Does Addressing Mental Health During a Musculoskeletal Specialty Care Visit Affect Patient-rated Clinician Empathy?

Abstract

Background

Unhelpful thoughts and feelings of worry or despair about symptoms account for a notable amount of the variation in musculoskeletal symptom intensity. Specialists may be best positioned to diagnose these treatable aspects of musculoskeletal illness. Musculoskeletal specialists might be concerned that addressing mental health could offend the patient, and avoidance might delay mental health diagnosis and treatment. Evidence that conversations about mental health are not associated with diminished patient experience might increase specialist confidence in the timely diagnosis and initial motivation to treat unhelpful thoughts and feelings of worry or despair.

Questions/purposes

Using transcripts of videotaped and audiotaped specialty care visits in which at least one instance of patient language indicating an unhelpful thought about symptoms or feelings of worry or despair surfaced, we asked: (1) Is clinician discussion of mental health associated with lower patient-rated clinician empathy, accounting for other factors? (2) Are clinician discussions of mental health associated with patient demographics, patient mental health measures, or specific clinicians?

Methods

Using a database of transcripts of 212 patients that were audio or video recorded for prior studies, we identified 144 transcripts in which language reflecting either an unhelpful thought or feelings of distress (worry or despair) about symptoms was detected. These were labeled mental health opportunities. Patients were invited on days when the researcher making video or audio records was available, and people were invited based on the researcher’s availability, the patient’s cognitive ability, and whether the patient spoke English. Exclusions were not tracked in those original studies, but few patients declined. There were 80 women and 64 men, with a mean age of 45 ± 15 years. Participants completed measures of health anxiety, catastrophic thinking, symptoms of depression, and perceived clinician empathy. Factors associated with perceived clinician empathy and clinician discussion of mental health were sought in bivariate and multivariable analyses.

Results

Greater patient-rated clinician empathy was not associated with clinician initiation of a mental health discussion (regression coefficient 0.98 [95% confidence interval 0.89 to 1.1]; p = 0.65). A clinician-initiated mental health discussion was not associated with any factors.

Conclusion

The observation that a clinician-initiated mental health discussion was not associated with diminished patient ratings of clinician empathy and was independent from other factors indicates that generally, discussion of mental health does not harm patient-clinician relationship. Musculoskeletal clinicians could be the first to notice disproportionate symptoms or misconceptions and distress about symptoms, and based on the evidence from this study, they can be confident about initiating a discussion about these mental health priorities to avoid delays in diagnosis and treatment. Future studies can address the impact of training clinicians to notice unhelpful thoughts and signs of distress and discuss them with compassion in a specialty care visit; other studies might evaluate the impact of timely diagnosis of opportunities for improvement in mental health on comfort, capability, and optimal stewardship of resources.

Introduction

Background

Patients with undiagnosed and undertreated unhelpful thinking or feelings of worry and despair regarding symptoms experience greater discomfort and greater incapability in proportion to the intensity of those thoughts and feelings [30, 35, 48]. Musculoskeletal specialists may be the most qualified to notice that musculoskeletal discomfort and incapability are disproportionate to the patient’s pathophysiology—a strong indication of unhelpful thoughts (misinterpretations) or undue worry or despair about symptoms [11, 29, 35]. In other words, musculoskeletal clinicians can play a key role in the prompt diagnosis, treatment, and prioritization of a patient’s opportunities for improved mental health.

Rationale

Clinicians may intentionally avoid discussing mental health owing to a lack of training and practice (feelings of inadequacy), discomfort with mental health topics, and fear of offending the patient, among other reasons [3, 28, 32, 42, 45]. Although some patients may misinterpret a musculoskeletal specialist’s intentions when a discussion about thoughts and feelings is raised, there is some evidence that most people welcome it and may feel the lack of it [12, 14, 34]. A few bad experiences might skew surgeon behavior toward not addressing unhelpful thinking and feelings of worry or despair. Evidence that, on average, the patient’s experience of receiving musculoskeletal specialty care is not harmed by discussions of mental health could embolden musculoskeletal specialists to take interest in learning how to develop skills, compassion, and strategies for guiding people toward considering a comprehensive approach to health. Using a set of transcripts of specialty care visits from prior studies [2, 36], we had the opportunity to measure the relationship between patient ratings of clinician empathy (patient experience), patient language reflecting an unhelpful thought about symptoms or feelings of worry or despair (mental health opportunity), and the clinician addressing that mental health opportunity (mental health discussion).

Questions

Among patient transcripts of musculoskeletal specialty care visits with patients who had at least one documented unhelpful thought (misconception) or instance of worry or despair, we asked: (1) Is clinician discussion of mental health associated with lower patient-rated clinician empathy, accounting for other factors? (2) Are clinician discussions of mental health associated with patient demographics, patient mental health measures, or specific clinicians?

Materials and Methods

Study Design and Setting

This study was a retrospective, secondary analysis of transcripts of audio and video recordings collected for three other studies, two of which have been published [24, 38]. We analyzed the transcripts of video records of 144 adult patients seeking musculoskeletal specialty care. The transcripts were made by researchers who were not involved in this study. During the informed consent process, patients were assured their participation in this study was voluntary, their responses would be anonymous, and participation or lack thereof would not affect the care they received at our institution. After the visit, patients completed a survey including demographics and several questionnaires. Exclusions were not tracked, but few people declined.

Clinician discussions of mental health consisted of either noticing symptoms of worry or despair or gentle reorientation of an unhelpful thought (misconception) about symptoms, with or without offering social work services. Twenty-eight percent (40 of 144 transcripts) had at least one clinician discussion of mental health. These were rated by three observers (MN, SRP, and RL), with resolution of all disputes by a senior author (DR). The interrater reliability score was 0.77 for identifying patient clues and 0.87 for identifying a mental health discussion.

Participants

The original study inclusion criteria were adults aged 18 years and older, new and returning patients visiting a musculoskeletal specialist, and English-language fluency and literacy. The exclusion criterion was cognitive dysfunction precluding questionnaire completion.

The specific inclusion criterion for this study was at least one instance in which patient language indicated a mental health opportunity in the form of an unhelpful thought about symptoms or feelings of worry, despair, or stress. One hundred forty-four patients with a mean age of 45 ± 15 years were included in this study (Table 1). Eleven clinicians were involved in these encounters: Four were videoed with 18 or more patients (two surgeons, a nurse practitioner, and a physician assistant) and the other seven were videoed with seven or fewer patients.

Table 1.

Patient demographics, clinician factors, and baseline scores (n = 144)

Variable	Value
Age in years, mean ± SD	45 ± 15
Gender, % (n)
Women	56 (80)
Men	44 (64)
Race or ethnicity, % (n)
White	57 (82)
Latino	22 (31)
Black	11 (16)
Other	10 (15)
Marital status, % (n)
Married or partnered	50 (72)
Single	35 (50)
Divorced or widowed	15 (22)
Education, % (n) (n = 141)a
High school or lower	29 (41)
2-year college	20 (28)
4-year college	25 (35)
Postgraduate degree	26 (37)
Work status, % (n)
Employed	59 (85)
Unemployed	13 (19)
Disabled	12 (17)
Retired	7.6 (11)
Other	8.3 (12)
Extremity, % (n) (n = 133)a
Upper	63 (84)
Lower	37 (49)
Diagnosis of traumatic condition, % (n)
No	79 (114)
Yes	21 (30)
Clinician factors
Clinic type, % (n)
University employed	79 (114)
Other practice settings	21 (30)
Clinician, % (n)
1	26 (37)
2	19 (28)
3	19 (28)
4	13 (18)
5	4.9 (7)
6	2.8 (4)
Other	15 (22)
Clinician-initiated discussion about mental health, % (n)
No	72 (104)
Yes	28 (40)
Mental health and empathy measures
PROMIS Depression score (n = 138)a, mean ± SD	49 ± 10
SHAI-5 score (n = 140)a, median (IQR)	10 (8-12)
JSPPPE score (n = 142)a, median (IQR)	31 (26-35)

^aThe number of records is indicated in parentheses for all variables with missing values. PROMIS = Patient-Reported Outcomes Measurement Information System; SHAI-5 = Short Health Anxiety Inventory, five-item; JSPPPE = Jefferson Scale of Patient Perceptions of Physician Empathy.

Outcome Instruments Used

We used the five-item Short Health Anxiety Inventory to measure health anxiety (symptoms of hypochondriasis) [32]. Scores range from 0 to 20, with higher scores indicating greater health anxiety.

The four-item version of the Pain Catastrophizing Scale was used to measure unhelpful thoughts in the form of catastrophic thinking [9, 23]. Scores range from 0 to 16, with higher scores indicating greater catastrophic thinking.

The Patient-Reported Outcome Measurement Information System depression computer adaptive test was used to measure symptoms of depression. Patient-Reported Outcome Measurement Information System scores are scaled as t-scores, with a score of 50 representing the population mean. Each 10 points above or below 50 represent one standard deviation [7, 8].

The Jefferson Scale of Patient Perceptions of Physician Empathy was used to measure the patient’s perception of the clinician empathy. The Jefferson Scale of Patient Perceptions of Physician Empathy has five items rated on a seven-point Likert scale from 1 (strongly disagree) to 7 (strongly agree) [15, 17]. Higher scores correlate to greater perceived clinician empathy [18].

Statistical Analysis, Study Size

Descriptive statistical analyses were performed for all participants. In a bivariate analysis of factors associated with patient-rated clinician empathy, for categorical explanatory variables, we used Wilcoxon Mann-Whitney tests, and we used Kruskal-Wallis H tests, where appropriate, for continuous variables. Spearman rank correlation coefficients were calculated. To address the secondary null hypothesis (factors associated with a clinician mental health discussion and a researcher-rated mental health opportunity), Fisher exact tests and chi-square tests were performed for categorical variables, and t-tests or Wilcoxon Mann-Whitney tests were used for continuous explanatory variables. All variables that had a p value below 0.10 were moved to a multivariable regression analysis. Separate multilevel Poisson and logistic regression models were constructed. We accounted for nesting of patient-rated clinician empathy within clinicians because we expected a priori that their ratings would be intercorrelated. Regression coefficients, odds ratios, standard errors, and p values are reported. Alpha was set at 0.05. The variables of age, five-item Short Health Anxiety Inventory, and Patient-Reported Outcome Measurement Information System scores were used as auxiliary variables to predict missing values, and we used a Poisson regression imputation method.

An a priori sample size calculation demonstrated that 168 patients would yield 80% statistical power based on a linear regression with eight explanatory variables, given that a clinician-initiated discussion about mental health explains 8% of the variation in patient-rated empathy, and given that the full model explained 15% of the total variation. We did not perform a post hoc power analysis because of the strong relationship between observed (post hoc) power and detected p values [10].

Results

Is Clinician-initiated Discussion of Mental Health Associated With Lower Clinician Empathy Ratings?

Clinician-initiated discussion of mental health was not associated with a lower rating of clinician empathy. Controlling for factors from the bivariate analysis including symptoms of depression and work status (Supplemental Table 1; http://links.lww.com/CORR/A994), we found no relationship between greater patient-rated clinician empathy and clinician initiation of a mental health discussion (regression coefficient 0.98 [95% confidence interval 0.89 to 1.1]; p = 0.65) (Table 2).

Table 2.

Factors associated with patient-rated clinician empathy^a

Variable	Regression coefficient (95% CI)	Standard error	p value
Clinician-initiated discussion about mental health
No	Reference value
Yes	0.98 (0.89 to 1.1)	0.049	0.65
PROMIS Depressionb	1.0 (0.99 to 1.0)	0.0020	0.47
Work statusc
Employed	Reference value
Unemployed	0.91 (0.81 to 1.0)	0.057	0.14
Disabled	0.94 (0.82 to 1.1)	0.061	0.31
Other	0.93 (0.83 to 1.0)	0.053	0.19

A multilevel Poisson regression analysis was performed.

^aAccounting for nesting by clinicians.

^bSix patients (4.2%) had missing values for the PROMIS Depression and were omitted from the final model.

^cThe work status of “retired” was pooled with “other.” PROMIS = Patient-Reported Outcomes Measurement Information System.

Are Clinician Discussions of Mental Health Associated With Patient Factors or Specific Clinicians?

We found no patient related-factors were associated with clinician initiation of a mental health discussion. Accounting for potential confounding using the multivariable binomial regression analysis, we found no factor was associated with clinician-initiated mental health discussions (Table 3).

Table 3.

Factors associated with a clinician-initiated discussion about mental health^a

Variable	Odds ratio (95% CI)	Standard error	p value
Age	0.99 (0.95 to 1.0)	0.020	0.60
Race or ethnicity
White	Reference value
Black	0.85 (0.18 to 4.0)	0.67	0.84
Latino	1.1 (0.32 to 3.6)	0.66	0.90
Other	2.2 (0.42 to 12)	1.9	0.35
Marital status
Married or partnered	Reference value
Single	2.0 (0.65 to 6.3)	1.1	0.22
Divorced or widowed	2.5 (0.60 to 11)	1.8	0.21
Educationb
High school or lower	Reference value
2-year college	2.5 (0.66 to 9.5)	1.7	0.18
4-year college	0.66 (0.16 to 2.7)	0.47	0.56
Postgraduate degree	0.24 (0.046 to 1.2)	0.20	0.087
Extremityb
Lower	Reference value
Upper	0.83 (0.19 to 3.5)	0.61	0.80
Clinic type
University employed	Reference value
Other practice settings	0.070 (0.0047 to 1.0)	0.096	0.053
PROMIS Depressionb	1.0 (0.98 to 1.1)	0.023	0.35

A multilevel logistic regression analysis was performed.

^aAccounting for nesting by clinicians.

^bThirteen patients had missing data for “extremity” and six for “PROMIS Depression,” and 17 patients were omitted from the final model. PROMIS = Patient-Reported Outcomes Measurement Information System.

Discussion

Background and Rationale

There is evidence patients convey unhelpful thoughts about symptoms, symptoms of distress, and sources of stress via verbal and nonverbal cues [5, 27, 46], and specialists notice these cues [39]. However, specialists may hesitate to discuss mental health because of fear of offending patients, lack of training and confidence, and perceived lack of time [28, 42, 45]. This analysis of transcripts of musculoskeletal specialty care visits found that clinician-initiated discussions of mental health were not, on average, associated with diminished ratings of patient-rated clinician empathy. In other words, it is possible to discuss mental health without offending patients.

Limitations

First, the patients were predominantly English-speaking, so they may not represent patients in other settings. However, there is sufficient diversity in the response and explanatory variables for meaningful associations that are likely reproducible in other settings. Second, transcribers did not observe or describe body language, which might have added notable mental health opportunities. In our experience, verbal and nonverbal signs of unhelpful thoughts and symptoms of worry or despair are relatively concordant; thus, a study using hand and arm postures and other types of body language would likely have similar findings [4, 13, 20]. Third, people behave differently under observation and may not feel as open to share deep personal thoughts in front of a camera—a type of Hawthorne effect. However, there is evidence that Hawthorne effects have limited influence on experimental measurements of important associations [21, 22]. Fourth, there may be some subjectivity in the identification of unhelpful thoughts and feelings of distress. Our use of consensus of trained raters with resolution by an expert has precedent for this type of study, and there was little debate about the presence or absence of mental health opportunities and good reliability. Fifth, some readers might not be comfortable with the concept that some thoughts about symptoms are unhelpful, such as the thought that pain always means the condition is getting worse, or that a patient must be pain-free to maintain his or her cherished activities. However, there is notable evidence supporting these types of thoughts as largely inaccurate and strongly associated with greater symptom intensity and greater magnitude of incapability [30, 33, 41]. Sixth, the Jefferson Scale of Patient Perceptions of Physician Empathy has a Poisson-type distribution because of the strong ceiling effects that are typical of current patient-reported experience measures. This ceiling effect in the questionnaire leads to less variation in scores and less accurate measurement of the construct (patient-rated clinician empathy). Having a Gaussian distribution would help detect smaller effect sizes, but not having one does not diminish the results of our study, because large differences that are potentially important would still be noted, regardless of ceiling effects. Finally, this study addressed clinician behavior using patient measures. Building blocks such as this study can establish a foundation for future research regarding clinician strategies associated with better patient health, in part by allowing for comprehensive diagnosis and treatment.

Is Clinician-initiated Discussion of Mental Health Associated With Lower Clinician Empathy Ratings?

The observation that clinician initiation of a discussion about mental health was not associated with a lower rating of clinician empathy suggests that such discussions can be safe and appropriate. This addresses one of the major barriers to comprehensive biopsychosocial care of musculoskeletal illness and allows us to focus on others such as training and practice to develop confident and effective communication strategies and develop systems with psychosocial resources that are readily available to patients with musculoskeletal illness [26, 28, 40, 45]. Supportive evidence includes a cross-sectional study of patients seeking musculoskeletal specialist care that investigated the demographic and mental health factors associated with the duration of expertise transfer and factors associated with patient satisfaction and perceived clinician empathy [38]. In that study, greater perceived clinician empathy and satisfaction with the visit were correlated with less health anxiety, and longer expertise transfer was correlated with catastrophic thinking. Another study, a qualitative analysis of hospital clinicians, sought to describe patient-clinician admission encounters of hospital-based physician responses to patients’ verbal expressions of negative emotion and identify patterns of further communication associated with different responses and found that responding to expressions of negative emotion neutrally or with statements focusing on emotion elicits more clinically relevant information and is associated with more positive patient-clinician relationship and care outcomes [1]. There is also evidence from interviews of patients in primary care that discussing mental health can increase patient willingness to consider treatment [16], which may also apply to musculoskeletal specialty care.

Are There Factors That Influence Clinician Discussion of Mental Health?

The observation that more than one in four patients with a clinic visit had a clinician-initiated mental health discussion and no factors were associated with such discussions suggests clinicians can be prepared for a mental health discussion with each patient. Some degree of misinterpretation and distress regarding symptoms is a normal aspect of human illness behavior that can be anticipated and planned for as part of a comprehensive and effective health strategy [25, 35]. The observation that measures of patient unhelpful thinking and feelings of despair regarding symptoms are not associated with mental health discussions suggests clinician factors, rather than patient factors, may be the key drivers of whether these factors become conversation topics—a line of inquiry worth pursuing. For example, in a study that anticipated misconceptions and feelings of distress, primary care physicians randomized to communication skills training documented more psychosocial problems, engaged in more strategies for managing emotional problems, and scored higher in clinical proficiency, and their patients reported reduced emotional distress for as long as 6 months [31]. There is some evidence that clinician factors may account for more of the variation in patient experience than patient factors [37]. Perhaps because of the study design using transcripts, this study was not able to demonstrate relationships between patient factors and communication strategies identified in prior studies. Examples of such evidence include a correlation between longer duration of expertise transfer and greater patient catastrophic thinking [38] and correlations between experience measures and feelings of distress [6, 38, 43, 44]. There are likely important modifiable patient and clinician factors to be addressed to optimize experience and outcomes of care. Future studies can focus on clinician factors associated with more confident discussions of mental health and better patient experience.

Conclusion

The finding that specialist initiation of a discussion about mental health does not diminish patient experience can make musculoskeletal clinicians more comfortable cultivating and practicing communication strategies for discussions about common misinterpretations of symptoms and feelings of worry or despair about symptoms. The finding that discussion of mental health was not associated with measured patient unhelpful thinking and feelings of distress indicates there is likely important variation on the clinician side that can increase opportunities for improved communication and care strategies. Studies similar to this one can embolden health systems and their care teams to understand episodes of patient offense as important opportunities for improvement in care and communication strategies [19, 47], but not as representing the average patient’s openness to discuss thoughts and feelings. The next steps might include an investigation of the most effective tactics for musculoskeletal specialty clinicians to become aware of patient unhelpful thinking and feelings of distress regarding symptoms, make these comfortable topics of discussion, and guide people to the most effective comprehensive treatments that address pathophysiology, mindset, and circumstances.