Int.J. Behav. Med. DOI 10.1007/s12529-014-9425-3
The Changeability and Predictive Value of Dysfunctional Cognitions in Cognitive Behavior Therapy for Chronic Tinnitus Isabell Conrad & Maria Kleinstäuber & Kristine Jasper & Wolfgang Hiller & Gerhard Andersson & Cornelia Weise
# International Society of Behavioral Medicine 2014
Abstract Background Multidimensional tinnitus models describe dysfunctional cognitions as a complicating factor in the process of tinnitus habituation. However, this concept has rarely been investigated in previous research. Purpose The present study investigated the effects of two cognitive-behavioral treatments on dysfunctional tinnitusrelated cognitions in patients with chronic tinnitus. Furthermore, dysfunctional cognitions were examined as possible predictors of the therapeutic effect on tinnitus distress. Method A total of 128 patients with chronic tinnitus were randomly assigned to either an Internet-delivered guided self-help treatment (Internet-based cognitive-behavioral therapy, ICBT), a conventional face-to-face group therapy (cognitive–behavioral group therapy, GCBT), or an active control group in the form of a web-based discussion forum (DF). To assess tinnitus-related dysfunctional thoughts, the Tinnitus Cognitions Scale (T-Cog) was used at pre- and postassessment, as well as at the 6- and 12-month follow-up.
I. Conrad : K. Jasper : W. Hiller Department of Clinical Psychology, Johannes Gutenberg-University of Mainz, Mainz, Germany M. Kleinstäuber (*) : C. Weise Division of Clinical Psychology and Psychotherapy, Department of Psychology, Philipps-University Marburg, Gutenbergstr. 18, 35032 Marburg, Germany e-mail: [email protected] G. Andersson Department of Clinical Neuroscience, Psychiatry Section, Karolinska Institutet, Stockholm, Sweden G. Andersson : C. Weise Department of Behavioural Sciences and Learning, Swedish Institute for Disability Research and Linnaeus Centre HEAD, Linköping University, Linköping, Sweden
Results Multivariate ANOVAs with post hoc tests revealed significant and comparable reductions of dysfunctional tinnitus-related cognitions for both treatments (GCBT and ICBT), which remained stable over a 6- and 12-month period. Negative correlations were found between the catastrophic subscale of the T-Cog and therapy outcome for ICBT, but not for GCBT. This means a higher degree of catastrophic thinking at baseline was associated with lower benefit from ICBT directly after the treatment. Hierarchical regression analysis confirmed catastrophizing as a predictor of poorer therapy outcome regarding emotional tinnitus distress in ICBT. No associations were detected in the follow-up assessments. Conclusion Both forms of CBT are successful in reducing dysfunctional tinnitus-related cognitions. Catastrophizing significantly predicted a less favorable outcome regarding emotional tinnitus distress in ICBT. Clinical implications of these results are described. Dysfunctional cognitions could be targeted more intensively in therapy and in future research on tinnitus. Keywords Chronic tinnitus . Dysfunctional cognitions . Cognitive-behavioral therapy . Internet-based therapy . Cognitive-behavioral group therapy . Self-help treatment
Introduction Tinnitus has been defined as a subjective perception of sound in the absence of an external acoustic stimulation [1]. Davis and El Refaie [2] estimated that tinnitus affects 10 to 15 % of the adult population, with about 3 % experiencing severe impairment in the form of sleep disturbances, difficulties with concentration, or negative effects on mood. Multidimensional biopsychological tinnitus models (e.g., the habituation model by Hallam et al. [3], and the neurophysiological tinnitus model by Jastreboff and Hazell [4]) attempt to explain impairment in coping with tinnitus. They emphasize cognitive appraisal
Int.J. Behav. Med.
processes as one of the key factors for a missing tinnitus habituation, along with a heightened cortical arousal and an unfavorable fixation of attention. Andersson and McKenna [5] also proposed dysfunctional beliefs and attitudes regarding the tinnitus as a necessary condition for the development of tinnitus annoyance. In their model, they differentiated between three cognitive aspects of tinnitus: (a) cognitive deficits (e.g., attention, working memory), (b) cognitive bias (e.g., selective information processing), and (c) conscious appraisal of tinnitus (e.g., beliefs and attitudes regarding the tinnitus). All three aspects are assumed to be important steps on the pathway from tinnitus generation to annoyance. It can be inferred from the models mentioned that therapeutic work on negative tinnitus-related thoughts should be an essential approach in the treatment of tinnitus. The importance of cognition is also reflected in evidence that cognitive–behavioral therapy (CBT), which targets the modification of dysfunctional cognitions, has been shown to be effective in reducing tinnitus distress (see review by Hesser et al. [6]). A meta-analysis on psychological treatments of tinnitus [7] confirmed a predominant effectiveness of CBT, compared to other treatment methods such as relaxation, hypnosis, biofeedback, problem-solving, or educational sessions. Although different tinnitus models and CBT are based on cognitions, research investigating the role of dysfunctional cognitions—especially the conscious cognitive appraisals of tinnitus—and their influenceability by CBT is still rare [8, 5]. Studies referring to these cognitions mostly target the development and validation of questionnaires [9–13]. There are only a few studies that have focused on tinnitusrelated thoughts and appraisals as a main construct. One of these studies was conducted by Cima et al. [14]. Catastrophic misinterpretations of tinnitus were found to be highly associated with tinnitus-specific fear and with an increased attention toward the tinnitus. Moreover, catastrophizing about tinnitus was related to lower ratings of quality of life. Although this association between catastrophizing and quality of life was fully mediated by tinnitusspecific fear, their findings suggest that catastrophic misinterpretations are a central factor in the maintenance of tinnitus distress. While Cima et al. [14] investigated the role of catastrophizing in chronic tinnitus, a current study by Weise et al. [8] concentrated on catastrophic thinking in recent onset tinnitus. The authors showed that a higher level of catastrophic thoughts was associated with higher tinnitus distress. Catastrophizing contributed significantly to the prediction of tinnitus distress and explained 30 % of the variance. Additionally, catastrophic thinking was found to be the strongest predictor of tinnitus distress compared with tinnitus loudness and sound sensitivity. Despite these interesting results, no conclusions about causality can be made due to the cross-sectional design of the study. To our knowledge, there are only two studies that have dealt specifically with tinnitus-related appraisals and cognitions in the context of therapy. The first study, which was
conducted by Hiller and Haerkötter [15], examined the effects of CBT for tinnitus on dysfunctional cognitions. Tinnitus patients who received a cognitive-behavioral group therapy program improved significantly with regard to dysfunctional cognitions, tinnitus distress, and depression. Medium to strong effects on dysfunctional tinnitus appraisals were found directly after treatment (pre–post effect size: d=.62) and at follow-up (6- and 18-month follow-up: d=.75 and d=1.03). These effects were lower than the effects on tinnitus distress (.89≤d≤1.25), but higher than those on depression (.43≤ d≤.69). Results of the study showed a satisfactory responsiveness of dysfunctional cognitions to CBT. The second longitudinal study dealing with the role of dysfunctional cognitions in the treatment of tinnitus was conducted by Kröner-Herwig et al. [16]. The authors examined patient characteristics as predictors of outcome in an outpatient group treatment of chronic tinnitus and hypothesized that a high level of catastrophic tinnitus-related cognitions would predict a less favorable therapy outcome. However, this hypothesis was not supported by the data as distinct dysfunctional cognitions at baseline were not found to be significantly associated with the therapeutic effect on tinnitus distress. Pointing out the necessity of further exploration, the authors found at least a small negative correlation in the follow-up data. Nevertheless, it should be mentioned that the study had a small sample size of 57 tinnitus patients and that the predictive value of catastrophizing was only investigated by means of correlational analyses. As far as we know, the study by Kröner-Herwig et al. [16] is the only one examining associations between negative cognitions and therapy outcome in chronic tinnitus. Further longitudinal exploration in larger samples is therefore needed, along with the inclusion of regression analyses. As mentioned above, there is also a lack of intervention studies that specifically analyze the efficacy of CBT in decreasing dysfunctional tinnitus-related thoughts. In order to illuminate the role of cognitions in the treatment of chronic tinnitus, two main research questions were generated. First, with regard to the findings of Hiller and Haerkötter [15], the present study investigated the particular effect of CBT on dysfunctional cognitions. We expected a significant change in dysfunctional tinnitus-related cognitions after treatment in two different CBT settings: (a) in a group therapy (cognitive-behavioral group therapy, GCBT) and (b) in an Internet-delivered guided self-help treatment (Internet-based cognitive-behavioral therapy, ICBT). Previous studies have supported the efficacy of Internet-based treatments for chronic tinnitus [17, 18] and have shown ICBT and conventional faceto-face GCBT to be equally effective in reducing tinnitus distress [19]. On the basis of these findings, it was predicted that both GCBT and ICBT would significantly reduce the level of dysfunctional tinnitus-related cognitions and that these improvements would be significantly higher than those of an active control group. Comparable moderate to high
Int.J. Behav. Med.
effect sizes as found by Hiller and Haerkötter [15] were expected. Furthermore, the effects were expected to be maintained at the 6- and 12-month follow-up. The second research question evaluated different types of dysfunctional cognitive appraisals of tinnitus (catastrophic and avoidance cognitions) as predictors of therapy outcome in the treatment of chronic tinnitus. Although catastrophizing did not predict therapy outcome in the study by KrönerHerwig et al. [16], negative correlations in the follow-up data might indicate a long-term effect. Additionally, strong associations were found between catastrophic misinterpretations and tinnitus distress [8], as well as quality of life in tinnitus sufferers [14]. Therefore, it was hypothesized that a high level of dysfunctional cognition at baseline would predict a smaller reduction of tinnitus distress during treatment, when controlling for different demographic and tinnitus-related variables.
Methods Participants A total of 128 participants were included in the study (Trial registration ID: NCT01205906, www.ClinicalTrials.gov), which aimed to compare an Internet-based self-help treatment (ICBT) with a group therapy (GCBT) for chronic tinnitus, and participation in an online discussion forum (DF). Participants were recruited in the area of Mainz, Germany. Inclusion criteria were (a) tinnitus duration of at least 6 months, (b) scoring 18 or above in the Tinnitus Handicap Inventory (THI, [11]) or scoring 8 or above in the Mini-Tinnitus Questionnaire (Mini-TQ, [20]), (c) medical examination by an ear, nose and throat physician, (d) no current psychological treatment for tinnitus, (e) Internet access, (f) ability to attend weekly group sessions in the Outpatient Clinic of the Johannes GutenbergUniversity of Mainz, (g) tinnitus is not caused by a general medical condition or otologic disease (e.g., active Meniere's Disease), (h) no major medical or psychiatric condition, and (i) age of at least 18 years. Participants were recruited through the Outpatient Clinic of the University of Mainz, the German Tinnitus Association, daily newspapers and information flyers, as well as by cooperating ear, nose and throat physicians. Demographic variables and tinnitus characteristics of the sample are shown in Table 1. There were no significant pretreatment differences between the three groups for demographic variables, tinnitus characteristics, or main outcome measures (see Table 1). Treatments Internet-Based Cognitive-Behavioral Therapy This guided self-help treatment, which is an adapted and modified version of a treatment manual by Kaldo and Andersson [21], was
exclusively provided via Internet over a period of 10 weeks. The treatment consists of 18 modules including information texts, worksheets, and detailed practice instructions for helpful coping strategies for tinnitus. Twelve modules on the topics of applied relaxation, positive imagery, attention shift, tinnitus exposure, cognitive restructuring, avoidance behavior, and relapse prevention were mandatory for all participants. Six optional modules (sound enrichment, noise sensitivity, tinnitus reframing, sleep management, concentration management, hearing tactics) could be additionally chosen. At the end of each treatment week, participants contacted the therapist via email to inform them of their progress with the modules and if they had encountered any problems. The therapist provided feedback, support, and recommendations on how to proceed. Cognitive-Behavioral Group Therapy This manualized program is based on CBT and has been described in detail in a study by Hiller and Haerkötter [15]. It consists of ten weekly group sessions of 90 min and includes the following components: education, relaxation techniques, cognitive restructuring, attentional processes in tinnitus perception, analysis of avoidance and illness behavior, tinnitus and the health care system, as well as relapse prevention. For each session, participants received written materials, exercises, and homework assignments, to enhance understanding and to transfer the new information into their daily routine. Experiences, problems, or questions were continually discussed between the participants and the therapist in every session. A more detailed overview of all topics and goals of both treatment approaches (ICBT and GCBT) of the current study is reported elsewhere [22]. Discussion Forum A tinnitus-specific Internet discussion forum was employed as an active control group. Patients who were assigned to this group received access to the DF for 10 weeks and could discuss different topics on tinnitus. A new discussion topic was presented every week (e.g., “Is tinnitus distress underestimated by non-sufferers?”). Postings were monitored by a therapist, but no therapeutic participation or intervention was given in order to avoid nonspecific effects. Procedure and Study Design Ethical approval was obtained by the Ethics Committee of the Department of Psychology at the University of Mainz. Information about the study was presented on a study website. After online registration on the study website and giving their informed consent, participants received a study code and a password via E-mail to access the online pre-assessment questionnaire. Participants who passed the questionnairerelated inclusion criteria received both a short telephone screening and an extensive face-to-face interview, where comorbid disorders and inclusion criteria were checked again.
Int.J. Behav. Med. Table 1 Baseline demographic and tinnitus-related characteristics for each intervention group
Female, n (%) Age (in years), M (SD) Educational level, n (%) Primary or secondary school A-level or academic degree Tinnitus duration (in months), M (SD) Hearing impairment, n (%) Tinnitus distress, M (SD) THI total Mini-TQ Dysfunctional cognitions, M (SD) T-Cog total T-Cog-TCT T-Cog-TAC Anxiety and depression, M (SD) HADS-A HADS-D
ICBT (n=41)
GCBT (n=43)
DF (n=44)
Test statistics of the baseline comparisons
16 (39.0) 51.32 (9.78)
19 (44.2) 50.23 (13.13)
16 (36.4) 52.09 (8.99)
χ2(2, N=128)=.57, p=.751 F(2, 125)=.33, p=.723 χ2(2, N=128)=.45, p=.797
14 (34.1) 27 (65.9) 110.85 (94.59) 28 (68.3)
17 (39.5) 26 (60.5) 100.21 (82.21) 31 (71.1)
18 (40.9) 26 (59.1) 95.41 (84.90) 34 (77.3)
F(2, 125)=.35, p=.708 χ2(2, N=128)=.87, p=.647
40.34 (17.64) 12.20 (4.58)
44.33 (19.17) 14.19 (4.51)
40.23 (20.54) 12.50 (4.83)
F(2, 125)=.63, p=.532 F(2, 125)=2.27, p=.108
20.32 (9.29) 13.71 (7.65) 6.61 (3.37)
23.58 (11.79) 16.74 (9.64) 6.84 (3.52)
20.77 (11.95) 14.34 (9.14) 6.43 (3.79)
F(2, 125)=1.08, p=.343 F(2, 125)=1.39, p=.254 F(2, 125)=.14, p=.869
7.41 (3.56) 5.95 (4.21)
7.79 (3.73) 5.98 (3.80)
8.00 (4.24) 6.43 (4.48)
F(2, 125)=.25, p=.780 F(2, 125)=.18, p=.834
ICBT Internet-based cognitive-behavioral therapy, GCBT cognitive-behavioral group therapy, DF discussion forum, THI Tinnitus Handicap Inventory, Mini-TQ Mini-Tinnitus Questionnaire, T-Cog-TCT/-TAC Tinnitus Cognitions Scale-Tinnitus-related Catastrophic Thinking/-Tinnitus-related Avoidance Cognitions, HADS-D/-A Hospital Anxiety and Depression Scale-Depression/-Anxiety
Following the face-to-face interview, the included participants were randomized to one of the three study arms: ICBT, GCBT, or DF. Subsequent to the 10 weeks of participation in DF, a repeated randomization of the DF participants to ICBT or GCBT was performed. The study was run in three blocks because the participants of this trial could not all be recruited at the same time. According to the CONSORT criteria [23] for assigning participants randomly to one of the three treatments in each block, the randomization sequence was generated by using the Web site www.randomization.com [24], with a 1:1:1 allocation and block sizes of 18, 37, and 73. The applied online service for randomization procedures uses the method of randomly permuted blocks. In order to implement allocation concealment, the generation of the randomization sequence and the random assignment of participants to one of the three treatments were conducted by an independent psychologist with no clinical involvement in the study. The random assignment of participants was done shortly before treatment started and after all participants of one block had had their initial diagnostic interview and their eligibility for the study had been confirmed. Figure 1 shows a CONSORT flowchart with the complete progress through the phases of the study. Data were collected at baseline (pre-assessment), at post-assessment, at 6-month follow-up, and at 12-month follow-up. The majority of participants received no additional treatment during the follow-up
period. At the 12-month follow-up, three participants of the GCBT (8.33 %, nFU12 =36) reported the use of additional medication (herbal medical products such as ginkgo biloba). None of the ICBT participants reported additional treatment within this 12-month period (0 %, nFU12 =28). Measures Tinnitus Cognitions Scale To assess dysfunctional tinnitusrelated cognitions, the German version of the T-Cog [15] was used. This self-report measure consists of 22 items with two subscales: tinnitus-related catastrophic thinking (TCT) and tinnitus-related avoidance cognitions (TAC). In the present study, internal consistency was good for the total scale (Cronbach’s α=.90) and for TCT (α=.90; 16 items), as well as acceptable for TAC (α=.72; 6 items). Tinnitus Handicap Inventory Tinnitus distress was measured by the THI [11], which was used in its German version [25]. This self-report questionnaire consists of 25 items on three subscales: a functional subscale (12 items), an emotional subscale (8 items), and a catastrophic response subscale (5 items). The total scale showed an excellent Cronbach’s alpha of .92 in the current study. The THI subscales obtained Cronbach’s alpha values of .86 (emotional subscale), .85 (functional subscale), and .64 (catastrophic response subscale).
Int.J. Behav. Med. Enrollment
Assessed for eligibility (n = 174) Excluded (n = 41) Declined to participate (n = 5) Lost interest (n = 4) Too much effort to participate (n = 1)
1. Pre-assessment online questionnaire 2. Telephone screening 3. Face-to-face interview
THI < 18 or Mini-TQ < 8 (n = 20) Not able to attend group sessions (n = 11) Lack of time for participation (n = 4) No Internet access (n = 3) Not meeting other inclusion criteria (n = 3)
Randomized (n = 128)
Allocation
Postassessment
FU6
FU12
Allocated to ICBT (n = 41)
Allocated to GCBT (n = 43)
Allocated to DF (n = 44)
Lost before treatment (n = 0) Started treatment (n = 41)
Lost before treatment (n = 2) Started treatment (n = 41)
Lost before DF (n = 0) Started DF (n = 44)
Lost to post-assessment (n = 3) Completed post-assessment (n = 38)
Lost to post-assessment (n = 3) Completed post-assessment (n = 40)
Lost to post-assessment (n = 1) Completed post-assessment (n = 43)
Lost to FU6 (n = 4) Completed FU6-assessment (n = 34)
Lost to FU6 (n = 3) Completed FU6-assessment (n = 37)
Randomized (n = 44)
Lost to FU12 (n = 6) Completed FU12-assessment (n = 28)
Lost to FU12 (n = 1) Completed FU12-assessment (n = 36)
Allocated to ICBT (n = 22)
Allocated to GCBT (n = 22)
Fig. 1 CONSORT flowchart of the progress through the phases of the trial (ICBT =Internet-based cognitive-behavioral therapy, GCBT =cognitivebehavioral group therapy, DF =discussion forum, FU6=6-month follow-up, FU12=12-month follow-up)
Hospital Anxiety and Depression Scale The Hospital Anxiety and Depression Scale (HADS) [26] consists of the two subscales depression and anxiety, each including seven items. The German version of the HADS [27] was administered. In the present study, internal consistency was good with a Cronbach’s alpha of .89 for the depression subscale and .83 for the anxiety subscale. In addition, demographic variables, tinnitus characteristics (e.g., tinnitus duration, noise sensitivity, or hearing impairment), and self-reported additional treatments during the follow-up period were assessed. All questionnaires were filled out as online versions, which yield comparable results to paper-and-pencil administration [28, 29]. Statistical Analyses Change of Dysfunctional Tinnitus-Related Cognitions During Therapy In order to examine the pre–post effects of the treatments on dysfunctional cognitions, we first conducted a 3×2 multivariate ANOVA using the within-subject factor time (pre, post) and the between-subjects factor group (ICBT, GCBT, DF). In case of significant main or interaction effects in the initial omnibus test, univariate and pairwise comparisons with Bonferroni correction served as post hoc tests. In order to investigate whether the effects of both treatments (ICBT and GCBT) were stable over time, a separate 2×4 multivariate ANOVA with the within-subject factor time (pre, post-, 6- and 12-month follow-up) and the betweensubjects factor group (ICBT and GCBT) was applied. The
two subscales of the T-Cog (TCT and TAC) served as dependent variables in these analyses. In order to quantify the effects of ICBT and GCBT, as well as their long-term stability on dysfunctional appraisals, effect sizes were calculated using Cohen’s d. All of the reported analyses with regard to the change of dysfunctional cognitions during therapy are based on imputed data sets. In order to replace single missing values, the multiple imputation procedure offered by IBM SPSS Statistics 20 was used. The procedure produced five data sets using the monotone multiple imputation algorithm [30]. These five imputed data sets were analyzed by using standard procedures used for complete data and then by combining the results across these analyses. Additionally, completer analyses were conducted. In the following, results of the completer analyses are only reported if there were deviations from the results based on imputed data. Prediction of Therapy Outcome by Dysfunctional TinnitusRelated Cognitions For the second research question, only complete data sets were used. In order to evaluate the role of dysfunctional cognitions as a predictor variable of therapy outcome in terms of a reduction of tinnitus distress, residual gain scores (RGS, [31]) of the THI served as a measure of therapy outcome. The THI-RGS were computed for the pre– post difference and for the differences between baseline and both follow-ups, using the z score method as described by Steketee and Chambless [32]. Due to substantial overlap between the items of the T-Cog and the catastrophic subscale of the THI, only the RGSs of THI-FS and THI-ES were used
Int.J. Behav. Med.
for the following analyses. First, Pearson product-moment correlation coefficients were calculated between the THIRGSs and the baseline scores of the T-Cog subscales. Secondly, in the case of a significant correlation coefficient, hierarchical multiple regression analyses were conducted in order to examine the unique contribution of initial dysfunctional cognitions to the prediction of therapy outcome (THIRGS) beyond the different control variables. In the first step of this analysis, control variables were entered on the basis of correlational analyses. The variables that were most highly correlated with the THI-RGS were chosen out of a selection of different demographic and tinnitus-related variables (age, sex, educational level, hearing impairment, tinnitus duration, anxiety, and depression subscale of the HADS at baseline). Anxiety and depression were examined as control variables because of their strong association with tinnitus distress, which has also been empirically evidenced in previous research [33–35]. Inclusion criterion for the control variables was a correlation of |r|≥.20 with the THI-RGS. The variable “education” was entered as a dummy variable, where A level or academic degree indicated higher education. Secondary school indicated lower education. In the second step of the hierarchical regression analysis, the baseline value of the corresponding subscale of the T-Cog was added. Correlational analyses and—if necessary—regression analyses were performed for the complete intervention sample (ICBT + GCBT), as well as separately for each intervention group. All reported statistical analyses were performed with IBM SPSS Statistics 20.
Results Change of Dysfunctional Tinnitus-Related Cognitions During CBT Pre–post Effects of ICBT, GCBT, and DF The conducted 3×2 multivariate ANOVA revealed a significant main effect of time for T-Cog-TCT, F(1, 125)=50.96, p
The Changeability and Predictive Value of Dysfunctional Cognitions in Cognitive Behavior Therapy for Chronic Tinnitus Isabell Conrad & Maria Kleinstäuber & Kristine Jasper & Wolfgang Hiller & Gerhard Andersson & Cornelia Weise
# International Society of Behavioral Medicine 2014
Abstract Background Multidimensional tinnitus models describe dysfunctional cognitions as a complicating factor in the process of tinnitus habituation. However, this concept has rarely been investigated in previous research. Purpose The present study investigated the effects of two cognitive-behavioral treatments on dysfunctional tinnitusrelated cognitions in patients with chronic tinnitus. Furthermore, dysfunctional cognitions were examined as possible predictors of the therapeutic effect on tinnitus distress. Method A total of 128 patients with chronic tinnitus were randomly assigned to either an Internet-delivered guided self-help treatment (Internet-based cognitive-behavioral therapy, ICBT), a conventional face-to-face group therapy (cognitive–behavioral group therapy, GCBT), or an active control group in the form of a web-based discussion forum (DF). To assess tinnitus-related dysfunctional thoughts, the Tinnitus Cognitions Scale (T-Cog) was used at pre- and postassessment, as well as at the 6- and 12-month follow-up.
I. Conrad : K. Jasper : W. Hiller Department of Clinical Psychology, Johannes Gutenberg-University of Mainz, Mainz, Germany M. Kleinstäuber (*) : C. Weise Division of Clinical Psychology and Psychotherapy, Department of Psychology, Philipps-University Marburg, Gutenbergstr. 18, 35032 Marburg, Germany e-mail: [email protected] G. Andersson Department of Clinical Neuroscience, Psychiatry Section, Karolinska Institutet, Stockholm, Sweden G. Andersson : C. Weise Department of Behavioural Sciences and Learning, Swedish Institute for Disability Research and Linnaeus Centre HEAD, Linköping University, Linköping, Sweden
Results Multivariate ANOVAs with post hoc tests revealed significant and comparable reductions of dysfunctional tinnitus-related cognitions for both treatments (GCBT and ICBT), which remained stable over a 6- and 12-month period. Negative correlations were found between the catastrophic subscale of the T-Cog and therapy outcome for ICBT, but not for GCBT. This means a higher degree of catastrophic thinking at baseline was associated with lower benefit from ICBT directly after the treatment. Hierarchical regression analysis confirmed catastrophizing as a predictor of poorer therapy outcome regarding emotional tinnitus distress in ICBT. No associations were detected in the follow-up assessments. Conclusion Both forms of CBT are successful in reducing dysfunctional tinnitus-related cognitions. Catastrophizing significantly predicted a less favorable outcome regarding emotional tinnitus distress in ICBT. Clinical implications of these results are described. Dysfunctional cognitions could be targeted more intensively in therapy and in future research on tinnitus. Keywords Chronic tinnitus . Dysfunctional cognitions . Cognitive-behavioral therapy . Internet-based therapy . Cognitive-behavioral group therapy . Self-help treatment
Introduction Tinnitus has been defined as a subjective perception of sound in the absence of an external acoustic stimulation [1]. Davis and El Refaie [2] estimated that tinnitus affects 10 to 15 % of the adult population, with about 3 % experiencing severe impairment in the form of sleep disturbances, difficulties with concentration, or negative effects on mood. Multidimensional biopsychological tinnitus models (e.g., the habituation model by Hallam et al. [3], and the neurophysiological tinnitus model by Jastreboff and Hazell [4]) attempt to explain impairment in coping with tinnitus. They emphasize cognitive appraisal
Int.J. Behav. Med.
processes as one of the key factors for a missing tinnitus habituation, along with a heightened cortical arousal and an unfavorable fixation of attention. Andersson and McKenna [5] also proposed dysfunctional beliefs and attitudes regarding the tinnitus as a necessary condition for the development of tinnitus annoyance. In their model, they differentiated between three cognitive aspects of tinnitus: (a) cognitive deficits (e.g., attention, working memory), (b) cognitive bias (e.g., selective information processing), and (c) conscious appraisal of tinnitus (e.g., beliefs and attitudes regarding the tinnitus). All three aspects are assumed to be important steps on the pathway from tinnitus generation to annoyance. It can be inferred from the models mentioned that therapeutic work on negative tinnitus-related thoughts should be an essential approach in the treatment of tinnitus. The importance of cognition is also reflected in evidence that cognitive–behavioral therapy (CBT), which targets the modification of dysfunctional cognitions, has been shown to be effective in reducing tinnitus distress (see review by Hesser et al. [6]). A meta-analysis on psychological treatments of tinnitus [7] confirmed a predominant effectiveness of CBT, compared to other treatment methods such as relaxation, hypnosis, biofeedback, problem-solving, or educational sessions. Although different tinnitus models and CBT are based on cognitions, research investigating the role of dysfunctional cognitions—especially the conscious cognitive appraisals of tinnitus—and their influenceability by CBT is still rare [8, 5]. Studies referring to these cognitions mostly target the development and validation of questionnaires [9–13]. There are only a few studies that have focused on tinnitusrelated thoughts and appraisals as a main construct. One of these studies was conducted by Cima et al. [14]. Catastrophic misinterpretations of tinnitus were found to be highly associated with tinnitus-specific fear and with an increased attention toward the tinnitus. Moreover, catastrophizing about tinnitus was related to lower ratings of quality of life. Although this association between catastrophizing and quality of life was fully mediated by tinnitusspecific fear, their findings suggest that catastrophic misinterpretations are a central factor in the maintenance of tinnitus distress. While Cima et al. [14] investigated the role of catastrophizing in chronic tinnitus, a current study by Weise et al. [8] concentrated on catastrophic thinking in recent onset tinnitus. The authors showed that a higher level of catastrophic thoughts was associated with higher tinnitus distress. Catastrophizing contributed significantly to the prediction of tinnitus distress and explained 30 % of the variance. Additionally, catastrophic thinking was found to be the strongest predictor of tinnitus distress compared with tinnitus loudness and sound sensitivity. Despite these interesting results, no conclusions about causality can be made due to the cross-sectional design of the study. To our knowledge, there are only two studies that have dealt specifically with tinnitus-related appraisals and cognitions in the context of therapy. The first study, which was
conducted by Hiller and Haerkötter [15], examined the effects of CBT for tinnitus on dysfunctional cognitions. Tinnitus patients who received a cognitive-behavioral group therapy program improved significantly with regard to dysfunctional cognitions, tinnitus distress, and depression. Medium to strong effects on dysfunctional tinnitus appraisals were found directly after treatment (pre–post effect size: d=.62) and at follow-up (6- and 18-month follow-up: d=.75 and d=1.03). These effects were lower than the effects on tinnitus distress (.89≤d≤1.25), but higher than those on depression (.43≤ d≤.69). Results of the study showed a satisfactory responsiveness of dysfunctional cognitions to CBT. The second longitudinal study dealing with the role of dysfunctional cognitions in the treatment of tinnitus was conducted by Kröner-Herwig et al. [16]. The authors examined patient characteristics as predictors of outcome in an outpatient group treatment of chronic tinnitus and hypothesized that a high level of catastrophic tinnitus-related cognitions would predict a less favorable therapy outcome. However, this hypothesis was not supported by the data as distinct dysfunctional cognitions at baseline were not found to be significantly associated with the therapeutic effect on tinnitus distress. Pointing out the necessity of further exploration, the authors found at least a small negative correlation in the follow-up data. Nevertheless, it should be mentioned that the study had a small sample size of 57 tinnitus patients and that the predictive value of catastrophizing was only investigated by means of correlational analyses. As far as we know, the study by Kröner-Herwig et al. [16] is the only one examining associations between negative cognitions and therapy outcome in chronic tinnitus. Further longitudinal exploration in larger samples is therefore needed, along with the inclusion of regression analyses. As mentioned above, there is also a lack of intervention studies that specifically analyze the efficacy of CBT in decreasing dysfunctional tinnitus-related thoughts. In order to illuminate the role of cognitions in the treatment of chronic tinnitus, two main research questions were generated. First, with regard to the findings of Hiller and Haerkötter [15], the present study investigated the particular effect of CBT on dysfunctional cognitions. We expected a significant change in dysfunctional tinnitus-related cognitions after treatment in two different CBT settings: (a) in a group therapy (cognitive-behavioral group therapy, GCBT) and (b) in an Internet-delivered guided self-help treatment (Internet-based cognitive-behavioral therapy, ICBT). Previous studies have supported the efficacy of Internet-based treatments for chronic tinnitus [17, 18] and have shown ICBT and conventional faceto-face GCBT to be equally effective in reducing tinnitus distress [19]. On the basis of these findings, it was predicted that both GCBT and ICBT would significantly reduce the level of dysfunctional tinnitus-related cognitions and that these improvements would be significantly higher than those of an active control group. Comparable moderate to high
Int.J. Behav. Med.
effect sizes as found by Hiller and Haerkötter [15] were expected. Furthermore, the effects were expected to be maintained at the 6- and 12-month follow-up. The second research question evaluated different types of dysfunctional cognitive appraisals of tinnitus (catastrophic and avoidance cognitions) as predictors of therapy outcome in the treatment of chronic tinnitus. Although catastrophizing did not predict therapy outcome in the study by KrönerHerwig et al. [16], negative correlations in the follow-up data might indicate a long-term effect. Additionally, strong associations were found between catastrophic misinterpretations and tinnitus distress [8], as well as quality of life in tinnitus sufferers [14]. Therefore, it was hypothesized that a high level of dysfunctional cognition at baseline would predict a smaller reduction of tinnitus distress during treatment, when controlling for different demographic and tinnitus-related variables.
Methods Participants A total of 128 participants were included in the study (Trial registration ID: NCT01205906, www.ClinicalTrials.gov), which aimed to compare an Internet-based self-help treatment (ICBT) with a group therapy (GCBT) for chronic tinnitus, and participation in an online discussion forum (DF). Participants were recruited in the area of Mainz, Germany. Inclusion criteria were (a) tinnitus duration of at least 6 months, (b) scoring 18 or above in the Tinnitus Handicap Inventory (THI, [11]) or scoring 8 or above in the Mini-Tinnitus Questionnaire (Mini-TQ, [20]), (c) medical examination by an ear, nose and throat physician, (d) no current psychological treatment for tinnitus, (e) Internet access, (f) ability to attend weekly group sessions in the Outpatient Clinic of the Johannes GutenbergUniversity of Mainz, (g) tinnitus is not caused by a general medical condition or otologic disease (e.g., active Meniere's Disease), (h) no major medical or psychiatric condition, and (i) age of at least 18 years. Participants were recruited through the Outpatient Clinic of the University of Mainz, the German Tinnitus Association, daily newspapers and information flyers, as well as by cooperating ear, nose and throat physicians. Demographic variables and tinnitus characteristics of the sample are shown in Table 1. There were no significant pretreatment differences between the three groups for demographic variables, tinnitus characteristics, or main outcome measures (see Table 1). Treatments Internet-Based Cognitive-Behavioral Therapy This guided self-help treatment, which is an adapted and modified version of a treatment manual by Kaldo and Andersson [21], was
exclusively provided via Internet over a period of 10 weeks. The treatment consists of 18 modules including information texts, worksheets, and detailed practice instructions for helpful coping strategies for tinnitus. Twelve modules on the topics of applied relaxation, positive imagery, attention shift, tinnitus exposure, cognitive restructuring, avoidance behavior, and relapse prevention were mandatory for all participants. Six optional modules (sound enrichment, noise sensitivity, tinnitus reframing, sleep management, concentration management, hearing tactics) could be additionally chosen. At the end of each treatment week, participants contacted the therapist via email to inform them of their progress with the modules and if they had encountered any problems. The therapist provided feedback, support, and recommendations on how to proceed. Cognitive-Behavioral Group Therapy This manualized program is based on CBT and has been described in detail in a study by Hiller and Haerkötter [15]. It consists of ten weekly group sessions of 90 min and includes the following components: education, relaxation techniques, cognitive restructuring, attentional processes in tinnitus perception, analysis of avoidance and illness behavior, tinnitus and the health care system, as well as relapse prevention. For each session, participants received written materials, exercises, and homework assignments, to enhance understanding and to transfer the new information into their daily routine. Experiences, problems, or questions were continually discussed between the participants and the therapist in every session. A more detailed overview of all topics and goals of both treatment approaches (ICBT and GCBT) of the current study is reported elsewhere [22]. Discussion Forum A tinnitus-specific Internet discussion forum was employed as an active control group. Patients who were assigned to this group received access to the DF for 10 weeks and could discuss different topics on tinnitus. A new discussion topic was presented every week (e.g., “Is tinnitus distress underestimated by non-sufferers?”). Postings were monitored by a therapist, but no therapeutic participation or intervention was given in order to avoid nonspecific effects. Procedure and Study Design Ethical approval was obtained by the Ethics Committee of the Department of Psychology at the University of Mainz. Information about the study was presented on a study website. After online registration on the study website and giving their informed consent, participants received a study code and a password via E-mail to access the online pre-assessment questionnaire. Participants who passed the questionnairerelated inclusion criteria received both a short telephone screening and an extensive face-to-face interview, where comorbid disorders and inclusion criteria were checked again.
Int.J. Behav. Med. Table 1 Baseline demographic and tinnitus-related characteristics for each intervention group
Female, n (%) Age (in years), M (SD) Educational level, n (%) Primary or secondary school A-level or academic degree Tinnitus duration (in months), M (SD) Hearing impairment, n (%) Tinnitus distress, M (SD) THI total Mini-TQ Dysfunctional cognitions, M (SD) T-Cog total T-Cog-TCT T-Cog-TAC Anxiety and depression, M (SD) HADS-A HADS-D
ICBT (n=41)
GCBT (n=43)
DF (n=44)
Test statistics of the baseline comparisons
16 (39.0) 51.32 (9.78)
19 (44.2) 50.23 (13.13)
16 (36.4) 52.09 (8.99)
χ2(2, N=128)=.57, p=.751 F(2, 125)=.33, p=.723 χ2(2, N=128)=.45, p=.797
14 (34.1) 27 (65.9) 110.85 (94.59) 28 (68.3)
17 (39.5) 26 (60.5) 100.21 (82.21) 31 (71.1)
18 (40.9) 26 (59.1) 95.41 (84.90) 34 (77.3)
F(2, 125)=.35, p=.708 χ2(2, N=128)=.87, p=.647
40.34 (17.64) 12.20 (4.58)
44.33 (19.17) 14.19 (4.51)
40.23 (20.54) 12.50 (4.83)
F(2, 125)=.63, p=.532 F(2, 125)=2.27, p=.108
20.32 (9.29) 13.71 (7.65) 6.61 (3.37)
23.58 (11.79) 16.74 (9.64) 6.84 (3.52)
20.77 (11.95) 14.34 (9.14) 6.43 (3.79)
F(2, 125)=1.08, p=.343 F(2, 125)=1.39, p=.254 F(2, 125)=.14, p=.869
7.41 (3.56) 5.95 (4.21)
7.79 (3.73) 5.98 (3.80)
8.00 (4.24) 6.43 (4.48)
F(2, 125)=.25, p=.780 F(2, 125)=.18, p=.834
ICBT Internet-based cognitive-behavioral therapy, GCBT cognitive-behavioral group therapy, DF discussion forum, THI Tinnitus Handicap Inventory, Mini-TQ Mini-Tinnitus Questionnaire, T-Cog-TCT/-TAC Tinnitus Cognitions Scale-Tinnitus-related Catastrophic Thinking/-Tinnitus-related Avoidance Cognitions, HADS-D/-A Hospital Anxiety and Depression Scale-Depression/-Anxiety
Following the face-to-face interview, the included participants were randomized to one of the three study arms: ICBT, GCBT, or DF. Subsequent to the 10 weeks of participation in DF, a repeated randomization of the DF participants to ICBT or GCBT was performed. The study was run in three blocks because the participants of this trial could not all be recruited at the same time. According to the CONSORT criteria [23] for assigning participants randomly to one of the three treatments in each block, the randomization sequence was generated by using the Web site www.randomization.com [24], with a 1:1:1 allocation and block sizes of 18, 37, and 73. The applied online service for randomization procedures uses the method of randomly permuted blocks. In order to implement allocation concealment, the generation of the randomization sequence and the random assignment of participants to one of the three treatments were conducted by an independent psychologist with no clinical involvement in the study. The random assignment of participants was done shortly before treatment started and after all participants of one block had had their initial diagnostic interview and their eligibility for the study had been confirmed. Figure 1 shows a CONSORT flowchart with the complete progress through the phases of the study. Data were collected at baseline (pre-assessment), at post-assessment, at 6-month follow-up, and at 12-month follow-up. The majority of participants received no additional treatment during the follow-up
period. At the 12-month follow-up, three participants of the GCBT (8.33 %, nFU12 =36) reported the use of additional medication (herbal medical products such as ginkgo biloba). None of the ICBT participants reported additional treatment within this 12-month period (0 %, nFU12 =28). Measures Tinnitus Cognitions Scale To assess dysfunctional tinnitusrelated cognitions, the German version of the T-Cog [15] was used. This self-report measure consists of 22 items with two subscales: tinnitus-related catastrophic thinking (TCT) and tinnitus-related avoidance cognitions (TAC). In the present study, internal consistency was good for the total scale (Cronbach’s α=.90) and for TCT (α=.90; 16 items), as well as acceptable for TAC (α=.72; 6 items). Tinnitus Handicap Inventory Tinnitus distress was measured by the THI [11], which was used in its German version [25]. This self-report questionnaire consists of 25 items on three subscales: a functional subscale (12 items), an emotional subscale (8 items), and a catastrophic response subscale (5 items). The total scale showed an excellent Cronbach’s alpha of .92 in the current study. The THI subscales obtained Cronbach’s alpha values of .86 (emotional subscale), .85 (functional subscale), and .64 (catastrophic response subscale).
Int.J. Behav. Med. Enrollment
Assessed for eligibility (n = 174) Excluded (n = 41) Declined to participate (n = 5) Lost interest (n = 4) Too much effort to participate (n = 1)
1. Pre-assessment online questionnaire 2. Telephone screening 3. Face-to-face interview
THI < 18 or Mini-TQ < 8 (n = 20) Not able to attend group sessions (n = 11) Lack of time for participation (n = 4) No Internet access (n = 3) Not meeting other inclusion criteria (n = 3)
Randomized (n = 128)
Allocation
Postassessment
FU6
FU12
Allocated to ICBT (n = 41)
Allocated to GCBT (n = 43)
Allocated to DF (n = 44)
Lost before treatment (n = 0) Started treatment (n = 41)
Lost before treatment (n = 2) Started treatment (n = 41)
Lost before DF (n = 0) Started DF (n = 44)
Lost to post-assessment (n = 3) Completed post-assessment (n = 38)
Lost to post-assessment (n = 3) Completed post-assessment (n = 40)
Lost to post-assessment (n = 1) Completed post-assessment (n = 43)
Lost to FU6 (n = 4) Completed FU6-assessment (n = 34)
Lost to FU6 (n = 3) Completed FU6-assessment (n = 37)
Randomized (n = 44)
Lost to FU12 (n = 6) Completed FU12-assessment (n = 28)
Lost to FU12 (n = 1) Completed FU12-assessment (n = 36)
Allocated to ICBT (n = 22)
Allocated to GCBT (n = 22)
Fig. 1 CONSORT flowchart of the progress through the phases of the trial (ICBT =Internet-based cognitive-behavioral therapy, GCBT =cognitivebehavioral group therapy, DF =discussion forum, FU6=6-month follow-up, FU12=12-month follow-up)
Hospital Anxiety and Depression Scale The Hospital Anxiety and Depression Scale (HADS) [26] consists of the two subscales depression and anxiety, each including seven items. The German version of the HADS [27] was administered. In the present study, internal consistency was good with a Cronbach’s alpha of .89 for the depression subscale and .83 for the anxiety subscale. In addition, demographic variables, tinnitus characteristics (e.g., tinnitus duration, noise sensitivity, or hearing impairment), and self-reported additional treatments during the follow-up period were assessed. All questionnaires were filled out as online versions, which yield comparable results to paper-and-pencil administration [28, 29]. Statistical Analyses Change of Dysfunctional Tinnitus-Related Cognitions During Therapy In order to examine the pre–post effects of the treatments on dysfunctional cognitions, we first conducted a 3×2 multivariate ANOVA using the within-subject factor time (pre, post) and the between-subjects factor group (ICBT, GCBT, DF). In case of significant main or interaction effects in the initial omnibus test, univariate and pairwise comparisons with Bonferroni correction served as post hoc tests. In order to investigate whether the effects of both treatments (ICBT and GCBT) were stable over time, a separate 2×4 multivariate ANOVA with the within-subject factor time (pre, post-, 6- and 12-month follow-up) and the betweensubjects factor group (ICBT and GCBT) was applied. The
two subscales of the T-Cog (TCT and TAC) served as dependent variables in these analyses. In order to quantify the effects of ICBT and GCBT, as well as their long-term stability on dysfunctional appraisals, effect sizes were calculated using Cohen’s d. All of the reported analyses with regard to the change of dysfunctional cognitions during therapy are based on imputed data sets. In order to replace single missing values, the multiple imputation procedure offered by IBM SPSS Statistics 20 was used. The procedure produced five data sets using the monotone multiple imputation algorithm [30]. These five imputed data sets were analyzed by using standard procedures used for complete data and then by combining the results across these analyses. Additionally, completer analyses were conducted. In the following, results of the completer analyses are only reported if there were deviations from the results based on imputed data. Prediction of Therapy Outcome by Dysfunctional TinnitusRelated Cognitions For the second research question, only complete data sets were used. In order to evaluate the role of dysfunctional cognitions as a predictor variable of therapy outcome in terms of a reduction of tinnitus distress, residual gain scores (RGS, [31]) of the THI served as a measure of therapy outcome. The THI-RGS were computed for the pre– post difference and for the differences between baseline and both follow-ups, using the z score method as described by Steketee and Chambless [32]. Due to substantial overlap between the items of the T-Cog and the catastrophic subscale of the THI, only the RGSs of THI-FS and THI-ES were used
Int.J. Behav. Med.
for the following analyses. First, Pearson product-moment correlation coefficients were calculated between the THIRGSs and the baseline scores of the T-Cog subscales. Secondly, in the case of a significant correlation coefficient, hierarchical multiple regression analyses were conducted in order to examine the unique contribution of initial dysfunctional cognitions to the prediction of therapy outcome (THIRGS) beyond the different control variables. In the first step of this analysis, control variables were entered on the basis of correlational analyses. The variables that were most highly correlated with the THI-RGS were chosen out of a selection of different demographic and tinnitus-related variables (age, sex, educational level, hearing impairment, tinnitus duration, anxiety, and depression subscale of the HADS at baseline). Anxiety and depression were examined as control variables because of their strong association with tinnitus distress, which has also been empirically evidenced in previous research [33–35]. Inclusion criterion for the control variables was a correlation of |r|≥.20 with the THI-RGS. The variable “education” was entered as a dummy variable, where A level or academic degree indicated higher education. Secondary school indicated lower education. In the second step of the hierarchical regression analysis, the baseline value of the corresponding subscale of the T-Cog was added. Correlational analyses and—if necessary—regression analyses were performed for the complete intervention sample (ICBT + GCBT), as well as separately for each intervention group. All reported statistical analyses were performed with IBM SPSS Statistics 20.
Results Change of Dysfunctional Tinnitus-Related Cognitions During CBT Pre–post Effects of ICBT, GCBT, and DF The conducted 3×2 multivariate ANOVA revealed a significant main effect of time for T-Cog-TCT, F(1, 125)=50.96, p
