Anger, Interpersonal Relationships, and Health-Related Quality of Life in Bullying Boys Who Are Treated With Outpatient Family Therapy: A Randomized, Prospective, Controlled Trial With 1 Year of Follow-up

Selection and Description of Participants

A total of 320 families within a radius of 20 km from the clinic were selected for this study with random numbers from an Excel table. Between 1998 and 2003, trained staff contacted 302 of these families by telephone. The male adolescents from these families, if indeed there were any, were asked about any possible, subjectively felt aggressiveness as a result of intrafamilial conflicts. They were specifically questioned about their situation at home, their relationships to their parents and siblings, and their conflict behavior. In addition, the first aggression screening by means of the State-Trait Anger Expression Inventory (STAXI; see below) was conducted by telephone. Nonresponders were those who could not be contacted, although up to 5 attempts were made to do so at different times of day, or who refused an interview.

The study's inclusion criteria were age between 14 and 16 years and bullying behavior of at least 6 months' duration. Exclusion criteria were psychotic illnesses, liability to prosecution, use of psychotropic medication and/or psychotherapy, or current use of narcotics. Sixty-nine male youths, who were assessed after the telephone interview as potentially meeting the inclusion criteria (Fig 1), were invited with their families to participate in a face-to-face interview and were tested with STAXI, the Inventory of Interpersonal Problems (IIP-D), and the SF-36 Health Survey (SF-36; see below).

• Download figure
• Open in new tab
• Download powerpoint

Fig 1.

Flow chart of patients' progress through the phases of the trial.

The necessary sample size was calculated for a type I error of 5% (z₁ = 1.96) and a power analysis of 80% (z₂ = 0.842), based on the mean value (m₁ = 19.5 and m₂ = 14.5) and SD (s₁ = 6.4 and s₂ = 5.3) for the Anger-Out scale of STAXI, which were obtained from a small pilot study. The formula is n (per group) = [(z₁ + z₂)² × (s₁² + s₂²)]/(m₁ − m₂).²⁵ This resulted in a group size of n = 44. Twenty-two boys were randomly selected for a family therapy program (FamTh-G), and 22 were selected for the control group (CG), using Excel tables with random numbers (Fig 1). Both randomizations were performed confidentially by the clinic administration.

All families appeared for the first therapy date (the FamTh-G [38 parents, 22 adolescent subjects, and 11 younger siblings]; the CG [39 parents, 22 adolescent subjects, and 12 younger siblings]). When the parents were separated and the distances involved were reasonable, we tried to invite both parents. The subjects were reminded about the appointment 3 to 5 times by telephone.

The boys in both groups were Germans from a small-town, rural population, were in good physical health, and had a comparable mean age (FamTh-G: 15.2 ± 0.5 years; CG: 14.9 ± 0.5 years), and comparably many of them lived together with both parents (FamTh-G: n = 15 [68.2%]; CG: n = 16 [72.7%]). The principal wage earners in the families were white-collar workers or civil servants (FamTh-G: n = 14 [63.6%]; CG: n = 16 [72.7%]) and blue-collar workers (FamTh-G: n = 8 [36.4%]; CG: n = 6 [27.3%]); a number of them were unemployed at the time of the pretherapeutic survey (FamTh-G: n = 5 [22.3%]; CG: n = 6 [27.3%]; at this point in time, the unemployment rate in Bavaria was calculated at ∼3.5%). Conduct disorder (FamTh-G: n = 11 [50.0%]; CG: n = 10 [45.4%]), oppositional defiant disorder (FamTh-G: n = 9 [40.9%]; CG: n = 9 [40.9%]), borderline personality disorder (FamTh-G: n = 4 [18.2%]; CG: n = 5 [22.7%]), bulimia (FamTh-G: n = 2 [9.1%]; CG: n = 2 [9.1%]), and attention-deficit/hyperactivity disorder (FamTh-G: n = 2 [9.1%]; CG: n = 1 [4.5%]) were diagnosed according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition. Most boys (65.9%) met the criteria for >1 disorder.

Assessment

At each examination, the Adolescents' Risky-Behavior Scale (ARBS), the STAXI, the IIP-D, and the SF-36 were used to assess anger, interpersonal relationships, and health-related quality of life. Siblings under the age of 14 were not tested but nevertheless took part in the family therapy or placebo intervention program (see below).

The ARBS is a procedure for assessing risky behavior in adolescents and consists of 7 scales: Drug Use (DU), Smoking (S), Binge Drinking (BD), Excessive Media Use (EMU), Sex Without a Condom (SW/OC), Sex While Using Drugs and Alcohol (SWDA), Sexual Disinhibition (SDI), and Index (I). Each scale is scored from 0 to 4, where 4 represents the strongest expression of the respective risky behavior. Thus, subjects can score between 0 and 28 points on the index. The validation of the ARBS (a test that we developed) has not been concluded definitively. For this reason, we are unable to provide any values or data from the literature.

The STAXI is a procedure for assessing anger and expression of anger (for male test subjects: Cronbach α = .76–0.89; retest correlation 8 weeks later = .55–.75) and consisting of 44 items that form 5 scales:

• State-Anger (S-A): subjective state of anger at the time of measurement

• Trait-Anger (T-A): readiness to react with anger (norm value: 18.1; SD: 5.34)

• Anger-In (AI): tendency to suppress anger (norm value: 16.0; SD: 4.04)

• Anger-Out (AO): tendency to direct anger outward (norm value: 13.0; SD: 4.02)

• Anger-Control (AC): tendency to keep anger under control (norm value: 22.4; SD: 5.29)

The values for S-A and T-A range from 10 to 40, and the others range from 8 to 32.²⁶

The IIP-D is primarily a questionnaire for individual status and change in clinical and personality psychology. The scales of the questionnaire are derived from theory and describe 8 subdimensions of the interpersonal circumplex: overly autocratic/dominant (PA), overly quarrelsome/competitive (BC), overly distant/cold (DE), overly introverted/socially avoiding (FG), overly subassertive/submissive (HI), overly exploitable/compliant (JK), overly nurturing/friendly (LM), and overly expressive/importunate (NO). With the IIP-D, one can obtain a complex picture of a person's interpersonal problems with relatively little effort. The consistency coefficients (Cronbach α) are ∼.60. The IIP-D is designed as a self-rating scale. The answers are coded into a rating scale on the questionnaire corresponding to 0 to 4. The raw values for the individual items then are added for each scale, and the sums yield the scales' raw values.²⁷

The SF-36²⁸ was developed to obtain individual reports from patients on their health-related quality of life, irrespective of their current state of health and age. It consists of a questionnaire with 36 items that are categorized according to several subject areas. The items record 8 dimensions of subjective health: physical functioning (PHFU), role limitations as a result of physical health (role-physical; ROPH), bodily pain (BOPA), general health perceptions (GEPE), vitality (VITA), social functioning (SOFU), role limitations as a result of emotional problems (role-emotional; ROEM), and mental health (PSYC). The items and scales of the SF-36 were calculated so that a higher score corresponds to a better state of health. A higher score in functionality, for instance, indicates better functionality in the subject, and a higher score on the pain scale means freedom from pain. The scales' raw values are converted into transformed scale values (T values). Reliability testing indicates an internal consistency ranging approximately from .77 to .93.²⁹

Procedure

The FamTh-G was treated with integrative family therapy for 6 months. Each treatment was conducted between November 1998 and September 2003. During the first 2 months, a 90-minute session took place once a week. Between the third and sixth months, therapy was conducted every 2 weeks.

The family therapy was not set up to address individual people but families exclusively, as communication systems. The focus was on communication, the family rules, and each individual member's behavioral latitude. The aggressive youths were seen as symptom bearers, eg, for the communication problems in the entire family's network of “playing rules.” In the course of therapy, another goal was to ascertain any family “games” that contributed to maintenance of the symptoms. The aim was to knock the tried and tested connections off balance, to change them, and to have families adopt new rules. Each therapeutic session was conducted by 2 therapists. Elements from systematic therapy, psychodynamic-oriented therapy, behavioral therapy, gestalt therapy, and psychodrama were included.

The CG was treated during the same period and with the same frequency as the FamTh-G, but with a placebo intervention. This intervention consisted of a structured and detailed survey of the psychological state of health, daily routine, and events. We rigorously checked that none of the above-described family therapeutic interventions took place. The trainer for this group was specially trained.

Testing was conducted by blinded staff every 2 weeks, independent from the therapy meetings. The boys were merely reminded 3 to 4 times about the next testing appointment. Three boys/families in the FamTh-G and 4 in the comparison group dropped out. They were absent >3 times for therapy and/or >3 times for assessment. Follow-up testing took place 12 months after the final family therapy session.

The STAXI, IIP, and SF-36 questionnaires were filled out in face-to-face interviews. Six trained interviewers were randomly assigned to each examination. The data were fed to the computer twice independently and automatically checked for deviations. A total of 2.8% of the entries were accordingly identified as erroneous and adjusted. The study was concluded according to plan.

Source of Funding and Ethical Considerations

The study was planned and performed in accordance with the Declaration of Helsinki and ethical laws pertaining to the medical professions. The design of this trial was approved by the clinic's “Ethikkomission” (the German equivalent of the Committee on Human Subjects). The study was conducted independent of any institutional influence and was not funded. Parents gave written informed consent.

Data Analysis

We used the statistical program SPSS, Version 11 (SPSS Inc Chicago, IL). Mann-Whitney U test and Fisher exact test were performed. We used SDs, differences in change between the 2 groups (DI) with their 95% confidence intervals (range of the mean differences for a certain parameter between both groups), probability (P), and rank correlation coefficients according to Spearman (R²) for reporting the treatment results according to the intention-to-treat principle.²⁵

Methodologic Limitations and Directions for Additional Research

This study had several methodologic limitations. First, the sample size (despite a valid power analysis) was relatively small. Second, the sample consisted only of male youths; whether these results could also be replicated with aggressive female youths is unknown. Third, the sample was composed of moderately aggressive individuals who had not engaged in criminal conduct and who still had the resources to enter therapy. Fourth, the length of this trial was only 6 months. That probably reduced the dropout rate, particularly in the CG. Sixth, the extent to which the study can be generalized is restricted by its relatively select population, conditioned by the hospital's small-town, rural setting. Seventh, we did not measure the family cohesion. Cohesion was simply estimated on the basis of actual participation in the therapy sessions. Additional studies are planned to shed light on the permanence of positive treatment effects such as these.