Reduced Alcohol Use Is Sustained in Patients Provided Alcohol-Related Counseling During Direct-Acting Antiviral Therapy for Hepatitis C

Abstract

Background

Patients with chronic hepatitis C and risky/harmful alcohol use experience poor outcomes. Granular data evaluating whether alcohol counseling during hepatitis C treatment impacts longitudinal alcohol consumption are lacking.

Aims

To evaluate whether provider-delivered counseling in the context of direct-acting antiviral hepatitis C treatment associates with decreased longitudinal alcohol consumption.

Methods

We performed secondary data analysis from the Hep ART study including adults with hepatitis C who underwent provider-delivered counseling during direct-acting antiviral treatment between October 2014 and September 2017. Demographics and disease characteristics were summarized. Alcohol consumption, abstinence, and heavy drinking were evaluated in periods before, during, and after direct-acting antiviral treatment. Multivariate regression analyses were performed to evaluate the association of alcohol consumption with each 12-week time period for all patients and a subsample with cirrhosis.

Results

One hundred twenty-three patients were included; 41 had cirrhosis. Most patients were male (74.0%) and Black (58.5%). Alcohol consumption improved during direct-acting antiviral treatment and was notably sustained (< 12 weeks before treatment 32.5 g/day; during treatment 20.0 g/day; and 12–24 weeks after treatment 23.7 g/day). Multivariable analyses showed significantly improved alcohol consumption metrics during and after antiviral treatment compared to < 12 weeks before treatment (during treatment 13.04 g/day less, p = 0.0001; > 24 weeks after treatment 15.29 g/day less, p = 0.0001). The subsample with cirrhosis showed similar results (during treatment 13.21 g/day less, p = 0.0001; > 24 weeks after treatment 7.69 g/day less, p = 0.0001).

Conclusions

Patients with chronic HCV and risky/harmful alcohol use given provider-delivered alcohol-related counseling during HCV treatment sustain decreased alcohol consumption patterns during and after treatment.

Introduction

Chronic hepatitis C virus (HCV) infection afflicts an estimated 71 million people worldwide [1] and can result in life-shortening conditions such as cirrhosis and hepatocellular carcinoma. Evidence has shown that alcohol use can accelerate fibrosis progression in chronic HCV patients and worsen both liver-related mortality and all-cause mortality [2]. Therefore, clinical guidelines do recommend alcohol abstinence and the consideration of interventions to achieve abstinence in those with active HCV infection [3, 4]. Notably, even after curative therapy for HCV, alcohol abstinence is stressed for those with advanced fibrosis in particular, as they are at risk for continued liver disease progression if alcohol use continues unabated. However, in the USA, heavy alcohol use is 1.3 times more common for patients with chronic HCV compared to those without infection [5]. As such, there is a need to focus clinical efforts on this segment of vulnerable chronic HCV patients, and even those who are cured, that consume considerable alcohol to optimize outcomes.

“Screening, Brief Intervention, and Referral to Treatment” (SBIRT) is an evidence-based alcohol reduction practice that follows the principles of motivational interviewing [6]. As a 5- to 10-min intervention, SBIRT establishes a dialogue with patients about their alcohol use and has a strong evidence base in reducing hazardous drinking patterns [7, 8]. Though SBIRT has been adopted for use in a variety of healthcare settings [9], there is evidence to suggest under-receipt of alcohol-related counseling for the drinking chronic HCV population [10]. Furthermore, there is a lack of granular data evaluating the impact of providing brief alcohol counseling during treatment for HCV with direct-acting antiviral (DAA) medications for those with chronic HCV infection and heavy alcohol use. Though alcohol use in this population has been shown to decrease in the context of DAA therapy and counseling [11, 12], there is clinical concern that such decreased consumption behavior may not be sustained over time once Hepatitis C is effectively cured. In this study, we therefore sought to evaluate whether providing alcohol-related counseling in the context of treating HCV with DAA medications decreased alcohol consumption even after DAA therapy was completed.

Methods

For this study, we used a behavioral clinical trial cohort including 181 participants across three academic centers enrolled from October 2014 to September 2017 described in our group’s original publication [11]. Inclusion criteria required previously confirmed chronic HCV via laboratory testing irrespective of current HCV treatment status, an Alcohol Use Disorders Identification Test (AUDIT) score of ≥ 4 in women and ≥ 8 in men, at least one alcoholic drink in the past 60 days, age ≥ 18 years, ability to understand and speak English, lack of current engagement in effective substance use therapy, a willingness to be involved in a 12-month study, and ability to access at least one individual therapy session in person. Potential participants with active psychosis as determined by medical providers were excluded. All participants at minimum received the evidence-based alcohol reduction intervention, SBIRT [7, 8, 13], delivered by their liver provider. In addition, 95 participants also received up to 6 months of colocated integrated care including individual and group therapy providing motivational, cognitive, and behavioral strategies to reduce alcohol consumption (the active arm in the original trial). Notably, in the original trial, significant reductions in alcohol consumption were seen in both the SBIRT and SBIRT plus integrated HCV–alcohol care arms with no significant difference between arms.

For the purposes of this study, we conducted a secondary analysis using a subpopulation (n = 123) of participants that received a single course of DAA therapy during the study period. We described the sample’s demographics and disease characteristics at Hep ART study baseline using summary statistics: percentage and count for the binary and categorical variables and mean and standard deviation for the continuous variables. Self-reported alcohol use data had been collected using validated Timeline Followback methods, capturing daily alcohol use across four interview time points for 15 months [14]. To evaluate characteristics of alcohol consumption over time in relation to DAA therapy, we divided the participants’ time course into > 24 weeks pre-DAA, 12–24 weeks pre-DAA, < 12 weeks pre-DAA, during DAA, < 12 weeks post-DAA, 12–24 weeks post-DAA, and > 24 weeks post-DAA. For each time period and each participant, we calculated (1) mean alcohol consumption in grams per day over all the observed days in the time period, (2) the percentage of days abstained from alcohol over all the observed days in the time period, and (3) the percentage of heavy drinking days over all the observed days in the time period. Then, for each time period, we calculated means and standard deviations of (1)–(3) over all participants that were observed in the time period, both for the entire analysis sample and for the subsample of participants who had cirrhosis at the Hep ART study baseline.

Since alcohol consumption in grams per day was not normally distributed in our data, a longitudinal multivariable regression model, using the generalized estimating equation (GEE) approach, was performed to evaluate the association of alcohol consumption with time periods in reference to the DAA timeline, controlling for socio-demographic characteristics, depression, anxiety, and Hep ART administration variables. The GEE approach does not require the normality assumption be satisfied as does standard linear regression [15].

To obtain estimates of prevalence ratios, a modified Poisson model [16] was also performed to evaluate the association of the two binary alcohol outcome variables, any alcohol consumption on a day and heavy drinking on a day, with time periods based on the DAA timeline, controlling for the same covariates. These model approaches were further used on a subsample of participants with cirrhosis. Because marital status was missing in a fraction of participants (n = 4), none of whom had cirrhosis, multiple imputation was performed for the multivariable regressions in the entire sample, but not the subsample.

Results

Most participants were of male gender (74.0%) and Black race (58.5%); the median income was $1090 per month with a substantial portion of participants being unemployed (20.3%) or on disability (30.1%) (Table 1). The mean baseline enrollment date for this study population was October 18, 2016; the study recruitment period ranged from October 2014 to September 2017. One-third of participants had cirrhosis at baseline, and more than half of patients (51.6%) reported a history of heavy alcohol use (binge drinking for 5 or more days in the past 30 days) during the study period (Table 1). Participants were observed before, during, and/or after DAA treatment, up to 334 days post-DAA plus 24 weeks. Mean alcohol consumption in grams/day decreased from before DAA treatment to during DAA treatment [< 12 weeks pre-DAA 32.5 (SD = 40.6) g/day versus during DAA 20.0 (SD = 34.2) g/d], and the decreased consumption pattern was sustained in the post-DAA period [12–24 weeks post-DAA 23.7 (SD = 34.1) g/d] (Fig. 1). The percentage of abstinence days similarly improved from before DAA treatment compared to during DAA treatment [< 12 weeks pre-DAA 56.7% (SD = 0.353) versus during-DAA 70.7% (SD = 0.356)], with a generally sustained improved abstinence pattern also noted post-DAA [12–24 weeks post-DAA 64.4% (SD = 0.362)] (Fig. 2). The percentage of heavy drinking days (days with five or more drinks for men, and four or more drinks for women) additionally was noted to improve during DAA treatment [< 12 weeks pre-DAA 25.8% (SD = 0.325) versus during DAA 15.8% (SD = 0.301)], and this response was likewise generally sustained in the post-DAA period [12–24 weeks post-DAA 18.1% (SD = 0.300)] (Fig. 3). In multivariate analyses controlling for demographic, socioeconomic, and study-related variables, participants demonstrated significant improvement in alcohol use behavior (consumption, abstinence, heavy drinking) during DAA and in the post-DAA period compared to less than 12 weeks before DAA (Table 2). In a subset of patients with cirrhosis (n = 41), multivariate analyses controlling for similar variables likewise demonstrated statistically significant improvement in consumption behaviors in the during-DAA and post-DAA periods (Table 3). A subset of patients that received SBIRT without integrated HCV–alcohol care (as in the original trial) further noted sustained reduction in alcohol use in grams/day in the post-DAA period (Supplemental Table 1).

Table 1.

Participant demographics and other characteristics

Participant characteristics (n = 123)	Mean (SD), % (n), or median (IQR)
Age (in years), mean (SD)	55.0 (10.5)
Gender, % (n)
Male	74.0% (91)
Female	26.0% (32)
Married (or cohabitating), % (n)	30.3% (36)
Race, % (n)
Single-racial Black, not Hispanic	58.5% (72)
Single-racial White, not Hispanic	32.5% (40)
Other (including bi- and multiracial)	8.9% (11)
Income ($/month), median (IQR)	$1090 (700, 2000)
Employment status, % (n)
Full-time	30.1% (37)
Part-time	9.8% (12)
Unemployed	20.3% (25)
Disabled	30.1% (37)
Other	9.8% (12)
Housing status, % (n)
Own/rent	74.8% (92)
Someone else’s housing	17.9% (22)
Shelter	3.3% (4)
Street/outdoors	1.6% (2)
Other	2.4% (3)
Hep ART treatment arm, % (n)
SBIRT + Hep ART alcohol therapy	55.3% (68)
SBIRT only	44.7% (55)
Study site, % (n)
Duke	59.3% (73)
UNC Chapel Hill	30.1% (37)
Durham VA	10.6% (13)
Presence of cirrhosis at Hep ART baseline, % (n)	33.3% (41)
Presence of conditions during Hep ART study period, % (n)
Heavy alcohol use*	51.6% (63)
Depression	48.8% (60)
Anxiety	43.9% (54)

^*Binge drinking for 5 or more days during the study period

Fig. 1.

Alcohol consumption over time

Fig. 2.

Abstinence over time

Fig. 3.

Heavy drinking days over time

Table 2.

Multiple imputed multivariable model of alcohol consumption outcomes with time periods on DAA timeline (n = 123)

	Generalized estimating equation, coefficient (95% CI)	Modified Poisson model, prevalence ratio (95% CI)
	Alcohol consumption (in g/day)	Any consumption on a day	Heavy drinking on a day
Time period on DAA timeline (reference: < 12 weeks pre-DAA)
> 24 weeks pre-DAA	14.86 (13.49, 16.24); p < 0.001	1.62 (1.19, 2.19); p = 0.002	1.53 (1.04, 2.24); p = 0.029
12–24 weeks pre-DAA	4.07 (2.87, 5.27); p < 0.001	1.13 (0.99, 1.30); p = 0.080	1.17 (0.97, 1.40); p = 0.106
During DAA therapy	− 13.04 (− 14.11, − 11.96); p < 0.001	0.65 (0.55, 0.78); p < 0.001	0.56 (0.43, 0.73); p < 0.001
< 12 weeks post-DAA	− 12.54 (− 13.65, − 11.43); p < 0.001	0.73 (0.59, 0.90); p = 0.004	0.61 (0.45, 0.85); p = 0.003
12–24 weeks post-DAA	− 12.31 (− 13.48, − 11.15); p < 0.001	0.72 (0.57, 0.92); p = 0.007	0.60 (0.41, 0.87); p = 0.008
> 24 weeks post-DAA	− 15.29 (− 16.47, − 14.11); p < 0.001	0.66 (0.47, 0.93); p = 0.016	0.53 (0.34, 0.83); p = 0.005
Female (ref = male)	− 15.57 (− 27.36, − 3.79); p = 0.010	0.74 (0.46, 1.18); p = 0.204	0.59 (0.34, 1.02); p = 0.060
Income at Hep ART baseline (in $1000/month)	7.09 (2.34, 11.84); p = 0.003	1.15 (1.00, 1.32); p = 0.057	1.40 (1.11, 1.76); p = 0.004
Housing at Hep ART baseline (reference: own/rent)
Shelter	− 15.38 (− 43.65, 12.88); p = 0.286	0.69 (0.17, 2.76); p = 0.596	0.56 (0.25, 1.24); p = 0.152
Street/outdoors	25.98 (− 13.45, 65.42); p = 0.197	1.55 (0.77, 3.13); p = 0.215	2.54 (0.94, 6.90); p = 0.067
Someone else’s housing	− 2.47 (− 16.64, 11.70); p = 0.733	0.64 (0.14, 2.95); p = 0.570	0.66 (0.25, 1.73); p = 0.397
Other housing status	55.66 (19.83, 91.49); p = 0.002	3.36 (1.28, 8.78); p = 0.014	4.63 (0.99, 21.67); p = 0.051

Marital status was imputed for four participants. Control variables were included in the model but not noted above given lack of statistical significance: depression, anxiety, age, married or cohabitation, race/ethnicity, employment, Hep ART baseline date, Hep ART treatment arm, study site. For the GEE model, identity link function and exchangeable covariance structure have been specified. Number of data point observations: model of alcohol consumption in grams 55,824; model of any consumption 55,839; and model of heavy drinking 55,825

Table 3.

Alcohol consumption for participants with cirrhosis at baseline (n = 41)

Time period on DAA timeline	Alcohol consumption (in g/day)		Heavy drinking on a day
	Grams per day over the time period, mean (SD)	GEE multivariate model, coefficient (95% CI)	Percentage of heavy drinking days over the time period, mean % (SD)	Modified Poisson multivariate model, prevalence ratio (95% CI)
> 24 weeks pre-DAA (n = 25)	34.8 (37.5)	17.68 (15.54, 19.82); p < 0.001	28.1% (.334)	2.52 (0.27, 23.48); p = 0.417
12–24 weeks pre-DAA (n = 38)	37.5 (48.5)	7.98 (5.81, 10.15); p < 0.001	30.3% (.380)	1.51 (0.77, 2.95); p = 0.230
< 12 weeks pre-DAA (n = 39)	26.8 (37.6)	Reference	18.6% (.275)	Reference
During DAA therapy (n = 39)	13.9 (27.8)	− 13.21 (− 15.22, − 11.20); p < 0.001	8.4% (.223)	0.43 (0.21, 0.90); p = 0.026
< 12 weeks post-DAA (n = 35)	16.2 (30.7)	− 12.80 (− 14.96, − 10.63); p < 0.001	9.1% (.165)	0.43 (0.21, 0.89); p = 0.022
12–24 weeks post-DAA (n = 35)	29.5 (44.9)	− 5.27 (− 7.68, − 2.86); p < 0.001	17.6% (.313)	0.71 (0.38, 1.34); p = 0.291
> 24 weeks post-DAA (n = 20)	32.4 (52.9)	− 7.69 (− 10.22, − 5.16); p < 0.001	20.5% (.359)	0.70 (0.31, 1.58); p = 0.391

Control variables included in the GEE model and Modified Poisson model: depression, anxiety, age, gender, married or cohabitation, race/ethnicity, income, employment, housing, Hep ART baseline date, Hep ART treatment arm, study site. For the GEE model, identity link function and exchangeable covariance structure have been specified. Number of data point observations: model of alcohol consumption 18,875; model of heavy drinking 18,875

Discussion

Patients with chronic HCV infection and significant alcohol consumption are particularly at risk for accelerated liver damage and poor outcomes, including liver decompensation and death. A previous study has also shown that adults with HCV are three times more likely to consume more than one alcoholic drink per day (35% vs. 14%) and around eight times more likely to consume three drinks per day (19% vs. 2%) compared to adults without HCV, underscoring the need to focus on this high-risk population [17]. Though alcohol-related counseling is evidence-based and recommended as standard of care for patients with risky or harmful alcohol use [18], current studies suggest under-receipt of recommended alcohol-related care, including brief intervention counseling, for patients with chronic HCV and risky alcohol use [10]. With the advent of highly efficacious DAA treatment for chronic HCV comes a refreshed opportunity for liver providers to systematically engage in alcohol-related counseling in the context of a therapeutic patient–provider relationship. Though there is evidence that alcohol use metrics may decrease generally during DAA treatment when counseling is provided [11], it is unclear whether these results are sustained after DAA treatment and effective cure of HCV. Our study is the first to granularly examine the change in alcohol consumption patterns for patients with chronic HCV infection and risky alcohol use who are provided behavioral counseling in the context of DAA treatment. Results from our study demonstrate the powerful and sustained impact behavioral counseling plus the opportunity to be cured of HCV can have on alcohol-related outcomes for this vulnerable population.

Our study by design is a secondary analysis of our group’s previously published clinical trial [11]. A key finding in this original study was that provider-delivered alcohol counseling through SBIRT paired with integrated colocated alcohol therapy fared no better than provider-delivered SBIRT alone in improving alcohol consumption metrics, although both improved alcohol use to a clinically meaningful degree. It is important to underscore that practical implementation of provider-driven SBIRT into liver clinic practices is feasible, if prioritized. In our study, liver providers were provided a 2-h training session regarding SBIRT, as well as National Institute of Alcohol Abuse and Alcoholism-developed handouts for SBIRT. Clinical flow required the clinic receptionist to distribute an alcohol screener to patients and medical providers to engage in brief alcohol counseling during clinic visits (Fig. 4). A referral out for alcohol treatment occurred at the end of the visit. Though in our study the study team conducted these referrals, in the absence of research funding, someone in the liver clinic structure such as a case manager would need to take responsibility for arranging referrals for specialty addiction care. Given the relative lack of logistical barriers and the vulnerability of this particular population to poor health outcomes, we believe our study highlights the importance of providing guideline-driven alcohol-related behavioral counseling in a systematic fashion during liver clinic visits. Such efforts for alcohol-use monitoring and alcohol-related counseling would ideally be performed serially during liver clinic visits, even after HCV treatment is completed, to minimize risk of harmful/risky alcohol use over time.

Fig. 4.

Clinical workflow

In addition, a key finding in our study is in regard to the subpopulation of patients with cirrhosis, as this population is particularly at increased risk for liver decompensation if alcohol use continues, even if HCV is effectively cured. Our study found that compared to a pre-DAA baseline, alcohol consumption was statistically significantly reduced even at > 24 weeks post-DAA treatment. Compared to a reference of < 12 weeks pre-DAA, in the < 12 weeks to > 24 weeks post-DAA period, consumption estimates ranged from a decrease of 5.3 to 12.8 g/day, representing a clinically meaningful reduction. A further reduction in heavy drinking days was also noted in the during-DAA and early post-DAA period. This change in longitudinal drinking behavior for patients with known advanced liver disease can have a particularly striking impact on the natural history of liver function for these patients.

Our study’s primary strength relates to the rich dataset of 123 patients and their alcohol use reports by day across 15 months, allowing us to granularly dissect the change in alcohol consumption patterns over time by grams/day, percentage abstinence days, and percentage heavy drinking days. More than half of our patients were noted to be heavy alcohol users. As this was a secondary analysis of a clinical trial, another strength is that all clinical providers were given the same standardized SBIRT training session and materials. Our population was also served in three different clinical care sites (Duke, UNC, Durham VA), and data that may influence alcohol consumption behavior such as socioeconomic factors were meticulously collected to allow for inclusion in multivariable analysis.

Limitations in our study are due to its nature of being a secondary analysis. Though all patients received SBIRT for behavioral counseling, in a randomized fashion about half also received a more intensive colocated alcohol treatment by an addiction therapist. However, because in the original study and in our secondary analyses the addition of a colocated alcohol treatment did not prove to be superior to SBIRT only, the current study’s findings may be applicable to patients with HCV who do not receive intensive alcohol treatment. Also, the original trial was not designed specifically to answer the current study’s questions, and therefore alcohol use data were not available for a full 24 weeks before and after DAA treatment on most participants; the current analyses report aggregate snapshots for each time period. Although unlikely, it is possible that a different alcohol use pattern in relation to DAA treatment would emerge if full pre- and post-DAA data were available for each individual participant. Finally, alcohol use was self-reported at just four time points and participants had to recall their daily alcohol use. A study of various alcohol measures in over 1700 participants found when using the Timeline Followback method as we did, self-reported alcohol use agreed with that of a significant other at baseline 97% of the time and 85% at a 15-month follow-up, with participants reporting more drinking than the significant others knew to report [19]. This Timeline Followback method has been shown to be superior to biochemical tests (gamma-glutamyl transpeptidase, aspartate aminotransferase, alanine aminotransferase, and carbohydrate deficient transferrin all underreported drinking in Project MATCH), which would be biased by preexisting liver disease in our population [19]. In our study, we did collect urine ethyl glucuronide and urine alcohol level at four time points for each patient when feasible. In total, we had testing data on 112 of 123 participants (91%). Only five of 112 participants (4%) had discrepant results where alcohol use was denied though testing was positive. We believe this demonstrates the robustness of the Timeline Followback methodology.

In conclusion, our study demonstrated that provider-led alcohol counseling in the context of treating HCV with DAA therapy was associated with a substantial and lasting impact on reduced alcohol consumption patterns over time. Participants were able to reduce alcohol use immediately prior and during DAA treatment, and importantly they generally chose to sustain reductions in alcohol use even >24 weeks post-DAA treatment. This was furthermore noted in a subsample of participants with known cirrhosis, who are at particular risk for poor outcomes if alcohol use continues even if HCV is cured. Liver providers should systematically provide alcohol-related counseling to patients with heavy alcohol use within clinic sessions.