(141) Comorbidity clusters before and after adjuvant endocrine therapy in oldest-old women with breast cancer: An unsupervised machine learning approach
Assistant Professor Health Outcomes Division, College of Pharmacy, The University of Texas at Austin, United States
Background: Oldest-old women with breast cancer have substantial comorbidities. Although adjuvant endocrine therapy (AET) and comorbidities interact with each other, little is known about the comorbidity patterns in the health trajectory of oldest-older women treated with AET.
Objectives: To evaluate comorbidity patterns in oldest-older women with breast cancer during one year before and two years after the initiation of AET.
Methods: Using the 2016-2019 Surveillance, Epidemiology, and End Results-Medicare database, we included the oldest-old women aged ≥80 years who had hormone receptor-positive, stage I to III breast cancer and who initiated AET in 2017. The index date was the date of the first use for AET. We used 540 Clinical Classifications Software Refined categories and selected the top 50 categories during one year before and two years after the index date. We applied hierarchical clustering, an unsupervised machine learning algorithm, to identify comorbidity clusters.
Results: We included 2,008 oldest-old women who initiated AET (mean age: 83.8 years). We identified ten comorbidity clusters before and after AET initiation and found that after AET use, fluid and electrolyte disorders became included in the top 50 comorbidities and grouped with other conditions (first year: hypertension with complications, chronic kidney disease, and anemia; second year: various symptom conditions including malaise, fatigue, and nervous/genitourinary/digestive/abdomen symptoms). In the second year after AET initiation, heart failure became included in the top 50 comorbidities and grouped with coronary atherosclerosis, hypertension with complications, chronic kidney disease, and anemia.
Conclusions: We found that comorbidity patterns changed immediately after AET initiation. This finding of clinically meaningful groupings from a large number of comorbidities highlights the need for a patient-centered approach to managing conditions likely to occur together.
