Embedded Research & Rapid‑Learning Starter Protocol

Purpose

This starter protocol helps clinicians and operational teams run short, pragmatic embedded-research projects that generate credible local evidence and support timely practice change. It is intentionally minimal: focused on a clear problem, a small set of feasible measures, practical eligibility, ethical safeguards, and rapid dissemination to frontline adopters.

When to use this

• When a frontline team needs quicker local evidence than traditional trials provide.
• When an operational change (workflow, order set, staffing, device, or simple clinical decision rule) needs testing with real-world constraints.
• When learning must inform an imminent operational decision.

Starter Protocol Template (fill and adapt)

1. Problem statement

[Briefly describe the clinical or operational problem, current baseline performance, why it matters to patients/staff, and the decision this project will inform.]

2. Aim

SMART aim example: Reduce average ED length-of-stay for discharged low-acuity patients by 20% within 8 weeks of implementation without increasing 72-hour returns.

3. Pragmatic eligibility (keep narrow and practical)

Example inclusion: Adults (≥18) presenting to ED with triage acuity 4–5 and discharged home. Exclusion examples: admission required, language barrier preventing usual care pathway, enrollment in conflicting study.

4. Intervention (concise)

Describe exactly what is changing in routine care (order set, workflow script, checklist, EHR nudge). Include who performs it and when.

5. Outcomes (primary, balancing, secondary — keep short)

• Primary: One actionable measure (e.g., ED LOS in minutes for eligible patients).
• Balancing: Safety or access (e.g., 72-hour return visits, unplanned admissions within 7 days).
• Process: Fidelity metric (e.g., percent of eligible patients with the order set used).

6. Data collection plan (practical and minimal)

List data sources, variables, responsible person, and frequency. Prefer existing operational data (EHR fields, administrative timestamps). Limit to variables needed for measures and essential covariates.

7. Analysis plan (practical)

Prefer simple, robust approaches you can run quickly:

• Primary: Run chart or control chart of the primary metric by week (or shift) comparing pre-implementation baseline to post-implementation. Consider interrupted time series for longer data.
• Report medians/IQRs or means with 95% CIs as appropriate. Use simple regression adjusting for key covariates if needed.
• Pre-specify decision rules for stopping, adapting, or spreading (e.g., signal of improvement in run chart and no adverse signal in balancing measure for 2 consecutive weeks).

8. Sample size / duration guidance

Embedded pragmatic tests usually use time-based or convenience sampling. Choose a short pilot window (4–12 weeks) sufficient to observe baseline variability and detect meaningful operational change. If you need a formal sample size, consult a statistician—but favor feasibility and iterative learning for early pilots.

9. Consent & ethics considerations

Most low-risk pragmatic tests qualify for simplified consent, opt-out, or waiver of consent if they use standard of care changes and pose minimal risk. Prepare a short justification for the IRB that explains:

• Minimal incremental risk beyond routine care.
• Use of existing data where possible and plans to protect privacy.
• Operational necessity and benefit to patients/staff.

Example brief script for verbal opt-out (if used): "We’re testing a small change in how we [describe change] that aims to [expected benefit]. This is part of unit improvement—your care is not expected to change in risk. You may decline and receive usual care."

10. Governance & roles

• Project lead (clinician): accountable for clinical decisions.
• Operational lead (manager): coordinates workflow and training.
• Data lead: builds queries, produces weekly run charts.
• Ethics/Privacy contact: prepares IRB or waiver request.

11. Rapid-cycle dissemination & adoption

1. Weekly one-page run chart with brief interpretation sent to frontline huddle and stakeholders.
2. At 2–4 week milestones, convene short huddle to decide: continue, adapt (PDSA), stop, or scale.
3. When criteria met for spread: create an implementation pack (one-page summary, one-sentence training script, order set links, measurement dashboard).

12. Implementation checklist (pre-launch)

• Confirm eligibility logic and data query with data lead.
• Train frontline staff with a 5-minute script and job aids.
• Prepare IRB/waiver submission or documentation of waiver criteria.
• Schedule weekly measurement and huddle cadence.
• Define stop/adapt/spread decision rules.

Appendix A — Sample timeline (8 weeks)

Week 0: finalize protocol, IRB contact, data queries. Week 1: staff training and go-live. Weeks 2–7: collect data, weekly run-charts, huddles. Week 8: formal review and decision about adaptation or spread.

Appendix B — Quick reproducible analysis snippet (conceptual)

Produce weekly aggregates, plot run chart, annotate implementation date, assess for sustained shift (e.g., 6 consecutive points above baseline median). Share plot and brief interpretation.

Customize this template to your local context. Prioritize feasibility, protection of patients, and rapid learning that informs operational decisions.