Frederick R Kates1, Ryan Romero2, Daniel Jones3, Jacqueline Egelfeld4, Santanu Datta5. 1. Department of Health Services Research, Management and Policy, College of Public Health and Health Professions, University of Florida, Gainesville, Florida, USA. kates.rick@phhp.ufl.edu. 2. Bachelor of Public Health, College of Public Health and Health Professions, University of Florida, Gainesville, Florida, USA. 3. Bachelor of Science in Statistics, College of Liberal Arts and Sciences, University of Florida, Gainesville, Florida, USA. 4. Bachelor of Health Science, College of Public Health and Health Professions, Gainesville, Florida, USA. 5. Department of Health Services Research, Management and Policy, College of Public Health and Health Professions, University of Florida, Gainesville, Florida, USA.
Abstract
INTRODUCTION: To align patient preferences and understanding with harm-benefit perception, the Centers for Medicare & Medicaid Services (CMS) mandates that providers engage patients in a collaborative shared decision-making (SDM) visit before LDCT. Nonetheless, patients and providers often turn instead to the web for help making decisions. Several web-based lung cancer risk calculators (LCRCs) provide risk predictions and screening recommendations; however, the accuracy, consistency, and subsequent user interpretation of these predictions between LCRCs is ambiguous. We conducted a systematic review to assess this variability. DESIGN: Through a systematic Internet search, we identified 10 publicly available LCRCs and categorized their input variables: demographic factors, cancer history, smoking status, and personal/environmental factors. To assess variance in LCRC risk prediction outputs, we developed 16 hypothetical patients along a risk continuum, illustrated by randomly assigned input variables, and individually compared them to each LCRC against the empirically validated "gold-standard" PLCO risk model in order to evaluate the accuracy of the LCRCs within identical time-windows. RESULTS: From the inclusion criteria, 11 calculators were initially identified. The analyzed calculators also vary in output characteristics and risk depiction for hypothetical patients. There were 13 total instances across ten hypothetical patients in which the sample standard error exceeded the mean risk percentage across all general samples and set standard calculations. The largest measured difference is 16.49% for patient 8, and the smallest difference is 0.01% for patient 2. The largest measured difference is 16.49% for patient 8, and the smallest difference is 0.01% for patient 2. CONCLUSION: Substantial variability in the depiction of lung cancer risk for hypothetical patients exists across the web-based LCRCs due to their respective inputs and risk prediction models. To foster informed decision-making in the SDM-LDCT context, the input variables, risk prediction models, risk depiction, and screening recommendations must be standardized to best practice.
INTRODUCTION: To align patient preferences and understanding with harm-benefit perception, the Centers for Medicare & Medicaid Services (CMS) mandates that providers engage patients in a collaborative shared decision-making (SDM) visit before LDCT. Nonetheless, patients and providers often turn instead to the web for help making decisions. Several web-based lung cancer risk calculators (LCRCs) provide risk predictions and screening recommendations; however, the accuracy, consistency, and subsequent user interpretation of these predictions between LCRCs is ambiguous. We conducted a systematic review to assess this variability. DESIGN: Through a systematic Internet search, we identified 10 publicly available LCRCs and categorized their input variables: demographic factors, cancer history, smoking status, and personal/environmental factors. To assess variance in LCRC risk prediction outputs, we developed 16 hypothetical patients along a risk continuum, illustrated by randomly assigned input variables, and individually compared them to each LCRC against the empirically validated "gold-standard" PLCO risk model in order to evaluate the accuracy of the LCRCs within identical time-windows. RESULTS: From the inclusion criteria, 11 calculators were initially identified. The analyzed calculators also vary in output characteristics and risk depiction for hypothetical patients. There were 13 total instances across ten hypothetical patients in which the sample standard error exceeded the mean risk percentage across all general samples and set standard calculations. The largest measured difference is 16.49% for patient 8, and the smallest difference is 0.01% for patient 2. The largest measured difference is 16.49% for patient 8, and the smallest difference is 0.01% for patient 2. CONCLUSION: Substantial variability in the depiction of lung cancer risk for hypothetical patients exists across the web-based LCRCs due to their respective inputs and risk prediction models. To foster informed decision-making in the SDM-LDCT context, the input variables, risk prediction models, risk depiction, and screening recommendations must be standardized to best practice.
Entities:
Keywords:
Centers for Medicare and Medicaid Services, U.S; Early detection of cancer; decision-making, shared; risk calculators; tomography, x-ray computed
Authors: Yaron B Gesthalter; Elisa Koppelman; Rendelle Bolton; Christopher G Slatore; Sue H Yoon; Hilary C Cain; Nichole T Tanner; David H Au; Jack A Clark; Renda Soylemez Wiener Journal: Chest Date: 2017-02-20 Impact factor: 9.410
Authors: Denise R Aberle; Amanda M Adams; Christine D Berg; William C Black; Jonathan D Clapp; Richard M Fagerstrom; Ilana F Gareen; Constantine Gatsonis; Pamela M Marcus; JoRean D Sicks Journal: N Engl J Med Date: 2011-06-29 Impact factor: 91.245
Authors: Dirga Kumar Lamichhane; Hwan Cheol Kim; Chang Min Choi; Myung Hee Shin; Young Mog Shim; Jong Han Leem; Jeong Seon Ryu; Hae Seong Nam; Sung Min Park Journal: Yonsei Med J Date: 2017-11 Impact factor: 2.759