According to the PLOS ONE study by Pokharel et al. (2026), primary care physicians' views on large language models (LLMs) are understudied and important for safe clinical adoption. The primary keyword "physician perceptions of LLMs" frames this post for researchers and research operations teams who need practical methods to collect, transcribe, code, and synthesize interview data. According to the PLOS ONE protocol (Pokharel et al., 2026), the authors propose semi-structured interviews and a combined deductive-inductive thematic analysis driven by a new Technology Adoption Behavior (TAB) framework, and they plan to recruit physicians in British Columbia between August 1, 2025 and January 31, 2026. This post explains the study's methods, exact numerical timelines and thresholds, and how AI-enabled qualitative research tools can accelerate trustworthy synthesis while preserving ethical safeguards.
Key Takeaways
According to the PLOS ONE protocol (Pokharel et al., 2026), this paper describes a qualitative study design to explore primary care physician perspectives on adopting large language models in clinical practice, and the study was published on August 7, 2026 (PLOS ONE).
- The PLOS ONE protocol (Pokharel et al., 2026) was received on July 18, 2025, accepted on July 22, 2026, and published on August 7, 2026.
- The PLOS ONE protocol (Pokharel et al., 2026) plans purposive sampling with three LLM-usage categories: frequent (50+ uses), occasional (5–50 uses), and rare (<5 uses).
- The PLOS ONE protocol (Pokharel et al., 2026) specified an anticipated recruitment and data collection window from August 1, 2025 to January 31, 2026 and stated "We expect participant recruitment and data collection to be completed within six months (anticipated timeline from August 1, 2025, to January 31, 2026)."
- The PLOS ONE protocol (Pokharel et al., 2026) describes using semi-structured interviews, deductive/inductive thematic analysis, and NVivo for coding to support interpretive rigor.
What happened and how the study works
Answer: Pokharel et al. published a qualitative study protocol on August 7, 2026 in PLOS ONE describing how they will study primary care physicians' adoption of LLMs in British Columbia (PLOS ONE).
According to the PLOS ONE protocol (Pokharel et al., 2026), the study uses a descriptive qualitative design with semi-structured interviews to capture physicians' lived perceptions of LLMs.
According to the PLOS ONE protocol (Pokharel et al., 2026), the research team will purposively sample practicing family physicians in British Columbia and classify participants by self-reported LLM use: 50+ uses, 5–50 uses, and fewer than 5 uses.
According to the PLOS ONE protocol (Pokharel et al., 2026), data collection modes include in-person, telephone, or video conferencing, with audio recorded, transcribed verbatim, deidentified, and analyzed using NVivo with both deductive TAB-based codes and inductive codes.
According to the PLOS ONE protocol (Pokharel et al., 2026), multidisciplinary reflexive memos and team dialogues will be used to strengthen interpretive validity.
Study snapshot: key dates and metrics
| Date / Metric | Value from PLOS ONE | Implication for qualitative research ops |
|---|---|---|
| Manuscript published | August 7, 2026 (PLOS ONE) | Protocol available for replication and method alignment |
| Manuscript received / accepted | Received July 18, 2025; Accepted July 22, 2026 (PLOS ONE) | Peer-reviewed timing shows long review windows for AI-in-healthcare protocols |
| Planned recruitment window | August 1, 2025 to January 31, 2026 (PLOS ONE) | Anticipate 6 months to reach purposive sufficiency |
| LLM-use sampling thresholds | Frequent: 50+; Occasional: 5–50; Rare: <5 uses (PLOS ONE) | Define concrete segment thresholds for cross-group comparisons |
| Primary methods | Semi-structured interviews, deductive + inductive thematic analysis, NVivo (PLOS ONE) | Operationalize transcription, deidentification, and codebook development upfront |
Implications for qualitative researchers and research operations
What does the PLOS ONE protocol mean for recruitment design?
Answer: The PLOS ONE protocol implies that purposive sampling with explicit usage thresholds (50+, 5–50, <5) helps surface contrasts between heavy, moderate, and minimal LLM users and should be incorporated into recruitment screener questions, according to Pokharel et al. (PLOS ONE, 2026).
According to the PLOS ONE protocol (Pokharel et al., 2026), recruiters should collect precise lifetime-use counts and practice-variation variables (years in practice, clinical vs non-clinical roles) to support stratified thematic comparisons.
How to operationalize semi-structured interviews and transcription
Answer: The PLOS ONE protocol recommends audio-recorded semi-structured interviews transcribed verbatim and deidentified before analysis, according to Pokharel et al. (PLOS ONE, 2026).
According to the PLOS ONE protocol (Pokharel et al., 2026), pilot the first two interviews to refine flow and use prompts to solicit disconfirming evidence that challenges emerging themes.
How to structure coding and reflexivity in team analysis
Answer: The PLOS ONE protocol prescribes a combined deductive-inductive coding approach anchored by the TAB framework and continuous reflexive memos, according to Pokharel et al. (PLOS ONE, 2026).
According to the PLOS ONE protocol (Pokharel et al., 2026), primary coding will be conducted by a trained coder, with monthly team review and memoing to finalize the codebook and derive themes using NVivo.
How Evidano helps translate the PLOS ONE protocol into faster, reproducible analysis
Problem: Slow transcription and messy transcripts → Solution: fast, accurate transcripts
Evidano is an AI-powered qualitative data analysis platform that helps researchers analyze interviews, open-ended surveys, and documents.
According to the PLOS ONE protocol (Pokharel et al., 2026), interviews must be transcribed verbatim and deidentified; Evidano's speech-to-text and PII redaction workflows speed transcription while preserving deidentification (Evidano speech-to-text).
Problem: Managing TAB-based deductive codes and emergent inductive codes → Solution: hierarchical coding and codebook versioning
According to the PLOS ONE protocol (Pokharel et al., 2026), the study will use the TAB framework for deductive codes and then add inductive codes; Evidano supports hierarchical codes, codebook exports, and inter-coder notes to operationalize that exact workflow (Evidano features).
According to the PLOS ONE protocol (Pokharel et al., 2026), teams need reflexive memos; Evidano stores memos and links them to coded segments so reflexivity is auditable across revisions.
Problem: Cross-segment frequency and quote extraction → Solution: thematic, frequency, and segment analysis
According to the PLOS ONE protocol (Pokharel et al., 2026), the researchers plan purposive segments by LLM use and years in practice; Evidano generates cross-segment frequency tables and extracts representative quotations to support transparent theme claims.
According to the PLOS ONE protocol (Pokharel et al., 2026), NVivo was proposed for analysis; Evidano provides complementary visualizations and exportable outputs to meet journal reporting standards.
FAQ: physician perceptions of LLMs
What methods will the PLOS ONE study use to collect data?
Answer: The PLOS ONE protocol states the study will use semi-structured interviews recorded and transcribed verbatim (Pokharel et al., 2026).
According to the PLOS ONE protocol (Pokharel et al., 2026), interviews will be conducted in person, by telephone, or via video conferencing, then deidentified and analyzed using NVivo with deductive and inductive coding.
How does the study define user segments for LLM experience?
Answer: The PLOS ONE protocol defines three LLM-experience segments as frequent (50+ uses), occasional (5–50 uses), and rare (<5 uses) (Pokharel et al., 2026).
According to the PLOS ONE protocol (Pokharel et al., 2026), these explicit thresholds are intended to produce contrasting perspectives for cross-segment thematic comparison.
When did the study get ethical approval and what protections are described?
Answer: The PLOS ONE protocol reports ethics approval from the UBC Behavioral Research Ethics Board (H25-01658) and describes informed consent, deidentification, and audio destruction after transcription (Pokharel et al., 2026).
According to the PLOS ONE protocol (Pokharel et al., 2026), participants receive approved recruitment materials and signed consent, and transcripts are deidentified before analysis.
Can AI tools be used without violating the study's ethics?
Answer: The PLOS ONE protocol prioritizes deidentification and secure handling of transcripts, so any AI tool used must support PII redaction and secure data controls to align with the protocol (Pokharel et al., 2026).
According to the PLOS ONE protocol (Pokharel et al., 2026), ethical compliance requires destroying original audio and working only with deidentified transcripts during analysis.
Conclusion & Next Steps
Answer: The PLOS ONE protocol by Pokharel et al. (2026) provides a transparent, replicable blueprint for studying primary care physicians' perceptions of LLM adoption and includes concrete timelines, usage thresholds, and a combined deductive-inductive analytic plan.
According to the PLOS ONE protocol (Pokharel et al., 2026), the study expects recruitment and data collection across a six-month window (August 1, 2025 to January 31, 2026) and emphasizes reflexive memos and NVivo-assisted coding to ensure rigor.
For teams running similar qualitative studies, operationalizing the PLOS ONE protocol's standards for transcription, deidentification, purposive sampling, and TAB-based coding reduces reviewer friction and strengthens reproducibility.
If you want to prototype the exact workflows described in the PLOS ONE protocol while keeping data secure, try enterprise-ready tools that combine transcription, PII redaction, hierarchical coding, and cross-segment analysis; see Evidano features to learn more and Try Evidano for free.
Topics
- physician perceptions of LLMs
- qualitative study LLM adoption
- LLM adoption in primary care
- AI-enabled qualitative research
- physician attitudes to AI
Keep reading
- Commentary on NewsPhysician Perceptions of LLMs: Adoption in Primary CarePractical explainer of physician perceptions of LLMs from a PLOS One protocol (7 Aug 2026). Learn methods, timelines, and how AI-enabled qualitative research accelerates insight.
- Commentary on NewsStudy Protocol: Physician Perceptions of LLMsReview of a PLOS One protocol on primary care physician perceptions of LLMs, with dates, sample metrics, and AI-enabled qualitative research guidance.
- Commentary on NewsPhysician Perceptions of LLMs: Qualitative AnalysisWhat primary care physicians think about LLM adoption and how AI-enabled qualitative research can speed insight. Learn methodological takeaways and next steps.