Psychosocial interventions in comorbid AUD and ARLD: contingency management

Interventions targeting engagement and abstinence outcomes may provide an opportunity for improving patient outcomes and health-related quality of life in comorbid AUD and ARLD. A systematic review comprised of 13 studies with 1945 patients with ARLD evaluated the efficacy of psychosocial interventions to induce and maintain abstinence in comorbid AUD and ARLD (Khan et al., 2016). The included studies trialed combined psychotherapy with cognitive behavioral therapy (CBT), Motivational Enhancement Therapy (MET), motivational interviewing, supportive therapy, and psychoeducation alone or in combination with treatment as usual. The review found that psychotherapy combined with CBT, MET, and integrated care with CBT induced abstinence; no psychosocial intervention was effective in maintaining abstinence. However, returning to drinking was lower with integrated care. The findings noted that integrated, multidisciplinary approaches appear promising in addressing this clinical population's unique needs and preferences. This research also pointed to the need to trial novel, tailored psychosocial interventions to improve patient outcomes in comorbid AUD and ARLD.

Contingency management (CM) is a psychosocial intervention that can support positive behavior change, with its principles founded upon the underpinnings of behaviorism and operant conditioning. By employing positive reinforcement, CM is based on the notion that a desired behavior reinforced in close temporal proximity to its occurrence will increase in frequency (Petry, 2011). This can increase the frequency of voluntary, health-promoting behaviors such as abstinence, reduced intake, or treatment engagement. A growing body of empirical evidence has supported the effectiveness of CM across a range of target behaviors (substance or non-substance-focused), clinical populations receiving substance use treatment (cocaine, opiates, polysubstance use, cannabis, alcohol or nicotine), modes of incentive (clinical privileges, prizes, vouchers) and schedules (Lussier et al., 2006; Pfund et al., 2024).

Psychosocial interventions such as CM can address barriers associated with pharmacological approaches and low treatment engagement. Accordingly, a multi-centre, three-arm parallel-group randomized controlled trial (RCT) has supported the scope of CM to support medication adherence (Donoghue et al., 2023). However, no previous research has evaluated a CM intervention among a clinical population with comorbid AUD and ARLD, whose effective management often involves complex, multidisciplinary treatment regimens (Khan et al., 2016).

Evidence gap and current study

While CM has been well established in substance use treatment, limited research has been conducted in AUD and clinical settings outside of the USA (Cohen et al., 1971; Petry et al., 2000). In addition, no previous research has evaluated a CM intervention among a clinical population with comorbid AUD and ARLD, whose effective management often involves complex, multidisciplinary treatment regimens (Hemrage et al., 2023; Khan et al., 2016). This gap reflects a need for novel interventions simultaneously targeting AUD and ARLD outcomes to be trialed in a real-world context (Lee et al., 2024; Mellinger et al., 2021; Singal et al., 2022).

To address this gap, a pilot trial was conducted to explore the feasibility of delivering CM to promote engagement with multidisciplinary treatment. The primary aim was to characterize the demographic profile of the clinical population of interest and generate data on feasibility to inform progression to a definitive RCT. The pilot trial also aimed to generate exploratory data on treatment engagement, alcohol use, and liver function following a CM intervention.

Design and procedures

A single-centre, randomized pilot trial of Integrated Liver Care (ILC) only (control) versus ILC + CM (intervention) was conducted at King's College Hospital NHS Foundation Trust (KCH) in partnership with the Alcohol Care Team (ACT) and Liver Outpatient Department (LOPD). The study aimed to explore the acceptability and feasibility of delivering CM to promote treatment engagement among the target clinical population. The present article reports on feasibility data (quantitative) and a study flow diagram can be found in Figure 1.

Qualitative data generated on acceptability are reported in a separate article (Hemrage et al., 2024c). Through a theory-informed approach, the qualitative component of the trial demonstrated positive views toward CM (intervention coherence, ethicality, self-efficacy, perceived effectiveness and affective attitudes), with its core components matching participants' preferences and needs. CM was considered an appropriate and person-centred intervention for this clinical population.

The pilot trial has been peer-reviewed following the regulatory and governance requirements set by King's College London and King's College Hospital Research and Innovation (R&I). A favorable opinion was granted by Camden and Kings Cross NHS Research Ethics Committee (reference 22/LO/0744). The trial has been registered on the Open Science Framework (identifier DOI 10.17605/OSF.IO/AVKYJ) and ClinicalTrials.gov database (identifier: NCT06183710) (ClinicalTrials.gov, 2023; Hemrage, 2022). As part of the protocol, a record of serious adverse events (SAEs) was kept and reported to the sponsor. The CONSORT checklist for the trial is provided as Appendix S1.

The trial was conducted under the principles of the Declaration of Helsinki, Good Clinical Research Practice (GCP) and General Data Protection Regulation (GDPR) and additional relevant regulatory requirements (UK Government, 2018; World Health Organization, 2005; World Medical Association, 2013). Written informed consent was sought before any data was collected. To compensate participants for their time, a £15 honorarium was offered both at baseline and post-ILC data collection points.

Study population, sampling and randomization

Consecutive sampling was used to recruit 30 adult participants presenting a comorbid diagnosis of AUD and ARLD from an inpatient setting, admitted with an acute liver episode and not referred for liver transplantation at the time of recruitment. A sample size of 30 participants was decided upon after reviewing available guidance for feasibility and pilot studies (Eldridge et al., 2016). Recruitment was set for an initial period of 12 weeks. Due to suboptimal recruitment rates, additional sites were added (the emergency department at KCH and Princess Royal University Hospital, King's College Hospital NHS Foundation Trust), and the recruitment period was extended until the target sample size was achieved (35 additional weeks).

Cognitive impairment, as evidenced by the inability to provide voluntary informed consent, was an exclusion criterion. Patients who were dependent upon substances other than alcohol, tobacco, or cannabis were not eligible. Pregnancy also precluded participation in the pilot trial.

Participants were randomized following a string randomization approach (1:1 ratio) using a Stata code developed by the study statistician (StataCorp, 2021). The allocation sequence was printed on separate pages and sealed into individual, numbered (1–30) envelopes. The allocation was concealed until the entry of a new participant. After enrollment, the researcher (SH) opened an envelope corresponding to the participant's assigned number. After this stage, double blinding was not possible due to the design of the study. Given the nature of the CM intervention, the participants, the ILC consultants (NJK, NS), and the researcher (SH) conducting the assessments were not blinded. The remaining members of the research team (SP, PD, CD) were blinded to treatment allocation.

Interventions

All participants were offered four ILC appointments at the LOPD over 12 weeks. These involved the provision of hepatology and addictions psychiatry within the same facility. Multidisciplinary treatment and co-location of healthcare providers have been noted as optimal in managing ARLD (López-Pelayo et al., 2019). Hepatology care entailed the delivery of regular liver health education, management of decompensation-related symptoms, referral to dietetics services, and elective paracentesis, provided by the ILC hepatology consultant. Psychiatric care comprised motivational interviewing, relapse prevention techniques, referral to support services, and prescription of anti-craving pharmacotherapy. These were provided by the ILC addictions psychiatry consultant. The consultants are part of the ACT and met daily for clinical supervision to discuss clinical cases and ensure treatment fidelity.

The employed CM intervention has been described in detail in a previous publication and is summarized in Table 1 (Hemrage et al., 2024c). The CM intervention aimed to promote treatment attendance and was delivered by the ILC consultants during appointments in the clinic. The incentives consisted of multi-outlet vouchers exchangeable for material goods. Participants in the intervention group could receive a maximum of £120 if attending all four ILC visits, with the total amount corresponding to the number of completed sessions. For each visit, a £15 was given, with gradually increasing bonuses (£10, £20, £30) for attending at least two ILC visits (the benchmark for engagement defined in the protocol). Engagement with treatment was measured by confirming attendance at the four sessions in the participants' respective electronic health records. Upon completion of the study, the incentives were discontinued, but participants were offered further ILC on a case-by-case basis.

The CM intervention was informed by a previous multicenter, three-arm, parallel-group, randomized controlled trial assessing the effectiveness of adjunctive medication management and CM to enhance adherence to acamprosate for alcohol dependence (Donoghue et al., 2023). The adaptation of the CM intervention was informed through the available literature, consultation with healthcare providers, and members of the public from the earliest stages of research planning.

Data collection and outcomes

Statistical analysis

SPSS Statistics version 29.0 was used to perform statistical analyses (IBM Corp, 2023).

Baseline sample characteristics were analyzed through descriptive statistics, and group differences were calculated (independent samples t-test for continuous outcomes, Chi-Square test for categorical outcomes).

Primary outcome data on feasibility (recruitment, randomization, study retention, completeness of the data, protocol fidelity) were reported through summary statistics, and additional analyses by arm were conducted where appropriate.

Given the nature of this pilot trial, hypothesis testing was not performed; per-protocol analyses of secondary outcome data were carried out for exploratory purposes. Secondary outcome data (engagement, alcohol intake and liver outcomes) were reported through summary statistics, and additional analyses by arm were conducted where appropriate.

Baseline participant characteristics

The demographic profile, clinical history, and health services utilization of the sample are presented in Table 3. The mean age of the cohort was 48.47 (SD 10.08); 30% (n = 9) self-identified as female and 70% (n = 29) as male. Regarding ethnicity, 13.30% (n = 4) identified as Asian, Asian British, or Asian Welsh, 6.70% (n = 2) as Black, Black British, Black Welsh, Caribbean, or African, and 80% (n = 24) as White, White British, or White Welsh. The demographic profile of participants enrolled in the pilot trial generally matches that of previous similar research, although the prevalence of participants of Asian, Asian British, or Asian Welsh ethnicity was higher (Mehta et al., 2024; Mellinger et al., 2023).

Most participants (80%, n = 24) were known to the ACT from previous admissions (an average of 3 admissions pre-randomization), but only a small proportion had engaged with community treatment services (36.70%, n = 11), peer support initiatives (23.30%, n = 7) or were supported by an assertive outreach (comprehensive, intensive delivery of community mental health care for service users who have not engaged effectively with treatment in the past) team (26.70%, n = 8). The median length of inpatient stay was 7 days (range 2–127).

In the month preceding admission, participants had an alcohol intake of 179.20 units per week (SD 70.22) and the mean percentage of days abstinent was 1.42% (SD 4.33). The mean APQ score was 16.16 (SD 7.52) and SADQ was 32.58 (SD 16.65), suggestive of severe alcohol dependence.

Over half of the participants (53.30%, n = 16) presented to the hospital with decompensated liver disease, 40% (n = 12) with cirrhosis, and 6.70% (n = 2) with non-severe alcohol-related hepatitis. Upon admission, 33.30% (n = 10) presented with ascites and 16.70% (n = 5) with jaundice. Hepatic encephalopathy was present in 16.70% (n = 5), presented with upper gastrointestinal bleeding in 13.30% (n = 4), and 30% (n = 9) with variceal hemorrhage, which are symptoms suggestive of portal hypertension. No participants had an active infection of hepatitis C at the time of recruitment. At baseline, MELD 3.0 and CPG scores were calculated for participants with cirrhosis and decompensated liver disease (n = 28) and had a median of 14.50 (range 6–30) and 8 (range 5–11) respectively. Additional data on liver biomarkers (albumin, AST, bilirubin, creatinine, GGT, INR, sodium) can be found in Table 3.

Feasibility outcomes

The overall feasibility of the trial was evaluated as amber (when there are remediable aspects that need to be addressed before proceeding) in line with Avery's traffic light system.

Recruitment and study-retention post-randomization

Between January and December 2023, 116 patients (Figure 2) were identified for the pilot trial, and 41 were assessed for eligibility (eligibility rate 35.40%). Thirty patients provided informed consent (recruitment rate 73.20%); 15 were randomized to ILC only and 15 to ILC + CM. Following randomization, eight participants died in hospital before discharge (ILC n = 3, 20%; ILC + CM n = 5, 33.30%), highlighting the severity of this patient population; a further 11 did not engage. Overall, this accounted for a total loss of 19 participants (ILC n = 11, 73.30%; ILC + CM n = 8, 53.30%). One participant (n = 1, 6.60%) allocated to the control group withdrew from the study before receiving ILC. A total of 11 participants (n = 11; retention rate 36.70%) engaged with the trial. Of these, four (26.70%) received ILC only and seven (46.60%) received ILC + CM. Despite the small sample size, no differences in retention were observed between groups (χ²(1) = 1.29, p = 0.25), although these values should be interpreted with caution. No severe adverse events related to the trial protocol occurred. In line with Avery's traffic light system, recruitment and study retention were considered amber (when there are remediable aspects that need to be addressed before proceeding). Reasons for suboptimal engagement with the trial and treatment have been reported in two separate qualitative investigations (Hemrage et al., 2024a, 2024b). In these investigations, a combination of interplay between individual, organizational, and structural factors, exacerbated by the COVID-19 pandemic and socioeconomic landscape, was found to condition access to clinical research (Hemrage et al., 2024a). Additionally, the authors also found gaps in the care continuum, complex health needs, accessibility, and stigma to hinder treatment engagement among this clinical population (Hemrage et al., 2024b). For patients with comorbid AUD and ARLD, these barriers further contribute to a gap in effective evidence-based treatment, exacerbating the health inequalities experienced by this clinical population.

Exploratory outcomes

Alcohol use

Across both groups, the overall weekly alcohol intake had a 76% reduction trend to 38.69 units (SD 105.71, range 0–350) post-ILC. Specifically, participants in the control group were abstinent, and participants in the CM group reduced their intake by 64.30%, with an intake of 60.80 units (SD 130.60). Figure 3 provides a graphical representation of the change in weekly alcohol intake.

At the longest follow-up, an 85.90% (23.95 units, SD 50.18) reduction trend on the weekly was observed compared to baseline. Participants in the control group sustained abstinence, and those receiving CM showed a 78.80% decreasing trend compared to baseline (35.93 units, SD 59.27). A corresponding trending increase in the percentage of days abstinent was observed for both groups, which was higher in the control group (100%) compared to CM (75.48% post-ILC, 71.40% 12-weeks post-ILC).

Across both groups, trending reductions in APQ and SADQ scores were observed post-ILC and at the longest follow-up. Within the APQ, improvements were mostly noted in the social, physical, and psychological domains, suggesting that engagement with ILC may support the management of AUD and ARLD. At the longest follow-up, a trend for improved SADQ scores was noted across both groups (ILC = 0.33, SD 0.57; ILC + CM = 11, SD 10.78). Given the increasing trend for abstinence observed in the control arm, the reduction in SADQ scores was less visible in the CM arm. Summary measures of alcohol outcomes are presented in Table 4.

TABLE 4. Alcohol use outcomes for participants completing the pilot trial (n = 11).

Alcohol use outcomes	Baseline (n = 11)				Post-ILC (n = 11)				12 Weeks post-ILC (n = 9)
	Overall (n = 11)	Control (n = 4)	CM (n = 7)	Group difference	Overall (n = 11)	Control (n = 4)	CM (n = 7)	Group difference	Overall (n = 9)	Control (n = 3)	CM (n = 6)	Group difference
TLFB weekly units, mean (SD, range); median	161.36 (74.36, 80–350); 140	145 (53.22, 80–210); 145	170.71 (86.71, 90–350); 140	p = 0.55	38.69 (105.71, 0–350); 0	0 (0); 0	60.80 (130.60, 0–350); 0	p = 0.38	23.95 (50.18,0–75.60); 0	0 (0); 0	35.93 (59.2, 0–75.60); 0	p = 0.34
TLFB units per drinking day, mean (SD, range); median	25.72 (11.38, 11.40–50); 20	25.98 (11.13, 6–50); 25	27.17 (9.41, 12.85–50); 30	p = 0.83	5.52 (15.10, 0–50); 0	0 (0); 0	8.68 (18.65, 0–50); 0	p = 0.26	3.42 (7.16, 0–20)	0 (0); 0	5.13 (8.46, 0–20)	p = 0.19
Percentage days abstinent, % mean (SD, range); median	2.58 (5.74, 0–14.20); 0	1.89 (4.99, 0–14.20); 0	0.94 (3.66, 0–14.20); 0	p = 0.72	84.40 (35.28, 0–100); 100	100 (0, 100); 100	75.48 (42.66, 0–100); 0	p = 0.17	80.93 (38.49, 0–100); 100	100 (0, 100); 100	71.40 (45.20, 0–100); 100	p = 0.18
Return to any drinking, % (n)	—	—	—	—	18.20 (2)	0 (0)	28.60 (2)	χ2(1) = 1.397, p = 0.23	22.20 (2)	0 (0)	33.30 (2)	χ2(1) = 1.286, p = 0.25
Abstinence, % (n)	—	—	—	—	81.80 (9)	100 (4)	71.40 (5)	χ2(1) = 1.397, p = 0.23	77.80 (7)	100 (0)	66.70 (4)	χ2(1) = 1.286, p = 0.25
APQ score, mean (SD, range); median	18.81 (8.54,4–30); 18	13 (7.39, 4–22); 13	22.14 (7.66, 10–30); 24	p = 0.08	5.81 (6.16,0–21); 4	4.25 (1.89,3–7); 3.50	6.71 (7.67, 0–21); 5	p = 0.55	4.55 (5.83, 0–18); 1	3.66 (4.61, 1–9); 1	5 (6.72, 0–18); 3	p = 0.77
SADQ score, mean (SD, range); median	19.72 (14.73, 0–50); 18	10 (9.55,0–21); 8.50	25.28 (14.73,8–50); 19	p = 0.09	9.09 (12.07,0–39); 6	5.25 (5.12,0–12); 4.50	11.28 (14.64,0–39); 6	p = 0.45	7.44 (10.06,0–26); 1	0.33 (0.57, 0–1); 0	11 (10.78,0–26); 10	p = 0.14

• Abbreviations: APQ, Alcohol Problems Questionnaire; CM, contingency management; ILC, integrated liver care; n, number; SADQ, Severity of Alcohol Dependence Questionnaire; SD, standard deviation; TLFB, timeline followback.

Liver function and biomarkers

Compared to baseline, the proportion of participants presenting with signs of ascites improved over time, observed in three post-ILC (ILC n = 1, ILC + CM n = 2) and in two participants (ILC n = 1, ILC + CM n = 1) at 12 weeks post-ILC. No participants presented with hepatic encephalopathy following discharge.

Among those completing the trial (n = 11), a trend for improvements on the MELD 3.0 and CPG scores was observed over time (Figures 4 and 5). The median MELD 3.0 decreased from 18 (range 3–6) to 9.50 (range 7–14) at the longest follow-up. The median CPG score improved from 7 (range 5–11) to 5 (range 5–7), representing a shift from Class B to Class A.

Summary measures of additional data on liver biomarkers can be found in Table 5.

TABLE 5. Liver health outcomes for participants completing the pilot trial (n = 11).

Liver function and biomarkers	Baseline (n = 11)				Post-ILC (n = 11)				12 Weeks post-ILC (n = 9)
	Overall (n = 11)	Control (n = 4)	CM (n = 7)	Group difference	Overall (n = 11)	Control (n = 4)	CM (n = 7)	Group difference	Overall (n = 9)	Control (n = 3)a	CM (n = 6)	Group difference
Albumin (g/L), mean (SD, range); median	35.63 (8.02, 27–51); 34	37 (9.52, 27–47); 37	34.85 (7.75, 27–47); 34	p = 0.69	40.54 (7.52, 26–53); 41	41 (2.16, 38–43); 41.5	40.28 (9.58, 26–53); 40	p = 0.85	45.62 (5.06, 37–53); 45	47 (5.65, 43–51); 47	45.16 (5.34, 37–53); 45	p = 0.69
AST (U/L), mean (SD, range); median	114.72 (99.06, 4–339); 81	73.5 (52.44, 4–128); 81	138.28 (114.89, 19–339); 81	p = 0.32	57.27 (44.52, 22–158); 38	36.25 (11.84, 22–46); 38.50	69.28 (52.64, 35–158); 38	p = 0.25	92.87 (74.37, 34–211); 59	53 (2.82, 51–55); 53	106.16 (83.02, 34–211); 72	p = 0.42
Bilirubin (μmol/L), mean (SD, range); median	114.36 (106.717,6–314); 64	129 (134.95, 6–314); 98	106 (98.23, 6–282); 64	p = 0.75	47.81 (78.55, 5–276); 64	24.75 (13.57, 5–35); 29.50	61 (98.16, 8–276); 20	p = 0.49	16.87 (11.11, 6–34); 26	14.50 (12.02, 6–23); 14.50	17.66 (11.877, 6–34); 12.50	p = 0.75
Creatinine, (μmol/L), mean (SD, range); median	58.81 (15.35, 38–91); 53	63.25 (12.60, 45–73); 67.50	56.28 (17.11, 38–91); 50	p = 0.49	66.54 (22.25, 36–106); 67	76 (20.21, 62–106); 68	61.14 (22.95, 35–98); 56	p = 0.31	79.25 (21.15, 52–103); 78	103 (0); 103	71.33 (18.05, 52–95); 63	p = 0.05
GGT (U/L), mean (SD, range); median	872.45 (546.29, 114–1804); 536	519.25 (426, 114–1108); 553	1074.28 (523.32, 381–1804); 436	p = 0.10	314 (219.30, 56–791); 272	193.50 (162.87, 56–418); 150	382.85 (227.48, 152–791); 277	p = 0.18	491.62 (241.418, 172–969); 416	502.50 (232.63, 338–667); 502.50	488 (265.91, 72–969); 416	p = 0.94
INR, mean (SD, range); median	1.22 (0.14, 1–1.4); 1.20	1.25 (0.17, 1.1–1.4); 1.25	1.21 (0.14, 1–1.40); 1.20	p = 0.74	1.16 (0.17, 1–1.5); 1.10	1.12 (0.09, 1–1.2); 1.15	1.03 (0.08, 1–1.50); 1.10	p = 0.60	1.05 (0.75, 1–1.20); 1	1.10 (0); 1.10	1.03 (0.08, 1–1.20); 1.10	p = 0.31
Sodium (mmol/L), mean (SD, range); median	135.81 (6.03, 123–142); 138	136.75 (7.27, 126–142); 139.50	135.28 (5.86, 123–140); 137	p = 0.36	137.81 (4.21, 130–145); 139	136.50 (4.50, 130–140); 138	138.58 (4.19, 132–145); 139	p = 0.23	136.87 (5.40, 130–143); 137	135.50 (6.36, 131–140); 135.50	137.33 (5.64, 130–143); 138	p = 0.71
Ascites, % (n)	63.60 (7)	75 (3)	57.10 (4)	χ2(1) =0.35, p = 0.50	27.30 (3)	25 (1)	28.60 (2)	χ2(1) = 0.01, p = 0.89	22.20 (2)	33 (1)	16.70 (1)	χ2(1) =0.32, p = 0.57
Hepatic encephalopathy, % (n)	36.40 (4)	25 (1)	42.90 (3)	χ2(1) = 0.35, p = 0.55	—	—	—	—	—	—	—	—
MELD 3.0 score, mean (SD, range); medianb	17.72 (7.21, 3–6); 18	17.50 (7.5, 9–26); 18	17.85 (7.64, 6–30); 17.50	p = 0.94	11.54 (5.27, 6–24); 11	11 (3.55, 8–16); 10	11.85 (6.30, 6–24); 11	p = 0.81	10.12 (2.16, 7–14); 9.50	11.50 (3.52, 9–15); 11.50	9.66 (1.75, 7–12); 9.50	p = 0.33
CPG score, mean (SD, range); medianb	8.27 (2.10, 5–11); 7	8.75 (2.62, 5–11); 8.50	8 (1.91, 6–10); 7	p = 0.59	6 (1.09, 5–8); 6	5.50 (1, 5–7); 5	6.28 (1.11, 5–8); 6	p = 0.27	5.37 (0.74, 5–7); 5	5.5 (0.70, 5–6); 5.50	5.33 (0.81, 5–7); 5	p = 0.80

• Note: Independent samples t-test calculated for continuous outcomes and Chi-Square tests for categorical outcomes. Significance level p < 0.05.
• Abbreviations: AST, aspartate aminotransferase; CM, contingency management; CPG score, Child-Pugh score; GGT, gamma-glutamyl transferase; INR, international normalized ratio; MELD score, Model for End-Stage Liver Disease 3.0 score; n, number; SD, standard deviation.
• ^a Data missing for one participant (n = 1).
• ^b Calculated for participants presenting cirrhosis and decompensated liver disease.

Strengths and limitations

Embedding the experimental protocol in a clinical setting enabled an organic understanding of real-world implementation and delivery of voucher-based CM, which is a strength of this research. Particularly, this allowed the identification of factors that shape research design and those inherent to care delivery in an under-served clinical population and can inform future trials aimed at developing novel interventions for this patient group.

In this pilot trial, CM was delivered as an adjunctive to ILC. No previous research has attempted to deliver CM to promote engagement with multidisciplinary ARLD care (Hemrage et al., 2023; Khan et al., 2016). ILC has been increasingly recommended to bridge the treatment gap in comorbid AUD and ARLD (DiMartini et al., 2022; Haque & Leggio, 2024; Singal et al., 2024). The integration of hepatology and addiction psychiatry in study design has also been recommended to generate higher quality research in this area (Lee et al., 2024).

From a healthcare provider perspective, pragmatic barriers to the wider uptake and implementation of CM in clinical settings include ensuring the supply of incentives and adapting standard clinical workflow to deliver CM (Sinclair et al., 2011). The administrative aspects related to CM delivery were handled by the research team, which circumvented these barriers to implementation. Attendance-based CM is also logistically simpler and less resource-intensive compared to abstinence-based CM. For the management of clinically heterogeneous populations, attendance-based CM can serve as an adjunct to treatment matching and person-centered care (Hell et al., 2021; Witkiewitz et al., 2010). This can enable a more tailored, individual approach to treatment compared to abstinence-focused interventions.

This research also had some limitations. The first limitation relates to the demographic profile of the cohort and its representativeness. Most participants self-identified as male and were of White, White British, or White Welsh ethnicity. This may impact the generalizability of the findings. Several lines of evidence have highlighted gender and ethnic differences in ARLD prevalence, severity, treatment access, and mortality (Levy et al., 2015; McElroy et al., 2020; Mellinger et al., 2019; Niu et al., 2023; Probst et al., 2015). Addressing these variations is key for the development of person-centered approaches tailored to subgroups that are more susceptible to alcohol-related harm. Additionally, data on socioeconomic status was not collected; this was aimed at minimizing the research burden upon participants. Socioeconomic status is a known modifier of alcohol consumption, harm, and ARLD incidence (Askgaard et al., 2024; Probst et al., 2020; Roblero et al., 2021). Therefore, it could be relevant to consider measurements of socioeconomic status to inform the development of approaches mitigating health inequalities across sociodemographic groups within the context of comorbid AUD and ARLD.

As a single-centre study, the findings may not extend to all service users with comorbid AUD and ARLD across the UK or internationally. This is particularly relevant given the geographical variation between areas of high prevalence of ARLD compared to research hotspots in the UK (Tavabie et al., 2024). Participants were also recruited from an inpatient setting. This may prevent the understanding of treatment experience and health-promoting behavior among service users who are not formally admitted to the hospital and those who do not voluntarily engage with services. In addition, a longer timeframe would allow for a better understanding of CM's long-term impact on health-promoting behavior. Many service users return to drinking within 6 months of admission, and multiple attempts at recovery might be needed before meaningful treatment engagement and improvements in alcohol intake and liver function outcomes are noticeable (Green et al., 2023). In addition, exploratory data is reflective only of those participants engaging with the trial, so the findings need to be interpreted with caution. As a result, multi-centre studies with longer follow-up times are needed to further ascertain the clinical significance of CM to improve patient outcomes in comorbid AUD and ARLD.