Annual Alcohol Screening and Brief Alcohol Interventions in the Veterans Health Administration

Abstract

Background: Alcohol screening and brief alcohol interventions (SBI) are ranked the third highest prevention priority for US adults. Although the US Preventive Services Task recommends routine SBI, the optimal screening interval is unknown and brief alcohol interventions (BI) have been challenging to implement. Objectives: To 1) evaluate the yield of repeat annual alcohol screening for patients who initially screen negative on 1-4 prior negative alcohol screens in the population as a whole and in subgroups of patients; and 2) estimate the prevalence of BI with alcohol misuse before, during, and after implementation of a national performance measure (PM) and dissemination of an electronic clinical reminder (CR) for BI. Methods: 1) Using VA electronic medical record data for more than 360,000 outpatients (2004-2008) the probability of a positive final screen after 1 -4 prior negative screens was estimated using adjusted logistic transition regression models that included prior negative screen scores as explicit predictors. Results are also presented as the number needed to screen (NNS) for one patient to screen positive on the final screen. 2) Among 6,788 VA outpatients whose medical records were reviewed for quality improvement (7/06-9/08) the adjusted prevalence of documented BI (advice and feedback) across the 4 phases of BI implementation was estimated using a generalized log-linear regression. Results: Age, gender and prior negative screen scores were the strongest predictors of a positive final screen. As the number of prior negative screens increased from 1-4, the probability of a positive final screen decreased and those who consistently reported no drinking the past year had the lowest probabilities. Depending on prior negative screen scores, age, and gender, the NNS ranged from 3 - 159. Among patients with alcohol misuse, the prevalence of BI increased significantly over successive phases of BI implementation, from 5.5% during the baseline year to 29.0% after CR dissemination. Conclusions: Extending the screening interval beyond a year may be appropriate for some patient groups with a low likelihood of a positive screen in settings that lack the resources to screen and follow-up on all patients annually. The VA's strategy of implementing BIs with a PM and CR meaningfully increased documentation over a 1-year period.