A Randomized Cluster Trial to Evaluate a Mobile Mammography Unit in Breast Cancer Screening in France (Mammobile)

Context In most EU member states, breast cancer screening is organised with a mammography screening. However, there are still differences in the way screening programmes are implemented. In France, the organized breast cancer screening (OBCS) has been conducted, nationwide since 2004. A screening mammography is offered every two years to women aged 50 to 74 at average risk for this cancer. They receive from management structures in charge of this screening (SMS), an invitation to visit an accredited radiologist's office. A radiological imaging of the breast with two views plus a clinical breast examination is realized. Two different radiologists assure readings. In the last three year, the national participation rate remains stable around 51%. In addition, studies, using ecological indices of deprivation, highlight social and territorial inequalities: women living in disadvantaged area or far from a radiologist's office less participate. Different strategies can be implemented to increase participation to breast cancer screening, especially in populations with low-income or in rural areas, who have less access to screening. Among these strategies, mobile mammography units (MMU) seem effective. Therefore, in the French context, there is very little evidence on how to intervene to reduce health socio-territorial inequalities in screening, so public health decision-makers are unable to base proposals on evidence.

Intervention The main intervention will be a proposition for an appointment at the MMU in complement to the current OBCS. This complementary mode of screening offers the possibility for women furthest away from the radiologist's office to undergo screening in the MMU. All women eligible for OBCS and living in the zone selected for the intervention will be invited to participate either in a radiologist's office or in the MMU, keeping the choice of their place of screening. The breast screening in the MMU will follow the same protocol as screening at the radiologist's office according to national recommendations, so that the quality level of screening proposed in the MMU will the same as in any radiologist office. The MMU will be equipped with a latest-generation digital scenographer as well as an ultrasound system according to the compliance rules of the specifications published by the National Institute of Cancer. The quality of the mammographic radiographic equipment will also be certified by the Nuclear Safety Agency.

About two weeks before the MMU is parked, local actors, will carry out actions to inform women on the organized breast cancer screening and on the MMU. Women will receive the timetable of these actions with the invitation to participate in screening.

Design This intervention will be conducted during two years in 2022-2024, as a prospective randomized controlled cluster trial in the general population in the departments of Calvados, Manche, Eure and Seine-Maritime (Normandy Region in the north west of France). The cluster of the study is a group of IRIS (Ilots Regroupés pour l'Information Statistique). This geographical unit is the smallest scale for which census data are available and represents an average of 2000 inhabitants.

An algorithm was developed to constitute clusters integrating some constraints in a regional scenario.

Aggregation of IRIS was done according to the travel time to the radiology centres, the most distant IRIS has been selected and then merged with neighbouring IRIS, still by distance run, until reaching areas of the expected population size. This except size is around 100 women. Expecting a 40% participation to the MMU, 32 to 48 mammograms daily should be performed.

This algorithm was applied to all IRIS according to distance rank, until no more aggregation was possible. 91,982 women (95.6%) and 1,067 IRIS (94.3%) were selected in the final population in 356 clusters (with 258 created by the algorithm).

All the women constituting the target population is geolocated and geocoded. Thus, for each woman, history of screening, the current date of screening invitation, IRIS of residence, level of social deprivation according to an ecological deprivation index (the European Deprivation Index) and the distance between the woman's house and the nearest approved radiologist's office are known.

A randomization in parallel group have been perform to constitute the intervention arm (n=178 clusters) and the control arm (n=178 clusters), corresponding to 45275 in intervention arm vs 46707 in control arm.

Evaluations The overall objectives of the study are to evaluate the intervention's ability a) to reduce the socio-territorial inequalities of participation in breast cancer screening in a regional area (Normandy) in setting remote from radiologist's office b) to increase participation rate and c) understand how intervention interact with contextual factors and which causal mechanisms leading to these results in order to identify the optimal modalities for extending the MMU at the national level.

Concerning the first two objectives, the main evaluation criterion will be participation in screening, measured and compared between the "intervention" and "control" arm, at the aggregate level (cluster) and individual level in intent to treat. At the aggregate level, the comparison of screening participation will allow us to measure the raw and age-standardized increase in participation due to the intervention. At the individual level, multilevel logistic regressions, taking into account cluster data, allow to assess the increase in the probability of participating in screening after adjustment of the social deprivation, age and other available individual characteristics.

Although the calculation of the number of subjects in the study came from a pragmatic approach, the intraclass correlation factor was estimated at 0.0083, which with an average area size of 318.6 women gives a design effect equal to 3.63. Thus, the minimum significant difference in participation that can be proven will be 1.5%.

Actors information action, will be prospectively registered in a dedicated database and will be include in the global model analysis.

To furthermore explore contextual effect and to provide elements of knowledge on the causal mechanisms contributing to the effectiveness of the intervention the quantitative analyses will be completed by a qualitative approach inspired by the combined approaches that have emerged in realistic randomised controlled trials, using a theory that goes beyond logic models to describe contextual mechanisms and contingencies. An intervention theory was developed and three auto questionnaires will respectively explore, the informed choice, satisfaction to realized the mammography in the mammobile and the ComB Model construct implied in the screening behaviour.