From Delegation to Measurable Safety: A Narrative Review and Core Outcome Framework for Evaluating Physician Assistant Integration in German Ambulatory Care

Physician assistants (PAs) are increasingly considered for physician-led ambulatory care teams in Germany, but activity, satisfaction, or direct labor cost alone cannot establish safe integration. This structured narrative review searched PubMed/MEDLINE in seven focused PA-specific blocks and supplemented the search with two documented Google Scholar sensitivity searches. After deduplication, all 1,508 unique records were screened for evidence concerning safety, diagnostic processes, supervision, role transparency, implementation, workload, and German transferability. Sources used in the synthesis were appraised for design, setting, case mix, supervision ascertainment, outcome validity, and directness. The evidence supports feasibility and positive patient or physician experience in selected, supervised, and often protocolized pathways, but it remains heterogeneous and does not establish equivalence across undifferentiated ambulatory case mix. Case complexity, physician supervision time, escalation performance, role understanding, and downstream utilization are frequently unmeasured. An evidence-informed framework is therefore proposed with nine domains: patient and diagnostic safety, diagnostic process quality, escalation and red-flag handling, supervision burden and hidden physician work, role transparency and patient understanding, patient experience, workflow and productivity, economic and resource impact, and implementation, governance, team effects, and sustainability. Tier 1 contains mandatory safety and implementation indicators, whereas Tier 2 contains extended service and economic outcomes. Every report should stratify results by setting, diagnostic stage, case complexity, supervision model, and PA experience. The framework is an author-proposed starting structure, not a validated core outcome set. German multi-stakeholder consensus, prospective feasibility testing, and validation are required before it can support benchmarking or policy thresholds.