3.1.2. Study Descriptions
Aboussouan et al. [
123] conducted a non-randomized longitudinal study to examine treatment outcomes in women suffering from chronic pelvic pain (CPP) after a 3–4 week Interdisciplinary Chronic Pain Rehabilitation Program (ICPRP), which included medication management, psychotherapy (individual, family and group), psychoeducation, physical and occupational therapy, weaning from medications and optional monthly aftercare. Fifty-eight women with CPP were age-matched with 58 women with non-pelvic chronic pain (NPCP) as controls. Primary outcomes were operationalized by pain severity (NRS-11; Numeric Rating Scale), depressivity (DASS-21; Depression, Anxiety, and Stress Scale), alexithymia (TAS-20) and impairment in sexual functioning (PDI; Sexual Behavior subscale of Pain Disability Index). The authors conclude that generalized interdisciplinary pain management programs are beneficial for CPP patients and NPCP patients by improvements in impairment in sexual functioning, depressivity and alexithymia. Even though changes within depressivity, alexithymia, and pain were independently associated with improvements in sexual functioning in CPP patients, the authors point out that further research is needed to reveal relationships between these comorbid conditions and sexual functioning within CPP patients. Thereby, they ask for further longitudinal research assessing more time points.
Melin et al. [
71] conducted a non-randomized longitudinal study to examine treatment outcomes in patients suffering from chronic benign pain after an Affect School consisting of eight weekly group therapy interventions—with a special theme for every session—and 10 individual sessions of Script Analysis (SA) afterwards to investigate benefits and harms caused by this treatment program. Group sessions focused on recall of occasions when specific affects occurred and how this affect was sensed in mind and body. Fifty-nine patients participated within this study. Primary outcomes were operationalized by alexithymia (TAS-20), anxiousness (HADS-A), depressivity (HADS-D), pain severity (VAS; Visual Analogue Scale), health-related quality of life (EQoL; European Quality of Life health barometer), and stress symptoms (SCI-93; Stress and Crisis Inventory-93). Thirty-six percent of participants matched criteria for alexithymia (TAS-20 ≤ 61). Due to the conducted treatment, which focused on identifying and describing affects and emotions, the authors explain the improvements within DIF and DDF and discuss EOT as a more trait-prone factor. Non-improvements within pain severity were attributed to (a) assessment, (b) emotional, and (c) neurological factors.
Saariaho et al. [
124] conducted a non-randomized longitudinal study to examine changes in alexithymia, depressivity, pain intensity and pain disability in patients suffering from chronic non-malignant pain for at least three months. The authors assessed primary outcomes at baseline and at eight-year follow-up. They did not conduct their own treatment approaches. Undergone treatment interventions were assessed retrospectively according to patient’s reports of undergone treatments which were based on a biomedical concept. Out of the initially assessed 271 patients, 83 patients participated at both measurement timepoints. Outcomes were operationalized as alexithymia (TAS-20), depressivity (BDI-II; Beck’s Depression Inventory), pain intensity (VAS; Visual Analogue Scale) and pain disability (PDS; Pain Disability Scale). The authors showed that alexithymia was associated with lessened improvements in pain intensity and disability. They further pointed out that alexithymia and male gender were associated with poorer outcomes after treatment in a sample of patients suffering from chronic pain. Alexithymic patients scored higher on pain disability and depressivity at baseline. At follow-up alexithymic patients scored higher on pain intensity, pain disability, and depressivity. Significant decreases were found in pain intensity, pain disability and depressivity in result of assessed treatment protocols, but they did not find any changes within alexithymia.
Saariaho et al. [
125] conducted a non-randomized longitudinal study to examine the effects of alexithymia, depressivity, pain and treatment options in patients suffering from chronic non-malignant pain for at least three months. Primary outcomes were assessed at baseline (T1) and one-year follow-up (T2). Patients did not perform a specific study-based treatment. Instead, the authors assessed the patient records to see if patients had undergone any interventions which were based on a biomedical concept. Out of the initially assessed 271 patients, 154 patients participated at both measurement timepoints. Primary outcomes were alexithymia (TAS-20), depressivity (BDI-II), pain intensity (VAS), and pain disability scores (PDS). Alexithymic patients scored higher on pain disability and depressivity at baseline and follow-up. Significant decreases were found in pain intensity and pain disability in result of assessed treatment protocols. Alexithymia (TAS-20, DIF, and DDF) decreased significantly within the whole sample. In conclusion baseline alexithymia TAS-20 score was identified as significant predictor for pain disability.
Porcelli et al. [
126] conducted a non-randomized longitudinal study to examine whether alexithymia was a predictor of treatment outcome in patients suffering from functional gastrointestinal disorders (FGID). They assessed primary outcomes at baseline and at six-month follow-up. Treatments were conducted according to the patient’s symptoms and consisted of combinations of gastrointestinal medications, diet modifications, psychotropic medications, and psychological counseling or brief psychotherapy. Out of the initially assessed 130 patients, 112 patients participated at both measurement timepoints. At baseline, 56% of the whole sample were categorized as alexithymic (TAS-20 ≥ 61). Primary outcomes were gastrointestinal symptoms (GSRS; Gastrointestinal Symptom Rating Scale), alexithymia (TAS-20), anxiousness (HADS-A), and depressivity (HADS-D). On the basis of (a) change within overall gastrointestinal symptoms and (b) low levels of gastrointestinal symptoms at follow-up, the whole sample was divided into “improved” and “unimproved” subgroups. The authors identified alexithymia as the most powerful predictor for recovery status and overall reduction in GSRS symptoms after treatment, even after controlling for baseline GSRS, depressivity, and anxiousness. The TAS-20 total score was shown to be a reliable and stable predictor of treatment outcome.
Reese [
122] conducted a randomized, wait-list-controlled longitudinal study to examine the effects of an individual 10-session CBT, focusing on reduction of physiological arousal through relaxation techniques, enhancing activity regulation, pace activities, increasing awareness of emotions, modifying dysfunctional beliefs, enhancing communication of thoughts and emotions and reducing spousal reinforcement of illness behavior, on alexithymia in patients suffering from somatization disorders. Another aim of this study was to examine whether improvements of alexithymia after CBT predicted improvements in somatization symptoms and functioning.
Primary outcomes were defined as severity of somatization, measured by Clinical Global Impression for Somatization Disorder (CGI-SD) as well as daily symptom diaries to record maximum severity of somatoform symptoms every day. Other assessed outcomes were defined as mental health (SF-36), physical functioning (Mental Health scale of MOS 26-Item Short-Form Health Survey, SF-36), alexithymia (TAS-20), defensiveness (MCS; Marlowe-Crowne Social Desirability Scale), and somatosensory amplification (SSAS; Somatosensory Amplification Scale). Outcomes were assessed at baseline (T1), 3-months-follow-up (T2), 9-months-follow-up (T3), and 15-months-follow-up (T4).
Participants not randomized into CBT in combination with psychiatric consultation letter (PCL) treatment group (n = 43) were randomized into the control group whose physicians only received a psychiatric consultation letter (PCL; n = 41). Out of 367 patients who completed a telephone screening interview, 142 patients participated in a face-to-face screening interview. Finally, 84 patients participated.
Overall, results revealed a decrease in medically unexplained physical symptoms. A decrease in alexithymia further correlated with improvements in the severity of somatization symptoms. Thus, the authors concluded that CBT successfully improves alexithymia in patients suffering from somatization disorders. Reese [
122] attributes these results to a possible shift away from externally oriented thinking towards an awareness of inner feelings and thoughts. He argues that EOT can be seen as ‘the cognitive component’ of alexithymia, whereas DIF and DDF represent ‘the emotional components’.
Probst et al. [
121] conducted a randomized controlled longitudinal study to examine treatment outcomes in patients suffering from multisomatoform disorders after a 12-week manualized brief psychodynamic–interpersonal therapy (PIT) which included establishment of a therapeutic alliance, treatment of somatoform symptoms, behavioral, emotional, and interpersonal correlates, and termination issues. One hundred and seven patients were randomized to treatment condition, and 104 were allocated to the control condition. Out of 107 patients within the treatment condition, 83 remained for statistical analyses due to missing data. Primary outcome was defined as patient-reported physical quality of life at 9-month follow-up after treatment was undertaken and was operationalized by the Physical Component Summary of the SF-36 Health Survey. They did not find significant interactions between the patients′ alliance ratings and alexithymia on physical quality of life at follow-up. Interaction between the therapists′ alliance ratings and the patients′ alexithymia scoring on physical quality of life was significant, but after controlling for depressivity this interaction did not reach significance anymore.
Saedi et al. [
120] conducted a randomized longitudinally study to evaluate the efficacy of behavioral–cognitive therapy (CBT) on alexithymia and self-effectiveness of pain in patients suffering from chronic musculoskeletal pain. Out of 45 patients, 30 patients were simple-randomized into two treatment intervention groups and 15 patients to one control group. Treatment consisted of eight sessions of 90 min CBT. Primary outcomes were defined as alexithymia (TAS-20) and self-effectiveness of pain (PSEQ; Pain self-effectiveness questionnaire). Patients underwent CBT decreased significantly more in alexithymia and pain self-efficacy than patients within control group. CBT led to increased self-efficacy of pain, reduced alexithymia, and reduced harmful effects of pain. The authors conclude by pointing out CBT as a beneficial treatment for decreasing alexithymia and increasing the self-effectiveness of pain in patients suffering from chronic musculoskeletal pain. They ascribe the decrease in alexithymia to (a) therapeutic distinction between physical and emotional sensations, as well as (b) emphasis on the role of bodily sensations in stimulating negative thoughts.
We identified three main foci within these included studies: (1) analyses examining associations between alexithymia and other psychological variables at baseline, (2) analyses examining alexithymia as an outcome variable itself, and (3) analyses examining associations between alexithymia and non-alexithymia-related treatment outcomes.
Table 6 provides an overview of types of analyses performed within the included studies.
3.1.4. Examination of Alexithymia as Outcome Variable
Aboussouan et al. [
123] reported a significant improvement of alexithymia, measured by TAS-20, over all patients (Δ = 6.17,
SE = 1.19). They did not find any group differences (CPP vs. NPCP) prior to (CPP:
MT1 = 50.81,
SDT1 = 13.86; NPCP:
MT1 = 48.76,
SDT1 = 12.91;
F (1, 104) = 0.62,
p = 0.43,
d = 0.15) and after (CPP:
MT2 = 45.31,
SDT2 = 12.63; NPCP:
MT2 = 42.31,
SDT2 = 10.59;
F (1, 104) = 1.75,
p = 0.19,
d = 0.26) treatment.
Melin et al. [
71] reported improvements of alexithymia following treatment. These improvements of alexithymia, measured by TAS-20, remained significant after Bonferroni correction (
p = 0.004). Across TAS-20 subscales, significant improvements were revealed for DIF and DDF. Changes within TAS-20 (
R2 = 0.03,
p > 0.05) as well as in DDF (
R2 = 0.02,
p > 0.05) were independent from a decrease in depressivity. Conversely, however, depressivity explained the variance within DIF (
R2 = 0.11,
p < 0.05) significantly.
Saariaho et al. [
124] did not find any significant change of alexithymia (TAS-20;
MT1 = 46.6,
SDT1 = 13.1;
MT2 = 46.3,
SDT2 = 12.6;
p = 0.84,
d = 0.02) or its subscales DIF (
MT1 = 15.2,
SDT1 = 6.7;
MT2 = 14.6,
SDT2 = 6.6;
p = 0.20,
d = 0.09), DDF (
MT1 = 10.9,
SDT1 = 4.6;
MT2 = 11.2,
SDT2 = 4.1;
p = 0.40,
d = 0.07), or EOT (
MT1 = 20.5,
SDT1 = 5.2;
MT2 = 20.7,
SDT2 = 5.1;
p = 0.70,
d = 0.04). At baseline, males scored significantly higher on the TAS-20 total score (
Mmales = 50.5,
SDmales = 11.8;
Mfemales = 44.3,
SDfemales = 13.6;
p = 0.035) and EOT (
Mmales = 22.7,
SDmales = 4.2;
Mfemales = 18.9,
SDfemales = 5.1;
p = 0.001). At follow-up, males scored higher on the TAS-20 total score (
Mmales = 49.6,
SDmales = 11.1;
Mfemales = 44.0,
SDfemales = 13.2;
p = 0.057), and EOT (
Mmales = 22.0,
SDmales = 3.5;
Mfemales = 19.7,
SDfemales = 6.0;
p = 0.057), but neither reached significance.
In a subsequent study, Saariaho et al. [
125] found that the percentage of patients categorized as non-alexithymic decreased significantly (T1: 85.0%; T2: 76.5%;
p = 0.015). Significant changes within outcome variables of the TAS-20 total score (
MT1 = 47.6,
SDT1 = 12.2;
MT2 = 49.7,
SDT2 = 13.1;
p = 0.005,
d = 0.166), DIF (
MT1 = 15.7,
SDT1 = 6.3;
MT2 = 16.7,
SDT2 = 7.0;
p = 0.017,
d = 0.150), and DDF (
MT1 = 11.1,
SDT1 = 4.4;
MT2 = 12.1,
SDT2 = 4.3;
p < 0.001,
d = 0.230) reached significance, but EOT (
MT1 = 20.7,
SDT1 = 4.6;
MT2 = 20.9,
SDT2 = 4.4;
p = 0.63,
d = 0.044) did not. Within subgroups of alexithymic-categorized and non-alexithymic-categorized patients, changes were found within the TAS-20 total score (alexithymic:
MT1 = 68.0,
SDT1 = 6.3;
MT2 = 66.6,
SDT2 = 9.7;
p = 0.034,
d = 0.171; non-alexithymic:
MT1 = 43.8,
SDT1 = 8.7;
MT2 = 46.7,
SDT2 = 11.3;
p = 0.001,
d = 0.288), DIF (alexithymic:
MT1 = 25.5,
SDT1 = 4.7;
MT2 = 25.3,
SDT2 = 5.5;
p > 0.9,
d = 0.039; non-alexithymic:
MT1 = 13.9,
SDT1 = 4.7;
MT2 = 15.2,
SDT2 = 6.1;
p < 0.010,
d = 0.239), DDF (alexithymic:
MT1 = 17.6,
SDT1 = 3.3;
MT2 = 16.9,
SDT2 = 3.9;
p = 0.28,
d = 0.194; non-alexithymic:
MT1 = 10.0,
SDT1 = 3.5;
MT2 = 11.3,
SDT2 = 3.8;
p < 0.001,
d = 0.356), and EOT (alexithymic:
MT1 = 24.9,
SDT1 = 3.5;
MT2 = 24.3,
SDT2 = 2.7;
p = 0.22,
d = 0.192; non-alexithymic:
MT1 = 20.0,
SDT1 = 4.4;
MT2 = 20.3,
SDT2 = 4.4;
p = 0.35,
d = 0.068).
Relative stability of alexithymia was revealed by Porcelli et al. [
126] through test–retest reliabilities and resulted in a reliability coefficient of
r = 0.76,
p < 0.001. Within the whole sample, the TAS-20 total score (
MT1 = 58.89,
SDT1 = 13.50;
MT2 = 55.85,
SDT2 = 12.32;
t (222) = 3.56,
p < 0.01,
d = 0.68) decreased significantly. The authors showed that, at baseline, unimproved patients scored significantly higher on the TAS-20 total score (
t (111) = 7.89,
p < 0.001). Analyses revealed that improved (
MT1 = 52.19,
SDT1 = 12.10;
MT2 = 48.66,
SDT2 = 8.91;
t (134) = 2.89,
p < 0.01,
d = 0.71) and unimproved patients (
MT1 = 69.25,
SDT1 = 7.81;
MT2 = 66.95,
SDT2 = 7.75;
t (86) = 2.10,
p < 0.01,
d = 0.71) did significantly improve within the TAS-20 total score.
Reese [
122] identified 24% of participants as alexithymic (TAS-20 ≥ 61) at T1. Analyses revealed baseline alexithymia as well as its subscales to be highly stable over the period of treatment (
rTAS-20 = 0.76,
pTAS-20 < 0.0001;
rDIF = 0.70,
pDIF < 0.0001;
rDDF = 0.72,
pDDF < 0.0001;
rEOT = 0.80,
pEOT < 0.0001). A significant effect for the interaction between time and condition was found for the TAS-20 total score (
F (2, 68) = 3.36,
p = 0.04). There was also found a significant effect of time within the TAS-20 total score (
F (2, 68) = 3.20,
p = 0.05). Analyses of TAS-20 subscales revealed a significant interaction between time and condition on DIF (
F (2, 68) = 4.23,
p = 0.02), indicating that the CBT + PCL group improved more than the PCL group. No significant interaction was found for DDF, whereas the interaction between time and condition proved to be significant within EOT (
F (2, 138) = 3.04,
p = 0.05). Both groups differed significantly on changes from T1 to T2 within the TAS-20 total score (
F (1, 75) = 5.59,
p = 0.02), DIF (
F (1, 75) = 6.19,
p = 0.02), and EOT (
F (1, 75) = 4.73,
p = 0.03). Subtracting the alexithymia scores of T2 from alexithymia scores of T4, generated another change score of alexithymia. Analysis revealed that there were no significant group differences within this change score. Alexithymia change score calculated by subtracting alexithymia scores of T1 from alexithymia scores of T4, indicated a significant difference between both groups within DIF (
F (1, 69) = 5.41,
p = 0.02), but not within the TAS-20 total score, DDF or EOT. Analyses of change scores of alexithymia, calculated by subtracting alexithymia scores at T2 from alexithymia scores from T4, revealed that treatment conditions differed significantly on EOT (
F (1, 81) = 4.08,
p < 0.05), indicating that PCL controls decreased more than participants within the CBT + PCL group. There were no significant difference on the TAS-20 total score, DIF, or DDF. It was also shown that treatment condition predicted changes within the TAS-20 total score (
F (1, 75) = 5.59, β = −0.26,
p = 0.02), DIF (
F (1, 75) = 6.19, β = −0.28,
p = 0.02), and EOT (
F (1, 75) = 4.73, β = −0.24,
p = 0.03), but not DDF.
Analyses conducted by Saedi et al. [
120] revealed that patients who underwent CBT improved significantly more in alexithymia (
F (1) = 50.48,
p ≤ 0.001) than patients within the control group. Within the intervention group, alexithymia (
MT1 = 69.67,
SDT1 = 9.65;
MT2 = 58.07,
SDT2 = 8.40;
MT3 = 46.67,
SDT3 = 8.86) decreased. Within the control group, alexithymia (
MT1 = 71.07,
SDT1 = 10.44;
MT2 = 70.53,
SDT2 = 10.54;
MT3 = 66.87,
SDT3 = 12.07) remained relatively stable.
3.1.5. Associations between Alexithymia and Non-Alexithymia-Related Treatment Outcomes
Aboussouan et al. [
123] found 20.69% of participants matching criteria for alexithymia (TAS-20 ≤ 61). They did not find significant group differences between CPP patients and NPCP patients in alexithymia at baseline and after treatment. In unadjusted analyses, CPP patients remained significantly more impaired in sexual functioning than NPCP patients (
µCPP = 4.41 ± 3.36;
µNPCP = 2.98 ± 2.87;
F = 6.03, 1,
p = 0.016,
d = 0.46). Linear mixed models revealed significant main effects for time within outcome improvements after treatment regardless of whether participants suffered from CPP (
p < 0.01; Δ
impairment in sexual function = 3.75,
SEimpairment in sexual function = 0.27; Δ
depressivity = 13.86,
SEdepressivity = 1.16; Δ
alexithymia = 6.17,
SEalexithymia = 1.19; Δ
pain = 3.45,
SEpain = 0.23). CPP patients decreased significantly more in depressivity than NPCP patients (
p < 0.05), but did not in impairment in sexual functioning, alexithymia or pain severity. Hierarchical linear regression within CPP patients revealed that marital status and baseline scores of outcome variables together explained 35% of the variance in post-treatment impairment in sexual functioning (
F (5, 47) = 3.0,
p = 0.02). Adding change scores of outcome variables increased explanation of the variance in impairment in sexual functioning by 29% (
F (3, 40) = 8.79,
p = 0.00).
As Saariaho et al. [
124] did show in their study, pain intensity (
MT1 = 5.7,
SDT1 = 1.2;
MT2 = 4.6,
SDT2 = 2.0;
p < 0.001,
d = 0.67), pain disability (
MT1 = 16.3,
SDT1 = 4.7;
MT2 = 11.2,
SDT2 = 6.2;
p < 0.001,
d = 0.93), and depressivity (
MT1 = 14.7,
SDT1 = 11.0;
MT2 = 10.8,
SDT2 = 9.1;
p < 0.001,
d = 0.36) decreased within the whole sample. At follow up alexithymic subjects reported more pain intensity (
Malexithymic = 5.6,
SDalexithymic = 1.2;
Mnon alexithymic = 4.4,
SDnon alexithymic = 2.1;
p = 0.034,
d = 0.70), pain disability (
Malexithymic = 15.2,
SDalexithymic = 6.0;
Mnon alexithymic = 11.1,
SDnon alexithymic = 6.0;
p = 0.015,
d = 0.068) and depressivity (
Malexithymic = 20.8,
SDalexithymic = 11.2;
Mnon alexithymic = 8.5,
SDnon alexithymic = 6.8;
p < 0.001,
d = 1.33) than non-alexithymic subjects. Male gender (improvers: 5/18; non-improvers: 28/31;
p = 0.33,
φ = 0.24), baseline TAS-20 score (
Mimprovers = 41.1,
SDimprovers = 8.1;
Mnon-improvers = 48.8,
SDnon-improvers = 14.0;
p = 0.016,
d = 0.67) and DDF (
Mimprovers = 9.2,
SDimprovers = 2.5;
Mnon-improvers = 11.6,
SDnon-improvers = 5.0;
p = 0.034,
d = 0.61) were significantly higher in the group without improvement within pain intensity. DIF (
Mimprovers = 13.0,
SDimprovers = 4.1;
Mnon-improvers = 16.2,
SDnon-improvers = 7.3;
p = 0.052,
d = 0.54) and EOT (
Mimprovers = 18.9,
SDimprovers = 5.0;
Mnon-improvers = 21.0,
SDnon-improvers = 5.1;
p = 0.097,
d = 0.42) did not differ significantly between improvers and non-improvers within pain intensity. A significant correlation was found between the TAS-20 total score and BDI-II at follow-up (
r = 0.743,
p < 0.001). At follow-up males scored higher on pain intensity (
Mmales = 5.2,
SDmales = 1.8;
Mfemales = 4.3,
SDfemales = 2.1;
p = 0.051), and pain disability (
Mmales = 13.6,
SDmales = 5.7;
Mfemales = 10.6,
SDfemales = 6.3;
p = 0.033).
Saariaho et al. [
125] showed that over the period of their study the percentage of patients with acceptable pain intensity (VAS ≤ 4; baseline: 7.2%, follow-up: 26.1%,
p < 0.001) and pain disability (PDS ≤ 13; T1: 28.9%, T2: 44.7%,
p < 0.001) increased. Over all patients significant changes within outcome variables of pain intensity (
MT1 = 5.9,
SDT1 = 1.3;
MT2 = 5.1,
SDT2 = 1.9;
p < 0.001,
d = 0.491) and pain disability (
MT1 = 16.3,
SDT1 = 4.9;
MT2 = 14.3,
SDT2 = 6.0;
p < 0.001,
d = 0.365) reached significance, but depressivity (BDI-II;
MT1 = 15.7,
SDT1= 10.8;
MT2 = 15.4,
SDT2 = 11.6;
p = 0.58,
d = 0.027) did not. Within subgroups of alexithymic-categorized and non-alexithymic-categorized patients changes were found within pain intensity (alexithymic:
MT1 = 6.3,
SDT1 = 1.4;
MT2 = 5.7,
SDT2 = 1.8;
p = 0.006,
d = 0.372; non-alexithymic:
MT1 = 5.9,
SDT1 = 1.3;
MT2 = 4.9,
SDT2 = 1.9;
p < 0.001,
d = 0.614), pain disability (alexithymic:
MT1 = 19.1,
SDT1 = 3.2;
MT2 = 17.4,
SDT2 = 5.3;
p = 0.034,
d = 0.388; non-alexithymic:
MT1 = 15.8,
SDT1 = 5.0;
MT2 = 13.7,
SDT2 = 6.0;
p < 0.001,
d = 0.362), and BDI-II (alexithymic:
MT1 = 27.3,
SDT1 = 9.8;
MT2 = 28.7,
SDT2 = 12.9;
p = 0.46,
d = 0.122; non-alexithymic:
MT1 = 13.6,
SDT1 = 9.5;
MT2 = 13.1,
SDT2 = 9.6;
p = 0.39,
d = 0.052).
Porcelli et al. [
126] found that within the whole sample TAS-20 total score (
MT1 = 58.89,
SDT1 = 13.50;
MT2 = 55.85,
SDT2 = 12.32;
t (222) = 3.56,
p < 0.01,
d = 0.68), anxiousness (
MT1 = 10.15,
SDT1 = 4.98;
MT2 = 7.67,
SDT2 = 5.08;
t (222) = 6.42,
p < 0.01,
d = 1.22), depressivity (
MT1 = 10.63,
SDT1 = 5.68;
MT2 = 8.02,
SDT2 = 5.61;
t (222) = 5.59,
p < 0.01,
d = 1.25), and GSRS total score (
MT1 = 11.00,
SDT1 = 5.18;
MT2 = 5.77,
SDT2 = 6.14;
t (222) = 12.12,
p < 0.01,
d = 2.30) decreased significantly. At baseline, unimproved patients scored significantly higher on anxiousness (
t (111) = 1.98,
p = 0.04), depressivity (
t (111) = 4.39,
p < 0.001, and GSRS total score (
t (111) = 3.62,
p = 0.005). Analyses revealed that improved patients significantly improved on anxiousness (
MT1 = 9.41,
SDT1 = 4.50;
MT2 = 5.53,
SDT2 = 3.65;
t (134) = 8.52,
p < 0.01,
d = 2.08), depressivity (
MT1 = 8.91,
SDT1 = 5.20;
MT2 = 5.18,
SDT2 = 3.91;
t (134) = 7.37,
p < 0.01,
d = 1.80), and GSRS total scores (
MT1 = 9.65,
SDT1 = 4.15;
MT2 = 1.75,
SDT2 = 1.26;
t (134) = 18.32,
p < 0.01,
d = 4.48). Unimproved patients did not improve on anxiousness (
MT1 = 11.30,
SDT1 = 5.49;
MT2 = 10.95,
SDT2 = 5.27;
t (86) = 0.61,
d = 0.19), depressivity (
MT1 = 13.30,
SDT1 = 5.40;
MT2 = 12.41,
SDT2 = 4.99;
t (86) = 1.61,
d = 0.49), and GSRS total scores (
MT1 = 13.09,
SDT1 = 5.93;
MT2 = 11.98,
SDT2 = 5.45;
t (86) = 3.12,
d = 0.95).
The authors further conducted logistic regressions that specified treatment outcome (improved/unimproved) as dependent variable and baseline TAS-20, anxiousness, depressivity, and GSRS as independent variables. Results revealed the TAS-20 as strongest predictor of treatment outcome (R2 = 0.38, X2(1) = 53.64, p < 0.01), followed by depressivity (Cox and Snell R2 = 0.14, X2(1) = 16.84, p < 0.01), GSRS total score (Cox and Snell R2 = 0.10, X2(1) = 12.21, p < 0.01), and anxiousness (Cox and Snell R2 = 0.03, X2(1) = 3.93, p < 0.05). Within hierarchical regression analyses the first step including anxiousness, depressivity, and GSRS total score showed a significant fit of the model (X2(3) = 28.97, p < 0.001, Cox and Snell R2 = 0.23). Adding baseline TAS-20 showed an increase in overall fit of the model (X2(4) = 61.48, p < 0.001, Cox and Snell R2 = 0.42) and thereby increased prediction of improved patients from 84% to 85% and unimproved patients from 66% to 82%. Conducting a hierarchical regression, adding only TAS-20 as first step, the model revealed significance (X2(1) = 53.64, p < 0.001, Cox and Snell R2 = 0.38). By adding anxiousness, depressivity and GSRS total scores as second step the fit of the model as still significant (X2 (1) = 7.84, p < 0.05, Cox and Snell R2 = 0.42). Percentage changes within GSRS were significantly predicted by the TAS-20 total score (F (1, 110) = 45.36, p < 0.05, R2 = 0.29), depressivity (F (1, 110) = 22.14, p < 0.01, R2 = 0.17), and anxiousness (F (1, 110) = 4.93, p < 0.05, R2 = 0.04). Within another conducted hierarchical regression anxiousness and depressivity added as step 1 the model showed significant prediction (F (2, 109) = 11.47, p < 0.01, R2 = 0.17) of percentage change of GSRS. By adding TAS-20 to the regression as step 2 a significant increase in incremental variance could be shown (F (3, 108) = 18.55, p < 0.01, R2 = 0.34).
Reese [
122] showed that improvement in somatization symptoms at T2 correlated with changes within the TAS-20 total score (
r = 0.33,
p < 0.01), DIF (
r = 0.26,
p < 0.01), and EOT (
r = 0.34,
p < 0.01), but not DDF (
r = 0.11,
p = n.s.) from T1 to T4. Improvement in somatization symptoms at T4 were associated with changes within the TAS-20 total score (
r = 0.34,
p < 0.01), DIF (
r = 0.32,
p < 0.01), DDF
r = 0.27,
p < 0.05), but not EOT (
r = 0.13,
p = n.s.) from T1 to T4. There were also associations between physical functioning at T2 and changes from T1 to T4 within the TAS-20 total score (
r = −0.26,
p < 0.05), DIF (
r = −0.27,
p < 0.05), DDF (
r = −0.23,
p < 0.05), but not EOT (
r = −0.01,
p = n.s.) and between physical functioning at T4 and changes from T1 to T4 within the TAS-20 total score (
r = −0.32,
p < 0.01), DIF (
r = −0.37,
p < 0.01), but not DDF (
r = −0.20,
p = n.s.), and EOT (
r = −0.09,
p = n.s.). Analyses also revealed correlations between work status and changes in the TAS-20 total score (
r = 0.26,
p = 0.02), and DDF (
r = 0.32,
p = 0.003) at T2, and changes within EOT between T1 and T2 were associated with the duration of symptoms (
r = −0.31,
p = 0.005), meaning that the longer the symptoms lasted, the more EOT decreased. Analyses comparing both groups revealed that the CBT + PCL group improved more in somatization symptoms than the PCL group from T1 to T2 (
F (1, 82) = 36.06,
p < 0.0001) and from T1 to T4 (
F (1, 82) = 27.85,
p < 0.0001). Analyses focusing on secondary outcomes revealed a significant interaction between time and condition in diary scores, meaning that the CBT + PCL group decreased more in diary scores of severity of somatoform symptoms (
F (2, 69) = 5.52,
p = 0.005). Another significant interaction between time and condition was found in physical functioning (
F (2, 69) = 5.39,
p = 0.007), meaning that the CBT + PCL group improved more than the PCL group. It was also shown that patients identified as alexithymic (TAS-20 ≥ 61) scored lower on mental health than patients defined as non-alexithymic (TAS-20 ≤ 51) did at T1 (
F (1, 67) = 9.74,
p = 0.003); this difference was no longer significant at T2 and T4. Patients identified as alexithymic scored higher on anxiousness than patients identified as non-alexithymic at T1 (
F (1, 67) = 14.73,
p = 0.0003) and T4 (
F (1, 59) = 5.07,
p = 0.02). Groups differed significantly on physical functioning, when controlling for physical functioning at T1, meaning that patients identified as alexithymic reported greater physical functioning at T4 (
F (1, 62) = 10.29,
p < 0.002). Alexithymic patients scored lower on defensiveness at T1 (
F (1, 67) = 4.04,
p < 0.05). There were no significant differences between groups within CGI-SD at T1, on improvement in CGI-SD, and somatosensory amplification. The CBT + PCL group improved more in somatization symptoms from T1 to T2 (
F (1, 82) = 30.51,
p < 0.0001), and at T4 (
F (1, 82) = 27.85,
p < 0.0001).
Mediational analyses revealed treatment condition as significant predictor for improvements within somatization symptoms at T2 (F (1, 75) = 36.06, β = −0.57, p < 0.0001) and T4 (F (1, 70) = 28.89, β = −0.54, p < 0.0001). Improvements within somatization symptoms at T2 were predicted by changes in the TAS-20 total score (F (1, 75) = 9.25, β = 0.33, p = 0.003), DIF (F (1, 75) = 5.61, β = 0.26, p = 0.02), and EOT (F (1, 75) = 9.68, β = 0.34, p = 0.003) and improvements in somatization symptoms at T4 were predicted by the TAS-20 total score (F (1, 70) = 9.06, β = 0.34, p = 0.004), DIF (F (1, 70) = 8.21, β = 0.34, p = 0.006), and DDF (F (1, 70) = 5.33, β = 0.27, p = 0.02). Model including treatment condition and the TAS-20 total score as predictors for improvements in somatization symptoms at T2, reached significance (F (2, 74) = 20.81, β = 0.19, p < 0.05). In final model changes within the TAS-20 total score was no longer a significant predictor for improvement in somatization symptoms when controlling for mental health, at T2, severity of somatization at T1, and defensiveness at T2, and somatosensory amplification. Final model reached significance (F (5, 71) = 9.39, p < 0.0001) and accounted for 40% of the variance in improvement in somatization symptoms at T2. Within final model just treatment condition proved to be a significant predictor (β = −0.49, p < 0.0001). The interaction between treatment condition and changes within the TAS-20 total score significantly predicted improvement in somatization symptoms at T2 (F (1, 75) = 5.48, p = 0.02) and accounted for 7% of the variance but was no longer significant when condition and treatment added separately to the model (p = 0.32). Interaction between condition and changes within EOT reached significance when entered alone into the regression model (F (1, 75) = 8.38, p = 0.005) and accounted for 10% of the variance in somatization symptoms at T2. Interactions between condition and DIF or DDF did not reach significance. Even though changes within DDF significantly predicted improvement in somatization symptoms at T4, DDF cannot seen as a mediator, because treatment condition did not predict changes within DDF. The TAS-20 total score and DIF did not reach significance as predictors for improvements in somatization symptoms at T4.
Mediational analyses focusing on physical functioning as outcome revealed treatment condition as a significant predictor for physical functioning at T2 (F (2, 74) = 158.53, β = 0.16, p = 0.01) and T4 (F (2, 69) = 48.92, β = 0.20, p = 0.01) when controlling for physical functioning at T1. Physical functioning at T2 was predicted by changes within the TAS-20 total score and changes within DIF when controlling for physical functioning at T1, defensiveness at T2, mental health at T2, and somatosensory amplification at T2. Mediational status of the TAS-20 total (z = 1.40, p = 0.16) score and DIF (z = 1.41, p = 0.16) for physical functioning at T2 did not reach significance. Physical functioning at T4 was significantly predicted by changes within the TAS-20 total score (F (2, 69) = 50.84, β = −0.23, p = 0.01) and changes within DIF (F (2, 69) = 52.79, β = −0.25, p = 0.002) when controlling for physical functioning at T1. Within the final model, physical functioning at T4 was significantly predicted by physical functioning at T1 (F (6, 58) = 15.62, β = 0.69, p < 0.0001) and changes within the TAS-20 total score (F (6, 58) = 15.62, β = −0.21, p < 0.05).
Mediational analyses focusing on daily symptom diary scores as outcome revealed treatment condition as a significant predictor of daily symptom diary scores at T2 (F (2, 74) = 11.45, β = −0.29, p = 0.006) and T4 (F (2, 69) = 17.53, β = −0.31, p = 0.002) when controlling for daily symptom diary scores at T1. EOT significantly predicted daily symptom diary scores at T2 when controlling for daily symptom diary scores at T1 (β = 0.26, p = 0.02) and proved to be a significant predictor for daily symptom diary scores at T2 when controlling for daily symptom diary scores at T1, defensiveness at T2, mental health at T2, and somatosensory amplification at T2 (β = 0.30, p = 0.01), but the Sobel test did not reach significance (z = −1.74, p = 0.08). For daily symptom diary scores at T4, no variable proved to be a significant predictor.
Analyses by Probst et al. [
121] revealed a significant association between TAS-20 total scores and PHQ-9 scores (
r = 0.414,
p < 0.01). Moderation analyses investigating alexithymia as a moderator of the association between the patient’s alliance ratings and physical quality of life revealed a non-significant interaction effect (
t = 0.58,
p = 0.57), but alexithymia was identified as a significant moderator for the association between the therapist’s alliance ratings and physical quality of life (
t = 2.01,
p < 0.05). The moderating value defining the Johnsons–Neyman significance region was shown to be a TAS-20 total score of 61.21. Adding baseline depressivity as covariate, the association between the therapist’s alliance and physical quality of life after treatment no longer reached significance (
t = 1.83,
p = 0.07).
Saedi et al. [
120] revealed that patients who underwent CBT improved significantly more in self-effectiveness of pain (
F (1) = 89.89,
p ≤ 0.001). Self-effectiveness of pain (
MT1 = 17.60,
SDT1 = 3.90;
MT2 = 25.13,
SDT2 = 4.03;
MT3 = 35.93,
SDT3 = 6.0) increased over time. Within the control group, the self-efficacy of pain (
MT1 = 17.47,
SDT1 = 1.80;
MT2 = 16.73,
SDT2 = 2.40;
MT3 = 16.40,
SDT3 = 1.92) remained relatively stable.