Determinants associated with longitudinal adherence to annual lung cancer screening: a - PowerPoint Presentation

1. Determinants associated with longitudinal adherence to annual lung cancer screening: a retrospective analysis of claims dataErin A. Hirsch1, Stephen P. Malkoski2, Anna E. Baron1, Betsy C. Risendal3, Melissa L. New2, and Jamie L. Studts4 1 Colorado School of Public Health, Department of Biostatistics and Informatics, University of Colorado Anschutz Medical Campus2 School of Medicine, Division of Pulmonary and Critical Care, University of Colorado Anschutz Medical Campus3 Colorado School of Public Health, Department of Community and Behavioral Health, University of Colorado Anschutz Medical Campus4 School of Medicine, Division of Medical Oncology, University of Colorado Anschutz Medical CampusObjectivesIdentify demographic, comorbidity, healthcare usage, and geographic factors associated with annual lung cancer screening adherence.Determine population-based adherence rates to annual lung cancer screening in Colorado.Describe adherence rates to subsequent annual lung cancer screening rounds.BackgroundLung cancer is the leading cause of cancer death in the US1, with lethality largely driven by the preponderance of metastatic disease at diagnosis. Despite recent treatment advances, the 5-year survival of metastatic lung cancer remains under 10%.2 Though the 5-year survival of early-stage lung cancer exceeds 55%, only 17% of patients present with localized disease.2 Lung cancer screening (LCS) efficacy by low-dose CT has been described in several large trials, including a 20% reduction in lung cancer specific mortality in the landmark National Lung Screening Trial (NLST)3, and a 25% reduction (for men) in the Dutch-Belgian Lung Cancer Screening (NELSON) trial4. Adherence to annual screening in these two large clinical trials exceeded 90%,3,5 directly contributing to lung cancer mortality reductions. Annual adherence rates in academic,6.7 community,8 and federal9 lung cancer screening programs have overwhelmingly failed to achieve half the adherence rates observed in these trials. Determinants of LCS adherence in a real-world, unselected population, outside of a study have not been described. We used a state-based, administrative claims dataset to quantify population level LCS adherence and identify demographic, comorbidity, healthcare usage, and geographic factors associated with adherence. MethodsWe obtained all available health claims from 01/2012 to 12/2018 from the Colorado All Payer Claims Dataset (CO APCD) for individuals with a LCS low dose CT claim (codes G0297 and S8032) between 10/2014 and 12/2018We defined annual adherence as a second claim for a screening CT 10-18 months after the index claim and non-adherence as not having a second CT claim and >18 months had elapsed since the first claim. Individuals with 10-18 months of follow-up time and no second CT claim were censored.We excluded individuals with <10 months of follow-up after the index CT, individuals with a second CT claim 3-9 months after the index CT, individuals with a lung cancer diagnosis using ICD 9 and 10 codes, out-of-state residents, and individuals with missing predictor variable data. We assessed the following associations with LCS adherence: sex, age at index CT, residence (urban vs rural/frontier), insurance type linked with the index CT claim, health care utilization (number of outpatient visits in a 5 yr period), and comorbidity burden by Charlson Comorbidity Index (CCI).Cox proportional hazards regression was performed to characterize the relationship between annual adherence and predictor variables. Data analysis was generated using SAS/STAT software, Version 9.4 of the SAS System for Windows (SAS Institute Inc., Cary, NC, USA).The Colorado Multiple Institutional Review Board determined the study to be exempt for use of secondary dataResults1) American Cancer Society, Cancer Facts and Figures 2020. 2) SEER cancer stat facts: lung and bronchus cancer. National Cancer Institute website.3) Aberle DR, Adams AM, Berg CD, et al. National Lung Screening Trial Research Team. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med. 2011; 365 (5): 395-409.4) Koning HJ, van der Aalst CM, de Jong PA, et al. Reduced Lung-Cancer Mortality with Volume CT Screening in a Randomized Trial. N Engl J Med 2020; 382:503-513.5) Yousaf-Khan U, van der Aalst C, de Jong PA, et al. Final Screening Round of the NELSON Lung Cancer Screening Trial: The Effect of a 2.5-year Screening Interval. Thorax. 2017 Jan;72(1):48-56.6) Hirsch EA, New ML, Brown S, Báron AE, Malkoski SP. Patient Reminders and Longitudinal Adherence to Lung Cancer Screening in an Academic Setting. Ann Am Thorac Soc. 2019 Oct; 16(10):1329-1332.7) Catteneo SM, Meisenberg BR, Geronimo MCM, Bhandari B, Maxted JW, and Brady-Copertino CJ. Lung Cancer Screening in the Community Setting. Ann Thorac Surg 2018; 105:1627-32.8) Spalluto LB, Lewis JA, LaBaze S, et al. Association of a Lung Screening Program Coordinator with Adherence to Annual CT Lung Screening at a Large Academic Institution. J Am Coll Radiol. 2020 Feb;17(2):208-215.9) Tanner NT, Brasher PB, Wojciechowski B, Ward R, Slatore C, Gebregziabher M, Silvestri GA. Screening Adherence in the Veterans Administration Lung Cancer Screening Demonstration Project. Chest. 2020 Oct;158(4):1742-1752ResultsAges 55-64 and 75-79, rural residence, and Medicare FFS and Medicaid insurance are associated with reduced adherence to annual LCS. While higher healthcare usage and increased comorbidity burden are both associated with increased LCS adherence, the effects of these two variables are inter-dependent. Using claims data from the Colorado All Payer Claims Database we observed a 46% population-based adherence rate to annual lung cancer screening guidelines. Quantifying population-based adherence rates and understanding factors predictive of annual adherence is a critical first step in improving screening adherence and ultimately reducing lung cancer death.References   Adherent  Non-adherent CensoredUnivariate p-value  n = 3,072 n = 3,570 n = 2,414 Sex  Male1694 (55)1847 (52)1228 (51)p=0.003 Female1378 (45)1723 (48)1186 (49) Age at index LDCT 55-59496 (16)676 (19)446 (19)p=0.0160-64676 (22)769 (21)552 (23)65-691026 (33)1079 (30)752 (31)70-74670 (22)772 (22)486 (20)75-79204 (7)274 (8)178 (7) Residence  UrbanUrban2973 (97)3307 (93)2162 (90)p<0.0001 Rural/Frontier Rural 99 (3)263 (7)252 (10)Insurance at index LDCT  Commercial939 (30)770 (22)479 (20)p<0.0001 Medicare Advantage1340 (44)1023 (29)698 (29) Medicare FFS571 (19)1269 (35)840 (35) Medicaid222 (7)508 (14)397 (16) Number of outpatient visits 0521 (17)1184 (33)862 (36)p<0.000111060 (34)1206 (34)911 (38)2842 (27)668 (19)376 (16)≥3649 (21)512 (14)265 (11) CCI score 01227 (40)1611 (45)1133 (47)p<0.00011951 (31)1142 (32)768 (32)≥2 894 (29) 817 (23) 513 (21)ConclusionsDetermination of the Study PopulationCharacteristics of the Study PopulationResults of Cox Proportional Hazards RegressionScreening Adherence by Year of Index CTFigure 1:After removing exclusions and duplicate claims we had a final dataset of 9,056 records with 3,072 adherent individuals, 3,570 non-adherent individuals, and 2,414 individuals who were censored.Table 1: The adherent group had more males, individuals aged 65-69, urban residents, individuals with commercial or Medicare Advantage insurance, outpatient visits, and comorbidities. Figure 2A:After adjustment for covariates, age, residence, and insurance were significantly associated with annual lung cancer screening adherence. Within the multivariate model, outpatient visits and CCI had a significant interaction (p < 0.0001), as presented in Figure 2B.A)B)Figure 3A: Overall adherence in this cohort using an 18-month definition was 46%; this increased to 53% if any return for screening was considered adherent. Adherence to the first annual CT increased over the duration of this study (p = 0.0009). Figure 3B: Of individuals who were adherent to the first round of screening, adherence to the second annual LDCT was approximately 50%; due to data maturity, adherence to additional screening rounds could not be assessed.*For 2017, individuals who could not be classified due to data maturity are not included. †Overall numbers do not include unclassifiable individuals screened in 2017 or any individuals screened in 2018.AcknowledgementsThis research was supported by the University of Colorado Cancer Center Support Grant (NCI/NIH P30CA046934) and by NIH/NCATS Colorado CTSA Grant Number UL1 TR002535. We thank the Center for Improving Value in Health Care, administrator of the CO APCD, for assistance in acquisition of the dataset.