It has been estimated that around 20% of patients attending their GP are at risk from their drinking or have an alcohol use disorder. Without using specific screening tools GPs may typically detect about 40% of cases they see, but miss the majority.

It has been suggested that by using more structured screening questions GPs might identify more patients with alcohol problems, enabling them to signpost more at-risk patients to alcohol brief interventions or treatment services.

However, time within the routine consultation is short, and there has been little evidence to support which is the best screening approach to take.

Mitchell and colleagues performed a meta-analysis of studies that examined the diagnostic accuracy of simple verbal questions in identifying at-risk alcohol use in the primary care setting.

Their literature search revealed 17 studies that had been carried out in primary care using a robust interview standard. Fifteen studies used a single screening question and two studies tested a two question approach. The total study population was 5,646 individuals.

Across the 15 studies a single question identified 55% of those with problematic drinking, sensitivity 54.5% (95% CI:43.0-65.5%), and correctly excluded 87% of patients with no alcohol problems, specificity 87.3% (95% CI:75.7-83.6%).

The single question with the highest sensitivity was a modification of the Single Alcohol Screening Question (M-SASQ): ‘How often do you have six or more standard drinks on one occasion?’ with monthly or more frequently taken as a positive result.

The two studies investigating a two question approach showed an overall sensitivity of 87.2% (95% CI:69.9-97.7%) and a specificity of 79.8% (95% CI:75.7-83.6%).

It is currently recommended practice that a positive simple alcohol screening test is followed up with a more in-depth screening questionnaire, the so-called algorithm approach. Looking at two of the studies, the authors investigated the effectiveness of a one or two question initial screen followed by use of either the CAGE or AUDIT questionnaire.

They concluded that the most efficient approach was to use a two question screen followed by the CAGE or AUDIT questionnaire, with these systems having almost exactly the same accuracy: 0.909 (95% CI:0.891-0.926) and 0.894 (95% CI:0.875-0.913) respectively.

Applying these numerical data to a real life situation the authors estimate that taking 100 hypothetical patients, and assuming a prevalence of problematic alcohol use of 21%, a GP using a single question to screen for alcohol use disorder would correctly identify 11 cases and miss ten. The GP would correctly identify 69 non-cases and falsely diagnose ten. The overall accuracy would be 80.4%.

Using an algorithm approach with the same assumed prevalence, a GP using a single question followed by the CAGE test would correctly identify eight cases and miss 13. The GP would correctly identify 78 non-cases and misdiagnose one. Overall accuracy would be 86.1%.

The authors are wise to balance extra questions and time used in alcohol misuse screening against improved sensitivity and specificity to bring their research into the real world of general practice consultations.

Their cautious recommendation is a single or two question approach to initial screening, followed by either the CAGE or AUDIT test for those who are positive, and onward referral for those who test positive on the more in-depth questionnaire.