Evidano is an AI-powered qualitative data analysis platform that ingests, codes, and visualizes qualitative research data. This post refracts a PLOS Medicine report (published 29 June 2026) through the lens of qualitative analysis of childhood cancer training. Between January-Sep 2024 a multilevel, onsite training plus 6-month mentorship in Northwest Ethiopia trained 18 hospital clinicians, 29 primary providers and 1, 020 health extension workers (HEWs), with chart reviews of 100 pediatric cancer cases. The study reported large gains (knowledge medians rose to ~90.9, diagnostic interval fell 54.3% to 25.8 days) but also persistent treatment delays and contextual barriers. For research teams and health-policy analysts, the qualitative dataset (18 IDIs, multiple FGDs, CFIR-coded transcripts) is a rich resource for understanding what changed and why. Read on for a concise playbook to reproduce the study’s thematic findings, turn interview transcripts and KAP texts into cross-segment evidence, and operationalize fixes faster using Evidano (Evidano). Note: this analysis is research-focused and non-diagnostic; see the original study at PLOS Medicine.
Key Takeaways
The PLOS Medicine study found that a context-tailored, multilevel onsite training plus 6-month mentorship (Jan-Sep 2024) produced large clinician knowledge gains and a 54.3% reduction in diagnostic interval in Northwest Ethiopia, while treatment initiation remained constrained by supplies and infrastructure.
- The intervention trained 18 Level I clinicians, 29 Level II providers, and 1, 020 HEWs, and used onsite courses plus a 6-month mentorship.
- Diagnostic interval fell 54.3% from 56.5 to 25.8 days (chart review n=100, p < 0.001) while treatment start times lagged due to reagent and pathology shortages.
- Key qualitative themes that explained outcomes were leadership and ownership, diagnostic supply constraints, HEW trust and community beliefs, and adaptive mentorship fidelity.
- A mixed-methods approach (KAP surveys, chart review, 18 IDIs, HEW FGDs, CFIR coding) supports rapid-cycle evaluation and replication in fragile settings.
Fast take + source
A context-tailored, multilevel onsite training and mentorship model (Jan-Sep 2024) was associated with big provider knowledge gains and a 54.3% drop in diagnostic interval in Northwest Ethiopia, but treatment start times lagged due to infrastructure constraints, as reported in PLOS Medicine.
- Study design: quasi-experimental pre-post mixed-methods (no contemporaneous control).
- Key sample: 18 Level I clinicians, 29 Level II providers, 1, 020 HEWs; chart review n=100.
- Implementation window: Jan-Sep 2024; published 29 June 2026. Read the full paper at PLOS Medicine.
Findings snapshot
| Metric | Value | Source / Note |
|---|---|---|
| Training period | Jan-Sep 2024 | Intervention + 6-month mentorship |
| Participants trained | 18 (Level I), 29 (Level II), 1, 020 HEWs | Onsite workshops + pictorial modules |
| Knowledge (median) | ↑ to 90.9 (post) | Level I/II combined; normalized 0-100 |
| Diagnostic interval | ↓ 54.3% (56.5 → 25.8 days) | Chart review n=100; p < 0.001 |
| HEW referrals | 0.4 → 1.2 referrals/month (tripled) | District registers |
| Budget | USD 52, 762 | Total reported implementation cost |
What happened: study design & qualitative inputs
The University of Gondar implemented immersive onsite courses and a 6-month mentorship that combined monthly onsite visits and tele-support to improve recognition and referral.
- The University of Gondar ran 10-day tertiary and 7-day primary onsite courses, a 5-day pictorial module for HEWs, then a 6-month mentorship with monthly onsite visits and tele-support.
- Quantitative KAP surveys and chart reviews were triangulated with 18 in-depth interviews and multiple HEW focus groups coded using CFIR.
- Primary qualitative materials included IDIs with Level I/II clinicians, FGDs with HEWs, mentor logs, and field notes; analysis used inductive thematic analysis, CFIR mapping, and NVivo coding with dual coders.
- Contextual constraints included active regional conflict since April 2023, seasonal rains, reagent shortages, and staff rotations.
Qualitative analysis of childhood cancer training: top themes
1) Leadership & ownership
Supportive hospital leadership enabled referral-form adoption and monthly case conferences and was linked to faster decision-making and higher fidelity mentoring.
Facilitator evidence: interview quotes linked leadership engagement to faster decision-making and higher fidelity mentoring.
2) Diagnostic supply constraints
Regular shortages of reagents and pathology constrained the translation of faster recognition into faster treatment initiation, explaining why diagnosis shortened but treatment start rose approximately 11.9%.
Barrier evidence: stockouts and limited pathology services limited timely treatment initiation after diagnosis.
3) Community beliefs & HEW trust
HEWs were trusted local actors and pictorial modules helped overcome literacy constraints, which boosted referrals from 0.4 to 1.2 per month.
Barrier evidence: persistent cultural beliefs and traditional healer use still delayed first contact despite increased HEW referrals.
4) Mentorship fidelity & adaptation
Adaptive mentorship, combining biweekly tele-support and cascade mini-trainings, produced higher knowledge retention (95.2% vs 88.3%) and fewer referral errors.
Implementation evidence: cascade trainings were used when travel was blocked and maintained mentoring fidelity under fragile conditions.
So what for researchers and program teams
The mixed-methods archive from this study is a template for rapid-cycle evaluation and scale-up for teams designing implementation research or evaluating training interventions.
- Use CFIR-coded transcripts to link micro-level barriers, such as reagent stockouts, to macro outcomes like treatment delays.
- Triangulate KAP score changes with thematic shifts in interviews, since practice gains often precede attitude shifts.
- Plan for adaptive mentorship (tele plus cascade) where security and transport are unstable.
Do more, faster with Evidano
Ingest and clean heterogeneous text
Evidano ingests interview transcripts, translated FGDs, KAP survey text, and mentorship logs and supports transcription, translation, and PII redaction so teams can centralize sources safely.
Operational note: use Evidano to apply custom dictionaries and redaction rules before analysis.
Automate thematic coding + CFIR mapping
Evidano runs AI-assisted coding to surface CFIR constructs and inductive themes, then enables interactive codebook refinement and reproducible exports.
Operational note: Evidano exports hierarchical codes and subcodes for reporting and audit trails.
Cross-segment and frequency analysis
Evidano compares segments such as Level I, Level II, and HEWs side-by-side, computes theme frequencies, and links themes to quantitative outcomes like knowledge scores and diagnostic intervals.
Operational note: cross-segment analysis can identify facilities with strong leadership and faster diagnostic intervals.
Visualize and share decision-grade evidence
Evidano creates co-occurrence networks, word clouds, and hierarchical code trees that make findings actionable for stakeholders and provide clickable quotes for dissemination briefs.
Operational note: use visuals to populate policy memos and stakeholder presentations.
Secure, research-ready governance
Evidano uses encrypted storage and proprietary LLMs tuned for qualitative research, and customer data is not used to train third-party models, supporting IRB-compliant projects in sensitive contexts.
Governance note: de-identify at upload and log access for auditability.
Checklist: reproduce this analysis in 7 steps
A seven-step checklist converts qualitative materials into decision-ready evidence.
- 1) Centralize transcripts, KAP free-text, mentorship logs, and de-identify at source.
- 2) Upload to Evidano and apply transcription/translation rules with a custom dictionary for local terms.
- 3) Run an initial unsupervised thematic extraction to surface common motifs such as leadership, supplies, and cultural beliefs.
- 4) Import a CFIR codebook and run AI-assisted coding; review conflicts with dual coders.
- 5) Link coded themes to quantitative measures (knowledge scores, diagnostic intervals) using Evidano cross-segment analysis.
- 6) Create visuals (co-occurrence networks, hierarchical code maps) for decision meetings.
- 7) Export quotes and an executive brief for policymakers; schedule a follow-up rapid survey to track attitude changes.
FAQ: Qualitative analysis of childhood cancer training
How do you compare segments reliably?
Use normalized KAP scores and code frequency per 1, 000 words to compare segments reliably.
Practical tip: Evidano automates normalization and significance testing across segments, enabling fair comparisons between Level I, Level II, and HEWs.
Can AI handle local-language idioms in HEW transcripts?
Yes, AI can handle local-language idioms when you use custom dictionaries and translation memory to preserve clinical terms and local phrases.
Practical tip: apply a custom dictionary in Evidano to preserve clinical terms and local idioms, for example labels for lymphadenopathy.
Is the approach IRB-friendly?
Yes, the approach is IRB-friendly when data are de-identified at upload, access is logged, and analysis focuses on systems and training evaluation rather than diagnostics.
Practical tip: de-identify at upload, use encrypted storage, and keep analysis non-diagnostic to meet IRB and ethical requirements.
Conclusion & next steps
The PLOS Medicine study demonstrates that targeted, multilevel training plus mentorship can rapidly improve recognition and shorten diagnostic timelines for childhood cancer in low-resource settings, while qualitative data reveal bottlenecks such as supply chains, cultural beliefs, and leadership variation.
- If you manage implementation research, convert transcripts and KAP text into linked thematic and quantitative evidence for the fastest path to action.
- Start a pilot by uploading a sample of interviews and KAP free-text to Evidano and run a CFIR-aligned thematic analysis to produce an executive brief in days.
Ready to try it? Try Evidano for free.