The longitudinal role of overweight and obesity women in mammographic breast screening participation: retrospective cohort study using linked data

BACKGROUND

Breast cancer is the leading cause of cancer death among women globally, with seven women on average dying each day in Australia from the disease [1]. Breast cancer places a heavy burden on healthcare systems, with expenditure on breast cancer treatment forecast to rise in coming years due to increasing incidence, an aging population and an increase in risk factors, such as obesity [2]. Early detection and intervention for breast cancer is key in achieving better survival rates with biennial mammographic screening and earlier detection shown to be associated with a reduction in mortality from breast cancer of between 21–28% [2]. In Australia, like other developed settings, mammography screening is population-based and provided via a free organised screening program for all women aged over 40 in Australia. Women between the ages of 50-74 are routinely invited to biennial screening as part of this BreastScreen program, delivered across all states and territories, including through BreastScreen New South Wales (NSW). Women aged 40-49 can self-refer.

Despite the established benefits and availability of screening, rates of participation in Australia remain suboptimal, with approximately 55% of women (number of women screened in 2-years divided by the eligible index cohort) regularly participating in the national program in 2017-2018 [3], well below BreastScreen’s National Accreditation Standard target of ≥70% [4]. Screening participation is also lower in certain areas and among certain populations, with lower rates in rural and remote regions, among people who experience socioeconomic disadvantage, individuals from certain ethnic backgrounds, and those of Aboriginal or Torres Strait Islander status. However, these sub-populations are also at higher risk of breast cancer due to the higher prevalence of risk factors including obesity [5, 6]. Obesity is a well-established risk factor for the development of post-menopausal breast cancer due to adipose tissue acting a major reservoir for oestrogen biosynthesis following menopause [7–9]. Higher body mass index (BMI) has also been associated with more aggressive clinical presentations of breast cancer [10, 11], is an adverse prognostic factor in response to adjuvant chemotherapy [12], and is associated with higher mortality rates due to breast cancer [13, 14].

In NSW, the population focus of the current study, the BreastScreen NSW program currently has one of the lowest participation rates (53.3%, 2018-2019) of all breast-screening programs in Australia [15], and is also a population where 1 in 3 women (29.96%) in the target screening age group have obesity [16]. This suggests there are a number of higher risk women in NSW who may not be accessing timely or regular mammography screening. However, the true incidence proportion of women who choose not to screen, or do not return for their biennial screening mammogram, because of issues associated with their weight is unclear. It is also not known how weight might interact with other potential screening influencers such as cultural background or level of education.

Exploratory research conducted in NSW among women with obesity has identified possible reasons for low participation in mammographic screening [17], including poor prior screening experiences, limited desire to prioritise personal health needs and low knowledge around the heightened need to screen. Suboptimal mammographic screening participation has also been reported among women with obesity in other settings, largely established through cross sectional studies based on self-reported screening behaviour [18–20]. Only one study in Australia, based on self-reported screening data, found obese women were 8% less likely to screen than ‘healthy’ weight women [21] and may be over-estimating of the proportion of women actually screening [21]. Associations like these are useful in informing further research directions but are unable to demonstrate a clear temporal association between BMI and adherence to screening. Research is needed to examine how changes in BMI and other co-variates determine subsequent mammographic screening adherence, using objectively collected screening data and accounting for longitudinal changes in BMI. Accordingly, this study investigates the association between prospective changes in BMI and longitudinal adherence to mammographic screening among overweight or obese women residing in NSW, Australia.

METHODS

RESULTS

At baseline, most participants included in the study (N=2,822) were married (75.8%) and born in Australia (80.6%) (Table 2). Most participants had a BMI of ≤25 kg/m² (43.5%), followed by a BMI of =25 kg/m² to ≤30 kg/m² (32.8%), then a BMI of >=30 kg/m² (22.8%). Nearly 40% of participants had attained a School Certificate qualification, and one-third a Certificate Diploma, university degree or higher degree. Nearly 46% lived in inner regional areas followed by major cities (36.5%), with the majority employed (72.6%) at the baseline. More than half (57%) had experienced any chronic disease with 14.8% reporting a personal history of cancer. Nearly 65% reported their ease of having a mammogram as either excellent or very good (Table 2).

Participants who had overweight/obesity were more likely to be non-adherent to mammography screening, compared to normal or underweight participants (OR=1.29, 95% CI=1.07, 1.55) according to MSM models adjusted for baseline confounders (Table 3). Stratified analyses indicated the association between overweight/obesity and non-adherence was slighter higher among those who ever had private health insurance (OR=1.30, 95% CI=1.05, 1.61) compared to those who never had private health insurance (OR=1.19, 95% CI=0.83, 1.71) (Table 3). Similarly, stratified analyses also indicated the association between overweight/obesity and non-adherence was slightly higher among those with lower educational background (OR=1.38, 95% CI=1.08, 1.75) compared to those with higher educational background (OR=1.27, 95% CI=0.93, 1.73) (Table 3). Similar associations were evident between overweight or obese and breast screening adherence according to the multilevel mixed effect models (Supplementary Tables 1 & 2). Additionally, those more likely not to adhere to routine breast cancer screening were older (>=65 years) (OR=4.81, 95% CI=3.87, 5.98), have attained a higher degree (OR=1.74, 95% CI=1.08, 2.81), were divorced (OR=1.62, [95% CI=1.23, 2.15), had ever had a cancer (OR=1.29, 95% CI=1.09, 1.54), were current smokers (OR=1.54, 95% CI=1.15, 2.08), had ever had an abnormal mammogram test result (OR=1.71, 95% CI=1.41, 2.08), and had previously had a breast examination by a doctor or nurse (OR=1.26, 95% CI=1.08, 1.47). Individuals currently using hormone replacement therapy were more likely to adhere routine breast cancer screening (OR=0.7, 95% CI=0.56, 0.87) (Supplementary Table 1). Further, compared to those who responded as ‘excellent’ to ease of having a mammogram, those who responded ‘good’ (OR=1.53, [95% CI=1.26, 1.85), ‘fair’ (OR=3.41, [95% CI=2.51, 4.63), poor (OR=3.07, [95% CI=1.86, 5.07) and ‘don’t know’ (OR=3.63, [95% CI=2.42, 5.45) were more likely to be non-adherent to routine breast cancer screening (Supplementary Table 1). Similar associations between the above exposures and non-adherence to routine breast cancer screening were observed in multilevel models stratified by education status and private health insurance (Supplementary Table 2).

DISCUSSION

This study investigated the association between BMI and subsequent participation in mammographic breast screening among a population-based cohort of NSW, Australia women. This is the first study to clearly delineate the temporal relationship between changes in BMI and longitudinal adherence to mammographic screening among women.

Our findings support the hypothesis that higher BMI has a long-term effect on participation in routinely organised breast screening, independent of individual risk factors. Women with a higher BMI should therefore be a focus of efforts to improve breast screening participation. This is particularly important given that women with a higher BMI have increased risk of post-menopausal breast cancer, more aggressive clinical presentations and histopathology of breast cancers, poorer treatment responses and metastases rates [10, 11, 27, 28]. Women with obesity have also been shown to have a one-third increased risk of breast cancer mortality and 41% increased mortality risk overall when compared to women diagnosed with breast cancer and who have a normal weight [29]. The need for focussed efforts is also highly warranted given knowledge relating to postmenopausal obesity as a risk factor for breast cancer appears to be limited [17, 30], highlighting the urgent need to educate women on their differing risk profiles.

Our findings also showed higher non-adherence to screening among women who ever had private health insurance compared to those who never had private health insurance. This could reflect misclassification bias, however, as the apparent higher non-adherence may in fact be due to women with private health insurance having mammograms outside of the BreastScreen NSW program via Medicare (the Australian Government-funded universal health insurance scheme). This type of de facto ‘screening’ requires general practitioner referral and requires co-payment by the patient as mammograms are partially refundable if clinically indicated, unlike the fully funded mammograms conducted via BreastScreen. De facto screening through alternate providers leading to late or lapsed attendance at BreastScreen has previously been associated with higher health insurance coverage, and higher education, and thought to be reflective of private screening outside publicly provided services [31–33]. Women with private health insurance (which is also a proxy for higher income) are likely to have more familiarity with the private sector and may perceive privately provided screening to be better quality.

The association between BMI and non-participation in screening was also observed to be higher among women with lower (compared to higher) educational attainment. This finding is consistent with literature examining adherence to breast screening among general populations which has found women with higher education are more likely to take part in regular breast screening when compared to women with lower education [34–37]. These findings suggest that while broad approaches aimed at all women with higher BMI may be useful, targeted, and tailored approaches to women who have a higher BMI and who also have lower education levels is warranted. This is particularly important in the context of the general lower awareness of risk breast cancer due to excess weight, given lower health literacy is likely to be a contributing factor to this issue [17, 30]. This type of approach is particularly relevant given the possibility that risk stratified screening is being trialled and considered in several settings globally underscoring the need for a well thought out approach to education [30, 38].

There are several methodological limitations when interpreting the findings from the current study. First, the baseline cohort were selected to be broadly representative of the general Australian population of women, however there was some loss to follow up in successive survey waves (from n = 13714 in 1996 to n = 7956 in 2016). At baseline, women had on average a higher BMI (54% with overweight/obesity) than similarly aged women in representative data from that time period (42% with overweight/obesity) [39]. The weight of women in the eight (final) survey wave was higher than baseline, with 63% of the remaining cohort having overweight or obesity. This proportion was lower than similarly aged women from the general Australian population (73.3%) on representative data from approximately the same time period [40], potentially under-estimating the true association between obesity and adherence to routine breast screening in the general population.

Additionally, BMI was based on self-reported estimates of height and weight, which may be a source of measurement bias in the exposure. Participants often under-estimate their weight and over-estimate their height [41].There may also be under-enumeration in the outcome measure of adherence to screening. As noted above, it is likely that a proportion of women (predominantly from higher SES groups) may have engaged in de facto screening through Medicare. It is not possible to determine the extent of this, however, previous estimates indicate that 36% of women who have never attended for a mammogram at BreastScreen reported a referral was required for breast screening, even though no referral is required for BreastScreen [31]. Significantly more women in this study who never attended BreastScreen screening stated they needed a doctor’s referral for screening when compared to attenders (OR 3.19, 95% CI 2.06-4.95, p<0.001). This suggests a number of women may be accessing Medicare partially funded private mammograms rather than BreastScreen mammograms.

Given the de facto screening may be more common among higher SES women, who are also likely to be of lower weight on average based on population estimates [40], then relative differences in adherence to screening between overweight/obese and normal weight women may be over-estimated. However, there is likely to be limited under-enumeration of subsequent BreastScreen mammograms given that all participants were able to be linked to the NSW state-wide BreastScreen clinical records up to the most recent period.

There are also some important strengths in this study. This is a large population-based cohort of women who have been followed longitudinally for 20 years allowing for the assessment of changes to BMI over time. The study also had objective enumeration of the outcome (adherence to organised breast screening) via individual-level record linkage. The analysis was also able to clearly delineate the temporal relationship between BMI and subsequent screening mammogram and incorporated a range of fixed and time-varying confounders in estimating controlled direct effects between the exposure and outcome. No previous study has investigated prospective associations between changes in BMI among women and subsequent screening behaviour.

CONCLUSION

This study investigated the association between prospective changes in BMI and longitudinal adherence to mammographic screening among overweight or obese women residing in NSW. Findings show long-term impacts on screening participation with higher BMI women being less likely to participate in routinely organized breast screening. Women with a higher BMI should be a focus of efforts to improve breast screening participation, particularly given their increased risk of post-menopausal breast cancer and the association of higher BMI with more aggressive clinical presentations and histopathology of breast cancers.

COMPETING INTERESTS

All authors declare no financial relationships with any organisations that might have an interest in the submitted work in the previous three years and no other relationships or activities that could appear to have influenced the submitted work.