Effectiveness of Artificial Intelligence–Assisted Decision-making to Improve Vulnerable Women’s Participation in Cervical Cancer Screening in France: Protocol for a Cluster Randomized Controlled Trial (AppDate-You)

Background The French organized population-based cervical cancer screening (CCS) program transitioned from a cytology-based to a human papillomavirus (HPV)–based screening strategy in August 2020. HPV testing is offered every 5 years, starting at the age of 30 years. In the new program, women are invited to undergo an HPV test at a gynecologist’s, primary care physician’s, or midwife’s office, a private clinic or health center, family planning center, or hospital. HPV self-sampling (HPVss) was also made available as an additional approach. However, French studies reported that less than 20% of noncompliant women performed vaginal self-sampling when a kit was sent to their home. Women with lower income and educational levels participate less in CCS. Lack of information about the disease and the benefits of CCS were reported as one of the major barriers among noncompliant women. This barrier could be addressed by overcoming disparities in HPV- and cervical cancer–related knowledge and perceptions about CCS. Objective This study aimed to assess the effectiveness of a chatbot-based decision aid to improve women’s participation in the HPVss detection-based CCS care pathway. Methods AppDate-You is a 2-arm cluster randomized controlled trial (cRCT) nested within the French organized CCS program. Eligible women are those aged 30-65 years who have not been screened for CC for more than 4 years and live in the disadvantaged clusters in the Occitanie Region, France. In total, 32 clusters will be allocated to the intervention and control arms, 16 in each arm (approximately 4000 women). Eligible women living in randomly selected disadvantaged clusters will be identified using the Regional Cancer Screening Coordinating Centre of Occitanie (CRCDC-OC) database. Women in the experimental group will receive screening reminder letters and HPVss kits, combined with access to a chatbot-based decision aid tailored to women with lower education attainment. Women in the control group will receive the reminder letters and HPVss kits (standard of care). The CRCDC-OC database will be used to check trial progress and assess the intervention’s impact. The trial has 2 primary outcomes: (1) the proportion of screening participation within 12 months among women recalled for CCS and (2) the proportion of HPVss-positive women who are “well-managed” as stipulated in the French guidelines. Results To date, the AppDate-You study group is preparing and developing the chatbot-based decision aid (intervention). The cRCT will be conducted once the decision aid has been completed and validated. Recruitment of women is expected to begin in January 2023. Conclusions This study is the first to evaluate the impact of a chatbot-based decision aid to promote the CCS program and increase its performance. The study results will inform policy makers and health professionals as well as the research community. Trial Registration ClinicalTrials.gov NCT05286034; https://clinicaltrials.gov/ct2/show/NCT05286034 International Registered Report Identifier (IRRID) PRR1-10.2196/39288


NÂT N IONÂtL vu CANCER PROJET RISP-21-009
The proposal is clearly written and well thought through. There is an innovative approach applying use of chatbots to cervical screening. Improving access to cervical screening for deprived women is relevant to the 2014-2019 cancer plan and 2018-2022 National Health Strategy. There is a strong rationale based on the existing literature. The intervention seeks not only to improve screening uptake but also improve informed decision making. It is future-orientated -embracing technology and benefits from being multilanguage and designed for women with lower education attainment. It is excellent that the information will be provided in text or spoken language to further address literacy barriers. It is carried by a strong team of national and international researchers. The description of outcome measures is clear. The project draws on self care framework and Ottawa Decision Support Framework. It incorporates health economics.
The aim is to increase HPV self-samples -which is a behaviour -however, a behavioural scientist does not currently appear to be included in the research team.

Conclusions.
I strongly support this application. It is innovative, addresses inequalities in cervical screening uptake, supports informed decision-making and have the potential to overcome literacy and language barriers which are significant issues in cancer screening.

Rapporteur 3
Points forts. This is an innovative project that will develop and evaluate a digital decision aid involving a chatbot delivered via multiple media to try to increase women's response to vaginal self-sampling in HPV testing (HPVss). The focus on women of lower socio-economic status is a particular strength of the project. The project has been well thought through and represents a coherent programme of research. Collaborators have been well chosen, and the project seems highly feasible. The project incorporates an economic evaluation.
The justification for using a decision aid to try to increase women's response to vaginal self-sampling (HPVss) could be strengthened. Although much is known about the barriers to cervical screening uptake, we know much less about the barriers to HPVss and, in particular, why less than 20% of women in two French studies returned a sample when a self-sampling kit was sent to their home (as stated on page 18 of the project description).The application states that the reasons will be clarified by an ongoing qualitative study. However, unless we understand what the barriers are, it is difficult to make a strong case for using a decision aid. If some women do not regard cervical screening as important or have other things to worry about, they may not use a decision aid.So it is important to ensure that the proposed intervention (decision aid) does actually address the key barriers. Related to the above point, the development of the decision aid needs to be informed closely by the views of the target population i.e. women of low socio-economic status who have not responded to a previous invitation for cervical screening. This is crucial. The views of the -« INSTITUT NATIONAL PROJET RISP-21-009 ' Du CANCER target population are more important than those of clinicians in this respect. Although the project description implies that the views of the target population will be taken into account, it is not clear that they will be given the highest priority.
The committee states it is a well thought, innovative and future orientated project, promising to contribute to the reduction of literacy barriers.