S574 I. J. Radiation Oncology d Biology d Physics Volume 78, Number 3, Supplement, 2010

Materials/Methods: Non-metastatic prostate cancer cases, diagnosed 1996-2003, who received a complete course of radiationtherapy (RT) in first 6 months post-diagnosis, were identified in the SEER-Medicare data file using CPT and SEER treatment codes.Five types of RT had sufficient number of patients for analysis: brachytherapy alone (B); 3DCRT; IMRT; 3DCRT plus B; andIMRT plus B. Logistic regression modeling assessed use of non-orchiectomy ADT 6 months after last radiation dose by RTtype, using 3DCRT as the reference, adjusting for age, race, grade, and co-morbidities.

Results: Over 90% of RT patients received at least 1 dose of ADT during RT course; only those who received ADT during RT wereincluded, and those receiving ADT in 6 months post-treatment were excluded, resulting in a sample size of 7,089. 3DCRT was themost common RT type (n = 3,156), and use increased with age. Non-white patients were more likely to receive 3DCRT; whitepatients were more likely to receive B; use of IMRT did not vary by race. Poorly differentiated tumors were more likely to receive3DCRT plus B. Compared to 3DCRT, the odds of receiving ADT 6 months after RT completion were lowest for B (OR 0.27, 95%CI 0.21-0.37), followed by IMRT (OR 0.46, 95% CI 0.32-0.69). Poor grade was the only significant predictor for use of ADT 6months after RT completion (OR 2.37; 95% CI 1.99-2.80).

Conclusions: Medicare data allowed RT treatment differentiation in the SEER database. Logistic regression, accounting for adverserisk factors, found treatment with IMRT or brachytherapy for patients with non-metastatic prostate cancer resulted in improvedoutcome as compared to 3DCRT as measured by hormone use more than 6 months post RT.

Author Disclosure: A.A. Konski, None; C. Yee, None; K. Schwartz, None.

2829 Error Reduction Model in Radiation Oncology

J. W. Sweet1, E. B. Kline2

1Urologic Associates of S.E. PA, Bala Cynwyd, PA, 2Healthtronics, Inc., Austin, TX

Purpose/Objective(s): Medical errors related to radiation oncology treatments are receiving increased public and governmentscrutiny. This work examines the results of implementing a software-based program to reduce the overall number of incidents,adverse events, and associated regulatory infractions. Findings are compared with error rates found at other institutions.

Materials/Methods: A software-based medical error reduction/compliance program was implemented at radiation oncology cen-ters A and B over 2 years and 1 year, respectively. The application was used to identify, categorize, evaluate, and correct the fol-lowing: preventable systems-related errors, violations of regulatory requirements, failures to meet recommended practicestandards, and billing mistakes. Errors were classified based on type, category, attribute, and significance. Reports were launchedfollowing a root-cause analysis. Errors were automatically routed for designated review. Template procedures provided custombenchmarking for center education and performance measurements.

Results: A total of 1,122 and 196 treatment-related errors were identified at centers A and B, respectively. For each center, theidentification of pre-treatment errors compared as follows: 51% vs. 9% for record and verify system, 14% vs. 3% for computertreatment planning, 9% vs. 38% for CT simulations, and 4% vs. 6% for portal images. In regards to post-treatment errors, centersA and B compared as follows: 40% vs. 11% for patient documentation/notes, 26% vs. 48% for billing, 15% vs. 35% for portalimages, and 12% vs. 1.8% for patient treatment delivery. Center A identified 24 errors that affected patient treatments and requiredsome level of clinical intervention. Treatment delivery errors at centers A and B showed error rates of 3.2% vs. 1.2% per patient,0.11% vs. 0.03% per fraction, and 0.0012% vs. 0.0042% per field. Results shown by Frass, et al., indicate error rates of 0.44% perfraction and 0.13% per field. Results from Grace, et al., show error rates of 1.87% per patient and 0.29% per field. Results discussedby French show error rates of 0.32% per fraction and 0.037% per field. Looking at error rates in the entire treatment process (patientregistration thru completion) at centers A and B, the combined pre- and post-treatment errors in this work showed an error rate perpatient of 27.33% vs. 38.5%, per fraction of 0.92% vs. 1.03%, and per field of 0.01% vs. 0.13%. Results from Kline, et al., showa combined error rate of 0.052% per fraction. When comparing medical event rates per fraction, center A showed 0.0022% and theUnited States Nuclear Regulatory Commission reported 0.0042%.

Conclusions: A model designed to reduce risk in the overall treatment process can be used as an effective tool in quantifying errorreduction and compliance improvement.

Author Disclosure: J.W. Sweet, None; E.B. Kline, None.

2830 Clinical Outcome of Hypofractionated Radiation Therapy for Lung, Liver, and Pancreatic Cancer

T. Xia, H. Li, J. Wang, Y. Wang, P. Li, D. Chang, J. Wang, W. Wu

Air Force General Hospital, Beijing 100142, China

Purpose/Objective(s): To analyze the characteristics of physical dose, explore new dose fractionation mode, and evaluate theefficacy and radiation side effect of patients with non-small-cell lung cancer (NSCLC), liver cancer and pancreatic carcinomatreated with hypofractional g- ray stereotactic body radiotherapy (SBRT).

Materials/Methods: A total of 154 patients with stage I and II NSCLC, liver cancer, or pancreatic carcinoma treated using SBRTfrom June, 2000 to May, 2006 were analyzed, including 43 cases of non-small-cell lung cancer, 52 cases of liver cancer and 59cases of pancreatic carcinoma. Reproducible fixation and exact tumor localization were obtained by using Vacuum bag. EnhancedCT or PET/CT simulation was conducted, and 3-dimensional treatment plans were designed. The target volumes, including GTV,CTV and PTV were delineated on CT images. It is demanded that 50% isodose line cover 100% of PTV, 60% isodose line coverover 90% of CTV, and 70% isodose line cover over 80% of GTV. Different prescription dose was delivered to the target in differentposition, with 3 to 5Gy per fraction at 50% isodose line, 5 fractions a week. The total dose was 40 to 51Gy with 10 to 17 fractions.The treatment was finished in 2 to 3 weeks. The follow-up duration time was 24 to 81 months and the median follow-up durationtime was 36 months.

Results: The beams of whole body g knife are focused, which results in several physical dose characteristics, including a high doseplateau, tightly concentric distribution of isodose lines, a small 50% isodose line coverage, and sharply decreasing of 50% isodoseline to 30%. It is suitable to give prescription dose at 50% isodose line with hypofractional mode. The biological equivalent