Nine studies were included in the review (6,353 patients). Sample size ranged from 103 to 5,065.
Detecting alcohol abuse and dependence: CAGE (three studies) at a cut-off of 1 or more was shown by Receiver Operator Curves from two of the studies to be the most efficient. Sensitivity ranged from 79.1% to 88% and specificity ranged from 55.8% to 88%.
Variations of MAST (four studies): The Receiver Operator Curves from two studies showed a cut-off of 4 or more was most efficient. However, only one study used used this cut-off, which gave a sensitivity of 91.4% and a specificity of 83.9%. A Receiver Operator Curve for the MAST-G suggested a cut-off of 5 or more. Two studies that used the MAST-G found that sensitivity ranged from 69.8% to 91% and specificity ranged from 80.5% to 84%.
Detecting hazardous or excessive drinking:
AUDIT (one study): a cut-off of 8 or more gave a sensitivity of 33.3% and a specificity of 90.7%.
CAGE (three studies): two studies looked at a cut off of 1 or more and found that sensitivity ranged from 31% to 60% and specificity ranged from 92% to 100%; two studies that looked at a cut off of 2 or more found that sensitivity ranged from 14% to 38.9% and specificity ranged from 97% to 97.1%.
SMAST (one study): SMAST performed poorly at a cut off of 2 or more, with a sensitivity of 48% and a specificity of 100%.
Variations of AUDIT (two studies): AUDIT at a cut-off of 8 or more showed a sensitivity of 66.7% and a specificity of 95.3%; AUDIT C at a cut-off of 3 or more showed a sensitivity of 100% and a specificity of 80.7%.
ARPS and shARPS in comparison to AUDIT and SMAST-G (one study): the study did not report the cut-off for ARPS and shARPS. Sensitivity was 93% (ARPS) and 92% (shARPS) compared to 28% for the AUDIT and 52% for the SMAST-G. Specificity was not so good at 63% and 51% for the ARPS and shARPS compared with 100% and 96% for the AUDIT and SMAST G.