Fast take: A mixed-method study published 23 June 2026 shows strong baseline knowledge and positive attitudes among Cambodian community health workers (CHWs), but important gaps on specific T2D risk factors and clear health-system barriers to scale-up. Read the original paper at PLOS One. If you work on NCD scale-up, policy design, or implementation research, this post explains how to turn the study’s transcripts, survey data (n=153), and key-informant notes into reproducible, segment-aware insights using AI qualitative analysis, and how Evidano speeds that work (Evidano). Soft CTA: Want a 2-week pilot that reproducibly maps CHW KAP gaps to training and financing recommendations? Read the workflow below.
Key Takeaways
Evidano is an AI-powered qualitative data analysis platform that helps teams convert CHW surveys and interview notes into reproducible, segment-aware insights quickly.
The PLOS study published 23 June 2026 found that Cambodian CHWs have strong general NCD knowledge but important gaps in specific T2D risk recognition and multiple health-system barriers to scale-up.
- Study headline: 153 active CHWs surveyed (96% response rate) and 11 key-informant interviews reveal strong core NCD knowledge, with item-level gaps: family history recognised by 23.5% and tobacco recognised by 48.4%.
- System barriers: financial incentives, training, supervision, supply constraints, and role ambiguity emerged as major themes in key-informant interviews.
- Practical next steps: clarify CHW roles, target training on overlooked risk factors, and build sustainable incentive and supervision models before large-scale task-shifting.
- Evidano workflow: the post outlines a reproducible two-week pilot to transform surveys and KI notes into prioritized, segment-specific actions using AI-assisted coding and visuals.
FAQ: AI qualitative analysis of community health workers
What did the PLOS study find about CHW knowledge and practice?
The study found strong general NCD knowledge but specific gaps in T2D risk recognition and routine screening advice.
The study surveyed 153 active CHWs across six operational districts with a 96% response rate, and reported general NCD knowledge of about 90% (mean 6.3/7). The study also found low recognition of family history as a T2D risk (23.5%) and only 48.4% recognising tobacco as a T2D risk. Self-reported practice included advising blood glucose screening (83.0%) and blood pressure screening (85.6%).
How were qualitative findings collected and analysed in the study?
The qualitative component used 11 semi-structured key-informant interviews, coded inductively.
Qualitative notes were coded inductively in NVivo and identified system-level barriers such as financial incentives, training needs, supervision, supply constraints, and role ambiguity, which triangulated with the survey findings.
How can the study’s data be turned into policy-ready insights quickly?
The study’s survey CSV and KI notes can be harmonized, AI-assisted coded, and analysed to produce prioritized recommendations within a two-week pilot.
The post provides a two-week pilot checklist that includes import, auto-transcription/translation, PII redaction, AI-assisted coding, cross-segment frequency and co-occurrence analyses, visuals, and an executive brief for MOH or partners.
What are the main barriers to scaling CHW-led T2D/HTN activities reported in the study?
The main barriers reported were financial incentives, insufficient training, weak supervision, supply constraints, and role ambiguity.
Key-informant interviews emphasised that support is often ad-hoc, coming from user fees or donor funds, and recommended piloting blended incentives (small stipend plus non-financial recognition) and clarifying roles before adding responsibilities.
What ethical safeguards are recommended when re-analysing CHW or patient data?
Ensure informed consent, local approvals, data minimization, and appropriate PII handling when working with CHW or patient data.
The post notes that Evidano supports PII redaction and encrypted storage to help meet local governance and ethical requirements.
Findings snapshot
| Date / Item | Metric | Value | Source / Note |
|---|---|---|---|
| Study period | Fieldwork | Dec 2019 – Apr 2020 | Mixed-methods (survey + KI interviews) |
| Publication | Published | 23 June 2026 | PLOS One |
| Sample | Active CHWs surveyed | 153 (96% response rate; 153/160) | Survey in 6 operational districts |
| Demographics | Mean age | 49 years (SD 14) | Table 1 |
| Demographics | Female | 51.0% | Table 1 |
| Experience | Mean years as CHW | 8.6 years (SD 6.5) | Table 1 |
| Knowledge (general NCDs) | Correct responses | ≈90% (M = 6.3/7) | Table 2 |
| Knowledge (T2D risk factors) | Family history correct | 23.5% | Low recognition of family history |
| Knowledge (T2D risk factors) | Tobacco recognised | 48.4% | Gap on tobacco as risk |
| Practice | Advised blood glucose screening | 83.0% | Self-reported practice |
| Practice | Advised blood pressure screening | 85.6% | Self-reported practice |
| Barriers | Major themes | Financial incentives, training, supervision, supply constraints, role ambiguity | Key-informant interviews |
What the study did (plain English)
The study ran a cross-sectional survey of 153 active CHWs across six purposively selected operational districts and conducted 11 semi-structured key-informant interviews.
Design and methods: The authors ran a cross-sectional survey of 153 active CHWs across six purposively selected operational districts and conducted 11 semi-structured key-informant interviews. Quantitative answers generated frequencies, and qualitative notes were coded inductively in NVivo.
- Mixed-methods strength: high survey response (96%) allowed triangulation with key-informant interviews to surface system-level barriers.
- Key quantitative signals: strong core knowledge on prevention and NCD basics, weak recognition of family history (23.5%) and only half recognising tobacco as a T2D risk (48.4%).
- Qualitative signals: stakeholders emphasised both financial incentive gaps and non-financial needs (training, supervision, supplies), plus inconsistent role definitions and potential CHW overload.
So what for researchers and program managers (implications)
The study implies three immediate priorities for CHW-based NCD programmes: clarify roles, target training on overlooked risk factors, and build sustainable incentive and supervision models before large-scale task-shifting.
If you design CHW-based NCD programmes, these findings imply three immediate priorities: clarify roles, target training on overlooked risk factors, and build sustainable incentive and supervision models before large-scale task-shifting.
- Target training: Use the study’s item-level gaps (family history, tobacco) to design short KAP modules and measure pre/post change.
- Workload and role mapping: 44.5% of CHWs had one role; others had multiple. Map tasks and travel distances before adding T2D/HTN responsibilities.
- Finance and sustainability: stakeholders reported current support is ad-hoc (user fees, donor funds). Pilot blended incentives (small stipend plus non-financial recognition) and measure retention.
Do more, faster with Evidano
Ingest & harmonize (transcripts, surveys, Khmer support)
Evidano ingests and harmonizes structured survey CSVs and KI interview notes or audio, supporting transcription, translation with a custom dictionary, and PII redaction.
Import the study’s structured survey and the KI interview notes (or audio if available). Evidano supports transcription and translation with a custom dictionary and PII redaction, useful when working with Khmer notes or mixed-language inputs.
Output: a cleaned corpus aligned by respondent ID and segment (OD, role, years of service).
Automate coding and thematic extraction
Evidano applies AI-assisted coding to an uploaded codebook, proposes subcodes, and surfaces theme frequencies and salient quotes for each code.
Upload an initial codebook (e.g., NCD basics, risk factors, incentives, supervision). Evidano applies AI-assisted coding, proposes subcodes, and surfaces theme frequencies and salient quotes for each code.
Output: reproducible theme counts (e.g., percent mentions of tobacco as T2D risk) and an exportable codebook for audits.
Cross-segment analysis & visuals
Evidano runs cross-tab analyses and generates co-occurrence networks, hierarchical code trees, and word clouds for policy briefs.
Run cross-tab analyses (e.g., knowledge gaps by years of service or by OD) and produce co-occurrence networks, hierarchical code→subcode trees, and word clouds that you can include in policy briefs.
Output: visual evidence to justify training priorities and targeted interventions.
Iterative follow-up with AI avatar interviews
Evidano can deploy AI avatar interviewers for consent-driven autonomous follow-ups, and integrate new transcripts into the project for comparison to baseline themes.
Need more qualitative depth? Deploy Evidano AI avatar interviewers to run autonomous follow-ups (consent-driven) to probe themes that emerged (e.g., incentives vs. motivation).
Output: new transcripts automatically integrated into the project and compared to baseline themes.
Security & reproducibility
Evidano encrypts data and does not use customer data to train third-party models, providing audit-ready exports and reproducible analysis logs.
Data is encrypted, and Evidano does not use customer data to train third-party models, critical when health data and local governance rules restrict data sharing.
Output: audit-ready exports, reproducible analysis logs, and stakeholder-ready slide decks.
Two-week pilot: checklist to reproduce the study’s insights
A focused two-week pilot converts raw surveys and interview notes into prioritized actions using import, auto-transcription, AI-assisted coding, cross-segment analysis, visuals, and an executive brief.
Run this pilot to convert raw materials (surveys and interview notes) into prioritized actions.
- Day 1: Import survey CSV and KI notes into Evidano; set segments (OD, role, years active).
- Day 2–3: Auto-transcribe and translate and run a quick QA; apply PII redaction if needed.
- Day 4–6: Upload or create an initial codebook; run AI-assisted coding; review borderline codes.
- Day 7–9: Run cross-segment frequency and co-occurrence analyses; export top 10 themes and supporting quotes.
- Day 10–11: Build visuals (word cloud, network, code hierarchy) and draft targeted training recommendations.
- Day 12–13: If gaps remain, deploy a short AI avatar follow-up to 10 CHWs in a target OD.
- Day 14: Produce an executive brief and a reproducible analytics package to hand to MOH or local partners.
Ethics note
When working with CHW or patient data ensure informed consent, local approvals, and data minimization, and use PII redaction and encrypted storage where required.
This content is research-focused and non-diagnostic. When working with CHW or patient data ensure informed consent, local approvals, and data minimization. Evidano supports PII redaction and encrypted storage to help meet these requirements.
Wrapping up & next steps
The PLOS study (23 June 2026) shows CHWs in Cambodia are a strong basis for T2D/HTN scale-up but require targeted training, clarified roles, and sustainable support.
The PLOS study (23 June 2026) shows CHWs in Cambodia are a strong basis for T2D/HTN scale-up but need targeted training, clarified roles, and sustainable support. If you are designing that intervention, use AI qualitative analysis to turn KAP items and KI notes into prioritized, segment-specific actions faster and reproducibly.
- Ready to test this on your corpus? Start a pilot at Evidano and use the checklist above to deliver an evidence pack in two weeks.
- Read the original study: PLOS One.
- Try Evidano for free