The VA screens over 95% of outpatients annually for alcohol misuse with the AUDIT-C which yields a scaled score (0-12 points). Since 2008, outpatients who screen positive with scores 5-12 are expected to be offered preventive brief alcohol interventions. VA clinicians, clinical leaders, and researchers are interested in knowing whether AUDIT-C scores can be used to measure drinking outcomes, but the AUDIT-C has never been validated as a measure of change. Resolution of alcohol misuse on the AUDIT-C could reflect biased reporting whereby patients purposely screen negative at follow-up to avoid stigma or being counseled about their drinking.
This study used objective clinical outcomes known to be associated with alcohol consumption to test whether resolution of alcohol misuse based on screening with the AUDIT-C was associated with expected decreases in objective alcohol-related measures. Main aims included determining whether resolution of alcohol misuse at follow-up screening, among VA patients who had screened positive for alcohol misuse on their 1st AUDIT-C, was associated with (1) decreases in high density lipoprotein cholesterol (HDL) - an alcohol biomarker that decreases as alcohol consumption decreases, and (2) lower risk of care for acute alcohol-related gastrointestinal (GI) conditions or trauma.
This retrospective cohort study used secondary data from the VA's Corporate Data Warehouse (CDW), National Patient Care Databases, Vital Status Master files, and VA-Medicare Datasets. Eligible patients received care at 24 VA facilities that had AUDIT-C data from 2004-2007 in the CDW and were screened with the AUDIT-C on two occasions at least 12 months apart during the study. Patients were excluded if they died in the year after their 2nd AUDIT-C or had no VA care during that year. Samples for main analyses were VA outpatients with scores of 5 or more on their 1st AUDIT-C, considered a "positive" screen in this study. Aim 1 analyses evaluated the association of changes in AUDIT-C scores with changes in HDL measured in the year after each AUDIT-C, based on a linear regression model that adjusted for baseline HDL and accounted for correlated data within facilities. By design, change versus change models also control for all time-invariant patient characteristics. Aim 2 analyses evaluated whether patients who resolved alcohol misuse on their 2nd AUDIT-C had lower risk of two alcohol-related medical conditions in the year after their 2nd AUDIT-C, compared to those who did not resolve misuse, using adjusted logistic regression. The two outcomes were VA or Medicare documentation of (1) "trauma," defined as a primary inpatient trauma diagnosis (including fractures) or outpatient fracture diagnosis, and (2) "GI hospitalization," defined as a primary inpatient diagnosis of liver disease, pancreatitis or upper GI bleeding. Patients who stopped drinking (AUDIT-C score 0) and two groups that continued drinking but decreased AUDIT-C scores below 5 (1-3 men/1-2 women and 4 men/3-4 women) were considered separately because prior studies have suggested that these groups vary in their risk for these outcomes (nondrinkers at higher risk than drinkers). The a priori defined referent group was patients who remained positive on their 2nd AUDIT-C (scores 5-12). Main analyses were adjusted for age, gender, race, marital status, VA eligibility, days between screens and facility, and accounted for correlated data within facilities.
The study sample included 44,146 patients who screened positive on their 1st AUDIT-C and had another AUDIT-C at least a year later, with 56% resolving alcohol misuse on their 2nd AUDIT-C. In the sub-sample of 25,287 VA patients who had an HDL in the year after each screen, patients who decreased their AUDIT-C scores at repeat screening were more likely to decrease their HDL, as hypothesized. For each 1-point decrease in AUDIT-C score: HDL dropped 0.40 mg/dl (0.34-0.45); for a 12-point decrease in AUDIT-C score: HDL dropped 4.76 mg/dl (4.11-5.40).
Compared to those with persistent alcohol misuse, patients who resolved alcohol misuse on their 2nd screen were less likely to have "GI hospitalizations" in the subsequent year if they reported drinking on the 2nd screen (AUDIT-C score 1-4; p-values<0.01), but not if they reported abstaining on the 2nd screen (score 0). Adjusted rates of GI hospitalization in the year after the 2nd AUDIT-C were: 1.00% (0.78-1.23) in nondrinkers; 0.76% (0.60-0.93) in low-level drinkers (1-3 men/1-2 women); 0.80% (0.65-0.96) in those with borderline AUDIT-C scores (4 men/3-4 women); and 1.15% (1.05-1.26) in those with persistent alcohol misuse. Exploratory analyses indicated that a subgroup of patients with persistent alcohol misuse on the AUDIT-C increased their AUDIT-C scores from 5-8 to 9-12, and that these patients and those with persistent scores of 9-12 accounted for the increased risk of GI hospitalizations among patients with persistent alcohol misuse.
Compared to those with persistent alcohol misuse, patients who resolved alcohol misuse on their 2nd screen did not, however, have lower risk of trauma in the year after the 2nd screen, irrespective of the 2nd AUDIT-C score. Adjusted rates of trauma were 4.29% (3.84-4.74) in nondrinkers (AUDIT-C 0); 3.57% (3.14-4.00) and 3.70% (3.43-3.97) in the groups with AUDIT-C scores 1-4; and 3.85% (3.68-4.03) in those who did not resolve alcohol misuse. When analyses were restricted to patients with no prior trauma, the expected association of between decreased alcohol misuse and decreased risk of trauma was observed.
Planned secondary analyses revealed that patients who increased alcohol screening scores from negative AUDIT-C screens to positive AUDIT-C screens had significant increases in HDL, GI hospitalizations and trauma in the year after their 2nd AUDIT-C compared to those who did not develop alcohol misuse, further supporting the validity of changes in AUDIT-C scores.
Increases and decreases in AUDIT-C scores were associated with the expected changes in three objective alcohol-related outcomes, with the exception that resolution of alcohol misuse was not associated with decreased risk of trauma at follow-up in main analyses. Overall findings indicate that resolution of alcohol misuse at follow-up screening is a patient-reported outcome measure that has predictive validity. These results suggest that persistently high AUDIT-C scores might be used to identify patients for more intensive interventions.
External Links for this Project
Grant Number: I01HX000115-01
- Bradley KA, Rubinsky AF, Lapham GT, Williams EC, Au DH, Blough DK, Berger D, Johnson-Chavez M, Hawkins EJ. Changes in AUDIT-C Scores Reflect Changes in Drinking - Results of a Biomarker Study. Presented at: Research Society on Alcoholism Annual Scientific Meeting; 2012 Jun 25; San Francisco, CA. [view]
- Williams EC, Rubinsky AF, Lapham GT, Achtmeyer CE, Rittmueller S, Bradley KA. No Association between Documented Brief Intervention and Resolution of Alcohol Misuse in 30 VA Medical Centers. Poster session presented at: International Network on Brief Interventions for Alcohol Problems Annual Conference; 2012 Sep 28; Barcelona, Spain. [view]
- Williams EC, Rubinsky AF, Lapham GT, Achtmeyer CE, Rittmueller S, Bradley KA. No Association between Documented Brief Intervention and Resolution of Alcohol Misuse in 30 VA Medical Centers. Poster session presented at: Addiction Health Services Research Conference; 2012 Oct 18; New York, NY. [view]
- Bradley KA, Rubinsky AD, Berger D, Lapham GT, Williams EC, Chavez LJ, Hawkins EJ. Resolution of Alcohol Misuse on the AUDIT-C and Risk of Hospitalization for Acute Gastrointestinal (GI) Conditions. Paper presented at: Addiction Health Services Research Conference; 2012 Oct 1; New York, NY. [view]
- Lapham GT, Rubinsky AD, Williams EC, Hawkins EJ, Kivlahan DR, Maynard C, Bradley KA, Heagerty PJ. The Association between a Positive Screen for Alcohol Misuse and Prior Negative Screen Scores. Paper presented at: Addiction Health Services Research Conference; 2012 Oct 18; New York, NY. [view]
- Chavez LJ, Rubinsky AD, Clark B, Hebert P, Maynard C, Au D, Bradley KA. The Association between Alcohol Misuse and Increased Inpatient Utilization among Veterans Affairs Patients. Paper presented at: Addiction Health Services Research Conference; 2012 Oct 1; New York, NY. [view]
Mental, Cognitive and Behavioral Disorders, Substance Use Disorders, Health Systems
Diagnosis, Treatment - Observational, Prognosis
Alcohol, Implementation, Screening