Qualitative Analysis of Telemedicine Acceptability

Hypertension care in low-resource settings often stalls on access, trust, and familiarity, exactly the barriers explored in the JMIR qualitative study published 25 Aug 2025. This qualitative analysis of telemedicine acceptability (www.jmir.org/2025/1/e72568) summarizes what 40 semistructured interviews in Soweto (Jul–Sep 2023) reveal and shows how research teams can transform interview transcripts into action in days (not months) using AI-enabled qualitative research tools like www.evidano.com. Read on to learn the four emergent themes (trust & credibility; comfort with tech; prior experience; in-person preference), concrete implications for UX/research/policy teams, and a tight 2‑week Evidano pilot you can run to produce thematic, frequency, and cross-segment analysis ready for stakeholders.

Fast take & source

In brief: JMIR (published 25 Aug 2025) interviewed 40 community members in Soweto sampled from a 2, 041-household screening to assess openness to web-based health modalities for BP management. Full study: www.jmir.org/2025/1/e72568.

Design: Random sample of 40 semistructured interviews (20 elevated BP, 20 high BP) conducted Jul–Sep 2023.
Core finding: openness to web-based care but persistent preference for in-person visits driven by limited prior experience, trust concerns, and data/airtime constraints.
Why it matters: these are implementation barriers that qualitative teams can quantify and monitor across segments (age, internet skills, BP level) using AI-enabled pipelines.

Findings snapshot (key numbers)

Metric	Value	Source / note
Households screened (original)	2, 041	Home health screenings (May 2023)
Invited to interview	178	Random selection to reach n=40
Interviews completed	40 (20 elevated, 20 high BP)	Jul–Sep 2023; published 25 Aug 2025
Smartphone ownership (interview sample)	28/40 (70%)	Table 1
Internet skills (self-reported)	15/40 (38%)	Table 1
Gender (interview sample)	28 female, 12 male	Table 1
Main emergent themes	Trust/credibility; Comfort with tech; Experience; In-person preference	Deductive thematic analysis (UTAUT2 mapped)

What happened: methods in plain English

Researchers randomly sampled participants identified during home screenings and ran 12–20 minute semistructured interviews in participants’ preferred language. The interview guide was mapped to the Extended UTAUT2 model and translated/back-translated into isiZulu.

Analysis method: deductive thematic analysis using an a priori codebook mapped to UTAUT2; transcripts coded in Dedoose; Braun & Clarke's 6-phase framework applied.
Data note: many Likert-type responses were nonnumeric and excluded; qualitative answers formed the analytic core.
Ethics: informed consent, deidentified data, R50 compensation, University of the Witwatersrand ethics approvals.

So what for research, UX, and health teams

For qualitative researchers

Trust emerged as a cross-cutting driver of intention, include trust-specific codes and quote-level provenance when you build codebooks.

Segment reporting matters: compare age, smartphone ownership, and internet-skill strata (the study found older cohorts had lower internet skills and different facilitating conditions).

Operational tip: pre-test Likert instruments in-context, the JMIR team excluded many numeric replies; capture free-text alternatives and code them systematically.

For UX / product teams

Design for limited data/airtime: support low-bandwidth flows (SMS, asynchronous chat, brief audio notes) and offline-first experiences.

Offer hybrid pathways: the community favored web-based care as a supplement, not a replacement, prototype ‘clinic + virtual follow-up’ journeys.

Measure perceived diagnostic adequacy: add tasks in usability testing that estimate user confidence in remote diagnosis vs in-person.

For health ops & policy teams

Rollout strategy: start web-based services as add-ons to routine care to build familiarity and trust over time.

Address digital inclusion: subsidize airtime/data or deliver community Wi‑Fi kiosks and digital literacy sessions for older adults.

Evaluation: track uptake, trust indicators, and BP outcomes (e.g., control rates) across segments to detect inequitable adoption.

Do more, faster with Evidano (mapped to this study)

Problem: scattered transcripts & missed themes → Solution: ingest + thematic analysis

Upload interview audio or transcripts and run automatic thematic extraction aligned to UTAUT2 plus custom trust-related codes.

Evidano produces hierarchical codes → subcodes and surfaces representative quotes for each theme so you can justify recommendations with verbatim evidence.

Problem: inconsistent coding across analysts → Solution: codebook import + AI-assisted coding

Import the JMIR study’s codebook or create one in-app; Evidano applies it consistently across transcripts and flags low-confidence codes for human review.

Problem: segment comparisons (age, internet skills, BP level) are manual → Solution: cross-segment analysis

Run cross-segment frequency tables and statistical summaries automatically (e.g., smartphone ownership by BP group) to replicate and extend the study’s segment tests.

Problem: language barriers & PII risk → Solution: transcription + translation + PII redaction

Evidano transcription supports custom dictionaries and isiZulu translation, with built-in PII redaction so transcripts remain research-safe.

Data is encrypted and never used to train third-party models, helpful for ethics and institutional review.

Problem: stakeholder-ready outputs take weeks → Solution: visuals & AI chat over your corpus

One-click visualizations (word clouds, co-occurrence networks, hierarchical code maps) and an AI chat over your documents let you generate executive briefings and slide-ready quotes in hours.

Advanced: run rapid follow-ups with AI avatars

If the JMIR team wanted more post-study probes (e.g., monetization barriers like airtime), Evidano can deploy AI-avatar interviews to collect structured qualitative follow-ups at scale.

2‑week pilot checklist (reproduce & extend the study)

Run this checklist to go from raw audio to a stakeholder-ready brief in two weeks using Evidano and minimal analyst time.

Day 0–2: Collect transcripts (or upload JMIR open dataset request) and metadata (age, BP group, smartphone ownership).
Day 3–4: Auto-transcribe with custom dictionary + PII redaction; translate where needed (isiZulu→English).
Day 5–7: Import or build UTAUT2 + trust codebook; run AI-assisted coding and review low-confidence items.
Day 8–10: Run thematic frequency and cross-segment comparisons (age, internet skills, BP level); produce visualizations (co-occurrence network, hierarchical codes).
Day 11–12: Draft a 1-page exec brief with representative quotes via AI chat and export slides.
Day 13–14: Run a small AI-avatar follow-up (n=20) for targeted probes (airtime, hybrid preferences) and append incremental analysis.

Limitations & ethics note

The JMIR study asked about willingness more than observed behavior; responses may not equal real-world uptake. Use pilot metrics (actual booking, completion rates) to validate stated intent.

Ethics: clinical context, findings are research-focused and non-diagnostic. When working with sensitive health transcripts, maintain consent, deidentification, and IRB oversight.

Conclusion, next steps

If your team needs to convert interviews like those in the JMIR Soweto study into prioritized, segment-aware recommendations, Evidano shortens the timeline from months to days while preserving qualitative rigor and auditability.

Run the 2‑week pilot above and compare uptake across segments to design hybrid telemedicine workflows that respect trust and access constraints.
Start a free pilot or request a demo at www.evidano.com; we can ingest transcripts, run UTAUT2-aligned analyses, and produce stakeholder-ready visuals in under two weeks.

Keep reading

Browse all articles