Table 3 Predictive utility of psychological factors on the outcome after conservative treatment for shoulder complaints. Bold font indicates high quality studies. Bold font indicates high quality studies based on the SIGN review

Berk et al. 1977 [33]

(+)

Pain (VAS)

–

Acupuncture in a negative and a positive milieu resulted in similar pain reduction (p = 0.053).

(++)

Pain and disability (SPADI, QuickDASH)

–

–

Patient expectation of ‘complete recovery’ compared to a ‘slight improvement’ as ‘a result of physiotherapy treatment’ (Beta 12.43, 95% CI 8.2–16.67 for 6 months). Depression and anxiety: no consistent association in the multivariate models.

Chester et al. 2019 [34]

(++)

Pain and disability (SPADI, QuickDASH)

+

+

Additional analysis using the risk groups as Chester et al. 2016. Using Classification and Regression Tree (CART) analysis, the authors categorized the subjects into three groups (based on the predictor analysis Chester 2016) – those with high, moderate or low levels of pain and disability. There was a positive association.

Between pain and disability at baseline and at follow-up. Those with high pain and disability and high self-efficacy scores (PSEQ≥48) were less likely to have continued high levels of pain. Patients with moderate levels of baseline pain and disability and high expectation of recovery had better outcomes at 6 months than those with low expectations of recovery. Patients with low baseline pain/disability and low pain self-efficacy (PSEQ < 41) had increased likelihood of persistent pain. No external validation of the CART.

Ekeberg et al. 2010 [35]

Pain and disability (SPADI)

Shoulder complaint (^aADI)

-

-

-

-

Distress (HSCL-25) and self-efficiency for pain (single item question) not associated with SPADI and shoulder complaint as measured by Global Assessment Score at 6 weeks.

Engebretsen et al. 2010 [36]

(++)

Pain and disability (SPADI)

Work status (^aADI)

-

-

-

-

Self-efficacy was significant in the univariate analysis but not in the final model for disability and not significant for return to work. Distress (Hopkins Symptoms Checklist) was not significant in the univariate analysis.

Engebretsen et al. 2020 [37]

(++)

Pain and disability (SPADI)

+

TSK predicted outcome SPADI scores at 6 months based on multivariate regression (Beta 0.76, 95% CI 0.27–1.2, p = 0.003). HSCL-10 did not predict pain and disability outcomes, however high baseline scores were associated with high HSCL follow up scores.

Örebro screening questionnaire predicted sick-leave at follow-up: OR 1.075 (95% CI (1.03 to 1.12)), p = 0.001 however no specific psychological factors were extracted in the analysis.

Geraets et al. 2005 [38]

(+)

Shoulder disability (SDQ)

+

–

Coping measured by PCCL, and FABQ and TSK measured but not used in the model; DSQ N.S.; Significant relationship between reduction of severity of main complaint (graded exercise group vs usual Care group Beta 7.6, 0.9–14.3) at 12 weeks and pain reduction (26.8, 95% CI 19.3–34.4) and baseline depression scores (8.3, CI 0.1–16.6, 4DSQ); Anxiety (4DSQ) N.S.

Karel et al., 2017 [39]

(++)

Perceived recovery (^aADI)

–

No significant association between psychological factors and perceived recovery: OR for patient reported “no anxiety/depression” in (EQ-5D) 1.8 (95% CI 0.9–3.6), p = 0.06.

Note: on the anxiety/depression dimension of the EQ-5D, only one patient scored “very anxious/depressed”, 83% reported “not anxious/depressed”, and 16% reported “moderately anxious/depressed”.

Kennedy et al. 2006 [40]

(++)

Pain and disability (SPADI)

+

Four patterns of response were found: cluster A had high disability at baseline and less improvement over a long course; cluster B had high disability at baseline but had a quick, steep improvement course; cluster C had moderate disability at baseline and, like A, a slow course with less improvement; and finally cluster D with lower disability at baseline and a short swift change to very low disability. Clusters C and D had a higher baseline Mental Component Score (SF-36 MCS, higher score indicates better health) than clusters A & B.

In the final model, one unit increase on the MCS is associated with approximately a 1.1 increase in the odds ratio of being in clusters C and D vs. clusters A and B. Therefore, a 10-unit increase on the MCS would be associated with approximately a 2.6 increase in the odds ratio of being in clusters C and D vs. clusters A and B.

Kromer et al. 2014 [41]

(+)

Pain and disability (SPADI)

–

–

Catastrophizing, measured by PCS, did not influence the baseline disability and change score in disability at 3 months; FABQ-P contributed significantly to baseline disability but not to the change score in disability at 3 months.

Kuijpers et al. 2006 [42]

(+)

Perceived recovery (^aADI)

–

–

–

–

Coping with pain (PCCL) N.S.; FABQ and TSK N.S. Univariate analysis for pain at 6 weeks but not for 6 months (4DSQ) significant, but not in the multivariate model; In univariate analysis for pain at 6 weeks but not for 6 months (4DSQ), not in the multivariate model; Anxiety (4DSQ) significant in univariate analysis for pain at 6 weeks but not for 6 months, not in the multivariate model.

Kvalvaag et al. 2018 [43]

(++)

Pain and disability (SPADI)

Work status (^aADI)

+

-

-

-

Univariate significant: SPADI baseline score, age, gender, work status, marital status, education, duration of pain, medication, self-efficacy for pain, outcome expectations, general health status, number of PT sessions and emotional distress.

Multivariate: low patient expectations were the strongest predictor of a negative outcome (Beta −4.2, 95% CI −7.2 to −1.1, p <  0.01). Self-efficacy, distress (HSCL-25) were no longer significant.

Outcome expectation, self-efficacy, distress univariate not significant.

Laslett et al. 2015 [44]

(++)

Pain and disability (SPADI)

+

–

Six months follow-up FABQ, OR 1.03 (95% CI 1.00–1.07), and 12 months FABQ OR 1.01 (95% CI 1.03–1.17) in the multivariate analysis.

SF-8 lower SF mental score in the multivariate model OR 0.93 (95% CI 0.85–1.01) 3 weeks, not significantly associated with outcome (3, 6, 12 months follow-up) in the univariate analysis.

O’Malley et al. 2004 [45]

(++)

Function (FLEX-SF)

+

In the final statistical model, patients with higher outcome expectancies (Patient Shoulder Expectancy Fulfilment measure) reported better 3-month shoulder functioning (Beta 0.46, p = 0.002).

Reilingh et al. 2008 [46]

(++)

Pain (NRS, in acute group)

Pain (NRS, in chronic group)

-

+

Catastrophizing (PCCL per point increase) univariate analysis (Beta 1.0, CI 0.44–1.57 (positive = more pain reduction) for decrease in pain at 6 months in acute shoulder pain patients but not in chronic shoulder pain patients. In the multivariate analysis catastrophizing is a negative predictor (less decrease of pain) in the chronic shoulder pain patients (Beta-0.62, CI −1.03- (−0.20)) and was no longer included in the acute pain patients; 4DSQ N.S.

Ryall et al. 2007 [47]

(++)

Pain (^aADI)

–

–

–

Belief that problem is likely to be causing difficulties in 3 months N.S.; Brief Symptom Inventory (BSI) >2points N.S.; Depression Scale (HADS)-D > 7 for continuing pain at 12 months, frequent continuing pain, unremitting pain N.S.; HADS-A > 7 continuing pain at 12 months, frequent continuing pain, unremitting pain N.S.

Sindhu et al. 2012 [48]

(+)

Shoulder function (Computerized Adaptive Test)

+

FABP-P > 16 high FAB: the improvement of function was greater in low fear avoidance groups after adjustment for 8 disease categories. No difference was found for arthropathies, fractures, sprains and strains, postsurgical conditions.

Smedbråten et al, 2018 [49]

(++)

Pain (NRS)

Function (PSFS)

+

-

In final multiple regression model, emotional distress (HSCL-25) associated with more pain (Beta 1.06, 95% CI 0.44–1.68, p = 0.001). Other significant predictors: pain intensity before treatment, duration of pain > 12 months.

Emotional distress univariate significant, not included in the multiple regression model. Significant predictors were higher pre-treatment disability, pain duration > 12 months, concomitant neck pain, and a lower level of education.

Van der Windt et al. 2007 [50]

(++)

Perceived recovery (^aADI)

Shoulder disability (SDQ)

-

-

-

-

-

-

Perceived recovery was measured by Likert scale. Catastrophizing (PCCL score) > 40 adjusted OR 0.94 (95% CI 0.52–1.68) for persisting symptoms, OR 1.32 (CI 0.78–2.24) for < 30% disability reduction; FABQ-P > 75 (0–100) adjusted OR 1.08 (CI 0.63–1.85) for persisting symptoms, OR 1.12 (0.568–1.85) for disability reduction; Somatization, measured by 4DSQ > 30 adjusted OR 1.46 (CI 0.63–3.42) for persisting symptoms, OR 1.49 (CI 0.74–3.01) for disability reduction; Distress 4DSQ > 12 adjusted OR 0.71 (CI 0.42–1.19) for persisting symptoms, OR 0.76 (CI 0.48–1.23) for disability reduction.

Wolfensberger et al., 2016 [51]

(++)

Shoulder disability (DASH)

Pain (Patient Global Impression of Change)

+

+

-

+

+

+

+

+

In the multivariable analysis factors were combined: HADS-A, HADS-D, and Pain Catastrophizing Scale (PCS) were associated with more disability (DASH, Beta 0.64 (95% CI 0.25–1.03, p = 0.002). Also, less Patient Global Impression of change associated with combination of: HADS-D + A + PCS + TSK (Beta 0.93, 95% CI 0.87–0.99, p = 0.026).