MP44-11 CREATION AND VALIDATION OF A CLAIMS-BASED ALGORITHM FOR IDENTIFYING IN-VITRO FERTILIZATION SERVICES

Abstract

You have accessJournal of UrologyGeneral & Epidemiological Trends & Socioeconomics: Practice Patterns, Quality of Life & Shared Decision Making I1 Apr 2018MP44-11 CREATION AND VALIDATION OF A CLAIMS-BASED ALGORITHM FOR IDENTIFYING IN-VITRO FERTILIZATION SERVICES James Dupree, Zachary Levinson, Angela Kelley, Jessica Dozier, Marsha Manning, Michael Lanham, Vanessa Dalton, Helen Levy, and Richard Hirth James DupreeJames Dupree More articles by this author , Zachary LevinsonZachary Levinson More articles by this author , Angela KelleyAngela Kelley More articles by this author , Jessica DozierJessica Dozier More articles by this author , Marsha ManningMarsha Manning More articles by this author , Michael LanhamMichael Lanham More articles by this author , Vanessa DaltonVanessa Dalton More articles by this author , Helen LevyHelen Levy More articles by this author , and Richard HirthRichard Hirth More articles by this author View All Author Informationhttps://doi.org/10.1016/j.juro.2018.02.1430AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookTwitterLinked InEmail INTRODUCTION AND OBJECTIVES Infertility is a disease, and understanding practice patterns for in-vitro fertilization (IVF) is important for improving access to fertility treatments. However, studying IVF in claims data is challenging because it is a rarely covered benefit, although other services provided in conjunction with IVF are often covered. Therefore, we sought to create and validate an algorithm for identifying IVF in administrative claims data. METHODS We utilized claims data from a large university employer that began offering IVF as a covered benefit in January 2015. We used employee claims data from September 2014 to January 2017 for women 22-46 years old to evaluate multiple algorithms that combined covered services or medications to predict IVF cycles. We defined our gold standard as Current Procedural Terminology (CPT) code 58970 (follicular puncture for oocyte retrieval). We then generated sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy characteristics for each algorithm. Finally, we performed a primary chart review to validate the final algorithm. RESULTS The best performing claims-based algorithm used a combination of two or more pelvic ultrasounds (CPT 76830, 76856, or 76857) within two days of each other plus a prescription for either menotropin or ganirelix within 58 or 88 days of the most recent ultrasound, respectively. In addition, the algorithm excluded episodes of fertility preservation and intrauterine insemination. The final algorithm had high sensitivity, specificity, NPV, and accuracy for predicting IVF (Table). The PPV was moderately high, reflecting the rate of false positives and relatively low volume of IVF. Of the 51 false positives, primary chart review identified 22 (43%) as canceled IVF cycles and 18 (35%) as IVF cycles likely occurring outside the healthcare network covered by the insurance policy. CONCLUSIONS A claims-based algorithm using paired pelvic ultrasounds and hormonal medications commonly used for IVF has high sensitivity, specificity, NPV, and accuracy for predicting IVF cycles. This algorithm will be useful for researchers, policymakers, and clinicians who aim to better understand large scale IVF practice patterns using administrative claims data. © 2018FiguresReferencesRelatedDetails Volume 199Issue 4SApril 2018Page: e591-e592 Advertisement Copyright & Permissions© 2018MetricsAuthor Information James Dupree More articles by this author Zachary Levinson More articles by this author Angela Kelley More articles by this author Jessica Dozier More articles by this author Marsha Manning More articles by this author Michael Lanham More articles by this author Vanessa Dalton More articles by this author Helen Levy More articles by this author Richard Hirth More articles by this author Expand All Advertisement Advertisement PDF downloadLoading ...