Estimating risk of alcohol dependence using empirically validated ordinal risk zones versus recommended binary risk zones of the RAPS4: a validation study using stratum-specific likelihood ratio analysis.

BACKGROUND: Effective treatment options for alcohol dependence exist; yet, only 10% of people with alcohol dependence receive treatment. The objective of the current study was to examine the performance of previously recommended Rapid Alcohol Problem Screen 4 (RAPS4) risk zones, based on single binary cut-points (RAPS4 ≥ 1; RAPS4 ≥ 2), and empirically identified RAPS4 risk zones to identify people with alcohol dependence so that further diagnostic assessment or interventions can be offered. METHOD: Stratum-specific likelihood ratio (SSLR) and receiver operating characteristic analyses were used to compare the screening performance of empirically identified "risk zones" on the RAPS4 to previously recommended binary cut-points in a general population sample of current drinkers in Ireland (N = 4,267). SSLRs were also used along with the pretest prevalence of alcohol dependence to estimate posttest probabilities of alcohol dependence for the recommended and empirically identified risk zones. RESULTS: The weighted prevalence estimate of alcohol dependence among current drinkers was 6.9% (9.3% men; 4.5% women). The SSLR analysis identified multiple risk zones in the RAPS4, with each of the individual scores (0, 1, 2, 3, 4) retained as 5 separate ordinal risk zones for both men and women. A comparison of the area under the receiver operating characteristic curve showed that the ordinal RAPS4 risk zones performed better than recommended binary thresholds for both men and women. Based on the pretest probability of 9.3% and the identified SSLRs for the ordinal risk zones, the posttest probability of alcohol dependence for men ranged from 1.6% for those in the lower risk zone (RAPS4 = 0) to 86.7% for those in the highest risk zone (RAPS4 = 4). The posttest probability of alcohol dependence for women ranged from 0.4% for those in the lower risk zone to 80% for those in the higher risk zone. CONCLUSIONS: The detection of alcohol dependence may be improved using the empirically identified ordinal RAPS4 risk zones for both men and women. The application of the identified SSLRs, particularly if integrated into a clinical decision support system, may be helpful for clinicians in providing feedback to patients regarding their risk of alcohol dependence