Practical Guide: Qualitative Analysis of Deprescribing

Qualitative analysis of deprescribing is critical for understanding why antithrombotic and other medications continue at end of life. A realist synthesis published 25 Aug 2025 screened 17, 036 citations (230 articles in Stage 1) to surface clinician, patient and system-level mechanisms that shape deprescribing decisions. This post translates those findings into practical qualitative research workflows you can run with Evidano (www.evidano.com) to accelerate thematic synthesis, segment comparisons, and stakeholder-ready outputs.

Fast take + source

In brief: The PLOS Medicine realist synthesis (published 25 Aug 2025) explored how shared decision-making affects deprescribing of antithrombotic therapy for people with cancer in their last phase of life. The authors report a large screening (main search = 17, 036 citations; Stage 1 screened 230 articles) that informed a set of 15 initial programme theories about clinician behaviour, patient readiness, hope, communication and system inertia. Read the original study: journals.plos.org/plosmedicine/article?id=10.1371%2Fjournal.pmed.1004663.

Why this matters for qualitative teams: the study maps mechanisms (e.g., clinician equipoise, patient hope, specialist boundaries) that are coded qualitatively but benefit from reproducible, cross-segment analysis.
Quick payoff: capture coded themes tied to role, timing, and medication class; quantify co-occurrence (e.g., 'hope' + 'oncologist deference') and export stakeholder summaries in hours, not weeks.

Findings snapshot

Item	Value	Note / implication	Source
Publication date	25 Aug 2025	Realist synthesis in PLOS Medicine	journals.plos.org/plosmedicine/article?id=10.1371%2Fjournal.pmed.1004663
Main search (post-duplication)	17, 036 citations	Large candidate pool; broad literature sweep	Study methods
Stage 1 screened	230 articles	Narrowed evidence to focused papers for programme theories	Study methods
Initial Programme Theories	15 IPTs	Mechanisms include clinician equipoise, patient hope, specialist deference	Supporting information

What the study did and why it matters

The authors used realist synthesis to generate causal programme theories (IPTs) explaining when deprescribing is likely to occur or be resisted. Key mechanisms include:

Clinician clinical equipoise and comfort with prognostic uncertainty (IPT1)
Recognition of transition to end of life and timing of conversations (IPT2, IPT8)
Meaning of medications for patients, hope and identity (IPT5)
Fear of offending original prescribers and specialist boundaries (IPT13)
Communication framing: specialized language and empathic delivery shape uptake (IPT10, IPT9)

For qualitative researchers and policy teams, these are testable nodes, each IPT can be operationalized as a code, a code co-occurrence hypothesis, and segment comparison (e.g., oncologist vs palliative clinician).

What this means for researchers: qualitative analysis of deprescribing

UX / Qualitative teams

Turn IPTs into a codebook: map IPT labels (e.g., 'hope', 'equipoise', 'deference') to in-vivo quotes and clinician notes so you can quantify prevalence and co-occurrence across transcripts.

Compare readiness over time: code for 'patient readiness' across longitudinal interviews to test IPT8 (readiness increases).

Policy & Clinical Governance

Use coded evidence to identify system-level barriers (e.g., specialist ownership of prescriptions) and design targeted interventions (training, PIMs lists) informed by IPT13 and IPT15.

Produce succinct evidence briefs linking theme frequency to recommended policy levers.

Mixed-methods teams

Triangulate qualitative IPT-derived themes with prescribing datasets (e.g., PIMs flagged prescriptions) to measure downstream outcomes of deprescribing conversations.

Do more, faster with Evidano (mapped to this use case)

Problem: Large, messy literature and transcripts

Solution: Bulk ingest PDFs, interview transcripts, and meeting notes into Evidano; run automated thematic extraction to surface candidate codes aligned to the 15 IPTs. Evidano preserves your codebook and allows iterative refinement.

Problem: Multilingual / clinical terms and jargon

Solution: Use Evidano transcription and translation with custom dictionaries so clinical terms (e.g., 'antithrombotic', 'PIMs') are transcribed consistently and mapped to codes.

Problem: Reproducing coder decisions and cross-segment counts

Solution: AI-assisted coding with hierarchical code→subcode visualizations, co-occurrence networks, and frequency tables, exportable for policy briefs and stakeholder decks.

Problem: Follow-up data and stakeholder engagement

Solution: Run AI avatar interviews to collect focused follow-ups (e.g., clinician attitudes after a training) and feed results directly back into the analysis corpus.

Security & compliance

Solution: Data is encrypted and not used to train third-party models, an important safeguard for sensitive EOL conversations.

Checklist: 7-step reproducible workflow

Run-book to operationalize the PLOS realist synthesis IPTs into a study you can replicate:

1) Collect: gather transcripts, meeting notes, PIMs lists and relevant articles (start with the PLOS synthesis).
2) Ingest: upload docs to Evidano and apply a project-specific custom dictionary for clinical terms.
3) Seed codebook: import the 15 IPT labels as top-level codes; add expected subcodes (e.g., 'hope: cure', 'hope: prolongation').
4) Auto-code + review: run AI-assisted coding, then validate 10–15% of excerpts manually to set reliability checks.
5) Analyze: run thematic frequency, co-occurrence networks, and cross-segment comparisons (role, phase of illness).
6) Visualize: export hierarchical code maps and clickable quote reports for clinicians and policy makers.
7) Iterate & intervene: deploy targeted interventions (training, specialist huddles), collect post-intervention interviews via AI avatars, and re-run the pipeline.

Ethics note

This content is research-focused and non-diagnostic. When working with end-of-life interviews, follow institutional consent, anonymization and data governance protocols.

Conclusion & next steps

The 25 Aug 2025 realist synthesis gives qualitative teams a clear set of mechanisms to code and test. Use an AI-enabled platform to scale code application, quantify patterns (e.g., how often 'hope' co-occurs with 'specialist deference'), and produce stakeholder-ready outputs faster.

Ready to turn IPTs into reproducible insights? Start a pilot: upload a small batch of transcripts to Evidano and run the IPT codebook across your corpus (www.evidano.com).

Keep reading

Browse all articles