Validation of Bone Metastases Coding in an Oncology Electronic Medical Records Database and a Commercial Claims Database

Abstract

s / Annals of Epidemiology 24 (2014) 682e702 684 Methods: Using SEER 18 cancer incidence rates and county-level %rurality, Two-way ANCOVA was performed on data stratified at the State and Rural Urban Continuum Code (RUCC) levels and adjusted for obesity, smoking, alcohol use, inactivity, and race. Results: RUCC and %rurality were positively correlated (0.63; p 50 mg/mL, receipt of scintigraphy, or receipt of hospital-administered antiresportive therapy. Chart review was conducted by a research nurse blinded to patient metastasis status. Results: Of 212 patients, 113 (53%) had evidence of metastases according to the algorithms and 43 (20%) had metastases according to chart review. The positive predictive value (PPV) of all algorithms was low. The highest performing algorithm to detect any metastases was a PSA value >50 mg/mL and receipt of scintigraphy, in the absence of antiresportive therapy, with a PPV of 28% (95% CI 14%-47%). Conclusion: In their present form, algorithms based on PSA, antiresorptive treatment, or scintigraphy were not suitable for identification of additional cases of metastases. Further investigation of alternate components, cutoffs, or timing is needed. Additionally chart reviews may not represent a “gold standard” for validation of metastases, a condition that can be asymptomatic and therefore prone to under or delayed registration. P08. Prostate-Specific Antigen Testing and Shared Decision Making for Prostate Cancer Screening Jun Li MD, PhD, Zahava Berkowitz, Thomas Richards, Ingrid Hall. CDC Purposes: To describe the prevalence of prostate specific antigen (PSA) testing from 2001-2010 and assess associations between shared decisionmaking (SDM) on PSA testing. Methods: We estimated prevalence of PSA testing using data from the 2000, 2003, 2005, 2008, and 2010 National Health Interview Surveys. Changes in the prevalence by age group, race/ethnicity, and year were assessed with logistic regression models and general linear contrasts of the predicted margins. Bivariate analyses were conducted to assess associations between demographic and health-related characteristics and PSA testing. Results: Men aged 75years had a small but significant decrease in PSA testing from 2008 to 2010 (50.8% vs. 43.9%, P < 0.05). However, in 2010, they continued to have a significantly higher prevalence of PSA testing (43.9%) than men aged 50-74 (38.4%), and men aged 40-49 (9.7%). Menwho had prescreening discussions about the advantages and disadvantages of PSA testing (54.9%) and about disadvantages alone (44.1%) were more likely to receive PSA testing than men who had no discussion (17.7%), respectively (both P <0.05). Conclusions: Men aged 75 years continued to receive more PSA testing than youngermen. Pre-screening discussionswere positively associatedwith PSA testing. Additional research is needed on how SDM is being conducted in clinical practice and the role played by patient’s values and preferences in decisions about PSA testing. P09. Supplemental Folic Acid in Pregnancy and Maternal Cancer Risk Jan Helge Seglem Mortensen MD, Nina Oyen Professor, PhD, Tatiana Fomina PhD, Mads Melby Professor, MD, DMSc, Stein Emil Vollset Professor, DrPH, Tone Bjorge Professor, PhD, MD. University of Bergen Background: The World Health Organization has recommended women planning pregnancy to consume a daily dose of folic acid to reduce offspring risk of neural tube defects and possibly other congenital malformations. However, there are concerns that folic acid supplementation may have a carcinogenic effect. Purpose: The present population-based study examined the later risk of maternal cancer after intake of supplemental folic acid in pregnancy. Methods: Among 430,000womenwith one ormore childbirths in the period 1999-2010, 3,800 women developed cancer. Data was retrieved from national health registries in Norway. Cox proportional hazards regression models were used to estimate hazard ratios for maternal cancer according to folic acid use prior to and during one or more pregnancies as compared to no supplement use. Results: Folic acid had no overall effect on cancer risk in women using folic acid supplementation. In exploratory sub-group analyses, two cancer types showed a positive association with folic acid intake in pregnancy, colorectal cancer (two times higher risk) in women using folic acid in two or more pregnancies, and lung cancer (three times higher risk) in non-smoking women who used folic acid in one or more pregnancies. Conclusion: Folic acid substitution before and during pregnancy had no effect on overall cancer risk. However, exploratory sub-group analyses suggested that pregnancy related folic acid use increased the risk of some cancer types. Acknowledgments: This study was supported by the Norwegian Cancer Society, and the Regional Health Authorities Western Norway.