Are Psychologic Factors Associated With Shoulder Scores After Rotator Cuff Surgery?

Abstract

Background

Psychologic factors are associated with pain and disability in patients with chronic shoulder pain. Recent research regarding the association of affective psychologic factors (emotions) with patients’ pain and disability outcome after surgery disagrees; and the relationship between cognitive psychologic factors (thoughts and beliefs) and outcome after surgery is unknown.

Questions/purposes

(1) Are there identifiable clusters (based on psychologic functioning measures) in patients undergoing shoulder surgery? (2) Is poorer psychologic functioning associated with worse outcome (American Shoulder and Elbow Surgeons [ASES] score) after shoulder surgery?

Methods

This prospective cohort study investigated patients undergoing shoulder surgery for rotator cuff-related shoulder pain or rotator cuff tear by one of six surgeons between January 2014 and July 2015. Inclusion criteria were patients undergoing surgery for rotator cuff repair with or without subacromial decompression and arthroscopic subacromial decompression only. Of 153 patients who were recruited and consented to participate in the study, 16 withdrew before data collection, leaving 137 who underwent surgery and were included in analyses. Of these, 124 (46 of 124 [37%] female; median age, 54 years [range, 21-79 years]) had a complete set of four psychologic measures before surgery: Depression, Anxiety and Stress Scale; Pain Catastrophizing Scale; Pain Self-Efficacy Questionnaire; and Tampa Scale for Kinesiophobia. The existence of clusters of people with different profiles of affective and cognitive factors was investigated using latent class analysis, which grouped people according to their pattern of scores on the four psychologic measures. Resultant clusters were profiled on potential confounding variables. The ASES score was measured before surgery and 3 and 12 months after surgery. Linear mixed models assessed the association between psychologic cluster membership before surgery and trajectories of ASES score over time adjusting for potential confounding variables.

Results

Two clusters were identified: one cluster (84 of 124 [68%]) had lower scores indicating better psychologic functioning and a second cluster (40 of 124 [32%]) had higher scores indicating poorer psychologic functioning. Accounting for all variables, the cluster with poorer psychologic functioning was found to be independently associated with worse ASES score at all time points (regression coefficient for ASES: before surgery -9 [95% confidence interval {CI}, -16 to -2], p = 0.011); 3 months after surgery -15 [95% CI, -23 to -8], p < 0.001); and 12 months after surgery -9 [95% CI, -17 to -1], p = 0.023). However, both clusters showed improvement in ASES score from before to 12 months after surgery, and there was no difference in the amount of improvement between clusters (regression coefficient for ASES: cluster with poorer psychologic function 31 [95% CI, 26-36], p < 0.001); cluster with better psychologic function 31 [95% CI, 23-39], p < 0.001).

Conclusions

Patients who scored poorly on a range of psychologic measures before shoulder surgery displayed worse ASES scores at 3 and 12 months after surgery. Screening of psychologic factors before surgery is recommended to identify patients with poor psychologic function. Such patients may warrant additional behavioral or psychologic management before proceeding to surgery. However, further research is needed to determine the optimal management for patients with poorer psychologic function to improve pain and disability levels before and after surgery.

Level of Evidence

Level II, therapeutic study.

Introduction

Psychologic factors are associated with pain and disability in a range of musculoskeletal disorders [15, 31, 47, 61, 63, 69] including the shoulder [10, 11, 59, 76]. Affective psychologic factors, including depression and anxiety, are associated with longer duration of shoulder symptoms [10], higher levels of shoulder disability [11, 75], and poorer quality of life [11]. Cognitive psychologic factors, including negative pain beliefs and catastrophizing [19, 20, 28, 71, 75], kinesiophobia [19, 28, 29, 36], and low pain self-efficacy [9, 43], are associated with higher levels of shoulder pain and disability [19, 36, 43] and predictive of poor outcome or nonrecovery after conservative management [9, 19, 28, 29, 71]. For orthopaedic surgery, affective psychologic factors, including depression and anxiety, are associated with worse outcomes for hip [18, 21], knee [18, 21], and spinal [56] surgery. For shoulder surgery, recent studies investigating psychologic factors have differed in their findings. Three studies reported no association of affective factors before surgery and pain and disability levels after surgery [12, 34, 58] and one study reported an association of affective factors with greater pain and disability after surgery [17]. The association of cognitive factors such as pain beliefs, catastrophizing, kinesiophobia, and self-efficacy with outcome after surgery has not been reported; however, one study has reported an association of affective and cognitive factors with shoulder pain [75]. Conceptual overlap between affective and cognitive factors has been reported in chronic musculoskeletal pain [65], low back pain [7, 61], and knee osteoarthritis [15] and may reflect a level of general psychologic distress. Alternatively, there may be important distinctions between these two factors, as identified in a study by Rabey et al. [61], in which people with chronic low back pain were grouped statistically (clustered) according to their individual pattern across a number of different measures of affective and cognitive psychologic functioning. Three psychologic clusters were identified: one cluster scoring low on all affective and cognitive measures representing better psychologic functioning, another scoring high on cognitive measures only, and one-third scoring high across all affective and cognitive measures, representing poorer psychologic function.

To date, most shoulder surgery studies have only explored the association of affective psychologic factors on outcome. The association of poor psychologic function with outcome after shoulder surgery remains controversial. Differential patterns of affective and cognitive psychologic factors in patients undergoing shoulder surgery may be evident and potentially associated with pain and disability levels before and after surgery; however, this remains unexplored. In view of the limited high-quality evidence to support some shoulder surgery procedures such as subacromial decompression surgery [3], greater focus on the biopsychosocial dimension as a contributor to shoulder pain and disability is imperative.

Therefore, we asked: (1) Are there identifiable clusters (based on psychologic functioning measures) in patients undergoing shoulder surgery? (2) Is poorer psychologic functioning associated with worse outcome (American Shoulder and Elbow Surgeons [ASES] score) after shoulder surgery?

Patients and Methods

A prospective longitudinal observational study of adult patients undergoing shoulder surgery was undertaken at one private and two public hospitals in Perth, Western Australia, during an 18-month period from January 2014 to July 2015. Study approval was attained from the Human Research Ethics Committees at Curtin University and Sir Charles Gairdner Hospital, Perth, Western Australia. All patients scheduled for shoulder surgery (n = 184) by one of six participating surgeons (MH, JS, GB, SG, PK, AT) were invited to participate. Three surgeons (JS, PK, AT) operated across the three hospital settings. Data collection at 3 and 12 months after surgery was completed between April 2014 and July 2016. Inclusion criteria were patients scheduled for surgery for rotator cuff repair with or without subacromial decompression for partial or full-thickness tears (n = 55) and arthroscopic subacromial decompression only (n = 43). The type of surgery was unable to be ascertained for 26 participants as a result of inaccessible surgical records. All participants underwent surgery on one shoulder only. Exclusion criteria were prior neck surgery, prior surgery on the same shoulder, presence of rheumatoid arthritis, fibromyalgia, local or malignant cancer, glenohumeral joint osteoarthritis, or an inability to comprehend English. Paper copies of all questionnaires were mailed for completion during the week before their scheduled surgery and the ASES questionnaire was mailed for completion at 3 and 12 months after surgery. A physical assessment was undertaken during the week before surgery in the surgical or physiotherapy clinic. Of the 184 patients who were invited to take part in the study, 14 were excluded, six were not able to be contacted, and 11 declined to participate, leaving 153 who were recruited and consented to participate. Of these, 16 patients withdrew before data collection as a result of cancellation of surgery, conservative management, or they changed their mind about study participation, leaving 137 patients who underwent surgery and were included in analyses. Of these, 124 had a complete set of psychologic measures before surgery and were included in the primary analyses of this study (Fig. 1). Median age was 54 years (range, 21-79 years) and 46 of 124 (37%) were women.

Fig. 1.

Flowchart illustrating the recruitment of participants into the study with psychologic indicator measures and ASES score.

Measurement of ASES Score

The primary outcome variable was the ASES score [62] that was completed before surgery and 3 and 12 months after surgery. The ASES score contains a pain subscale with one pain item (10-cm visual analog scale) and a function subscale with 10 functional items (questions are rated on a 4-point Likert scale for level of difficulty) with total score 0 to 100 (pain subscale 0-50; function/disability subscale 0-50) with lower scores indicating greater pain and disability. The ASES score has been reported to have good reliability and validity [4, 42, 44], is a robust shoulder pain and disability measure able to differentiate between patients making small versus large gains in pain and disability level [66], and change scores in the range of 12 to 17 are considered a minimal clinically important change [68].

Measurement of Psychologic Function

Four self-report psychologic questionnaires were completed before surgery. The Depression, Anxiety and Stress scale (DASS) [40] is a 42-item, valid [5, 39], and reliable [14] measure containing three subscales of depression, anxiety, and stress with scores ranging from 0 to 42. Higher scores reflect higher levels of depression (moderate 14-20; severe 21-27; extremely severe 28+), anxiety (moderate 10-14; severe 15-19; extremely severe 20+), or stress (moderate 19-25; severe 26-33; extremely severe 34+). The Pain Catastrophizing Scale (PCS) [67] is a 13-item, valid and reliable [54, 55] measure of thoughts and feelings that may be experienced in the presence of pain with scores ranging from 0 to 52 with higher scores reflecting greater catastrophizing. The Pain Self-Efficacy Questionnaire (PSEQ) [51] is a 10-item, valid and reliable questionnaire [2, 27, 51] evaluating the pain self-efficacy of an individual for a range of issues with scores ranging from 0 to 60 with lower scores reflecting poorer self-efficacy beliefs. The Tampa Scale of Kinesiophobia (TSK-11) [74] is an 11-item, valid and reliable [23, 45, 70] questionnaire that examines fear of movement or reinjury with scores ranging from 11 to 44 with higher scores reflecting greater pain-related fear. Recommendations regarding missing data management were followed where available. If there were two or less missing items, the average of the other items was imputed to calculate a total score; otherwise, totals were recorded as missing.

Statistical Analysis

Latent class analysis is a statistical technique used to subgroup people according to their scoring patterns across a number of variables (indicators) in such a way that people within a particular subgroup have a similar scoring pattern across the variables and the difference in scoring patterns across subgroups is as distinct as possible [33]. This technique was chosen to investigate the existence of clusters of people with differential profiles of psychologic functioning using measures of pain self-efficacy (PSEQ), pain catastrophizing (PCS), fear of movement (TSK), and depression, anxiety, and stress (DASS) as indicators. Thus, the resultant subgroups can be considered a “latent” variable identified by the indicator variables, which represents a more complex construct, in this case combined affective and cognitive functioning. Latent class analysis was used because it has advantages over other statistical techniques for subgrouping individuals in that it uses maximum likelihood estimation to allow a statistical evaluation of the optimal number of clusters, allows inclusion of variables with nonnormal distributions, and provides classification probabilities for each person [13]. All four indicator variables (PSEQ, PCS, DASS, and TSK) were used in continuous form in the latent class analysis with PCS and DASS set as zero truncated to adjust for right-skewed sample distributions of these variables and PSEQ reversed so that higher scores represented worse functioning across all four measures. Further details regarding latent class analysis are provided (Appendix, Supplemental Digital Content 1).

Confounding Variables

Potential confounding variables of the relationship between psychologic cluster membership and ASES scores were taken from the demographic, health and lifestyle, and surgical dimensions and based on previously established associations with shoulder pain and disability [10, 11, 19, 28, 29, 36, 43, 71], low back pain [47, 61], and knee pain [15].

Height and weight measures were taken during the physical examination before surgery and used to calculate body mass index. The self-report questionnaire booklet before surgery recorded age, gender, smoking (yes/no), alcohol use (never, monthly or less, 2-4 days per month, 2-3 days per week, > 4 days per week), health comorbidities (none, one or more comorbidities), education level (high school or less, tertiary college, university), occupation (lifting tasks/no lifting), an active workers’ compensation claim (yes/no), duration of symptom history (< 1 year, ≥ 1 year), hospital setting (public/private), primary surgical procedure group, and confidence in recovery from surgery. Confidence in recovery from surgery was determined on a 0 to 10 Likert scale in response to the question “How confident are you that your shoulder symptoms will improve after surgery?” with 0 representing “not confident” and 10 representing “completely confident.” Responses were subgrouped into a binary variable, around the median score of 8, with higher confidence denoted by a score of ≥ 8 and lower confidence represented by a score of < 8.

Assessment of Potential Confounding Variables

Clusters were profiled according to potential confounding variables using an independent t-test (normally distributed data), chi-square analysis (categorical data), or Wilcoxon rank-sum (Mann-Whitney) test (ordinal/count data). Linear mixed-effects regression models with random intercept and time as factor variables were used to assess the association of each variable with ASES score. For each variable, an interaction with time was assessed to test if the association differed according to time and in the absence of interaction estimates was pooled over time. Variables demonstrating an association with ASES score at p < 0.100 were included as potential confounders in subsequent analyses testing the association between cluster membership and ASES score over time. Univariate associations between potential confounding variables and ASES scores are detailed (Appendix, Supplemental Digital Content 2) and differences in potential confounding variables between psychologic clusters are detailed (Appendix, Supplemental Digital Content 3).

Assessment of the Association Between Cluster Membership and ASES Score

A linear mixed-effects regression model with random intercept and time and psychologic cluster membership as factor variables were used to assess the association of psychologic cluster membership with ASES score over time. A time-by-psychologic cluster interaction term was included to allow separate estimates for each group at each time point and thus test the difference between these measures according to psychologic cluster membership at each time point as well as assess differential changes between groups over time. The model was adjusted for variables identified as potential confounders as described previously. Ninety-five percent confidence intervals and p values are provided for all contrasts of interest. Linear mixed-effects regression models have advantages over traditionally used repeated-measures analysis of variance in these circumstances because they (1) allow the correlation of the within-subject repeated measures to be explicitly accounted for; (2) allow flexible modeling of time; and (3) allow use of all cases including those with missing data at one or more time points, which allows for unbiased estimates for time points with missing data providing data are at least missing at random [60]. Cases were used in this final analysis if they had (1) measures of all four psychologic indicator variables at baseline; and (2) ASES scores for least at one of the three time points. A sensitivity analysis was performed by imputing cluster membership for those cases missing psychologic indicator variables at baseline based on any available psychologic variables and variables associated with cluster membership.

Analyses were performed using LatentGold Version 5.0 (Statistical Innovations Inc, Belmont, MA, USA) and STATA Version 14.1 (StataCorp, College Station, TX, USA).

Results

Are There Psychologic Clusters in Patients Undergoing Shoulder Surgery?

Two distinct clusters were identified (Fig. 2): one cluster with better psychologic functioning (84 of 124 [68%]) and a second cluster with poorer psychologic functioning (40 of 124 [32%]). Akaike’s information criterion, Bayesian information criterion, and consistent Akaike’s information criterion were used to assess and compare the statistics of the one- to five-cluster models (Table 1). These statistics supported three-, five-, and two-cluster models, respectively, but graphic inspection showed a leveling out of all three statistics from the two-cluster model upward. The classification accuracy of the two-cluster model was good with classification error and entropy R² value equal to 0.05 and 0.80, respectively. The average (SD) posterior probability for better psychologic function was 0.96 (0.08) and for poorer psychologic function was 0.93 (0.13), which is well above the minimum value of 0.7 recommended for model adequacy [48], and the odds of correct classification were 11.8 and 27.0 for better and poorer psychologic function, respectively, also well above the value of ≥ 5 suggested to indicate high assignment accuracy [48]. Two- to five-cluster models all demonstrated varying distinctions of better and poorer psychologic functioning across all measures similarly with no patterns indicating a class with high scores on some indicator variables and not others. Given this and the acceptable fit of the two-cluster model, a two-cluster solution was chosen as the most parsimonious solution and participants assigned to the cluster for which they displayed the maximum posterior probability of membership for cluster profiling. The cluster with poorer psychologic function exhibited moderate levels of depression and stress, anxiety within normal limits, high levels of kinesiophobia, mild to moderate levels of catastrophizing, and low levels of pain self-efficacy (Table 2).

Fig. 2.

Cluster profiles were identified from the psychologic indicator variables used in latent class analysis. Dark line represents cluster with poorer psychologic function (32% of sample) and light line represents cluster with better psychologic function (68% of sample). PSEQ = Pain Self-Efficacy Questionnaire: (reversed score) possible range: 0-60 (higher score = lower pain self-efficacy); PCS = Pain Catastrophizing Scale; possible range: 0-52 (higher score = greater pain catastrophizing); DASS = Depression Anxiety Stress Scale; possible range: 0-126 (higher score = greater psychologic distress); TSK = Tampa Scale for Kinesiophobia: possible range: 1-44 (higher score = greater pain-related fear of movement). Data are presented as medians and interquartile range.

Table 1.

Latent class analysis measures of model fit and classification accuracy for one to six cluster models of psychologic functioning

Table 2.

Individual psychologic measures for the two clusters derived using latent class analysis

Association of Psychologic Functioning With ASES Score

The adjusted ASES scores were consistently better for those patients with better psychologic functioning, where the mean (SD) adjusted ASES score before surgery and 3 and 12 months after surgery was 54.3 (18.0), 71.6 (16.3), and 86.2 (17.6), respectively, for the cluster with better psychologic function and 39.5 (15.2), 52.2 (18.4), and 74.9 (19.7), respectively, for the cluster with poorer psychologic function (Fig. 3). At all time points, differences were above or close to the minimal clinically important difference [1, 68]. Similar changes over time were evident between clusters with both clusters displaying a significant improvement in ASES score from before surgery to 3 months, from 3 to 12 months, and in total improvement over 12 months, but there was no difference in the amount of ASES score improvement between clusters (between-cluster difference in overall change before to 12 months after surgery = -0.1 points; 95% confidence interval [CI], -9.1 to 8.9) (Table 3). A sensitivity analysis was conducted whereby the latent class membership for the 13 participants missing some or all of the psychologic indicator variables was estimated, utilizing the available baseline psychologic data and those covariates associated with class membership. This resulted in almost identical findings regarding the differences between the clusters at and between the time points with both clusters displaying a significant improvement over 12 months (change of 32 points for cluster with better psychologic function and 31 for cluster with poorer psychologic function) and no difference in the amount of ASES score improvement between clusters (between-cluster difference in overall change before to 12 months after surgery = -1.6 points; 95% CI, -9.9 to 6.7). The association between confounding variables and adjusted ASES score for all other variables showed that women had worse pain and function (regression coefficient for ASES: -6.3 [95% CI, -11.1 to -1.5], p = 0.010) as did patients with workers’ compensation (regression coefficient for ASES: -7.5 [95% CI, -13.5 to -1.5], p = 0.014), whereas patients with a greater increment of alcohol use had better pain and function (regression coefficient for ASES: 2.2 [95% CI, 0.2-4.3], p = 0.031) as did patients with a score of ≥ 8 for confidence in surgical outcome (regression coefficient for ASES: 5.6 [95% CI, 0.7-10.6], p = 0.026) (Table 4).

Fig. 3.

The adjusted predictions of ASES score over time by psychologic cluster. Dark line represents cluster with poorer psychologic function (32% of sample) and light line represents cluster with better psychologic function (68% of sample). ASES = American Shoulder and Elbow Score possible range: 0-100 (higher score = less pain and disability). Data are presented as mean (95% CI).

Table 3.

Change in ASES before surgery and 3 months and 12 months after surgery by psychologic cluster adjusted for potential confounders for gender, workers’ compensation status, alcohol use, and confidence in surgical outcome

Table 4.

Estimates for the independent association between confounders with ASES, pooled over time points, for all confounding variables in the multivariable model

Discussion

Previous studies exploring the association of affective psychologic factors with outcome after shoulder surgery have reported conflicting findings [12, 17, 34, 58], and cognitive factors have not been considered. Therefore, we aimed to assess if differential patterns of affective and cognitive factors in people undergoing shoulder surgery were evident and if there was an association of these factors with pain and disability levels before and after surgery. This study identified two clusters of people, one with poorer psychologic function before surgery that was associated with statistically and clinically important higher levels of shoulder pain and disability [1, 68], both before and up to 1 year after shoulder surgery, adjusting for gender, workers’ compensation claim, alcohol consumption, and confidence in surgical outcome. However, both clusters showed very similar improvements in ASES score over time. The inclusion of both affective and cognitive psychologic measures in this study encompasses a broader analysis of psychologic function than has previously been explored for shoulder surgery. The study findings support recent reports that both depression and catastrophizing are associated with higher levels of pain and disability in the shoulder [75] and other musculoskeletal conditions [7, 61, 65].

There were several limitations in this study. Participants in this study were recruited from surgical lists that may result in some patients scoring favorably on some cognitive measures as a result of a belief that surgery would “fix” their shoulder. Selection bias may have occurred during the recruitment phase, when not all patients undergoing surgery by the participating surgeons were recruited, and patients who declined to participate in the study may have scored differently on the psychologic measures. Psychologic measures were not explored at 3- and 12-month followup, so it is unknown whether psychologic measures change over time after shoulder surgery and how this links to changes in pain and disability. The confidence in surgical outcome with the Likert scale used in this study has not been previously validated. Because this study investigated patients undergoing shoulder surgery, there was no comparison to conservative management or no treatment, so the comparative course of recovery over time for conservative management or no treatment is unknown. Because this study included patients who underwent subacromial decompression only as well as patients who underwent rotator cuff repair, the findings cannot be attributed to one type of shoulder surgery; however, with the numbers available, there was no evidence that the proportions of people receiving each surgery type were different according to cluster membership or ASES scores. This study represents a sample of convenience, although the sample size obtained was sufficient to show statistically significant and clinically meaningful differences. Although there was a number of limitations in this study, our important findings support an association of poor psychologic function with higher levels of shoulder pain and disability before and after shoulder surgery. Psychologic factors warrant consideration before shoulder surgery in view of recent literature that questions the benefit of subacromial decompression surgery [3], a procedure commonly undertaken for impingement, which is a debatable condition [37] with limited surgical indication or efficacy [57]. Surgical management targeting the biologic dimension of shoulder pain only may fail to address factors such as poor psychologic function, resulting in unnecessary surgery or poorer outcomes after surgery.

Two psychologically derived clusters were identified in this sample of patients who underwent either subacromial decompression only or rotator cuff repair with or without subacromial decompression. One cluster displayed better scores across all psychologic measures, which reflected confidence in undertaking daily activity tasks despite pain, limited fear beliefs, or catastrophizing and normal limits of depression, anxiety, and stress [40]. In contrast, a second cluster of patients had poorer psychologic function with lower pain self-efficacy, mild to moderate levels of pain catastrophizing, moderately elevated levels of depression and stress, and high levels of kinesiophobia. The difference in kinesiophobia scores between clusters (6.8 points) was clinically meaningful [74] with the median score of 28.8 in the poorer psychologic function cluster aligned with previous studies reporting elevated kinesiophobia and catastrophizing in patients with shoulder pain and disability [16, 28, 43]. Patients with poorer psychologic function showed levels of moderate depression and stress but lower levels of anxiety, similar to a previous study of patients scheduled for rotator cuff repair [11].

For all study participants, being female, having an active workers’ compensation claim, less alcohol consumption, and lower levels of confidence in surgery were associated with lower ASES scores at all time points. However, psychologic cluster membership was associated with ASES score independent of these variables. These findings are consistent with current evidence reporting that older age [22, 53], female gender [50, 53], smoking [35, 41, 46], stiffness before surgery [12], insulin-dependent diabetes [8], obesity [73], lower education level [30], and an active workers’ compensation claim [26, 32] are negatively associated with pain and disability outcome after shoulder surgery, whereas greater expectations are positively associated with outcome [24]. This study did not find a significant association between age, education level, or comorbidities and ASES score after surgery. It is unknown why our findings showed more frequent alcohol consumption was associated with better ASES scores. Although participants with poorer psychologic function reported significantly lower levels of confidence that surgery would relieve symptoms, a lower level of confidence in surgical outcome was associated with poorer scores on ASES score at all data collection time points independent of psychologic cluster membership. A less optimistic outlook regarding surgical outcome is unsurprising in a group scoring poorly on psychologic questionnaires, potentially associated with a more negative affect. However, the results show that even in those with better psychologic function, optimism regarding surgery is associated with a better outcome. These findings are consistent with previous research suggesting that an optimistic outlook is strongly associated with improved outcome of surgery [64] and physiotherapy treatment [9]. In this study, duration of symptoms of > 1 year was not significantly associated with ASES score, which is consistent with findings from other studies [49, 53]. This study found no overall association with hospital setting (public/private) or primary surgical procedure and ASES score.

Participants who scored poorly on a range of psychologic measures before shoulder surgery displayed worse ASES scores at 3 and 12 months after surgery, although both clusters showed similar improvements in ASES score over time. Screening of psychologic factors before surgery is recommended to identify patients with poor psychologic function with simple screening tools such as the Orebro [38] or modified STarT Back musculoskeletal tool [6, 25] that capture fear, mood, anxiety, and beliefs. It could be speculated that patients with poor psychologic function may benefit from pharmacology, psychologic, or behavioral interventions before consideration of surgery; however, this an area that requires further research in the shoulder. Targeted interventions directed to psychologic factors before or after shoulder surgery may improve clinical outcomes. Cognitive-behavioral therapy has been reported to be an effective treatment for the management of other musculoskeletal conditions such as low back pain [52] and idiopathic hand and arm pain [72]. Future randomized controlled trials are recommended to test behavioral and psychologic intervention in patients with poor psychologic function who are undergoing shoulder surgery.

This study supports that a comprehensive assessment of affective and cognitive psychologic factors before shoulder surgery may identify patients with poor psychologic function. Alternative management pathways may be beneficial to improve clinical outcomes; however, further research is needed to determine the optimal management for patients with poorer psychologic function to improve pain and disability levels before and after surgery.