Are Psychosocial Factors Associated With Patient-reported Outcome Measures in Patients With Rotator Cuff Tears? A Systematic Review

Abstract

Background

Psychosocial factors are key determinants of health and can influence patient-reported outcomes after rotator cuff tears. However, to our knowledge, a systematic review of published studies has not been conducted to determine the degree of consistency and strength of the relationship between psychosocial factors and patient-reported outcomes in this patient population.

Questions/purposes

(1) Are psychosocial factors associated with patient-reported measures at initial clinical presentation in patients with rotator cuff tears? (2) Are psychosocial factors associated with patient-reported outcomes after treatment in patients with rotator cuff tears?

Methods

A systematic review of cross-sectional and prospective observational studies was performed in MEDLINE/PubMed, EMBASE, CINAHL, PsycINFO, and Web of Science from each database’s inception to June 2016. We included studies examining associations between psychosocial factors and patient-reported measures in patients with rotator cuff tears. We excluded studies not reporting on this relationship, involving patients with nonspecific shoulder pain, and written in a language other than English. Two independent reviewers performed the search, extracted information, and assessed methodological quality. Study quality was assessed using the Methodological Items for Non-Randomized Studies (MINORS) criteria. The primary outcomes for the review were associations between each psychosocial factor and patient-reported measures of function or disability, pain, or quality of life. Associations were interpreted based on significance, strength, and direction of the relationship. A total of 10 studies (five cross-sectional and five prospective) in 1410 patients (age range, 46-62 years, 60% [571 of 958] men) were included in the review. Pooling of results for meta-analyses was not possible as a result of study heterogeneity.

Results

Weak to moderate cross-sectional associations were found for emotional or mental health with function or disability and pain in multiple studies. Lower emotional or mental health function was associated with greater pain or disability or lower physical function at initial evaluation. Only one psychosocial factor (patient expectation) was weak to moderately associated with patient-reported outcomes after treatment in more than one study. In the two studies that examined expectations, the higher the expectation of benefit, the greater the perceived benefit after surgical intervention.

Conclusions

At the initial evaluation of patients with rotator cuff tear, there was an association between self-reported function and pain and emotional or mental health. However, these factors were not associated with patient-reported outcomes after intervention. This finding could be attributed to the lack of large prospective studies in this area or complex phenotypes within this patient population. Preoperative patient expectation is an important predictor of patient-reported outcomes in patients after rotator cuff surgery and may be a modifiable target for enhancing recovery.

Level of Evidence

Level III, therapeutic study.

Introduction

Rotator cuff tears are a common musculoskeletal injury with a prevalence that increases with age. For example, rotator cuff tears are found with increasing prevalence from 9% in individuals younger than 21 years, 30% to 50% in individuals older than 60 years, and up to 62% in individuals 80 years of age or older [54, 71, 82], yet not all individuals with rotator cuff tears report pain or functional limitations. Furthermore, rotator cuff tears are often present in the asymptomatic shoulder of individuals with symptomatic rotator cuff tears [90]. Approximately 35% of rotator cuff tears are symptomatic and lead to individuals seeking medical management [55, 91]. In patients with symptomatic rotator cuff tears, treatment includes nonoperative treatment (like physical therapy) or surgical intervention. Patient-reported outcomes such as disability and pain show variability in response after treatment, including after successful surgical repair of the anatomic defect [23, 49].

Historically, biomedical factors such as the patient’s age [15, 42, 70], rotator cuff tear size [24, 35, 36, 84], muscle atrophy and fat infiltration [29, 30], history of smoking [68], and medical comorbidities [4, 72] including diabetes [9] and cardiovascular disease [47] have been considered important individual variables explaining variability in patient-reported outcomes. However, recent evidence has supported a biopsychosocial model in which psychosocial factors, including general constructs such as psychologic distress and emotional health, and specific constructs such as anxiety and depression [87] and fear avoidance beliefs [22] are associated with patient-reported outcomes. Prior systematic reviews have reported on the association between psychosocial factors and patient-reported outcomes in patients with nonspecific shoulder pain [8, 37, 40, 78]. For example, Struyf et al. [78] recently examined prognostic associations between a variety of factors, including psychosocial factors, and chronicity in patients with acute or subacute nontraumatic shoulder pain. Struyf et al. [78] reported that psychosocial factors were not associated with future chronicity. To our knowledge, a similar systematic review has not been conducted for determining the degree of consistency and strength of psychosocial associations with patient-reported outcomes in patients with rotator cuff tears.

The aim of this study was to conduct a systematic review to answer the following two questions: (1) Are psychosocial factors associated with patient-reported measures at initial clinical presentation in patients with rotator cuff tears? (2) Are psychosocial factors associated with patient-reported outcomes after treatment in patients with rotator cuff tears?

Search Strategy and Criteria

This systematic review was conducted in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [46, 56]. To accomplish our aim, we assessed cross-sectional (Question 1) and prospective observational studies (Question 2). The protocol for this systematic review was developed a priori and included a systematic review. Electronic databases were searched from each database’s inception to June 2016 for relevant studies meeting eligibility criteria. These databases included MEDLINE/PubMed, EMBASE, Cumulative Index to Nursing and Allied Health Literature (CINAHL), PsycINFO, and Web of Science. Limits were placed for articles involving human participants and published in English and within peer-reviewed journals. A hand search of reference lists was conducted to capture potentially missed articles.

A systematic search strategy was developed and included key clinical terms such as “rotator cuff”, “shoulder”, “subacromial”, “impingement”, “surgery”, “arthroscop^*”, “psychosocial”, “psycholog^*”, “catastrophiz^*”, “fear”, “affect”, “depress^*”, “stress”, “anxiety”, “pain”, “disability”, “recovery”, “function”, “quality of life”, and “health”. Terms related to study methods were added and included “cohort”, “prospective”, “longitudinal”, “cross-sectional”, “retrospective”, “prognos^*”, “regression”, and “predict^*”. Medical subject headings (MeSH) and key terms were used as applicable. The full search strategy of each database is presented (see Appendix, Supplemental Digital Content 1).

Search results from each database were compiled and imported into EndNote X7 bibliographic software (Thomson Reuters, New York, NY, USA). The initial screening involved removing duplicate and non-English articles. All titles and abstracts were screened for inclusion by two independent reviewers (RAC, EP). In cases in which the title and abstract were not informative enough for inclusion, the full text of articles was accessed. Final lists of potentially eligible articles were compared by the two reviewers and any disagreements were resolved in collaboration with a third reviewer (NBJ).

Articles were considered eligible for inclusion based on study design, population, psychosocial factor, and outcome. The specific inclusion and exclusion criteria are described subsequently. Studies examining associations between psychosocial factors and outcome were of primary interest. Thus, cross-sectional, prospective cohort, and retrospective studies were targeted, whereas other designs were excluded. The population of interest was participants with documented or diagnosed rotator cuff tears, either on clinical examination or imaging findings, or a combination of the two. Articles were excluded if the sample included patients with mixed shoulder pathology or nonspecific shoulder pain. No limit was placed on the type of intervention received within the cohort study and could include nonoperative or surgical intervention. Only studies assessing the influence or association of a psychosocial factor were eligible. Studies were excluded if statistical associations were not reported. There are a variety of psychosocial factors, including general constructs such as psychologic distress and emotional health, and specific constructs such as anxiety, depression, and fear avoidance beliefs. We did not limit the inclusion to preidentified psychosocial factors. The outcomes of interest were reliable and valid patient-reported measures of disability, pain, and function or a more broad measure of general health such as quality of life. For prospective studies, no limit was placed on the minimum followup. Articles were excluded if not written in English and if full texts were unavailable.

The systematic review yielded a total of 4166 articles from the five databases and individual reference lists (Fig. 1). After duplicate articles were removed, there were 2530 unique articles that were screened. Thirty-nine full-text articles were identified for review. Of these, 28 were excluded for not including a specific sample of patients with rotator cuff tears [3, 5, 6, 39, 63, 64, 67, 73, 74, 77, 86, 92], not measuring a psychosocial factor [4, 12, 13, 21, 32, 34, 41, 52, 57, 58, 65, 79, 81], and not incorporating a relevant clinical outcome [7, 18, 93]. Eleven studies met the eligibility criteria for inclusion into this review [2, 10, 11, 31, 59-62, 80, 88, 89]. Three sets of studies included data from the same center (set 1 [10, 11], set 2 [31, 80], and set 3 [61, 62, 89]). For the first two sets of data, the analyses in each study involved either cross-sectional (preoperative only) or followup longitudinal analyses of the same sample data and, therefore, were not duplicate aims. The data from these studies were included in the review because each study was summarized separately to answer each of our review questions. For the third set, two studies involved samples that included some overlap, as determined by contact with the studies’ senior authors [62, 89]. Both studies investigated cross-sectional associations, but with different psychosocial variables. Regardless, we included the more recent study with the larger sample size [89] and excluded the former study [62]. Additionally, within the third set, Potter et al. [61] included a cross-sectional analysis of data within their longitudinal study. This aspect of the analysis was not included in the review because this was a similar analysis to their previous paper [62]. Ten studies were ultimately included in this review [2, 10, 11, 31, 59-61, 80, 88, 89].

Fig. 1.

This figure displays the flow diagram for the systematic review.

Methodological quality of individual studies was assessed using the Methodological Index for Non-Randomized Studies (MINORS) tool [76]. The 12-item tool is a reliable and valid measure for grading individual study quality [76]. For noncomparative studies, only the first eight items are scored. Each item is scored on a 0 to 2 scale with 0 meaning “not reported,” 1 meaning “reported but inadequate,” and 2 meaning “reported and adequate.” Scores can range from 0 to 16 with higher scores indicating higher methodological quality. Quality assessment was performed by two independent reviewers (RAC, ALS). Consistency of scores between the two raters was strong (intraclass correlation coefficient = 0.75). Discrepancies were resolved through discussion until consensus was achieved. Quality scores for the 10 articles ranged from 8 to 14 on the MINORS tool (Table 1). All studies adequately reported a clearly stated aim and appropriate endpoints for the study. No study adequately reported on an unbiased assessment of the study endpoint. Only two studies provided an a priori calculation of sample size [10, 11].

Table 1.

Quality scoring of included studies (N = 10) based on Methodological Items for Non-Randomized Studies (MINORS) criteria

A standardized data extraction form was used for obtaining relevant data from each article. Two authors (RAC, EP) independently extracted data. Extracted data included descriptive study characteristics including study design, country of origin, sample data (inclusion/exclusion, number of participants, age, symptom duration, rotator cuff tear size), psychosocial and outcome measures, statistical modeling, and measurement time points (Table 2).

Table 2.

Characteristics of included studies (N = 10)

Of the 10 studies, five were cross-sectional [2, 10, 60, 80, 89] and five were prospective cohort studies [11, 31, 59, 61, 88]. Four of the five prospective studies assessed outcomes at least 12 months after surgery [11, 31, 59, 61], whereas one study measured outcome at 6 months after surgery [88]. Cross-sectional associations were also examined in one prospective study [59]. Six studies were conducted in the United States [2, 31, 61, 80, 88, 89], three in South Korea [10, 11, 59], and one in Finland [60]. All patients were recruited in orthopaedic centers and within a preoperative context. A total of 1410 patients (range of sample size, 46–452) was represented across all studies. Of the studies reporting sex, men were predominant (60% [571 of 958]). The age of patients ranged in studies from 46 to 62 years.

Psychosocial factors varied across all studies and included measures of depression, anxiety, emotional or mental health, expectation, psychologic distress, somatic perceptions, preoperative concerns (eg, concerns about hospital environment/staff, medical procedures, potential complications, recovery), and fear avoidance beliefs (Table 3). Depression was the most examined psychosocial factor in four studies. Two studies used the Depression subscale of the Hospital Anxiety and Depression Scale [10, 11], one study used the Zung Depression Scale [61], and one study used the Clinical Epidemiologic Studies Depression Scale [88]. Anxiety, emotional or mental health, and expectation were all assessed in three studies. The Anxiety subscale of the Hospital Anxiety and Distress Scale was used in two studies [10, 11] and the Beck Anxiety Inventory in a single study [88]. Emotional or mental health was assessed in individual studies using the Western Ontario Rotator Cuff Index [2], SF-36 [60], or SF-12 [89]. Expectation was assessed using questions within the Musculoskeletal Outcomes Data Evaluation and Management System (MODEMS) [31, 59, 80]. Psychologic distress and somatic perceptions were assessed in one study using the Distress Risk Assessment Method and Modified Somatic Perception Questionnaire, respectively [61]. Preoperative concerns were assessed in one study using a 64-item questionnaire within MODEMS [59]. Fear avoidance beliefs were assessed in one study with the Fear-Avoidance Beliefs Questionnaire [88].

Table 3.

Psychosocial factors examined within included studies

All 10 studies included an upper extremity or shoulder-specific function or disability measure. The most common measures used in six studies were the American Shoulder and Elbow Surgeons Scale [10, 11, 60, 61, 89] and Simple Shoulder Test [31, 59, 61, 80, 89]. Three studies used the Disabilities of the Arm, Shoulder, and Hand [31, 80, 88], whereas fewer than three studies used the Korean Shoulder Scale [10, 11, 43], visual analog scale [31, 89], Single Assessment Numeric Evaluation [2], Constant-Murley Shoulder Score [59], or Western Ontario Rotator Cuff Index [88]. Pain intensity was measured in seven studies using a visual analog scale [2, 10, 11, 31, 61, 80, 89]. Quality of life was assessed in only four studies using the SF-36 [31, 59, 80], World Health Organization Quality of Life scale [10], or visual analog scale [31, 80].

Outcome associations between psychosocial factors and patient-reported outcomes were extracted and summarized for cross-sectional (Table 4) and prospective cohort studies (Table 5). Where available, strength of association values (ie, correlation coefficients) was recorded. Correlations < 0.3 were considered weak, between 0.3 and 0.5 moderate, and > 0.5 as strong [14]. As a result of study heterogeneity, meta-analytical procedures were not conducted.

Table 4.

Cross-sectional associations between preoperative psychosocial factors and preoperative outcomes in patients with rotator cuff tears.

Table 5.

Associations between preoperative psychosocial factors and postoperative clinical outcome after rotator cuff surgery

Results

Scores suggesting poorer emotional or mental health, preoperative concerns, and depression were consistently associated with worse psychosocial and physical functioning before surgery (Table 4). Cross-sectional associations between a psychosocial factor and patient-reported measures at initial presentation were found in five of six studies for function or disability [2, 10, 59, 60, 89], two of three studies for pain [2, 89], and three of three studies for quality of life [10, 59, 80]. Multiple studies showed consistent associations between emotional or mental health with function or disability and pain (Table 6) [2, 10, 89]. The direction of all associations suggested that the lower the level of emotional or mental health function, the worse the outcome at the preoperative time point. The range of correlation values (absolute r values: 0.33-0.51) suggested these psychosocial factors are weak to moderately correlated with patient-reported outcomes at the preoperative time point. All other specific psychosocial constructs did not show consistent associations across studies, were shown not to be associated with outcome, or were only assessed in one cross-sectional study. No included studies examined the cross-sectional association between (1) psychologic distress, fear avoidance beliefs, or somatic perceptions and patient-reported outcome; (2) preoperative concerns and pain; or (3) emotional or mental health, psychologic distress, or somatic perceptions and quality of life.

Table 6.

Summary of psychosocial factors and the relation to patient-reported outcomes

Preoperative expectation was the only psychosocial factor associated with postoperative outcome in more than one prospective study [31, 59] (Table 5). Longitudinal associations between a psychosocial factor and treatment outcome were found in three of five studies for function or disability [31, 59, 88], one of three studies for pain [31], and two of two studies for quality of life [31, 59]. In a single study, preoperative fear avoidance beliefs were associated with postoperative function [88], and preoperative concerns were associated with postoperative quality of life [59]. Anxiety and depression were found not to be associated with postoperative function or disability and pain in multiple prospective studies with either univariable or multivariable analyses (Table 6). All other psychosocial factors did not show an association with postoperative outcome in the one study examining this relationship [61].

Discussion

Psychosocial factors have an important influence on symptoms and functional limitations, which favor adoption of a biopsychosocial model over a biomedical one for many health conditions [20], yet the degree to which published studies support this assumption in patients with rotator cuff tears has not been previously explored in a systematic review. Determining the consistency of associations across studies is a necessary step toward advancing clinical recommendations regarding psychosocial screening and providing a basis for treatment modification. Furthermore, this systematic review characterizes the prominent psychosocial factors that have received attention and those that are in need of further study in this population. The results of this review suggest that psychosocial factors, namely emotional or mental health, are associated to a weak to moderate degree with initial function or disability and pain in patients seeking operative treatment for rotator cuff tears, whereas expectation was the only factor associated with postoperative patient-reported outcomes. However, these findings were only observed in at most three studies. The lack of well-designed prospective studies in this area limits conclusions about the potential prognostic value of psychosocial factors.

We acknowledge several limitations in this review. The heterogeneity of included studies did not allow for pooling of individual study results for meta-analysis. Thus, this review was limited in its ability to establish a more definitive estimate and precision of effect for psychosocial associations. The heterogeneity observed came in the variety of constructs examined with a range of psychosocial factors and outcomes included in the few studies found. However, there was consistency in our summary results supporting cross-sectional associations between emotional or mental health and patient-reported outcomes, but we did not find this to be the case for prognostic associations other than expectation. More prospective studies are needed to draw firm conclusions about the potential influence of psychosocial factors on patient-reported outcomes in this patient population. We did not find studies examining the association of psychosocial factors on patient-reported outcomes of pain, function or disability, or quality of life in populations seeking nonoperative treatments such as rehabilitation or in populations not actively seeking treatment. Thus, the current knowledge on the relationship of psychosocial factors on patient-reported outcomes is biased toward patients who did not benefit from first-line nonoperative treatment including injections, physical therapy, exercise, manual therapy, nonsteroidal antiinflammatories, or other nonoperative interventions. Three of five cross-sectional studies and four of five cohort studies excluded patients who had prior shoulder surgery or potentially revision procedures. Thus, there is no evidence supporting the use of patient expectations for treatment decision-making in patients in whom initial surgery was not perceived as beneficial.

Emotional or mental health was shown to be consistently associated with function or disability and pain before a surgical intervention. Depression is an important determinant of emotional functioning and was also found to be associated with patient-reported outcomes in a single study [10]. It was therefore not surprising that depression was also shown to be associated with preoperative measures. Roh et al. [66] also found depression to be moderately correlated with pain and disability in a sample of 109 patients with chronic shoulder pain. In the multivariable model for disability, Roh et al. [66] found depression to be a stronger predictor of disability than clinical variables of ROM or pain. These findings reflect similar observations of cross-sectional associations between depression and clinical measures in a variety of patients with musculoskeletal pain [1, 25]. Future studies should address limitations of existing evidence by longitudinal assessment of patients with rotator cuff tears and assessing a causal relationship between emotional or mental health and patient-reported outcomes.

Surprisingly, no cross-sectional studies were identified that examined associations between fear avoidance beliefs and pain catastrophizing on patient-reported outcomes. These factors are widely known risk factors within the Fear-Avoidance Model [44, 85] and have been extensively studied in patients with low back pain. In patients with nonspecific shoulder pain, associations between pain catastrophizing and fear have been previously reported [26, 45, 53]. The findings from these studies cannot be directly generalized to patients with specific rotator cuff tear pathology because these studies include mixed samples of patients. The importance of shoulder pathology on psychosocial and outcome associations is not clear, but some evidence exists related to potential subgroup effects based on shoulder pathology [75]. There appears to be a clear need for examining the extent to which Fear-Avoidance Model factors explain pain and function variability in patients presenting with symptoms associated with rotator cuff tear pathology.

In general, psychosocial factors were not shown to be associated with patient-reported outcomes at 6 months or up to 1 year after initial evaluation. It is unclear the extent to which small sample sizes influenced these findings, because the studies examining anxiety and depression within longitudinal designs included only 46 [88], 47 [11], and 70 patients [61]. Koorevaar et al. [38] reported a similar lack of association among preoperative psychologic distress, depression, and somatization with 12-month outcomes in a mixed sample of patients after shoulder surgery. Although the observation suggesting a lack of prognostic utility of preoperative psychosocial factors on postoperative outcomes is consistent with this evidence [38], findings across postoperative conditions are conflicting [17, 33, 83]. There are a few potential reasons for this observation including the presence of more complex phenotypes within patient samples and the relative importance of preoperative versus postoperative psychosocial functioning. Psychosocial risk is more complex than the sum total of negative feelings, attitudes, or beliefs. Positive psychosocial functioning, like self-efficacy and resilience, may facilitate beneficial responses after surgery by protecting against negative psychosocial influence. The consideration of both positive and negative psychosocial factors as well as genetic predisposition is integral within a more comprehensive model of recovery [19]. George and colleagues [27, 28] have identified risk phenotypes based on specific psychosocial and genetic factor interactions that predict outcomes in patients with nonspecific shoulder pain. Further investigations in line with research on complex psychosocial factor interactions are needed for predicting outcome and informing potential treatment strategies. In terms of timing of psychosocial assessment, Koorevaar et al. [38] found postoperative psychosocial status to be more relevant for predicting clinical outcome. This finding is consistent with recent studies in patients after spine surgery [16, 69]. Determining the prognostic value of immediate postoperative psychosocial risk on rotator cuff treatment outcome is a relevant area for future study.

The only preoperative factor associated with treatment outcome in more than one study was patient expectation. Henn et al. [31] reported relationships between preoperative expectation and 1-year postoperative outcome in 125 patients with rotator cuff tears seeking surgical repair after not being satisfied with nonoperative treatment. Similarly, Oh et al. [59] examined 128 patients seeking surgical rotator cuff repair after not being satisfied with at least 3 months of nonoperative treatment. Patients with high compared with low preoperative expectations, defined as the patients in the top and bottom thirds of mean expectation scores in the sample, had greater improvement in function or disability. The studies by Henn et al. [31] and Oh et al. [59] represent independent findings in > 250 patients from two different populations who were not satisfied with nonoperative treatment and then sought surgical treatment. This finding is consistent with the predictive value of patient expectations on patient-reported outcomes in other surgical orthopaedic populations including hip and knee arthroplasty [48] and lumbar surgery [51]. Thus, preoperative procedures to assess and identify patients with discordant patient expectations would be extremely valuable to enhance patient-reported surgical treatment outcomes of pain, function, or disability. Patient expectations of recovery appear to be modifiable before surgical intervention and can be delivered with surgeon counseling and preoperative educational techniques [50]. As such, assessment of and interventions to align patient treatment expectations, address misconceptions or “magical thinking,” or offer suggestions for improving coping strategies should be explored as a requisite to enhance patient outcomes.

No study in this review examined patient expectations as a predictor of patient outcomes after nonoperative treatment for rotator cuff tear. However, a recent study by Dunn et al. [18] showed that of the multiple factors (including tear severity, age, pain) considered for predicting an unsatisfactory response to nonoperative treatment, patient expectation was the strongest predictor [18]. Although this study was not included in the current review (because it did not include a priori outcome), it further corroborates results of this review suggesting patient expectation is an important factor that should be used for treatment decision-making in patients with rotator cuff tears. A major gap in the available evidence is the lack of assessment of a full biopsychosocial model in a single analysis. This is the result of lack of measurement of a priori-designated variables such as patient expectations that might contribute to such a biopsychosocial model.

Outcomes in patients with rotator cuff tears are variable and identifying factors that explain this variability is important for enhancing recovery. Psychosocial factors are considered important risk factors in nonspecific shoulder pain. For patients with rotator cuff tears, emotional or mental health is associated with initial pain and function although it is unknown whether shoulder pain and functional disability causally lead to an impact on emotional or mental health. Few studies have examined the prognostic role of preoperative psychosocial factors on treatment outcome, and so this precludes definitive conclusions in this patient population. The lack of associations observed with treatment outcomes could be attributed to limited sample size or the presence of complex phenotypes within this patient population. Preoperative patient expectation of treatment benefits appears to be an important factor related to postoperative recovery. Future research in this area should consider more comprehensive psychosocial modeling (such as inclusion of fear avoidance model variables and more focus on positive psychosocial factors) with sufficient sample sizes.