Evidano is an AI-powered qualitative data analysis platform that accelerates reproducible coding and privacy-preserving synthesis. This post walks researchers and program teams through what user feedback from an electronic immunization registry (EIR) pilot in Rajshahi City Corporation, Bangladesh (published July 2, 2026) actually tells us and how to extract actionable insight with AI. The study (July–September 2024, n=321: 305 caregivers, 16 HCPs) reports high satisfaction but flags connectivity and device problems. Read the full source at PLOS One. In the next sections you’ll get a concise methods recap, a numbers snapshot, practical implications for researchers and implementers, and a two-week workflow to reproduce the qualitative analysis using AI-enabled tools like Evidano.
Key Takeaways
The PLOS One pilot shows broadly positive user perceptions but also clear operational constraints, and it is a feasibility snapshot not a scalability evaluation.
Re-analyzing open-ended responses with privacy-preserving AI can surface actionable themes (connectivity, device reliability, training) to prioritize fixes. The study period was July–September 2024 and the article was published 2 July 2026.
- Sample and main numbers: n=321 (305 caregivers; 16 HCPs); facility-based registration 87%; caregiver satisfaction 59–78% satisfied (16–22% highly satisfied); HCP high satisfaction 75–88%.
- Design note: cross-sectional descriptive survey, open responses reviewed but not published due to identifiability; treat findings as feasibility/acceptability signals rather than scalability proof.
- Practical priority areas: offline-capable UI and lightweight sync, device maintenance workflows, connectivity investments, and targeted refresher training.
Fast take: what the PLOS study actually found
The study found that caregivers and health workers in an urban EIR pilot reported broadly positive perceptions alongside operational constraints. Caregiver satisfaction ranged 59–78% satisfied with 16–22% highly satisfied across items, health-care-provider high satisfaction ranged 75–88% across items, and electronic registration occurred mostly at EPI centers (87%). Operational constraints centered on intermittent internet and device limitations, and the authors frame the work as a feasibility/acceptability snapshot not an effectiveness or scalability evaluation (study period July–September 2024; published 2 July 2026). See the original article at PLOS One.
- Study design: cross-sectional descriptive survey (n=321; 305 caregivers, 16 HCPs).
- Key numbers: 87% facility-based registration; reported registration timeliness ≈88% within 45 days (mentioned in discussion).
- Qualitative hint: open-ended responses (not publicly released) referenced connectivity and device issues alongside perceived convenience of reminders and better record access.
Findings snapshot (quick reference)
| Metric | Value | Source / Note |
|---|---|---|
| Total participants | 321 (305 caregivers; 16 HCPs) | PLOS One, July 2, 2026 |
| Facility-based registration | 87% | EPI centers primary registration |
| Caregiver satisfaction (satisfied range) | 59–78% satisfied; 16–22% highly satisfied | Likert items across service attributes |
| HCP high satisfaction | 75–88% highly satisfied (varies by item) | Small sample (n=16); interpret cautiously |
| Reported registration timeliness | ≈88% within 45 days | Mentioned in discussion; aligns with other LMIC pilots |
What happened & how the qualitative part was handled
The qualitative component used a pretested semi-structured questionnaire with closed Likert items and open-ended questions, and open responses were reviewed but withheld from publication for identifiability. Quantitative items were summarized with descriptive statistics (SPSS). Open responses informed the narrative on operational challenges but were not published because authors anonymized data to protect respondents.
- Sampling: purposive recruitment of caregivers attending EPI sessions, useful for experience-rich responses but can overestimate satisfaction.
- HCP data: small n (16) so qualitative quotes from providers could be high-leverage but are few.
- Ethics: data anonymized; open responses not publicly shared for confidentiality. Non-diagnostic; research-focused.
So what for researchers, UX and program teams?
Researchers
Researchers should use descriptive satisfaction metrics as a starting point, and not as a substitute for systematic thematic coding of open responses. To assess feasibility versus scalability researchers need longitudinal and cross-site qualitative work, and they should extract themes on barriers (connectivity, device reliability, training) and benefits (reminders, record access) and quantify co-occurrence across segments (caregiver vs. HCP, facility vs. outreach).
UX / Product teams
UX and product teams should treat high reported satisfaction as potentially masking friction points and prioritize offline-capable user interfaces, lightweight data sync, and clearer device maintenance workflows. Teams should instrument post-deployment micro-surveys and session logs to correlate reported issues with device and network telemetry.
Policy & implementation leads
Policy and implementation leads should invest in connectivity and phased rollout with refresher training, and validate registry completeness against household or coverage surveys. Decision-makers should note this was urban RCC data and plan targeted qualitative assessments for rural and informal settlements where patterns may differ.
Do more, faster with Evidano
Problem: open-ended answers contain PII or sensitive context
You can safely ingest open-ended responses for thematic analysis while preserving privacy, because Evidano provides automated PII redaction and encrypted storage. In the PLOS study the authors withheld open responses for identifiability; Evidano lets teams analyze those responses while preserving confidentiality.
Problem: small provider samples plus rich text are hard to generalize
You can quantify qualitative signals across segments to interpret small-n findings, because Evidano cross-segment analysis quantifies theme frequency by segment, surfaces co-occurrence networks, and shows which concerns cluster with high or low satisfaction. This helps interpret signals from groups like the n=16 HCP sample.
Problem: multilingual inputs and inconsistent terminology
You can translate and standardize terminology with custom dictionaries, because Evidano provides transcription and translation with domain-preserving dictionaries. If interviews or field notes mix Bengali and English, Evidano translates consistently and preserves domain terms such as EPI and EIR via a custom dictionary.
Problem: stakeholder alignment and rapid reporting
You can produce rapid, shareable outputs for decision-makers, because Evidano creates visualizations (theme frequencies, hierarchical code→subcode trees, co-occurrence networks) and downloadable executive briefs to share recommendations in minutes.
Security & compliance
Evidano uses end-to-end encryption and does not use your data to train third-party models, which supports ethically sensitive health research and compliance with IRB or institutional requirements.
Two-week workflow: reproducible qualitative analysis for an EIR pilot
This two-week workflow lays out reproducible qualitative analysis steps for an EIR pilot: Week 1 covers ingest and preparation, and Week 2 covers coding, synthesis, and reporting.
- Day 1: Import transcripts, survey open responses, and metadata (site, role, date) and enable PII redaction and custom dictionary entries (local vaccine names, EPI terms).
- Day 2–3: Auto-transcribe audio (if present) and run translation review, tagging low-confidence segments for human quality control.
Week 2: Code, synthesize, report
- Day 4–6: Run AI-assisted thematic coding and accept or adjust the suggested codebook, using hierarchical codes to separate operational barriers into 'connectivity', 'device', and 'training'.
- Day 7–9: Run cross-segment frequency and co-occurrence analyses (caregiver vs. HCP; facility vs. outreach), and export top quotes per theme with provenance for stakeholder slides.
- Day 10: Generate a one-page executive brief and a slide-ready visualization pack for implementation leads with clear recommended actions (e.g., offline sync rollout, targeted refresher training).
FAQ: qualitative analysis of EIR feedback
How do I compare themes between caregivers and HCPs?
Use cross-segment theme frequency and normalized co-occurrence to compare themes between caregivers and HCPs, and surface representative quotes for each segment. Evidano shows which themes are disproportionately reported by each group and provides representative quotes with provenance.
Can AI reliably redact PII in local languages?
AI can assist reliable PII redaction in local languages, but validation is required: use a custom dictionary and review a validation sample. Evidano supports configurable redaction rules and human-in-the-loop checks to validate redaction quality.
Are short cross-sectional pilots worth deep qualitative coding?
Short cross-sectional pilots can be worth deep qualitative coding because small samples (for example, n=16 HCPs) can reveal high-impact operational fixes when coded and triangulated with registry metadata. Triangulate qualitative findings with timestamps and registration location for richer interpretation.
Wrapping up: next steps
The immediate next steps are to treat open-ended data as the explanatory signal for quantitative metrics and to use AI to accelerate coding while preserving privacy and reproducibility. Step 1: Re-import your open responses and enable PII redaction. Step 2: Run thematic and cross-segment analysis to prioritize fixes such as offline mode, device maintenance, and refresher training. Step 3: Share an executive brief with visuals and representative quotes for decision-makers.
- Step 1: Re-import your open responses and enable PII redaction.
- Step 2: Run thematic plus cross-segment analysis to prioritize fixes (offline mode, device maintenance, refresher training).
- Step 3: Share an executive brief with visuals and representative quotes for decision-makers.
Ready to try this on your own EIR dataset? Try Evidano for free.