OpenEvidence Quality and Adoption Audit
Overview
Create a human-centered review loop without claiming access to private product telemetry. Keep inputs minimal, separate observed facts from assumptions, and leave consequential decisions with the named accountable owner.
Prerequisites
- A clearly bounded workflow, accountable clinical owner, and organizational policy
- Current first-party OpenEvidence documentation and applicable institution agreements
- Synthetic or properly authorized minimum-necessary data
Tool Discipline
Use Read, Glob, and Grep to inspect supplied policies, plans, and evidence. Use WebFetch only for current first-party OpenEvidence documentation. Use Write or Edit only when the user requests a named deliverable with an approved destination. Never expose credentials, PHI, recordings, or unrestricted environment output.
Current Contract
- No public observability API or standard metrics export is documented.
- Measurements must come from authorized account/institution records and redacted quality samples.
- Volume and speed are insufficient without citation, applicability, and safety review.
Authentication
Use only the official OpenEvidence web/mobile sign-in or an institution-approved access path. Do not invent API keys, OAuth clients, SDK credentials, service accounts, or private endpoints. Never ask a user to reveal a password, session token, cookie, or recovery code.
Instructions
- Define the monitored workflow, cohort, period, data authority, clinical owner, and escalation thresholds.
- Select measures for completion, citation traceability, unsupported claims, reviewer overrides, time burden, incidents, and training gaps.
- Sample the minimum authorized records and de-identify evidence used outside the care record.
- Have qualified reviewers score outputs with a stable rubric and record inter-reviewer disagreement.
- Trend results without inferring patient outcomes or vendor-wide performance from a local sample.
- Publish findings, limitations, actions, owners, thresholds, and the next review date.
Approval Boundaries
Do not create or share accounts; change access, roles, agreements, consent, retention, or security settings; enter PHI; record a conversation; copy content into another system; contact a patient; make a diagnosis or treatment decision; submit billing; transmit a support packet; run a production pilot; or represent vendor capabilities without explicit approval from the accountable owner. A qualified professional remains responsible for clinical decisions.
Output
Return scope, current first-party evidence and date, data classification, workflow or findings, citations reviewed, assumptions rejected, clinical and governance owners, approval state, unresolved risk, and the exact next action. Redact patient and credential data.
Error Handling
| Condition | Response |
|---|---|
| Telemetry unavailable | Use an approved sample or survey and label coverage. |
| Metric hides harm | Add safety and override measures before reporting success. |
| Sample contains PHI | Keep it in the governed system or stop the audit export. |
Examples
This compact example shows the minimum reviewable handoff; adapt fields to the approved workflow without adding sensitive data.
Input:
period=30d; workflow=Ask; sample=25 de-identified reviews
Expected handoff:
traceability=measured; overrides=recorded; incidents=1; actions=3