Associations between area-level deprivation, rural residence, physician density, screening policy and late-stage colorectal cancer in Canada

Background: Important social disparities in colorectal screening exist in Canada. Few studies have assessed disparities further along the cancer control continuum. Here we assess the associations between social and material deprivation, physician density, rural residence, and the absence of provincial mail-based screening programs and colorectal cancer (CRC) stage at diagnosis. Methods: Colorectal cancer cases and stage data (local or regional if stage 0, I, or II, late if stage III or IV) were obtained through the Canadian Cancer Registry (2011-2015, N = 54,745). Cases were linked to 2006 Canadian Census Dissemination Area-level data on rural/urban status, exposure to a provincial mail-based screening program, and social and material deprivation (Pampalon Index quintile groups); and to Scott's Medical Database 2011 physician density data (< 7 vs. >= 7/10,000). Age, sex, and predictor-adjusted Generalized estimating equation (GEE) Poisson models were used to determine independent associations between predictors and late-stage at diagnosis. Results: Half of CRC cases are diagnosed at stage III or IV (51 %), with younger age groups experiencing higher late-stage prevalence (57 % among those aged 18-49). The covariate-adjusted late-stage prevalence was 2-percentage points higher in most materially- and socially-deprived areas (95 % CI: 1 %, 4 %, in both, respectively) and in provinces with no mail-based screening programs (95 % CI: 1 %, 2 %). No significant differences were observed according to rural residence or physician density. Conclusions: Social disparities in late-stage CRC diagnosis are modest. Continued surveillance of these disparities may be warranted as provinces continue to promote early cancer detection through screening, and stage distributions may change overtime.