This repeated cross-sectional study showed that the “13 en Santé” health mediation program successfully targeted individuals facing cumulative socioeconomic barriers to accessing health care (i.e., non–French native speakers, those with the lowest levels of education, and those with the highest rates of unemployment or homemaking)²³. These individuals reported the lowest access to a primary care physician and the lowest receipt rates of cancer screening invitation letters. The reduced receipt rate may reflect an actual lower likelihood of receiving the letter — for example, due to unstable housing — or difficulties in understanding the content of the invitation. Additionally, this study shows that health mediation may significantly increase individual uptake of breast and colorectal cancer screening, with a 20-percentage-point gain. As demonstrated by the multivariate analysis, this effect may suggest a complementary role of health mediation alongside access to a primary care physician and receipt of the invitation letter.

The 20-percentage-point increase in cancer screening uptake associated with health mediation is consistent with previous studies, although slightly higher than reported median effects. A 2023 systematic review of CHW interventions effectiveness reported a median increase of 11.5 percentage points for breast cancer screening (interquartile range: 5.5–23.5) and a slightly lower median increase of 10.5 percentage points for colorectal cancer screening (interquartile range: 4.5–17.5)¹³. In contrast, despite an upward trend, no significant effect was observed in our study, although this review also reported a median increase of 12.8 percentage points (interquartile range: 6.4–21.0) for cervical cancer screening. A 2017 meta-analysis similarly demonstrated beneficial effects of community-based interventions across breast, colorectal, and cervical cancer screenings (OR=1.90 [1.60–2.26])²⁴. Reported encounter with a health mediator was associated with receipt of screening invitation only for cervical cancer, which may suggest less precise targeting toward women already engaged with preventive care. Consistently, women aged 30–39 — who are more likely to have regular gynaecological follow-up related to childbearing — appeared more often up to date than other eligible women, indicating that targeting younger and older age groups may enhance intervention effectiveness.

By combing multicomponent interventions, the “13 en Santé” health mediation program may promote cancer screening through multiple mechanisms of action and by mobilizing diverse levers. First, providing tailored educational sessions or information — using motivational interviewing techniques — may enhance intrinsic motivation by addressing three fundamental needs (competence, autonomy, and relatedness) and can be further reinforced through navigation support when needed^24,25. Second, navigation support helps address contextual barriers — whether perceived or real, social or related to accessing the healthcare system — that may hinder individuals from translating intention into action^25–27. As a matter of fact, health mediation appeared associated with increased awareness only for colorectal cancer screening, whereas breast cancer awareness was already very high in our sample. This suggests differential mechanisms of action: for colorectal cancer, both raising awareness through tailored education and providing navigation support to overcome barriers appear necessary, while for breast cancer, the intervention may act primarily by helping individuals resolve practical barriers rather than by increasing awareness.

No effect of health mediation or primary care physician access was observed on cervical cancer screening uptake. The absence of an effect of primary care physicians on cervical cancer screening in French urban settings is consistent with litterature²⁸. This finding highlights the central role of specialists — gynaecologists and midwives — in this screening pathway and underscores that its uptake driving mechanisms differ from those for breast and colorectal cancers. Further research is needed to identify the specific mechanisms that should be integrated into health mediation strategies to improve their effectiveness in promoting cervical cancer screening.

This study has a major strength in enabling assessment of the intervention through exposure defined at both the individual and neighbourhood levels, allowing simultaneous evaluation of individual- and population-level effects.

It also has several limitations. First, although the intervention was concentrated within the designated intervention neighbourhoods, it may not have been sufficient to prevent dilution effects across time and space. In the northern intervention neighbourhoods — where activities were primarily conducted through workshops and focus groups — health mediators may have interacted with individuals residing in adjacent areas, including control or non-included neighbourhoods, thereby diluting the measurable effect of the intervention. Second, although 4,523 individuals were interviewed, it represented only about 6% of the total population of the included neighbourhoods. Third, the T1 follow-up survey was conducted in two phases for operational reasons (i.e., completion of missing questionnaires in Phase 2 due to time constraints during Phase 1), which may have influenced observed screening uptakes. Although survey phase was accounted for in the statistical models, this two-phase data collection may have reduced statistical power to detect population-level effects. Combined with the potential dilution effects described above and the limited population coverage, this may have constrained the study’s ability to identify an effect at the population level. Further studies using more comprehensive medico-administrative data from the French National Health Insurance are underway to confirm these findings.

In addition, questionnaire administration relied on door-to-door visits, which aligned with the main outreach strategy in the city centre but not in the northern neighbourhoods. This discrepancy may partly explain the lower reported encounter rates in Northern areas, as interviewed individuals may have been less representative of the populations reached by the intervention, thereby potentially limiting the observed effectiveness of the program.

Other limitations are inherent to surveys based on self-reported outcomes. First, recall bias may have led to underestimation of health mediator encounters, limiting the assessment of intervention effects. Second, social desirability bias may have resulted in overestimation of screening uptake rates. Finally, a limitation of this study is inherent to observational quasi-experimental design, which does not allow for establishing a causal relationship between exposure to the intervention and increasing screenings uptake.