Changes in mammography screening in Ontario and Alberta following national guideline dissemination: an interrupted time series analysis

Background: In November 2011, the Canadian Task Force on Preventive Health Care released guidelines for screening women at average breast cancer risk. Weak recommendations (framed using GRADE methodology) were made for screening women aged 50 to 74 years every two to three years, and for not screening women aged 40 to 49 years. Methods: We conducted an interrupted time series analysis using administrative data to examine bilateral mammography use before and after a release of a national breast screening guideline. Women aged 40 to 74 years living in Ontario or Alberta from 30th November 2008 to 30th November 2014 were included. Strata included age, region of residence, neighbourhood income quintile, immigration status, and education level. Results: In both provinces, mammography use rates were lower in the post-intervention period (527 vs. 556 and 428 vs. 465/1000 women in Ontario and Alberta, respectively). In Ontario, mammography trends decreased following guideline release to align with recommendations for women aged 40 to 74 (decrease of 2.21/1000 women, SE 0.26/1000, p<0.0001). In Alberta, mammography trends decreased for women aged 40 to 49 years (3/1000 women, SE 0.32, p<0.001) and 50 to 69 (2.9/1000 women, SE 0.79, p<0.001), but did not change for women aged 70 to 74 (0.7/1000 women, SE 1.23, p=0.553). In both provinces, trends in mammography use rates were sustained for up to three years after guideline release. Conclusions: We observed a decrease in screening for women aged 40-49. Additional research to explore whether shared decision making was used to optimize guideline-concordant screening for women aged 50-74 is needed.


Introduction
The Canadian Task Force on Preventive Health Care (Task Force) was reconstituted in 2009 to develop and disseminate evidence-based clinical practice guidelines to support preventive practices in primary care. Breast cancer screening was the first topic selected for guideline development because the U.S. Preventive Services Task Force 1 had produced recommendations in 2009 favouring screening for women aged 40 years and older, yet questions about costs and benefits remained given the lower rate of breast cancer at younger ages and the potential risk of harms. 2 Breast cancer is the most common cancer in Canadian women with one in eight women receiving a diagnosis over her lifetime and one in thirty-three dying from the disease. In 2020, there were an estimated 27,400 new cases and 5,100 deaths from breast cancer, which is 13% of all cancer deaths in women in Canada. Mammography can identify asymptomatic breast cancer, but there are harms associated with screening, including overdiagnosis. 3 Several provinces and countries have implemented breast cancer screening programs with established targets; for instance, Alberta Health Services identified a target of screening 70% of eligible women aged 50 to 74 years every two years, which is consistent with the Canadian target of screening 70% of eligible women every 30 months.
In 2011, the Task Force aimed to provide a recent review of evidence to inform breast cancer screening practices. In keeping with the use of Grading of Recommendations Assessment, Development and Evaluation (GRADE) methodology, the Task Force reviewed available evidence using a systematic review to develop screening guidelines for women at average risk of breast cancer (see Box 1 for recommendations). 4 The Task Force developed weak recommendations (defined using GRADE methodology) based on moderate to low quality evidence. In keeping with GRADE methods, weak recommendations are those for which the benefits probably outweigh the harms, or the harms probably outweigh the benefits, but uncertainty exists. 4, 5 The guidelines also highlighted the need for clinicians to used shared decision making in order to help patients make a decision that is consistent their informed values and preferences. 4 This guideline • For women aged 50-69 years we recommend routinely screening with mammography every two to three years. (Weak recommendation; moderate quality evidence) • For women aged 70-74 years we recommend routinely screening with mammography every two to three years. (Weak recommendation; low quality evidence)

REVISED Amendments from Version 1
The following changes were made to the manuscript following the peer review comments: we clarified data sources (specifically how the OCR cancer registry was used and which women were excluded from analysis), we corrected how the interrupted time series was presented (now shown as rates per 1000 women), we added clarifications around the use of GRADE methodology classifications of 'weak recommendations', we have clarified the study purpose and aims, we have reorganized the presentation of the methods section to improve clarity, we have changed the presentation of p values (0.0000 revised to <0.001 in study tables), and we have made minor edits to wording throughout the manuscript to ensure consistency of presentation in the abstract, results, discussion and conclusion.
Any further responses from the reviewers can be found at the end of the article was an update to the 2001 guideline published by the Task Force which recommended screening with mammography every one to two years for women aged 50 to 69. 6,7 The Task Force developed a tailored guideline dissemination strategy in partnership with stakeholder groups, including primary care physicians and women in the target age groups. We evaluated the impact of this dissemination strategy by investigating mammography use in Ontario and Alberta before and after the strategy was implemented.
The objective of this study was to determine, using an interrupted time series, the impact of the 2011 Task Force breast cancer screening guidelines and corresponding national dissemination strategy on rates of mammography screening by age group in Ontario and Alberta. We hypothesized that there would be a significant change in mammography screening rates over time to align with the Task Force breast cancer recommendations, and that effects would be sustained for up to three years post-guideline release.

Setting and participants
We performed an interrupted time series (ITS) analysis to examine mammography use in Ontario Women with a history of breast cancer and/or a bilateral mastectomy before each period were excluded from the analysis. Bilateral mastectomy was identified using the Canadian Classification of Health Intervention (CCI) procedure codes. In Alberta, the Alberta Kidney Disease Network (AKDN) database (incorporating data from Alberta Health), which includes >99% of adult Albertans, was used to identify participants. 10 MT had access to de-identified data in the AKDN database; specifically, no health care numbers or names were available. For Alberta data, a validated administrative algorithm was used to identify and exclude women with a history of breast cancer (i.e., exclusion of women with hospitalization of breast cancer or malignant neoplasm of the breast, using ICD-9 174 and ICD-10 C50 codes and 1 hospitalization or 2 claims in 2 years). 11,12 Procedure codes from claims data (CCP codes 97.12A, 97.12B, and 97.22A) were used to exclude women with previous bilateral mastectomies (extended data).

Ethics statement
This study was approved by the Unity Health Toronto Research Ethics Board (REB# 12-220). Participant consent was waived by the REB, given all data analyses were conducted by ICES and Alberta Health Services using aggregate, nonidentifiable data.

Intervention
The intervention was the Task Force's guideline dissemination strategy, which included theory-and evidence-based printed and online education materials and a mass media campaign. [13][14][15][16][17][18] The content of the education materials was based on the guideline recommendations, discussions with stakeholders about barriers to implementing the recommendations, and barriers to guideline implementation. The mass media campaign focused on key messages for the target audiences as identified from discussions with the Task Force and its stakeholders. The detailed approach to developing and testing the education materials is available online. The guidelines were published in the Canadian Medical Association Journal (CMAJ) on 21 st November 2011 accompanied by a mass media launch. 4 The print CMAJ (including the guidelines) was mailed to over 60,000 primary care and specialty physicians nationally. The printed education materials provided in this mailing included a one-page decision algorithm for patients and clinicians, and a one-page 'Frequently Asked Questions' document for patients and clinicians. These education materials were also available on the Task Force's website along with a patient decision aid infographic explaining the risks and benefits of mammograms, a video and script that role modelled shared decision making (the process for patients and physicians to collectively make a decision regarding treatment), a presentation on the methods used to create the guidelines, and the systematic review used to inform guideline development. 19 Outcomes Data were cleaned to ensure that participants were adult and female. The Ontario and Alberta datasets were analysed separately and were not linked. NW conducted and oversaw the analysis of the Alberta data and RN oversaw the analysis of the Ontario data. Use of bilateral mammogram was the primary outcome, identified from the Ontario Health Insurance Program (OHIP, fee code in 'X185' and 'X178') and Ontario Breast Screening Program (OBSP) databases in Ontario and the AKDN claims data (CCP codes X27C, X27D, X27E) 10 in Alberta. For both Ontario and Alberta data, a mammogram was considered a duplicate if repeated within seven days and was then removed from the data; the date of the first mammogram within each time period was used for analysis. Additional variables collected included age group, setting (rural versus non-rural), income quintile, immigrant status, visible minority status 1 , and education status. Urban area was defined as having a population of at least 1,000 and a density of 400 or more people per square kilometre. Income quintile was used as a proxy for the women's socioeconomic status and was based on their postal code (neighbourhood) using Statistics Canada census data. Immigrant status and visible minority status were defined as the percentage of immigrants and visible minorities in the postal code, respectively. Education was specified as the proportion of women with a bachelor's degree as the highest level of educational attainment. Details on immigrant, minority, and education status were not available for Alberta participants.

Statistical analysis
We used an ITS to longitudinally assess how the outcome of mammography use was impacted by the guideline release.
Pre-guideline and post-guideline data were summarized descriptively using mammography counts and rates per 1000 women eligible at the end of each period (30 th November 2011 and 30 th November 2014, respectively). Data from each province were aggregated monthly to calculate rates per 1000 women defined as (number of monthly screening tests among eligible women)/(total number of women eligible each month * 1000, and ITS was performed using segmented regression. The model was defined as: where Rate t is the rate of breast cancer screening at each month t (t = 1 to 72 months) and e t is the random error at time t. β 0 is the intercept or pre-intervention baseline level of the rate; β 1 is the slope of the rate until the intervention is introduced (slope prior to the intervention); β 2 is the change in rate level that occurs in the period immediately following the introduction of intervention; and β 3 is the change in rate trend (i.e. difference between post-intervention and preintervention slopes). We used this approach to test the significance of β 2 (an immediate intervention effect) and β 3 (significant intervention effect over time to determine sustained impact). 21 Subgroup analyses were performed for age categories. In Ontario, subgroup analyses were also separately performed by rural or urban setting, neighbourhood income quintile (mean income per person equivalent in an area obtained from census data), education categories (population in a neighbourhood with a bachelor's degree), immigration categories (population in a neighbourhood of immigrant status), and minority categories (created as a % of visible minority in a total population by postal code). The cut off values for categories were determined using distributions of the 1 st and 3 rd quartiles (interquartile range) of the neighbourhood distributions at pre-and post-intervention periods.
For Ontario data, regression parameters were estimated by maximum likelihood using SAS Enterprise Guide, version 7.15 software procedure AUTOREG (SAS Institute Inc., Cary, NC, USA), which adjusts for autocorrelation. 22,23 For Alberta data, analysis was conducted using Stata MP 15.1 software command 'itsa'; however, the same models and parameters were used. 24 Canadian mammography screening is recommended every 24 to 36 months 25 ; final models included autogressive terms for only up to 12 lags, selected by backward elimination. The Durbin-Watson statistic was calculated to test autocorrelation terms. All statistical tests were two-sided and statistical significance was defined as a p-value less than 0.05.

Ontario Descriptive
In Ontario, a total of 2,935,241 women were included before guideline dissemination and 3,098,205 women were included after dissemination; the study populations had similar demographic characteristics (Table 1). Overall, the number of women who had a mammogram decreased in the post-intervention period compared to pre-intervention (527 versus 556 per 1000 women, respectively). Similar trends were observed in women aged 40 to 49 and 50 to 69 years. 1 'Visible minority' is a term used in the ICES database. 20 The proportion of women aged 70 to 74 years who underwent a mammogram was slightly higher in the post-intervention period (Table 1). Both at pre-and post-guideline release, women in higher income quintiles (compared to lower quintiles) and women in rural areas (compared to urban areas) had higher screening mammography use.

ITS results
Among women aged 40 to 49 years, monthly screening mammography rates increased before the guideline release (slope 0.06/1000, SE 0.01/1000, p < 0.0001). Immediately following the guideline release, there was a significant rate decrease of 2.21 per 1000 women (SE 0.26/1000, p < 0.0001) who received a mammogram. This rate was sustained; however, the slope was not, meaning the rate of women who received a mammogram did not continue to decrease further during this period (Table 2, Figure 1). Among women aged 50 to 69 years, monthly screening mammography rates decreased before the guideline release ( Table 2). Immediately following the guideline release, there was a significant increase of 3.82 per 1000 women who received a mammogram. This rate was sustained; however, the slope was not.
Among women aged 70 to 74 years, monthly screening mammography rates decreased before the guideline release ( Table 2). Immediately following the guideline release, there was a significant increase of 7.50 per 1000 women who received a mammogram. This rate was sustained three years following the intervention.
Among all women, rates of mammography screening were significantly declining pre-guideline release; following the release, a significant decrease in mammography use was observed in all income quintiles and across all education levels. There were no significant changes to the slopes in the three years following the guideline release. Similar trends (significantly decreasing slope, increase in mammography use, no significant difference in slope three years after guideline release) were observed across neighbourhoods with a low, moderate, or high percentage of women who were immigrants or identified as a visible minority and among women in urban and rural areas ( Table 2).

Alberta Descriptive
A total of 788,131 women were included pre-intervention and 861,014 women were included post-intervention; demographic details are provided in Table 1. Overall, mammography use decreased post-intervention compared to pre-intervention (428 versus 465 per 1000 women, respectively).
These trends were observed in women aged 40 to 74 years. In both the pre-and post-intervention periods, women in higher income quintiles had increased rates of mammography use compared to women in low income quintiles (Table 1). In Alberta, women residing in urban areas had higher rates of mammography use pre-and post-intervention as compared to women residing in rural areas.

ITS results
Among women aged 40 to 49 years, monthly screening mammography rates in Alberta were slightly decreasing before guideline release. Immediately following guideline release, there was a significant rate decrease of 3 per 1000 women who received a mammogram. This rate was sustained; however, the rate of women who received a mammogram did not decrease further during this period (post-minus pre-intervention slope difference 0.0015/1000, SE 0.02/1000, p = 0.933).
Among women aged 50 to 69 years, monthly screening mammography rates were stable. Immediately following the guideline release, there was a significant decrease of 2.9 per 1000 women who received a mammogram; this rate was sustained for the remainder of the study period ( Figure 2, Table 3).
Among women aged 70 to 74 years, monthly screening mammography rates were stable before the guideline release. Immediately following the guideline release, there were no significant changes in the rate of women aged 70 to 74 who received a mammogram.
All strata (by age, income quintile, rural/urban region) experienced significant drops in mammography rates at intervention, with the exception of women 70 to 74 years. Prior to the intervention, most slopes were stable pre-and post-intervention with the exception of women 40 to 49 years (rate decreased immediately post-intervention) and women in high income quintiles, who demonstrated a slight decrease in mammography use prior to guideline release (Income Q5, see Table 3). Trends were stable three years following guideline release, with the exception of women living in rural areas, for whom mammography rates continued to decrease over time.

Summary of results
In comparing rates of mammography screening following guideline dissemination, we observed trend differences in the two provinces. Overall, Alberta women underwent less mammography screening both pre-and post-guideline dissemination compared to Ontario women. In both provinces, fewer women underwent mammography in the 36 months after guideline dissemination compared to the same period before guidelines were released. In Ontario, mammography rates among women aged 40 to 49 years decreased immediately following guideline release and increased for women aged 50 to 74, in keeping with the guideline recommendations. In Alberta, mammography use decreased for women aged 40 to 69 and did not significantly change among women aged 70 to 74. In Ontario, mammography use decreased overall following guideline release across all income quintiles and education levels and among women living in both urban and rural areas. Similarly, in Alberta, mammography decreased among women in all income quintiles and decreased for women living in both urban and rural areas; the rate of mammography use continued to decrease among women living in rural areas in Alberta.

Discussion
In both Ontario and Alberta, mammography among women aged 40 to 49 years immediately decreased following guideline release. This decrease may correlate with the difference in screening recommendations in the 2011 guideline compared to the previous 2001 Task Force guideline, which suggested that upon reaching the age of 40, women should consider the benefits and risks of screening to determine whether to begin routine mammography between 40 and 49 or whether to delay until age 50. 7 Similarly, rates of mammography among women aged 50 to 59 in Alberta may have decreased following the recommendation for extending screening intervals to 2 to 3 versus 1 to 2 years. 25,26 Our Higher income levels were associated with more screening. These results are similar to those from other studies of breast cancer screening conducted in the U.S. 30-34 and Canada. 35 In Ontario, women living in rural areas had higher mammogram screening rates compared to women living in urban areas; this may indicate the effect of organized screening programs in rural communities. Additional studies further examining trends in Canadian breast cancer screening, particularly following the most recent 2018 release of the Task Force's breast cancer guideline update, may provide further insights into these trends.

Limitations
There are several limitations to this study. Firstly, individual level data (such as exact age, education status, ethnicity) were not available; rather, these data were available as categories, and education and immigration status were only available at a neighbourhood level. Consequently, we were not able to investigate the impact of patient level factors on screening activities. Secondly, there may have been other interventions targeting breast cancer screening in Ontario and Alberta during the study periods. In Ontario, for instance, Cancer Care Ontario launched a campaign to increase screening for breast, cervical, and prostate cancer in November 2012. This media campaign encouraged people to send cards to friends and family to undergo cancer screening. However, the Task Force engaged provincial partners on the guidelines and we were not able to identify any large-scale strategies that may have overlapped with dissemination. An additional confounder may be that pay-per-performance models (meaning practitioners receive compensation based on their activities, such as screening, and not a standard salary) were introduced in Ontario at this time; however, the models don't definitively explain the trends observed in this study, as the literature is not conclusive on the impact of these models on rates of breast cancer screening. 36,37 In addition, we were not able to assess the use of shared decision making, which was recommended in the guideline, as these data are not captured in administrative databases. Finally, these data were not created to answer the specific research question, which is a limitation of all population-level administrative database studies.

Conclusion
In summary, the decline in mammography for women aged 40 to 49 years was in alignment with guideline recommendations. Future strategies should focus on optimizing implementation of the recommendations for women aged 50 to 74 years. This requires understanding barriers to behavioural changes at the clinician and patient level and investigating the use of shared decision making.

Underlying data
The datasets generated and analysed during the current study are not publicly available. They could potentially be made available with assistance from the corresponding author, if for the purpose of further research. Obtaining Ontario data which are owned by the Ontario Ministry of Health and Long-Term Care would require providing a research protocol and consultation with ICES, who will also request a budget be submitted for data access and analysis. Researchers could alternatively request to obtain a similar dataset for Alberta or Ontario data.

Disclaimer
This study is based in part by data provided by Alberta Health and Alberta Health Services. The interpretation and conclusions contained herein are those of the researchers and do not represent the views of the Government of Alberta or Alberta Health Services. Neither the Government of Alberta nor Alberta Health Services express any opinion in relation to this study. The purpose of the paper is not well-developed. There is mention of a dissemination strategy, which is discussed in the methods, but does not seem to be tested for or mentioned in the conclusion. If there was no comparison for differences in dissemination strategies, then it appears that the manuscript evaluates the effect of the guideline change on mammography rates, but not dissemination strategy of the guidelines. We recommend changing the second to last paragraph in the introduction to reflect that dissemination 2.
We confirm that we have read this submission and believe that we have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however we have significant reservations, as outlined above. 2. The purpose of the paper is not well-developed. There is mention of a dissemination strategy, which is discussed in the methods, but does not seem to be tested for or mentioned in the conclusion. If there was no comparison for differences in dissemination strategies, then it appears that the manuscript evaluates the effect of the guideline change on mammography rates, but not dissemination strategy of the guidelines. We recommend changing the second to last paragraph in the introduction to reflect that dissemination strategy was not assessed and the methods section to reflect that it was the change in guidelines that was assessed rather than the dissemination strategy.
Thank you for this important point. The dissemination strategy was passive and was disseminated via avenues considered 'routine' for primary care practitioners (including via a leading Canadian medical journal, at conferences, etc.) The aim was to determine mammography trends following release of the guideline and the dissemination strategy, rather than test the dissemination strategy in a controlled study design. We recognize that the findings may not be causally linked to the dissemination strategy, and we have amended our introduction to clarify the purpose and outline potential confounders in our limitations section. We have also clarified the study objective in the introduction. Abstract 1. Amend description of "weak recommendations" to "recommendations based on weak evidence." In order to remain consistent with GRADE language, we have revised this sentence to "weak recommendations based on GRADE terminology (i.e., the benefits and risks are closely balanced and/or uncertain).

2.
Recommend removing information about dissemination strategy in line with recommendations above.
Thank you, we have removed "dissemination strategy targeting primary care physicians" from the abstract and the sentence in the abstract now reads "We conducted an interrupted time series analysis using administrative data to examine bilateral mammography use before and after release of a national breast cancer screening guideline".
3. Recommend clarifying that "trends in mammography use rates" be changed to "declining trends in mammography use rates" in last sentence of results in abstract. Consider removing standard errors to stay within word limit.
Thank you for this comment. The trends did not decline in all cases. While we saw a decrease in screening for all women in Ontario, and for women aged 40-69 in Alberta, we did not see a decrease among women aged 70-74 in Alberta.
4. Recommend changing conclusion to indicate that it was the guideline that changed uptake, rather than the dissemination strategy.
Thank you, we have revised the concluding sentence.
5. The conclusion does not match the results in the abstract. The results indicate a decrease in mammography trends, but the conclusion indicates that there was an increased uptake in guideline-concordant screening practice. As this was not actually assessed (would require individual level data) the conclusion would be more accurate if discussed along the lines of, "decreasing trends in mammography rates among 40-49 year old women indicated that this group may be receiving fewer screenings, possibly in response to the new guidelines." Thank you for this comment. We have revised the wording as suggested by the reviewer.

Introduction
1. The dissemination strategy was not elaborated on in the introduction. Since this study did not test the dissemination strategy, it is recommended that the description of the strategy be moved from Methods section to the Introduction section.
Thank you; our goal was to evaluate the impact of guideline release, which was disseminated using a national strategy. We thus have kept the description of the strategy in the methods section. We recognize there are limitations to drawing conclusions of impact of the strategy, which we have outlined in the limitations.
2. Change "weak recommendations" to "recommendations based on weak evidence." Please see comments above regarding consistency with GRADE language.

Methods
We recommend reorganization of the methods section.
1. Begin with the setting and participants with integration of the dates of data collection in the current first sentence, followed by the ethics statement then Outcomes and statistical analysis.
Thank you for this suggestion. We have reorganized the methods section accordingly.
2. We recommend that the description of the dissemination strategy be shortened and included in the Introduction section.
Thank you, please see above comment for our rationale.
3. We recommend that the statistical analyses used (ITS) be moved to the statistical analyses section. The method used for this study looked like a retrospective population surveillance study.
Thank you, we have provided details on the ITS in the statistical analysis section.
4. Provide a justification for using interrupted times series analyses in the statistical analyses section.
Thank you, this has been provided.

1.
Were the women included in the study cohorts? This was mentioned in sentence 1 of Results section. With shifting ages across the timeline, it seems they could be described as the "study population," as women would shift into different age groups across the times assessed.
Thank you for this comment. We have revised "cohorts" to "study populations".
2. Recommend using p-value designation of <0.001 in last column of tables where currently 0.0000 is listed, as these are more accepted statistical notations in the literature.
Thank you for this suggestion, we have updated the tables accordingly.
3. In the Alberta results section, Alberta was mentioned in the text as well as subheading.
Would be more clear if the Ontario section was written in similar fashion. At first I thought that the total population estimates were for the entire study, not just Ontario.
Thank you for this suggestion, we have added "in Ontario" to provide clarity for this section.

Discussion
1. It appears new results are added in first paragraph. We saw no evidence of direct comparisons of mammography rates between Alberta and Ontario in the results section.
Recommended to amend statistical analyses section and results section accordingly. The first paragraph of the Discussion section could then be the final paragraph of the Results section.
Thank you. We have moved the first paragraph of the discussion to the results.
2. Last sentence in 3rd paragraph not clear. If fewer women decided not to undergo screening, wouldn't the trend have been positive after guideline implementation?
We observed a decrease in screening rates for all women, except for those aged 70-74 in Alberta.

Consider subheadings for Limitations and Conclusion sections.
Thank you for this suggestion. We have added subheadings for Limitations and Conclusions.
4. The Conclusion section in the Discussion appears to differ from the conclusion in the Abstract.
Thank you; we have revised our wording to ensure consistency and clarity.

5.
Given the importance placed on shared decision-making in the conclusion, more context could be given in the Introduction and/or Discussion section(s).
Thank you; we have added included wording on shared decision making in the introduction, and have also described the concept in the 'intervention' section of the methods.
Competing Interests: No competing interests.