Introduction
Persistent somatic symptoms (PSS) are a central problem in various medical disciplines (Leiknes et al., 2007). Back pain, headache, gastrointestinal symptoms and fatigue are among the most frequently reported symptoms in primary care settings of the Western world (Kitselaar et al., 2023). PSS can be more or less medically explained, with about 30% of PSS being medically unexplained or unrelated to a distinct disease (Haller et al., 2015). PSS are particularly common in individuals who are diagnosed with chronic diseases (Fletcher et al., 2022; Halpin & Ford, 2012; Kohlmann et al., 2013). In addition to being highly prevalent, PSS are associated with significant psychological impacts, increased healthcare visits, and disability (Haller et al., 2015; Rask et al., 2015). Numerous psychosocial factors, such as recent stressful life events and attentional biases can contribute to the development and maintenance of such symptoms (Witthöft & Hiller, 2010).
Patients experiencing a new somatic symptom often visit their primary care physician first, expecting to learn why they are experiencing their symptoms and to learn about potential treatments (Murray et al., 2016). Unfortunately, PSS patients are often dissatisfied with their healthcare as they perceive their providers’ agendas to be unhelpful at times (Kenny, 2004; Salmon et al., 2008). When they present with symptoms that medical tests cannot sufficiently explain, physicians are unable to provide adequate helpful treatments because of the unknown origin of the symptoms and patients often leave appointments feeling disappointed and discouraged (Houwen et al., 2017). Physicians who do not feel comfortable treating PSS typically refer these patients to other specialty providers, beginning what can sometimes turn into a cycle of endless referrals and misunderstandings between clinician and patient (Kenny, 2004; Murray et al., 2016). The distress related to navigating the healthcare system adds to the already significant distress these patients are experiencing (Houwen et al., 2017; Kenny, 2004).
Given that psychological and behavioral factors play an important role in the development and perpetuation of PSS, it follows that psychological support and treatment would be helpful in addressing the distress these patients experience. Cognitive behavior therapy (CBT) has emerged as the psychotherapy with the strongest evidence of efficacy for treating patients with PSS (van Dessel et al., 2014). CBT for PSS focuses on shifting maladaptive thoughts and behaviors related specifically to the symptoms through a re-attribution process in which a patient can hopefully adopt a more biopsychosocial model of understanding their somatic symptoms (Kleinstäuber et al., 2011; van Dessel et al., 2014). This allows patients to reframe beliefs, reduce associated distress, and lower the disability they experience as a result of their symptoms (van Dessel et al., 2014). A meta-analysis of 27 studies with 1781 participants demonstrated that psychotherapy, including CBT, had positive results for individuals with PSS (Kleinstäuber et al., 2011). For example, patients showed improvements in PSS symptoms, PSS-related cognitions, and healthcare utilization (Kleinstäuber et al., 2011). A different meta-analysis of 21 studies including 2658 patients found that CBT significantly reduced somatic symptom severity by the end of treatment (van Dessel et al., 2014). At a one-year follow-up assessment, this effect persisted (van Dessel et al., 2014).
An important element of psychological treatment is the therapeutic alliance, or the relationship between patient and therapist. Patients with PSS often encounter invalidating experiences, so developing a strong therapeutic alliance with PSS patients becomes an essential task for therapists hoping to maximize the efficacy of their treatment (Mander et al., 2017).
While schools of thought vary in how they characterize the therapeutic alliance, all can agree that the relationship is one co-created by therapist and patient, meaning that both members of the dyad contribute in some ways to the relationship. One widely accepted theory suggests that the therapeutic alliance is both the emotional bond or relationship between a therapist and patient, and agreement on tasks and goals within therapy (Bordin, 1979). Despite this, most studies focus on the bond rather than therapist-patient agreement on tasks and goals, and very few focus on therapist and patient perspectives together, with many studies simply exploring positive and negative alliance perceptions from either the therapist or patient perspective; a meta-analytic review of 79 therapeutic alliance studies found that 37 studies captured patient ratings, 26 captured therapist ratings, and 25 captured observer ratings (Martin et al., 2000). What may be more important than just positive or negative alliance (bond), however, is whether a therapist and patient can find agreement, or congruence, with one another about their relationship and the therapy (tasks and goals). Further, the direction of this congruence may also be relevant, whether both parties agree things are going well or agree things are going poorly.
Congruence has been shown to be associated with treatment outcomes in mood disorders, eating disorders, and borderline personality disorder (Atzil-Slonim et al., 2015; Jennissen et al., 2020; Kivity et al., 2020). Congruence may be an especially important element of treatment for PSS patients, as they tend to experience incongruence regarding how to treat symptoms in their interactions with medical providers (Kenny, 2004; Mander et al., 2017). Congruence in medical settings for PSS patients may look like a physician considering the patient’s agenda for the appointment and vice versa, allowing for collaborative or shared decision making (Gobat et al., 2015; Guassora et al., 2021; Kenny, 2004). However, physicians often experience frustration with PSS patients, particularly when they perceive the patient to be stubborn or rigid about their condition or when the patient “expects answers” from them (Kenny, 2004). This emotional experience has been shown to undermine potential congruence in the doctor-patient relationship (Kenny, 2004). Additionally, medical settings have time pressures and doctors are trained to be directive, potentially interfering with opportunities for doctors to build congruence with their patients (Applbaum, 2017; Yahanda & Mozersky, 2020).
In CBT for PSS, the intervention focuses on psychoeducation (an opportunity for therapists and clients to find shared agreement of somatic symptom etiology), rapport building (developing comfort within the therapist-patient relationship), validation (therapist expressing knowledge that the symptoms are real), and skill-building (reappraisal techniques, relaxation training, stress management, etc.; Kleinstäuber et al., 2011; van Dessel et al., 2014). A psychological care setting appears to inherently include more opportunities for a patient and therapist to reach congruence relative to medical settings.
Perceptions about how therapy is going are of particular interest when studying the PSS population, given that CBT for PSS has demonstrated profound efficacy, yet also has high drop-out rates (van Dessel et al., 2014). Since congruence has been associated with greater treatment adherence, building congruence in therapy could likely help address the high drop-out rates in PSS patients (Gobat et al., 2015; Houwen et al., 2017; Kenny, 2004; van Dessel et al., 2014; Waumans et al., 2023). Unfortunately, no congruence studies consisting solely of PSS patients have been completed.
One congruence study including somatoform disorder patients, a diagnosis characterized by PSS, found that greater congruence contributed to lower symptom distress (Jennissen et al., 2020). However, only 15 somatoform disorder patients were included within a larger sample of mood disorder patients, so further research is warranted to expand the understanding of the role of congruence in treatment for patients with PSS, particularly because the preliminary research suggests that stronger congruence may contribute to better treatment outcomes (Atzil-Slonim et al., 2015; Jennissen et al., 2020; Kivity et al., 2020).
The importance of this research is clear, yet many barriers exist in conducting this research. In particular, the existing congruence studies conducted in other clinical populations typically have large sample sizes and many measurement points that can support complex analyses of congruence such as response surface analysis, truth-and-bias modeling, or growth mixture modeling (Atzil-Slonim et al., 2015; Kivity et al., 2020; Marmarosh & Kivlighan, 2012; Rubel et al., 2018). PSS patients, unfortunately, are a population who have been historically difficult to recruit, making it difficult for the data to support such analyses (van Dessel et al., 2014). As such, alternate statistical methods are required to conduct congruence research in a PSS population.
For the present analysis we applied the grid sequence approach, a more fitting analytic method for exploring congruence and examining the following three research questions in a PSS population: (1) What congruence patterns between patients and therapists regarding therapy progress perceptions exist in this population?; (2) Which baseline patient characteristics are associated with which congruence patterns?; (3) Are congruence patterns associated with treatment outcomes?
The first question aimed to explore congruence patterns, defined as the ways in which therapists and patients agree with one another, regarding therapy progress perception. Moreover, we aimed to understand whether a dyad would move toward greater congruence over the course of therapy, or toward incongruence, or whether their congruence fluctuated. The second question examined how the level of somatic symptom disability, symptom severity, symptom distress, health anxiety, and depression in patients at the beginning of therapy was associated with the identified congruence patterns. These variables were selected because they were either primary treatment outcomes or because they were variables previously found to be associated with treatment outcomes in individuals with PSS (Sarter et al., 2021, 2022). The third question examined whether certain congruence patterns were associated with end-of-treatment somatic symptom severity and disability.
If more is known about the baseline patient characteristics associated with certain congruence patterns, and about what types of congruence patterns are associated with treatment outcomes, this information can inform therapists how to improve therapy for PSS patients by considering congruence. For example, it may be helpful for therapists to intentionally cultivate congruence by asking patients about their agenda or offering an open discussion of whether a patient believes a particular skill would be relevant or useful. It would also be helpful for therapists to know whether their patient is likely to develop a particular kind of congruence pattern based on baseline characteristics, so that they can tailor congruence building to that particular patient.
Methods
Dataset Overview
The present analysis is a secondary analysis of existing data from a clinical trial for 254 participants with PSS. All data for this analysis had been previously anonymized. As such, no ethics approvals were required. This randomized multicenter trial compared 20 sessions of cognitive behavior therapy (CBT) to 20 sessions of CBT complemented with emotion regulation training (ENCERT; Kleinstäuber et al., 2019). This trial found clinically significant improvement of all outcomes in both groups by the end of therapy (Kleinstäuber et al., 2019). The two treatment groups did not differ significantly from one another in these outcomes at the end of therapy (Kleinstäuber et al., 2019). As such, these two groups were merged for the purpose of the present analysis. Out of the 254 participants of the original RCT, we gained data for 174 participant-therapist dyads that were used for the analyses of the current study. The original RCT had been approved by a University IRB and was conducted in accordance with the principles of the Declaration of Helsinki.
Participant Recruitment and Eligibility
Participants were recruited in Germany and, if eligible, received treatment at one of seven outpatient university mental health clinics. In order to participate in this research trial, individuals had to meet the following criteria: (a) between ages 18 and 69; (b) reported at least three distressing persistent somatic symptoms that could not be sufficiently explained by medical examination; (c) scored ≥ 4 on the modified Pain Disability Index (mPDI; Tait et al., 1990) and ≥ 5 on the Patient Health Questionnaire-15 (PHQ-15; Kroenke et al., 2002); (d) met at least one of the three B-criteria of somatic symptom disorder (SSD) in the DSM-5 (American Psychiatric Association, 2013): (1) disproportionate and persistent thoughts about the seriousness of one’s symptoms; (2) persistently high level of anxiety about health or symptoms; and (3) excessive time and energy devoted to these symptoms or health concerns; (e) experienced somatic symptoms for at least six months, and (f) provided a documented medical check for medical diseases as a potential cause for the somatic symptoms.
Individuals were excluded if they (a) had a primary mental disorder requiring other treatments; (b) had acquired brain injuries; (c) took benzodiazepine, anti-psychotic, or opioid medications at the time of the study; (d) had a change in antidepressant medication during the four weeks prior to treatment until six months after the end of therapy; and (e) had received outpatient CBT targeting the PSS during the past two years.
Study Procedure
Eligible persons provided written informed consent before being randomly assigned to one of the two treatment groups in a 1:1 ratio. Individuals provided baseline characteristic data just prior to randomization. Participants then completed three to five preparatory intake sessions followed by 20 manualized, highly structured therapy sessions. Survey data were collected after treatment sessions 1, 10, and 20. The preparatory sessions included a thorough exploration of the participant’s physical and mental health history, social history, and the precipitating and perpetuating factors of the somatic symptoms. All sessions were 50 minutes long.
Treatments
Participants were either randomized to CBT or ENCERT. CBT focused on the precipitating and perpetuating factors of the participant’s PSS and how to change these factors. ENCERT focused on negative emotions as both a cause and a consequence of PSS. In the ENCERT treatment arm, participants learned a variety of emotion regulation strategies, including acceptance- and mindfulness-based strategies as well as CBT- and change-oriented strategies, such as cognitive reappraisal. The primary goal was for participants to learn both traditional CBT strategies and acceptance-based strategies and to successfully apply these to their individual problems. As previously stated, the original trial found both of these treatments comparably to be effective at treating PSS, so the two study groups were analyzed together in the present study. In this trial, 50 therapists provided the interventions, with some therapists treating just one participant and others treating up to 11. All therapists who provided interventions were either trainees or early-career therapists. Please refer to the study protocol for more details about this trial (Kleinstäuber et al., 2016) and to Table 1 for a summary of therapist characteristics.
Measures
Therapy Progress Perceptions
The authors created a measure including four items to assess therapy progress perceptions, which were answered on a 7-point Likert scale ranging from 0 (strongly disagree) to 6 (strongly agree). The scale included the following four questions: (1) Therapy helps me to understand my problems and gain perspective about my problems; (2) Therapy has given me new ideas to address my problems; (3) Therapy has helped me make changes; (4) I have learned to better deal with my problems. The patient and therapist versions both focused on the participant’s experience. For example, the patient version said, “The therapy has helped me make changes,” while the therapist version said, “The therapy has helped my patient make changes.” Both therapists and participants completed this questionnaire at session 1, session 10, and session 20.
The authors of the original trial created this measure to add a monitoring instrument to the measures that patients and therapists were completing. The authors of the original trial did not complete a validation study on this measure as it was initially created to be a simple monitoring measure. However, a factor analysis indicated that these items loaded onto a single factor with factor loadings ranging from 0.543 to 0.949. Additionally, item correlations suggested that these items demonstrated strong internal consistency at each of the three measurement time points for both the therapist version (asession 1 = 0.88; asession 10 = 0.91; asession 20 = 0.94) and patient version (asession 1 = 0.85; asession 10 = 0.90; asession 20 = 0.92).
Outcome Measures
This study included two primary outcome measures: somatic symptom severity and somatic symptom-related disability. Severity refers to the intensity of somatic symptoms or the number of somatic complaints. Disability refers to the functional impairment these symptoms may cause. The two outcome measures and their psychometric properties are further described below.
Somatic Symptom Severity
Somatic symptom severity was assessed with the summed score of the Patient Health Questionnaire-15 (PHQ-15; Kroenke et al., 2002). The PHQ-15 consists of 15 somatic symptoms and respondents rate how much each symptom has bothered them over the past week on a 3-point scale with higher scores indicating higher symptom severity (Kroenke et al., 2002). This measure demonstrates high internal reliability (a = 0.80; Kroenke et al., 2002). The PHQ-15 score at session 20 was used as an outcome measure in this analysis.
Somatic Symptom-Related Disability
Disability caused by somatic symptoms was assessed with the sum score of a modified version of the Pain Disability Index (mPDI; Tait et al., 1990). Respondents rate their level of disabilty on a 10-point numeric rating scale in seven life domains, including family/home responsibilities, recreation, social activity, occupation, sexual behavior, self-care, and life support activity (Tait et al., 1990). Higher scores indicate a higher level of disability. This measure demonstrates high internal reliability (a = 0.86; Tait et al., 1990). The mPDI score at session 20 was used as an outcome measure in this analysis.
Baseline Participant Characteristics
To examine baseline participant characteristics, a subscale of the modified version of the Pain Coping Questionnaire focused on somatic symptom distress and symptom coping (mPCQ; Geissner, 1999). Health anxiety was assessed with a sum score of the modified version of the Short Health Anxiety Scale (mSHAI; Salkovskis et al., 2002). Depression was measured with the sum score of the Beck Depression Inventory-II (BDI-II; Beck et al., 2011). The PHQ-15 and mPDI were administered at baseline as well.
Statistical Analyses
Statistical analyses were conducted with R using RStudio and with IBM SPSS Statistics 27 (SPSS Version 27.0, 2021; RStudio Team, 2022). Descriptive statistics were conducted on demographic information, including age, sex, education, number of co-morbid mental disorders, and treatment group to examine the demographic characteristics present in the sample.
As mentioned previously, longitudinal data analytic methods to assess congruence, such as response surface analysis or truth-and-bias models, require a minimum number of participants or measurement time points that may be difficult to achieve in a PSS population. The present study did not have enough measurement time points to use one of these more established longitudinal methods. Additionally, the authors of the present analysis were interested not only in congruence, but also the direction or valence of this congruence. Thus, a grid sequence analysis was conducted to answer the first research question. This approach uses repeated-measures dyadic data to learn more about within-dyad dynamics and allows comparison between dyads (Brinberg et al., 2018). Grid sequence analysis is a unique analytic method that has the ability to not only capture congruence (whether or not a dyad agrees), but also can capture the valence of this congruence (whether both members of a dyad agree that things are going well vs. poorly). Other established methods for assessing congruence are not easily able to capture the valence of the congruence.
Grid sequence analysis does not allow for missing data, so any cases with missing data were first removed (excluded participants n = 81). It was not possible to use a technique such as multiple imputation to address this missing data, since one cannot meaningfully predict the fluctuations of treatment trajectories for a PSS population. Thus, chi-squared tests and t-tests were conducted to ensure there were no significant characteristic differences between the included participants and the full sample.
Grid sequence analysis tracks a dyad’s movement across a four-quadrant grid where the x-axis represents patient scores and the y-axis represents therapist scores. For each of the three measurement time points (session 1, session 10, session 20) of the four-item therapy progress perception scale, a single point was plotted that captured the outcome of the measure for each member of the dyad. See Fig. 1 for a sample grid. Quadrant I is characterized by the therapist reporting greater therapy progress than the patient, quadrant II is characterized by high scores from both the therapist and the patient, quadrant III is characterized by low scores from both the therapist and the patient, and quadrant IV is characterized by the patient reporting greater therapy progress than the therapist.
Fig. 1
Sample grid
×
![]()
The movement of each dyad across this grid over the course of therapy was tracked and converted into a sequence. For example, a dyad may be in quadrant III at session 1, quadrant IV at session 10, and quadrant II at session 20; their sequence would be III, IV, II. Another common sequence might be IV, II, II, in which a dyad is in quadrant IV at session 1 and quadrant II at sessions 10 and 20. Once each dyad had their unique sequence, dyads with similar sequences (or congruence patterns) were identified and grouped together. However, grid sequence analysis offers results similar to a cluster analysis that can be interpreted in more than one way. The present study interpreted the results of the grid sequence analysis in two different ways: first grouping dyads by the valence of patients’ therapy progress perception at the end of treatment (positive or negative therapy progress perception), then by grouping dyads by the four most frequent congruence patterns to offer two different interpretations of the data.
To answer the second research question, aiming to explore which baseline patient characteristics are associated with congruence patterns, logistic regression models were conducted. These models included baseline somatic symptom disability (mPDI), symptom severity (PHQ-15), symptom distress (mPCQ), health anxiety (mSHAI), and depression (BDI-II) as continuous predictor variables and congruence patterns as a categorical outcome variable.
Last, linear regression models were conducted to examine the relationship between congruence patterns and treatment outcomes. These models included congruence patterns as a categorical predictor variable and end of treatment somatic symptom disability (mPDI) and somatic symptom severity (PHQ-15) as continuous outcome variables. These models controlled for baseline mPDI, baseline PHQ-15, and for any baseline participant characteristics that were significantly associated with congruence patterns as found in the second research question.
Results
Summary of the Findings of the Original RCT and Participant Characteristics
At baseline, the individuals in the two treatment groups, CBT versus ENCERT, did not differ regarding sociodemographic variables, clinical characteristics, or somatic symptom severity or disability (0.054 ≤ p ≤.870). The trial found clinically significant improvement of all outcomes, including PHQ-15 and mPDI, in both groups by the end of therapy (Kleinstäuber et al., 2019). The two treatment groups did not differ significantly from one another in these outcomes at the end of therapy (Kleinstäuber et al., 2019). As such, these two groups were merged for the purpose of the present analysis. Please refer to Kleinstauber et al. (2019) to see the full results of this clinical trial. Please refer to Table 1 for a summary of the demographic characteristics, both for the original trial and the current analysis.
Table 1
Summary of participant characteristics
Patient variables | Included in present analysis | Total sample original trial | Test statistics |
|---|---|---|---|
n = 174, M (SD) | n = 254, M (SD) | ||
Age (in years) | 43.26 (12.88) | 43.38(12.92) | t(427) = 0.95, p =.925 |
Years of Education | 14.51 (3.16) | 14.55(2.94) | t(427) = 0.13, p =.893 |
Sex | n(%) | n(%) | |
Female | 115(66.10) | 163(64.20) | X2(1) = 0.17, p =.682 |
Male | 59(33.90) | 91(35.80) | |
Main diagnosis | |||
Somatization disorder | 39(22.40) | 55(21.70) | X2(2) = 1.28, p =.528 |
Undifferentiated somatoform disorder | 90(51.70) | 144(56.60) | |
Somatoform pain disorder | 45(25.90) | 55(21.70) | |
Number of comorbid mental disorders | |||
0 comorbid disorders | 92(52.90) | 126(49.60) | X2(2) = 0.44, p =.802 |
≥ 1 comorbid disorder | 59(33.90) | 92(36.20) | |
≥ 2 comorbid disorders | 23(13.20) | 36(14.20) | |
Randomization | |||
CBT | 83(47.70) | 127(50.00) | X2(1) = 0.22, p =.640 |
ENCERT | 91(52.30) | 127(50.00) |
Baseline variables | M(SD) | M(SD) | |
|---|---|---|---|
BDI | 20.25(9.14) | 21.16(10.04) | t(427) = 0.96, p =.340 |
mPCQ | 68.73(17.26) | 67.87(18.10) | t(427) = 0.62, p =.860 |
mSHAI | 29.92(13.15) | 29.69(13.47) | t(427) = 0.18, p =.861 |
mPDI | 33.21(11.49) | 33.64(11.93) | t(427) = 0.37, p =.710 |
PHQ-15 | 13.01(4.02) | 12.76(4.13) | t(427) = 0.62, p =.534 |
Outcome variables | M(SD) | M(SD) | |
|---|---|---|---|
mPDI | 18.04(13.97) | 19.20(14.24) | t(427) = 0.83, p =.404 |
PHQ-15 | 7.71(4.61) | 7.59(4.60) | t(427) = 0.26, p =.791 |
Therapist variables | n = 50, M(SD) | ||
|---|---|---|---|
Age (in years) | 30.92(7.20) | ||
Years of Experience | |||
Trainee | 2.16(1.11) | ||
Licensed psychologist | 9.89(7.67) | ||
Training Status | n(%) | ||
Trainee | 40(80%) | ||
Licensed Psychologist | 10(20%) | ||
Sex | |||
Female | 42(84%) | ||
Male | 8(16%) | ||
Analysis of Participants Lost to Follow-Up
Analysis of the 81 excluded participants indicated that these participants did not significantly differ from the included participants of the original RCT in age, gender, baseline depression, symptom coping, health anxiety, symptom disability, or symptom severity (0.051 ≤ p ≤.837). The median number of therapy sessions was 1 (IQR: 0-7.5) for the excluded participants and 20 for the included participants (IQR: 20–20).
Grid Sequence Analysis (GSA)
As previously stated, GSA results in something similar to a cluster analysis, which requires a researcher to make a subjective decision about how to define clusters, aiming to reach relatively equal distribution within groups. Within the data analyzed for the present study, the researchers found two potential ways to cluster the groups when considering theoretical factors and data-related factors for grouping. These two different groupings were each analyzed to see whether consistent findings would emerge, helping to minimize bias in an otherwise more subjective analytical method. The results of the two grouping methods are each described below.
Grouping Dyads by the Valence of Patients’ Therapy Progress Perceptions at the end of Therapy
Grid Sequence Analysis: Congruence between Patients and Therapists Regarding Therapy Progress Perceptions
The 174 included dyads of patients and therapists demonstrated 29 unique congruence patterns. Given the high number of unique sequences, the patterns were further grouped together. Two groups were created: (1) dyads who ended therapy in quadrants I or III, and (2) dyads who ended therapy in quadrants II or IV. This sorting of dyads separated the 15 dyads in which the patient scores were low at the end of therapy, indicating a more negative view of therapy (quadrants I and III) from the 159 dyads in which the patient scores were high at the end of therapy, indicating a more positive view of therapy (quadrants II and IV). The decision to create a grouping based on positive vs. negative patient views was a theoretical one— research has long demonstrated that patient perceptions of various aspects of therapy, but particularly elements of the therapeutic relationship such as congruence, are related to treatment outcomes (Jennissen et al., 2020; Kenny, 2004; Rubel et al., 2018; Wright & Davis, 1994).
Prediction of Congruence Patterns through Baseline Characteristics of Patients
The logistic regression model revealed no significant effects, indicating that the patients in these two groups did not differ in their baseline participant characteristics (Table 2).
Table 2
Logistic regression model (grouping dyads by the valence of patients’ therapy progress perceptions at the end of therapy): comparison between dyads ending treatment in quadrants I/III (negative patient therapy progress perception) and II/IV (positive patient therapy progress perception) regarding participant characteristics at baseline
b(SE) | OR[95% CI] | p-value | |
|---|---|---|---|
X2(6) = 10.58, p =.102, Nagelkerke’s R2 = 0.13 | |||
Predictors | |||
Baseline BDI-2 | -0.11(0.06) | 0.89[0.79, 1.01] | 0.064 |
Baseline mPCQ | 0.03(0.02) | 1.03[0.99, 1.07] | 0.110 |
Baseline mSHAI | 0.03(0.03) | 1.03[0.98, 1.09] | 0.195 |
Baseline PDI | 0.05(0.03) | 1.06[0.99, 1.12] | 0.078 |
Baseline PHQ-15 | -0.02(0.09) | 0.98[0.82, 1.18] | 0.817 |
Prediction of the Therapy Outcome through Congruence Patterns
The two linear regression models (one for symptom severity and one for disability as the outcome) indicated no difference in end-of-treatment symptom-related disability between the two groups, but did indicate that patients who had a more positive view of therapy had lower symptom severity at the end of treatment (p =.019; Table 3).
Table 3
Linear regression models (grouping dyads by the valence of patients’ therapy progress perceptions at the end of therapy): comparison between dyads ending treatment in quadrants I/III (negative patient therapy progress perception) and II/IV (positive patient therapy progress perception) regarding therapy outcomes (symptom disability and severity)
b(SE) | β | t | p | |
|---|---|---|---|---|
Outcome: mPDI (end of therapy) | ||||
mPDI (baseline) | 0.50(0.09) | 0.44 | 5.45 | < 0.001 |
I/III vs. II/IV | -3.89(2.21) | -0.14 | -1.76 | 0.081 |
Outcome: PHQ-15 (end of therapy) | ||||
PHQ-15 (baseline) | 0.52(0.08) | 0.45 | 6.62 | < 0.001 |
I/III vs. II/IV | -2.63(1.11) | -0.16 | -2.37 | 0.019 |
Grouping Dyads by Most Common Congruence Patterns
Grid Sequence Analysis: Congruence between Patients and Therapists Regarding Therapy Progress Perceptions
Given the high number of dyads ending in a positive quadrant, a deeper analysis via a second grouping of the dyads was completed. When counting the most frequent sequences, the four most common sequences all included dyads ending therapy in quadrant II, the “positive congruence” quadrant. Ten dyads moved from quadrant I at the beginning of therapy to quadrant II by the end of therapy, 26 dyads remained in quadrant II during the entire course of therapy, 31 dyads moved from quadrant III at the beginning of therapy to quadrant II by the end of therapy, and 49 dyads moved from quadrant IV at the beginning of therapy to quadrant II by the end of therapy. These dyads together accounted for 116 of the 174 total dyads.
Prediction of Congruence Patterns through Baseline Characteristics of Participants
The four logistic regression models (one for each of the four most common congruence patterns) revealed that the patients moving from quadrant I at the beginning of therapy to quadrant II by the end of therapy had higher baseline depression than the patients moving from quadrant IV to quadrant II (Table 4). These two groups of dyads both ended in quadrant II, the strong positive congruence quadrant in which patients and therapists agreed that therapeutic progress was being made. They differed in their starting points— those beginning in quadrant I were dyads in which the therapist gave a high therapy progress perception score and the patient gave a low one. In other words, these therapists were optimistic relative to their patients about how therapy would go. This was the group that included patients with higher baseline depression scores. The dyads who began in quadrant IV were ones in which the patient gave a high therapy progress perception score and the therapist gave a low one, indicating that the patients were more optimistic than the therapist in this group. These were the patients who had lower baseline depression scores. No other significant effects were identified.
Table 4
Logistic regression models (grouping dyads by four most common congruence patterns): comparison between dyads beginning in quadrants I/II/III/IV and ending in quadrant II regarding baseline participant characteristics
b(SE) | OR[95% CI] | p-value | |
|---|---|---|---|
X2(18) = 29.08, p =.047, Nagelkerke’s R2 = 0.225 | |||
Quadrant I vs. II | |||
Predictors | |||
Baseline BDI-2 | -0.14(0.08) | 0.87[0.74, 1.02] | 0.080 |
Baseline mPCQ | 0.00(0.02) | 1.00[0.96, 1.05] | 0.909 |
Baseline mSHAI | 0.04(0.03) | 1.04[0.97, 1.11] | 0.242 |
Baseline mPDI | 0.05(0.04) | 1.05[0.97, 1.13] | 0.203 |
Baseline PHQ-15 | -0.14(0.13) | 0.87[0.67, 1.13] | 0.287 |
Quadrant I vs. III | |||
Predictors | |||
Baseline BDI-2 | -0.13(0.08) | 0.88[0.75, 1.03] | 0.098 |
Baseline mPCQ | -0.03(0.02) | 0.97[0.93, 1.02] | 0.205 |
Baseline mSHAI | -0.00(0.03) | 1.00[0.94, 1.07] | 0.982 |
Baseline mPDI | 0.03(0.04) | 1.03[0.96, 1.11] | 0.390 |
Baseline PHQ-15 | -0.17(0.13) | 0.84[0.65, 1.08] | 0.175 |
Quadrant I vs. IV | |||
Predictors | |||
Baseline BDI-2 | -0.20(0.08) | 0.82[0.70, 0.96] | .012 |
Baseline mPCQ | -0.00(0.02) | 1.00[0.96, 1.04] | 0.932 |
Baseline mSHAI | 0.02(0.03) | 1.02[0.96, 1.09] | 0.512 |
Baseline mPDI | 0.05(0.04) | 1.05[0.98, 1.12] | 0.195 |
Baseline PHQ-15 | -0.03(0.12) | 0.97[0.77, 1.23] | 0.796 |
Quadrant II vs. III | |||
Predictors | |||
Baseline BDI-2 | 0.01(0.05) | 1.01[0.92, 1.11] | 0.843 |
Baseline mPCQ | -0.03(0.02) | 0.97[0.94, 1.00] | 0.063 |
Baseline mSHAI | -0.04(0.02) | 0.96[0.92, 1.01] | 0.090 |
Baseline mPDI | -0.02(0.03) | 0.98[0.93, 1.04] | 0.533 |
Baseline PHQ-15 | -0.03(0.10) | 0.97[0.80, 1.17] | 0.724 |
Quadrant II vs. IV | |||
Predictors | |||
Baseline BDI-2 | -0.06(0.05) | 0.95[0.86, 1.03] | 0.215 |
Baseline mPCQ | -0.00(0.02) | 1.00[0.97, 1.02] | 0.768 |
Baseline mSHAI | -0.02(0.02) | 0.98[0.94, 1.03] | 0.396 |
Baseline mPDI | -0.00(0.03) | 1.00[0.95, 1.05] | 0.923 |
Baseline PHQ-15 | 0.11(0.09) | 1.12[0.94, 1.32] | 0.211 |
Quadrant III vs. IV | |||
Predictors | |||
Baseline BDI-2 | -0.07(0.04) | 0.94[0.86, 1.02] | 0.134 |
Baseline mPCQ | 0.03(0.01) | 1.03[1.00, 1.06] | 0.071 |
Baseline mSHAI | 0.02(0.02) | 1.02[0.98, 1.07] | 0.297 |
Baseline mPDI | 0.02(0.02) | 1.02[0.98, 1.06] | 0.533 |
Baseline PHQ-15 | 0.14(0.08) | 1.15[0.98, 1.35] | 0.080 |
Prediction of the Therapy Outcome through Congruence Patterns
The two linear regression models (one for symptom severity, one for disability) revealed that there were no significant differences in end-of-treatment symptom-related disability between these four groups. The linear regression models showed that the dyads who moved from quadrant IV at the beginning of therapy to quadrant II by the end of therapy had lower end of treatment symptom severity than the dyads who moved to quadrant II from either quadrants I (p =.045) or III (p =.014; Table 5). Again, all dyads in this grouping landed in quadrant II, the strong positive congruence quadrant, by the end of treatment. Quadrant I dyads are characterized by therapist scores being higher than patients, quadrant III dyads are characterized by negative congruence in which both patients and therapists gave low therapy progress perception scores, and quadrant IV dyads are characterized by patient scores being higher than therapists. This particular finding suggests that the dyads in which patients have higher therapy progress perception scores have lower end of treatment symptom severity than dyads in which patients have lower therapy progress perception scores.
Table 5
Linear regression models (grouping dyads by four most common congruence patterns): comparison between dyads beginning in quadrants I/II/III/IV and ending in quadrant II regarding therapy outcomes (symptom disability and severity)
b(SE) | β | t | p | |
|---|---|---|---|---|
Outcome: mPDI (end of therapy) | ||||
Beginning in QI vs. II | -2.04(4.79) | -0.07 | -0.43 | 0.673 |
Beginning in QI vs. III | 0.65(2.24) | 0.04 | 0.29 | 0.771 |
Beginning in QI vs. IV | -0.55(1.30) | -0.06 | -0.42 | 0.676 |
Beginning in QII vs. III | 4.21(3.40) | 0.14 | 1.24 | 0.220 |
Beginning in QII vs. IV | 0.13(1.47) | 0.01 | 0.09 | 0.931 |
Beginning in QIII vs. IV | -4.40(2.57) | -0.17 | -1.71 | 0.091 |
Outcome: PHQ-15 (end of therapy) | ||||
Beginning in QI vs. II | -3.11(1.65) | -0.31 | -1.89 | 0.067 |
Beginning in QI vs. III | -0.17(0.77) | -0.03 | -0.22 | 0.824 |
Beginning in QI vs. IV | -0.89(0.44) | -0.26 | -2.05 | 0.045 |
Beginning in QII vs. III | 1.94(1.08) | 0.195 | 1.80 | 0.077 |
Beginning in QII vs. IV | -0.05(0.44) | -0.01 | -0.12 | 0.905 |
Beginning in QIII vs. IV | -2.08(0.83) | -0.24 | -2.52 | 0.014 |
Discussion
The present study explored congruence patterns regarding therapy progress perceptions between participants with persistent somatic symptoms (PSS) and their therapists. The first research question aimed to understand the congruence patterns present within a PSS population. For a study of a smaller sample size with few measurement time points, grid sequence analysis offered a creative and well-fitting method for examining congruence subgroups within the sample.
The first important result of this analysis is that regardless of how dyads were grouped, an overwhelming majority of dyads in this study ended therapy in quadrant II, characterized by strong positive congruence. This indicates that both patients and therapists described that therapy was helpful, addressed problems effectively, gave new perspective, and increased coping. This shift toward positive congruence by the end of therapy was present irrespective of whether the patient-therapist dyad was congruent (either positive or negative) or incongruent at the beginning of therapy. It would be important for future studies to examine whether completing many sessions of therapy improves congruence, or perhaps a shift toward positive congruence over the course of therapy encourages patients to continue attending therapy.
The second important result of this analysis showed that baseline participant characteristics did not consistently relate to congruence patterns. For the first interpretation of the grid sequence analysis with groups characterized by positive or negative valence, no particular baseline participant characteristics were associated with either group. However, for the second interpretation of the grid sequence analysis in which groups were characterized by the four most common congruence patterns, the dyads that began therapy in the two incongruent quadrants differed from one another in their baseline depression scores. In particular, dyads who began therapy in quadrant I and ended therapy in quadrant II had higher depression than the dyads who began therapy in quadrant IV and ended therapy in quadrant II. The dyads who began in quadrant I were those in which the therapist had very positive therapy progress perceptions while the patient had rather negative ones. The dyads in quadrant IV were the opposite, with the patient having very positive therapy progress perceptions and the therapist having low ones.
In other words, the more depressed patients in quadrant I were underestimating therapeutic progress relative to their therapists when compared to the less depressed patients in quadrant IV who were overestimating therapeutic progress relative to their therapists. This finding is not surprising, as previous research has demonstrated that depressive symptoms are associated with more negative cognitive biases, which may explain the underestimation of therapeutic progress seen in the more depressed patients in this sample (Kaya Lefèvre et al., 2019). However, despite this negative perception, depressed patients in our sample still saw a meaningful reduction in their somatic symptom severity and associated disability. Therapists of depressed PSS patients should keep this in mind and consistently orient their patients and themselves to observable progress throughout therapy.
Given that health anxiety, pain coping, and symptom disability and severity were not associated with any particular congruence patterns, we may interpret that baseline participant characteristics are generally not related to patients’ and therapists’ abilities to find congruence with one another. This has important implications for therapists hoping to work with PSS patients since these patients often struggle to develop congruence in medical settings (Kenny, 2004; Murray et al., 2016). The findings of the present analysis indicate that congruent interactions are possible and even common as the duration of therapy increases. Given this, if initial incongruence is encountered when working with PSS patients, therapists should not assume that the incongruence will persist. Rather, therapists should continue to maintain hope and potentially focus on factors that may improve congruence, such as intentionally attuning to their patients and effectively addressing ruptures, and making a point to agree on the content and goals of therapy together with the patient as a collaborative progress (Atzil-Slonim et al., 2015; Rubel et al., 2018).
The third important result of this analysis is that congruence patterns were not found to be associated with end-of-treatment somatic symptom disability, regardless of the grouping method used for the grid sequence analysis. However, as previously mentioned, study patients on average saw a significant decrease in their somatic symptom disability by the end of treatment. This suggests that a significant reduction in disability or improvement in function is possible regardless of congruence, and appears to happen via a mechanism potentially unrelated to congruence. This mechanism is likely related to something specific within the intervention itself, as cognitive behavioral interventions for PSS focus explicitly on decreasing disability and increasing function (van Dessel et al., 2014). Since providers often worry about their ability to successfully treat PSS patients, this should be a comforting finding: that patients are likely to see a clinically significant decrease in disability regardless of whether the relationship with their provider is positively congruent, negatively congruent, or incongruent.
Congruence patterns were found to be associated with end-of-treatment somatic symptom severity. In particular, based on the first way of grouping dyads by positive or negative valence, the patients who ended therapy with a more positive view of the treatment outcome had lower symptom severity at the end of treatment than those who held a more negative view. These findings support existing literature that suggests strong positive congruence and positive patient perceptions may be associated with more favorable treatment outcomes (Jennissen et al., 2020; Rubel et al., 2018). Of course, it is also possible that the positive patient perceptions are a result of the symptom reduction, or, perhaps more likely, that there is a bidirectional relationship between patients’ perspectives and their outcomes. It appears that there is mixed evidence of the relationship between congruence and treatment outcomes in that some treatment outcomes are related to congruence while others are not.
Based on the second way of grouping dyads by the most frequent congruence patterns, the dyads who began therapy in quadrant IV (therapist rating low, patient rating high) had lower end-of-treatment somatic symptom severity than the dyads who began therapy in quadrant I (therapist rating high, patient rating low) and III (therapist rating low, patient rating low), even though all of these dyads ended therapy in quadrant II. This suggests that patients’ perceptions of therapy outcomes from the beginning of therapy is a relevant factor. This finding is again supported by previous research that suggests positive patient perceptions are associated with better treatment outcomes (Jennissen et al., 2020).
Strengths and Limitations
The present analysis expands the understanding of patient-therapist congruence in a PSS population, an area in which minimal research has been previously conducted. Congruence is a nuanced factor in therapy and continues to appear to be a factor particularly relevant in the treatment of PSS patients. A strength of the analysis is that grid sequence analysis was able to assess positive or negative valence in addition to simple congruence or incongruence between patient and therapist. Knowing the direction of the incongruence, whether therapists are over- or underestimating treatment progress relative to their patients, or whether dyads agree things are going well or poorly, offers a more detailed view of the therapeutic relationship. Future studies should aim to capture valence in addition to congruence as this offers a more well-rounded view of a dyad’s dynamics. An additional strength of this analysis is that 50 therapists provided treatment in the study, which offers quite a broad sample of therapists, potentially strengthening the generalizability of these findings. Future studies should aim to further explore what factors impact symptom disability and severity.
The present analysis also has a few limitations to consider. First, the authors relied on a self-created measure of therapy progress perceptions. Though a factor analysis and an analysis of internal consistency were performed, there may be issues related to validity and reliability when using a non-validated measure.
Another particular challenge is that only three time points were available for analysis. While the beginning, middle, and end of treatment all serve as important landmarks in therapy, having only three time points simplifies the complex and dynamic nature of therapeutic progress and congruence. This complexity would be true of any therapy but seems particularly reductive in this population as PSS are a dynamic and fluctuating experience. Additionally, given the dyadic analysis, both therapists and patients needed to complete the measures at the beginning, middle, and end of therapy in order to be included, 81 dyads with missing data were removed from the analysis, limiting the sample size. However, an analysis of these 81 dyads demonstrated no characteristic differences between these dyads and the ones that were included for the analysis. Despite no major differences between the included and excluded participants, the present analysis includes a large and notable proportion of missing data. It is possible that this degree of missing data causes certain congruence patterns to appear to more common than they really are, thus, results should be interpreted cautiously given this limitation. However, even with a conservative interpretation of the results, it is clear that congruence plays an important role in the psychological treatment of individuals with PSS, warranting further research in this area.
Based on the first grouping strategy, end-of-treatment congruence was examined. Though this is arguably the richest time point— with patients and therapists having completed all therapy sessions to reflect back on and approximately four months’ time to understand their therapeutic relationship, examining only one time point is similarly reductive. For this reason, results should be interpreted cautiously.
Unfortunately, the present study was unable to investigate or control for therapist effects. As previously stated, the 50 therapists had caseloads ranging from one to 11. Typically, five patients per therapist are required in order to control for therapist effects in a meaningful way (Schiefele et al., 2017). The potential effects of therapists will be helpful to consider in future research.
Additionally, since GSA offers results that can be interpreted in more than one way, there are challenges associated with deciding how to cluster groups. The goal is to create only a few groups that are theoretically plausible with relatively equal distribution. Given that this is a subjective decision, there is a chance that bias has been introduced into these results. The authors aimed to minimize this bias by offering analyses of two plausible interpretations. With regard to the second grouping strategy, about one-third of dyads were excluded as they did not follow one of the four most common congruence patterns. Given that this amount of data was removed, interpretations of the results of this grouping strategy should be made cautiously.
Finally, the trial provided 3–5 preparatory intake sessions before providing 20 therapy sessions. It is uncommon for patients in the general population to receive this many therapy sessions, and it is possible that different results would be found if the same analysis were conducted with fewer therapy sessions.
Conclusion
The research that has been conducted on patient-therapist congruence in this population is extremely limited, so the present research has valuable implications both on a clinical and public scale despite the limitations listed above. Though there were many unique congruence patterns identified in this analysis, both congruent and incongruent, patients on average saw a meaningful improvement in their somatic symptom disability and severity by the end of treatment, which may be an encouraging finding for healthcare professionals with reservations regarding the success of treatment for PSS. The improvement in symptom disability and severity was found even in patients with higher baseline severity, disability, or other mental health comorbidities. A notable majority of dyads ended in strong positive congruence, indicating that even patients who enter therapy with high levels of symptom severity can have a positive outlook on therapy, and develop congruence regarding positive therapy progress perceptions. Though the findings remain inconsistent with regard to which kinds of treatment outcomes are most impacted by congruence, the preliminary evidence suggests that congruence and positive treatment perceptions are worthwhile and important for therapists to cultivate at the beginning of therapy, especially for PSS patients who may be entering therapy with more hopelessness.
Future studies should aim to understand and analyze the dynamic nature of PSS in relation to congruence and treatment outcomes. Studies with more time points will be better able to capture the typical fluctuations of somatic symptoms and make it easier to address cases of missing data. More time points may help researchers better understand whether a positive view of therapy causes symptom reduction, or whether symptom reduction causes a positive view. Additionally, further research in this area can help develop the understanding of PSS patient experiences, which can help professionals provide more validating care in the future, particularly for those professionals who feel less confident treating this population.
Declarations
Competing Interests
The authors declare no competing interests.
Conflict of Interest
None to report.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.