Evaluation of long-term follow-up non-respondence in a prospective study of financial toxicity of patients after radiation therapy.

Abstract

e23266 Background: Barriers to enrollment and retention of socioeconomically vulnerable patients limit the diversity and generalizability of longitudinal oncology studies. Financial toxicity (FT) may be underestimated if patients who are at highest risk are less likely to complete long-term follow-up (FU). We evaluated socioeconomic and clinical factors associated with non-respondence at 1-year follow-up in a prospective FT survey study of patients undergoing radiation therapy (RT). Methods: All patients ≥18 years that were undergoing RT for any malignancy were eligible. Surveys were completed at baseline (pre-RT), 1-month, and 1-year post-RT, which included the Comprehensive Score for Financial Toxicity (COST) tool and questions regarding demographics and social determinants of health (SDOH). Lower COST scores indicate higher FT. Respondents to the 1-year FU were classified as those who completed the survey between 6-18 months post-RT. Non-respondents were those who did not respond or passed away within the response window. Patients who passed prior to the 1-year FU survey window were ineligible. Chi-square, Fisher’s exact, and Wilcoxon rank-sum tests were used to compare factors associated with respondents and non-respondents. Results: Of 305 patients completing survey, 241 were eligible for analysis: 36% completed the 1-year FU survey, while 64% did not. Non-respondents were more likely to report lower COST scores (higher FT), have a non-English primary language, not be married/partnered, have an educational level of high school or less, hold either private insurance or MassHealth, and have an annual income of < $40,000 (all P< 0.05). Non-respondents also received a greater number of fractions of RT (P< 0.05). When excluding patients who passed during the response window, all variables remained statistically significant except for partnership status. Cancer and treatment characteristics, other than the number of fractions of RT, were not significantly different between the groups. Conclusions: Long-term FU nonresponse was associated with specific SDOH and number of RT fractions. This suggests multiple socioeconomic factors and certain treatment aspects could potentially impact long-term study FU data. As randomized controlled trials follow patients across a longitudinal period, these challenges may similarly affect long-term outcome assessment. These findings can inform strategies to reduce attrition, which will ultimately improve diversity and representativeness in clinical trials.