**2. Methods**

The CCR is a population-based database comprised of data annually collected and reported to Statistics Canada by each provincial and territorial cancer registry. Demographic information regarding the individual diagnosed with cancer, and characteristics of the cancer itself, are available for each new primary case. Cancer-specific information, including stage at diagnosis, is available for common cancers, including BC [42]. Individual provinces and territories have varying practices for screening women aged 40–49, ranging from organised screening programs with annual recall to recommendations against screening in several provinces [43]. The Canadian Community Health Survey (CCHS) is a national crosssectional survey that allows the determination of screening mammography activity [44] with acknowledged inherent bias [45]. Because health care in Canada is publicly funded, regular screening activity was mostly captured in the population-based screening programs, unless screens were performed outside of the organised programs or recorded as diagnostic mammograms. Although information about the method of detection of BC is not available, combining provincial-level BC data with the presence of organised programs and screening activity allows for a comparison of screening policies on stage at diagnosis in age cohorts 40–49 and 50–59.

This study was a secondary analysis of nationally de-identified data collected by Statistics Canada, and as such, ethics approval was not required. Female BC incidence data were obtained from the CCR file released on 19 May 2021 [46]. This version of the CCR included primary invasive cancer cases diagnosed among Canadian residents from 1992 to 2018, although cases diagnosed in Quebec from 2011 onward had not ye<sup>t</sup> been submitted. The analytic file used followed the multiple primary coding rules of the International Agency for Research on Cancer (IARC) [47]. Full staging data were available from 2010 to 2017. Stage data used collaborative stage; a comprehensive standardised system sponsored by the American Joint Committee on Cancer (AJCC) which is compatible with the other staging systems in use during that period [48]. Unstaged BC cases were excluded from our analysis. There was no information on breast density or ethnicity.

Results for women aged 40–49 at diagnosis were compared to those for women aged 50–59 to assess the impact of screening policies in younger women. The 50–59 age group was chosen because it is the closest age group to 40–49 for which women in all jurisdictions may undergo regular screening mammography

Provincial and territorial screening practices varied across the country (Table 1). Those jurisdictions with screening programs that allowed women to access BC screening in their 40s by self-referral, and subsequently followed these women with annual recall, were designated as screeners [49,50]. Five screener jurisdictions were identified: Nova Scotia, British Columbia, Alberta, Prince Edward Island, and Northwest Territories. Alberta allowed self-referral in 2007, but by 2012 required a physician referral for the first screen. BC changed from annual to biennial recall in 2014. The other jurisdictions collectively formed the comparator group. Quebec was necessarily excluded from this group due to the absence of incidence data from this province in the CCR for the study period. Manitoba had biennial recall, and after the study period Yukon began to send anual reminder letters. In provinces that required a physician referral, some 40–49-year-old women screened may have had a family history of breast cancer that led to the referral. Data from Statistics Canada's nationally representative CCHS were used to determine the percentage of women aged 40–49 who reported having a screening mammogram in the previous two years [51]. This yielded screening participation rates that were independent of provincial/territorial screening programs. Despite this, these jurisdictions were not included in the "screener" group based on the a priori definition.


**Table 1.** Breast cancer screening information and participation rates by province and territory, women aged 40–49, Canada, selected years.

n/a = no screening data available. Note: screening percentage rates refer to the percentage of women 40–49 with a screening mammogram in the previous two years, based on data from the Canadian Community Health Survey. Shaded provinces and territories denote screeners while the non-shaded ones are the comparators. Source: Canadian Community Health Survey, Cycle 2.1 (2003), Annual Component (2008, 2012, 2017); provincial and territorial screening practices [49,50]. Screening program information changed in some jurisdictions throughout the study period: \* BC changed from annual to biennial recall in 2014. \*\* Alberta changed from self-referral to requiring an MD referral for the first screen in 2012.

Annual average percent changes in age- and stage-specific female BC incidence rates between 2011 and 2017 were calculated using JoinPoint 4.9.0.0 [52], which fit a piecewiselinear regression model that assumed a constant rate of change in the logarithm of the annual incidence rate. The year 2011 was chosen as the starting point for this trend analysis because it corresponded with the CTFPHC recommendation against screening women aged 40–49 years. Because the incidence of BC is lower in women in their 40s than in their 50s, the relative proportions of BC stages at diagnosis for the period from 2010 to 2017 were compared between screener and comparator jurisdictions, and statistical significance was calculated using z-tests. Standard errors for incidence rates were derived directly from the Poisson distribution, whereas those for proportions were calculated using the Agresti–Coull method [53].

A linear regression analysis was used to assess the relationship between self-reported jurisdictional screening percentages for 2012 and BC incidence rates for the period from 2011 to 2013. The analysis was restricted to early stage BC, defined by TMIST (AJCC 8th ed) as stage I, and the most advanced BC as stage IV. This time period encompasses the 2012 two-year time frame for reported mammography. Stage migration in women aged 50–59 was investigated by using provincial screening status for women aged 40–49 and comparing proportions of BC stage at diagnosis for women aged 50–59 in screener and comparator provinces. P-values correspond to two-sided tests of the null hypothesis that there was no difference in stage distribution, with a significance level of 0.05.
