Enhancing Therapy Adherence Among Metastatic Breast Cancer Patients: the Study Protocol
Enhancing Therapy Adherence Among Metastatic Breast Cancer Patients: the Study Protocol
Background: Emerging evidence indicates that patients with advanced cancer, such as those with MBC, often exhibit significant levels of nonadherence to oral anticancer treatments. Leveraging of the machine learning models in clinical practice enables the provision of personalized predictions on medication adherence for individual patients, thereby supporting adherence and facilitating targeted interventions.
Objective: The current protocol aims to assess the efficacy of the DSS, a web-based solution named TREAT (TREatment Adherence SupporT), and a machine learning web application in promoting adherence to oral anticancer treatments within a sample of MBC patients.
Methods and Design: This protocol is part of a project titled "Enhancing Therapy Adherence Among Metastatic Breast Cancer Patients" (Tracking Number 65080791). A sample of 100 MBC patients is enrolled consecutively and admitted to the Division of Medical Senology of the European Institute of Oncology. 50 MBC patients receive the DSS for three months (experimental group), while 50 MBC patients not subjected to the intervention receive standard medical advice (control group). The protocol foresees three assessment time points: T1 (1-Month), T2 (2-Month), and T3 (3-Month). At each time point, participants fill out a set of self-reports evaluating adherence, clinical, psychological, and QoL variables.
Conclusions: our results will inform about the effectiveness of the DSS and risk-predictive models in fostering adherence to oral anticancer treatments in MBC patients.
Metastatic breast cancer (MBC) represents an incurable condition wherein pharmacological interventions are directed towards deferring disease progression and alleviating symptoms, thereby extending survival rates and preserving the quality of life (QoL) and psychological well-being. Clinical advancements in anticancer treatments have notably augmented survival rates among MBC patients. However, accruing evidence reported that adherence to medications is a critical issue in the disease trajectory of breast cancer patients, particularly in the context of oral anticancer treatments (OATs). Emerging evidence indicates that patients with advanced cancer, such as those with MBC, often exhibit significant levels of nonadherence. MBC patients encounter various barriers to the daily management of OATs, including emotional and physical distress associated with side effects, dosage variations, treatment interruptions, and a lack of disease-related knowledge. Prediction models for adherence have been previously developed and tested across diverse scenarios and diseases. Evidence suggested that leveraging of the machine learning models in clinical practice enables the provision of personalized predictions on medication adherence for individual patients, thereby supporting adherence and facilitating targeted interventions. Even so, existing studies have yet to systematically address medication adherence among MBC patients by designing and implementing a decision support system (DSS) that integrates risk predictive models alongside educational and training tools.
The current protocol aims to assess the efficacy of the DSS, a web-based solution named TREAT (TREatment Adherence SupporT), and a machine learning web application in promoting adherence to oral anticancer treatments within a sample of MBC patients. This protocol is part of a project titled "Enhancing Therapy Adherence Among Metastatic Breast Cancer Patients" (Tracking Number 65080791). The overarching goal of this project is to develop a predictive model of nonadherence, an associated DSS, and guidelines to enhance patient engagement and therapy adherence among MBC patients.
The web-based DSS was developed in the first year of the Pfizer Project (65080791) using a patient-centric approach and comprises four sections: i) Metastatic Breast Cancer; ii) Adherence to Cancer Therapies; iii) Promoting Adherence; iv) My Adherence Diary. Moreover, a machine learning web-based application was designed to focus on predicting patients' risk factors for adherence to anticancer treatment, specifically considering physical status, comorbid conditions, and short- and long-term side effects. This machine learning web-based application was developed through a retrospective study employing physiological, clinical, and quality of life data available in the European Institute of Oncology (Milan, Italy) (R1595/21-IEO 1704). Specifically, multi-modal retrospective data has been retrieved from the Patient Electronic Health Records (EHR) using natural language processing (NLP) in a sample of 2.750 MBC patients (from 2010 to 2020).
Methods/Design
Main objectives
Evaluating the effectiveness of the DSS web-based solution and machine learning web application (TREAT - "TREatment Adherence SupporT") in fostering adherence to oral anticancer treatments within a cohort of 100 Metastatic Breast Cancer (MBC) patients over a three-month period. Adherence is assessed by calculating the number of pills taken divided by the prescribed amount.
Secondary Objectives
Identify clinical factors (comorbidities, pain presence, tumor type, treatment type), psychological parameters (personality traits, anxiety, depression, self-efficacy for coping with cancer, sense of coherence, and risk perception), and QoL variables that serve as predictors for patients' adherence to OATs. These predictors are utilized to assess nonadherence to OATs among MBC patients and enhance the initial version of a machine learning model developed in the retrospective study (R1595/21-IEO 1704). Data for the secondary endpoints are collected using the European Organization for Research and Treatment of Cancer Quality of Life questionnaire (EORTC-QLQ-C30), the European Organization for Research and Treatment of Cancer 23-item Breast Cancer-specific Questionnaire (EORTC-QLQ-BR23), and the Brief Pain Inventory (BPI). Furthermore, to evaluate psychological variables, the following measures are used: the State-Trait Anxiety Inventory (STAI-Y), the Beck Depression Inventory-II (BDI-II), the Big Five Inventory (BFI), the Cancer Behavior Inventory CBI Short form (CBI-B/I), the Sense of Coherence (SOC-13), and Risk Perception (utilizing two Visual Analog Scales).
Trial Duration and Study Design
The study is designed as a 3-month randomized controlled study conducted at the European Institute of Oncology (IEO). More specifically, a sample of 100 patients is enrolled consecutively and admitted to the Division of Medical Senology with an MBC diagnosis. Patients who signed the informed consent are given a unique identifier and assigned to either the control or intervention arm in a 1:1 ratio. Earliest, the system asks to confirm all inclusion and exclusion criteria. Then, an independent researcher generates a random sequence using the statistical language R (R Core Team 2020).
Experimental Group - TREAT (TREatment Adherence SupporT): 50 MBC patients receive the DSS for three months. Patients are instructed to use the DSS ad libitum. Further, Patients are explicitly informed that TREAT does not replace clinical consultations, but it is designed to assist in managing oral treatment and enhancing adherence through education based on evidence-based information. Control Group: 50 MBC patients not subjected to the intervention receive standard medical advice.
The protocol foresees three assessment time points: T1 (1-Month), T2 (2-Month), and T3 (3-Month). At the baseline (T0), all patients fill out validated questionnaires to measure adherence, clinical, psychological, and QoL variables. The expected time to complete all the given questionnaires at baseline is approximately 40 minutes. Furthermore, all patients have to fill a weekly adherence medication diary for three months. Each month, all participants receive a brief telephone interview in which they are monitored for compliance with the research protocol. At T1, T2, and T3, all behavioral, psychological, and QoL measures are filled out, and an interview (online or vis-à-vis) is performed. Variables that are not sensitive to change, such as personality and anxiety trait, are collected only at T0.
Inclusion Criteria:
Exclusion Criteria: