How to run qualitative analysis of LLM adoption

This post summarizes the PLOS One protocol and shows how AI-enabled qualitative research can accelerate and scale thematic insights on physician adoption of LLMs. PLOS One is the source for the protocol and timeline below.

• Pokharel et al., PLOS One (published August 7, 2026) plan semi-structured interviews with purposive sampling across LLM usage strata (50+, 5–50, <5 uses) and analysis using the TAB framework.
• The PLOS One protocol reports that the number of Canadians without a family doctor rose from 4.5 million in 2019 to about 6.5 million in 2023, motivating the focus on primary care (Pokharel et al., 2026).
• Recruitment in the protocol was planned between August 1, 2025 and January 31, 2026, with the team expecting results within a year after data collection begins (Pokharel et al., 2026).
• The authors note, “The study may provide timely input into relevant policies, practices, and behaviors, ” and they commit to thematic analysis using both deductive and inductive coding (Pokharel et al., 2026).
• The protocol will use NVivo for coding and a multidisciplinary reflexive team to maintain analytic rigor (Pokharel et al., 2026).

Answer: Pokharel et al. published a qualitative study protocol in PLOS One on August 7, 2026 that outlines how to study primary care physicians' perceptions of LLM adoption.

According to Pokharel et al., PLOS One (2026), the protocol uses a descriptive qualitative design with semi-structured interviews, purposive sampling in British Columbia, and the Technology Adoption Behavior (TAB) framework to guide questions and coding.

According to Pokharel et al., PLOS One (2026), interviewees will be sampled across three LLM-usage strata defined as 50+ uses, 5–50 uses, and fewer than 5 uses in clinical contexts, and interviews will be recorded, transcribed, and deidentified before analysis.

According to Pokharel et al., PLOS One (2026), the team will combine deductive codes derived from the TAB framework with inductive codes from the data, run iterative reflexive memos, and finalize themes with monthly team reviews.

Answer: The PLOS One protocol shows three practical implications for researchers designing qualitative studies of LLM adoption.

First, Pokharel et al., PLOS One (2026) demonstrate that sampling across usage strata (frequent, occasional, rare) clarifies differences between hands-on users and skeptics; researchers should predefine strata such as 50+, 5–50, and <5 uses when relevant.

Second, Pokharel et al., PLOS One (2026) use the Technology Adoption Behavior (TAB) framework to combine perceived ease of use, perceived usefulness, and task-technology fit; UX teams should adapt TAB or an equivalent theoretical scaffold to keep coding deductively aligned with implementation levers.

Third, Pokharel et al., PLOS One (2026) plan reflexive memos and multidisciplinary team reviews to improve trustworthiness; qualitative teams should schedule recurring codebook reviews and log intersubjective decisions.

Recap: The PLOS One protocol from August 7, 2026 provides a concrete blueprint for qualitative analysis of LLM adoption in primary care using semi-structured interviews, purposive strata, the TAB framework, and combined deductive and inductive coding.

Next step: replicate the protocol’s sampling strata and coding discipline while using AI-enabled tools to speed transcription, code suggestion, and cross-segment counts.

If you want to pilot the workflow described here, Try Evidano for free to transcribe, code, and generate thematic summaries that map directly to implementation questions and policy briefs.