A Life-Stress, Emotional Awareness, and Expression Interview for Primary Care Patients With Medically Unexplained Symptoms: A Randomized Controlled Trial

Abstract

Objective

Lifetime trauma, relationship adversities, and emotional conflicts are elevated in primary care patients with medically unexplained symptoms (MUS), and these risk factors likely trigger or exacerbate symptoms. Helping patients disclose stressors, increase awareness and expression of inhibited emotions, and link emotions to physical symptoms may improve health. We developed an emotional awareness and expression interview that targets stressful life experiences and conflicts and then tested its effects on primary care patients with MUS.

Method

Patients (N = 75) with MUS were recruited at a family medicine clinic and randomized to an interview condition or treatment-as-usual (TAU) control condition. In a single 90-min interview in the clinic, the interviewer elicited disclosure of the patient’s stressors, linked them to the patient’s symptom history, and encouraged emotional awareness and expression about unresolved relationship trauma or conflict. At baseline and 6-week follow-up, patients completed self-report measures of their physical and psychological health.

Results

Analyses of covariance, controlling for baseline symptoms, compared patients in the interview condition with TAU at 6-week follow-up. Compared with TAU, the interview led to significantly lower pain severity, pain interference, sleep problems, and global psychological symptoms.

Conclusions

This study provides preliminary evidence for the value of integrating a disclosure and emotional awareness and expression interview into the primary care setting for patients with MUS.

Up to half of primary care visits are for patients who report somatic symptoms that are not adequately explained by structural, organic, or disease processes assessed by current biomedical tests and observations. Such patients may report individual somatic symptoms, such as musculoskeletal pain, head pain, abdominal pain, urogenital pain, dizziness, fatigue, cardiac palpitations, or dyspnea; or they may present with multiple comorbid symptoms and be diagnosed with irritable bowel syndrome (IBS), fibromyalgia (FM), chronic fatigue syndrome, or other syndromes (Abbass, Kisely, & Kroenke, 2009; Kroenke, 2003; Roelofs & Spinhoven, 2007; Steinbrecher, Koerber, Frieser, & Hiller, 2011). Such symptoms and syndromes have been given a range of labels, including “somatoform,” “somatic symptom,” “central sensitization,” “functional,” “psychosomatic,” “mind-body,” “psychophysiological,” or “medically unexplained.” The varied and sometimes pejorative nature of the nomenclature reflects the complexity of these presentations, the lack of understanding of mechanisms, and, perhaps, the challenges of clinicians in working with patients with these conditions. In this article, we adopt the term “medically unexplained symptoms” (MUS), because this label is in current use by many authors (Chew-Graham et al., 2017; Isaac & Paauw, 2014; Rosendal et al., 2017; Steinbrecher et al., 2011) and reflects how these conditions are typically identified in actual practice—by the failure of medical tests to identify structural or organic causes to explain them (Kroenke, 2003).

Although MUS appear to be heterogeneous, there are commonalities among them. MUS often overlap and have similar psychiatric comorbidity, family history, and other risk factors (Aaron & Buchwald, 2001; Kroenke & Rosmalen, 2006; Yunus, 2007). High rates of anxiety and depression have been observed in numerous clinical and epidemiologic samples (Barsky & Borus, 1999; Henningsen, Zimmermann, & Sattel, 2003). A meta-analytic review concluded that the association of MUS with depression and anxiety exceeds that observed in healthy controls as well as in patients with similar diseases that are medically explained (Henningsen et al., 2003).

Histories of psychological trauma, abuse, stressful life events, or interpersonal difficulties are relatively common among patients with MUS (Abbass et al., 2009; Barsky & Borus, 1999; Häuser et al., 2013; Jones, Power, & Macfarlane, 2009; Lampe et al., 2003). Indeed, a meta-analysis of 71 studies found that individuals reporting exposure to trauma were 2.7 times more likely to have medically unexplained syndromes than various control or comparison samples (Afari et al., 2014). For example, people with FM report elevated major life events and traumatic experiences (Häuser et al., 2013; Häuser, Kosseva, Üceyler, Klose, & Sommer, 2011), and sexual and physical abuse throughout life predict various pain conditions in adulthood (Häuser et al., 2011). Almost half of women with chronic urogenital pain or interstitial cystitis report physical or sexual abuse at some point in their lives, which is double the U.S. population estimate (Peters, Carrico, Ibrahim, & Diokno, 2008). Patients with IBS report higher rates of early life adversity, including trauma and physical, emotional, and sexual abuse, than healthy controls (Bradford et al., 2012; Drossman, 2011). Children exposed to physical and social adversity experience a 50% to 100% increase in the risk of chronic widespread pain in adulthood, and physically traumatic events and psychosocially challenging environments during development are associated with long-term increased risk of chronic widespread pain. Indeed, emotional stressors have been found to precipitate, exacerbate, or prolong medically unexplained conditions (Clauw & Chrousos, 1997; Häuser et al., 2011).

A few studies of trauma and stressful life experiences have compared patients with MUS with those with medically explained or organic diseases that have similar somatic physical symptom profiles. For example, patients with FM have higher levels of abuse and other childhood traumas than patients with rheumatoid arthritis (Castro et al., 2005). Similarly, patients with IBS report more childhood and adulthood abuse of various forms than patients with inflammatory bowel disease (Ali et al., 2000; Beesley, Rhodes, & Salmon, 2010). These comparisons suggest that psychological stress is linked to MUS specifically rather than simply having a physically symptomatic condition.

Several theoretical models linking psychosocial adversity or stress to MUS have been proposed (Gracely & Schweinhardt, 2015; van Houdenhove & Luyten, 2007; Yunus, 2007). Stress-induced changes in health behavior (e.g., diet, exercise, sleep) and stress-induced changes in physiology (e.g., musculoskeletal tension or sympathetic-adrenal medullary activation) could trigger or augment somatic symptoms. Importantly, a model that is garnering substantial support views MUS as a manifestation of central nervous system sensitization and augmentation, in which the brain and spinal cord are altered by stressful experiences, becoming sensitized to peripheral input and either generating or augmenting somatic symptoms (Clauw & Chrousos, 1997; Yunus, 2007).

Several factors appear to interfere with the resolution of people’s stressful experiences. Processes related to experiential avoidance—including difficulties identifying and labeling emotions, emotional suppression or inhibition, and failure to actively engage or cognitively work through negative emotional experiences—can prolong and maintain an excessive stress response, including posttraumatic stress disorder (Lind, Delmar, & Nielson, 2014; Lumley et al., 2011). Given that such avoidance processes prolong stress responses, engaging in the opposite behaviors—such as experiencing, expressing, and processing traumatic or conflicted experiences—may help resolve stress and reduce somatic symptoms. Several lines of research support this. Written or verbal emotional disclosure about private stressful experiences has small but positive benefits on health (Frattaroli, 2006), particularly for people with medically unexplained chronic pain conditions (Lumley, Sklar, & Carty, 2012) such as FM (Broderick, Junghaenel, & Schwartz, 2005) and chronic pelvic pain (Norman, Lumley, Dooley, & Diamond, 2004). Short-term, intensive psychodynamic psychotherapy—an intervention that encourages patients to confront avoided emotions and psychological conflicts—also has benefits for individuals with somatic symptom disorders (Abbass et al., 2009). There also are positive health effects of increasing emotional awareness (Hsu et al., 2010) and trauma processing (Lumley et al., 2008) for FM, eye movement and desensitization for back pain (Tesarz et al., 2013), anger awareness and expression for headaches (Slavin-Spenny, Lumley, Thakur, Nevedal, & Hijazi, 2013), and emotional awareness and expression training for IBS (Thakur et al., 2017).

Interventions such as these have typically been tested on self-selected people from the community who are seeking stress-reduction interventions to improve their symptoms. The primary care setting, in contrast, is where most people with MUS initially present for health care, and recent estimates suggest that MUS account for 40% to 49% of primary care visits (Haller, Cramer, Lauche, & Dobos, 2015). Yet the role of psychosocial stressors is rarely assessed or treated by medical staff in this setting. One early study revealed that only a small minority of patients shared their abuse histories with their physicians (Drossman et al., 1990). Interestingly, however, even mental health professionals located in primary care typically do not assess or treat psychological trauma and adversities. Rather, these professionals target psychiatric symptoms and diagnoses (e.g., anxiety, depression) and unhealthy behaviors, and their interventions are usually symptom management, motivational enhancement, and behavior change techniques. Consequently, the many primary care patients whose unresolved stress, trauma, and emotional conflicts may be contributing to their MUS are not adequately assessed or treated, thereby prolonging both physical and psychological symptoms and continued high health care utilization (Escobar, Waitzkin, Silver, Gara, & Holman, 1998).

We developed a novel interview that targets emotional disclosure, awareness, and expression about psychosocial trauma and life stress for patients with MUS, and we conducted a randomized controlled clinical trial of the effects of this interview on physical and psychological symptoms in a sample of primary care patients. We hypothesized that patients who received this interview would experience more improvement in physical symptoms (i.e., pain severity and interference, sleep problems) and psychological symptoms (i.e., depression, anxiety, and interpersonal sensitivity) than patients receiving primary care treatment as usual (TAU).

Method

Study Design

The study was a single-site, two-arm, randomized controlled trial comparing a life-stress interview with a no-interview, TAU control condition on health measures at a 6-week follow-up.

Participants

Adults (Ages 18 to 70) with MUS were recruited during routine visits at a family medicine clinic that is affiliated with a major university and houses a family medicine residency program. Consistent with the practice used elsewhere (Martin, Rauh, Fichter, & Rief, 2007), we identified patients with MUS by referral from clinic physicians or by combining a widely used somatization scale with further exclusions of disease processes that might account for the elevated symptoms. All patients were screened with the Patient Health Questionnaire-15 (Kroenke, Spitzer, & Williams, 2002), and those patients who scored above 10 (moderate range) were contacted by a research assistant to further determine their eligibility, prior to the patient leaving the clinic. The assistant then interviewed patients who scored above the cutoff for additional exclusion criteria: (a) the presence of disease or injury (e.g., autoimmune disease, bodily injury, serious infection, cancer, heart disease, chronic obstructive pulmonary disease, stroke) that might account for the elevated somatic symptoms, (b) planning to have a major medical procedure (e.g., surgery) in the next 2 months, or (c) various conditions that could interfere with the successful conduct of the interview (non-English-speaking, serious mental illness, dementia, intellectual impairment). Medical staff or medical records were consulted as needed to obtain more information on the results of medical testing, and medical staff also excluded several patients for additional reasons (e.g., presumed drug seeking, serious legal problems).

Procedure

The study was institutional review board approved by Wayne State University and registered with Clinicaltrials.gov (NCT02151500) before recruitment, which occurred from September 2014 through September 2015, with follow-up assessments completed in November 2015. Prior to recruitment, a person not associated with recruitment or data collection created a computer-based randomization scheme. Randomization was stratified by patient gender and specific interviewer in randomized blocks of three or six and conducted in a 2:1 ratio (interview: control) to ensure a larger sample in the interview condition, which would support later secondary analyses of interview content.

Prior to leaving the primary care clinic, patients who met study criteria and were interested in participating reviewed the study procedures and provided written, informed consent. Participants were informed that they would attend an interview aimed at exploring the links between their stress and health and learn tools for managing their stress. Later that day, patients received an e-mail asking them to complete baseline questionnaires, at home, using an online system (Qualtrics). Those patients who completed the baseline measures were subsequently contacted by the research assistant to reconfirm study criteria (e.g., still interested and available; no upcoming surgeries). If the patient still met criteria, the assistant opened a sealed envelope with the randomized condition assignment and informed the patient whether he or she was in the interview condition or in TAU condition (called the “delayed interview” for these patients, who were offered the interview after the follow-up measures were completed).

Patients assigned to the interview condition were asked to return to the clinic within 1 week of completing baseline measures, where they had the life-stress interview in an examination room. The interview consisted of one 90-min session with an interviewer—one of two female doctoral students in clinical psychology. Interviewers were trained and supervised by a licensed clinical psychologist, and all interviews were audio recorded for supervision and to ensure fidelity to the interview protocol. Patients in the TAU condition simply engaged in their regular health care during the 6-week period. Six weeks after randomization, all participants (interview and TAU) were asked to complete the same online measures that had been completed at baseline. Participants were paid $20 for completing each of the two assessment sessions, and $10 for the time and transportation needed to come to the clinic to conduct the interview.

Life-Stress Interview

The goals of the interview were to help patients (a) disclose stressful experiences and emotional conflicts that might be contributing to their symptoms, (b) explore associations between their stress and physical symptoms, and (c) experience and express their emotions related to these stressful situations. The interview followed a written manual or protocol (available upon request) and progressed through these four phases:

1. Review of lifetime medical symptoms (~10 min). Patients were briefly interviewed about their medical history, including the onset and development of their symptoms over their lives. As the patients spoke, interviewers sat next to them and took notes by creating a visual timeline or graph of the patients’ medical history over their lives.

2. Review of life stress and emotional conflicts (~30 min): Patients were asked about their history of stressful life events, including traumas, family and other relationship conflicts, and internal struggles or secrets. These events were added to the patients’ timeline of health problems, so that the patient could see temporal linkages between stressful experiences and the experience of symptoms or other health problems.

3. Emotional awareness and expression exercises (~35 to 40 min): Using a pictorial handout, the interviewer explained to the patients the links between stressful experiences, emotional avoidance, and symptoms. The interviewer then led patients through a series of exercises that targeted two core emotional/relational processes that are often conflicted: dominance (e.g., power, agency, including anger) and vulnerability (e.g., love, sadness, guilt, and other connecting emotions). These exercises encouraged participants to experience their emotions in their bodies, especially their muscles, and then express these emotions in the room, with vocal tone, facial, and physical expression, as if communicating in a direct and healthy way to the offending or conflict-related person.

4. Summary and discussion (~10 to15 min): The participants discussed their experience with the interviewer, including what they learned about their stress, emotions, and links to their symptoms.

In addition, several times during the interview, patients were asked to evaluate their pain or other symptoms, so that they might note any links between the topics they were discussing, the emotions they were activating and expressing, and immediate changes in their physical symptoms.

Outcome Measures

At baseline and at 6-week follow-up, patients completed reliable and valid self-report measures of primary outcomes (pain severity and pain interference) and secondary outcomes (sleep problems and psychological symptoms).

Pain severity and pain interference were assessed with the short form of the Brief Pain Inventory (Cleeland & Ryan, 1994). Pain severity was rated from 0 (no pain) to 10 (pain as bad as you can imagine) regarding current level, as well as highest, lowest, and average levels during the past week. The four items were internally consistent in this sample (baseline, α = .87; follow-up, α = .96) and averaged to yield a single pain severity score. The seven pain interference items were rated from 0 (no interference) to 10 (interferes completely), and were internally consistent in this sample (baseline, α = .90; follow-up, α = .97); the items were averaged to yield a single pain interference score. This is one of the most widely used pain measures and has excellent psychometric data, including reliability and a wealth of validity data demonstrating that its scales correlate as expected with other measures of pain and functioning (Cleeland, 2009; Cleeland & Ryan, 1994).

Sleep problems were assessed with the seven-item Insomnia Severity Index (Bastien, Vallières, & Morin, 2001), which measures the global severity of insomnia, including perceived daytime consequences. Items were rated from 0 (not at all) to 4 (extremely) and summed for a total score. This measure has acceptable internal consistency (α = .76 –.78) and concurrent validity against other measures of sleep (Tang, Wright, & Salkovskis, 2007), and had excellent internal consistency in this sample (baseline, α = .89; follow-up, α = .92).

Psychological symptoms were assessed with 18-item short form of the Brief Symptom Inventory (Derogatis & Melisaratos, 1983), which assesses symptoms of depression, anxiety, and interpersonal sensitivity over the past week. Items were rated from 0 (not at all) to 4 (severely) and averaged to yield a global symptom severity score. This measure is widely used and its reliability and validity have been demonstrated repeatedly (Derogatis, 1993), and it had excellent internal consistency in this sample (baseline α = .91; follow-up, α = .93).

Statistical Analyses

Sample size was determined by power analysis to test the difference between the interview and TAU conditions. Based on prior studies of emotional awareness and expression interventions (Burger et al., 2016; Slavin-Spenny et al., 2013), we estimated an effect size (ES) for this comparison halfway between medium (0.50 SD) and large (0.80 SD; in G-Power, an F value of 0.325). To obtain 80% power using an analysis of covariance (ANCOVA) with two-tailed alpha of .05 required a total sample size of 77 participants.

To assess the success of randomization, preliminary analyses compared the two conditions (interview and TAU) on demographics and baseline levels of outcome measures using t- and chi-square tests. Primary analyses were intent-to-treat of the full randomized sample, including participants who missed the interview or who did not provide follow-up data. Missing data were replaced via multiple imputation, which was conducted using all variables in this study as predictors—interview condition, demographics (age, sex, education, race, marital status), and baseline and follow-up values of the four outcome measures. Primary analyses on the multiply imputed data set compared the two conditions on each outcome measure using ANCOVA, covarying the baseline value of the outcome. Analyses of within-condition changes over time were calculated using paired t tests. ESs were calculated both within and between conditions. Within condition, d was calculated by subtracting the baseline M from the follow-up M and dividing by the SD of the change scores. Between conditions, ES was the standardized difference in change between conditions: ([interview follow-up minus baseline] minus [TAU follow-up minus baseline]) divided by the pooled standard deviation of change scores. All analyses were conducted with SPSS 23.0 and used two-tailed tests with alpha set at .05.

Results

Patient Characteristics and Preliminary Analyses

Figure 1 depicts patient flow through the study. A total of 310 patients were screened, but the majority did not meet study criteria (61.3%) or were not interested or able to participate (12.9%). Of the 80 patients who enrolled in the study and completed baseline measures, we did not randomize five patients, because they were unavailable, having surgery, or excluded by staff (including several patients after we stopped recruitment, resulting in us falling below the target of 77 randomized patients). A final sample of 75 patients met criteria and were randomized (interview, n = 49; control, n = 26).

Figure 1.

Flow of participants through the study.

As shown in Table 1, the randomized sample was primarily female, young to middle age, European American, not partnered, and had, on average, 2 years of college education (only 29.7% had a bachelor’s degree or higher). Patients had a wide range of somatic symptoms, usually involving pain, but only a minority reported any diagnosed syndrome (n= 12 had FM; n = 9 had IBS). The two conditions did not differ on demographics (see Table 1) or baseline levels of any outcome measures (see Table 2), suggesting that randomization created equivalent conditions.

Table 1.

Comparison of Conditions on Demographic Measures at Baseline

Variable	Full sample (n = 75)	Interview (n = 49)	TAU control (n = 26)	t/χ2	p
Age in years, M (SD)	39.16 (13.66)	39.12 (13.72)	39.23 (13.83)	0.03	.97**
Years of education, M (SD)	14.04 (1.85)	14.08 (1.99)	13.96 (1.57)	−0.27	.79*
Gender				0.11	.74*
Male, n (%)	10 (13.3)	7 (14.3)	3 (11.5)
Female, n (%)	65 (86.7)	42 (85.7)	23 (88.5)
Ethnicity				0.84	.66**
European American, n (%)	59 (78.7)	37 (75.5)	22 (84.6)
African American, n (%)	12 (16.0)	9 (18.4)	3 (11.5)
Other, n (%)	4 (5.3)	3 (6.1)	1 (3.8)
Marital status				2.15	.14*
Married, n (%)	18 (24.0)	11 (22.4)	7 (26.9)
Living together, n (%)	11 (14.7)	5 (10.2)	6 (34.6)
Divorced/separated, n (%)	21 (28.0)	17 (34.7)	4 (15.4)
Widowed, n (%)	2 (2.7)	2 (4.1)	0 (.0)
Never married, n (%)	23 (30.7)	14 (28.6)	9 (34.6)

Note. All tests were two-tailed. Chi-square test for ethnicity compared European American with African American and Other combined. Chi-square tests for marital status compared Partnered (married or living together) with all others combined. TAU = treatment as usual.

^*p < .01.

^**p < .001.

Table 2.

Comparison of Interview and TAU Control Conditions on Outcomes From Baseline to 6-Week Follow-Up, Between-Condition Analyses of Covariance, and Both Within- and Between-Condition Effect Sizes

Outcome measure	Time point	Interview (n = 49)	d within	TAU (n = 26)	d within	ES between	F	p
Pain severity	Baseline, M (SD)	4.91 (2.17)		4.82 (2.24)
	6-week, M (SD)	4.12 (2.46)		4.95 (1.83)
	6-week adj., M (SE)	4.09 (.22)		5.00 (.29)			6.08	.014
	Change, M (SD)	−.80 (1.40)	−0.54**	.13 (1.49)	0.08	−0.62
Pain interference	Baseline, M (SD)	5.23 (2.81)		4.55 (2.39)
	6-week, M (SD)	4.12 (2.97)		5.12 (2.79)
	6-week adj., M (SE)	3.96 (.34)		5.44 (.46)			6.53	.011
	Change, M (SD)	−1.10 (2.41)	−0.47*	.45 (1.63)	0.19	−0.71
Sleep problems	Baseline, M (SD)	16.04 (5.95)		15.58 (7.69)
	6-week, M (SD)	13.41 (7.70)		17.36 (6.32)
	6-week adj., M (SE)	13.28 (.75)		17.61 (1.00)			12.03	.001
	Change, M (SD)	−2.42 (5.08)	−0.47*	1.23 (3.20)	0.23	−0.85
Psychological symptoms	Baseline, M (SD)	1.27 (.83)		1.27 (.85)
	6-week, M (SD)	.87 (.87)		1.29 (.70)
	6-week adj., M (SE)	.87 (.10)		1.29 (.14)			6.08	.014
	Change, M (SD)	−.43 (.68)	−0.58**	.01 (.75)	0.01	−0.54

Note. 6-wk adjusted (adj.) M is adjusted for the baseline value of the outcome measure. d within is the effect size of change within each condition, which was calculated by subtracting the baseline M from the follow-up M and dividing by the SD of the change scores. Between conditions, ES was the standardized difference in change between conditions: ([interview follow-up minus baseline] minus [TAU follow-up minus baseline]) divided by pooled SD of change scores. TAU = treatment as usual; ES = effect size; SE = standard error.

^*p < .01.

^**p < .001.

Almost all patients assigned to the interview condition completed the interview, although two patients did not respond to scheduling attempts and did not receive the interview. Fully 67 of the 75 patients (89.3%) completed the trial, but eight patients (10.6%) did not complete the follow-up assessment (interview condition: 12.2%, n = 6, including the two patients who did not do the interview; TAU: 7.7%, n = 2). The only other missing data were the measure of sleep problems from one patient at follow-up only. The 67 completers did not differ significantly on demographic or baseline levels of the outcome measures from the eight noncompleters.

Primary Analyses

Table 2 presents the means and standard deviations for each outcome measure by condition at baseline and follow-up, the baseline-adjusted means and standard errors at the 6-week follow-up, and the results of the ANCOVAs comparing conditions. Also presented in Table 2 are the ESs for within- and between-condition comparisons.

The primary outcomes—pain symptom severity and interference—differed significantly between the two conditions at follow-up. The interview condition had lower pain severity than TAU, with a medium to large between-condition effect. The interview condition reduced pain severity significantly from baseline to follow-up—nearly 1 point on the 0 to 10 rating scale, whereas the controls did not change over time. The interview condition also had significantly lower pain interference than the control condition at follow-up, with a large between-condition effect. The interview condition significantly reduced pain interference over time, whereas the increase observed in controls was not significant.

Sleep problems showed a similar pattern. The interview condition had significantly lower sleep problems than TAU at follow-up, with a large between-condition effect. Again, the interview condition decreased significantly over time in sleep problems, whereas the increase observed in controls was not significant.

The interview led to significantly less psychological symptoms than TAU at follow-up, with a medium effect. The interview condition decreased significantly in psychological symptoms over time, whereas controls were unchanged.

Sensitivity Analyses

To test the sensitivity of the analyses to the imputation of missing data, we repeated the primary analyses on only those patients who completed the trial (n = 67, but n = 66 for sleep problems). The interview condition remained significantly superior to TAU on pain intensity, F(1, 64) = 5.33, p = .024, ES = −0.56; pain interference, F(1, 64) = 7.63, p = .007, ES = −0.70; sleep problems, F(1, 63) = 10.81, p = .002, ES = −0.78; and psychological symptoms, F(1, 64) = 6.98, p = 01, ES = −0.54.

Discussion

This randomized controlled trial assessed the effects of a single interview that targeted the awareness and expression of trauma, stress, and conflict-related emotions among family medicine patients with MUS. The results are encouraging. Compared with TAU, the interview reduced pain severity, pain interference, and sleep problems 6 weeks later, with medium to large ESs. These results suggest that a stress- and emotion-focused interview, conducted in the family medicine clinic, is a useful technique for reducing MUS. It appears that helping people confront rather than avoid their memories and emotions related to stressful experiences can reverse key maladaptive processes (e.g., emotional avoidance, failure to process and resolve stress) that are responsible for maintaining and prolonging stress-related physical symptoms. This finding—that directly targeting unresolved trauma, stress, and conflict reduces somatic symptoms such as pain—provides experimental support to the many correlational studies that show that stressful experiences are associated with, and likely trigger or exacerbate, pain and other symptoms (Abbass et al., 2009; Burger et al., 2016; Hsu et al., 2010; Lampe et al., 2003; Slavin-Spenny et al., 2013).

The interview also reduced psychological symptoms (anxiety, depression, and interpersonal sensitivity), with a medium ES. Research has shown that emotional expression can improve psychological symptoms by helping people regulate internal states (Ekman & Davidson, 1993), differentiate emotions to reduce stress (Davis, Zautra, & Smith, 2004), and enhance interpersonal functioning and prosocial behavior (Lopes, Salovey, Coté, Beers, & Petty, 2005). As such, an approach that enhances emotional awareness and expression may increase the ability of patients with MUS to preserve positive engagement, potentially leading to resolution of conflict, reduced fear of emotions, and improved relationships.

It is often a challenge for primary care physicians to help patients with MUS shift from pursuing medical explanations, tests, and treatments to adopting psychological models and self-management. Simply telling patients that their symptoms are “stress-related” is likely ineffective and perhaps experienced by patients as dismissive or invalidating. In contrast, the interview conducted in this study explored links between patients’ difficult life experiences and the onset or exacerbation of their symptoms. Moreover, the experience and expression of avoided emotions often changed patient’s somatic symptoms in the moment, providing the patient with immediate, firsthand proof of the mind–body connection. Such a personal, experiential demonstration of the role played by emotions in a patient’s somatic symptoms may be more convincing to the patient than trying to explain a mind–body model.

Several potential mechanisms may have led to the reduction in symptoms following the interview. First, the interview attempted to shift patients’ perspective and attributions about their somatic symptoms from a biomedical model to a central nervous system sensitization or augmentation model. Research has shown that adopting such a new explanatory model leads to reductions in chronic pain (Moseley & Butler, 2015). Second, the interview encouraged patients to disclose rather than keep private their stressful life experiences. Whereas secrecy or inhibition is generally detrimental to one’s health (Gross, 2013; Imami et al., 2017), reversing inhibition by disclosing such experiences has a range of health benefits (Clarke, 2016; Pennebaker & Smyth, 2016). Third, the interview sought to elicit the expression of specific primary or adaptive emotions related to stressful or conflicted experiences, not only anger but also vulnerable emotions such as sadness, guilt, or love. Such emotional expression in a private and safe venue was sometimes difficult for patients, some of whom experienced anxiety and occasionally temporary increases in somatic symptoms in the process. However, when conducted by a committed and empathic interviewer, this technique likely helped patients more clearly recognize links between their emotions and physical symptoms, overcome their fears of expressing their emotions, and possibly motivate them to change in the way they interact with others, such as being more honest and direct about their feelings and needs. These possible mechanisms, however, need to be tested.

This study challenges current approaches to working with MUS patients in primary care by presenting an alternative approach of directly targeting the stress or psychological factors underlying physical symptoms. Given the time and training limitations of most primary care physicians, there likely is value to integrating psychological specialists in these settings to promote comprehensive patient care. Current psychological interventions seek to manage or accept pain and other symptoms, typically by decreasing physiological and emotional arousal. This study of a one-session interview that enhances emotional awareness and expression for patients with MUS suggests that this technique can serve as a beneficial alternative to current practices, leading to substantial symptom reduction.

The study has several limitations. Although we recruited patients directly from a family medicine clinic and conducted the interview in the clinic, the sample was limited by self-selection into the study. Patients who chose not to participate, although relatively small in number, likely differ from those who did participate, such as being less open to psychological processes. It would be valuable to test this interview in routine clinical practice to see how it works on all patients with MUS. Also, the modest sample size (including falling two patients short of targeted sample size), purposeful imbalance in condition sizes, and slight loss of follow-up data limit the trial’s statistical power, and replication with larger samples is indicated. Future research should also include other health outcomes that may be affected by this interview, such as measures of interpersonal functioning, health care utilization, and self-management, as well as a formal assessment of possible adverse reactions to the interview. A longer follow-up would also be ideal to determine whether the benefits are transitory and short-lived, or perhaps even strengthened, because adaptive changes to patients’ emotional and relational systems spurred by this interview probably take additional time to occur. Although the sample’s education level reflects that of the larger population, and the predominance of women in the sample is consistent with other studies of MUS (Nimnuan, Hotopf, & Wessely, 2001), research is needed to test this interview with men and ethnic minorities specifically, who had lower representation in the current sample.

Moreover, given the heterogeneity of MUS, it is likely that only a subset of people will benefit from such an intensive, emotion-focused interview. Some patients may not be open to exploring and discussing their lives or may be limited in their ability to see psychological connections. Therefore, future research should examine patient characteristics that may moderate the effects of the interview so that clinical care can be appropriately tailored. Research on potential mediators (e.g., self-efficacy, insight, changes in emotional awareness and expression, changes in relationships) can illuminate how this interview achieves its outcomes. Finally, this study used a TAU control condition, which controlled for many threats to internal validity but did not control for nonspecific aspects of the interview (e.g., support, interaction with interviewer). A credible and active control condition (e.g., educational interview) or alternative comparison condition (e.g., relaxation training) is needed to determine the specificity of the emotional awareness and expression components to the improved symptoms.

In conclusion, this randomized controlled trial provides evidence that an interview that helps patients with MUS disclose their stressful experiences, become aware of the links between their stress and health, and engage in adaptive emotional expression can substantially improve both somatic and psychological symptoms. Patients who had this novel interview experienced reductions in their pain severity, pain interference, sleep problems, and psychological symptoms. These findings suggest the importance of integrating a more intensive focus on life stressors and emotional processes in assessment and intervention with MUS in primary care. This life-stress, emotional awareness, and expression interview may be a beneficial alternative to current approaches that seek only to reduce negative emotions and physiological arousal or simply manage pain or other symptoms.