Research Study Attrition and Participant-Informed Retention Strategies Among Cancer Survivors.

IMPORTANCE

Attrition in longitudinal cancer cohorts compromises data completeness, introduces bias, and limits generalizability.

OBJECTIVE

To quantify nonresponsiveness over time in a contemporary cohort of cancer survivors, identify individual- and center-level factors associated with nonresponsiveness, and elicit participant-informed strategies to strengthen long-term engagement and retention.

DESIGN, SETTING, AND

PARTICIPANTS

This cohort study used a mixed-methods design integrating cohort analyses among patients enrolled in a longitudinal study from 2012 to 2018 and a national cross-sectional survey and qualitative thematic analyses among participants enrolled in 2023. Cohort analyses were conducted among patients recruited from the Cancer Toxicity (CANTO) study, a multicenter, prospective, longitudinal cohort of survivors of stage I to III breast cancer, with assessments at diagnosis and years 1, 2, 4, and 6. The national survey was distributed online through the Seintinelles collaborative research network to cancer survivors and individuals without cancer. Remote interviews and focus groups participants were recruited from CANTO. The final analyses were performed in June 2026.

EXPOSURE: Baseline sociodemographic, behavioral, treatment-related, and center-level characteristics; survey-reported preferences, barriers, and facilitators; and qualitative accounts of participation experiences.

MAIN OUTCOMES AND MEASURES: The outcome of interest was nonresponsiveness at each follow-up, defined as completion of none of the protocol-specified clinical, socioeconomic, or patient-reported outcome assessments. Multivariable logistic regression estimated factors associated with nonresponsiveness. Survey responses and qualitative themes characterized participant-informed retention strategies.

RESULTS

Among 11 413 CANTO participants, mean (SD) age was 56.4 (11.3) years and 4019 (42.4%) reported a monthly household income of €3000 or higher. Nonresponsiveness increased from 149 participants (1.3%) at year 1 to 1016 participants (9.1%) at year 2, 1738 participants (16.1%) at year 4, and 2512 participants (25.8%) at year 6. Factors associated with higher nonresponsiveness at the 4-year follow-up included lower socioeconomic status (income <€1500 vs ≥€3000: adjusted odds ratio [aOR], 1.51 [95% CI, 1.18-1.93]), unhealthy behaviors (current/former smokers vs never smokers: aOR, 1.37 [95% CI, 1.19-1.58]), and enrollment at smaller centers (vs larger: aOR, 1.88 [95% CI, 1.43-2.48]), with consistent findings at year 6. Among 2383 survey respondents (1295 respondents [55.8%] aged 40-65 years; 2203 [96.5%] female), 1982 (94.8%) preferred remote questionnaire completion, 1994 (95.4%) preferred digital completion, 1606 (67.4%) valued flexible scheduling, and 1494 (71.4%) reported that regular study updates or result-sharing events would encourage participation. Qualitative findings among 30 participants further highlighted reduced burden, decentralized participation, navigation support, clearer communication, and feedback on study progress and findings as key facilitators.

CONCLUSIONS AND RELEVANCE: In this mixed methods cohort study of survivors enrolled in a large longitudinal breast cancer cohort, nonresponsiveness increased over time and was socially, clinically, and organizationally patterned. Survey and qualitative findings suggested that participant-centered retention strategies should combine procedural flexibility, remote follow-up, navigation support, targeted communication, and regular feedback and informed the design of a digitally enabled, multimodal retention intervention.