Protocol for Outcome Evaluation of Ayahuasca-Assisted Addiction Treatment: The Case of Takiwasi Center

Rush and colleagues frame their work within a resurgence of interest in psychedelic and traditional medicines as potential treatments for substance use and other mental disorders, driven by limitations in mainstream addiction medicine and gaps in global mental health coverage. The introduction summarises historical, cultural and clinical interest in ayahuasca — a two-plant Amazonian brew containing DMT and harmala alkaloids — and notes observational and qualitative evidence suggesting reductions in problematic substance use, alongside plausible neuropharmacological and psychological mechanisms. The authors emphasise the importance of non-pharmacological factors, commonly referred to as set (individual mindset, intentions and preparation) and setting (social, ritual and therapeutic context), which may interact with neurobiology to shape outcomes. The paper describes the Ayahuasca Treatment Outcome Project (ATOP) and presents the study protocol for its first site, the Takiwasi Center in Peru. The stated aim is to evaluate treatment outcomes and explore therapeutic mechanisms of an ayahuasca-assisted, integrative residential model for addiction rehabilitation using a mixed-methods, longitudinal cohort design with extended follow-up. The authors position ATOP as a multisite umbrella with common core measures and methods, and present Takiwasi as a naturalistic case-study setting in which to investigate how elements of set and setting relate to a range of recovery-oriented outcomes.

The ATOP-Takiwasi evaluation uses a prospective, observational mixed-methods cohort design. Quantitative assessments occur at baseline (program intake), discharge and at 3, 6, 12, 18 and 24 months post-discharge. Baseline diagnostic assessment employs the MINI International Neuropsychiatric Interview to identify alcohol/drug abuse or dependence and co-occurring mental disorders for eligibility and subgroup description. Core quantitative instruments include the Addiction Severity Index (ASI-5), selected subscales from the Global Appraisal of Individual Needs (GAIN-I) adapted to the local drug nomenclature, the Beck Depression and Anxiety Inventories (BDI, BAI), the WHOQOL-SRPB for spirituality-related quality of life, and the Treatment Entry Questionnaire (TEQ-9) to capture treatment motivation; most instruments are validated in Spanish or translated and back-translated where necessary. The qualitative component comprises semi-structured interviews with patients at the same time points as the quantitative measures (entry, during treatment, discharge, and follow-ups), together with interviews of managers, staff, therapists and healers, and an ethnographic-style contextual description of the centre. Interview topics probe personal histories of substance use and treatment, motivations and expectations, subjective experiences of ayahuasca and related ritual elements (e.g., icaros, purging, dietas), perceived importance of integration activities, and perceptions of recovery. A separate sub-project was initiated to investigate reasons for premature program exit. Participant recruitment and exposure measurement are integrated into the protocol. Core inclusion criteria limit the study to Latin American sites integrating Amazonian plant medicine with western psychosocial care and to substance-related problems (including tobacco but excluding behavioural/process addictions); exclusions include pregnancy and concurrent MAOI antidepressants or other contraindicated medications. Treatment exposure is measured both categorically (completed treatment; left voluntarily; suspended) and continuously via a program participation index extracted from patient files (number of ayahuasca ceremonies, days of dieta, integration sessions, etc.). Follow-up contacts are coordinated by a research coordinator independent of program staff. Ethical approval was obtained from Comité de Ética PRISMA, Lima. For analysis, the quantitative team prepares a unified person-level data file, conducts descriptive analyses, then reduces measures through a priori conceptual modelling. Directed Acyclic Graphs (DAGs) will be constructed to guide causal thinking and multivariate regression modelling. Primary longitudinal analyses will use generalized linear mixed effects models with random intercepts (and slopes when needed) to account for within-participant correlation and time-varying covariates. The qualitative team uses grounded theory methods, coding interview data collaboratively (via Dedoose) to allow emergent themes to inform iterative theory-building. Integration of findings involves stratifying participants by consensus-rated overall outcome into three groups (very poor/poor; moderate; very good/excellent) and cross-referencing qualitative themes with quantitative predictors and outcomes.

The extracted text reports study recruitment status and early cohort metrics rather than final clinical outcomes. As of 17 December 2020, 128 individuals had completed intake at Takiwasi and were potentially eligible for ATOP. Forty-two of these (32.8%) left the programme before participating in any ayahuasca ceremony and therefore were not eligible for inclusion in the study cohort. Baseline interviews were completed with 86 participants. Of the 60 participants scheduled for one-year follow-up, 48 had completed both the one-year qualitative and quantitative interviews, representing an 80% one-year completion rate. Follow-up sample sizes at later time points were reported as ongoing: 18-month data collection had 30 participants and 24-month had 25 participants at the time of reporting, though the text cautions these figures reflect scheduling and are not comparable to final retention figures. For the 12-month sample of 48 clients, subgroup counts were 30 completers (60%), 14 who left voluntarily (29%), and 4 suspended (8%); weeks of participation are to be calculated for each patient for use as a moderating variable. The team had conducted 17 semi-structured interviews with managers, staff, therapists and healers by the same date. The paper notes historically high non-completion rates for therapeutic community models and cites prior Takiwasi data indicating a non-completion rate of 51.8% in earlier studies. Beyond these recruitment and retention figures, the extracted text does not provide quantitative effect estimates, symptom-change statistics, or other primary outcome results from the cohort.

Rush and colleagues interpret the ATOP-Takiwasi protocol as a comprehensive, complexity-aware approach to studying ayahuasca-assisted residential treatment for substance use disorders, with particular attention to the interplay of set and setting. They argue that the mixed-methods, longitudinal cohort design — combining validated quantitative instruments, detailed measures of programme participation, ethnographic description and grounded qualitative inquiry — is well suited to identify which contextual and experiential ingredients are most closely associated with longer-term outcomes. The authors highlight strengths of the protocol including repeated measures up to 24 months, a broad outcome domain beyond abstinence (e.g., mental health, quality of life, spirituality), and an analysis plan that uses DAGs and mixed-effects models to limit inappropriate significance testing and to address confounding. The authors also position Takiwasi as an informative naturalistic site given its long history, structured nine-month residential programme, and syncretic therapeutic tripod (Amazonian medicine, psychotherapy, and community living). They anticipate that findings will inform other ayahuasca-assisted modalities and broader psychedelic-assisted therapy, identifying contextual elements — such as music, therapeutic integration, and treatment milieu — that might be tested in more controlled designs or adapted to outpatient settings. The discussion reiterates the study team's deliberate choice not to measure the immediate psychedelic "mystical" experience with standard scales, arguing this was not a good fit for a longitudinal naturalistic evaluation where ayahuasca is administered repeatedly from variable batches. Key limitations acknowledged by the authors include the observational design that precludes strong causal claims, lack of precise data on ayahuasca composition and per‑kg dosing, inability to measure phenomenological intensity of individual ceremonies within the protocol, absence of neuroimaging due to facility constraints, and substantial programme non-completion that may bias results. The team describes mitigation strategies such as a follow-up sub-project targeting those who left prematurely and inclusion of treatment exposure variables in analyses. Finally, the authors suggest the protocol and shared measurement set may serve as a usable framework for multisite comparative work and for evaluating context-dependent psychedelic interventions more generally, while emphasising the centrality of context to therapeutic effect.

In their concluding remarks the authors underscore that, despite limitations, the ATOP-Takiwasi protocol provides a comprehensive model for evaluating a context-dependent ayahuasca-assisted residential intervention. They acknowledge that uncontrolled factors (e.g., variable brew composition, unmeasured subjective intensity of individual sessions) and the observational design limit causal inference, and that high programme non-completion presents a risk of bias. To address some of these concerns the study includes a sub-project to follow up with early leavers and incorporates measures of length and intensity of participation as moderating variables. Overall, Rush and colleagues present the protocol as an important example of how mixed-methods, complexity-informed evaluation can advance understanding of both ayahuasca-assisted treatment and the broader therapeutic community model for substance use disorders. They propose that data generated by the study will inform quality improvement at Takiwasi, contribute to multisite comparisons under the ATOP umbrella, and offer insight into which contextual ingredients may be most valuable to translate or test in other psychedelic-assisted treatment models.