Adolescent psychiatric patients' interactions with their mothers.

Lasko, David ; Field, Tiffany ; Bendell, Debra 等

Few studies have focused on the interactions between adolescents with psychiatric problems and their parents. One might expect that these adolescents would not only have disturbed interactions with their parents, but would also have distorted perceptions of their interaction behaviors. Research in this area has focused on perceptions of relationships rather than actual interactions. Studies examining the interactions of adolescent psychiatric inpatients have been limited to audio rather than videotaped interactions, and did not assess the subjects' perceptions of the interactions. In a study by Hauser, Houlihan, Powers and Jacobsen (1987), for example, detailed analyses were made of audiotaped parent-child interactions. Conversation patterns differentiated psychiatric and nonpsychiatric adolescents. More disturbed mother-adolescent interactions were noted in the psychiatric sample.

The interactions of normal adolescents and their parents could serve as a model for studying disturbed adolescents. In a study by Callan and Noller (1986), normal adolescents and parents coded their own and their partners' behaviors during a video replay of dyadic interactions. The authors reported differences in how the adolescents and their parents behaved and in how they perceived their interactions. The adolescents rated their parents as more anxious, less involved, and less dominant than the parents rated themselves. The adolescents were rated as less involved and less dominant than their parents. In a second study, outside coders were added to the procedure (Noller & Callan, 1988). These observers rated family members more negatively than family members rated themselves, raising the possibility that the adolescents and their parents had distorted perceptions of their inter-actions.

The purpose of the present study was to determine (1) whether adolescents with psychiatric problems and their mothers interact differently as a function of symptom profile (internalizing/externalizing, depressed/nondepressed, and high and low anxiety); (2) whether adolescents rate their mothers similarly to the way their mothers rate them; and (3) whether their perceptions of the interactions are congruent with ratings by an objective observer. The findings would indicate whether brief interactions should be videotaped as a portion of the psychiatric intake process. The videotape would help familiarize the clinical staff with the dynamics of parent-adolescent interaction, which they might not otherwise have an opportunity to observe. In addition, the ratings would give them information on adolescents' and parents' perceptions of each other's behaviors.

METHOD

Subjects

Thirty-eight adolescent-mother dyads were recruited from two clinical sites; twenty dyads from an inpatient adolescent psychiatric unit in Miami and 18 from an outpatient unit in Northern California. Although these were convenience samples, it was expected that recruiting from two psychiatric settings in different areas of the country would offer greater generalizability of the results. Both of these sites served children and adolescents with the typical range of psychiatric diagnoses, including depression, conduct disorder, adjustment disorder, attention deficit hyperactivity disorder, and psychoses. On average, children remained in these treatment programs for three months.

The adolescents (63% female, 37% male) ranged in age from 11 to 18 (M = 14.3 years) and included a broad representation of family income levels (median income = $20,000-30,000 per year, range = $10,000-90,000 per year). However, most of the patients would be categorized as low socioeconomic status. The ethnic distribution was 45% white (non-Hispanic), 35% Hispanic, and 20% black. The average education of the parents was 11.3 years (range = 8 to 16). The exclusion criteria for recruiting were the unavailability of the mother, inappropriate diagnoses, or active psychosis in the adolescent. Approximately 40% of the subjects who were approached refused to be in the study or failed to meet the inclusion criteria.

Procedure

Subjects were recruited immediately after intake to the inpatient or outpatient facility and prior to treatment. This was the first treatment situation for all subjects. Although the adolescents' problems presumably had occurred in advance of their admission to these treatment programs, the adolescents' and mothers' reports on duration of problems varied considerably.

The procedure for this study consisted of the following: (a) psychiatric interview with the parent and adolescent, (b) questionnaires administered to the mother and adolescent, (c) the videotaped interaction, (d) the videotape playback for independent coding by the mother and adolescent, and (e) the coding of the tapes by a trained observer. The videotaping procedure was expected to give psychiatric staff greater familiarity with the adolescents' interactions with their mothers.

Psychiatric interview. The admitting psychiatrist conducted a structured interview with the mothers and observed and talked with the adolescents at intake. Based on these interviews, DSM-III-R diagnoses of externalizing (conduct disorder) or internalizing (depression) disorders were assigned to the adolescents. Two psychiatrists (in the case of the inpatient unit) and one psychiatrist (in the case of the outpatient unit) were responsible for these diagnoses. Interrater reliability was checked on the diagnoses of 30% of the patients and was found to be high (r = .87). Although the comorbidity rate was high, as might be expected, the primary diagnosis was used for classification.

Questionnaires. Each mother was asked to complete the following instruments.

1. Demographic/Social Support Questionnaire. This 20-item questionnaire measures family socioeconomic status and the mother's overall perception of her relationship with her adolescent (Field, 1990). For example, participants are asked how frequently they touch, hug, or kiss their adolescents, and how frequently they engage in one-to-one interactions. They are also asked to rate, from poor to very good, the quality of these interactions. The questions are scored from i (representing the most negative answer) to 4 (the most positive answer).

2. Center for Epidemiological Studies Depression (CES-D) Scale (Adult Form). The adult version of this 20-item, 4-point Likert scale assesses current level of depressive symptomatology, with emphasis on the affective component (Radloff, 1977). Subjects are asked to recall the extent to which they experienced depressive symptoms during the past week. This scale has shown good internal consistency and test-retest reliability.

3. Interaction Behavior Checklist (Parent Form). The 20-item IBC asks parents to report on various characteristics of their relationship with their adolescents (Farber & Jenne, 1963). The true-or-false items on this scale include such questions as "My child and I sometimes end our arguments calmly" and "I understand my child." This questionnaire was used as the focus for the first five-minute interaction between mother and adolescent.

The adolescents completed the following instruments.

1. Activity Checklist. This scale includes 20 items on school activities and grades, as well as extracurricular activities and drug use (Field, 1990). In addition, adolescents are asked the same questions as their mothers about the quality of their interactions and about how often they touch, kiss, and hug their mothers. The items are scored from 1 (most negative answer) to 4 (most positive answer).

2. Center for Epidemiological Studies Depression Scale (Child Form). This is the adolescent and child version of the CES-D adult form (Radloff, 1977). The mean score in a normal sample of high school students is 16.8 and the standard deviation is 9.9. The scale has shown good reliability, and internal consistency has been documented with several large samples (Radioif, 1991). This scale was included to tap the adolescents' depressive symptoms.

3. Interaction Behavior Checklist (Child Form). This is the same as the parent form, except that adolescents rate their perceptions of the relationship.

4. Social Anxiety Scale for Children-Revised (Adolescent Version). The 22-item SASC measures children's self-perceived anxiety in social situations (LaGreca, 1989). It yields three subscales: fear of negative evaluations (FNE), social avoidance and distress in generally distressing situations (SAD-G), and social avoidance and distress in situations that are new or unfamiliar (SAD-N). Items are scored from never (1) to most of the time (5). Higher scores indicate greater social anxiety.

Videotaped interaction. Following completion of the questionnaires, the mothers and adolescents participated in two videotaped interactions of five minutes each. The first interaction involved a discussion of agreements and disagreements on the individual items of the Interaction Behavior Checklist. For the second interaction, the dyads received the following instruction: "Discuss the one rule you would choose for your family or household if only one rule were allowed and why you would choose that rule." The two formats were designed to introduce slight conflict into the conversation.

Videotape playback. Immediately after the interactions, subjects were asked to rate each conversation. The ratings were on a 6-point Likert scale for four behavioral dimensions: anxious/calm, unfriendly/ friendly, uninvolved/involved, and pushover/bossy (adapted from the Callan and Noller studies, 1986, 1988). Each interaction was replayed with pauses every 30 seconds, during which subjects rated the preceding segment on these four scales. At each interval, subjects first rated their own behavior and then the other person's behavior. This resulted in two sets of ratings for each member of the dyad for each interaction. Ratings were then averaged across the ten segments.

Observer ratings. Later, a trained observer, who was not informed of the purpose of the study and the psychiatric status of the subjects, coded the interaction tapes using the identical method employed by the subjects. Thus, at each of the ten 30-second intervals, the observer rated both the mother and adolescent on each of the four dimensions. Interobserver reliability was assessed based on 30% of the videotapes; kappa coefficients averaged .82.

Data analyses. Preliminary analyses were undertaken (1) to determine whether the sample was different from the larger treatment population, and (2) to determine whether outpatient/inpatient, sex, ethnic group, and SES needed to be entered as covariates. No differences were found on demographic variables. The data from the inpatient and outpatient samples were then compared; no differences emerged (see Table 1). Thus, data were collapsed across the two samples.

The adolescents were then grouped as internalizers or externalizers based on the primary diagnosis (DSM-III-R) made by the psychiatrist at the time of their admission to treatment. For the purposes of the present study, internalizing diagnoses included depression and anxiety disorders, and externalizing diagnoses included adjustment disorders with behavioral components and disruptive behavior disorders. High and low adolescent depression and social anxiety groups were defined by using the standard cutoff scores on the CES-D (median = 18; Radloff, 1977) and the SASC-R (median = 47; LaGreca, 1989). Table 1 Demographic and Interaction Variables for Inpatient and Outpatient Samples Inpatient Outpatient Mean SD Mean SD Demoqraphic Variables Mother Education 12.70 2.8 13.10 2.0 Interactions 6.70 1.1 6.90 1.1 CES-D Score 18.45 11.5 12.72 8.9 Adolescent Age 13.6 2.3 15.1 1.6 # of friends 4.10 1.2 4.10 1.1 # of problems 0.90 0.9 0.64 0.8 Interactions 5.10 2.2 4.70 2.3 CES-D score 17.14 8.9 25.17 13.8 SASC-R score 49.37 13.3 51.89 19.2 Interaction Variables Calmness 4.49 1.2 4.45 1.1 Friendliness 4.78 1.1 4.87 1.0 Involvement 4.96 1.1 4.81 1.1 Bossiness 3.47 0.9 3.51 0.8

A randomized block factorial design was employed, which is a modification of the method used by Noller and Callan (1988) to analyze their data for normal adolescents. The randomized block design was used here instead of a completely randomized design because the adolescents, the mothers, and the independent observers rated each member of the interaction. Therefore, the assumption of independence of observations for a completely randomized design would have been violated. No such assumption exists in the randomized block design, which is suited for multiple observations of the same individual or block (Kirk, 1982). The blocking variables included symptoms (internalizer/externalizer, high/low depression, and high/low social anxiety), target of the coding (adolescent versus mother), and coder (adolescent, mother, and independent observer). Because each member of the dyad was coded by three people and each person's rating was considered a separate observation, the degrees of freedom were greater than the number of people who actually participated. This is similar tea repeated measures analysis of variance, although the repeated measures in this case were collected from different persons. The dependent variables were the four behavioral dimensions coded by the adolescent, mother, and independent observer (calmness, friendliness, involvement, and bossiness).

RESULTS

An initial analysis revealed no gender or ethnicity differences in the internalizer/externalizer diagnoses or the depression and Anxiety scores.

Psychiatric Diagnosis

For the internalizer/externalizer grouping, 53% of the adolescent externalizers were males. The only background variable difference was in number of close friends, with externalizers reporting significantly more close friends (M = 4.5 versus 3.9, p [less than] .05) (see Table 2). Analysis of variance yielded significant main effects for calmness, F(1, 190) = 8.45, p [less than] .005; friendliness, F(1, 190) = 8.92, p [less than] .005; and involvement, F(1, 190) = 5.04, p [less than] .03. All coders judged both members of the internalizing adolescent dyads to be calmer, friendlier, and more involved than were those in the externalizing adolescent dyads (see Table 3). No interaction effects emerged between the internalizer/externalizer diagnosis and the target of the coding (mother/adolescent).

Depression

For the depression grouping, 75% of the higher depression group were females. As expected, the higher depression adolescents had significantly higher depression scores (M = 31.5 versus 11.8, p [less than] .001). These subjects also reported significantly greater social anxiety (M = 59.1 versus 41.4, p [less than] .005) (see Table 2). Significant main effects emerged for calmness, F(1,170) = 6.36, p [less than] .02; friendliness, F(1, 170) - 5.96, p [less than] .05; and involvement, F(1, 170) = 5.80, p [less than] .05. The low adolescent depression dyads were calmer, friendlier, and more involved than were the high adolescent depression dyads (see Table 3).

Social Anxiety

The higher social anxiety group (55% females) reported greater depression (M = 27.2 versus 15.3, p [less than] .005) (see Table 2). No significant main effects emerged for anxiety level (see Table 3). However, a significant interaction between anxiety grouping and the target of coding emerged for calmness, F(1, 196) = 8.44, p [less than] .005. The low anxiety adolescents and their mothers received similar ratings on calmness (adolescents, M = 4.51; mothers, M = 4.54), whereas the high anxiety adolescents received significantly lower ratings than did their mothers (adolescents, M = 3.95; mothers, M = 4.91).

Target of the Coding

The analyses yielded main effects for the target of the coding on all four dimensions. In each case, mothers, as compared with adolescents, were rated as calmer, F(1, 194) = 9.52, p [less than] .005; friendlief, F(1, 194) = 5.93,p [less than] .05; more involved, F(1,194) = 12.92,p [less than] .001; and bossier, F(1, 194) = 4.53, p [less than] .05 (see Table 4).

DISCUSSION

That adolescent psychiatric patients and their mothers differed in their conversation behaviors is not surprising. Differences could be due to less pathology in the mothers or simply to the lower responsivity and sociability typically observed in adolescents. Because these effects were derived from the scores of three different raters, they probably do not represent a rater bias. This incongruence in interaction behavior may have an adverse effect on the mother-adolescent relationship. The literature suggests that complementarity in interactions is beneficial to friendships.

The internalizers in this study were rated more positively than were the externalizers. Not surprisingly, they were rated as calmer, although calmness could have been confused with withdrawal. The finding that the internalizers (and their mothers) were rated as friendlier, more involved, happier, and having better overall interaction skill is more difficult to explain. These findings may relate to the nature of the interaction. Withdrawal (which is characteristic of internalizers in the face of conflict) may have seemed more appropriate in the interaction situation of this study. Table 2 Demographic and Questionnaire Variables Independent Dependent Mean (SD) Mean (SD) Variable Variable Psychiatric Diaqnosis Internalizer externilizer Mother Education 2.86 (0.9) 2.58 (1.0) Interactions 6.81 (1.3) 6.31 (0.6) CES-D score 14.72 (10.1) 14.89 (11.7) Adolescent Interactions 4.83 (2.2) 4.31 (1.8) Age 14.78 (1.9) 13.69 (1.8) Number of friends(*) 3.86 (1.1) 4.54 (0.8) Number of problems 0.74 (0.9) 0.85 (0.7) CES-D score 21.85 (12.3) 19.72 (12.5) SASC-R score 50.30 (14.3) 49.00 (18.9) Below Above Cutoff Cutoff Depression Mother Education 2.60 (0.7) 2.75 (1.2) Interactions 6.31 (1.4) 6.67 (1.1) CES-D score 12.15 (7.1) 16.80 (12.2) Adolescent Interactions 4.50 (2.0) 4.63 (2.2) Age 14.69 (1.8) 14.81 (2.0) Number of friends 4.13 (1.1) 3.81 (1.1) Number of problems 0.56 (0.7) 1.00 (0.9) CES-D score(***) 11.81 (4.7) 31.50 (9.5) SASC-R score(**) 41.44 (13.8) 59.13 (15.6) Below Above Cutoff Social Anxiety Mother Education 2.73 (0.9) 2.75 (1.1) Interactions 6.20 (1.4) 6.80 (1.0) CES-D score 15.17 (9.9) 14.25 (10.8) Adolescent Interactions 4.24 (2.0) 4.85 (2.1) Age 14.88 (1.5) 14.05 (2.2) Number of friends 4.35 (0.9) 3.79 (1.2) Number of problems 0.59 (0.6) 1.00 (1.0) CES-D score(**) 15.33 (8.6) 27.24 (12.8) SASC-R score(***) 36.35 (8.6) 62.70 (10.3) * p[less than].05, ** p[less than].005, *** p[less than].001 Table 3 Main Effects Adolescent Internalizer versus Externalizer Dyads Internilizer Externalizers Mean SD Mean SD Calmness(**) 4.65 1.2 4.16 1.1 Friendliness(**) 4.98 1.1 4.52 1.1 Involvement(*) 4.99 1.1 4.64 1.1 Bossiness 3.43 0.8 3.59 0.8 Low versus High Adolescent Depression Dyads Low High Mean SD Mean SD Calmness(*) 4.69 1.1 4.25 1.2 Friendliness(*) 5.08 1.0 4.71 1.1 Involvement(*) 5.09 1.0 4.72 1.2 Bossiness 3.51 0.8 3.47 0.9 Low versus High Adolescent Anxiety Dyads Low High Mean SD Mean SD Calmness 4.53 1.0 4.43 1.3 Friendliness 4.78 1.1 4.86 1.1 Involvement 4.89 1.1 4.87 1.2 Bossiness 3.48 0.8 3.50 0.8 * p[less than].05, ** p[less than].005, *** p[less than].0005 Table 4 Target of Coding Adolescents Mothers Mean SD Mean SD Calmness(**) 4.19 1.2 4.75 1.1 Friendliness(*) 4.64 1.1 5.02 0.9 Involvement(***) 4.59 1.2 5.17 1.0 Bossiness(*) 3.35 0.8 3.63 0.8 * p[less than].0, ** p[less than].01, *** p[less than].0005

Despite their demographic similarity, the internalizer and externalizer dyads differed on several behavioral dimensions, which seems to validate the internalizer/externalizer psychiatric diagnosis groupings. However, psychiatric diagnosis may relate more to behavior than to serf-report (i.e., depression and anxiety levels) - the diagnostic process takes into account specific behaviors in order to meet criteria rather than the feelings and beliefs that are central to self-report. The greater reliance on behavioral criteria for the internalizer/externalizer diagnosis may therefore have accounted for the behavior differences between groups.

Another important factor relates to comorbidity, which often occurs in this population. Replication with less comorbid groups of adolescents may identify more subtle differences between depressed and nondepressed adolescents.

Although one interaction effect emerged for the social anxiety grouping, the lack of main effects could relate to the narrow distinction between the two groups. In order to examine this hypothesis, the data were reanalyzed using cutoff scores that have been developed for the SASC-R more recently (LaGreca, 1993, personal communication). The low anxiety group included adolescents who scored below 39 and the high anxiety group included those who scored above 49. Those scoring between 39 and 49 were excluded from the analysis. However, no main effects were identified, suggesting that it was not the cutoff between the groups that led to the lack of effects.

The lack of differences between the high and low social anxiety groups is puzzling because several differences were found between the depression groups. The comorbidity of depression and anxiety has been well established, and in the present study CES-D and SASC-R scores were highly correlated (r = .64, p [less than] .01). One possible explanation relates to the high anxiety group having a higher proportion of males than the overall sample and a lower proportion of females. Thus, the sample may not be representative with regard to the social anxiety variable, explaining the absence of high-flow anxiety group differences on the CES-D (i.e., females in this sample who had higher CES-D scores were underrepresented in the high anxiety group). Therefore, it would be inappropriate to generalize the findings on this variable to a clinical population without further research.

Finally, the absence of interaction effects in all three sets of data analyses suggests that the patient is not the only one affected. Adolescent psychopathology may affect mothers as well. Parents are often uninvolved in treatment, making it difficult for clinicians to observe interactions. The findings indicate that the brief-videotaping procedure used in this study might be productively implemented as part of the psychiatric intake process in order to help clinical staff understand family dynamics.

The authors gratefully acknowledge the parents and teenagers who participated in this study, and Cami McBride and Jeff Harding, who assisted in data collection and coding. This research was completed in partial fulfillment of the master's degree by David Lasko and was supported by NIMH Research Scientist Award #MH00331 and NIMH Basic Research Grant #MH46586 awarded to Tiffany Field.

REFERENCES

Callan, V. J., & Noller, P. (1986). Perceptions of communicative relationships in families with adolescents. Journal of Marriage and the Family, 48, 813-820.

Farber, B., & Jenne, W. C. (1963). Family organization and parent-child communication: Parents and siblings of a retarded child. Monographs of the Society for Research in Child Development, 28 (Serial No. 91).

Field, T. M. (1990). Demographic/Social Support Questionnaire. Unpublished questionnaire.

Field, T. M., Sandberg, D., Goldstein, S., Garcia, R., Vega-Lahr, N., Porter, K., & Dowling, M. (1987). Play interactions and interviews of depressed and conduct disorder children and their mothers. Child Psychiatry and Human Development, 17, 213-234.

Hauser, S. T., Houlihan, J., Powers, S. I., & Jacobsen, A.M. (1987). Interaction sequences in families of psychiatrically hospitalized and nonpatient adolescents. Psychiatry, 50, 308-319.

Kirk, R. E. (1982). Experimental design. Chapter 6, pp. 237-261.

LaGreca, A.M. (1989). Social Anxiety Scale for Children - Revised. Unpublished manuscript.

Noller, P., & Callan, V. J. (1988). Understanding parent-adolescent interactions: Perceptions of family members and outsiders. Developmental Psychology, 24, 707-714.

Radloff, L. S. (1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385-401.

Radloff, L. S. (1991). The use of the Center for Epidemiologic Studies Depression Scale in adolescents and young adults. Journal of Youth and Adolescence, 20(2), 149-165.