209 Background: The success of quality improvement (QI) efforts is usually assessed initially within a year of implementation but performance may later decline. We sought to determine the longer-term effectiveness of practice improvement interventions and the characteristics of interventions associated with persistent improvement. Methods: The Durham VA Hem-Onc fellowship program affiliated with Duke University participated in QOPI twice yearly in 2007 to 2010 and Spring (Sp) 2012. Four interventions were implemented focused on treatment planning (TP; Sp07), treatment summary (TS; Fa07), consent (C; Sp07), and aprepitant use (A; Sp09). Effectiveness was determined by reassessment of 4, 3, 1, and 1 measures, respectively. The TP, TS, and C interventions were policy-dictated use of templated documentation or written consent, and, for TP and C, required conformance prior to chemotherapy administration; TS is not linked to a process. The A intervention modified order set templates to include A for highly emetogenic regimens. For each measure, we compared conformance pre-intervention with post-intervention conformance and, for quality indicators not targeted for improvement, the baseline with each subsequent assessment using chi-square. Results: Considering only measures in the core or symptom management modules with unchanged or similar definitions throughout the study period, 13 of 28 measures had conformance > 85% in the baseline 2007 data. TP conformance increased from 64% to 88, 94, 92, and 96% in subsequent years (p < 0.01). TS was 45% at baseline, 43% (p = NS) in the first post-intervention year, decreased to 24% for two years (p < 0.01) and then increased to 38% (p = NS). C increased from 64% to 97, 91, 94, and 100% (p < 0.01). Conformance with A increased from 9% to 78 and 86% (p < 0.01). Among measures not targeted by intervention, smoking assessment and counseling increased in later years. Conclusions: In this single practice experience, QI improvement interventions integrated into physician processes resulted in substantial and persistent improvement while no improvement was observed as a result of policy-required use of standard documentation alone.