Evidano is an AI-powered qualitative data analysis platform that helps teams ingest transcripts, surveys, and lab records and turn them into reproducible thematic insights. Problem: Thailand’s 2022 cannabis decriminalization has coincided with rising use among young adults. Payoff: learn how researchers can convert the PLOS ONE mixed-methods dataset (n=200; published June 26, 2026) into reproducible thematic and cross-segment insights that inform youth-centred harm reduction. This article shows a concise qualitative analysis workflow and how to operationalize it in Evidano (Evidano). First, read the original study: PLOS ONE.
Key Takeaways
This post shows how to convert a PLOS ONE mixed-methods study (n=200, published June 26, 2026) into reproducible thematic and cross-segment insights and how to run a 2-week Evidano pilot to reproduce the paper’s qualitative synthesis.
- The PLOS ONE study (n=200; mean age 21.45) found 22% past-month cannabis use and a significant association between cannabis use and substance- or alcohol-influenced sex (Poisson PR = 1.24, 95% CI: 1.09–1.41).
- The original study combined tablet surveys, routine clinic lab results, and 30 semi-structured Thai-language interviews, which supports integrated mixed-methods coding and cross-segment analysis.
- The article maps a reproducible 2-week workflow to ingest surveys and transcripts, run auto-theme extraction and cross-segment summaries, then generate stakeholder deliverables using Evidano.
Fast take + source
Fast take: A mixed-methods PLOS ONE study published June 26, 2026 (n=200; mean age 21.45) found 22% past-month cannabis use among clinic-attending young adults in Bangkok and a significant association between cannabis use and substance- or alcohol-influenced sex (Poisson PR = 1.24, 95% CI: 1.09–1.41).
- Why this matters: qualitative quotes reveal perceived disinhibition, but users often cite personal responsibility, a split that matters for messaging and intervention design.
- What you’ll get from this post: a compact thematic analysis plan, role-specific implications, and a reproducible 2-week pilot you can run in Evidano.
- Source: PLOS ONE.
Findings snapshot
| Date | Sample (n) | Past-month cannabis | Key association | HIV prevalence | Source |
|---|---|---|---|---|---|
| June 26, 2026 | 200 (age 18–24; mean 21.45) | 22% | Substance-/alcohol-influenced sex (PR=1.24; 95% CI 1.09–1.41) | 1.5% | PLOS ONE |
What the study did (plain English)
The study used a sequential mixed-methods design combining a tablet-based self-report survey, routine clinic lab results, and semi-structured interviews to explain numeric associations with context.
Design: sequential mixed-methods, a tablet-based self-report survey (Nov 27, 2023–Mar 29, 2024) of 200 clinic clients plus 30 semi-structured interviews (May–Sep 2024).
- Quant: demographics, sexual behaviour, PrEP adherence, CUDIT-R scores, and routine lab results (HIV/STIs).
- Qual: 30 Thai-language interviews transcribed and coded; thematic analysis used to interpret contexts behind numeric associations.
- Main quantitative outcome: cannabis users more likely to report alcohol- or substance-influenced sex; no statistical link to condomless sex, PrEP adherence, or STI results in this sample.
Implications for researchers, UX teams & clinics
For qualitative researchers and UX teams
Qualitative researchers and UX teams should integrate qualitative coding with quantitative prevalence so interview excerpts can explain mechanisms behind numeric associations.
Don’t stop at prevalence: the study shows why integrated qualitative coding matters, interview excerpts explain the mechanism (lowered inhibition vs. asserted responsibility).
Use cross-segment analysis (users vs non-users; gender identity; co-use patterns) to surface contradictory narratives that will change how you frame messages and probes in follow-up research.
For clinic teams and public-health practitioners
Clinic teams and public-health practitioners should prioritise brief, non-stigmatizing harm-reduction conversations that address polysubstance contexts rather than blanket risk warnings.
Actionable takeaway: prioritize brief, non-stigmatizing harm-reduction conversations that address polysubstance contexts (61% of cannabis users also co-used alcohol) rather than blanket risk warnings.
Screening: integrate short substance-use screens (AUDIT-C, CUDIT-R) into sexual health visits and link to tailored counselling when co-use is present.
Limitations & ethics note
The study’s limitations and ethics considerations mean findings are research-focused and not diagnostic guidance.
Limitations: convenience sample, self-report bias, limited power to detect associations for some outcomes; qualitative narratives are contextual and not causal proof.
Ethics note: findings are research-focused and non-diagnostic. Sensitive datasets are controlled-access per the authors; replicate only under approved protocols and informed consent.
Do more, faster with Evidano, mapped to this study
Import and harmonize mixed inputs
To import and harmonize mixed inputs, ingest survey spreadsheets and Thai transcripts while preserving timestamps and metadata so coding aligns across sources.
Problem: surveys, Thai interview transcripts, and clinic lab records live in different formats.
Evidano: ingest survey spreadsheets and Thai transcripts, preserve timestamps and metadata, and apply custom dictionary entries (drug terms, local venue names) so coding aligns across sources.
Thematic + cross-segment analysis
To run thematic and cross-segment analysis faster, auto-generate themes and compute frequencies by subgroup to triangulate quotes with subgroup metrics.
Problem: manually triangulating quotes with subgroup metrics is slow and error-prone.
Evidano: auto-generate themes, surface co-occurrence networks (e.g., 'cannabis' + 'bar' + 'unplanned sex'), and compute frequency and prevalence by segment (users vs non-users; transgender women; students).
Transcription, translation & PII handling
To handle multilingual interviews and protect identities, run transcription with a custom dictionary, translate, and redact PII before analysis.
Problem: multilingual interviews need reliable transcription and safe redaction.
Evidano: built-in transcription with custom dictionary, Thai→English translation, and PII redaction before analysis, all encrypted and never used to train third-party models.
From insight to deliverable
To move from insight to deliverable, generate visual exports and use an AI chat over your corpus to draft harm-reduction scripts or briefing notes.
Problem: stakeholders need concise, defensible evidence for clinic protocols.
Evidano: one-click visual exports (word clouds, hierarchical codes → subcodes, quote exports) and AI chat over your corpus to draft harm-reduction scripts or briefing notes.
2-week pilot workflow (reproduce the paper’s qualitative synthesis)
This 2-week pilot workflow reproduces the paper’s qualitative synthesis by ingesting survey and lab data, transcribing interviews, extracting themes, and producing stakeholder-ready outputs.
Week 1, ingest & prep:
- Day 1: Upload survey CSV and lab results; map participant IDs to transcripts.
- Day 2–3: Upload audio files; run transcription with custom Thai dictionary and PII redaction.
- Day 4: Run automated theme extraction and configure codebook (inspired by study codes: disinhibition, personal responsibility, co-use).
Week 2, analyse & deliver:
- Day 6–8: Review auto-coded segments, adjust rules, and merge subcodes (e.g., 'alcohol-influenced sex').
- Day 9: Run cross-segment frequency and PR-like summaries to compare themes by cannabis-use status.
- Day 10–11: Generate visuals and export a stakeholder brief; use Evidano chat to draft clinic messaging and a one-page harm-reduction checklist.
FAQ: qualitative analysis of cannabis use
Q: How do I compare themes across subgroups reliably?
Use code harmonization and compute theme frequency plus conditional co-occurrence to compare themes across subgroups reliably.
Use code harmonization (imported codebook) and compute theme frequency + conditional co-occurrence. Evidano provides cross-segment tables and significance flags to prioritize differences.
Q: Can AI handle Thai transcripts accurately?
AI can assist with Thai transcripts but you should validate auto-transcripts against human transcripts before bulk processing.
Use custom dictionaries for local slang and drug terms; validate a sample of auto-transcripts against human transcripts before bulk processing.
Q: How is sensitive health data protected?
Sensitive health data should be encrypted and redacted; Evidano supports encryption and PII redaction and does not share your data to train third-party models.
Evidano encrypts data end-to-end, supports PII redaction, and does not share your data to train third-party models, suitable for clinic and policy research.
Wrapping up & next steps
This article recommends replicating the study’s triangulation strategy on your own corpus, pairing quick surveys with targeted interviews, then running thematic and cross-segment analyses to reveal mechanisms behind numerical associations.
- If you run clinic-based research, start with the 2-week pilot above and use outputs to design non-stigmatizing harm-reduction messages.
- Ready to try this in Evidano? Try Evidano for free and bring your transcripts, surveys, and codebook; we’ll help you go from raw data to stakeholder-ready insights in days.