Veterans with hepatitis C (HCV) and other liver disease report high rates of current or prior alcohol use disorders. Alcohol consumption is particularly problematic among this population as it can negatively impact eligibility for antiviral therapy and worsen liver disease progression. Despite abundant data demonstrating effectiveness, person-delivered Brief Alcohol Interventions (BAIs) are not always implemented in VA liver clinics due to many potential barriers (e.g., scarce provider time, limited clinic resources, and logistical concerns). This results in a gap in care for many Veterans with liver disease who would likely experience improved health if they reduced or eliminated their alcohol use.
The objectives of this study are: 1. Redesign an existing web-based BAI for use with Veterans with HCV and other liver diseases, and assess patient, provider, and system factors that may impact the adoption of this intervention within Liver clinics. 2. Implement and examine the effectiveness of the redesigned BAI in Liver clinics at two VA medical centers. 3. Conduct budget impact analyses to estimate the short-term costs of adoption and diffusion of the web-based BAI within these clinics and estimate the trajectory of health care spending for study participants.
Objective 1: We conducted qualitative semi-structured interviews with Veterans and clinic staff at both VA sites to identify potential facilitators or barriers to the initial adoption of the web-based BAI, and to obtain patient and provider feedback on the content and presentation of the existing BAI to help inform its redesign for use with Veterans with HCV and other liver disease.
Objective 2: We conducted a randomized controlled trial utilizing a hybrid (Type 1) design that tests the redesigned web-based BAI, and also gathers information on implementation. Patient-level clinical outcome data and formative evaluation data on the feasibility of implementation will be collected at both study sites.
Objective 3: We are estimating the budgetary impact from the VA's perspective using a one year time horizon, without discounting. We are estimating the costs of the intervention and the subsequent health spending for participants during their enrollment in the study (six months) and six months after enrollment for a total of 1 year.
To achieve study objective 1, we conducted qualitative interviews with Veterans with hepatitis C (N = 30) and liver clinic staff (N = 9) at both study sites. Qualitative interviews were conducted to obtain feedback on the elements of a computer-delivered BAI for the purpose of re-versioning this program for use in VA liver clinics. Findings from the qualitative interviews revealed several domains of content to be revised including: (a) general format and structural modifications and (b) content revisions including the addition of information about the health effects of alcohol misuse for persons with hepatitis C. A second step in the revision process included presenting qualitative findings to expert consultants, finalizing the list of revisions based on meetings with these individuals, and collaborating with a programmer to execute these modifications into a revised version of the web-based BAI to be used among Veterans with hepatitis C. (c) A second round of qualitative interviews were conducted with the first 10 participants in the second aim. Participants responded positively to the changes and requested no new revisions.
For Aim 2, we completed a randomized clinical trial of the brief alcohol intervention versus usual care. Final enrollment was below original projects but still high enough (n=138 with over 80% follow-up at both waves) to allow rigorous assessment of outcome.
Preparatory to generalized estimating equation (GEE) models, we assessed differences between conditions at baseline. Two variables (gender and hepatitis C diagnosis) were significantly associated with condition (intervention or control) at baseline and therefore were included as covariates in all the models. Alcoholic cirrhosis was also significantly associated with number of unhealthy drinking days and therefore was included in the model for this outcome. Ethnicity was significantly associated with number of drinks per drinking day and was included in this model. No other covariates were significant at .05 level in bivariate analysis to be included.
Number of days with any drinking. Number of drinking days decreased over time for Veterans in both the intervention and the control group. The interactions between condition (intervention or control) and time (both 3-month and 6-month) were not significant. However, with these interaction terms present in the model, significant differences in the number of drinking days were observed between the two conditions at the 6-month follow-up only. Specifically, Veterans in the intervention condition reported significantly fewer drinking days (predicted mean = 4.96) than Veterans enrolled in the control condition (predicted mean = 8.88) at 6-month follow-up (mean ratio = 0.56, p=.031).
Number of unhealthy drinking days. In contrast to the number of drinking days, we observed a significant interaction between condition (intervention or control) and time at 3-month interview (p<.05) but not at 6-month interview for unhealthy drinking days. Specifically, Veterans in the intervention group reported significantly fewer unhealthy drinking days than those in the control group at the 3-month interview (predicted means were 2.30 and 5.24 respectively for the intervention and the control groups with the mean ratio of 0.44, p=.049). Even though a similar trend was observed at the 6-month follow-up, the difference did not reach significance at the .05 level. In addition, number of unhealthy drinking days significantly reduced at each follow-up period when compared to the baseline for Veterans in the intervention group (p<0.05 for the comparisons between baseline and each of the follow-ups) but not for the control group.
Number of drinks per drinking day. Both the intervention and the control group reported significantly fewer drinks per drinking day at each of the follow-ups when compared to the baseline (p<.05 for all the comparisons). The interactions between condition (e.g., intervention or control group) and time (3 and 6 months) were not significant. No significant intervention effect was found at either the follow-up periods indicating no support for the hypothesis that the intervention affected this outcome at any point.
At this writing, we have run preliminary models on outcomes reflecting overall physical and psychological well-being. Although we have not finalized these analyses, they currently point to the intervention having no effect on these secondary outcomes.
Aim 3, the budget impact analysis, is currently in progress.
For individuals with HCV and other liver disease, alcohol misuse can worsen disease progression and be a significant barrier to receiving antiviral therapy for HCV. Our study developed a tool that all clinics in VA can now use to intervene with problem drinking among liver disease patients. The trial showed that web-based BAIs are a feasible and effective approach for reducing alcohol consumption in Veterans with HCV and other liver disease, and by extension could be tried in other care settings as well (e.g., surgery). If our web-based BAI is cost-effective, which we are now assessing, Office of Mental Health and Suicide Prevention, and the VA Office of Public Health are interested in implementing this approach across VA medical centers nationally.
External Links for this Project
Grant Number: I01HX000974-01
None at this time.