DAILY – A personalized circadian Zeitgeber therapy as an adjunctive treatment for alcohol use disorder patients: results of a pilot trial

Alcohol use disorder (AUD) is a major contributor to global morbidity and mortality. Evidence suggests a bidirectional relationship between circadian rhythms and alcohol consumption: disrupted circadian rhythms increase risk for AUD, and AUD is associated with circadian dysregulation (sleep-wake disturbances, altered molecular and neuroendocrine rhythms). The suprachiasmatic nucleus (SCN) coordinates circadian timing, and consistent environmental time cues (Zeitgebers) such as light and meals are necessary for stable synchronization. Irregular time cues lead to desynchronization and potential behavioral and physiological deficits relevant to addiction. DAILY (Depression Alcohol Illness Therapy) seeks to identify personal circadian characteristics and strengthen rhythms by stabilizing the timing of key daily activities (meals, bed and wake times) using individualized structure plans during withdrawal therapy. Primary hypotheses: (1) The DAILY intervention reduces temporal variability of daily activities and reduces alcohol craving, alcohol consumption, and relapse. (2) Elevated temporal variability increases craving and relapse risk; therefore, the intervention’s effects on variability and outcomes are interrelated. Secondary hypotheses: (3) DAILY improves accompanying symptoms of AUD (sleep quality, depressive symptoms, self-efficacy, alcohol-related blood markers). (4) DAILY improves compliance and study retention. (5) DAILY reduces temporal variability without shifting chronotype.

Prior research links circadian disruption to increased risk of AUD, including associations with late chronotype, shift work, and polymorphisms in core clock genes (CLOCK, PER1, PER2, PER3). AUD patients frequently display disturbed sleep timing and quality, altered molecular rhythms, and dysregulated neuroendocrine rhythms, including during withdrawal. The SCN synchronizes peripheral clocks, and consistent Zeitgebers (notably light and food) are required for stable entrainment. Desynchronization impairs inter-regional brain communication and may contribute to addictive behaviors. These findings motivate circadian-based interventions aimed at stabilizing daily routines to support abstinence and reduce relapse risk in AUD.

Design: Six-week (occasionally up to seven weeks for scheduling) monocentric, randomized, controlled, single-blinded, parallel-group pilot trial conducted at the Clinic for Psychiatry and Psychotherapy, University Hospital of Munich. Participants were randomized by computer-generated list to intervention group (IG) or sham control group (CG). Participants: 85 screened; 54 recruited; 13 excluded post hoc for meeting exclusion criteria; 41 randomized (CG n=20; IG n=21). Inclusion: ICD-10 F10.2 alcohol dependence, age 15–75; other substance dependence (except nicotine) required ≥12 months abstinent. MDD diagnosis allowed (excluding psychosis/suicidality). Key exclusions: inability to consent, pregnancy, shift work, medical barriers to independent intervention adherence, blindness, psychiatric diagnoses other than AUD/MDD, use of benzodiazepines, agomelatine, or prescribed cannabinoids. Procedures: Baseline visit with informed consent, questionnaires, and blood draw for all participants. IG received psychoeducation on circadian biology and AUD, then completed a 1-week test diary. Personalized daily structure plans were co-created from diary time clusters (meals, bed and wake times), considering personal constraints (e.g., work). IG were asked to adhere to the plan for 4–5 weeks and continue daily diaries; plans were adjusted if care setting changed. Weekly contact (phone or in-person) to support adherence. CG attended matched-time sessions on advertising and marketing of legal substances in addiction; completed diaries but received no feedback or structure guidance; contacted every 1–2 weeks for compliance. Data collection: Paper diaries recorded daily times for meals (breakfast, lunch, dinner), going to bed, falling asleep, waking, getting up, and sleep quality. The last 22 participants (CG:11, IG:11) reported daily peak alcohol craving; a secure messenger app (Wire) was optionally used late in the study (n=4) for real-time entries and reminders. Questionnaires: alcohol use (AUDIT, EuropASI, TLFB), depression (HAMD, IDS-SR), self-efficacy (SWE), sleep quality (PSQI), chronotype (MCTQ). Blood markers: GGT, De Ritis ratio (GOT/GPT), CDT, MCV (aP, Hb, B12 not assessed). Statistics: Significance threshold p<0.05. Baseline differences between completers and drop-outs: Kruskal–Wallis with Dunn’s multiple comparisons. Group differences in categorical variables (drop-outs, relapse, sex, recruitment site): chi-square. Two-group comparisons: unpaired t-tests. Relationships to craving: Pearson correlation of subject-level means; linear mixed models (subject and time random effects; Satterthwaite p-values, normality checked by QQ plot); logistic regression modeling relapse using decreasing pre-relapse time windows (days −42 to −1). Repeated measures: mixed-effects models and Bonferroni post hocs for group×time effects (baseline vs final). Two-way ANOVA for sex×group effects. Blinding: Single-blinded (participants were assigned to IG or CG; outcome assessors utilized standardized procedures; interventions differed in content but were matched in contact/time). Duration and follow-up: Target 6 weeks; some extended to 7 weeks due to scheduling; final appointment included repeat questionnaires and blood draw; remote completion was permitted when necessary.

- Variability reduction: IG exhibited significantly lower variability in eating times and bedtimes compared to CG; overall variability score also lower (unpaired t-tests, p<0.05 to p<0.01; n CG/IG: 12/17). Increased regularity of meals was mainly driven by lunch (and trend for dinner); sleep regularity did not hinge on specific sleep timing components. No significant sex effects; both sexes appeared to benefit similarly. - Relapse outcomes: Among participants with reliable relapse data, IG: 2/17 relapsed (<12%); CG: 10/16 relapsed (62.5%); chi-square p<0.01. Total days with alcohol consumption: IG 3 days vs CG 52 days. IG consumption occurred later (days ~25 and ~39) whereas CG consumption occurred throughout the study period. - Abstinence vs variability: Abstinent participants had lower average variability (meals, bedtimes, total) than relapsed participants. - Craving associations: No group difference in mean craving (subset n CG/IG: 9/11). Higher average variability in getting-up time correlated with higher average craving (Pearson p=0.005). Day-level linear mixed model: same-day drift in getting-up time predicted craving (p=0.0432). Other parameters were not significant predictors in the mixed model. - Pre-relapse dynamics and prediction: Variability of lunch and dinner times increased progressively in the ~14 days before relapse. Craving rose sharply before relapse, exceeding prior ranges. Logistic regression showed change in craving significantly predicted relapse up to ~12 days in advance; lunchtime variability approached significance over wide windows, whereas breakfast, getting-up, going-to-bed variability and sleep quality were less discriminative. - Secondary outcomes: AUDIT improved in both groups, more in IG (between-group difference at study end). HAMD improved only in IG. IDS-SR improved in both, more in IG. SWE improved more in IG. PSQI improved in IG only. Blood markers: GGT improved significantly only in IG; De Ritis ratio, MCV, and CDT showed no significant changes in either group. Chronotype (MCTQ) did not change. - Retention: Drop-outs were higher in CG (12/20; 60%) than IG (4/21; ~19%); chi-square p=0.0072; IG drop-outs occurred early, before or just as intervention began. Sex and recruitment site did not significantly influence relapse.

The DAILY intervention, which stabilizes timing of key daily activities, was associated with markedly fewer relapses, fewer consumption days, and later occurrence of any consumption during early withdrawal when relapse risk is high. The mechanism appears linked to reduced temporal variability of daily routines: abstinent participants exhibited lower variability, and day-to-day drift in getting-up time related to same-day craving. Temporal variability of lunch and dinner increased in the days preceding relapse, while craving surged and significantly predicted relapse up to about 12 days in advance. Combining craving monitoring with variability metrics (notably lunch/dinner timing) may aid relapse prediction and timely therapeutic response. Beyond relapse prevention, DAILY enhanced depressive symptoms, perceived self-efficacy, and sleep quality without shifting chronotype, aligning with the aim to strengthen, not retime, intrinsic rhythms. GGT improvement in IG supports potential reductions in acute liver stress, though causality and specificity remain to be clarified. The intervention also appeared to enhance engagement and retention. These findings support continued evaluation of DAILY and integration of objective circadian measures to substantiate endogenous rhythm improvements.

This pilot randomized trial suggests that a personalized circadian Zeitgeber therapy (DAILY) adjunctive to standard AUD treatment reduces temporal variability of routines, lowers relapse rates and alcohol consumption days, improves mood, sleep quality, and self-efficacy, and enhances retention, without altering chronotype. Future research should: (1) incorporate objective circadian assessments (e.g., 24-h profiles of hormones/temperature, saliva/blood sampling, actimetry) to verify endogenous rhythm changes; (2) implement digital data capture (apps) to improve completeness and timeliness; (3) focus on outpatient settings to maximize autonomy and between-group contrasts in structure; (4) extend intervention duration to examine long-term abstinence and relapse timing; (5) increase sample size to enable individualized modeling of which activities’ variability most strongly predicts craving/relapse and to explore chronotype-specific tailoring (e.g., addressing risks associated with eveningness).

- No direct measures of endogenous circadian phase or amplitude; only behavioral timing was assessed. - Self-reported diary data (paper-based for most) prone to underreporting and compliance issues; limited use of app-based reporting (n=4). - Heterogeneous care settings (inpatient and day-care/outpatient) with varying external structure may have reduced between-group variability differences and generalizability. - Short intervention duration (6–7 weeks) limits assessment of longer-term relapse dynamics. - Small sample size reduces statistical power and limits subgroup/individualized analyses; some models had unbalanced designs (mixed-model p-values to be interpreted cautiously). - Missing end-of-study blood data for some CG participants (particularly with high baseline GGT), and potential non-alcohol-related liver factors were not assessed. - Limited craving data (subset of last 22 participants) restricts generalizability of craving-related analyses.