Integrating genetic ancestry into clinical care: Accuracy, utility, and stakeholder views

How to integrate genetic ancestry into clinical care remains unsettled. We assessed (i) agreement between two commercial ancestry providers, (ii) concordance between multiple ancestry algorithms and self‑identified race/ethnicity (SIRE) recorded in the electronic health record (EHR), and (iii) participant perspectives on accuracy, clinical utility, discrimination risk, and EHR integration. Using data from 451 adults in the UCSF 3D Health Study, we harmonized ancestry outputs to five continental categories and quantified cross‑provider agreement and ancestry–SIRE concordance at ≥ 50%, ≥ 75%, and ≥ 90% thresholds; we also analyzed survey responses from 166 participants. Mean agreement between the two commercial providers was 58.4% overall (95% BCa CI: 57.4–59.4%), differing by ancestry: 54.7% European, 73.1% Asian, 57.6% Mixed American, and 76.2% African (overall p = 3.2 × 10⁻²⁰). At ≥ 90% concordance with SIRE, algorithms varied substantially: 2.6% for Ancestry 1 (11/428), 80.1% for Ancestry 2 (353/441), 88.0% for Ancestry 3 (388/441), and 89.3% for Ancestry 4 (394/441). When stratified by SIRE group, European and Asian participants showed consistently high concordance for Ancestries 2–4, while African and Mixed American subgroups showed lower and more variable concordance across all methods. Surveyed participants were largely nondistressed (only 1.2% reported distress); 51.2% agreed ancestry reports are accurate and 50.0% agreed ancestry could support more personalized care, while 52.4% anticipated discrimination. Opinions on storing ancestry in the EHR were mixed (agree: 38.6%; disagree: 30.1%; neutral: 28.9%). Findings show vendor/algorithm choice strongly shapes ancestry outputs and alignment with SIRE. Community genetics programs should avoid treating ancestry as a proxy for race, adopt vendor‑aware workflows, and implement transparent consent and safeguards before linking ancestry to the EHR.