Young people with hereditary cancer syndromes face difficult questions about their future, prompting researchers to test whether tailored digital conversations could strengthen the information they receive in specialist care.
Study: Chatbot Intervention and Cancer Risk Knowledge in Adolescents and Young Adults With Cancer Predisposition. Image Credit: Stock-Asso / Shutterstock
In a recent study published in JAMA Network Open, researchers assessed how a chatbot-based intervention affected cancer risk understanding among adolescents and young adults (AYAs) with hereditary cancer predisposition.
AYAs with a hereditary predisposition to cancer have an elevated lifetime risk of cancer. Genetic testing and counseling are increasingly recommended for AYAs diagnosed with cancer. Models for cancer care and risk communication for these young people are still emerging. Previously, the authors developed AYA–Risk Information and Screening Education (AYA-RISE), a chatbot-based intervention delivered after a visit to a cancer predisposition clinic to educate AYAs about cancer risk.
About the study
In the present randomized trial at four US academic centers, English-speaking AYAs aged 12–24 years diagnosed with any one of the nine cancer predisposition syndromes were recruited, whereas those on active cancer therapy were excluded. Participants were randomized to receive a cancer predisposition clinic visit plus AYA-RISE (intervention) or a clinic visit alone (control). Knowledge of cancer risk before age 30 was the primary outcome; screening knowledge and psychological distress were secondary outcomes.
Participants completed a survey before the clinic visit. Researchers asked about the average risk of being diagnosed with cancer before age 30 for those with their syndrome. A similar question was used for lifetime cancer risk. AYAs were also asked which cancer screening methods were recommended at their current age.
Psychosocial problems were assessed using the Psychosocial Aspects of Hereditary Cancer (PAHC) questionnaire. A separate distress thermometer measured overall distress during the previous week. After the clinic visit, controls completed the same survey measures, whereas the intervention group interacted with AYA-RISE before completing the survey measures in the clinic or at home. AYA-RISE was accessible via phone, computer, or tablet, and its homepage offered six options (learn, plan, share, grow, chat, and connect).
The chat option led to a chatbot that provided information on cancer risk, screening, and prevention. Researchers developed syndrome-specific content from literature and screening guidelines; no material came from generative artificial intelligence. The trial evaluated the full AYA-RISE package, including its resource pages. Researchers assessed changes in distress and knowledge of cancer risk and screening from baseline to post-visit (follow-up) using generalized linear regression.
Findings
The study recruited 106 AYAs, with a mean age of 18.3 years. Most participants were girls or women (64%) and White (93%). DICER1 (19%) and Li-Fraumeni syndromes (18%) were the most common cancer predisposition syndromes. At baseline, 52% of the intervention group and 38% of controls had accurate knowledge of cancer risk by age 30; at follow-up, these estimates were 78% and 37%, respectively, reflecting a significantly greater improvement in the intervention group.
Lifetime cancer risk results did not differ significantly between groups. Initially, 33% and 44% of AYAs in the intervention and control groups had accurate knowledge of lifetime cancer risk, increasing to 52% and 50%, respectively, at follow-up.
Likewise, knowledge of recommended screening improved in the intervention group in the main analysis, but the difference was not significant. At baseline, 85% of responses about individual screening methods in controls and 90% in the intervention group were accurate, leaving limited room for improvement. Across the assessed PAHC categories, participants reported problems in 37% of responses in controls and 36% in the intervention group at baseline.
Both groups had fewer post-visit problems: 30% in controls and 25% in the intervention group, with no significant difference in problem reduction between groups. Distress about hereditary predisposition was the most common problem domain, with more than half of participants in each group reporting problems in this domain at baseline. Additional analyses found that participants who initially answered incorrectly showed significantly greater knowledge gains in the intervention group than in controls for risk before age 30, screening methods, and body areas requiring screening.
Enrollment fell short of the original 300-person target, limiting statistical power, and only about one-quarter of approached, eligible AYAs enrolled. The predominantly White sample from specialist centers limits generalizability. The reported results do not show whether knowledge gains persisted or led to better screening adherence or health outcomes.
Conclusions
Taken together, the chatbot-based intervention was more effective at improving knowledge of cancer risk before age 30 than a standalone clinic visit among AYAs with cancer predisposition. The intervention did not increase psychological distress, but it also did not show an additional well-being benefit. Combining in-person clinic interaction with electronic information can support AYAs with cancer predisposition syndromes by improving short-term risk knowledge; benefits for other conditions require further study.
Sources:
Journal reference: