lab-trend

## Instructions Use this skill when the user asks for a trend, history, or progression of specific lab values (e.g. "Show me Eleanor's last 3 A1Cs", "George's eGFR trend", "How has Tomás's lipid panel changed?"). ### 1. Resolve the patient `epic_search_patients_by_name` → `epic_get_patient` if needed. ### 2. Pull observations for each lab code For each lab code the user named, call `epic_list_observations(patient_id, code: '<code>', limit: 10)`. Common codes in our fixtures: `A1C`, `BP`, `LDL`, `EGFR`, `TSH`, `FT4`, `HR`, `TEMP`, `SPO2`, `PEF`. ### 3. Render side-by-side For one lab: ```markdown # George Achterberg (PAT-100006) — HbA1c trend | Date | HbA1c | Interpretation | |---|---|---| | 10 Mar 2026 | 7.8% | Above goal | ``` For multiple labs: ```markdown # George Achterberg (PAT-100006) — Lab trends ## HbA1c | Date | Value | Interpretation | |---|---|---| | 10 Mar 2026 | 7.8% | Above goal | ## eGFR | Date | Value | Interpretation | |---|---|---| | 10 Mar 2026 | 42 mL/min/1.73m² | CKD stage 3 | ## Reading A1C is drifting up against an eGFR that puts him in CKD-3. Consider the SGLT-2 indication for renal protection (he's already on empagliflozin) and whether adding a GLP-1 is appropriate. Recheck both in 3 months. ``` ### Constraints - Quote values + units verbatim. - Order newest-first by `effective_at`. - If the patient has no readings for the requested code, say so directly ("No HbA1c on file in the last 12 months."). - The "Reading" section is clinical context, not a recommendation — keep it short and frame as considerations.