Purpose
Evaluate a health program, policy, or intervention for who it actually reaches versus who it leaves behind—surfacing structural barriers, scoring the design against established health and racial equity frameworks, and producing ranked, fundable recommendations to make the design more equitable. Replaces vague "we should be more equitable" intentions with a concrete diagnosis and action plan.
When to Use
Invoke this skill when the user asks to:
- Assess whether a program or intervention reaches the populations with the greatest need
- Identify which communities are served versus excluded by a current design
- Build an equity justification for a grant proposal, funder, or board
- Redesign an existing program to reduce disparities or close access gaps
- Apply a racial equity or health equity lens to a policy or initiative
- Diagnose why outcomes differ across racial, geographic, income, or language groups
Inputs Required
Before starting, collect:
- Program description: What the intervention does, how it's delivered, and its stated goal (2–4 sentences)
- Intended population: Who the program is designed to serve (demographics, geography, condition)
- Delivery model: How people access it—location, channel, cost, eligibility, referral pathway
- Population context: Known disparities, demographic breakdown, or health data for the service area
- Stage: Whether this is a new design, a pilot, or an established program being reviewed
- Equity goals: Any stated equity objectives or priority populations the funder/org has named
- Available data: What outcome or utilization data exists, disaggregated by subgroup if possible
If population context or disaggregated data is missing, proceed using publicly recognized disparity patterns for the named population and flag every assumption explicitly.
Instructions
Step 1: Reach Mapping — Who Is Served vs. Left Out
Construct a two-column map of the population.
- Served well: Subgroups for whom the current design works (by race/ethnicity, language, income, geography, disability status, age, immigration status, insurance status)
- Underserved or excluded: Subgroups the design structurally misses For each excluded subgroup, name the specific design feature that excludes them (e.g., "weekday daytime hours exclude hourly workers," "online-only enrollment excludes households without broadband"). Distinguish intended exclusions (eligibility by design) from unintended exclusions (design artifacts no one chose).
Step 2: Structural Barrier Diagnosis
Assess barriers across the social determinants of health, not just the clinical touchpoint:
- Access barriers: Transportation, hours, location, cost, insurance, documentation requirements
- Communication barriers: Language, literacy level, cultural relevance, channel (assumes smartphone/email)
- Trust barriers: History of mistreatment, immigration enforcement fears, lack of representation among staff
- Economic barriers: Time off work, childcare, indirect costs, wage loss
- Upstream barriers: Housing, food security, neighborhood conditions that the program assumes away For each barrier present, rate severity (High / Medium / Low) and note which subgroup it most affects.
Step 3: Equity Framework Scoring
Score the design against three lenses, rating each High / Moderate / Low equity alignment with a one-sentence justification:
- Health Equity (WHO / social determinants): Does the design allocate effort proportional to need, or treat everyone identically regardless of disparity? Flag whether it is equality-based (same for all) or equity-based (more for those with greater need).
- Racial Equity (procedural, distributional, structural): Were affected communities involved in design (procedural)? Does benefit distribution narrow or widen racial gaps (distributional)? Does it address root causes or only symptoms (structural)?
- Targeted Universalism: Is there a universal goal with targeted strategies for the groups furthest from it—or a one-size-fits-all approach that predictably underserves the margins?
Step 4: Disparity Risk Assessment
Identify where this design is likely to widen existing gaps if implemented as-is. Name the mechanism—for example, a digital-first model adopted faster by higher-resource groups, producing a "rising tide that lifts some boats higher." Flag any inverse care law dynamics (those with greatest need getting least access).
Step 5: Ranked Redesign Recommendations
Produce 6–8 recommendations ranked by equity impact and feasibility. For each:
- The recommendation (specific design change)
- Which exclusion or barrier it closes
- Expected equity effect (which subgroup benefits, gap it narrows)
- Feasibility (Quick win / Moderate lift / Structural change) Lead with high-impact quick wins; place structural changes last with honest effort framing.
Step 6: Community Engagement & Measurement Plan
- Engagement: Name 2–3 concrete ways affected communities should inform the redesign (advisory roles, co-design sessions, compensated community reviewers)—move beyond "gather input."
- Equity metrics: Recommend 3–5 disaggregated measures to track whether gaps narrow—not just aggregate reach. Specify the subgroups to disaggregate by and the baseline-vs-target framing.
Output Format
- Structure: Six labeled sections following the steps; reach map as a two-column table; recommendations as a ranked numbered list with sub-fields; framework scoring as labeled ratings with justification
- Length: 1,000–1,400 words
- Format: Structured document—tables for the reach map and metrics, ranked list for recommendations, prose for diagnoses
- Sections: Reach Mapping → Structural Barriers → Equity Framework Scoring → Disparity Risk → Ranked Recommendations → Engagement & Measurement Plan
- Tone: Public health strategist—rigorous, evidence-grounded, candid about tradeoffs; names exclusions plainly without euphemism and without overclaiming beyond the data
Quality Checklist
Before delivering output, verify:
- [ ] Reach map names specific excluded subgroups AND the design feature causing each exclusion—not generic "underserved communities"
- [ ] Every structural barrier is rated for severity and tied to a specific affected subgroup
- [ ] All three equity frameworks are scored with a one-sentence justification each
- [ ] Disparity risk section names at least one mechanism by which the design could widen a gap
- [ ] 6–8 recommendations are ranked and each names the specific exclusion it closes and its feasibility tier
- [ ] Measurement plan specifies disaggregated metrics by subgroup—not aggregate reach numbers
- [ ] Every assumption made due to missing data is explicitly flagged
Notes
- Equity is not equality: a design that treats everyone identically will usually widen gaps for the groups starting furthest behind. The scoring should reward proportional, not uniform, allocation.
- The most common failure mode is a program that reaches its target population in aggregate while missing the highest-need subgroups within it—always disaggregate before concluding reach is adequate.
- Procedural equity (who was in the room during design) often predicts distributional equity; if affected communities had no voice in the design, flag it as a leading risk indicator even when outcome data is unavailable.
- When data is thin, base disparity reasoning on recognized patterns and label it as inferred—do not present inference as measured fact.
How to use this skill
This file is a Claude Code skill in Anthropic's standard format — markdown with YAML frontmatter at the top.
Installing in Claude Code:
~/.claude/skills/<folder-name>/SKILL.md— auto-loads in every session anywhere on your machine (recommended for skills you'll use across multiple projects)<your-project>/.claude/skills/<folder-name>/SKILL.md— auto-loads only when working in that project (recommended for project-specific skills)
Restart Claude Code or open a new session to register.
Using in Claude.ai, ChatGPT, Gemini, or other AI tools:
- Drag this file into a new chat, or copy and paste its contents
- Send your task or request as your next message — the AI will apply the skill where it fits
For Claude Code skill details: https://docs.anthropic.com/en/docs/claude-code/skills