Do Unhelpful Thoughts or Confidence in Problem Solving Have Stronger Associations with Musculoskeletal Illness?

Abstract

Background

Measures of unhelpful thoughts and distress correlate with the intensity of pain and the magnitude of incapability among people seeking musculoskeletal specialty care. In this evolving knowledge area, we want to be sure we have not neglected other important mental health factors. This study addressed how measures of confidence in problem solving as well as past and current ability to achieve goals account for variation in symptoms and capability independent of unhelpful thoughts and distress.

Questions/purposes

(1) Are measures of confidence in problem solving ability and past and current ability to achieve goals regarding future outcomes associated with variation in capability, independent of measures of symptoms of depression and anxiety (distress) and measures of unhelpful thoughts (worst-case thinking, negative pain thoughts)? (2) Are these measures independently associated with variation in pain intensity? (3) Are these measures associated with measures of symptoms of depression, symptoms of anxiety, and unhelpful thoughts?

Methods

Over a 7-month period during the pandemic, we enrolled sporadically from the offices of four surgeons treating patients who sought care for various upper and lower extremity conditions. We invited approximately 200 adult new and returning patients to participate (the number of invitations was not formally tracked) and 187 accepted. Thirty-one were excluded due to markedly incomplete entries (related to a problematic attempt to use the patient’s cell phone to complete questionnaires as a pandemic work around), leaving 156 for analysis. Patients completed an 11-point ordinal rating of pain intensity, two measures of unhelpful thoughts (the Pain Catastrophizing Scale and the Negative Pain Thoughts Questionnaire), the Adult Hope Scale to measure past and current ability to achieve goals, the Personal Optimism and Self-Efficacy Optimism Scale to measure confidence in problem solving ability, the Patient-reported Outcomes Measurement Information System (PROMIS) computer adaptive test to measure symptoms of anxiety, the PROMIS computer adaptive test to measure symptoms of depression, and the PROMIS physical function computer adaptive test to assess the magnitude of capability. All questionnaires were validated in previous studies. We used bivariate analyses to identify factors associated with magnitude of capability, pain intensity, confidence in problem solving ability, and past and current ability to achieve goals. All factors with a p value of less than 0.1 were included in multivariable analyses to seek associations between these measures accounting for confounders. We reported partial η² as a measure of effect size for all multivariable regression models. The following rules of thumb are used to interpret values for partial η²: a value of 0.01 = small, 0.06 = medium, and values of 0.14 and higher show large effect size.

Results

Greater capability was modestly associated with fewer negative pain thoughts (β = -0.63 [95% CI -1.0 to -0.22]; standard error = 0.20; partial η² = 0.06; p = 0.003) and no self-reported comorbidities (β = 2.6 [95% CI 0.02 to 5.3]; standard error = 1.3; partial η² = 0.03; p = 0.048) after controlling for education, symptoms of depression and anxiety, worst-case thinking, as well as past and current ability to achieve goals. In a similar multivariable model, greater pain intensity was modestly associated with greater worst-case thinking (β = 0.33 [95% CI 0.20 to 0.45]; standard error = 0.06; partial η² = 0.16; p < 0.001) and established patients (β = -1.1 [95% CI -1.8 to -0.31]; standard error = 0.38; partial η² = 0.05; p = 0.006). In another similar multivariable model, having more confidence in problem solving ability had a limited association with higher ratings of past and current ability to achieve goals (β = 0.15 [95% CI 0.09 to 0.21]; standard error = 0.03; partial η² = 0.13; p < 0.001). In a final multivariable model, lower past and current ability to achieve goals was independently associated with having greater symptoms of depression (β = -0.45 [95% CI -0.67 to -0.23]; standard error = 0.11; partial η² = 0.1; p < 0.001) and more negative pain thoughts (β = -0.49 [95% CI -0.89 to -0.09]; standard error = 0.20; partial η2 = 0.04; p = 0.02).

Conclusion

The observation that unhelpful thoughts about symptoms are more strongly associated with symptom intensity than past and current ability to achieve goals and confidence in problem solving ability add to the evidence that attentiveness to unhelpful thinking is an important aspect of musculoskeletal health. Musculoskeletal specialists can prioritize communication strategies such as relationship building and motivational interviewing that develop trust and facilitate reorientation of common unhelpful thoughts.

Level of Evidence

Level II, prognostic study.

Introduction

Many psychological and social factors are associated with symptom intensity and the magnitude of incapability; these metrics of symptom intensity can be quantified using patient-reported outcome measures (PROMs). For instance, unhelpful thoughts and distress account for a large amount of the variation in symptom intensity and magnitude of capability measured by PROMs [15]. Unhelpful thoughts about symptoms include worst-case thinking (commonly measured using the Pain Catastrophizing Scale) [22], fear of painful movement (Tampa Scale of Kinesiophobia) [18, 20], and misinterpretation of symptoms (cognitive errors or negative pain thoughts) [10]. The concept of “distress” as we are using it includes symptoms of worry (anxiety) and despair (depression).

In this evolving knowledge area, we want to be sure we have not neglected other important mental health factors that might help people get and stay healthy. There is evidence that an individual’s perceived agency (success with goal achievement in the past, present, and future) [34] is associated with greater engagement in healthy behaviors, more effective coping strategies, lower pain intensity, and adherence to prescribed medication regimens [33]. Likewise, an individual’s confidence in solving problems is associated with satisfaction with life and fewer depression symptoms [17]. If these measures are associated with variation in symptom intensity and magnitude of capability in patients with musculoskeletal illness independent of what is accounted for by measures of unhelpful thoughts and distress, it could lead to more complete assessment and treatment of mental health and improved health.

We therefore asked: (1) Are measures of confidence in problem solving ability and past and current ability to achieve goals regarding future outcomes associated with variation in capability, independent of measures of symptoms of depression and anxiety (distress) and measures of unhelpful thoughts (worst-case thinking, negative pain thoughts)? (2) Are these measures independently associated with variation in pain intensity? (3) Are measures of past and current ability to achieve goals and confidence in problem solving ability associated with measures of symptoms of depression, symptoms of anxiety, and unhelpful thoughts?

Patients and Methods

Study Design and Setting

This cross-sectional study was performed at musculoskeletal specialty offices in an urban setting in the United States during a 7-month period (August 2020 to March 2021). Enrollment was paused and slowed by the coronavirus 2019 pandemic. As this study occurred during the COVID pandemic, 80% of visits were in-person and 20% were video telemedicine. A research team not involved in patient care invited all new and established patients older than 18 years seeking musculoskeletal care to participate in the study. Studies consistently demonstrate little or no difference between new and returning patients in cross-sectional studies testing relationships between measures [1, 6, 21, 29] (Supplementary Material 1; http://links.lww.com/CORR/A634), and so we pooled new and returning patients in our analyses.

Two hundred patients were invited, and less than 10% of invitees declined participation. We excluded patients if they did not speak English or Spanish or had cognitive dysfunction precluding completion of the questionnaires. One hundred eighty-seven patients evaluated by one of four surgeons for various upper and lower extremity conditions were enrolled in the study, and 31 were excluded because of incomplete forms, likely influenced by the changes in enrollment processes because of the pandemic. The remaining 156 records were included in the analysis. During the height of the COVID pandemic, we used a QR code to access the survey on the patient’s phone instead of using shared tablets. This workflow led to a higher number of incomplete surveys. We think people may have been more likely to leave before completing the survey when they did not have the obligation to return the tablet.

Patients were assured that participation in this study was voluntary and that responses were anonymous and would not affect the care they received at our institution. They were informed that they could discontinue the study at any time during the process.

Description of Study Population

One hundred fifty-six patients were included in the analysis, 56% (88) of whom were women. The median (range) age was 51 years (35 to 61); 55% (86) were new patients (Table 1). We included both new and established patients in this study after noticing that the visit status was rarely a factor in our previous studies. We compared factors associated with being a new patient versus an established one and did not find any significant difference between two groups in terms of past and current ability to achieve goals and confidence in problem solving ability.

Table 1.

Patient demographic and clinical characteristics (n = 156)

Variable	% (n)
Categorical variables
Gender
Women	56 (88)
Education level
High school diploma or less	18 (28)
Some college or more	81 (127)
Marital status
Single	34 (53)
Married or partner	47 (74)
Divorced or separated	15 (23)
Other	3.2 (5)
Work status
Full- or part-time employed	47 (74)
Not working	52 (81)
Income in USD
Less than 15,000	24 (37)
15,000-29,999	10 (15)
30,000-49,999	16 (25)
50,000-99,999	12 (19)
More than 100,000	37 (57)
Comorbidities
None	43 (67)
One or more	57 (89)
New patient
Yes	55 (86)
Continuous variables	Median (IQR)
Age in years	51 (35-61)
PROMIS depression CAT	51 (46-57)
PROMIS anxiety CAT	56 (51-62)
PROMIS physical function CAT	44 (38-50)
PCS-4	4 (1-8.5)
NPTQ-4	3 (0-7)
Adult Hope Scale	53 (45-59)
Optimism scale	25 (23-27)
Pain intensity	4 (1-6.5)

Categorical variables presented as % (n); continuous variables are presented as median (IQR); PROMIS = Patient-reported Outcomes Measurement Information System; CAT = computer adaptive test; PCS-4 = Pain Catastrophizing Scale (four-item version); NPTQ-4 = Negative Pain Thoughts Questionnaire (four-item version).

Outcomes Instruments Used

Participants who opted to participate were instructed to complete the following surveys: demographics, 11-point ordinal measure of pain intensity, Pain Catastrophizing Scale, Negative Pain Thoughts Questionnaire, Adult Hope Scale, Personal Optimism and Self-efficacy Optimism Scale, Patient-reported Outcomes Measurement Information System (PROMIS) Anxiety computer adaptive test (CAT), PROMIS Depression CAT, PROMIS Physical Function CAT, and additional information (including diagnosis, clinic name, and care provider). Participants also completed a demographics survey (age, gender, marital status, highest education level, work status, yearly household income in USD, visit type [new or established], and other health conditions [none or present]).

Patients rated their current level of pain intensity on a validated 11-point ordinal scale from 0 (no pain) to 10 (worst pain ever) [4].

We used two instruments to measure unhelpful thoughts regarding symptoms. Based on knowledge to date, we have found it interesting and worthwhile to track the cognitive (thought) aspects of mental health separately from the emotional (distress) aspects of mental health. Thoughts that represent at least partial misinterpretation of symptoms have a notable and consistent association with greater pain intensity and lower capability and are therefore described as unhelpful. Symptoms of distress (like symptoms of depression and anxiety) also have a strong, consistent, and somewhat independent association with symptom intensity. We used the validated short form of the four-question version of Pain Catastrophizing Scale to assess the degree of worst-case thinking regarding symptoms [22]. The questions are rated on a 5-point Likert scale, with higher scores indicating greater worst-case thinking. The validated short form of the Negative Pain Thoughts Questionnaire quantifies the extent to which people are experiencing unhelpful thoughts in the form of common cognitive errors (things that seem true but are not such as “pain indicates damage” or “this is taking too long”). The short Negative Pain Thoughts Questionnaire consists of four questions scored on a 6-point Likert scale, with higher scores indicating greater negative thoughts about pain [10].

Participants completed the Adult Hope Scale, a 12-item, validated questionnaire that measures an individual’s perceived success achieving goals in the past, present, and future on an 8-point Likert scale from 1 = definitely false to 8 = definitely true [34]. The total score is determined by summing the values of eight of 12 questions, with higher scores reflecting greater past and current ability to achieve goals.

The validated 9-item version of the Personal Optimism and Self-efficacy Optimism Scale was used to measure confidence in problem solving ability. Each question was scored on a 4-point Likert scale from 0 = completely incorrect to 3 = completely correct [14]. The total score is derived by summing all nine questions, with higher scores reflecting greater confidence.

Three validated PROMIS CATs—Anxiety, Depression, and Physical Function—were completed by participants to measure symptoms of anxiety, depression, and capability, respectively [30, 31, 37]. A score of 50 represents the mean of the general population in the United States, with every 10 points above or below representing one SD [23].

Additional survey questions included clinic site, name of care provider, diagnosis given by the care provider, and whether the condition was caused by trauma.

Ethical Approval

The protocol for this study was approved by the institutional review board at the University of Texas at Austin, Austin, TX, USA (protocol number 2020-03-0084).

Statistical Analysis

An a priori power analysis indicated that a minimum sample size of 135 patients would provide 80% statistical power with alpha set at 0.05. This was based on a regression analysis with seven predictors and the assumption that change in either measure of confidence in problem solving ability would account for 5% or more of the variability in outcomes and the complete model would account for at least 15% variability. We enrolled 187 patients in the study to ensure that the sample size would be met after exclusion of incomplete forms. All analyses were performed using Stata software (StataCorp.). A total of 17% (31 of 187) of participants did not complete the questionnaire; in all instances, they left more than 50% unfinished, so we could not use multiple imputation to account for missing data.

Continuous variables are reported as the median (interquartile range), and categorical variables are reported as the percentage (number). The following categorical variables were dichotomized: education (high school diploma or less versus some college or higher), work status (full-time or part-time employed versus not working), and comorbidities (none versus one or more).

To address the primary null hypothesis, we tested for bivariate associations between the PROMIS-physical function score and categorical variables using Mann-Whitney U tests (Supplementary Table 1; http://links.lww.com/CORR/A635), and between the PROMIS-physical function score and continuous variables using the Spearman rank-order correlation tests (Supplementary Table 2; http://links.lww.com/CORR/A636). For the variables that had a significant (p < 0.1) association with PROMIS-physical function scores, we performed a multivariable regression analysis with the PROMIS-physical function as the dependent variable.

For the secondary null hypothesis, we tested for bivariate associations between pain intensity and categorical variables using Mann-Whitney U tests (Supplementary Table 1; http://links.lww.com/CORR/A635), and between pain intensity and continuous variables using the Spearman rank-order correlation tests (Supplementary Table 2; http://links.lww.com/CORR/A636). For the variables that had a significant (p < 0.1) association with pain intensity, we performed a multivariable regression analysis with pain intensity as the dependent variable.

In an unplanned analysis, we used the Spearman rank-order correlation to test for bivariate associations between the Adult Hope Scale (past and current ability to achieve goals) and continuous variables (age, magnitude of depression, anxiety, capability, pain intensity, catastrophic thinking in response to nociception, unhelpful thoughts about symptoms, and confidence in problem solving ability) and between the Personal Optimism and Self-efficacy Optimism Scale (confidence in problem solving ability) and other continuous variables (Supplementary Table 3; http://links.lww.com/CORR/A637). For the variables that had a significant (p < 0.1) association, we performed a multivariable regression analysis with the significant variables as the independent variables. We used the variance inflation factor to assess for collinearity among the measures. All multivariable models had a variance inflation factor of less than 2, indicating a moderate correlation at most.

We reported partial η² as a measure of effect size for all multivariable regression models. The following rules of thumb are used to interpret values for partial η²: a value of 0.01 = small, 0.06 = medium, and values of 0.14 and higher show large effect size [39].

Results

Factors Associated with the Magnitude of Capability

After accounting for potential confounding variables such as education, symptoms of depression and anxiety, worst-case thinking, as well as past and current ability to achieve goals, we found that a higher score on the PROMIS-physical function CAT was modestly associated with lower scores on the Negative Pain Thoughts Questionnaire (β = -0.63 [95% CI -1.0 to -0.22]; standard error = 0.20; partial η² = 0.06; p = 0.003) and no self-reported comorbidities (β = 2.6 [95% CI 0.02 to 5.3]; standard error = 1.3; partial η² = 0.03; p = 0.048). Note that the association between PROMIS-physical function and no self-reported comorbidities can be considered fragile given the p value just under 0.05 and a wide confidence interval (Table 2). Bivariate analysis results are available (Supplementary Table 1; http://links.lww.com/CORR/A635).

Table 2.

Multivariable regression modeling of the PROMIS physical function CAT

Variable	β (95% CI)	Standard error	p value
Education	1.5 (-1.9 to 4.9)	1.7	0.40
Comorbidities
None	2.6 (0.02 to 5.3)	1.3	0.048a
PROMIS-depression CAT	-0.07 (-0.30 to 0.15)	0.11	0.52
PROMIS-anxiety CAT	-0.16 (-0.43 to 0.12)	0.14	0.26
PCS-4	-0.18 (-0.61 to 0.25)	0.22	0.41
NPTQ-4	-0.63 (-1.0 to -0.22)	0.20	0.003b
Adult Hope Scale	-0.045 (-0.21 to 0.12)	0.08	0.58

Model 1: n = 155; r² = 0.34; variance inflation factor = 1.5.

^aThe association of self-reported comorbidities with capability is statistically significant with a small effect size (partial η² = 0.03).

^bThe association of NPTQ-4 with capability is statistically significant with a small effect size (partial η² = 0.06); PROMIS = Patient-reported Outcomes Measurement Information System; CAT = computer adaptive test; PCS-4 = Pain Catastrophizing Scale (four-item version); NPTQ-4 = Negative Pain Thoughts Questionnaire (4-item version).

Factors Associated with Pain Intensity

After controlling for education, symptoms of depression and anxiety, unhelpful thoughts about symptoms, as well as past and current ability to achieve goals, greater pain intensity was modestly associated with higher scores on the Pain Catastrophizing Scale (β = 0.33 [95% CI 0.20 to 0.45]; standard error = 0.06; partial η² = 0.16; p < 0.001) and established patients (β = -1.1 [95% CI -1.8 to -0.31]; standard error = 0.38; partial η² = 0.05; p = 0.006 for new patients) (Table 3). Bivariate analysis results are available (Supplementary Table 2; http://links.lww.com/CORR/A636).

Table 3.

Multivariable regression modeling for pain intensity

Variable	β (95% CI)	Standard error	p value
Education	-0.35 (-1.4 to 0.65)	0.51	0.49
New patient	-1.1 (-1.8 to -0.31)	0.38	0.006a
PROMIS-depression CAT	0.04 (-0.03 to 0.1)	0.03	0.26
PROMIS-anxiety CAT	-0.03 (-0.11 to 0.05)	0.04	0.46
PCS-4	0.33 (0.2 to 0.45)	0.06	< 0.001b
NPTQ-4	0.04 (-0.07 to 0.16)	0.06	0.49
Adult Hope Scale	-0.002 (-0.05 to 0.04)	0.02	0.94

Model 2: n = 155; r² = 0.44; variance inflation factor = 1.8; p < 0.001.

^aThe association of new patient status with pain intensity is statistically significant with a small effect size (partial η² = 0.05).

^bThe association of PCS-4 with pain intensity is statistically significant with a large effect size (partial η² = 0.16); PROMIS = Patient-reported Outcomes Measurement Information System; CAT = computer adaptive test; PCS-4 = Pain Catastrophizing Scale (4-item version); NPTQ-4 = Negative Pain Thoughts Questionnaire (4-item version).

Correlations of a Measure of Past and Current Ability to Achieve Goals with a Measure of Confidence in Problem Solving Ability and with Other Mental Health Measures

After controlling for symptoms of depression, the Personal Optimism and Self-efficacy Optimism Scale (confidence in problem solving ability) had a limited association with the Adult Hope Scale (β = 0.15 [95% CI 0.09 to 0.21]; standard error = 0.03; partial η² = 0.13; p < 0.001) (Table 4). Bivariate analysis results are available (Supplementary Table 3; http://links.lww.com/CORR/A637).

Table 4.

Multivariable regression modeling for confidence in problem solving ability (Optimism Score)

Variable	β (95% CI)	Standard error	p value
PROMIS-depression CAT	-0.005 (-0.07 to 0.06)	0.03	0.87
Past and current ability to achieve goals (Adult Hope Score)	0.15 (0.09 to 0.21)	0.03	< 0.001a

Model 3: n = 156; r² = 0.19; variance inflation factor = 1.2; p < 0.001.

^aThe association of the Adult Hope Score with the Optimism Score is statistically significant with a medium effect size (partial η² = 0.13); PROMIS = Patient-reported Outcomes Measurement Information System; CAT = computer adaptive test.

After controlling for symptoms of anxiety and worst-case thinking, the Adult Hope Scale (past and current ability to achieve goals) had a modest association with symptoms of depression (β = -0.45 [95% CI -0.67 to -0.23]; standard error = 0.11; partial η² = 0.1; p < 0.001) and negative pain thoughts (β = -0.49 [95% CI -0.89 to -0.09]; standard error = 0.20; partial η² = 0.04; p = 0.02) (Table 5).

Table 5.

Multivariable regression modeling for past and current ability to achieve goals (Adult Hope Scale)

Variable	β (95% CI)	Standard error	p value
PROMIS-depression CAT	-0.45 (-0.67 to -0.23)	0.11	< 0.001a
PROMIS-anxiety CAT	-0.13 (-0.40 to 0.14)	0.14	0.35
PCS-4	0.29 (-0.12 to 0.70)	0.21	0.17
NPTQ-4	-0.49 (-0.89 to -0.09)	0.20	0.02b

Model 4: n = 156; r² = 0.35; variance inflation factor = 1.5; p < 0.001.

^aThe association of symptoms of depression with the Adult Hope Scale is statistically significant with a modest effect size (partial η² = 0.1).

^bThe association of NPTQ-4 with capability is statistically significant with a small effect size (partial η² = 0.04); PROMIS = Patient-reported Outcomes Measurement Information System; CAT = computer adaptive test; PCS-4 = Pain Catastrophizing Scale (4-item version); NPTQ-4 = Negative Pain Thoughts Questionnaire (4-item version).

Discussion

Extensive research consistently identifies unhealthy thoughts and distress as mental health factors that explain a notable percentage of the variation in magnitude of capability and pain intensity among patients undergoing musculoskeletal specialty care [1, 7, 20]. Part of furthering this line of evidence is identifying a set of brief measures that can detect mental health opportunities while taking care not to overlook additional factors that independently account for additional variations in symptom intensity. Our goal is to streamline routine assessment of mental health in daily musculoskeletal specialty practice and ensure it feels relevant and not burdensome to patients. This study addressed the potential for measures of past and current ability to achieve goals as well as confidence in problem solving ability to account for additional variation in PROM scores. Our study found that measures of past and current ability to achieve goals and optimism are associated with magnitude of capability and pain intensity, but they do not account for variation independent of measures of unhelpful thoughts and distress. This adds to the growing evidence that a few brief measures of unhelpful thoughts and distress may be sufficient to characterize mental health opportunities, which can help focus investigation of the best way to use such measures at the point of care to help individual patients.

Limitations

This study has several limitations. First, most participants were people seeking upper extremity and sports specialty care at musculoskeletal specialty offices in one city, which may limit the generalizability of the study’s findings. On the other hand, the psychosocial aspects of illness tend to apply across settings, and there is sufficient diversity in our sample that the relationships will likely be similar in other samples. Second, the number of incomplete forms was greater than expected, which may be attributable to the change, in part, to a recruitment strategy using a QR code that allowed people to complete questionnaires on their phone, which was implemented for safety during the pandemic. People were more likely to leave without completing the questionnaires when they did not have to return a tablet. Questionnaire fatigue may also have influenced the number of incomplete forms [16, 28], although we attempted to limit the total time spent completing questionnaires and the number of questions answered by using CATs and short forms. We identified the issue and enrolled more patients to ensure adequate power. Third, we included both new and established patients in this study because we have found doing so has a limited effect on associations identified in cross-sectional research [1, 6, 21, 29]; in addition, we compared new and return patients for factors that might have influenced our study’s outcomes, and found no differences between those groups of patients except for pain intensity (Supplementary Table 1; http://links.lww.com/CORR/A635). Based on the experience in numerous studies and our confirmatory analysis here (Supplementary Table 1; http://links.lww.com/CORR/A635), we anticipate that a study of all new or all returning patients would have similar findings to the current study. Fourth, we did not collect information on clinician type (such as surgeon or nurse practitioner). In prior research, we have identified differences by clinician, but not by clinician type (Supplementary Material 2; http://links.lww.com/CORR/A638), so we no longer routinely track clinician type. The key elements in prior studies—in addition to specific clinician—have related mostly to relationship building, exploration of values, reorientation of common unhelpful thoughts, and time, but not clinician type [1, 7, 12, 24].

We believe that the relatively high level of worry and despair during the COVID-19 pandemic should not influence the result of an experiment that depends on correlation/association, and there may even be an advantage if it resulted in additional variation in distress. Another possible limitation could be differences between in-person and telemedicine visits. Given that this study occurred during the COVID pandemic, 80% of visits were in-person and 20% were video telemedicine. In our opinion, the setting is unlikely to have a substantial influence on how people answered the questionnaires.

We note that this study included patients with a variety of conditions with a range of pathological severity and illness onset. Again, we see this variation as an advantage and anticipate few if any differences in a more homogenous cohort. Our reasoning is that mental health factors tend to have similar influence at various anatomical regions and for various types and severities of pathology. Finally, some of the relationships are marginal in terms of effect size and are fragile (in the sense that they were close to the cutoff we used for statistical significance, such that a few responses in a different direction may have resulted in a finding of no difference). An example of this was the association of capability with comorbidities, which had a wide confidence interval and marginal p value. We chose to de-emphasize these results.

Factors Associated with the Magnitude of Capability

The finding that a higher score on the PROMIS-physical function questionnaire was associated with more negative pain thoughts but not with greater past and current ability to achieve goals or confidence in problem solving suggests that measures of unhelpful thoughts obviate the use of these additional measures. This can help us arrive at a brief set of questions for identifying and addressing mental health in musculoskeletal care. Similar observations have been made with respect to measures of general resiliency, locus of control, neuroticism [26, 27], and others [3, 11, 25]. The correlations are not as strong as when we specifically measured unhelpful thoughts and distress (such as symptoms of worry or despair), and they are not included in multivariable models. On the other hand, a measure of kinesiophobia was demonstrated to account for additional variation beyond what is accounted for by symptoms of distress and catastrophic thinking [9]. We interpret this line of research to indicate that although terms such as agency, resiliency, optimism, and self-efficacy may be useful when discussing healthy behaviors, measures specific to unhelpful thoughts and distress are better able to account for variation in the magnitude of capability. In future studies, we anticipate identifying a limited set of questions that can account for unhelpful thoughts and distress with limited ceiling and floor effects, good reliability, and no loss of ability to account for variation in symptom intensity and magnitude of capability. If this possibility is realized, it will facilitate measurement of mental health opportunities routinely at the point of care with limited burden. Combined with a line of research identifying mental and physical health phenotypes, and another line of research developing effective communication strategies, collectively, this evidence has the potential to improve the health of people seeking musculoskeletal specialty care.

Factors Associated with Pain Intensity

The observation that greater pain intensity was associated with greater worst-case thinking but not with levels of past and current ability to achieve goals or confidence in problem solving suggests that the same concepts hold true for symptom intensity. Specific measures of unhelpful thoughts (catastrophic thinking, in this case) may provide adequate information about mental health opportunities [5, 13, 35, 36, 38]. If additional research confirms these relationships, it may greatly simplify how we measure and address mental health in specialty care.

Correlations of a Measure of Past and Current Ability to Achieve Goals with a Measure of Confidence in Problem Solving Ability and with Other Mental Health Measures

The finding that greater past and current ability to achieve goals had a modest correlation with fewer symptoms of depression and a moderate correlation with negative pain thoughts (Table 5) suggests a measure of past and current ability to achieve goals is closely tied to unhelpful thoughts and distress. This relationship may explain why this measure does not account for additional variation in human illness behavior.

The observation that higher rating of confidence in problem solving ability was associated with higher past and current ability to achieve goals is similar to results of a cross-sectional study on university students that showed hope is associated with optimism and both are correlated with life satisfaction [2]. A study of 57 college students had similar findings [32]. The finding that greater confidence in problem solving ability was not associated with other mental health measures suggests that this measure might be too nonspecific. We have found that other nonspecific measures such as the Brief Resiliency Scale have a similarly limited utility [8, 10, 19]. We speculate that people may answer general questions such as “I master difficult problems” and “I always find a solution to a problem” positively in accordance with their self-image, even when they are experiencing unhelpful thoughts and worry related to pain. Indeed, they may feel that by seeking care, they are mastering the problem. It seems that measures that specifically quantify unhelpful thoughts (which can also be conceptualized as cognitive errors, misinterpretations, or misconceptions) and distress (symptoms of worry or despair) alone can account for most variations in musculoskeletal illness related to mental health opportunities.

Conclusion

We tested whether measures of past and current ability to achieve goals, as well as confidence in problem solving ability, are useful for evaluating people receiving musculoskeletal specialty care. There were modest associations with pain intensity and magnitude of capability that were not independent of assessment of unhelpful thoughts and distress, suggesting that these measures (past and current ability to achieve goals with a measure of confidence in problem solving ability) can be omitted from mental health assessment of patients with musculoskeletal conditions. Based on the findings of this study as well as collective evidence from other studies [8, 10, 19, 26, 27], we concluded that specific gauges of unhelpful thoughts and distress are more useful than more general ones, such as indicators of goal achievement, or confidence in problem solving ability and general resiliency for understanding the variation in human illness behavior and likely for developing targeted measures. A future goal ought to be to develop a small set of questions that identify key mental health opportunities such as unhelpful thoughts and distress in a way that feels relevant and not burdensome to patients. Then we would be able to tie these brief queries into effective relationship-building and communication strategies that build trust. This could support care pathways that prioritize reorientation of unhelpful thoughts and amelioration of distress in parallel with considering physical interventions. We believe doing so would help bolster patient readiness and ease recovery from discretionary surgery.