Dissertations, Theses, and Capstone Projects

Date of Degree

2-2023

Document Type

Dissertation

Degree Name

Ph.D.

Program

Social Welfare

Advisor

Daniel Gardner

Committee Members

Michael Lewis

Alexis Jemal

Jamie Ostroff

Subject Categories

Health Psychology | Social and Behavioral Sciences | Social Work

Keywords

Cancer, tobacco treatment, social work, machine learning

Abstract

Among patients diagnosed with cancer, persistent tobacco use is associated with adverse clinical outcomes such as worse treatment side effects, decreased effectiveness of cancer treatment (chemotherapy, radiotherapy, and surgery), all increasing risk of recurrence, second primary cancers, and poor survival. Despite the clinical importance of tobacco cessation in the context of high quality cancer care, engaging Black/African American and Hispanic/Latino cancer patients in tobacco treatment programs can be challenging. Prior studies with the general adult population demonstrate that Black/African American and Hispanic/Latino smokers are referred to and accept tobacco cessation treatment at lower rates compared to non-Hispanic White smokers. This observation raises concern about access and utilization of evidence-based tobacco treatment services among Black/African American and Hispanic/Latino cancer patients. The importance of this issue is underscored by the observation that Blacks and African Americans as well as Latinos/Hispanic patients have lower rates of cancer survival compared to non-Hispanic White cancer patients.

Prior research on engagement of Black/African American and Hispanic/Latino cancer patients in tobacco treatment is extremely limited. Identifying multi-level barriers (patient, provider, and systems-level) associated with Black/African American and Hispanic/Latino cancer patient acceptance and reasons for refusal in tobacco treatment is an essential step in improving the reach and effectiveness of evidence-based, socio-politically aware tobacco cessation treatment.

This study seeks to understand patterns and socio-demographic factors associated with tobacco treatment referral and treatment acceptance among a large, hospital-based, clinical population of adult cancer patients identified as currently smoking. With the overall objective of advancing understanding of specific barriers and facilitators that impact Black/African American and Hispanic/Latino cancer patients’ decision to accept or refuse tobacco cessation treatment, this project will examine the following complementary research questions: 1) What are the overall patterns of tobacco treatment referral and acceptance of referral among cancer patients who are identified as currently smoking. 2) What, if any, sociodemographic and cancer diagnosis disparities in treatment referral acceptance exist among cancer patients identified as currently smoking? To address the research questions this study will use archival data from a large clinical cohort of cancer patients treated at a single comprehensive cancer center who were identified as current smokers and received a referral for tobacco treatment between January 2021 – December 2021 (N=95,356). Aim 1 will use univariate descriptive statistics: 1) To provide a quantitative summary of treatment referral and acceptance patterns among cancer patients identified as current smokers; 2) To describe and summarize the self-reported reasons cancer patients identified as currently smoking refuse tobacco treatment. Aim 2 will use Random Forest, a machine learning algorithm, to identify sociodemographic and cancer diagnosis factors (i.e., age, sex, race, ethnicity, income, education and cancer type and stage, smoking history, geographic region, and insurance type) that most strongly predict tobacco treatment referral acceptance.

The results of this research will lay the groundwork for the development of socio-politically aware implementation strategies for increasing the rate of tobacco treatment acceptance and utilization among Black/African American and Hispanic/Latino cancer patients and survivors.

Download

Share

COinS