Six studies reporting a total of 16 diagnostic accuracy data sets were included. The studies included a total of 713 participants.
In terms of methodological quality, no study used the optimal reference standard (EMG and imaging combined) and only one was conducted in a primary care setting. The majority of the included studies (5 out of 6) were judged susceptible to disease progression bias and/or spectrum bias. Review bias, caused by the results of the reference standard being available to those interpreting the index text, or vice versa, was a concern for 3 of the 6 studies.
Shoulder abduction test: the sensitivity ranged from 0.17 (specificity 0.92) to 0.78 (specificity 0.75) and the specificity from 0.75 (sensitivity 0.78) to 0.92 (sensitivity 0.17), based on 3 studies.
Spurling's test: the sensitivity ranged from 0.30 (specificity 0.93) to 1.00 (specificity 0.94) and the specificity from 0.74 (sensitivity 0.50) to 1.00 (sensitivity 0.90), based on 4 studies and 7 data sets.
ULTT: the sensitivity ranged from 0.72 (specificity 0.33) to 0.97 (specificity 0.22) and the specificity from 0.11 (sensitivity 0.83) to 0.33 (sensitivity 0.72), based on 2 studies and 3 data sets.
Traction/neck distraction test: the specificity ranged from 0.90 to 0.97, based on 2 studies, with a sensitivity of 0.44 in both cases.
Valsalva manoeuvre: 1 study reported a sensitivity of 0.22 and a specificity of 0.94.