Trajectories of clinical encounters among long-term lung cancer survivors: A SEER-Medicare cluster analysis.

Abstract

e13633 Background: With improved lung cancer survival, there is a knowledge gap in the variations in clinical needs and healthcare utilization among long-term survivors, and how these variations may affect their survivorship. Methods: This study used 2010–2019 SEER–Medicare data to examine clinical trajectories among long-term lung cancer survivors (months 61–72 post-diagnosis) who were continuously enrolled in Medicare Parts A, B, and D and had no record of active treatment during the 12 months. Each clinical encounter was classified into one of six broad provider-specialty categories: primary care, oncology/hematology, hospitalist/ED/surgery, pain specialists, musculoskeletal/rehabilitation, and other specialties, then arranged chronologically to form each patient’s trajectory. We calculated a pairwise dissimilarity measure comparing the trajectories of any two individuals and applied hierarchical agglomerative clustering to group survivors with similar encounter patterns. Finally, we examined how demographics, comorbidities, and community socioeconomic profiles varied across these empirically derived clusters. Results: Among 17,256 lung cancer survivors, hierarchical clustering analysis revealed four distinct patterns of clinical encounters. The largest cluster, “low‐volume utilizers” (N = 10,800), had short encounter sequences (median = 11 encounters), the lowest prevalence of chronic comorbidities, and relatively lower opioid prescriptions (29.3%) in months 55–60 of survivorship. In contrast, the “primary care dominant” group (N = 1,377; median = 39) exhibited the greatest comorbidity burden, including high rates of cardiovascular disease, COPD, depression, and ADRD, and the highest opioid prescription rate (46.3%). The “specialist dominant” group (N = 3,494; median = 26) had substantial chronic disease burden but slightly lower opioid prescriptions, 30.9%, while the “musculoskeletal provider dominant” cluster (N = 1,585; median = 35) had a notable prevalence of musculoskeletal conditions and showed elevated opioid use (35.7%). Survivors in the primary care dominant cluster were more likely to reside in communities with lower income and education. Conclusions: Our pattern mining analysis found a complex interplay of comorbidity burden, provider specialty patterns, and disparities in chronic disease management across lung cancer survivorship clusters. Overall Low utilizers PCP dominant Specialist dominant Musculoskeletal dominant N 17,256 10,800 1,377 3,494 1,585 Opioid prescriptions 31.6% 29.3% 46.3% 30.9% 35.7% ADRD 11.0% 8.8% 28.2% 11.2% 10.0% Chronic Kidney Disease 23.8% 18.5% 42.1% 33.4% 23.2% COPD 47.7% 42.7% 68.9% 57.5% 42.1% Musculoskeletal conditions 40.9% 33.1% 58.8% 47.6% 64.0% Depression 21.5% 18.0% 44.3% 22.8% 23.1% Hypertension 68.9% 62.0% 88.6% 80.7% 72.8% Median household income $63,467 $61,399 $58,573 $67,739 $72,361