How to write Treatment Algorithms

Want to write clear and clinically actionable treatment algorithms that support oncologists and trainees in making informed decisions at the point of care? Read on!

Core Principles

1. Be Specific

• Clearly define the topic, cancer type, subtype and/or stage in the title and description

  • Example: "Stage IV Non-Small Cell Lung Cancer (EGFR-mutant)"

• Avoid overly broad algorithms that dilute clinical utility

• Include key stratifiers (e.g., biomarkers, performance status) where appropriate

2. Optimize for Clinical Decision Making

• Structure algorithms to reflect real-world workflow or a detailed part of the workflow

  • Initial diagnosis → staging → biomarker testing → first-line → subsequent lines

  • PET-CT → TNM classification  → Mediastinoscopy / EBUS → ECOG  → Tumor board discussion (oncology, radiology, pathology, surgery)

• Prioritize decisions and action steps

3. Think Like a Trainee

• Assume the reader is synthesizing large volumes of information quickly and may not have your depth of knowledge in a certain disease space

• Where relevant, include:

  • Brief context for each decision point

  • Key rationale (1-2 lines max)

  • Common pitfalls or exceptions

4. Incorporate Key Clinical Pathways

Include relevant sub-pathways when they materially affect management:

• Pathology workflows (e.g., biopsy adequacy, molecular testing)

• Biomarker-driven treatment pathways

• Surgical vs. systemic vs. radiation decision points

• Supportive care considerations when relevant

Content Best Practices

Titles and Structure

• Use specific, searchable titles (cancer + stage + key biomarker)

References (SEO + Credibility)

• Include up-to-date, high-quality references on specific nodes where relevant:

  • NCCN, ASCO, ESMO guidelines

  • Landmark trials on published on NEJM, Lancet, PubMed

• Improves discoverability in ChatGPT, Claude, Google and more

• No need to add references for every step; this is where your clinical judgment and experience is important

Visual Design

• Add images and diagrams to enhance engagement:

  • PFS curves from pivotal trials

  • Tables for regimen comparisons and risk assessments

  • Pathology or imaging examples when relevant

• Keep visuals simple and clinically focused

Algorithm Construction

• Use clear, concise nodes:

  • Node title: “If EGFR mutation” → connects to child node: "osimertinib"

  • Node title: "If no actionable biomarker and PD-L1 ≥50%" → connects to child Node: "pembrolizumab monotherapy"

• Avoid ambiguity; specify thresholds and criteria

• Limit node title length to a max of 180 characters to maintain readability

Here's an example from Dr. Paolo Tarantino at Dana-Farber Cancer Institute:

Early HER2 positive Breast Cancer algorithm by Paolo Tarantino, MD

Check out more algorithms on Open Medicine. 🙂