Sixteen studies (2,390, range 45 to 552) were included in the review. Overall study quality was rated as moderate-to-good. The main methodological concerns were time delay between clinical test and reference standard and lack of clarity regarding blinded test interpretation.
Ten studies of five clinical tests (1,684) were included in the meta-analyses.
Hawkins-Kennedy test (six studies, 1,029 patients): The pooled estimates of sensitivity and specificity were 74% (95% CI 57 to 85) and 57% (95% CI 46 to 67). The pooled positive likelihood ratio was 1.70 (95% CI 1.29 to 2.26) and the pooled negative likelihood ratio was 0.46 (95% CI 0.27 to 0.78).
Neer's sign (five studies, 1,127 patients): The pooled estimates of sensitivity and specificity were 78% (95% CI 68 to 87%) and 58% (95% CI 47 to 68). The pooled positive likelihood ratio was 1.86 (95% CI 1.49 to 2.31) and the pooled negative likelihood ratio was 0.37 (95% CI 0.25 to 0.55).
Empty can test (six studies, 695 patients): The pooled estimates of sensitivity and specificity were 69% (95% CI 54 to 81%) and 62% (95% CI 38 to 81). The pooled positive likelihood ratio was 1.81 (95% CI 1.16 to 2.83) and the pooled negative likelihood ratio was 0.50 (95% CI 0.40 to 0.63).
Drop arm test (five studies, 1,213 patients): The pooled estimates of sensitivity and specificity were 21% (95% CI 14 to 30) and 92% (95% CI 86 to 96). The pooled positive likelihood ratio was 2.62 (95% CI 1.60 to 4.30) and the pooled negative likelihood ratio was 0.86 (95% CI 0.79 to 0.94).
Lift-off test (four studies, 267 patients): The pooled estimates of sensitivity and specificity were 42% (95% CI 19 to 69) and 97% (95% CI 79 to 100). The pooled positive likelihood ratio was 16.47 (95% CI 1.43 to 185.61) and the pooled negative likelihood ratio was 0.59 (95% CI 0.37 to 0.97).
No results were reported for the six studies not included in the meta-analyses.