Evidano is an AI-powered qualitative data analysis platform that ingests transcripts (audio→text), survey spreadsheets, and PDF reports to produce harmonised datasets for coding and rapid synthesis. Late recognition and referral are key drivers of poor childhood-cancer survival in low-resource settings. This post uses a qualitative analysis of training to unpack a multilevel onsite training and 6-month mentorship program in Northwest Ethiopia (Jan–Sep 2024) that raised provider knowledge (median scores to ≈91) and cut diagnostic intervals by ~54%. We draw actionable lessons for implementation researchers, program leads, and qualitative teams and show how to reproduce the analytic steps using AI-enabled tools. Read the original study in PLoS Medicine for full methods and data.
Key Takeaways
A tiered onsite training combined with six months of monthly on-site and tele-mentorship in Northwest Ethiopia reduced the median diagnostic interval for childhood cancer by about 54% and raised median knowledge scores to approximately 91. The qualitative analysis explained how leadership engagement, adapted pictorial materials, cascade mini-trainings, and mentorship cadence produced behaviour change despite resource and security constraints.
- The intervention trained 18 hospital clinicians, 29 primary clinicians, and 1, 020 health extension workers between Jan–Sep 2024, with a reported budget of USD 52, 762.
- Median diagnostic interval fell ~54%, from 56.5 to 25.8 days (p < 0.001); patient delay fell 9.3% (27.0 → 24.5 days, p = 0.02).
- Study methods were quasi-experimental pre–post mixed-methods: KAP surveys, chart review (n = 100), and CFIR-guided interviews and FGDs; lack of a contemporaneous control limits causal certainty.
Findings snapshot
| Metric | Value | Source / Note |
|---|---|---|
| Published | June 29, 2026 | PLoS Medicine |
| Intervention period | Jan–Sep 2024 (6-month mentorship Apr–Sep) | Study methods |
| Trainees | 18 Level I clinicians; 29 Level II clinicians; 1, 020 HEWs | Onsite workshops + pictorial modules |
| Chart review | n = 100 pediatric cancer charts | Patient-journey intervals measured |
| Diagnostic interval change | ↓ ~54% (56.5 → 25.8 days) | p < 0.001 |
| Patient delay change | ↓ 9.3% (27.0 → 24.5 days) | p = 0.02 |
| Knowledge scores | Median rose to ≈90.9 at both levels | KAP surveys pre/post |
| Budget | USD 52, 762 | Reported cost for the intervention |
What happened: methods and mechanisms (plain English)
The team implemented tiered, hands-on trainings and mentorship, and used mixed methods to identify mechanisms that produced measurable referral improvements. The team ran 10-day workshops for tertiary/secondary clinicians, 7-day workshops for primary providers, pictorial 5-day modules for HEWs, and six months of mentorship (monthly on-site plus tele-support). Data collection combined pre/post KAP surveys, chart reviews, and CFIR-based interviews and FGDs.
Mechanisms that explain the quantitative gains included repeated contact via mentorship, peer-led cascade training, standardised referral forms, and monthly audits. Contextual barriers that attenuated impact included reagent shortages, staff rotations, transport and security constraints, and persistent cultural beliefs delaying care-seeking. The qualitative component mattered because interviews and FGDs illuminated how mentorship cadence and leadership buy-in produced behaviour change despite those systemic constraints.
Implications for research & implementation teams: qualitative analysis of training
For implementation researchers
Implementation researchers should use CFIR or a comparable implementation framework to code facilitators and barriers early, and triangulate KAP scores with interviews to detect false positives (for example, high knowledge but low uptake). Capture process fidelity metrics such as attendance and mentorship logs, because these explained dose–response patterns in the study.
For program managers & MOH teams
Program managers and MOH teams should design cascade trainings and budget for dedicated mentorship, because the study reached remote HEWs via mini-trainings and increased referrals threefold. Track simple, actionable indicators like referrals per HEW and referral-form error rates, and use monthly case conferences for rapid course correction.
For qualitative teams / UX researchers
Qualitative and UX teams should prioritise short FGDs and targeted IDIs to surface cultural beliefs and workload trade-offs, since HEWs reported diverted time. Codebooks should allow inductive themes so unexpected effects, such as interdisciplinary spillover, are captured and quantified for stakeholders.
Do more, faster with Evidano (map to this use case)
Problem: fragmented transcripts, surveys, and field notes → Solution
Evidano ingests transcripts (audio→text), survey spreadsheets, and PDF reports and produces harmonised datasets for coding, removing manual consolidation tasks that delay analysis.
Problem: CFIR coding is time-consuming → Solution
Import your CFIR codebook into Evidano, run AI-assisted coding across interviews, get inter-coder agreement metrics, and export coded extracts for rapid triangulation with KAP scores.
Problem: multilingual, pictorial HEW notes & local terms → Solution
Use Evidano transcription and translation with custom dictionaries (retain local terms like “nififit”) and PII redaction to produce clean, searchable transcripts.
Problem: stakeholder-ready outputs → Solution
Generate theme frequency tables, co-occurrence networks, hierarchical code→subcode visualizations, and clickable quotes for dashboards and monthly case-conference slides, reproducible and exportable.
Security & governance
Evidano uses encrypted storage and proprietary LLMs tuned for qualitative research, and data is not used to train third-party models, which supports handling sensitive health data and ethics approvals.
Two‑week reproducible workflow (checklist)
A reproducible qualitative synthesis following the study's approach can be executed in about 10–14 days with a disciplined workflow. Follow these steps to replicate the study's qualitative synthesis:
- Day 0–2: Gather inputs: audio files, translated transcripts, KAP survey spreadsheets, and chart-review CSVs.
- Day 3–4: Upload to Evidano; set a custom dictionary for local terms and import the CFIR codebook.
- Day 5–7: Run automated transcription and translation checks, then run AI-assisted initial coding and review edge cases.
- Day 8–10: Produce thematic frequency tables, segment comparisons (for example, HEWs vs clinicians), and co-occurrence maps.
- Day 11–12: Pull illustrative quotes and generate stakeholder slides; validate results with one or two senior analysts.
- Day 13–14: Export a reproducible report and raw coded data for archiving and ethics audits.
Ethics note
This post interprets published, de-identified research and is research-focused and non-diagnostic. When handling transcripts with identifiable health information, ensure local IRB approvals and use PII redaction tools (available in Evidano).
FAQ: qualitative analysis of training
What was the main effect of the training on diagnostic intervals?
The main effect was a reduction in the median diagnostic interval by about 54%, from 56.5 to 25.8 days. This change was statistically significant (p < 0.001).
Who was trained and when did the intervention run?
The intervention ran Jan–Sep 2024 and trained 18 Level I clinicians, 29 Level II clinicians, and 1, 020 health extension workers, with six months of mentorship from April–September 2024.
What methods did the study use to measure outcomes?
The study used a quasi-experimental pre–post mixed-methods design with KAP surveys, a chart review of n = 100 pediatric cancer charts, and CFIR-guided interviews and FGDs to measure outcomes and explain mechanisms.
How did the qualitative analysis explain why the intervention worked?
The qualitative analysis identified enabling mechanisms such as leadership engagement, adapted pictorial materials, cascade mini-trainings to reach HEWs in conflict zones, and rapid feedback loops via case conferences as drivers of improved referrals and diagnostic timeliness.
Wrapping up: next steps and CTA
If you are running implementation research or scaling training interventions, apply a disciplined qualitative analysis of training to expose mechanisms, not just outcomes. The PLoS Medicine study (published June 29, 2026) shows how combined training plus mentorship can halve diagnostic intervals under difficult conditions, and qualitative data explained how mentorship cadence, leadership, and cascade trainings produced that effect.
Ready to reproduce this kind of synthesis? Try Evidano for free to ingest transcripts, run CFIR coding, and generate stakeholder-ready visuals in days, not weeks.