Risk factors for lack of adherence with diagnostic follow-up care in breast cancer patients

Abstract Purpose/Objective: There is a need for innovative methods to provide high-quality care to vulnerable populations in an effort to reduce disparities. This is particularly important for women facing a new breast cancer diagnosis. The purpose of the present research was to investigate what social determinants may play a role in the existing disparities impacting adherence to care recommendations in communities with poor access to primary care and higher rates of morbidity and mortality. This study characterized adherence with breast health follow-up care in a diverse population of patients where inequalities exist that may negatively impact adherence to treatment recommendations. The objective of the study was to characterize factors associated with adherence to treatment among women with newly diagnosed breast cancer. Materials/Methods: Women diagnosed with stage I through IV breast cancer treated at the University of Florida College of Medicine-Jacksonville between 01/01/2014 and 12/31/2019 were included in the sample. Patterns of adherence were categorized using machine learning methods with data derived from the electronic medical record and the UF Health Jacksonville Tumor registry. Age, race, and Area Deprivation Index (ADI) state rank in 2019 as a disparity proxy were used to build a machine learning model and classify compliance to treatment. Included patients had a diagnostic procedure that identified breast cancer. Compliance to treatment was fulfilled if the patient received surgery following diagnostic confirmation. A machine learning model was used to stratify patients by risk of non-adherence to treatment following a diagnostic procedure. The models were evaluated using their area under the curve (AUC). Results: A total of 6,951 women were included, 629 who were adherent and 6322 non-adherent patients with breast cancer. The average age of the participants was 61.4 years, (Standard Deviation = 12.8 years). The majority of patients were Black (48%) or Caucasian (45%), 2% were Asian, and 5% were Other races. Payer type at diagnosis showed 45% had Medicare, 30% had commercial insurance, 17% were covered by Medicaid, 7% were charity, and 1% had other sources of pay. Most women were diagnosed with stage III breast cancer. Of 346 patients who received surgery that data was available, 127 (36.7%) had surgery within 30 days of diagnosis, 102 (29.5%) between 31 and 60 days, and 37 (10.7%) between 61 and 90 days. Fifteen models were compared using the PyCaret Python library. The ADI appeared as the most important factor to predict adherence in the model, followed by race and characterized by an AUC of 0.63. Conclusion: Our clinic treats predominantly more women diagnosed with biologically aggressive and advanced breast cancer especially in young African American population. The role social conditions play that precipitate and perpetuate health care disparities were investigated to determine their impact on adherence to treatment. At our safety net hospital, over one third were able to undergo surgery within 30 days of diagnosis. The ADI appeared as the most important feature to predict adherence, followed by race. This demonstrated the necessity to better understand the relation between socio-economical determinants and care received by patients. A more detailed description of the patients’ circumstances, such as access to transport, proximity of the hospital, and insurance status may further improve the model. There is a need for innovative methods of providing quality health care to vulnerable populations. Machine learning models can be used to stratify patients by risk of non-adherence to diagnostic follow-up and treatment following a diagnosis of breast cancer. Future research needs to move from identification of non-adherence risk factors to implementation of interventions to improve breast cancer outcomes. Citation Format: Guillaume Labilloy, Brian Celso, Bharti Jasra, Leigh Neumayer, Erin Mobley, Carmen Smotherman, Jennifer Brailsford. Risk factors for lack of adherence with diagnostic follow-up care in breast cancer patients [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P5-14-14.