Life Event Record was used, events were indicated only as occurring in the year prior to the EEG sleep studies and were not specificallydated. To the best of our knowledge, the current report represents the first full application of the Bedford College LEOS method for assessing acute and chronic stressors among depressed adolescents. As hypothesized, depressed adolescentsexperiencedsignificantlymore stressors during
Comparison of First Episode and Recurrent Depression
Of the 26 depressed adolescents , 9 had at least one prior episode of depression. Among these adolescents with recurrent depression, none had a prior episode occur during the I-year period prior to the onset of their most recent episode of depression. Depressed adolescents who had their first episode of depression reported experiencing a similar rate of refined severe events prior to onset (65%) compared with adolescents with recurrent depression (56%) (generalized Wilcoxon X2 = 0.62, P < .43). In contrast , significantly more depressed adolescents with a recurrent depressive episode reported a major difficulty (67% versus 6%, Fisher exact test, p ~ .0 1). Considering refined severe events and major difficulties together, adolescents with recurrent depression had similar rates of at least one acute or chronic stressor prior to the onset of their most recent depressive episode (78% versus 65%, Fisher exact test, not significant). DISCUSSION
Our interest in using the LEOS, an investigator-based interview schedule, to assess stressors in adolescents evolved out of our previous studies in which we examined the relations between stressful life events and cellu-
1054
to
2
J.
AM . AC AD. C HI LD A DOL ESC . PSYCHIATRY, 37 : 10 , OCTO BE R 19 9 8
LIFE EVENTS OF DEPRESSED ADOLESCENTS
the year prior to onset when compared with a comparable 12-month period in normal controls. A significantly greater number of depressed adolescents experienced a traditionally defined severe event (46% versus 20%, not significant), those rated I-marked or 2-moderate longterm threat and focused on the adolescent or jointly with another person, in the year prior to onset. However, probably because of the small sample size, this increase did not reach statistical significance (the power to detect a between-group difference was 0.41). Although severe events occurred across each of the 10 domains, typically they were centered around the adolescent's romantic relationships. Following is an example of one such severe event that occurred in the year prior to onset in a depressed adolescent female:
A I6-year-old female had had a confiding and sexual relationship with her boyfriendfor 72 months. Their relationship had a history ofongoing physical violence. While at a party, the subject's boyfriend hit her and she immediatelybroke offthe relationship. During the next 2 weeks he telephoned herseveral times, but she did not return hiscalls. Additional severe events in this sample of adolescents included the following: close friend moving away, first sexual intercourse, becoming pregnant, having an abortion, friend being raped, sister returning home to live, being sexually harassed by a stranger, argument and breakdown in relationship with mother, and a fight at school. Expanding severe events to include those events rated as 2-moderate on long-term threat and focused on others (parents, siblings, grandparents, friends, etc.) resulted in a significantly higher percentage of depressed adolescents having a refined severe event prior to onset (62% versus 27%, p < .02). Below is an example of an "otherfocused" event that was rated 2-moderate on long-term threat and reclassified by us as a refined severe event:
A I4-year-old malereported havinga very close relationship with his70-year-oldgrandfather; they livedon thesame street and saweach other at least every other day. Thegrandfather died ofkidneyfailure after having been hospitalized for 2 weeks; hehad been scheduled to undergo kidney dialysis. In this sample of adolescents, additional events that were other-focused and classified as refined severe events included the following: mother having surgery for breast cancer, father being involved in a car accident and taken by ambulance to the hospital, death of an uncle, exboyfriend shooting another person, and classmate being killed in an automobile accident.
J.
The finding that events rated 2-moderate on long-term threat and focused on others appear related to depression onset in adolescents differs from Brown's findings in depressed adults. In his original Camberwell sample of women (Brown and Harris, 1978), only events rated as I-rnarked or 2-moderate and focused on the woman herself or jointly with someone else were related to the presence of depression. From findings of the current study it appears that the type of events likely to precede the onset of a depressive episode during adolescence differs from the type of events found to precede the onset of depression in adults. In adolescents, stressful events occurring to primary attachment figures such as parents and grandparents seem to be centrally involved in the onset of depression in this age group. Future studies will need to confirm the importance of these other-focused events in the development of depression in adolescents. On the basis of their study of depression among women living in Camberwell, Brown and Harris (1978) defined provoking agents for depression as either a severe event or a major difficulty. Major difficulties are ongoing stressors lasting a minimum of 2 years, rated on the top 3 points of severity, and not purely involving health problems. Studies of depressed adults have shown that severe events account for most of the variance; however, major difficulties significantly add to the variance for depression (Brown and Harris, 1978). In the current study, 27% of the depressed adolescents had a major difficulty compared with none of the normal control adolescents. Following is an example of an ongoing stressor that was classified as a major difficulty:
A I5-year-old white female reported that she and her father did not get along. The subject and herfather fought constantly about hergrades, herfriends, and her behavior. She said that he did not showany interest in her life he rarely went to teacher conferences). He told her that he wouldnot besurprised ifshe were a tramp. The[ather often threatened topunch her and had attackedher in the past. He had slapped her in the face and about the head, had thrown a remote control device and typewriter at her, and had punched her in her back. These physical altercations resulted in the subject's sustaining bodily bruises and welts.
AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 37:10, OCTOBER 1998
««.
Additional major difficulties in this sample of adolescents included mother's health, brother's health, tension in relationship between parents, and relationship with parents. Sixty-nine percent of the depressed adolescents in the current study had a severe acute or chronic stressor, with
1055
W[LLIAMSON ET AL.
62% having a refined severe event and 7% having only a major difficulry. The finding that refined severe events account for most of the between-group variance is in agreement with studies of depressed adults (Brown and Harris, 1978) . In their study of Camberwell women, Brown and Harris found that 61 % of depressed patients had a provoking agent, with 49 % having a severe event and 12% having only a major difficulry. The results from this study suggest that the onset of depression in adolescents is similarly preceded by acute stressful events, although ongoing stressorsappear to contribute to the onset of depression in at least some adolescents. In the future, it will be interesting to examine the relation between refined severe events and major difficulties and their combined importance for the development of depression in adolescents. In this investigation, depressed adolescents were significantly more likely to have had two or more refined severe events prior to becoming depressed (50 versus 0%) compared with normal controls. Considered together with the high rate of nonsevere srressors experienced by both depressed and normal control adolescents, the se data suggest that depressed adolescents are experiencing higher levelsof stressors compared with depre ssed adults . Future investigations might fruitfully explore the source of these refined severe events to determine whether they evolve from similar or separate domains. Clinical Implications
The main clinical implication of this study is that adolescents who experience high levels of severe stressful events might be at increased risk for the development of depression. As reponed herein, 50% of the depressed adolescents reponed experiencing two or more severe events prior to onset compared with none of the normal controls! Therefore, clinical efforts focused on providing clinical intervention soon after the occurrence of the first severeevent might help to prevent the onset of depression in adolescents. Future studies should include prospective assessment of life events in nondepressed adolescents, some of whom will subsequently become depressed . In summary, this study represents one of the first applications of the LEOS to a sample of adolescents in the United States. The finding that LEOS-defined stressors are elevated in depressed adolescents prior to onset parallels the work of Brown and Harris (1978, 1989) in depressed adults . Future studies of adolescents are needed
1056
which use an investigator-based measure of acute and chronic stress, such as the LEOS , that also include other potential risk factors such as parental psychopathology and specifically examine the role of stressorsin the onset, maintenance, and recurrence of depression in this age group.
REFERENCES American Psychiatric Association ( 1987). Diagnostic and Statistical Manual of Mental Disorders, 3rd edition-reuised (DSM -III-R). Washington. DC: American Psychiatric Association Banez GA. Compas BE (1990 ), Children's and parent's daily stressful events and psychological symptoms. ] Abnorm Child Psycho! 18:591- 605 Beck AT. Ward C H. Mendel son M (196 1). An invento ry for measur ing depression . Arch Gm Psychiatry 4:53- 63 Birmaher B. Rabin BS. Garcia MR er al. (1994). Cellular im munity in depressed. cond uct disord er. and norm al adolescents: role of adverse life events. ] Am Acad Child Adolesc Psychiatry 33:671- 678 Brown GW. Harris TO. eds (1978), Socia! Origins of Depression: A Study of Psychiatric Disorder in Womrn. Lond on : Tavisrock Brown GW, H arr is TO. eds (1989). Lift Events and Illness. New York: G uilfo rd Brown GW. Sklair F. Harris T O . Birley JLT (1973) . Life-events and psychiatric disorders part 1: some methodological issues. Psycho! Med 3:74- 87 Cha mbers WJ. Puig-Ant ich J. Hirsch M et al. (1985) . T he assessment of affective d isorders in child ren and adolescents by semi-str uctu red interview: test-retest reliability of the Sched ule for Affective Disord ers and Schizophren ia for School-Age Children. Present Episode version . Arch
Gen Psychiatry 42:696- 702 Coc hran WG (1954). Some methods for strengthening the common X' test. Biometrics 10:417-44 I Co ddington RD (1972), The significance of life events as etiologic factors in diseasesof children. I: a survey of professionals.] Psychosom Res [6:205-2 13 Co hen S. W ills TA (1985), Stress. social suppo rt and the buffering hypothesis. Psy cho! BuI/9 8:31O-357 Dahl RE. Ryan N D . Matty MK et al. (1996 ). Sleep onset abnormalities in depressed ado lescents. Biol Psychiatry 39:400-410 Daniels D. Moos RH (1990), Assessinglife stressorsand social resources among adolescents: applications to depressed youth.] Adolesc Res 5:268-289 Goo dyer I. Kolvin I, Garzani s S ( 1985). Recent unde sirable life events and psychiatric disorder in childhood and adolescence. Br ] Psychiatry 147: 5 17- 523 Goo dyer [M . Wright C, Altharn PME ( 1988). Maternal adversity and recent stressful life events in anxious and depr essed children. ] Child Psycho! Psychiatry 29:6 5 1-667 Hollingshead AB (1960). Four-Factor Index ofSocia!Position. New Haven. CT: Yale Un iversity Dep artment of Sociology Jenki ns CD. Hu rst MW. Rose RM (1979). Life chang es: do peopl e really remember? Arch Gen Psychiatry 36:379-384 John son JH . McCutcheon SM (1980 ). Assessing life stress in older child ren and adolescent s: prelim inary finding s with the Life Events C hecklist. In : Stress and Anxiety. Sarason [G . Spielberger CD. eds. Washington. DC: Hemisphere Kashan i JH . Vaidya AF.Soltys SM. Dandoy AC, Reid JC (1990). Life events and major depression in a sample of inpatient childr en. ComprPsychiatry 31:266 -274 Monck E. Dobb s R (1985 ). Measuring life events in an adolescent popul ation : method ological issues and related finding s. Psycho! Med 15:841-850 Mullins LL. Siegel LJ. Hodges K (198 5). Co gnitive problem -solving and life event correlates of depre ssive symptoms in ch ildren. ] Abnorm Child Psycho! 13:30 5-314 O rvaschel H . Puig-Am ich J. C hambers WJ. Tabrizi MA. Johns on R ( 1982). Retrospect ive assessment of child psychopathology with the K-SADS-E. ] Am Acad Child Psychiatry 2 I :392-397
J. AM . ACA D . C H I1.D AD01. ESC. " SYCHI ATRY .
3 7 : 10 . OC T O BE R 199 8
LIFE EVENTS OF DEPRESSED ADOLESCENTS
Ryan NO, Dahl RE, Birrnaher B er al. (1994), Stimulatory tests of growth hormone secretion in prepubertal major depression: depressed versus normal children. JAm Acad ChildAdolesc Psychiatry 33:824-833 Swearingen EM, Cohen LH (1985), Life events and psychologicaldistress: a prospective study of young adolescents. DeuPsychol21: 1045-1054 Williamson DE, Birmaher B, Anderson BP, Al-Shabbour M, Ryan NO (1995a), Stressfullife events in depressed adolescents: the role of dependent events during the depressiveepisode. JAm AcadChildAdolesc Psychi-
atry 34:591-598
Williamson DE, Birmaher B, Dahl RE, Al-Shabbour M, Ryan NO (1996), Stressful life events influence nocturnal growth hormone secretion in depressed children. BioiPsychiatry 40: 1176-1180 Williamson DE, Dahl RE, Birmaher B, Goetz RR, Ryan NO (1995bl, Stressfullife events and EEG sleep in depressed and normal control adolescents. BioiPsychiatry 37:859-865 Williamson DE, Ryan NO, Dahl RE, Jeanette L (1992), Hamilton depression scorescan be extracted from the K-SADS-P in adolescents.J ChildAdolesc
Psychopharmacol2: 175-181
The Changing Pattern of Substance Abuse in Urban Adolescents. Cynthia Brasseux, MA, Lawrence J. D'Angelo, MD, MPH, Mark Guagliardo, PhD, Jocelyn Hicks, PhD
Objectives: To determine the prevalence of specific drug use in adolescents attending an adolescent health clinic and to compare current rates with a similar previous study. Design: Blinded and anonymous urine samples obtained from patients presenting for routine health care were tested for the presence of cannabinoids, phencyclidine (PCPl, amphetamines, opiates, and cocaine. Setting: Adolescent medicine outpatient clinic. Patients: Patients were between 12 and 21 years of age. Specimens from 1313 patients in 1995 to 1996 and 1312 patients in 1989 to 1990 were tested. Main Outcome Measures: Current drug use rates were compared with a similar screening of patients conducted in 1989 to 1990. Comparisons between studies were made on the basis of specific drug, age, and SeX. Results: For the most recent patient group, 14% Were positive for 1 or more drugs and 13% were positive for cannabinoids. Males were significantly more likely to test positive for drug use than females. The oldest adolescents Were more likely to test positive for drug use than younger adolescents. Comparing the 2 study year cohorts, patients tested recently were significantly more likely to have urine tests positive for at least 1 drug and cannabinoids in particular and less likely to have urine tests positive for cocaine. Conclusions: There has been an increase in positive urine tests in patients seen in our ambulatory clinic, with a strong shift toward cannabinoids and a shift away from cocaine. Practitioners need to be aware that drug use patterns in adolescents can shift relatively abruptly and counseling should be targeted to current drug USe patterns. Arch Pediatr Adolesc Med 1998;152:234-237 Abstracts selected by Michael]. Maloney, M.D., Assistant Editor.
j. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 37:10, OCTOBER 1998
1057