case study example breast cancer

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• My Account Login
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 23 November 2020

Clinical Study

A case-control study to evaluate the impact of the breast screening programme on mortality in England

• Roberta Maroni   ORCID: orcid.org/0000-0001-6420-2881 1   na1 ,
• Nathalie J. Massat   ORCID: orcid.org/0000-0002-1095-994X 1   na1 ,
• Dharmishta Parmar 1 ,
• Amanda Dibden   ORCID: orcid.org/0000-0002-0599-9840 1 ,
• Jack Cuzick 1 ,
• Peter D. Sasieni   ORCID: orcid.org/0000-0003-1509-8744 2   na2 &
• Stephen W. Duffy   ORCID: orcid.org/0000-0003-4901-7922 1   na2  

British Journal of Cancer volume  124 ,  pages 736–743 ( 2021 ) Cite this article

8131 Accesses

13 Citations

79 Altmetric

Metrics details

• Breast cancer
• Cancer screening

Over the past 30 years since the implementation of the National Health Service Breast Screening Programme, improvements in diagnostic techniques and treatments have led to the need for an up-to-date evaluation of its benefit on risk of death from breast cancer. An initial pilot case-control study in London indicated that attending mammography screening led to a mortality reduction of 39%.

Based on the same study protocol, an England-wide study was set up. Women aged 47–89 years who died of primary breast cancer in 2010 or 2011 were selected as cases (8288 cases). When possible, two controls were selected per case (15,202 controls) and were matched by date of birth and screening area.

Conditional logistic regressions showed a 38% reduction in breast cancer mortality after correcting for self-selection bias (OR 0.62, 95% CI 0.56–0.69) for women being screened at least once. Secondary analyses by age group, and time between last screen and breast cancer diagnosis were also performed.

Conclusions

According to this England-wide case-control study, mammography screening still plays an important role in lowering the risk of dying from breast cancer. Women aged 65 or over see a stronger and longer lasting benefit of screening compared to younger women.

Similar content being viewed by others

Effectiveness of the Korean National Cancer Screening Program in reducing breast cancer mortality

Eunji Choi, Jae Kwan Jun, … Kui Son Choi

Health benefits and harms of mammography screening in older women (75+ years)—a systematic review

Erin Mathieu, Naomi Noguchi, … Nehmat Houssami

Assessing trends of breast cancer and carcinoma in situ to monitor screening policies in developing settings

Érika de Abreu Costa Brito, Marcela Sampaio Lima, … Rosana Cipolotti

Following an evaluation of several randomised controlled trials (RCT) 1 that showed an overall reduction in mortality from breast cancer in women undergoing mammography screening, the National Health Service Breast Screening Programme (NHS BSP) was launched in the United Kingdom (UK) in 1988. At the time, it aimed to offer free routine screening to every woman aged 50–64 once every three years. It now invites women aged 50–70, with an age extension to younger and older women (47–73 years) being trialled. 2

Over the last thirty years, major advances have been made in the fields of cancer screening, treatment, and management (including effective adjuvant systemic therapies 3 and two-view mammography 3 , 4 ), with resulting lengthening of survival times after a breast cancer diagnosis. 5 Despite recent reductions in breast cancer mortality, breast cancer is still the cancer with the highest incidence 6 and the second most common cause of cancer death 7 in females in the UK.

Case-control studies are a useful tool to evaluate screening programmes in settings where lack of equipoise would mean that RCTs would be unethical, or as in this case, where the RCTs have already been done, but there remains a need to ensure that the service is delivering the expected clinical benefit. Case-control studies also overcome some limitations associated with other observational designs by taking into account changes in cancer incidence and use of treatments over time and adjusting for any imbalances in other factors that could affect breast cancer mortality.

Taking as an example a case-control study 8 that resulted in policy change within the NHS cervical screening programme by altering age at first screen and the screening interval, we designed a similar study focussing on the NHS BSP with the aim of:

Evaluating the effect of mammography screening in the NHSBSP on breast cancer mortality

Evaluating the effect of mammography screening on breast cancer incidence, and incidence of late stage disease

Estimating overdiagnosis

Analysing the interplay of early detection, pathology, and treatment on fatality of breast cancer.

The study protocol and results from two pilot studies have been published previously. 9 , 10 , 11 This paper reports on the first objective above (breast cancer mortality), making use of England-wide data. Effects on incidence etc. will be reported in future papers.

Definition of cases and controls

As the main objective was to evaluate the effect of mammography screening on breast cancer mortality, cases were defined as women whose primary cause of death was breast cancer, who were diagnosed at age 47 years or older and died at age 89 years or younger in 2010–2011. We chose the lower limit of 47 as there is a major trial of screening in ages 47–49 ongoing, 2 so substantial numbers of women have been screened in this age group. We chose the upper limit of 89 because above this age we would not expect a major effect of screening taking place mainly at ages 50–70, because we were less confident of the cause of death in the very old, and because screening is essentially aimed at preventing premature mortality, which one might reasonably interpret as death below age 90 years. Only diagnoses occurring after 1990 were included in the analysis. Their matched controls were women sampled from the general population of those invited for screening (99.9% of women eligible for screening in England 12 ) and alive at the time of their corresponding case’s death. Controls may have been diagnosed with breast cancer, but not before their case’s date of diagnosis. Where possible, two controls were selected per case and matched on date of birth (within one month of the case’s) and screening area at date of diagnosis.

For the purposes of the statistical analysis, controls were assigned a date of pseudodiagnosis, equal to the diagnosis date of their corresponding matched case. To be eligible as a case or a control, a woman had to have had at least one invitation to screening prior to the date of diagnosis/pseudodiagnosis.

The primary endpoint was to estimate, among those invited to breast screening, the effect of ever attending breast screening on mortality from breast cancer. Changes in this effect over time were also investigated. Secondary endpoints included the effect of measures of screening intensity, such as time between last screen and diagnosis/pseudodiagnosis, and their estimations in different age subgroups.

Data selection and linkage

Cases were identified from the National Cancer Registration and Analysis Service (NCRAS) database accessed through the Office for Data Release of Public Health England (PHE). This database contains Office for National Statistics date and cause of death data. NHS Digital used the National Health Application and Infrastructure Services (NHAIS) system to identify matched controls and provided breast and cervical screening histories within.

We excluded any breast screens occurring outside the usual call/recall system of the national screening programme. All the screening histories of the study subjects were considered up to and including their date of diagnosis/pseudodiagnosis.

The data were processed according to the NHS Information Governance guidelines. 13

Sample size

Sample size calculations for the pilot study showed that, assuming an OR for breast cancer mortality of 0.7 and a number of discordant pairs of 33%, two controls per case with 800 breast cancer deaths and 1600 controls would confer more than 90% power to detect such an effect size at the 5% significance level using a two-sided test. 10 As the data for this main phase encompassed the whole of England, we had ample power, not only for the primary outcome (8288 cases and 15,202 controls after exclusions), but also for subgroup analyses.

Statistical analysis

Data were analysed using Stata version 13 14 by matched (conditional) logistic regression with death from primary breast cancer as the outcome. Date of birth and screening area were accounted for by the matching process.

Ineligible subjects were excluded (see Fig.  1 ). For some of these, this resulted in a matched set containing only a case, or only controls, which could then no longer be used in the matched logistic regression. Sensitivity analyses using unmatched logistic regression and controlling for age at diagnosis/pseudodiagnosis and screening area were performed on the same dataset with fewer exclusions; in this case, the inclusion criteria considered were the same, but the fact that a case or a control was excluded did not imply discarding that matched set.

Asterisk indicates that these records were excluded for being in a 1:1 matched set where the case or the control was excluded or for being in a 1:2 matched set where the case or both controls were excluded. Hash indicates that these become 1:1 matched sets in the final dataset. Note: some records may be excluded for more than one reason.

Case-control studies used to evaluate population screening programmes are subject to a type of bias known as non-compliance or self-selection bias, which is based on the assumption that people who are already ill may be less likely to attend screening and those who do attend may be more health conscious, and therefore healthier, than those who do not take up the invitation. This may confer an artificially greater protective effect for screening, which was corrected in our analyses using a variant of the method by Duffy et al. 15

The effect of self-selection bias was estimated using data available on cervical screening attendance for the women in the study, on the basis that any observed protective effect of cervical screening on breast cancer death cannot be due to cervical screening (which does not include breast examination) and is therefore likely to be caused by self-selection bias. In particular, the odds ratio (OR) uncorrected for self-selection is an estimate of the relative risk:

An unbiased estimate of the effect of screening on risk of dying from breast cancer would be (refer to Duffy et al. 15 ):

The OR for death from breast cancer associated with attendance at cervical screening, i.e. the self-selection correction factor, can be considered an approximate estimate of the relative risk:

Therefore, we obtain an estimate of θ by dividing γ by φ . The fundamental assumption here is that the populations choosing to attend or not to attend cervical cancer screening have the same risk of dying of breast cancer a priori as those choosing or not choosing to attend breast cancer screening. We do not assume that the effects of self-selection are the same in the two programmes. This is referred to as our first method of correction in the Results section.

As there is considerable uncertainty in the extent of self-selection, and of course decisions to attend at two separate screening programmes are likely to be confounded with each other, we also corrected for this using the method of Duffy et al. 15 . This method estimates the effect of participation in screening in those who would participate if invited as:

where p is the proportion of the invited population who participate in screening and D r is the a priori relative risk of dying of breast cancer for someone who chooses not to attend compared to an uninvited general population member. We estimated D r as 1.19 (95% CI 1.11–1.27), from the cohort study of Johns et al. 16 Thus, this correction was based on a prospective estimate of the extent of self-selection bias in a cohort of 988,090 women in the NHS Breast Screening Programme. We estimated p as 73.4% from the annual report of the National Programme. 12 This method, referred to as our second method of correction in the Results section, also yields an estimate of the effect of invitation to screening as follows: 15

More details on the methods are available in the published study protocol 9 and pilot study analysis. 10

The study dataset had a total of 9550 cases and 17,993 controls. There were 1107 sets with matching ratio 1:1 (1 case to 1 control) and 8443 sets with matching ratio 1:2 (1 case to 2 controls). Records of 1262 cases and 2791 controls (15% of the total) were excluded for various reasons before the statistical analysis (see study flow diagram in Fig.  1 ). This left a final dataset of 8288 cases and 15,202 controls, divided into 1,374 matched sets of size 1:1 and 6914 of size 1:2.

Sensitivity analyses using unconditional logistic regression were performed including subjects without a matched case or control, leaving us with 8479 cases and 16,794 controls.

Table  1 shows patient demographics and screening histories. Median age at first diagnosis was 64 years for both cases and controls and median age at death for cases was 71 years. Whilst the distributions of the number of screening invitations in the two study groups were comparable, differences can be noted in screening attendance, with 72% of the cases versus 82% of the controls attending their first screening invitation; 64% of the cases versus 76% of the controls attending their last screening invitation before diagnosis/pseudodiagnosis; and 21% of the cases versus 12% of the controls never being screened. Median time between last screen and date of diagnosis/pseudodiagnosis for compliers was also slightly longer for cases. From the data available on cervical screening history up to the date of diagnosis/pseudodiagnosis, it can be noted that 22% of the cases compared with 19% of the controls never had a cervical screen.

Table  2 summarises the main results without and with correction for self-selection bias. Using data from cervical screening attendance, the self-selection correction factor was estimated to be 0.78 (95% CI 0.73–0.84). The primary endpoint, the association between attending one or more screens and death from breast cancer, had a resulting OR = 0.49 (95% CI 0.45–0.53) and, when corrected for self-selection, had OR = 0.62 (95% CI 0.56–0.69) by our first method and OR = 0.63 (95% CI 0.55–0.71) by our second. Using the second method, the estimate of the effect of invitation to screening was a 26% reduction in breast cancer mortality (OR = 0.74, 95% CI 0.68-0.81). The unmatched logistic regression on the larger dataset for sensitivity analyses showed a similar effect of screening on breast cancer mortality both before and after controlling for age at diagnosis/pseudodiagnosis and screening area (in both cases, uncorrected OR = 0.55, 95% CI 0.51–0.59).

In order to analyse changes of the effect of screening over time, we excluded women diagnosed before year 2000 (13% of the total records), which led to a corrected OR of 0.56 (95% CI 0.51–0.63) for the effect of ever attending mammographic screening on breast cancer mortality. Women diagnosed from year 2003 onwards had an even larger benefit from being screened (OR corrected by first method = 0.53, 95% CI 0.47–0.59). The estimated effect continued to increase as we restricted the year of diagnosis/pseudodiagnosis further in time (Supplementary Fig.  1 ).

Table  3 shows how the effect of screening varies depending on how much time has passed between a woman’s last screen and her diagnosis/pseudodiagnosis. Screen-detected cancers (assumed to be cancers diagnosed within three months of screening) showed a positive association with breast cancer fatality, after self-selection bias correction by our first method (OR = 1.93, 95% CI 1.68–2.22), while women screened in any other time interval were at reduced risk of dying from breast cancer. This was lowest for women screened in the last year (OR corrected by our first method = 0.19, 95% CI 0.17–0.23) and gradually increased, while still conferring a beneficial effect to screening, for women screened further back in time with respect to their date of diagnosis/pseudodiagnosis. Results using our alternative correction for self-selection were very similar (Table  3 ). Note that the time is from screening to diagnosis, not to death. The Table shows risk of subsequently dying of breast cancer increasing by the time between the screen and diagnosis/pseudodiagnosis.

A similar analysis is shown in Table  4 and Fig.  2 for different time intervals after stratifying for three different age categories at diagnosis/pseudodiagnosis (younger than 60 years, between 60 and 64 years, and 65 years or older). The results show that the protective effect of a screen is greater and lasts longer in the oldest group. The benefit of attending screening in the three years prior to diagnosis/pseudodiagnosis, the recommended interval for screening in the NHS BSP, is shown in the final row of Table  4 , and shows close to a halving of risk with screening within the recommended interval, following self-selection correction by our first method (OR = 0.51, 95% CI 0.46–0.57). Results using our second method of correction were very similar to those using the first (Supplementary Table  1 ). The estimated effect of invitation to screening within the last 36 months using our second method was a 33% reduction in breast cancer mortality (OR = 0.67, 95% CI 0.61–0.73).

Note: the coordinates on the x -axis are the midpoints of the time intervals: 0–3, 3–6, 6–18, 18–36, 36–54 and 54–72 months.

Despite the many improvements in treatments, diagnostic procedures and technologies over the last thirty years, and changes in baseline rate of breast cancer mortality, our data showed an overall reduction in the risk of dying from breast cancer of ~38% for women attending at least one mammography screen, after adjusting for self-selection bias. This is in line with the results obtained from the pilot phase of the study, 10 in which a mortality reduction of 39% was seen for women attending screening in London (deaths occurring in 2008–2009). Using the same calculation method as in the review by the Independent UK Panel on Breast Cancer Screening UK Independent Review, 17 this would correspond to approximately nine breast cancer deaths prevented for every 1,000 women attending screening at ages 50–69 years, larger than but in the same general scale as the six deaths estimated from the UK Independent review.

It should be noted that there is a wide range of estimates of the absolute mortality benefit of mammography screening 18 , 19 , 20 , 21 some finding considerably smaller benefits than above. The size of the estimated effect depends on sources used and assumptions made. However, it has been shown to depend more crucially on whether the effect pertains to screening per se or to invitation to screening only, and on the timescale envisaged. 22 Screening prevents deaths not this year or next, but 5, 10, 15 or 20 years from now. Considering the effect of screening on 10-year mortality will considerably underestimate the absolute benefit. Nevertheless, it should be acknowledged that while the body of evidence, randomised and observational, points to a substantial reduction in breast cancer mortality with screening, there is sufficient variation that different views are still possible.

Our first method of correction for self-selection caused a decrease of about 25% in the estimated protective effect of screening for women having at least one mammogram. The second method yielded similar results. This is a greater correction than the one estimated in the pilot phase, 10 where self-selection only played a minor role, despite the fact that the final risk reduction is very similar. London has a lower coverage than the rest of England for both breast and cervical screening, which is largely explained by factors like deprivation and ethnicity. 23 Such variations in coverage might be one of the causes for the different impact of self-selection between the two phases of the study. For example, a larger population of non-participants, such as in London, may be less different in health status than a smaller population. In the Swedish two-County trial, 24 where only 15% of the population were non-participants, the rate of death from breast cancer in this population was very high. It is also worth noting that, during the early 21st century, breast screening attendance was rapidly increasing in London, and the socioeconomic gradient in attendance was reducing with time nationally. 25 , 26

Case-control studies tend to give higher estimates of benefit than other evaluations, largely because they assess the effect of actually being screened rather than simply being invited to screening. 19 , 27 It should be noted that with our second correction for self-selection bias, we were able to estimate the effect of invitation, giving a 26% breast cancer mortality reduction, similar to the effect observed in the randomised trials in this age group and to the prospectively estimated effect of a 25% reduction in the Copenhagen screening programme. 28 As a comparison, in the review by the Independent UK Panel on Breast Cancer Screening, 17 a meta-analysis of 11 RCTs found that the relative risk reduction of breast cancer mortality for women invited to screening was 20%. Furthermore, in the same report, the panel stated that the case-control studies that they had analysed seemed to inflate the benefit of screening compared to the trials and postulated that this may have been caused by some residual bias unaccounted for by the authors. We believe that our adjustments for self-selection bias has largely accounted for this and that the greater effect of screening in this study is due to technical improvements in mammography since the RCTs were carried out, accompanied by improved treatment and strong quality assurance measures in the NHS BSP. 11

The greater benefit of screening observed for women diagnosed after year 2000 was similar to the pilot study, 10 but here we were able to restrict the analysis to later years of diagnosis and see the benefit getting larger (data not shown). We could conjecture that this improvement was due to the introduction of better procedures in the NHS BSP, such as two-view mammography at every attendance in year 2000 4 ; however, there may be a bias in comparing different times since diagnosis as we only have data on deaths in years 2010–2011. In the first place, cases diagnosed before 2000 have a long survival by definition, and there might therefore be an over-representation of screen-detected cancers. In other words, it is more likely that a case diagnosed before year 2000, for example, who had a breast cancer for more than 10–11 years before dying from it, had a screen-detected cancer rather than a symptomatic one. This confers a bias against screening in the analysis of cancers diagnosed prior to the year 2000. In the second place, there will be a bias in favour of screening if the analysis is restricted to cancers diagnosed within a short time before death, i.e. if we only consider women (pseudo)diagnosed a few years before 2010–2011. We are therefore unable to make any definitive conclusions on the impact of any improvements in the NHS BSP over time.

As shown in RCTs of breast screening, 24 measures of the benefit of screening are largely influenced by the consequent reduction in mortality from symptomatic cancers. This is due to the fact that screen-detected cancers (defined as the ones diagnosed within three months of a screen), despite being less fatal overall, represent a larger proportion of the cancer-related deaths in the immediate period after a screen as it can be seen from the spike in excess mortality in Fig.  2 .

The duration of the benefit of attending screening appears to be greater in older women (Table  4 and Fig.  2 ). Women aged 65 or more see the greatest and longer lasting benefit, which might suggest that they could be screened less often than younger women. This result is in agreement with the impact of ageing on breast cancer biology 29 and is also potentially important in light of the recent incident in the NHS BSP, where a number of women aged 69 and 70 years did not receive the scheduled invitation to their last screening appointment. 30 The exact number affected has been debated but an Independent Review concluded that 5000 women were not invited as scheduled, and that a further 62,000 could be interpreted as having missed their final invitation as defined in the service specification. 30 Our findings suggest that the effect of a delayed screen in older women has a lesser consequence for increased risk of breast cancer mortality than it would have had in younger women. While three years is a longer interval than other programmes in Europe and North America, and further slippage of the interval should be avoided if at all possible, these results could also be used as guidelines for screening units at times of capacity constraints, with the provision that all women receive an opportunity for a final screen around or shortly after age 70. There is interest in stratified screening and these results may inform further thinking on this subject.

A limitation of the study is the retrospective design and the potential for self-selection bias. We have corrected for this in two different ways and for one of these, an effect of invitation to screening was derived which was consistent with trials results and prospective studies for this age group. However, it must be acknowledged that there remains some uncertainty about the extent of self-selection bias. Furthermore, case-control studies for cancer screening programmes are subject to an inherent type of anti-screening bias known as screening opportunity bias. 27 As most of the controls do not have a breast cancer diagnosis, the only way they can be exposed to screening is if they attended a mammography appointment in the past. Cases, on the other hand, may have had a screen in the past, but some of them will also have an additional screen for when their cancer was diagnosed. This induces an artificially higher retrospective probability of screening exposure among cases. Screening opportunity bias was corrected for in the pilot study, 10 where a 10–15% increase in mortality reduction was seen following this, but here we preferred to keep a conservative approach and not adjust for it. To minimise biases with respect to age and opportunity to be screened, we matched very closely for age. This meant that in 1107 cases out of 9550, we could only find one control.

Although the effect of the NHS BSP in preventing breast cancer mortality has been assessed several times, 31 , 32 , 33 , 34 we are aware of only one other case-control study conducted using national data. 34 The latter relies on data up to year 2005 (diagnoses and deaths took place between 1991 and 2005), while ours uses more recent data up to year 2012, arguably more in the epoch of effective adjuvant systemic therapies. It is of interest that our more recent case base shows similar results in terms of the reduction in risk of breast cancer death with screening. In any case, we suggest that it would be of interest to repeat this type of analysis for years thereafter, to ensure that the programme continues to deliver its aims even with the introduction of new diagnostic technologies (e.g. digital mammography). Before the establishment of the NHS BSP in 1987, it was suggested that a routine case-control assessment could and should be part of an ongoing evaluation of a mass screening programme. 35 For this reason, we believe that this exercise should be held on a two-yearly basis.

The results of further national case-control studies (1) evaluating the effect of the NHS BSP on breast cancer incidence and incidence of late stage disease, (2) estimating overdiagnosis, and (3) analysing the interplay of early detection, pathology and treatment on fatality of breast cancer will be published shortly.

To conclude, this study showed that the breast screening programme in England continues to play an important role in the control of breast cancer. The effect of screening within the NHS BSP in England is stronger and longer lasting in women aged 65 or over, but it remains highly relevant for younger women.

Forrest, A. P. M. Breast cancer screening: report to the Health Ministers of England, Wales, Scotland and Northern Ireland by a working group . (HM Stationery Office, London, 1987)

Moser, K., Sellars, S., Wheaton, M., Cooke, J., Duncan, A., Maxwell, A. et al. Extending the age range for breast screening in England: pilot study to assess the feasibility and acceptability of randomization. J. Med. Screen 18 , 96–102 (2011).

Article   Google Scholar  

Plevritis, S. K., Munoz, D., Kurian, A. W., Stout, N. K., Alagoz, O., Near, A. M. et al. Association of screening and treatment with breast cancer mortality by molecular subtype in US Women, 2000-2012. Jama 319 , 154–164 (2018).

Patnick, J. NHS breast screening: the progression from one to two views. J. Med Screen 11 , 55–56 (2004).

Cancer Research UK. Breast cancer survival trends over time. https://www.cancerresearchuk.org/health-professional/cancer-statistics/statistics-by-cancer-type/breast-cancer/survival#heading-Two (2014).

Cancer Research UK. Cancer incidence for common cancers. https://www.cancerresearchuk.org/health-professional/cancer-statistics/incidence/common-cancers-compared#heading-Two (2020).

Cancer Research UK. Cancer mortality for common cancers. https://www.cancerresearchuk.org/health-professional/cancer-statistics/mortality/common-cancers-compared#heading-Two (2019).

Sasieni, P., Adams, J. & Cuzick, J. Benefit of cervical screening at different ages: evidence from the UK audit of screening histories. Br. J. Cancer 89 , 88–93 (2003).

Article   CAS   Google Scholar  

Massat, N. J., Sasieni, P. D., Parmar, D. & Duffy, S. W. An ongoing case-control study to evaluate the NHS breast screening programme. BMC Cancer 13 , 596 (2013).

Massat, N. J., Dibden, A., Parmar, D., Cuzick, J., Sasieni, P. D. & Duffy, S. W. Impact of screening on breast cancer mortality: The UK Program 20 Years On. Cancer Epidemiol. Biomark. Prev. 25 , 455–462 (2016).

Massat, N. J., Sasieni, P. D., Tataru, D., Parmar, D., Cuzick, J. & Duffy, S. W. Explaining the better prognosis of screening-exposed breast cancers: influence of tumor characteristics and treatment. Cancer Epidemiol. Biomark. Prev. 25 , 479–487 (2016).

The NHS Information Centre. Breast Screening Programme, England 2010-11. https://files.digital.nhs.uk/publicationimport/pub05xxx/pub05244/bres-scre-prog-eng-2010-11-rep.pdf (2012).

Department of Health. Information Governance Toolkit. https://www.igt.hscic.gov.uk (2020).

StataCorp. Stata Statistical Software: Release 13. (College Station, 2013).

Duffy, S. W., Cuzick, J., Tabar, L., Vitak, B., Chen, T. H.-H., Yen, M.-F. et al. Correcting for non-compliance bias in case–control studies to evaluate cancer screening programmes. J. R. Stat. Soc. Ser. C. 51 , 235–243 (2002).

Johns, L. E., Coleman, D. A., Swerdlow, A. J. & Moss, S. M. Effect of population breast screening on breast cancer mortality to 2005 in England and Wales: an individidual-level cohort study. Br. J. Cancer 116 , 246–252 (2017).

Independent, U. K. Panel on Breast Cancer Screening. The benefits and harms of breast cancer screening: an independent review. Lancet 380 , 1778–1786 (2012).

Gøtzsche, P. C., Jørgensen, K. Screening for breast cancer with mammography. Cochrane Database of Syst. Rev. 6 , CD001877 (2013).

International Agency for Research on Cancer. Breast Cancer Screening , Vol. 15. (IARC Handbooks of Cancer Prevention, 2016).

Dimitrova, N., Parkinson, Z. S., Bramesfeld, A., Ulutürk, A., Bocchi, G., López-Alcalde, J. et al. European Guidelines for Breast Cancer Screening and Diagnosis–the European Breast Guidelines . Joint Research Centre Technical Reports (Luxembourg Publications Office of the European Union, 2016).

Siu, A. L. Screening for Breast Cancer: U.S. Preventive Services Task Force Recommendation Statement. Ann. Intern Med. 164 , 279–296 (2016).

Duffy, S. W., Chen, T. H.-H., Smith, R. A., Yen, A. M.-F. & Tabar, L. Real and artificial controversies in breast cancer screening. Breast Cancer Manag. 2 , 519–528 (2013).

Massat, N. J., Douglas, E., Waller, J., Wardle, J. & Duffy, S. W. Variation in cervical and breast cancer screening coverage in England: a cross-sectional analysis to characterise districts with atypical behaviour. BMJ Open 5 , e007735 (2015).

Tabar, L., Fagerberg, G., Duffy, S. W., Day, N. E., Gad, A. & Grontoft, O. Update of the Swedish two-county program of mammographic screening for breast cancer. Radio. Clin. North Am. 30 , 187–210 (1992).

CAS   Google Scholar  

Eilbert, K. W., Carroll, K., Peach, J., Khatoon, S., Basnett, I. & McCulloch, N. Approaches to improving breast screening uptake: evidence and experience from Tower Hamlets. Br. J. Cancer 101 , S64–S67 (2009).

Douglas, E., Waller, J., Duffy, S. W. & Wardle, J. Socioeconomic inequalities in breast and cervical screening coverage in England: are we closing the gap? J. Med. Screen 23 , 98–103 (2016).

Walter, S. D. Mammographic screening: case-control studies. Ann. Oncol. 14 , 1190–1192 (2003).

Olsen, A. H., Njor, S. H., Vejborg, I., Schwartz, W., Dalgaard, P., Jensen, M.-B. et al. Breast cancer mortality in Copenhagen after introduction of mammography screening: cohort study. BMJ 330 , 220 (2005).

Benz, C. C. Impact of aging on the biology of breast cancer. Crit. Rev. Oncol. Hematol. 66 , 65–74 (2008).

The Independent Breast Screening Review 2018. https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/764413/independent-breast-screening-review-report.pdf (2018).

Moss, S. M., Summerley, M. E., Thomas, B. T., Ellman, R. & Chamberlain, J. O. A case-control evaluation of the effect of breast cancer screening in the United Kingdom trial of early detection of breast cancer. J. Epidemiol. Community Health 46 , 362–364 (1992).

Allgood, P. C., Warwick, J., Warren, R. M., Day, N. E. & Duffy, S. W. A case-control study of the impact of the East Anglian breast screening programme on breast cancer mortality. Br. J. Cancer 98 , 206–209 (2008).

van der Waal, D., Broeders, M. J., Verbeek, A. L., Duffy, S. W. & Moss, S. M. Case-control studies on the effectiveness of breast cancer screening: insights from the UK Age Trial. Epidemiology 26 , 590–596 (2015).

Johns, L. E., Swerdlow, A. J. & Moss, S. M. Effect of population breast screening on breast cancer mortality to 2005 in England and Wales: a nested case-control study within a cohort of one million women. J. Med. Screen 25 , 76–81 (2018).

Sasco, A. J., Day, N. E. & Walter, S. D. Case-control studies for the evaluation of screening. J. Chronic Dis. 39 , 399–405 (1986).

Download references

Acknowledgements

Data for this study is based on information collected and quality assured by the PHE National Cancer Registration and Analysis Service. Access to the data was facilitated by the PHE Office for Data Release. We would like to thank Rachael Brannan from the PHE Office for Data Release and David Graham from NHS Digital for their help with the data selection and matching of cases and controls. This work uses data provided by patients and collected by the NHS as part of their care and support.

Author information

These authors contributed equally: Roberta Maroni, Nathalie J Massat

These authors jointly supervised this work: Peter D Sasieni, Stephen W Duffy

Authors and Affiliations

Centre for Cancer Prevention, Wolfson Institute of Preventive Medicine, Queen Mary University of London, Charterhouse Square, London, EC1M 6BQ, UK

Roberta Maroni, Nathalie J. Massat, Dharmishta Parmar, Amanda Dibden, Jack Cuzick & Stephen W. Duffy

Faculty of Life Sciences and Medicine, Cancer Prevention Group, School of Cancer and Pharmaceutical Sciences, King’s College London, Guy’s Campus, Great Maze Pond, London, SE1 9RT, UK

• Peter D. Sasieni

You can also search for this author in PubMed   Google Scholar

Contributions

R.M. oversaw the first draft of the manuscript draft and submission. N.J.M., J.C., P.D.S. and S.W.D. designed the study. D.P. contributed to data collection and cleaning. A.D. assisted with data interpretation. R.M. and S.W.D. analysed the data and produced the figures. All the authors critically reviewed the paper.

Corresponding author

Correspondence to Stephen W. Duffy .

Ethics declarations

Ethics approval and consent to participate.

The study protocol was reviewed and approved by the Department of Health. Ethical approval was obtained from the London Research Ethics Committee of the National Research Ethics Service (reference: 12/LO/1041), and by the National Information Governance Board Ethics and Confidentiality Committee (reference: ECC 6–05 (e)/2012). The ethics committee agreed that informed consent to participate for the study subjects was not necessary. The study was performed in accordance with the Declaration of Helsinki.

Data availability

Data were saved on the servers of the Barts Cancer Institute, Queen Mary University of London, in a folder with restricted access to D.P. A clean, anonymised version of the data was produced and made available to R.M., A.D. and S.W.D. with restricted access to the staff of the Policy Research Unit in Cancer Awareness, Screening and Early Diagnosis at Queen Mary University of London. The data were obtained via the Office for Data Release at Public Health England. We do not have authority to share the data with others, but requests for access to data will be forwarded to the Office for Data Release.

Competing interests

P.D.S. reports personal fees from GRAIL Bio outside the submitted work. J.C. and S.W.D. are members of the editorial board of the British Journal of Cancer. The remaining authors declare no competing interests.

Funding information

his research is funded by the National Institute for Health Research (NIHR) Policy Research Programme, conducted through the Policy Research Unit (PRU) in Cancer Awareness, Screening and Early Diagnosis, PR-PRU-1217-21601. The PRU is a collaboration between researchers from seven institutions (Queen Mary University of London, University College London, King’s College London, London School of Hygiene and Tropical Medicine, Hull York Medical School, Durham University, and Peninsula Medical School). The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care. The funding body was not involved in design, data collection, analysis or interpretation. The funding body had sight of the paper prior to publication but has not had input to its content.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary materials, rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Maroni, R., Massat, N.J., Parmar, D. et al. A case-control study to evaluate the impact of the breast screening programme on mortality in England. Br J Cancer 124 , 736–743 (2021). https://doi.org/10.1038/s41416-020-01163-2

Download citation

Received : 18 March 2020

Revised : 21 October 2020

Accepted : 28 October 2020

Published : 23 November 2020

Issue Date : 16 February 2021

DOI : https://doi.org/10.1038/s41416-020-01163-2

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

This article is cited by

Artificial intelligence in the treatment of cancer: changing patterns, constraints, and prospects.

• Mohammad Ali
• Shahid Ud Din Wani
• Seema Mehdi

Health and Technology (2024)

Years of life lost due to cancer in the United Kingdom from 1988 to 2017

• Amar S. Ahmad
• Judith Offman

British Journal of Cancer (2023)

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Ohio State nav bar

The Ohio State University

• BuckeyeLink
• Find People
• Search Ohio State

Pathophysiology

Normal physiology of the human breast.

Prior to sexual maturity, male and female breasts are structurally and functionally similar; they are both comprised of small immature nipples, fatty and fibrous tissue and several duct-like arrangements beneath the areola. 

When puberty is underway in males and females, this is where the major structural development occurs. Male breasts remain unchanged due to the lack of high levels of estrogen and progesterone. Females, on the other hand, have significant changes occur due to an assortment of hormones (estrogen, growth hormone, insulin-like growth factor-1, progesterone, and prolactin) (p.739) that cause the female breast to develop into a lactating system. 

  The mature female breast’s foundational unit is the lobe (each breast contains 15-20) a system of ducts which is comprised of and supported by Cooper ligaments. Each lobe is made up of 20-40 lobules (glands that produce milk). The lobules contain alveolar cells, which are complicated spaces lined with epithelial cells that secrete milk and sub-epithelial cells that contract, passing milk into the arrangement of ducts that leads to the nipple (p. 739). 

The lobes and lobules are enclosed and separated by muscle strands and adipose connective tissue which varies in amount depending on weight, genetics, endocrine factors and contributes to the diversity of breast size and shape.

(Memorial Sloan Kettering Cancer Center)

During the reproductive years, breast tissue undergoes cyclic changes in response to hormonal changes of the menstrual cycle. After menopause, adipose deposits and connective tissue increases, glandular breast tissue becomes involuted, and breasts are reduced in size and form. Due to elevated aromatase (decreases circulating estrogen) there can be an  increase in white adipose tissue inflammation (p.740).

The function of the female breast is primarily to provide a source of nourishment for the newborn; however, breasts are also a source of pleasurable sexual sensation and in Western cultures have become a sexual symbol (p.741).

(McCance, K. L., & Heuther, S. E. (2019). Pathophysiology: The biological basis for disease in adults and children (8th ed., pp.739-741). St. Louis, MO: Mosby.)

Pathophysiology of Breast Cancer 

Except for skin cancer, breast cancer is the most common cancer in American women. Most breast cancer occurs in women older than 50 years. The major risk factors for breast cancer are classified as reproductive, such as nulliparity and pregnancy-associated breast cancer; familial, such as inherited gene syndromes; and environmental and lifestyle, such as hormonal factors and radiation exposure. Some examples of known carcinogenic agents with sufficient evidence in humans that contribute to the development of breast cancer are alcoholic beverages, diethylstilbestrol, estrogen-progestogen contraceptives, estrogen-progestogen menopausal therapy, X-radiation and γ-radiation (Rote, 2019). Other important factors are delayed involution of the mammary gland and increased breast density.

Overall, lifetime risk of breast cancer is reduced in parous women compared to nulliparous women, but pregnancy must occur at a young age. The influence of pregnancy on the risk of cancer also depends on family history, lactation postpartum, and overall parity. Breast gland involution after pregnancy and lactation uses some of the same tissue remodeling pathways activated during wound healing. The presence of macrophages in the involuting mammary gland contributes to carcinogenesis. 

Most breast cancers are adenocarcinomas and first arise from the ductal/lobular epithelium as carcinoma in situ. Carcinoma in situ is an early-stage, noninvasive, proliferation of epithelial cells that is confined to the ducts and lobules, by the basement membrane. About 84% of all in situ disease is ductal carcinoma in situ (DCIS); the remainder is mostly lobular carcinoma in situ (LCIS) disease.  Ductal carcinoma in situ (DCIS) refers to a heterogenous group of proliferative lesions limited to ducts and lobules without invasion to the basement membrane. DCIS occurs predominantly in women but can also occur in men. DCIS has a wide spectrum of risk for invasive cancers. Preinvasive lesions do not invariably progress to invasive malignancy. Lobular carcinoma in situ (LCIS) originates from the terminal duct-lobular unit. Unlike DCIS, LCIS has a uniform appearance; thus, the lobular structure is preserved. The cells grow in non-cohesive clusters, typically because of a loss of the tumor-suppressive adhesion protein E-cadherin . Also, unlike DCIS, LCIS is found as an incidental lesion from a biopsy and not mammography, is more likely to be present bilaterally.

(Winslow, 2012)

Breast cancer is a heterogeneous disease with diverse, molecular, genetic, phenotypic, and pathologic changes. Tumor heterogeneity results from the genetic, epigenetic, and microenvironmental influences (selective pressure) that tumor cells undergo during cancer progression. Cellular subpopulations from different sections of the same tumor vary in many ways including growth rate, immunogenicity, ability to metastasize, and drug response, demonstrating significant heterogeneity. The biological attributes of a tumor as a whole are strongly influenced by its subpopulation of cells with cellular populations communicating through paracrine or contact-dependent signaling (juxtacrine) from ligands and mediated from components of the microenvironment such as blood vessels, immune cells, and fibroblasts. 

Figure 1. Histological special types of breast cancer preferentially oestrogen receptor positive. (A) Tubular carcinoma, (B) cribriform carcinoma, (C) classic invasive lobular carcinoma, (D) pleomorphic invasive lobular carcinoma, (E) mucinous carcinoma, (F) neuroendocrine carcinoma, (G) micropapillary carcinoma, (H) papillary carcinoma, (I) low grade invasive ductal carcinoma with osteoclast-like giant cells. (Weigelt, Geyer, & Reis-Filho, 2010)

Figure 2. Histological special types of breast cancer preferentially oestrogen receptor negative. (A) Adenoid cystic carcinoma, (B) secretory carcinoma, (C) acinic-cell carcinoma, (D) apocrine carcinoma, (E) medullary carcinoma, (F) metaplastic carcinoma with heterologous elements, (G) metaplastic carcinoma with squamous metaplasia, (H) metaplastic spindle cell carcinoma, (I) metaplastic matrix-producing carcinoma. (Weigelt, Geyer, & Reis-Filho, 2010)

Gene expression profiling studies have identified major subtypes classified as luminal A, luminal B, HER2+, basal-like, Claudin-low, and normal breast. These subtypes have different prognoses and responses to therapy. Tumors can be stratified with gene expression profiles such as Oncotype Dx, Prosigna, and MammaPrint on the basis of genetic profiles. This information helps personalize breast cancer treatment and determine which women need aggressive systemic treatment for high-risk cancers versus close surveillance for indolent tumors.

Many models of breast carcinogenesis have been suggested and the expanding themes include (1) gene addiction, (2) phenotype plasticity, (3) cancer stem cells, (4) hormonal outcomes affecting cell turnover of mammary epithelium, stem cells, extracellular matrix, and immune function. 

Cancer gene addiction includes oncogene addiction, whereby these driver genes play key roles in breast cancer development and progression. In non-oncogene addiction, these genes may not initiate cancer but play roles in cancer development and progression. Examples of key driver genes include HER2 and MYC, and examples of tumor-suppressor genes include TP53, BRCA1, and BRCA2. Once a founding tumor clone is established, genomic instability may assist through the establishment of other subclones and contribute to both tumor progression and therapy resistance. 

Phenotypic plasticity is exemplified by a distinctive phenotype called epithelial-to-mesenchymal transition (EMT) . EMT is involved in the generation of tissues and organs during embryogenesis, is essential for driving tissue plasticity during development, and is hijacked during cancer progression. The EMT-associated programming is involved in many cancer cell characteristics, including suppression of cell death or apoptosis and senescence. It is reactivated during wound healing and is resistant to chemotherapy and radiation therapy. Remodeling or reprogramming of the breast during post-pregnancy involution is important because it involves inflammatory and “wound healing-like” tissue reactions known as reactive stroma or inflammatory stroma . The reactive stroma releases various signals and interleukins that affect nearby carcinoma cells, inducing these cells to activate their previously silent EMT programs. The activation is typically reversible (i.e., plasticity), and those EMT programs may revert through mesenchymal-epithelial (MET) to the previous phenotypic state before the induction of the EMT program. Reactive stroma increases the risk for tumor invasion and may facilitate the transition of carcinoma in situ to invasive carcinoma. Activation of an EMT program during cancer development often requires signaling between cancer cells and neighboring stromal cells. In advanced primary carcinomas, cancer cells recruit a variety of cell types into the surrounding stroma. Overall, increasing evidence suggests that interactions of cancer cells with adjacent tumor-associated stromal cells induce malignant phenotypes.

Figure 3. Putative EMT and MET in breast cancer progression. Normal epithelial cells undergo a series of transformational changes to become malignant tumor cells. Clonal proliferation of malignant cells gives rise to invasive carcinoma. Some of these cells undergo EMT and enter into the neighboring blood vessels or lymphatic vessels. These cells may remain in the circulation as circulating tumor cells or may extravasate at a distant site. The extravasated tumor cells form macrometastasis by a reverse mechanism known as MET. EMT, epithelial-mesenchymal transition; MET, mesenchymal-epithelial transition. (Liu, Gu, Shan, Geng, & Sang, 2016)

Using a mouse model of tumor heterogeneity, investigators demonstrated different clones within the heterogeneous population had distinct properties, such as the ability to dominate the primary tumor, or to contribute to metastatic populations, or to enter the lymphatic or vascular systems via vascular mimicry. 

(Wagenblast et al., 2015)

Figure 4. Two adjacent sections of a mouse breast tumor. Tissue at left is stained so that normal blood vessels can be seen (brown arrow). Extending from these vessels are blood filled channels (green arrows). On the right, the tissue is stained for a fluorescent protein expressed by the tumor cells. Here it is seen that blood-filled channels are actually formed by tumor cells in a process known as vascular mimicry.  The team demonstrate that the tumor cells lining these channels help drive metastasis, the process by which tumors spread. (Ravindran, 2019)

Invasion by primary tumor cells typically involves the collective migration of large cohesive groups into adjacent tissue rather than the scattering of individual carcinoma cells. However, still unknown are the precise events occurring at the invasive stage. Dormant carcinoma cells called minimal residual disease (MRD) appear to perpetuate carcinogenesis and form the precursors of eventual metastatic relapse and, sometimes, rapid cancer recurrence. Dormant cells have exited the cell cycle and are not proliferating. Thus current treatments that preferentially kill proliferating cells render dormant cells intrinsically more resistant and may remain after initial chemotherapy, radiotherapy, and surgery. 

Emerging evidence supports three main prerequisites that must be met for metastatic colonization to succeed: the capacity to seed and maintain a population of tumor-initiating stem cells; the ability to create adaptive, organ-specific colonization programs; and the development of a supportive microenvironmental niche. Metastases may also occur early in the process of neoplastic transformation.

Figure 5 . Extracellular matrix (ECM) changes in breast cancer progression and metastasis. The primary components of the ECM in normal mammary gland are significantly changed in breast cancer. A desmoplastic reaction is associated with breast cancer development, due to the increased production of fibrous ECM by activated fibroblasts and cancer cells. The increased collagen deposition and crosslinking by lysyl oxidase (LOX) enzymes, together with the increased production of fibronectin and other ECM components, stiffens the ECM, which in turn promotes tumor aggressiveness. The basement membrane surrounding the mammary gland epithelium is broken down by ECM remodeling enzymes like MMPs, heparanase and others. Matricellular proteins that promote cancer cell fitness such as tenascin C, periostin, osteopontin, SPARC and thrombospondin-1 are also upregulated. Breast cancer cells from the primary tumor, that include cells with the ability to establish metastatic colonies, enter the blood circulation, disseminate and can reach distant sites. While the vast majority of disseminated cancer cells are eliminated or undergo dormancy due to the adverse environment, few cancer cells are able to resist the selective pressure and establish a metastatic colony. These cells may rely on signals from the ECM such as type I collagen (collagen I), crosslinked by LOX. Tenascin C (TNC) and periostin (POSTN), which are crucial ECM proteins of the metastatic niche, promote stem/progenitor pathways and metastatic fitness in disseminated breast cancer cells. (Insua-Rodríguez & Oskarsson, 2016)

The first clinical manifestation of breast cancer is usually a small, painless lump in the breast. Other manifestations include palpable lymph nodes in the axillae, dimpling of the skin, nipple and skin retraction, nipple discharge, ulcerations, reddened skin, and bone pain associated with bony metastases. 

Treatment is based on the extent or stage of the cancer and includes surgery, radiation, chemotherapy, hormone therapy, and biologic therapy.

(Unless otherwise cited, all pathophysiology information was gathered from Danhausen, Phillippi, & McCance, 2019)

• Member Benefits
• Communities
• Grants and Scholarships
• Student Nurse Resources
• Member Directory
• Course Login
• Professional Development
• Institutions Hub
• ONS Course Catalog
• ONS Book Catalog
• ONS Oncology Nurse Orientation Program™
• Account Settings
• Help Center
• Print Membership Card
• Print NCPD Certificate
• Verify Cardholder or Certificate Status

• Trouble finding what you need?
• Check our search tips.

• Clinical Journal of Oncology Nursing
• Number 6 / December 2021
• Supplement, December 2021, Survivorship Care

Breast Cancer: Survivorship Care Case Study, Care Plan, and Commentaries

Amber Whitton-Smith

Rachael Schmidt

Kristie Howlett

Rachée Hatfield

This case study highlights the patient’s status in care plan format and is followed by commentaries from expert nurse clinicians about their approach to manage the patient’s long-term or chronic cancer care symptoms. Finally, an additional expert nurse clinician summarizes the care plan and commentaries, emphasizing takeaways about the patient, the commentaries, and additional recommendations to manage the patient. As can happen in clinical practice, the patient’s care plan is intentionally incomplete and does not include all pertinent information. Responding to an incomplete care plan, the nurse clinicians offer comprehensive strategies to manage the patient’s status and symptoms. For all commentaries, each clinician reviewed the care plan and did not review each other’s commentary. The summary commentary speaks to the patient’s status, care plan, and nurse commentaries.

Jump to a section

When delivering survivorship care, a nurse may start with incomplete information about the cancer survivor’s treatment and follow-up. Therefore, this breast cancer package includes an incomplete case study and care plan, and then follows with commentaries from expert nurses about their approach to comprehensive survivorship care.

A.B. is a 57-year-old second-generation Mexican American woman, diagnosed with breast cancer detected on a mammogram. She has completed four years of anastrozole and presents for routine follow-up with a new medical oncologist because her previous oncologist retired from clinical practice (see Figure 1).

•  Vital signs

•  Blood pressure: 152/84

•  Temperature: 97.9°F (36.6°C)

•  Height: 5’6” (1.676 m)

•  Weight: 69.4 kg (153 lb)

•  Body mass index: 24.69 kg/m 2

•  Last menses: age 53 years with chemotherapy

•  Care gaps: no history of colonoscopy

•  Oncology history: malignant neoplasm of overlapping sites of right breast; estrogen receptor–positive (C50.811; Z17.0)

•  Other relevant histories: allergies to dust, seasonal allergies, and some antibiotics; hypothyroidism

•  Tobacco use history: never smoker

•  Medication list

•  Fluticasone propionate: 50 mcg per spray, suspension nasal inhaler administered with two sprays in each nostril daily

•  Anastrozole: 1 mg by mouth daily

•  Cetirizine: 10 mg tablet by mouth daily

•  Levothyroxine: 50 mcg tablet by mouth daily

Challenges and Questions

Staging information is not always available or is incomplete. Clinical cancer staging was completed based on A.B.’s surgical procedure and findings when pathologic staging should have completed. Histology of breast cancer was not received. Stage is shown as III and IIB. The patient was diagnosed when the staging system changed; ideally, the staging system used should be noted.

•  The total dose of radiation would be important to have documented in case A.B. has local regional recurrence.

•  Radiation and surgery to axilla increases the risk of lymphedema.

•  Genetic counseling was not pursued at diagnosis. Rationale was not provided.

•  Considerations related to chemotherapy

•  Taxanes and the risk of peripheral neuropathy

•  Trastuzumab and the risk of cardiotoxicity (What was A.B.’s ejection fraction prior to treatment? Did ejection fraction drop during treatment?)

•  Aromatase inhibitors and the risk of hot flashes, vaginal dryness, osteoporosis, and hyperlipidemia

•  Fatigue and depression are common among cancer survivors; rule out other diagnoses that can contribute to fatigue.

•  Stress the need for continued screening for second cancers, including all general population cancer screenings.

•  Focus on assessing how A.B. currently feels emotionally and physically compared to when she finished treatment.

Commentary 1

Amber whitton-smith, ms, np-c, ocn ®, initial assessment and patient teaching.

I would begin by reviewing A.B.’s chart and treatment plans, analyzing the chemotherapy and antiendocrine therapy used and focusing on known side effects. I would assess neuropathy, skin or nail changes, and any potential late cardiovascular effects from trastuzumab treatment. Because of the aromatase inhibitor A.B. is on, I would assess for hot flashes, vaginal dryness, and/or dyspareunia. She received radiation, so I also want to assess any skin changes, evaluate for lymphedema of the breast and arm, and assess for cording (tightness, typically in the axilla, that can extend down the arm after surgery). I would review the dual-energy x-ray absorptiometry (DEXA) scan to determine if she needs a bone-modifying agent while on treatment.

I would provide education surrounding signs of heart failure, such as shortness of breath, exercise intolerance, edema, and/or cough. I would discuss management of neuropathy, and I would avoid inserting needles or taking blood pressure on the side where A.B. had a partial mastectomy and sentinel node biopsy. Because of aromatase inhibitor treatment, I would discuss hot flashes, joint aches, and sexual dysfunction; then, I would assess tolerance, emphasizing the need to communicate with providers to help manage symptoms. This is important; adhering to aromatase inhibitor treatment plans decreases breast cancer recurrence by 50% (Cuzick, 2005). We should also discuss the importance of bone health while A.B. is taking aromatase inhibitors, and I would encourage engaging in weight-bearing exercise and consuming calcium (1,000–1,200 mg daily in divided doses as she is postmenopausal) and vitamin D (typically a minimum of 400 IU daily or dose needed to achieve normal range). If her vitamin D level is normal, I would recheck it annually; if the value is low, I would recheck it every three to six months and titrate vitamin D doses. If A.B. was started on denosumab or biphosphonate, I would discuss the importance of dental visits every six months, as well as the importance of notifying her dentist that she is on these medications because of the risk of osteonecrosis of the jaw. A DEXA scan is indicated every two years while on aromatase inhibitors, and if significant bone loss occurs despite intervention, I would consider changing A.B. to tamoxifen. I would emphasize the importance of regular health maintenance, such as annual mammograms, colonoscopies, and vaccines.

Long-Term Symptom Management

A.B. reports depression, anxiety, fatigue, neuropathy, lymphedema, memory concerns, and sexual dysfunction. If I had seen the patient prior, I typically would know which issues are of most concern and what needs a follow-up. If I am seeing the patient for the first time, I would review the notes or recent questionnaire responses to determine the most important issue of concern to address at the visit. Oftentimes, I ask the questions, “How much does this issue affect you on a daily basis?” or “How much emotional distress does this cause you?”

Follow-Up Assessment and Patient Teaching

Referral to a lymphedema specialist may be helpful for assessment of A.B.’s skin and lymphedema. If neuropathy is affecting activities of daily living, I would review potential medications but also ensure that B12 and thyroid studies are normal. I would refer her for echocardiogram or multigated acquisition scan if concerned about late-onset cardiomyopathy. In addition, I would assess A.B.’s mental health and any possible depression by administering the Patient Health Questionnaire–9. If other factors are affecting her mental health, I would refer her to a social worker, counselor, or support group. Finally, I would assess her sexual health, making recommendations or appropriate referrals and discussing the possibility of a Breast Cancer Index test to see if A.B. qualifies for extended antiendocrine therapy.

Data and Screening to Establish as Baseline

I would monitor these vitals: height and weight, blood pressure, complete blood cell count, comprehensive metabolic panel, and tumor markers, such as carcinoembryonic antigen and cancer antigen 27-29 (these are controversial), as well as laboratories to confirm menopausal status and vitamin D levels. For A.B., I also would want a thyroid-stimulating hormone test because of her thyroid disease and complaints of fatigue. Postsurgery or radiation therapy mammograms are indicated. The National Comprehensive Cancer Network’s (2021b) guidelines recommend intervals for diagnostic mammograms following diagnosis; women who undergo bilateral mastectomy do not require imaging. Clinical examinations should be performed every 6–12 months.

Because breast cancer staging changed in 2018, it would helpful to have pathology to determine the size of the tumor and of metastases to the lymph node as well as A.B’s current staging. Histology is also important, as well as total dose of radiation. During chemotherapy, A.B. developed amenorrhea and subsequently was started on an aromatase inhibitor; this is concerning because she was either pre- or perimenopausal at the time of her treatment. I would order laboratory tests (e.g., follicle-stimulating hormone, estradiol) to confirm that she was menopausal before prescribing an aromatase inhibitor. Prior to starting a patient on an aromatase inhibitor, a baseline DEXA scan and bone health assessment should be completed. I would also like reports on most recent breast imaging, as well as the last reported echocardiogram from when she was on HER2-directed therapy. Given A.B.’s family history, I am concerned that she has not had genetic testing done, because those results could have directed whether she should have had mastectomy versus partial mastectomy.

Commentary 2

Rachael schmidt, dnp, fnp-c, aocnp ®.

While A.B. is waiting to be seen for her visit, I would have her complete the NCCN (2021c) survivorship assessment questionnaire to evaluate fatigue, depression, lymphedema, sexual function, exercise, nutrition, sleep, and other symptoms. I would review the responses and, for any indicating difficulty, I would probe deeper. Specifically, for this patient, I would do the following:

•  Ask more about menopausal symptoms: Is A.B. experiencing vaginal dryness, hot flashes, or pain with intercourse? What intervention has she tried?

•  Does she snore? This could contribute to fatigue and her high blood pressure or shortness of breath if she has sleep apnea.

•  When was her last Papanicolaou test (general population screening)?

•  Assess skin for lymphedema, as well as consideration of the early signs of skin cancer.

•  Is she up to date on vaccines?

•  Assess lifestyle behaviors, including exercise, nutrition, and alcohol use.

Recommendations Based on Patient History and Symptoms

•  Order an annual bilateral mammogram for breast cancer surveillance.

•  Order an echocardiogram related to new shortness of breath, hypertension, and risk of cardiac dysfunction after chemotherapy. Based on the findings, a referral to a cardio-oncologist for further management could be indicated.

•  Order lisinopril 10 mg for A.B.’s hypertension, or determine who is managing her hypertension. If referred to a cardio-oncologist, the provider could manage the cardiovascular issues. She will require ongoing blood pressure monitoring, so it would be important to clarify who is doing this.

•  Order a DEXA scan to establish a baseline for her bone health because of long-term aromatase inhibitor use. Repeat every two years, start supplementation of vitamin D 2,000 IU per day, and encourage including calcium in her diet as well as performing weight-bearing exercises.

•  Order a thyroid-stimulating hormone test and complete blood cell count to rule out other factors that could contribute to fatigue (e.g., hypothyroidism, anemia).

•  Order a lipid panel because of long-term aromatase inhibitor use and the risk for cardiac dysfunction.

•  Order a colonoscopy for general population screenings for colon cancer.

•  Order duloxetine for peripheral neuropathy and depression.

•  Order a home sleep study if positive for snoring.

•  Refer to physical therapy for evaluation and management of lymphedema.

•  Refer to genetics for counseling and testing because of a family history of pancreatic cancer.

Because of A.B.’s history, I would follow up in person in three months to reassess symptoms and blood pressure, see if any side effects occurred with duloxetine or is a dose adjustment is needed, and ensure referrals have been completed. However, if A.B. visited a cardiologist (or cardio-oncologist) and that provider is following her blood pressure, then an in-person visit would not be necessary if she preferred telehealth. Moving forward, I would recommend at least annual follow-up with in-person visits based on her diagnosis, provided her symptoms are controlled. That way, I could continue to address these symptoms, in addition to other quality-of-life factors and lifestyle changes, and ensure this patient is not lost to follow-up. Depending on her genetic testing results, it may be necessary to talk sooner if she has a pathogenic variant that requires additional screenings.

Commentary 3

Kristie howlett, ms, cns, np-c, aocnp ®, initial assessment.

As a new survivorship clinician to A.B., I would start off by explaining what my role is in relation to her healthcare team. Sometimes, as a survivorship clinician, I take a very clinical role to discuss laboratory work and diagnoses, and sometimes I take a health coaching role to help with behavior change. I always stress to patients that change can be very difficult but can also be very achievable with support. I find explaining the role beneficial because, in my clinical experience, many patients do not know what a survivorship clinic is or what survivorship means.

This initial assessment would consist of a complete history and physical. I would then assess A.B.’s ongoing symptoms and issues with several tools: the Functional Medicine Matrix to guide my questions; Medical Symptom Questionnaire; and patient completion of a diet, nutrition, and lifestyle journal over three to five days. This might include assessment of issues, such as how often and how severe her hot flashes are, how often she experiences fatigue (what makes it worse and what makes it better), and whether she has taken any medication or supplements for these side effects. I would also want to know how often A.B. experiences shortness of breath and heart racing. I would explain the importance of understanding her symptoms so that I can make the correct and best recommendations and referrals for care. For example, given that she has a history of trastuzumab treatment and is now experiencing shortness of breath and racing heart with walking upstairs, she may warrant cardiac referral or, at a minimum, an echocardiogram or multigated acquisition scan to evaluate left ventricular ejection fraction.

Baseline Data

After establishing rapport and completing a history and physical, I would want to obtain some baseline data, including basic laboratory values (complete blood cell count, complete metabolic panel, vitamin D) and metabolic laboratory results (hemoglobin A1c, fasting lipids, fasting glucose, fasting insulin). I find it important to evaluate metabolic laboratory results as metabolic health (i.e., diabetic or prediabetic?). This is related to the increased risk of recurrence and increased risk of cancer in postmenopausal women (Gunter et al., 2015). Functional medicine looks at the root cause of disease. In cancer, this can be a difficult dynamic but can be helpful when looking at overall health after diagnosis and treatment. Integrative medicine looks at the patient holistically, considering all lifestyle factors and allowing a deeper relationship with the patient. I have found that looking at the patient through the lens of integrative medicine and functional medicine helps me dive deeper into overall health, with the goal of improved health in survivorship.

I would want to obtain a DEXA scan because it does not appear it was done when starting anastrozole. Given A.B.’s age at diagnosis, family history of breast and/or ovarian cancer, and her ethnic background, I would inquire about genetic testing as I do not know if she was offered genetic counseling and testing and declined or if she was not offered genetic counseling and testing. Is she interested in genetic testing? If so, I would refer her to a genetic counselor for evaluation and testing.

Finally, I would assess A.B.’s lifestyle behaviors, including nutrition, exercise, sleep, and stress. Her body mass index is on the higher side of normal, and no information is included about her physical activity, nutrition, alcohol intake, sleep, stress level, or vitamins or supplements. She may benefit from a referral to a dietitian depending on her nutritional assessment (i.e., diet, nutrition, and lifestyle journal). It is important to assess standard and metabolic laboratory values at baseline and then again at surveillance visits. A.B. seems to be adherent to follow-ups and mammograms and takes her aromatase inhibitor. Depending on her DEXA scan results, she may benefit from a bisphosphonate. Given her age, A.B. should undergo a colonoscopy or at least a fecal immunochemical test, and I would make referrals to the gastrointestinal clinic.

Considerations

Based on the lifestyle behavior assessment and other assessment, A.B. may benefit from more frequent follow-ups, which could be scheduled as telehealth visits. To facilitate adherence to a plan, it is important for the patient to participate in decision-making about lifestyle changes and follow-up appointments.

Overall, it is important to look beyond any survivorship checklist because not everything a patient is experiencing may be related to long-term side effects. Symptoms could be labeled as long-term side effects because someone has had chemotherapy. However, with a deeper dive using a root cause functional medicine approach, I have seen many patients improve symptoms with lifestyle changes. Listening to the history and looking at the interactions among genetic, environmental, and lifestyle factors can influence long-term health. This approach to complex, chronic disease supports a unique expression of health and vitality for each individual (Institute for Functional Medicine, 2021).

Summary Commentary

Rachée D. Hatfield, MS, APRN-CNP

Each commentary is quite thorough, with many commonalities and some differences. In each case, the provider indicates the importance of reviewing the patient’s cancer treatment and other health history, as well as understanding current symptoms to develop the appropriate workup and treatment plan. Although perhaps obvious, this is the foundation of providing the highest quality patient care.

In Commentary 3, the clinician explains the role of a survivorship clinician to the patient. It is also my experience that most patients do not understand the meaning of cancer survivorship or what a survivorship clinician does. I agree that sometimes the role can differ depending on the needs of the patient at the time. Sometimes the role is clinical, sometimes it is psychological for a patient whose distress is more emotional in nature, and sometimes it is very much a health coaching role. Often, it is little bit of each. What resources are available may determine the degree to which a survivorship clinician takes on each of these roles. At a comprehensive cancer center, many supportive care resources may be available (e.g., physical therapist, occupational therapist, dietitian, counseling), but if the clinician works in a smaller organization or rural community, they may need to take on all or some of these roles and/or become familiar with local, state, or national resources to refer and support the patient.

As discussed in Commentary 1, some important clinical information is missing from the treatment summary, as well as some contradictory information that would be useful to clarify if possible (i.e., biomarkers). A copy of the surgical pathology report should provide histologic type, tumor size, margins and lymph nodes (total removed and how many with metastatic disease), and biomarkers. The American Joint Committee on Cancer (AJCC) updated breast cancer staging in 2018 (version 8), recognizing the prognostic importance of cell grade and biomarkers (Guiliano et al., 2018). Assuming the treatment summary provides the correct tumor, node, and metastasis classification of malignant tumors and grade (T2, N1, M0, grade 3), the patient’s cancer would be pathologic stage IB (AJCC, version 8, for ER/PR+ and HER2+) versus stage IIB (AJCC, version 7, and AJCC, version 8, for estrogen receptor/progesterone receptor–negative and HER2+). Understanding these details could inform a provider’s level of concern for distant recurrence, allowing for the provision of realistic information and guidance to the patient.

Evaluating causes for shortness of breath or heart racing is the highest priority. Although the possibility of cardiac dysfunction (given the history of trastuzumab) is appropriate to evaluate, I would also want to consider less likely causes, such as lung metastasis or long-term effects of radiation on the heart and lungs. I agree with each commentary about ordering an echocardiogram or multigated acquisition scan; I would also obtain an electrocardiogram based on the patient’s report of heart racing. In addition, I would consider obtaining chest computed tomography (perhaps pending results of cardiac testing). Clinical examination findings may also be useful to assess for signs of heart failure. Cardiac dysfunction could cause or contribute to fatigue.

Aligning with every expert commentary, I would order a complete blood cell count, thyroid-stimulating hormone test, vitamin D level, and complete metabolic panel. The findings could reveal or eliminate causes for fatigue (e.g., hypothyroidism, vitamin D deficiency, anemia). I agree with obtaining hemoglobin A1C (metabolic implications) and a fasting lipid profile (cardiac risk). Commentary 1 recommends obtaining hormone laboratory values, which is important because it is unclear whether menopausal status was established prior to starting chemotherapy and initiating aromatase inhibitor. I also agree that a DEXA scan is indicated for A.B., who is on an aromatase inhibitor. Regardless of the results, I would educate her on the daily recommended intake for calcium (600 mg twice daily) and vitamin D (800–1,000 IU daily) and the benefits of weight-bearing exercise (Bone Health and Osteoporosis Foundation, n.d.-a, n.d.-b). I would also inform her of the possible need for bisphosphonate therapy pending results of the DEXA scan.

In Commentary 1, the recommendation was to obtain vital signs, height, and weight at baseline and at surveillance visits. Height and weight are important in monitoring body mass index. Being overweight is a risk factor for cancer recurrence or the development of other cancers and increases the risk of other comorbidities (National Cancer Institute, 2017; National Institute of Diabetes and Digestive and Kidney Diseases, 2017). The patient’s blood pressure is elevated at this visit. It would be important to continue to monitor to see whether this is a trend or an isolated incident. For a patient experiencing shortness of breath, additional vitals to assess include respiratory rate and oxygen saturation.

Fatigue is by far the most common symptom I hear reported by cancer survivors. The causes can be widely varied and may or may not be related to cancer treatment. I agree with the recommendation to assess how the patient feels now compared to closer to when she completed treatment. I agree that too often symptoms are labeled as long-term side effects because someone has had chemotherapy. Lifestyle habits can play a big part in fatigue, so it is important to assess and educate the patient on the importance of sleep, regular exercise, good nutrition, and stress management.

Hot flashes are very common and can be present at varying degrees. I would assess how bothersome they are to the patient (affecting sleep and quality of life). For more mild symptoms, education can be provided on avoidance of triggers and helpful lifestyle habits. For more significant symptoms, prescription medications can be considered.

As indicated in each commentary, A.B. should be referred for genetic counseling; she meets NCCN (2022) criteria for testing. I concur that it would be helpful to clarify if referral was offered and declined or was not offered at all. I find that taking a few minutes to explain what to expect at a genetic counseling session and the implications of finding a genetic variant helps the patient better understand the benefit. At minimum, a genetic counselor can provide a comprehensive genetic assessment that can be used to make an informed decision about whether to pursue testing. Some patients misunderstand, believing a variant cannot be passed down from the paternal side, and some do not realize that some variants increase the risk for more than one kind of cancer.

I recommend using the NCCN (2021c) survivorship assessment questionnaire as a tool to more succinctly evaluate patient symptoms. Use of a tool like this can be valuable in gathering a comprehensive picture of what the patient is experiencing. In my workplace, we use a validated distress screening tool that asks the patient to rate distress experienced in the past week related to specific symptoms of emotional, spiritual, social, practical, cognitive, and physical health on a scale from 0 to 3 (0 = none, 1 = mild, 2 = moderate, and 3 = severe). It also asks about concerns related to sexual function; nutrition; use of tobacco, alcohol, or prescription drugs; how family and friends are coping; and interest in information on these topics, which really helps to guide my visits (Wells-Di Gregorio et al., 2013).

One question I have is whether this patient has a primary care provider. Although most cancer survivors will be followed by an oncology specialist for at least five years post-treatment, most will eventually be released to continue follow up with their primary care provider. It is important for that primary care provider to be aware of a patient’s cancer history, surveillance recommendations, and possible late or long-term side effects of treatment. This is where an accurate, complete treatment summary and survivorship care plan can be of great use; a copy can be provided to the primary care provider. In the long term, the primary care provider manages age-appropriate cancer and noncancer health screenings (e.g., colon, prostate, blood pressure, glucose, cholesterol) and health maintenance (i.e., vaccines), education on healthy lifestyle habits, and management of physical and mental health. If the patient does not have a primary care provider, I would refer her to establish care.

A couple of other comments that really stood out to me were (a) normalizing the experience and validating to the patient that what she is experiencing is common and treatable and (b) the importance of including the patient in decision-making. Each patient differs in their level of knowledge, receptiveness to education and counseling, and motivation to make changes for the betterment of their health. In my practice, I often challenge patients to set a goal or two for their health and wellness. We have an opportunity in cancer survivorship to empower patients to take charge of their future health, providing support in meeting their goals.

I think that what is most challenging in survivorship care (and perhaps this is the nature of being a nurse) is that I cannot fix everything. I realize that this is not possible and that most patients do not expect me to fix everything. Most patients appreciate that someone has listened to their fears and concerns, and are grateful to have some counseling or recommendations for resources that can help, if appropriate.

Additional Considerations

Suzanne M. Mahon, DNS, RN, AOCN ® , AGN-BC, FAAN

•  Regularly consider and assess the risk of cardiovascular disease secondary to aromatase inhibitor treatment or monitoring cholesterol levels. Also, clarify who is assessing and managing the patient’s cardiovascular disease (e.g., primary care, oncology team, cardio-oncologist).

•  Psychosocial distress and depression merit more intensive assessment to select the best management strategies (NCCN, 2021e).

•  NCCN (2021a) guidelines do not recommend checking tumor markers and have very specific guidelines for follow-up, presented in an algorithm format for healthcare providers. These are also available for patients in a bullet list format (NCCN, 2020). Sharing these with patients can help provide assurance that follow-up is adequate and guidelines-based.

•  Medical problems, such as hypertension, can be identified in follow-up visits. This creates the challenge of determining whether it is a long-term consequence of prior therapy, as well as clearly determining who will complete the diagnostic workup (oncology team or primary care).

•  Fatigue is very real for those in long-term follow-up, requiring assessment of sleep patterns. NCCN (2021d) provides guidelines for assessment and management of cancer-related fatigue.

•  Germline genetic testing affects treatment decisions and recommendations for long-term surveillance for the patient and possibly other family members. On at least an annual basis, review the family history. If the family history has changed or a pathogenic variant has been identified in the family, refer the patient to genetics for further evaluation. If a pathogenic variant has been previously identified, review recommendations for care to ascertain that recommendations are appropriate for that pathogenic variant because recommendations change over time, as well as the availability of additional testing.

About the Author(s)

Amber Whitton-Smith, MS, NP-C, OCN ® , is a nurse practitioner at Cypress Oncology/Centura Health in Parker, CO. The author takes full responsibility for this content and did not receive honoraria or disclose any relevant financial relationships. The article has been reviewed by independent peer reviewers to ensure that it is objective and free from bias. Whitton-Smith can be reached at [email protected] , with copy to [email protected] . (Submitted June 2021. Accepted August 26, 2021.) 

Rachael Schmidt, DNP, FNP-C, AOCNP ® , is a survivorship and cancer risk/prevention program director at Nebraska Medicine in Omaha. The author takes full responsibility for this content and did not receive honoraria or disclose any relevant financial relationships. The article has been reviewed by independent peer reviewers to ensure that it is objective and free from bias. Schmidt can be reached at [email protected] , with copy to [email protected] . (Submitted June 2021. Accepted August 26, 2021.) 

Kristie Howlett, MS, CNS, NP-C, AOCNP ® , is an oncology/survivorship nurse practitioner at Howlett Integrative Cancer Care PLLC in Roseville, CA. The author takes full responsibility for this content and did not receive honoraria or disclose any relevant financial relationships. The article has been reviewed by independent peer reviewers to ensure that it is objective and free from bias. Howlett can be reached at [email protected] , with copy to [email protected] . (Submitted June 2021. Accepted August 26, 2021.) 

Rachée D. Hatfield, MS, APRN-CNP, is a certified nurse practitioner at the Ohio State University Wexner Medical Center–The James in Columbus. The author takes full responsibility for this content and did not receive honoraria or disclose any relevant financial relationships. The article has been reviewed by independent peer reviewers to ensure that it is objective and free from bias. Hatfield can be reached at [email protected] , with copy to [email protected] . (Submitted June 2021. Accepted August 26, 2021.)

Suzanne M. Mahon, DNS, RN, AOCN ® , AGN-BC, FAAN, is a professor in the Department of Internal Medicine in the Division of Hematology/Oncology and in the Trudy Busch Valentine School of Nursing at Saint Louis University in Missouri. The authors take full responsibility for this content and did not receive honoraria or disclose any relevant financial relationships. The article has been reviewed by independent peer reviewers to ensure that it is objective and free from bias. Mahon can be reached at [email protected] , with copy to [email protected] . (Submitted June 2021. Accepted August 26, 2021.)

Related Articles

Sleep quality and associated factors among survivors of breast cancer: from diagnosis to one year postdiagnosis, cardiovascular disease incidence and cardiovascular health among diverse women with breast and gynecologic cancers, associations of demographic and social factors on health-related quality-of-life changes among older women with breast or gynecologic cancer.

• Around the Practice
• Between the Lines
• Contemporary Concepts
• Readout 360
• Insights from Experts at Mayo Clinic on Translating Evidence to Clinical Practice
• Optimizing Outcomes in Patients with HER2+ Metastatic Breast Cancer

• Conferences
• Publications

Clinical Case Presentation: A 36-Year-Old Woman with Breast Cancer and Brain Metastases

• Ruta Rao, MD

Ruta Rao, MD, presents the case of a 36-year-old woman with metastatic HER2+ breast cancer and brain metastases.

EP: 1 . Clinical Case Presentation: A 36-Year-Old Woman with Breast Cancer and Brain Metastases

EP: 2 . Treatment Options in the Frontline Setting for Metastatic HER2+ Breast Cancer

EP: 3 . Second-Line and Third-Line Treatment Options for Metastatic HER2+ Breast Cancer

Ep: 4 . tucatinib for metastatic her2+ breast cancer and brain metastases: patient selection, ep: 5 . treatment options for her2+ breast cancer after progression on tucatinib, ep: 6 . clinical case presentation: a 53-year-old woman with metastatic er/pr+ her2+ breast cancer and brain metastases, ep: 7 . neratinib for metastatic her2+ breast cancer and brain metastases, ep: 8 . leptomeningeal metastases in her2+ breast cancer, ep: 9 . novel agents under evaluation for her2+ breast cancer and brain metastases, ep: 10 . clinical case presentation: a 66-year-old woman with er+ her2+ invasive ductal carcinoma and brain metastases, ep: 11 . trastuzumab deruxtecan treatment for metastatic her2+ breast cancer with brain metastases, ep: 12 . role of neurooncologists in management of her2+ breast cancer and brain metastases.

EP: 13 . Recap: Updates in Treatment of HER2-Positive Breast Cancer and Brain Metastases

Real-World Data Support SLNB Omission in HR+ HER2– Breast Cancer

Oncologic outcomes from a real-world analysis appear to be comparable with those observed in the sentinel lymph node biopsy arm of the SOUND trial.

Managing CDK4/6 Inhibitor, ADC Toxicity in Metastatic Breast Cancer

Sarah Donahue, MPH, NP, speaks to the importance of communicating potential adverse effects associated with treatments such as CDK4/6 inhibitors to patients with breast cancer.

Diagnostic Delays in Younger Women With Breast Cancer Are Mostly Patient-Related

Young patients with breast cancer were more likely to be the reason for diagnostic delays vs system delays, partly due to less concern for their symptoms.

HER2CLIMB-02 Trial Shows ‘Interesting Data’ in HER2+ Breast Cancer

Tucatinib plus trastuzumab emtansine shows a progression-free survival improvement in HER2-positive breast cancer in the phase 3 HER2CLIMB-02 trial, says Sara A. Hurvitz, MD, FACP.

Novel Injection Earns FDA ODD in Breast Cancer Brain Metastases

Utidelone injection may cross the blood-brain barrier and help treat patients with breast cancer brain metastases.

FDA Accepts BLA for Dato-DXd in Metastatic HR+/HER2– Breast Cancer

Data from the phase 3 TROPION-Breast01 trial support the biologics license application for datopotamab deruxtecan in hormone receptor–positive, HER2-negative breast cancer.

2 Commerce Drive Cranbury, NJ 08512

609-716-7777

• Open access
• Published: 16 October 2021

Light at night and risk of breast cancer: a systematic review and dose–response meta-analysis

• Teresa Urbano   ORCID: orcid.org/0000-0003-1490-5508 1 ,
• Marco Vinceti   ORCID: orcid.org/0000-0002-0551-2473 1 , 2 ,
• Lauren A. Wise   ORCID: orcid.org/0000-0003-2138-3752 2 &
• Tommaso Filippini   ORCID: orcid.org/0000-0003-2100-0344 1  

International Journal of Health Geographics volume  20 , Article number:  44 ( 2021 ) Cite this article

7822 Accesses

28 Citations

40 Altmetric

Metrics details

Breast cancer is the most common malignancy in women and the second leading cause of cancer death overall. Besides genetic, reproductive, and hormonal factors involved in disease onset and progression, greater attention has focused recently on the etiologic role of environmental factors, including exposure to artificial lighting such as light-at-night (LAN). We investigated the extent to which LAN, including outdoor and indoor exposure, affects breast cancer risk. We performed a systematic review of epidemiological evidence on the association between LAN exposure and breast cancer risk, using a dose–response meta-analysis to examine the shape of the relation. We retrieved 17 eligible studies through September 13, 2021, including ten cohort and seven case–control studies. In the analysis comparing highest versus lowest LAN exposure, we found a positive association between exposure and disease risk (risk ratio [RR] 1.11, 95% confidence interval-CI 1.07–1.15), with comparable associations in case–control studies (RR 1.14, 95% CI 0.98–1.34) and cohort studies (RR 1.10, 95% CI 1.06–1.15). In stratified analyses, risk was similar for outdoor and indoor LAN exposure, while slightly stronger risks were observed for premenopausal women (premenopausal: RR 1.16, 95% CI 1.04–1.28; postmenopausal: 1.07, 95% CI 1.02–1.13) and for women with estrogen receptor (ER) positive breast cancer (ER + : RR 1.09, 95% CI 1.02–1.17; ER–: RR 1.07, 95% CI 0.92–1.23). The dose–response meta-analysis, performed only in studies investigating outdoor LAN using comparable exposure assessment, showed a linear relation up to 40 nW/cm 2 /sr after which the curve flattened, especially among premenopausal women. This first assessment of the dose–response relation between LAN and breast cancer supports a positive association in selected subgroups, particularly in premenopausal women.

Introduction

Breast cancer is the most common cancer in women in both developed and undeveloped countries [ 1 ]. In Italy, there were an estimated 55,000 new cases in 2020 [ 2 ], and while incidence is increasing, mortality rates have significantly decreased across the years. Several factors appear to be involved in both etiology and prognosis of this malignancy, including selected genes, ageing, family history, reproductive factors, long-term use of postmenopausal female hormones, lifestyle [ 3 , 4 ], and environmental factors such as exposure to chemical endocrine disruptors [ 5 , 6 , 7 , 8 , 9 ].

In 2007, the International Agency for Research on Cancer (IARC) defined night-shift work as ‘probably carcinogenic to humans’ i.e. belonging to category 2A [ 10 ], due to a possible link with breast cancer [ 11 , 12 ] and with prostate and colorectal cancer [ 13 ]. The definition of night-shift work, also identified as graveyard shift, refers to a work schedule involving the sleeping hours of the general population. Among the exposure linked to graveyard shift work there is light-at-night (LAN), which induces disruption of the circadian rhythm and oxidative stress [ 14 ]. In particular, LAN could be involved in breast cancer etiology through different mechanistic pathways including DNA damage, impairment of melatonin and estrogen secretion, inflammation, and disruption of metabolic function [ 15 ]. Exposure to LAN can cause circadian and sleep disruptions, which may adversely affect different inflammatory and immunological pathways, thereby decreasing production of circulating natural killer cells or enhancing pro-inflammatory effects [ 16 , 17 ]. Since sleep has antioxidant effects, its disruption may also cause increased production of reactive oxygen species [ 18 , 19 ]. When evaluating exposure to LAN, both outdoor (e.g., streetlamps, illuminated buildings, lights from vehicles) and indoor artificial sources (e.g., domestic lights, electric devices-derived illumination) are relevant in increasing circadian disruption and therefore the risk of developing cancer. Outdoor LAN is generally assessed using satellite-derived data, while indoor is often evaluated through surveys on night habits and bedroom light characteristics [ 20 ].

Two recent meta-analyses summarized data on the association between LAN and breast cancer risk, investigating the effects of the highest vs. the lowest LAN exposure categories [ 21 , 22 ]. Since the publication of these meta-analyses, three large studies of the LAN-breast cancer association have been published [ 23 , 24 , 25 ]. In the present report, we update the meta-analysis with these new studies, perform subgroup analyses by breast cancer subtype and other factors, and more comprehensively assess the epidemiological evidence about LAN and breast cancer risk. In addition, we apply dose–response meta-analyses to assess, for the first time, the shape of the association between LAN and breast cancer.

We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) 2020 statement [ 26 ] to perform this review.

Study identification and selection

The research question was configured according to PECOS statement (Population, Exposure, Comparator(s), Outcomes, and Study design): “Is exposure to light-at-night, as assessed through indoor and outdoor exposure to lighting sources, positively associated with risk of breast cancer in non-experimental studies?” and “Is there a dose–response association between LAN and breast cancer incidence?” [ 27 ]. Accordingly, we carried out a systematic literature search for publications available as of September 13, 2021 in the PubMed/MEDLINE, Embase and Web of Science (WoS) databases. We used search terms linked to “breast cancer” and “lighting” in PubMed, WoS, and Embase databases with no language restrictions (Additional file 1 : Table S1). We also performed citation chasing by scanning the reference list of included studies and of previous reviews, as well as backward/forward citation retrieval to identify additional relevant papers [ 28 ]. Inclusion criteria were as follows: titles including LAN and breast cancer; mentioning case–control/case-cohort/cohort studies; monitoring LAN from space according to the US Defense Meteorological Satellite Program (DMSP) Operational Linescan System or from the Visible Infrared Imaging Radiometer Suite Day-Night Band (DNB); evaluating indoor LAN based on self-reported questionnaires and mentioning LAN levels as low, medium or high, and darkness and nightlight levels, and habits of sleeping with lights on; reporting of risk estimates for breast cancer, along with their 95% confidence intervals, or availability of enough data to calculate them.

Two authors reviewed all titles and abstracts independently, and any conflicts were resolved with the help of third author. For each included study, we extracted information about design, population size and characteristics, country, study period and years of follow-up, risk estimates (either odds ratio, risk ratio, or hazard ratio) along with their 95% confidence interval (CI) of breast cancer, adjustment factors, type of exposure assessment, and dose of exposure.

Quality assessment

We assessed the quality in the included studies by using the Risk Of Bias In Non-randomized Studies of Exposures (ROBINS-E) tool [ 29 ]. We classified studies as having low, moderate, or high risk of bias according to seven domains: bias due to confounding; bias in selecting participants in the study; bias in exposure classification; bias due to departures from intended exposures; bias due to missing data; bias in outcome measurement; and bias in the selection of reported results. In Additional file 1 : Table S2, we report criteria for risk of bias evaluation, performed by two authors. In case of disagreement, a third author helped in the final decision. A study’s overall risk of bias was considered high or moderate if at least one domain was judged at high or moderate risk, otherwise it was classified as having a low risk of bias.

Statistical analysis

We performed a meta-analysis comparing breast cancer incidence in the highest versus lowest levels of LAN exposure using a restricted maximum likelihood random effect model, which bases estimates on a likelihood function calculated from a transformed set of data [ 30 ]. Additionally, whenever possible, we carried out a dose–response meta-analysis of breast cancer risk according to increasing LAN exposure through a random-effects model, using a one-stage approach as previously implemented in other fields [ 31 , 32 , 33 ]. Specifically, for each LAN category, we used the mean or the median value, or the midpoint for the intermediate categories, whichever was available. For the highest and lowest exposure categories, if the average values were not reported and were ‘open’, we used as boundary a value 20% higher or lower than the closest cut-point. We used a restricted cubic spline model with three knots at fixed percentiles (10th, 50th, and 90th) and we considered the correlation within each set of published effect estimates using generalized least-squares regression through a multivariate random-effect meta-analysis, incorporating the restricted maximum likelihood method [ 30 , 34 ].

Besides the overall group, we also performed stratified analyses according to menopausal status (pre and postmenopausal), body mass index-BMI (< 25 and ≥ 25), estrogen receptor-ER status (ER + and ER–) of cases, and type of LAN exposure (outdoor and indoor). Furthermore, we explored the role of possible effect modifiers, by dividing the studies according to the country-specific estimated annual sunshine hours [ 35 ] into the three subgroups (< 2000, 2000–3000, and > 3000 annual mean sunshine hours), and country solar ultraviolet B (UV-B) radiation [ 36 ].

We assessed the potential for small-study bias using funnel plots for studies reporting highest versus lowest exposure, and by performing Egger’s test [ 37 , 38 ] and trim-and-fill analysis [ 39 ]. We also evaluated the effect of variation across studies through the graphical overlay of study-specific predicted curves by using fixed and random effects [ 34 ]. Finally, we assessed heterogeneity by reporting I 2 statistics, and by carrying out stratified analyses whenever possible such as for LAN exposure assessment method, menopausal status, participants’ weight (normal vs overweight/obese), and ER status. We used Stata software (v 16.1, 2021—Stata Corp., College Station, TX), namely its ‘meta’ and ‘drmeta’ routines, for data analysis.

Overall, of the 494 individual studies identified after removal of duplicates, we excluded 465 studies due to title and abstract screening, and 13 additional studies after full-text evaluation, leaving 17 studies eventually fulfilling inclusion criteria (Fig.  1 ). Main reasons of exclusion were the following: insufficient data, commentaries, reviews or meta-analyses, editorials, ecological studies, or lack of LAN exposure assessment (reasons reported in detail in Additional file 1 : Table S3).

Flowchart summarizing the literature search and identification

Table 1 presents characteristics of the 17 included studies, three with case-cohort design [ 40 , 41 , 42 ], six cohort [ 23 , 24 , 25 , 43 , 44 , 45 ], and eight case–control [ 46 , 47 , 48 , 49 , 50 , 51 , 52 , 53 ]. The studies were published during 2001–2021, mainly in North America (N = 10), followed by Europe (N = 3), Israel (N = 2), Australia (N = 1), and China (N = 1). Follow-up duration was reported in only two studies (16 and 6.1 years) [ 23 , 42 ]. LAN was assessed according to two main methods: from outdoor (N = 7) [ 23 , 24 , 40 , 41 , 43 , 44 , 52 ] and indoor (N = 11) sources [ 41 , 42 , 43 , 45 , 46 , 47 , 48 , 49 , 50 , 51 , 53 ]. In all cohort and case-cohort studies assessing outdoor LAN, the unit of measurement for LAN was nano-Watt/square centimeters/steradian (nW/cm 2 /sr) [ 23 , 24 , 40 , 43 , 44 , 52 ]. The only exception was a Spanish study that assessed outdoor LAN using an index of outdoor blue light spectrum to calculate melatonin suppression index (MSI). MSI was estimated at each pixel of images of Madrid and Barcelona detected from the International Space Station (ISS) [ 41 ]. Indoor LAN data were evaluated based on self-reported questionnaires, for example, referring to use of lamps during the night or other sources of artificial light in the bedroom while sleeping. Additional characteristics of studies included in the systematic review are shown in the Additional file 1 : Table S4.

Risk of bias assessment (Additional file 1 : Table S5) showed that most studie s were at low risk of bias due to confounding, while four w ere at moderate risk of bias because they did not control for some breast cancer risk factors, typically family history of breast cancer, postmenopausal hormone use, or sm oking [ 24 , 40 , 47 , 48 ]. Concerning exposure assessment, studies assessing outdoor LAN exposure were at low risk of bias [ 23 , 24 , 25 , 40 , 44 , 52 ], while those assessing both outdoor and indoor LAN or only indoor LAN were generally at moderate risk due to possible misclassification bias [ 41 , 42 , 43 , 45 , 46 , 47 , 49 , 50 , 51 , 53 ]. One study was considered at high risk of bias because exposure assessment was based on a non-validated self-administered questionnaire [ 48 ]. All other domains were considered at low risk of bias in all studies, although four studies were judged at moderate risk of bias because some information (i.e., smoking or menopausal status) had been collected but not reported, no differences were found, or no data were presented [ 44 , 46 , 51 ].

Comparing the highest versus the lowest LAN exposure category, we consistently found positive associations with breast cancer risk (summary RR = 1.11, 95% CI 1.07–1.15). In subgroup analyses (Table 2 ), we found positive associations for outdoor (RR = 1.11, 95% CI 1.07–1.16) and indoor (RR = 1.08, 95% CI 1.00–1.17) LAN exposure, as well as both for case–control (RR = 1.11, 95% CI 0.97–1.28) and cohort studies (RR = 1.11, 95% CI 1.07–1.15) as shown in Fig.  2 and Additional file 1 : Figure S1.

Risk ratio (RR) with 95% confidence interval (CI) for the association between light at night exposure and r isk o f breast cancer (N = 17 studies) com paring the highest versus the lowest exposure category in studies assessing outdoor and indoor exposure. The squares represent point estimates of RR and horizontal lines represent their 95% confidence intervals (CIs). The area of each square is proportional to the inverse of the variance of the estimated log RR. The diamonds represent the combined RR for each subgroup and the overall RR for all studies. The solid line represents RR = 1

Eight studies assessed breast cancer risk among both pre and postmenopausal women at the moment of diagnosis, while one was restricted to postmenopausal women only. The summary RR was slightly stronger among premenopausal women (RR = 1.16, 95% CI 1.04–1.28) than postmenopausal women (RR = 1.07, 95% CI 1.02–1.13) (Fig.  3 ). A slight positive association among premenopausal women also emerged in the cohort/case-cohort study subgroup and for outdoor LAN exposure. Conversely, for the case–control category and indoor LAN exposure, the RR was slightly higher among postmenopausal women (Table 2 and Additional file 1 : Figure S2).

Risk ratio (RR) with 95% confidence interval (CI) for the association between light at night exposure and risk of breast cancer (N = 9 studies) among premenopausal and postmenopausal women, comparing the highest versus the lowest exposure category. The area of each grey square is proportional to the inverse of the variance of the estimated log RR. Black diamonds represent point estimates of RR and horizontal lines represent their 95% confidence intervals (CIs). The open diamonds represent the combined RR for each subgroup and the overall RR for all studies. The solid line represents RR = 1

In the dose–response meta-analysis, we found a positive linear relation up to 40 nW/cm 2 /sr of outdoor LAN exposure, after which a plateau was reached (Fig.  4 ). A substantially comparable pattern was noted for all women and in analyses stratified according to menopausal status.

Dose–response meta-analysis between light at night and risk of breast cancer among all women [ 23 , 24 , 25 , 40 , 43 , 44 , 52 ] and between light at night and risk of breast cancer in premenopausal [ 43 , 44 , 52 ] and postmenopausal women [ 23 , 43 , 44 , 52 ]. Spline curve (black solid line) with 95% confidence limits (black dashed lines), and linear trend estimation (long-dashed gray line). RR risk ratio

Concerning possible effect-modification by ER status, few differences emerged between women with ER + and E R–b r east cancer , for whom the summary RRs were 1.09 (95% CI 1.02–1.17) and 1.07 (95% CI 0.92–1.23), respectively (Table 2 and Additional file 1 : Figure S3). In addit ion, in the dose–response meta-analysis we found that the risk was slightly higher in the ER  −  subgroup compared to the ER + one. A difference also emerged above 30 nW/cm 2 /sr of outdoor LAN exposure, when the curve flattened in the ER + subgroup while continued to increase in the ER − one (Additional file 1 : Figure S4).

With regards to BMI status, the two studies of breast cancer risk among women with BMI < 25 or BMI ≥ 25 yielded similar positive summary RRs in both categories, though RRs were stronger in the normal-weight (BMI < 25) group (RR = 1.17, 95% CI 1.00–1.36 and RR = 1.07, 95% CI 0.87–1.32, respectively (Table 2 and Additional file 1 : Figure S5). A monotonic positive association emerged in the dose–response meta-analysis for the two BMI subgroups, despite the very low number of studies (Additional file 1 : Figure S6).

We also performed stratified analyses according to the annual sunshine hours’ map of the world (Table 2 and Additional file 1 : Figure S7), dividing the 17 studies by country latitude. We divided the countries in three different groups of annual sunshine hours (< 2000 h: Canada, UK, Denmark; 2000–3000 h: US, Western Australia, China; > 3000 h: Spain, Israel). Countries with annual sunshine hours < 2000 exhibited null association (RR = 0.97, 95% CI 0.87–1.08). Conversely, we found a stronger association between LAN in the subgroup of countries with 2000 to 3000 annual sunshine hours (RR = 1.12, 95% CI 1.08–1.17) as well as in countries with more than 3000 annual sunshine hours, i.e., Spain and Israel (RR = 1.30, 95% CI 1.11–1.52) (Additional file 1 : Figure S8). These findings were also observed in analyses stratified by postmenopausal status (Additional file 1 : Figure S9) and indoor/outdoor exposure, although for indoor exposure, the risk ratio was slightly higher in countries with 2000 to 3000 annual sunshine hours than those with more than 3000 (Table 2 and Additional file 1 : Figure S10). Finally, when we accounted for estimated equinoctial UV-B irradiance map (Additional file 1 : Figure S11) we found a null association between LAN and breast cancer risk in studies from countries with less UV-B irradiance (RR = 0.97, 95% CI 0.87–1.08), while an inverse association was observed in the outdoor exposure subgroup (RR = 0.91, 95% CI 0.76–1.09). Conversely, there was a positive association in studies from countries with higher (> 0.58 W/m 2 ) UV-B irradiance levels (Additional file 1 : Figure S12), as also confirmed when considering either outdoor or indoor exposure (Table 2 ).

Exclusion of the one study [ 48 ] considered at high risk of bias did not substantially alter the results (Additional file 1 : Table S6). Findings were also similar when we additionally excluded the two studies considered at moderate risk of bias in selection of reported results [ 46 , 51 ] (Additional file 1 : Table S7). To further limit the effect of potential biases, we then performed the analyses excluding three additional studies [ 24 40 , 47 ] considered at moderate risk of bias due to confounding. Still, there were little changes in the results, and the estimates were substantially confirmed (Additional file 1 : Table S8). Similarly, analysis of conditional study-specific lines arising from the estimated random-effects model yielded homogeneous results overall and among premenopausal women, while among postmenopausal women, slightly higher variation was noted (Additional file 1 : Figure S13). Finally, evaluation of small-study bias suggested no occurrence of bias due to symmetric distribution and no studies were added when running trim-an-fill analysis both in overall studies (Additional file 1 : Figure S14) and in analyses stratified by menopausal status (Additional file 1 : Figure S15) and exposure assessment (Additional file 1 : Figure S16).

Higher urbanization has prompted substantial changes in peoples’ lifestyles as compared with our ancestors. Nowadays, over 80% of the World’s population and close to 100% of the people in the United States and Europe live under skies polluted by light [ 54 ], one of the key environmental factors characterizing the Western world environment . Besides residence-related artificial light (i.e., urban light pollution), other sources of non-natural LAN are electronic devices (TVs, smartphones, tablets, computers, etc.) or lights turned on during night at home or at the workplace. The possible carcinogenic effects of LAN has been recently assessed also by the U.S. National Toxicology Program (NTP) cancer hazard assessment [ 55 ]. NTP concluded that there was moderate evidence for a causal relation between LAN exposure and human cancer, since LAN may act through different mechanisms of circadian disruption and its biological effects are the same of well-known recognized carcinogens [ 55 ].

Over the last twenty years, the association between LAN exposure and breast cancer risk has been assessed primarily in occupational settings, specifically among night-shift workers [ 56 , 57 , 58 , 59 , 60 ]. These studies generally found a slight to substantial excess for breast cancer in women working graveyard shifts. Most recently, epidemiological studies investigating LAN exposure, in most cases independently from nightshift work, and its association with risk of breast cancer in the general population have greatly increased. In longitudinal studies, metrics of outdoor LAN have been collected through sophisticated methods such as the US DMSP Operational Line-Scan System or the Visible Infrared Imaging Radiometer Suite DNB, and expressed as nW/cm 2 /sr, except for one study [ 41 ], which was based on a visual artificial light-at-night (ALAN) assessment to estimate ground-based spectrum of the light emission, and melatonin suppression index for outdoor blue light spectrum. A Canadian study used both DNB and DMSP data. Even if DNB has a higher resolution and a calibrated radiometer, DMSP was used in our analysis for comparison with other studies [ 52 ].

Most case–control studies, in turn, have assessed LAN exposure using self-administered questionnaires regarding sleep and/or night habits. Exposure assessment in these studies has included frequency of waking up and turning on lights during night, sleeping with the TV on or off, darkness level in the room, residency near strong artificial LAN sources, wearing a mask while sleeping, keeping lights on or off while sleeping.

There is some biological plausibility for a LAN breast cancer association, given the observation that repeated exposure to artificial light during night hours might induce DNA damage and oxidative stress, alter melatonin and estrogen synthesis and metabolism, inflammation and immune function, and disrupt metabolic function [ 18 ]. More specifically, three mechanisms have been proposed to explain the link between LAN and some types of cancers. LAN could inhibit melatonin secretion directly [ 61 ], through sleep deprivation (also affecting cell proliferation and cytokines production [ 62 ]), and through chronodisruption [ 63 , 64 ]. Along these lines, previous studies referred more generally to night shift work than to LAN exposure, but night shift work is a far more complex exposure, including among other changes in sleep habits, sleep deprivation, eating during the night [ 61 ]. For this reason, we focused our attention more specifically on exposure to LAN as a factor associated with higher incidence of breast cancer, rather than night shift work, though the latter may confound to some extent the association between LAN and breast cancer risk.

We found a consistent positive association between LAN and breast cancer risk overall and among several subgroups, including premenopausal women, those with BMI < 25, and those living in countries experiencing more than 3000 sunshine hours a year. The risk of developing breast cancer was almost monotonically associated with outdoor LAN up to the value of 40 nW/cm 2 /sr, above which the threshold of the curve flattened. The association was stronger among premenopausal women, suggesting that younger women or women with higher endogenous levels of estrogens may have greater susceptibility to the effects of LAN. Effect measure modification by menopausal status could be due to different underlying biological mechanisms. Previous studies have reported that the suppressive effect of LAN on melatonin secretion may be stronger among younger people, tending to decrease with age [ 62 , 65 ]. In addition, LAN may impact on the length of the menstrual cycle through endocrine-disrupting properties, thus leading to higher breast cancer risk in the premenopausal period [ 66 ].

With regard to confounding factors, solar UV-B radiation is thought to be protective for breast cancer development. The inverse association between cancer risk and UV-B radiation was hypothesized for the first time by the Garland brothers in 1980 [ 67 ] who theorized sunlight-induced increases in vitamin D 3 may confer protection. The final product of the vitamin D 3 metabolism is the calcitriol, which has many anti-carcinogenic properties including inhibiting cellular proliferation [ 68 ]. Across the years, many studies have investigated the potential protective role of the UV-B radiation against different types of cancer [ 69 , 70 , 71 , 72 ]. A remaining question is the relationship between LAN and ultraviolet radiation. We found a positive association between LAN and breast cancer risk in countries exposed to higher levels of UV-B radiation (> 0.58 W/m 2 ). Conversely, there was no association between LAN and breast cancer risk in countries with low UV-B irradiance (< 0.58 W/m 2 ).

We considered another confounding factor that also correlates with UV-B radiation and could influence the outcome risk: the influence of annual sunshine hours [ 73 , 74 ]. We found an increased breast cancer risk associated with LAN in countries where annual sunshine exposure exceeded 3000 h. A positive association was also found in countries where sunshine ranged 2000–3000 h/years, while no appreciable association was observed in countries where there were fewer sunshine hours. This could be explained by different habits of people living in different countries, which may reflect epigenetics adaptation [ 75 ]. As indicated by a 2014 Italian study, humans’ biological clocks may have adapted to different environmental conditions during migrations, consistent with studies on insects [ 76 ], birds [ 77 ], and fish [ 78 ] living at different latitudes. These studies analyzed, in particular, the evolution of circadian genes which may be related to selective pressure exerted from latitude, temperature, ultraviolet radiation flux [ 79 ]. Consequently, people living in countries exposed to less than 2000 annual sunshine hours are less susceptible to higher frequency of light during night compared to people living in countries exposed to more annual sunshine hours. An explanation may be the presence of different alleles of their circadian genes, which acted to adapt the organism to different living conditions such as different latitudes. [ 80 ]. Finally, we may hypothesize that if people residing in “darkest countries” have artificial light kept on also during the daytime, they may not be as strongly influenced by higher levels of LAN because of different environment-adaptive alleles which acted to adapt the organism to light regimes diverse from the natural ones, as has been shown in animals [ 81 ]. In the meantime, another hypothesis that may explain the stronger association in relation to the annual sunshine hours is represented by the cumulative effect of longer daily sunshine hours with LAN exposure. In fact, those living in “brightest countries” are exposed to higher levels of light, which may lead to greater melatonin suppression levels and chronodisruption. Hence, the combination of daily sunshine hours and LAN may increase breast cancer risk.

Our results are relatively consistent with two previous meta-analyses [ 21 , 22 ], with the exception of the subgroup analysis according to menopausal status, where our results were similar to those of Lai et al. [ 21 ] but conflicted with those of Wu et al. [ 22 ]. However, to our knowledge this meta-analysis is the first to have assessed the dose–response between LAN and breast cancer risk, particularly among premenopausal women. In addition, owing to three new, recently-published studies we could include in the present review, we could re-assess the LAN-breast cancer relation according to ER cancer type, also performing a dose–response meta-analysis in these subgroups. Though our findings support a harmful effect of LAN in both the ER + and ER − breast cancer subtypes, at high exposure levels i.e., above 30 nW/cm 2 /sr the curve flattened in women with ER + disease but still increased in the ER − subgroup.

Our review has some strengths and limitations. Firstly, we used a newly developed meta-analytic tool for exploring the full shape of the dose–response, enabling us to assess the shape of the relation between LAN exposure and breast cancer risk over a wide range of exposure and across population subgroups. Our approach also yielded some indications of the threshold exposure levels that can increase breast cancer risk. Moreover, we systematically used the most adjusted model from each included study, thereby accounting for major confounders of the association.

Nonetheless, we acknowledge that some summary estimates are still statistically unstable due to the low number of studies still characterizing some subgroups. In addition, we could not rule out that unmeasured confounding was still likely in the investigated studies, and therefore influenced the findings. An example of such potential identified confounder could be the possibility that air pollution is heavier in urban areas, where the highest levels of LAN exposure are also detectable. In particular, LAN exposure may correlate with higher levels of traffic-related pollutants, including noise, as indicated by its inverse correlation with greenness and green space diversity [ 82 , 83 ]. Unfortunately, only one study included traffic noise in the multivariable model, thus hampering the evaluation of any independent effects of this factor and its potential for confounding in LAN-related studies. Additionally, other confounders may be those related to the occupational night environment, especially for studies assessing LAN exposure among nightshift workers [ 55 ]. Another limitation could be the limited capacity of outdoor LAN to adequately reflect personal light exposure due to differences in indoor lightning, use of electronic devices, nighttime activities, or window treatments, being these only some of the potential other sources of exposure [ 83 , 84 ]. Therefore, future studies should ideally use validated questionnaires combined with satellite data to more accurately measure individual LAN exposure. Finally, funnel plots and trim-and-fill analysis suggest a negligible probability of small-study effects in overall and stratified analyses.

Conclusions

Our review suggests a positive association between LAN exposure and risk of breast cancer, particularly in some subgroups, especially in premenopausal women, while few differences substantially emerged according to ER status, thus ongoing efforts to minimize LAN exposure might contribute to decrease human burden of diseases [ 85 , 86 , 87 ].

Availability of data and materials

All data generated or analyzed during this study are included in this published article and its additional information files.

Abbreviations

Artificial Light At Night

Body Mass Index

Confidence Interval

Estrogen Receptor

US Defense Meteorological Satellite Program Operational Linescan System

Visible Infrared Imaging Radiometer Suite Day-Night Band

Heterogeneity

Light At Night

Ultraviolet B

Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, Bray F. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71(3):209–49.

Article   PubMed   Google Scholar  

AIRTUM: The numbers of cancer in Italy-2020. In . https://www.registri-tumori.it/cms/sites/default/files/pubblicazioni/new_NDC2020-operatori-web.pdf ; 2020.

Sun YS, Zhao Z, Yang ZN, Xu F, Lu HJ, Zhu ZY, Shi W, Jiang J, Yao PP, Zhu HP. Risk factors and preventions of breast cancer. Int J Biol Sci. 2017;13(11):1387–97.

Article   CAS   PubMed   PubMed Central   Google Scholar  

Lei S, Fan P, Wang M, Zhang C, Jiang Y, Huang S, Fang M, He Z, Wu A. Elevated estrogen receptor beta expression in triple negative breast cancer cells is associated with sensitivity to doxorubicin by inhibiting the PI3K/AKT/mTOR signaling pathway. Exp Ther Med. 2020;20(2):1630–6.

White AJ, Keller JP, Zhao S, Carroll R, Kaufman JD, Sandler DP. Air pollution, clustering of particulate matter components, and breast cancer in the Sister Study: a U.S. wide cohort. Environ Health Perspect. 2019;127(10):107002.

Article   PubMed   PubMed Central   Google Scholar  

Goldberg MS, Villeneuve PJ, Crouse D, To T, Weichenthal SA, Wall C, Miller AB. Associations between incident breast cancer and ambient concentrations of nitrogen dioxide from a national land use regression model in the Canadian National Breast Screening Study. Environ Int. 2019;133(Pt B):105182.

Article   CAS   PubMed   Google Scholar  

Filippini T, Torres D, Lopes C, Carvalho C, Moreira P, Naska A, Kasdagli MI, Malavolti M, Orsini N, Vinceti M. Cadmium exposure and risk of breast cancer: a dose-response meta-analysis of cohort studies. Environ Int. 2020;142:105879.

Gaudet HM, Christensen E, Conn B, Morrow S, Cressey L, Benoit J. Methylmercury promotes breast cancer cell proliferation. Toxicol Rep. 2018;5:579–84.

Adani G, Filippini T, Wise LA, Halldorsson TI, Blaha L, Vinceti M. Dietary intake of acrylamide and risk of breast, endometrial, and ovarian cancers: a systematic review and dose-response meta-analysis. Cancer Epidemiol Biomarkers Prev. 2020;29(6):1095–106.

IARC working group on the evaluation of carcinogenic risks to humans. Painting, firefighting, and shiftwork. IARC Monogr Eval Carcinog Risks Hum. 2010;98:1–818.

Straif K, Baan R, Grosse Y, Secretan B, El Ghissassi F, Bouvard V, Altieri A, Benbrahim-Tallaa L, Cogliano V, WHO International Agency For Research on Cancer Monograph Working Group. Carcinogenicity of shift-work, painting, and fire-fighting. Lancet Oncol. 2007;8(12):1065–6.

Papantoniou K, Castano-Vinyals G, Espinosa A, Aragones N, Perez-Gomez B, Ardanaz E, Altzibar JM, Sanchez VM, Gomez-Acebo I, Llorca J, et al. Breast cancer risk and night shift work in a case-control study in a Spanish population. Eur J Epidemiol. 2016;31(9):867–78.

IARC working group on the evaluation of carcinogenic hazards to humans. Night shift work. IARC Monogr Eval Carcinog Hazards Hum. 2020;124:1–381.

Rumanova VS, Okuliarova M, Zeman M. Differential effects of constant light and dim light at night on the circadian control of metabolism and behavior. Int J Mol Sci. 2020;21(15):5478.

Article   CAS   PubMed Central   Google Scholar  

Russart KLG, Nelson RJ. Light at night as an environmental endocrine disruptor. Physiol Behav. 2018;190:82–9.

Irwin MR, Olmstead R, Carroll JE. Sleep disturbance, sleep duration, and inflammation: a systematic review and meta-analysis of cohort studies and experimental sleep deprivation. Biol Psychiatry. 2016;80(1):40–52.

Castanon-Cervantes O, Wu M, Ehlen JC, Paul K, Gamble KL, Johnson RL, Besing RC, Menaker M, Gewirtz AT, Davidson AJ. Dysregulation of inflammatory responses by chronic circadian disruption. J Immunol. 2010;185(10):5796–805.

Samuelsson LB, Bovbjerg DH, Roecklein KA, Hall MH. Sleep and circadian disruption and incident breast cancer risk: an evidence-based and theoretical review. Neurosci Biobehav Rev. 2018;84:35–48.

McEwen BS, Karatsoreos IN. Sleep deprivation and circadian disruption: stress, allostasis, and allostatic load. Sleep Med Clin. 2015;10(1):1–10.

Cho Y, Ryu SH, Lee BR, Kim KH, Lee E, Choi J. Effects of artificial light at night on human health: a literature review of observational and experimental studies applied to exposure assessment. Chronobiol Int. 2015;32(9):1294–310.

Lai KY, Sarkar C, Ni MY, Cheung LWT, Gallacher J, Webster C. Exposure to light at night (LAN) and risk of breast cancer: a systematic review and meta-analysis. Sci Total Environ. 2021;762:143159.

Wu Y, Gui SY, Fang Y, Zhang M, Hu CY. Exposure to outdoor light at night and risk of breast cancer: a systematic review and meta-analysis of observational studies. Environ Pollut. 2021;269:116114.

Xiao Q, James P, Breheny P, Jia P, Park Y, Zhang D, Fisher JA, Ward MH, Jones RR. Outdoor light at night and postmenopausal breast cancer risk in the NIH-AARP diet and health study. Int J Cancer. 2020;147(9):2363–72.

Clarke RB, Amini H, James P, von Euler-Chelpin M, Jorgensen JT, Mehta A, Cole-Hunter T, Westendorp R, Mortensen LH, Loft S, et al. Outdoor light at night and breast cancer incidence in the Danish Nurse Cohort. Environ Res. 2021;194:110631.

Xiao Q, Gierach GL, Bauer C, Blot WJ, James P, Jones RR. The association between outdoor artificial light at night and breast cancer risk in black and white women in the Southern Community Cohort Study. Environ Health Perspect. 2021;129(8):87701.

Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, Shamseer L, Tetzlaff JM, Akl EA, Brennan SE, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. PLoS Med. 2021;18(3):e1003583.

Morgan RL, Whaley P, Thayer KA, Schunemann HJ. Identifying the PECO: A framework for formulating good questions to explore the association of environmental and other exposures with health outcomes. Environ Int. 2018;121(Pt 1):1027–31.

European network for Health Technology Assessment (EUnetHTA): Process of information retrieval for systematic reviews and health technology assessments on clinical effectiveness. Version 2.0, December 2019.

Morgan RL, Thayer KA, Santesso N, Holloway AC, Blain R, Eftim SE, Goldstone AE, Ross P, Ansari M, Akl EA, et al. A risk of bias instrument for non-randomized studies of exposures: a users’ guide to its application in the context of GRADE. Environ Int. 2019;122:168–84.

Orsini N, Spiegelman D: Meta-analysis of dose-response relationships. In: Handbook of Meta-Analysis. Edited by Schmid CH, Stijnen T, White I, 1st Edition. New York: Chapman and Hall/CRC; 2021.

Filippini T, Hatch EE, Rothman KJ, Heck JE, Park AS, Crippa A, Orsini N, Vinceti M. Association between outdoor air pollution and childhood leukemia: a systematic review and dose-response meta-analysis. Environ Health Perspect. 2019;127(4):46002.

Filippini T, Malavolti M, Whelton PK, Naska A, Orsini N, Vinceti M. Blood pressure effects of sodium reduction: dose-response meta-analysis of experimental studies. Circulation. 2021;143(16):1542–67.

Vinceti M, Filippini T, Wise LA, Rothman KJ. A systematic review and dose-response meta-analysis of exposure to environmental selenium and the risk of type 2 diabetes in nonexperimental studies. Environ Res. 2021;197:111210.

Crippa A, Discacciati A, Bottai M, Spiegelman D, Orsini N. One-stage dose-response meta-analysis for aggregated data. Stat Methods Med Res. 2019;28(5):1579–96.

De Jong S, Auping WL, Oosterveld WT, Usanov A, Abdalla M, Van de Bovenkamp A, Frattina C: The geopolitical impact of climate mitigation policies: How hydrocarbon exporting rentier states and developing nations can prepare for a more sustainable future. The Hague Centre for Strategic Studies 2017.

Mohr S, Garland C, Gorham E, Grant W, Highfill R, Garland F: Mapping Vitamin D deficiency, breast cancer, and colorectal cancer. 2005.

Egger M, Smith GD, Phillips AN. Meta-analysis: principles and procedures. BMJ. 1997;315(7121):1533–7.

Lin L, Shi L, Chu H, Murad MH. The magnitude of small-study effects in the Cochrane Database of Systematic Reviews: an empirical study of nearly 30 000 meta-analyses. BMJ Evid Based Med. 2020;25(1):27–32.

Duval S, Tweedie RL. A non-parametric “trim and fill” method of accounting for publication bias in meta-analysis. J Am Stat Assoc. 2000;95(449):89–98.

Google Scholar  

Bauer SE, Wagner SE, Burch J, Bayakly R, Vena JE. A case-referent study: light at night and breast cancer risk in Georgia. Int J Health Geogr. 2013;12:23.

Garcia-Saenz A, Sanchez de Miguel A, Espinosa A, Valentin A, Aragones N, Llorca J, Amiano P, Martin-Sanchez V, Guevara M, Capelo R et al : Evaluating the association between artificial light-at-night exposure and breast and prostate cancer risk in Spain (MCC-Spain Study). Environ Health Perspect 2018, 126(4):047011.

Johns LE, Jones ME, Schoemaker MJ, McFadden E, Ashworth A, Swerdlow AJ. Domestic light at night and breast cancer risk: a prospective analysis of 105 000 UK women in the Generations Study. Br J Cancer. 2018;118(4):600–6.

Hurley S, Goldberg D, Nelson D, Hertz A, Horn-Ross PL, Bernstein L, Reynolds P. Light at night and breast cancer risk among California teachers. Epidemiology. 2014;25(5):697–706.

James P, Bertrand KA, Hart JE, Schernhammer ES, Tamimi RM, Laden F. Outdoor light at night and breast cancer incidence in the Nurses’ Health Study II. Environ Health Perspect. 2017;125(8):087010.

White AJ, Weinberg CR, Park YM, D’Aloisio AA, Vogtmann E, Nichols HB, Sandler DP. Sleep characteristics, light at night and breast cancer risk in a prospective cohort. Int J Cancer. 2017;141(11):2204–14.

Davis S, Mirick DK, Stevens RG. Night shift work, light at night, and risk of breast cancer. J Natl Cancer Inst. 2001;93(20):1557–62.

Fritschi L, Erren TC, Glass DC, Girschik J, Thomson AK, Saunders C, Boyle T, El-Zaemey S, Rogers P, Peters S, et al. The association between different night shiftwork factors and breast cancer: a case-control study. Br J Cancer. 2013;109(9):2472–80.

Keshet-Sitton A, Or-Chen K, Yitzhak S, Tzabary I, Haim A. Can avoiding light at night reduce the risk of breast cancer? Integr Cancer Ther. 2016;15(2):145–52.

Kloog I, Portnov BA, Rennert HS, Haim A. Does the modern urbanized sleeping habitat pose a breast cancer risk? Chronobiol Int. 2011;28(1):76–80.

Li Q, Zheng T, Holford TR, Boyle P, Zhang Y, Dai M. Light at night and breast cancer risk: results from a population-based case-control study in Connecticut, USA. Cancer Causes Control. 2010;21(12):2281–5.

O’Leary ES, Schoenfeld ER, Stevens RG, Kabat GC, Henderson K, Grimson R, Gammon MD, Leske MC, Electromagnetic Fields and Breast Cancer on Long Island Study Group. Shift work, light at night, and breast cancer on Long Island, New York. Am J Epidemiol. 2006;164(4):358–66.

Ritonja J, McIsaac MA, Sanders E, Kyba CCM, Grundy A, Cordina-Duverger E, Spinelli JJ, Aronson KJ. Outdoor light at night at residences and breast cancer risk in Canada. Eur J Epidemiol. 2020;35(6):579–89.

Yang W, Shi Y, Ke X, Sun H, Guo J, Wang X. Long-term sleep habits and the risk of breast cancer among Chinese women: a case-control study. Eur J Cancer Prev. 2019;28(4):323–9.

Falchi F, Cinzano P, Duriscoe D, Kyba CC, Elvidge CD, Baugh K, Portnov BA, Rybnikova NA, Furgoni R. The new world atlas of artificial night sky brightness. Sci Adv. 2016;2(6):e1600377.

Article   PubMed   PubMed Central   CAS   Google Scholar  

US National Toxicology Program: NTP cancer hazard assessment report on night shift work and light at night. In: NTP Cancer Hazard Assessment Report on Night Shift Work and Light at Night. Research Triangle Park (NC); 2021.

Hansen J, Lassen CF. Nested case-control study of night shift work and breast cancer risk among women in the Danish military. Occup Environ Med. 2012;69(8):551–6.

Hansen J, Stevens RG. Case-control study of shift-work and breast cancer risk in Danish nurses: impact of shift systems. Eur J Cancer. 2012;48(11):1722–9.

Lie JA, Kjuus H, Zienolddiny S, Haugen A, Stevens RG, Kjaerheim K. Night work and breast cancer risk among Norwegian nurses: assessment by different exposure metrics. Am J Epidemiol. 2011;173(11):1272–9.

Megdal SP, Kroenke CH, Laden F, Pukkala E, Schernhammer ES. Night work and breast cancer risk: a systematic review and meta-analysis. Eur J Cancer. 2005;41(13):2023–32.

Pukkala E, Helminen M, Haldorsen T, Hammar N, Kojo K, Linnersjo A, Rafnsson V, Tulinius H, Tveten U, Auvinen A. Cancer incidence among Nordic airline cabin crew. Int J Cancer. 2012;131(12):2886–97.

Costa G, Haus E, Stevens R. Shift work and cancer - considerations on rationale, mechanisms, and epidemiology. Scand J Work Environ Health. 2010;36(2):163–79.

Haus EL, Smolensky MH. Shift work and cancer risk: potential mechanistic roles of circadian disruption, light at night, and sleep deprivation. Sleep Med Rev. 2013;17(4):273–84.

Touitou Y, Reinberg A, Touitou D. Association between light at night, melatonin secretion, sleep deprivation, and the internal clock: health impacts and mechanisms of circadian disruption. Life Sci. 2017;173:94–106.

Erren TC, Reiter RJ, Piekarski C. Light, timing of biological rhythms, and chronodisruption in man. Naturwissenschaften. 2003;90(11):485–94.

Herljevic M, Middleton B, Thapan K, Skene DJ. Light-induced melatonin suppression: age-related reduction in response to short wavelength light. Exp Gerontol. 2005;40(3):237–42.

Stevens RG, Rea MS. Light in the built environment: potential role of circadian disruption in endocrine disruption and breast cancer. Cancer Causes Control. 2001;12(3):279–87.

Garland CF, Garland FC. Do sunlight and vitamin D reduce the likelihood of colon cancer? Int J Epidemiol. 1980;9(3):227–31.

Moukayed M, Grant WB. The roles of UVB and vitamin D in reducing risk of cancer incidence and mortality: a review of the epidemiology, clinical trials, and mechanisms. Rev Endocr Metab Disord. 2017;18(2):167–82.

Altieri B, Grant WB, Della Casa S, Orio F, Pontecorvi A, Colao A, Sarno G, Muscogiuri G. Vitamin D and pancreas: the role of sunshine vitamin in the pathogenesis of diabetes mellitus and pancreatic cancer. Crit Rev Food Sci Nutr. 2017;57(16):3472–88.

Chen W, Clements M, Rahman B, Zhang S, Qiao Y, Armstrong BK. Relationship between cancer mortality/incidence and ambient ultraviolet B irradiance in China. Cancer Causes Control. 2010;21(10):1701–9.

Ong JS, Cuellar-Partida G, Lu Y, Australian Ovarian Cancer S, Fasching PA, Hein A, Burghaus S, Beckmann MW, Lambrechts D, van Nieuwenhuysen E et al : Association of vitamin D levels and risk of ovarian cancer: a Mendelian randomization study. Int J Epidemiol 2016, 45(5):1619–1630.

Grant WB. Low ultraviolet-B exposure may explain some of the link between night shift work and increased risk of prostate cancer. Int J Cancer. 2015;137(4):999.

IARC working group on the evaluation of carcinogenic risks to humans. Solar and ultraviolet radiation. IARC Monogr Eval Carcinog Risks Hum. 2012;100D:35–101.

Holick MF. Biological effects of sunlight, ultraviolet radiation, visible light, infrared radiation and vitamin D for health. Anticancer Res. 2016;36(3):1345–56.

CAS   PubMed   Google Scholar  

Hancock AM, Witonsky DB, Alkorta-Aranburu G, Beall CM, Gebremedhin A, Sukernik R, Utermann G, Pritchard JK, Coop G, Di Rienzo A. Adaptations to climate-mediated selective pressures in humans. PLoS Genet. 2011;7(4):e1001375.

Kyriacou CP, Peixoto AA, Sandrelli F, Costa R, Tauber E. Clines in clock genes: fine-tuning circadian rhythms to the environment. Trends Genet. 2008;24(3):124–32.

Johnsen A, Fidler AE, Kuhn S, Carter KL, Hoffmann A, Barr IR, Biard C, Charmantier A, Eens M, Korsten P, et al. Avian Clock gene polymorphism: evidence for a latitudinal cline in allele frequencies. Mol Ecol. 2007;16(22):4867–80.

O’Malley KG, Banks MA. A latitudinal cline in the Chinook salmon ( Oncorhynchus tshawytscha ) Clock gene: evidence for selection on PolyQ length variants. Proc Biol Sci. 2008;275(1653):2813–21.

CAS   PubMed   PubMed Central   Google Scholar  

Forni D, Pozzoli U, Cagliani R, Tresoldi C, Menozzi G, Riva S, Guerini FR, Comi GP, Bolognesi E, Bresolin N, et al. Genetic adaptation of the human circadian clock to day-length latitudinal variations and relevance for affective disorders. Genome Biol. 2014;15(10):499.

Putilov AA, Dorokhov VB, Poluektov MG. How have our clocks evolved? Adaptive and demographic history of the out-of-African dispersal told by polymorphic loci in circadian genes. Chronobiol Int. 2018;35(4):511–32.

Hancock AM, Alkorta-Aranburu G, Witonsky DB, Di Rienzo A. Adaptations to new environments in humans: the role of subtle allele frequency shifts. Philos Trans R Soc Lond B Biol Sci. 2010;365(1552):2459–68.

Stanhope J, Liddicoat C, Weinstein P. Outdoor artificial light at night: a forgotten factor in green space and health research. Environ Res. 2021;1:111012.

Article   CAS   Google Scholar  

Huss A, van Wel L, Bogaards L, Vrijkotte T, Wolf L, Hoek G, Vermeulen R. Shedding some light in the dark-a comparison of personal measurements with satellite-based estimates of exposure to light at night among children in the Netherlands. Environ Health Perspect. 2019;127(6):67001.

Jones RR. Exposure to artificial light at night and risk of cancer: where do we go from here? Br J Cancer. 2021;124(9):1467–8.

Helbich M, Browning M, Huss A. Outdoor light at night, air pollution and depressive symptoms: a cross-sectional study in the Netherlands. Sci Total Environ. 2020;744:140914.

Munzel T, Hahad O, Daiber A. The dark side of nocturnal light pollution. Outdoor light at night increases risk of coronary heart disease. Eur Heart J. 2021;42(8):831–4.

Walker WH 2nd, Bumgarner JR, Walton JC, Liu JA, Melendez-Fernandez OH, Nelson RJ, DeVries AC. Light pollution and cancer. Int J Mol Sci. 2020;21(24):9360

Download references

Acknowledgements

Not applicable.

This work was supported by a grant “Dipartimenti di Eccellenza 2018 − 2022, MIUR, Italy” to the Department of Biomedical, Metabolic and Neural Sciences (Drs. Filippini, Urbano and Vinceti) and by a grant FAR IMPULSO2020 no. 494/2020 (Dr. Filippini).

Author information

Authors and affiliations.

CREAGEN - Environmental, Genetic and Nutritional Epidemiology Research Center, Department of Biomedical, Metabolic and Neural Sciences, University of Modena and Reggio Emilia, Modena, Italy

Teresa Urbano, Marco Vinceti & Tommaso Filippini

Department of Epidemiology, Boston University School of Public Health, Boston, MA, USA

Marco Vinceti & Lauren A. Wise

You can also search for this author in PubMed   Google Scholar

Contributions

TU and TF designed the original study, and with MV and LAW analysed and interpreted the data and drafted the article. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Marco Vinceti .

Ethics declarations

Ethics approval and consent to participate, consent for publication, competing interests.

The authors declare that they have no competing interest.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supple mentary Information

Additional file 1..

Additional figures and tables.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Urbano, T., Vinceti, M., Wise, L.A. et al. Light at night and risk of breast cancer: a systematic review and dose–response meta-analysis. Int J Health Geogr 20 , 44 (2021). https://doi.org/10.1186/s12942-021-00297-7

Download citation

Received : 18 June 2021

Accepted : 01 October 2021

Published : 16 October 2021

DOI : https://doi.org/10.1186/s12942-021-00297-7

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Breast cancer
• Light-at-night
• Menopausal status
• Dose–response meta-analysis

International Journal of Health Geographics

ISSN: 1476-072X

• Submission enquiries: [email protected]

ORIGINAL RESEARCH article

Determinants of breast cancer among women attending oncology units in selected health facilities of hawassa city, sidama region, southern ethiopia, 2023: case-control study.

• 1 Public Health Department, Yanet-Liyana College of Health Sciences, Hawassa, Ethiopia
• 2 School of Public Health, College of Medicine and Health Sciences, Hawassa University, Hawassa, Ethiopia

Background: The incidence of breast cancer (BC) is rampantly increasing in developing countries particularly Ethiopia. Unfortunately, the morbidity and mortality rates are sharply increasing, and because of this, families are suffering from socioeconomic crises. Despite this, there is limited evidence on the determinants of breast cancer in Ethiopia as well as in the study area.

Objective: To identify the determinants of BC among women attending oncology units in selected hospitals in Hawassa City, Ethiopia, in 2023.

Method: A hospital-based, case-control study with 300 patients (75 cases and 225 controls) was carried out in Hawassa from June to July 2023. A simple random sampling technique was used to select cases and controls. Data were collected via pretested and structured digitally installed questionnaires with Kobo collection/smartphones. The data were exported from the server to SPSS version 27 for analysis. Descriptive analysis of univariate, bivariate, and multivariable logistic regression data was conducted to determine the associations between breast cancer incidence and independent factors.

Results: A total of 300 women participated in this study for a response rate of 100%. The mean ( ± SD) ages of the respondents were 37.2 ( ± 14.8) and 36.6 ( ± 15.1) years for the cases and controls, respectively. According to the multivariate logistic regression model, postmenopausal status [AOR: 2.49; 95% CI (1.18, 5.23)], family history of cancer [AOR: 2.33; 95% CI (1.12, 4.82)], oral contraceptives [AOR: 2.74; 95% CI (1.34, 5.99)], overweight and/or obesity [AOR: 2.29; 95% CI: (1.14, 4.59)], and consumption of solid oil [AOR: 2.36; 95% CI (1.20, 4.67)] were independently associated with BC risk.

Conclusion: This study revealed important risk factors for BC. Therefore, women should adopt healthier lifestyles through healthy nutrition and regular exercise to reduce the risk of developing BC. In addition, early detection and regular screening are proactive approaches for detecting BC.

1 Introduction

Breast cancer (BC) is the most prevalent cancer in women globally accounting for 2.3 million new cases and one in six cancer deaths in 2020 ( 1 ). Globalization has led to a greater incidence of BC in high-income countries (571/100,000) than in low-income countries (95/10,000) ( 2 ). Due to globalization, risk factors for cancer are appearing with our modern diet, which emphasizes super grain and more processed food; the use of addictive substances, toxic and pharmaceutical products; and waste exposure. However, the prevention of cancer still depends on the recognition and elimination of risks from carcinogens ( 3 ). The five most prevalent nations were all in Europe. The number of new cases in Africa is estimated to have been 92,600 in 2008 and 133,900 in 2012 ( 1 ). In 2020, 2.3 million women were diagnosed with BC, and 685,000 women died from BC worldwide ( 4 , 5 ).

In Africa, 8% of all BC cases are diagnosed, but the mortality rate is far higher than expected ( 1 ). Moreover, sub-Saharan Africa has the highest BC incidence and mortality rates attributed to Westernized lifestyles, food changes, and reduced physical activity among the African population ( 6 ). In 2020, there was significant geographical variation in the major cancers in sub-Saharan Africa ( 7 ).

In Ethiopia, BC is the most prevalent cancer accounting for 17.7% of cancer deaths and 22.6% of all cancer cases annually ( 8 , 9 ). In addition, every year, approximately 60,000 new cases of BC are diagnosed. Similarly, evidence from the Addis Ababa cancer registry report shows that BC is responsible for 23% of all cancer cases and 33% of all cancer cases in women ( 10 ). Moreover, it is the most common malignancy in Ethiopia with increased rates of mortality and morbidity. For instance, of the 4,139 new cancer patients diagnosed between 2012 and 2013, 67% were female, and 31.5% had BC. The age-standardized incidence rate of BC was 43.3 per 100,000 ( 11 ) females, and a yearly average of 216 incident cases of BC were reported between 1997 and 2012 ( 11 ).

Multiple factors, such as alcohol consumption, tobacco use, obesity, inactivity, diet, family history, early menarche, late menopause, late age, null parity, and nonbreastfeeding practices, are known to increase the risk of BC ( 8 ). In contrast, lifestyle modifications, targeted prevention programs, and population-based screening can significantly reduce BC incidence ( 12 ).

Unfortunately, even if all of the potentially modifiable risk factors could be mitigated, this would only reduce the risk of developing BC by at most 30% ( 13 ). In addition, 20%–30% o BCs can be ascribed to controllable characteristics, and 5%–10% of BCs can be related to factors such as genetic mutations and family history ( 14 ). Thus, the healthcare system should be strengthened, and the gradual adoption of universal health coverage should be encouraged especially for low- and middle-income nations where cancer and noncommunicable disease programs are frequently inaccessible and limited-resource citizens are critical ( 10 ).

The World Health Organization launched the Global BC Initiative to address the increasing burden of BC, which is estimated to kill 685,000 women globally in 2020 ( 1 ). Similarly, the Federal Ministry of Health of Ethiopia (FMOHE) developed and amended a strategy to reduce risk factors and encourage a healthy lifestyle to prevent and control noncommunicable diseases (NCDs), including BC ( 15 ). On the other hand, the population of Ethiopia is diverse, representing a wide range of lifestyles, cultures, socioeconomic statuses, and breastfeeding practices. These elements may have an impact on overall health, which may include BC risk ( 15 ).

However, to the best of our knowledge, most of the existing studies are cross-sectional, and little is known about BC determination in Ethiopia or in the study area. Therefore, this study aimed to identify determinants of BC among women attending oncology units in the Sidama Region in selected hospitals.

2 Materials and methods

2.1 study design.

A hospital-based unmatched case-control study design was employed at Hawassa City from 1 June– to 30 July 2023.

2.2 Study area and period

The study was conducted in the public and private health facilities of Hawassa City. The Hawassa City Administration is located in Sidama Region, Ethiopia, approximately 275 km south of Addis Ababa. The city had a total of 385,257 populations with male dominance in number. Of the total population, 89,765 were females of reproductive age.

Cities are administrative structures with dense grid populations of more than 1,500 people per km 2 . Within a city, there are small units of administrative structure called subcities. Therefore, in Hawassa City, there are eight small administrative units composed of the smallest administrative units called kebeles. Based on the above explanation, Hawassa City had eight subcities (small administrative units), and each subcity had a different number of smallest administrative units called kebeles accounting for 32 kebeles. Of these smallest administrative units, 11 were rural and 21 were urban kebeles. Hawassa City also has 8 hospitals (3 public and 5 private), 12 public health centers, and 18 health posts. The three public hospitals are Hawassa University Comprehensive Specialized Hospital (HUCSH), Adare General Hospital (AGH), and Hawela Tula Primary Hospital (HTPH) ( 11 ). According to the 2022 Sidama Regional State Health Bureau report, a flow of approximately 177 patients per month is expected. The study was performed from 1 June to 30 July 2023.

2.3 Population

2.3.1 source population and study population.

All women who were more than 15 years old and who presented at Hawassa City health facilities were the source population. All women with confirmed BC visiting the selected health facilities for cases and all women without BC visiting the selected health facilities for other services during the study period composed the study population.

2.4 Inclusion and exclusion criteria

The patients were all women over the age of 15 years who had BC that had been confirmed and were receiving chemotherapy. The controls were all women over the age of 15 years who visited selected hospitals for other unrelated to cancer disease in the period of the study. However, women who were mentally incompetent or seriously ill during the course of data collection were excluded from both the case and control groups.

2.5 Sample size determination

Using the statistical software Epi-info version 7, the sample size for unmatched case-control studies was calculated. The following assumptions were used: 95% confidence level, 80% power, a case-to-control ratio of 1:3, a percentage of exposure among control-exposed women (i.e., the percentage of overweight and obese women without BC), and a percentage of exposure among cases (i.e., the percentage of obese women with BC).

The proportion of obese individuals among the controls was 25%, and the proportion of obese individuals among the patients was 31.9%, based on research performed in Addis Ababa, Ethiopia. Based on the above assumptions, the sample size was 272 (66 cases and 204 controls), and after adding a 10% nonresponse rate, the final sample size for the study was 300 (75 cases and 225 controls) ( Table 1 ).

Table 1 Sociodemographic characteristics of the study participants in selected health facilities in Hawassa City, 2023.

2.6 Sampling technique and procedures

The study was performed in two hospitals, namely, Hawassa University Comprehensive (HUCSH) and Specialized Hospital and Yanet Internal Medicine Specialized Center (YIMSC), which provide chemotherapy services. The final calculated sample size was proportionally allocated to each hospital based on the number of patients who received chemotherapy services for BC treatment within the last year’s data from the health management information system (HMIS) report. Hence, 46 cases and 138 controls from HUCSH and 29 cases and 87 controls from YIMSC were included in the study. Since the cases are rare, we included all cases sequentially, but controls were included using a systematic random sampling technique. Therefore, once we included one patient, the next three controls were included in the study until the sample size was met.

2.7 Study variables

2.7.1 dependent variable: breast cancer.

Dependent variable is breast cancer.

2.7.2 Independent variables

Sociodemographic variables such as age, educational level, occupation, place of residence, and family history of cancer; behavior and/or alcohol intake, smoking status, fatty diet intake, and body mass index; reproductive variables such as age at first delivery, age at menarche, parity, breastfeeding practice, and use of hormonal contraceptives; diet and lifestyle variables such as vegetable intake, exercise, and fruit intake; and environmental and health-related variables such as pesticide contact, radiation exposure, breast trauma, and history of breast infection.

2.8 Data collection instruments and procedures

2.8.1 data collection instruments.

Primarily, the tool was developed in English after reviewing different literature from previous studies ( 16 – 18 ). Then, the questionnaire was translated into the local languages Sidamagna Afoo and Amharic by an expert and translated back to English by another expert to maintain consistency.

2.8.2 Data collection procedures

One day of training was given for two data collectors and one supervisor on the instrument. The interviewer administered a questionnaire supplemented with a checklist to collect data using the Kobo toolbox. Hence, during the interview, the data collector interviewed study participants about their sociodemographic, behavioral, clinical, and reproductive matters for both cases and controls. In addition, other pertinent determinant factors were identified from patients’ medical records. Pretests were performed on 5% of the sample before the actual data collection, and the necessary amendments were made.

The patient’s interview and the review of the patient’s records were used to gather information about the patient’s sociodemographic (age, residence, marital status, and level of education) and behavioral (smoking, BMI, alcohol consumption, and physical activity) BMI results. Patients were classified as underweight (<18.5), normal weight (18.5–24.9), overweight (25–29.9), or obese (>30). Reproductive (parity, OC use, menopausal status, breastfeeding practices) characteristics. Before the real data collection process, a pretest was conducted using the collected data.

2.9 Data quality assurance

To ensure data quality, first, the data collection tools were prepared in English, translated to Amharic, and then returned to English to ensure consistency. Appropriate training was given to the data collectors and supervisors. The training included a briefing on the general objectives of the study, approach to accessing study participants, clarity on each item in the instrument, data collection procedure, including or excluding the target data source, timeliness of data submission, data handling, and time management. Pretests were performed outside the study health institution on 5% of the sample before the actual data collection, and the necessary correction was made based on the pretest results to avoid any confusion and for better completion of the questions. Every day, the collected data were reviewed and cross-checked for completeness.

2.10 Data processing and analysis

The data were collected via Kobo data collection and exported to SPSS version 27 for analysis. Descriptive statistics, such as frequency and percent distribution, were used to present categorical variables. Means and standard deviations were used for continuous variables. We conducted an independent sample t test to evaluate the equality of variance and mean difference among cases and controls to exposure variables to evaluate the mean difference in exposure between cases and controls. To identify factors related to BC, a binary logistic regression analysis was performed. According to the bivariate analysis, variables with a p-value of 0.25 were candidates for multivariate analysis. To demonstrate the strength of the link, an adjusted odds ratio with a 95% confidence interval was calculated. The association of the dependent and independent variables was considered significant at a p-value of less than 0.05. The goodness of fit was checked by the Hosmer and Lemeshow test (p-value = 0.86). Moreover, multicollinearity was assessed for each variable considering variance inflation factor (VIF) values less than 10.

2.11 Operational definitions

BC: Women who had a confirmed BC diagnosis according to histological examination. Case: Women with BC and histologically confirmed cases ( 8 ). Controls: Women who visited gynecological OPD in selected hospitals for noncancer care. Cancer: This is a condition in which body cells multiply uncontrollably and become contagious to other parts of the body ( 19 ). BMI: Somebody’s mass divided by the square of height in meters ( 20 ). For menopausal status, women who met any of the following criteria were classified as postmenopausal: 1) had menstruation for no less than 1 year (any age) and 2) had undergone bilateral oophorectomy or estrogen deprivation therapy ( 21 ). Active smoking is a condition in which people (including former smokers) have a history of smoking within 6 months ( 22 ). Passive smoking: People who are not smokers but are exposed to tobacco smoke for more than 15 min at least 1 day per week ( 22 ).

3.1 Sociodemographic characteristics

A total of 300 women (75 cases and 225 controls) participated in the study for a response rate of 100%. Thirty (40%) of the cases and 139 (61.78%) of the controls were aged less than 40 years with the mean ( ± SD) age of the women being 37.2 ( ± 14.8) and 36.6 ( ± 15.1) for the cases and controls, respectively. More than half of the participants [47 (62.67%) cases and 132 (58.67%) controls] were urban dwellers. Seventeen (22.67%) patients and 41 (18.22%) controls had a college education or above. Regarding occupational status, 26 (34.67%) patients and 79 (35.11%) controls were housewives ( 1 ).

3.2 Reproductive health-related characteristics

Of the total respondents, 34.67% (n = 26) of the patients and 22.22% (n = 50) of the controls had experienced their first menstrual period before the age of 12 years, and the mean ( ± SD) ages of the women who experienced menarche were 13.5 ( ± 2.1) and 14 ( ± 1.92) years for the cases and controls, respectively. Approximately 10 (13.33%) patients and 8.44% (n = 19) of the controls had a history of abortion. Forty-two (56%) patients and 26.22% (n = 59) of the controls were premenopausal women. More than a quarter (29.33%) of the patients and 16.44% (n = 37) of the women had irregular menstrual periods. More than half (58.67%) of the patients and 36.89% (n = 83) of the controls used oral contraceptives. In the majority of the patients, 66.6% (n = 50) and 73% (n = 162) of the controls had never breastfed their infants. One-quarter 25.33%, n = 19) of the patients and 27.11% (n = 61) of the controls had a history of breast infection ( Table 2 ).

Table 2 Reproductive health-related characteristics of the women included in the study in selected health facilities in Hawassa City, 2023 (n = 300).

3.3 Behavioral and biological-related characteristics

Among the patients, 30.67% (n = 23) and 16.00% (n = 36) of the controls had a family history of cancer. Forty percent (n = 30) of the patients and 25.33% (n = 57) of the controls were overweight and/or obese. In the majority of the patients, 77.33% (n = 58) and 70.22% (n = 158) of the controls did not participate in regular physical exercise. In the majority of the participants, 88% (n = 66) of the patients and 82.22% (n = 185) of the controls ate fewer than seven servings of fruit per week. Regarding exposure to smoking, 18.67% (n = 14) of the patients and 22.22% (n = 50) of the controls were exposed. Approximately 40% (n = 30) of the patients and 30.22% (n = 68) of the controls used packed food or drinks. Approximately 50.67% (n = 38) of the patients and 33.78% (n = 76) of the controls consumed solid oil. Among the participants, 9.33% (n = 7) of the patients and 8.8% (n = 8) of the controls had ever undergone a mammogram. Approximately 45.33% (n = 34) of the patients and 44.4% (n = 100) of the controls had practiced self-breast examinations ( Table 3 ).

Table 3 Behavioral and biological characteristics of the study participant women in selected health facilities in Hawassa City, 2023.

3.4 The mean difference in exposure between cases and controls

In this study, we conducted an independent-sample t-test to evaluate the equality of variance and mean difference among cases and controls to exposure variables. To check the similarity of variance among cases and controls, Levene’s test for equality of variances was used (i.e., equal variances, p > 0.05, and unequal variances p < 0.05). Similarly, to identify the mean difference among cases and controls, a t-test for the equality of means with respective p-values was used [p < 0.05 indicates a significant difference in the means of the two sample populations tested (cases and controls)]. Since the variance is greater than 4 for almost all of the test variables, we assumed unequal variance and used a one-sample t-test ( 23 ). The factors associated with an increased risk of BC among women were family history of cancer (p < 0.014), consumption of packed food (p < 0.03), BMI (p < 0.04), fruit consumption (p < 0.03), menopausal status (p < 0.01), and smoking status (p < 0.01) ( Table 4 ).

Table 4 Independent-sample t-test of respondents for determinants of BC among women attending oncology units at selected health facilities of Hawassa City, Sidama Region, Ethiopia, 2023 (n = 75 cases and n = 225 controls).

3.5 Factors associated with BC

According to the bivariate logistic regression analysis, age, menopausal status, use of packed foods or drinks, history of abortion, regularity of menstruation, age at menarche, body mass index (BMI), family history of cancer, use of oral contraceptives, regular physical exercise, and consumption of solid oil were independently associated with BC. However, in the multivariate logistic regression analysis, menopausal status, overweight and obesity status, use of oral contraceptives, family history of cancer, and consumption of solid oil were independently associated with BC.

The odds of having BC were almost three times greater among postmenopausal women than among their male counterparts [adjusted odds ratio (AOR) = 2.5; 95% CI: 1.18–5.2, p = 0.01]. Women with a family history of cancer were 2.33 times more likely to develop BC than women without a family history of cancer (AOR = 2.33; 95% CI: 1.12–4.82, p = 0.02). The odds of developing BC were almost three times greater among oral contraceptive users than among their counterparts (AOR = 2.74; 95% CI: 1.34–5.99, p = 0.005). Similarly, the odds of having BC were almost two times greater among overweight women than among normal-weight women (AOR = 2.29; 95% CI: 1.14–4.59, p = 0.01). Finally, women who used palm/solid oil were 2.4 times more likely to have BC than were their male counterparts (AOR = 2.36; 95% CI: 1.20–4.67, p = 0.01) ( Table 5 ).

Table 5 Bivariate and multivariate logistic regression analysis of factors associated with BC risk among women who attended selected health facilities in Hawassa City, 2023.

4 Discussion

In this facility-based study, we evaluated the determinants of BC among women attending oncology health facilities in Hawassa City, Sidama Region, Ethiopia. Postmenopausal status (p < 0.016), BMI (p < 0.020), family history of cancer (p < 0.023), use of oral contraceptives (p < 0.02), and consumption of saturated fat (p < 0.013) were independently associated with BC.

Our study showed that menopausal status significantly increases BC risk with postmenopausal women having almost three times greater odds than premenopausal women. This finding is similar to that of a study in Addis Ababa ( 24 ). Another study from Malaysia showed that postmenopausal women had a 52% greater risk of BC ( 25 ). A meta-analysis from India comparable with this study showed that postmenopausal women have a 35% greater risk of developing BC than premenopausal women ( 26 ). A possible explanation could be that postmenopausal women face an increased risk of BC due to decreased estrogen levels, obesity, and a sedentary lifestyle, which can increase the risk of developing the disease. Thus, a systematic review and meta-analysis of evidence suggested that postmenopausal women should have regular physical activity and a healthy diet to prevent overweight and obesity to address BC ( 27 ).

Similarly, a family history of cancer significantly increases the risk of BC, and women with a history of BC are 2.3 times more likely to develop BC. This result is in line with research in Addis Ababa ( 8 ), which demonstrated that women with a family history of BC are substantially more likely to develop BC than other women. Early detection and screening could be key for BC progression. Families with a history of BC may also be more aware of the risks and symptoms leading them to seek regular screening and early detection. While this proactive approach to healthcare is important, it can create a perception of greater risk due to increased awareness and monitoring within the family.

The study revealed that oral contraceptives significantly increased BC risk with women using oral contraceptives having 2.74 times greater odds of developing BC than nonusers. This result is in line with a study conducted in Jordan and Ethiopia ( 28 , 29 ). Naturally occurring estrogen and progesterone stimulate the development and growth of some cancers (e.g., cancers that express the receptor for these hormones, such as BC). Birth control pills contain synthetic versions of these female hormones. In contrast, oral contraceptives and BC were not significantly linked in a study in Ethiopia ( 24 ). The possible reason for this slight increase in risk could be related to the hormonal components of oral contraceptives. Oral contraceptives contain synthetic versions of the hormones estrogen and progestin, which are used to prevent pregnancy by suppressing ovulation and altering the cervical mucus and the lining of the uterus. Therefore, maintaining regular BC screenings and self-exams is crucial for the early detection and treatment of BC regardless of whether a woman is using oral contraceptives.

Women’s nutritional status significantly impacts BC risk with overweight or obese women having a 2.29-fold greater risk than normal-weight women. This finding is comparable to that of a study conducted in Jordan and two other studies in Ethiopia ( 8 , 30 , 31 ). The possible justification could be unhealthy eating patterns that are marked by high consumption of refined carbohydrates, sugar, and saturated fats ( 32 ). This increased risk can be attributed to the higher levels of estrogen produced by adipose tissue and the fact that obese women are more physically inactive, both of which can promote the growth of BC cells.

Furthermore, consuming saturated oil significantly increases BC risk. Women consuming solid oil have 2.36 times greater odds of developing BC than nonusers. This result is in line with a study performed in the USA showing that eating saturated fat increases the risk of developing BC ( 33 ). The possible reason could be that saturated fatty acids increase low-density lipoprotein cholesterol, obesity, and free radicals, which increase the risk of BC ( 34 ). A meta-analysis revealed that a high-fat diet is a risk factor for BC ( 35 ). Another systematic review showed that an unhealthy high-fat diet may contribute to obesity and affect BC ( 36 ).

The strengths of this study included the use of an observational study design with a 1:3 case-to-control ratio, which enhances its ability to identify BC determinants and its validity due to its multicenter nature. The possible study limitations include the small number of patients in the health facility, recall bias, and over- or underreporting of self-reported data on dietary consumption and physical activity.

5 Conclusion

BC is a public health problem both globally and in Ethiopia. This study aimed to identify the determinants of BC. Therefore, postmenopausal status, BMI, family history of cancer, being overweight, using oral contraceptives, and consuming solid oil were identified as risk factors for BC. Therefore, women should adopt healthier lifestyles through healthy nutrition and regular physical exercise, which might contribute to reducing the risk of developing BC in women. In addition, early detection and regular screening are proactive approaches for identifying BC.

Data availability statement

The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding authors.

Ethics statement

Ethical clearance was obtained from the Ethical Review Committee of Yanet-Liyana College of Health Science (with Ref # of LHC/YLCHS/OGL/981/15 and Date: 29/05/2023 GC). The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

Author contributions

SK: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Writing – original draft, Writing – review & editing. TA: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Software, Supervision, Writing – original draft, Writing – review & editing. AM: Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Visualization, Writing – original draft, Writing – review & editing.

The author(s) declare that no financial support was received for the research, authorship, and/or publication of this article.

Acknowledgments

We would like to extend our sincere gratitude to Yanet-Liyana College of Health Science for their ethical letter provision. We would also like to acknowledge all our data collectors and supervisors who participated in the data collection at the field level. We would like to extend our gratitude to all the study participants for their important insights and contributions in providing information in this research process.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

1. Arnold M, Morgan E, Rumgay H, Mafra A, Singh D, Laversanne M, et al. Current and future burden of breast cancer: Global statistics for 2020 and 2040. Breast . (2022) 66:15–23. doi: 10.1016/j.breast.2022.08.010

PubMed Abstract | CrossRef Full Text | Google Scholar

2. Kashyap D, Pal D, Sharma R, Garg VK, Goel N, Koundal D, et al. Global increase in breast cancer incidence: risk factors and preventive measures. BioMed Res Int . (2022) 2022. doi: 10.1155/2022/9605439

3. Sasco AJ. Cancer and globalization. BioMed Pharmacother . (2008) 62:110–21. doi: 10.1016/j.biopha.2007.10.015

4. Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA: A Cancer J Clin . (2021) 71:209–49. doi: 10.3322/caac.21660

CrossRef Full Text | Google Scholar

5. Nguyen D, Yu J, Reinhold WC, Yang SX. Association of independent prognostic factors and treatment modality with survival and recurrence outcomes in breast cancer. JAMA Netw Open . (2020) 3:e207213–e. doi: 10.1001/jamanetworkopen.2020.7213

6. Mekonnen BD. tffBreast self-examination practice and associated factors among female healthcare workers in Ethiopia: a systematic review and meta-analysis. PloS One . (2020) 15:e0241961. doi: 10.1371/journal.pone.0241961

7. Bray F, Parkin DM. Cancer in sub-Saharan Africa in 2020: a review of current estimates of the national burden, data gaps, and future needs. Lancet Oncol . (2022) 23:719–28. doi: 10.1016/S1470-2045(22)00270-4

8. Tolessa L, Sendo EG, Dinegde NG, Desalew A. Risk factors associated with breast cancer among women in addis ababa, Ethiopia: Unmatched case–control study. Int J Women’s Health . (2021) 13:101–10. doi: 10.2147/IJWH.S292588

9. Bacha RH, Jabir YN, Asebot AG, Liga AD. Risk factors affecting survival time of breast cancer patients: The case of southwest Ethiopia. J Res Health Sci . (2021) 21:e00532. doi: 10.34172/jrhs.2021.65

10. Anderson BO, Abdel-Wahab M, Fidarova E. The Global Breast Cancer Initiative: a strategic collaboration to strengthen health care for noncommunicable diseases. Lancet Oncol . (2021) 22(5):578–81. doi: 10.1016/S1470-2045(21)00071-1

11. Bejitual K, Fikre R, Ashegu T, Zenebe A. Determinants of neonatal sepsis among neonates admitted to the neonatal intensive care unit of public hospitals in Hawassa City Administration, Sidama Region, Ethiopia, 2020: an unmatched, case–control study. BMJ Open . (2022) 12:e056669. doi: 10.1136/bmjopen-2021-056669

12. Mertens E, Barrenechea-Pulache A, Sagastume D, Vasquez MS, Vandevijvere S, Peñalvo JL. Understanding the contribution of lifestyle in breast cancer risk prediction: a systematic review of models applicable to Europe. BMC Cancer . (2023) 23(1):687. doi: 10.1186/s12885-023-11174-w

13. WHO. Breast cancer fact sheet (2021). Available online at: https://www.who.int/news-room/fact-sheets/detail/breast-cancer .

Google Scholar

14. Liu H, Shi S, Gao J, Guo J, Li M, Wang L. Analysis of risk factors associated with breast cancer in women: a systematic review and meta-analysis. Trans Cancer Res . (2022) 11:1344. doi: 10.21037/tcr

15. Khalis M, Charbotel B, Chajès V, Rinaldi S, Moskal A, Biessy C, et al. Menstrual and reproductive factors and risk of breast cancer: A case−control study in the Fez region, Morocco. PloS One . (2018) 13:e0191333. doi: 10.1371/journal.pone.0191333

16. Duche H, Tsegay AT, Tamirat KS. Identifying risk factors of breast cancer among women attending selected hospitals of Addis Ababa city: hospital-based unmatched case−Control study. Breast Cancer (Dove Med Press) . (2021) 13:189–97. doi: 10.2147/BCTT.S293867

17. Tolessa L, Sendo EG, Dinegde NG, Desalew A. Risk factors associated with breast cancer among women in Addis Ababa, Ethiopia: unmatched case−Control study. Int J Womens Health . (2021) 13:101–10. doi: 10.2147/IJWH.S292588

18. Tekeste Z, Berhe N, Arage M, Degarege A, Melaku YA. Cancer signs and risk factors awareness in Addis Ababa, Ethiopia: a population-based survey. Infect Agent Cancer . (2023) 18:1. doi: 10.1186/s13027-022-00477-5

19. Siegel RL, Miller KD, Wagle NS, Jemal A. Cancer statistics, 2023. CA: Cancer J Clin . (2023) 73:17–48. doi: 10.3322/caac.21763

20. García-Estévez L, Cortés J, Pérez S, Calvo I, Gallegos I, Moreno-Bueno G. Obesity and breast cancer: a paradoxical and controversial relationship influenced by menopausal status. Front Oncol . (2021) 11:705911. doi: 10.3389/fonc.2021.705911

21. Rothschild HT, Abel MK, Patterson A, Goodman K, Shui A, van Baelen K, et al. Obesity and menopausal status impact the features and molecular phenotype of invasive lobular breast cancer. Breast Cancer Res Treat . (2022) 191:451–8. doi: 10.1007/s10549-021-06453-8

22. Fakhri N, Chad MA, Lahkim M, Houari A, Dehbi H, Belmouden A, et al. Risk factors for breast cancer in women: an update review. Med Oncol . (2022) 39:197. doi: 10.1007/s12032-022-01804-x

23. Zimmerman DW. Teacher’s corner: A note on interpretation of the paired-samples t test. J Educ Behav Stati . (1997) 22:349–60. doi: 10.3102/10769986022003349

24. Hassen F, Enquselassie F, Ali A, Addissie A, Taye G, Tsegaye A, et al. Association of risk factors and breast cancer among women treated at Tikur Anbessa Specialized Hospital, Addis Ababa, Ethiopia: a case–control study. BMJ Open . (2022) 12:e060636. doi: 10.1136/bmjopen-2021-060636

25. Durham DD, Abraham LA, Roberts MC, Khan CP, Smith RA, Kerlikowske K, et al. Breast cancer incidence among women with a family history of breast cancer by relative’s age at diagnosis. Cancer . (2022) 128:4232–40. doi: 10.1002/cncr.34365

26. Vishwakarma G, Ndetan H, Das DN, Gupta G, Suryavanshi M, Mehta A, et al. Reproductive factors and breast cancer risk: A meta-analysis of case–control studies in Indian women. South Asian J cancer . (2019) 8:080–4. doi: 10.4103/sajc.sajc_317_18

27. Dehesh T, Fadaghi S, Seyedi M, Abolhadi E, Ilaghi M, Shams P, et al. The relation between obesity and breast cancer risk in women by considering menstruation status and geographical variations: a systematic review and meta-analysis. BMC Womens Health . (2023) 23:392. doi: 10.1186/s12905-023-02543-5

28. Bardaweel SK, Akour AA, Al-Muhaissen S, AlSalamat HA, Ammar K. Oral contraceptive and breast cancer: do benefits outweigh the risks? A case–control study from Jordan. BMC women’s Health . (2019) 19:1–7. doi: 10.1186/s12905-019-0770-x

29. Das D. Breast cancer: Risk factors and prevention strategies. World J Biol Pharma Health Sci. (2022) (12)(3):265280. doi: 10.30574/wjbphs.2022.12.3.0253

30. Ayoub NM, Yaghan RJ, Abdo NM, Matalka II, Akhu-Zaheya LM, Al-Mohtaseb AH. Impact of obesity on clinicopathologic characteristics and disease prognosis in preand postmenopausal breast cancer patients: a retrospective institutional study. J Obes . (2019) 2019:1–11. doi: 10.1155/2019/3820759

31. Duche H, Tsegay AT, Tamirat KS. Identifying risk factors for breast cancer among women attending selected hospitals of Addis Ababa city: Hospital-based unmatched case−control study. Breast Cancer: Targets Ther . (2021) 13:189–97. doi: 10.2147/BCTT.S293867

32. Seiler A, Chen MA, Brown RL, Fagundes CP. Obesity, dietary factors, nutrition, and breast cancer risk. Curr Breast Cancer Rep . (2018) 10:14–27. doi: 10.1007/s12609-018-0264-0

33. Dandamudi A, Tommie J, Nommsen-Rivers L, Couch S. Dietary patterns and breast cancer risk: a systematic review. Anticancer Res . (2018) 38:3209–22. doi: 10.21873/anticanres.12586

34. Xia H, Ma S, Wang S, Sun G. Meta-analysis of saturated fatty acid intake and breast cancer risk. Medicine . (2015) 94:e2391. doi: 10.1097/MD.0000000000002391

35. Uhomoibhi TO, Okobi TJ, Okobi OE, Koko JO, Uhomoibhi O, Igbinosun OE, et al. High-fat diet as a risk factor for breast cancer: a meta-analysis. Cureus . (2022) 14:e32309. doi: 10.7759/cureus.32309

36. Gopinath A, Cheema AH, Chaludiya K, Khalid M, Nwosu M, Agyeman WY, et al. The impact of dietary fat on breast cancer incidence and survival: a systematic review. Cureus . (2022) 14:e30003. doi: 10.7759/cureus.30003

Keywords: breast cancer, case–control, determinates, oncology units, Sidama Region Ethiopia

Citation: Kebede S, Alemu T and Mekonnen A (2024) Determinants of breast cancer among women attending oncology units in selected health facilities of Hawassa City, Sidama Region, Southern Ethiopia, 2023: case-control study. Front. Oncol. 14:1352191. doi: 10.3389/fonc.2024.1352191

Received: 07 December 2023; Accepted: 25 March 2024; Published: 12 April 2024.

Reviewed by:

Copyright © 2024 Kebede, Alemu and Mekonnen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Selamawit Kebede, [email protected] ; Tsegaye Alemu, [email protected] ; Ashenafi Mekonnen, [email protected]

† ORCID : Selamawit Kebede, orcid.org/0009-0000-1980-0938 Tsegaye Alemu, orcid.org/0000-0001-9511-9028 Ashenafi Mekonnen, orcid.org/0000-0001-5272-5829

Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• Br J Cancer
• v.124(4); 2021 Feb 16

A case-control study to evaluate the impact of the breast screening programme on mortality in England

Roberta maroni.

1 Centre for Cancer Prevention, Wolfson Institute of Preventive Medicine, Queen Mary University of London, Charterhouse Square, London, EC1M 6BQ UK

Nathalie J. Massat

Dharmishta parmar, amanda dibden, jack cuzick, peter d. sasieni.

2 Faculty of Life Sciences and Medicine, Cancer Prevention Group, School of Cancer and Pharmaceutical Sciences, King’s College London, Guy’s Campus, Great Maze Pond, London, SE1 9RT UK

Stephen W. Duffy

Associated data.

Data were saved on the servers of the Barts Cancer Institute, Queen Mary University of London, in a folder with restricted access to D.P. A clean, anonymised version of the data was produced and made available to R.M., A.D. and S.W.D. with restricted access to the staff of the Policy Research Unit in Cancer Awareness, Screening and Early Diagnosis at Queen Mary University of London. The data were obtained via the Office for Data Release at Public Health England. We do not have authority to share the data with others, but requests for access to data will be forwarded to the Office for Data Release.

Over the past 30 years since the implementation of the National Health Service Breast Screening Programme, improvements in diagnostic techniques and treatments have led to the need for an up-to-date evaluation of its benefit on risk of death from breast cancer. An initial pilot case-control study in London indicated that attending mammography screening led to a mortality reduction of 39%.

Based on the same study protocol, an England-wide study was set up. Women aged 47–89 years who died of primary breast cancer in 2010 or 2011 were selected as cases (8288 cases). When possible, two controls were selected per case (15,202 controls) and were matched by date of birth and screening area.

Conditional logistic regressions showed a 38% reduction in breast cancer mortality after correcting for self-selection bias (OR 0.62, 95% CI 0.56–0.69) for women being screened at least once. Secondary analyses by age group, and time between last screen and breast cancer diagnosis were also performed.

Conclusions

According to this England-wide case-control study, mammography screening still plays an important role in lowering the risk of dying from breast cancer. Women aged 65 or over see a stronger and longer lasting benefit of screening compared to younger women.

Following an evaluation of several randomised controlled trials (RCT) 1 that showed an overall reduction in mortality from breast cancer in women undergoing mammography screening, the National Health Service Breast Screening Programme (NHS BSP) was launched in the United Kingdom (UK) in 1988. At the time, it aimed to offer free routine screening to every woman aged 50–64 once every three years. It now invites women aged 50–70, with an age extension to younger and older women (47–73 years) being trialled. 2

Over the last thirty years, major advances have been made in the fields of cancer screening, treatment, and management (including effective adjuvant systemic therapies 3 and two-view mammography 3 , 4 ), with resulting lengthening of survival times after a breast cancer diagnosis. 5 Despite recent reductions in breast cancer mortality, breast cancer is still the cancer with the highest incidence 6 and the second most common cause of cancer death 7 in females in the UK.

Case-control studies are a useful tool to evaluate screening programmes in settings where lack of equipoise would mean that RCTs would be unethical, or as in this case, where the RCTs have already been done, but there remains a need to ensure that the service is delivering the expected clinical benefit. Case-control studies also overcome some limitations associated with other observational designs by taking into account changes in cancer incidence and use of treatments over time and adjusting for any imbalances in other factors that could affect breast cancer mortality.

Taking as an example a case-control study 8 that resulted in policy change within the NHS cervical screening programme by altering age at first screen and the screening interval, we designed a similar study focussing on the NHS BSP with the aim of:

• Evaluating the effect of mammography screening in the NHSBSP on breast cancer mortality
• Evaluating the effect of mammography screening on breast cancer incidence, and incidence of late stage disease
• Estimating overdiagnosis
• Analysing the interplay of early detection, pathology, and treatment on fatality of breast cancer.

The study protocol and results from two pilot studies have been published previously. 9 – 11 This paper reports on the first objective above (breast cancer mortality), making use of England-wide data. Effects on incidence etc. will be reported in future papers.

Definition of cases and controls

As the main objective was to evaluate the effect of mammography screening on breast cancer mortality, cases were defined as women whose primary cause of death was breast cancer, who were diagnosed at age 47 years or older and died at age 89 years or younger in 2010–2011. We chose the lower limit of 47 as there is a major trial of screening in ages 47–49 ongoing, 2 so substantial numbers of women have been screened in this age group. We chose the upper limit of 89 because above this age we would not expect a major effect of screening taking place mainly at ages 50–70, because we were less confident of the cause of death in the very old, and because screening is essentially aimed at preventing premature mortality, which one might reasonably interpret as death below age 90 years. Only diagnoses occurring after 1990 were included in the analysis. Their matched controls were women sampled from the general population of those invited for screening (99.9% of women eligible for screening in England 12 ) and alive at the time of their corresponding case’s death. Controls may have been diagnosed with breast cancer, but not before their case’s date of diagnosis. Where possible, two controls were selected per case and matched on date of birth (within one month of the case’s) and screening area at date of diagnosis.

For the purposes of the statistical analysis, controls were assigned a date of pseudodiagnosis, equal to the diagnosis date of their corresponding matched case. To be eligible as a case or a control, a woman had to have had at least one invitation to screening prior to the date of diagnosis/pseudodiagnosis.

The primary endpoint was to estimate, among those invited to breast screening, the effect of ever attending breast screening on mortality from breast cancer. Changes in this effect over time were also investigated. Secondary endpoints included the effect of measures of screening intensity, such as time between last screen and diagnosis/pseudodiagnosis, and their estimations in different age subgroups.

Data selection and linkage

Cases were identified from the National Cancer Registration and Analysis Service (NCRAS) database accessed through the Office for Data Release of Public Health England (PHE). This database contains Office for National Statistics date and cause of death data. NHS Digital used the National Health Application and Infrastructure Services (NHAIS) system to identify matched controls and provided breast and cervical screening histories within.

We excluded any breast screens occurring outside the usual call/recall system of the national screening programme. All the screening histories of the study subjects were considered up to and including their date of diagnosis/pseudodiagnosis.

The data were processed according to the NHS Information Governance guidelines. 13

Sample size

Sample size calculations for the pilot study showed that, assuming an OR for breast cancer mortality of 0.7 and a number of discordant pairs of 33%, two controls per case with 800 breast cancer deaths and 1600 controls would confer more than 90% power to detect such an effect size at the 5% significance level using a two-sided test. 10 As the data for this main phase encompassed the whole of England, we had ample power, not only for the primary outcome (8288 cases and 15,202 controls after exclusions), but also for subgroup analyses.

Statistical analysis

Data were analysed using Stata version 13 14 by matched (conditional) logistic regression with death from primary breast cancer as the outcome. Date of birth and screening area were accounted for by the matching process.

Ineligible subjects were excluded (see Fig.  1 ). For some of these, this resulted in a matched set containing only a case, or only controls, which could then no longer be used in the matched logistic regression. Sensitivity analyses using unmatched logistic regression and controlling for age at diagnosis/pseudodiagnosis and screening area were performed on the same dataset with fewer exclusions; in this case, the inclusion criteria considered were the same, but the fact that a case or a control was excluded did not imply discarding that matched set.

Asterisk indicates that these records were excluded for being in a 1:1 matched set where the case or the control was excluded or for being in a 1:2 matched set where the case or both controls were excluded. Hash indicates that these become 1:1 matched sets in the final dataset. Note: some records may be excluded for more than one reason.

Case-control studies used to evaluate population screening programmes are subject to a type of bias known as non-compliance or self-selection bias, which is based on the assumption that people who are already ill may be less likely to attend screening and those who do attend may be more health conscious, and therefore healthier, than those who do not take up the invitation. This may confer an artificially greater protective effect for screening, which was corrected in our analyses using a variant of the method by Duffy et al. 15

The effect of self-selection bias was estimated using data available on cervical screening attendance for the women in the study, on the basis that any observed protective effect of cervical screening on breast cancer death cannot be due to cervical screening (which does not include breast examination) and is therefore likely to be caused by self-selection bias. In particular, the odds ratio (OR) uncorrected for self-selection is an estimate of the relative risk:

An unbiased estimate of the effect of screening on risk of dying from breast cancer would be (refer to Duffy et al. 15 ):

The OR for death from breast cancer associated with attendance at cervical screening, i.e. the self-selection correction factor, can be considered an approximate estimate of the relative risk:

Therefore, we obtain an estimate of θ by dividing γ by φ . The fundamental assumption here is that the populations choosing to attend or not to attend cervical cancer screening have the same risk of dying of breast cancer a priori as those choosing or not choosing to attend breast cancer screening. We do not assume that the effects of self-selection are the same in the two programmes. This is referred to as our first method of correction in the Results section.

As there is considerable uncertainty in the extent of self-selection, and of course decisions to attend at two separate screening programmes are likely to be confounded with each other, we also corrected for this using the method of Duffy et al. 15 . This method estimates the effect of participation in screening in those who would participate if invited as:

where p is the proportion of the invited population who participate in screening and D r is the a priori relative risk of dying of breast cancer for someone who chooses not to attend compared to an uninvited general population member. We estimated D r as 1.19 (95% CI 1.11–1.27), from the cohort study of Johns et al. 16 Thus, this correction was based on a prospective estimate of the extent of self-selection bias in a cohort of 988,090 women in the NHS Breast Screening Programme. We estimated p as 73.4% from the annual report of the National Programme. 12 This method, referred to as our second method of correction in the Results section, also yields an estimate of the effect of invitation to screening as follows: 15

More details on the methods are available in the published study protocol 9 and pilot study analysis. 10

The study dataset had a total of 9550 cases and 17,993 controls. There were 1107 sets with matching ratio 1:1 (1 case to 1 control) and 8443 sets with matching ratio 1:2 (1 case to 2 controls). Records of 1262 cases and 2791 controls (15% of the total) were excluded for various reasons before the statistical analysis (see study flow diagram in Fig.  1 ). This left a final dataset of 8288 cases and 15,202 controls, divided into 1,374 matched sets of size 1:1 and 6914 of size 1:2.

Sensitivity analyses using unconditional logistic regression were performed including subjects without a matched case or control, leaving us with 8479 cases and 16,794 controls.

Table  1 shows patient demographics and screening histories. Median age at first diagnosis was 64 years for both cases and controls and median age at death for cases was 71 years. Whilst the distributions of the number of screening invitations in the two study groups were comparable, differences can be noted in screening attendance, with 72% of the cases versus 82% of the controls attending their first screening invitation; 64% of the cases versus 76% of the controls attending their last screening invitation before diagnosis/pseudodiagnosis; and 21% of the cases versus 12% of the controls never being screened. Median time between last screen and date of diagnosis/pseudodiagnosis for compliers was also slightly longer for cases. From the data available on cervical screening history up to the date of diagnosis/pseudodiagnosis, it can be noted that 22% of the cases compared with 19% of the controls never had a cervical screen.

Patient demographics and screening history by case-control status.

Table  2 summarises the main results without and with correction for self-selection bias. Using data from cervical screening attendance, the self-selection correction factor was estimated to be 0.78 (95% CI 0.73–0.84). The primary endpoint, the association between attending one or more screens and death from breast cancer, had a resulting OR = 0.49 (95% CI 0.45–0.53) and, when corrected for self-selection, had OR = 0.62 (95% CI 0.56–0.69) by our first method and OR = 0.63 (95% CI 0.55–0.71) by our second. Using the second method, the estimate of the effect of invitation to screening was a 26% reduction in breast cancer mortality (OR = 0.74, 95% CI 0.68-0.81). The unmatched logistic regression on the larger dataset for sensitivity analyses showed a similar effect of screening on breast cancer mortality both before and after controlling for age at diagnosis/pseudodiagnosis and screening area (in both cases, uncorrected OR = 0.55, 95% CI 0.51–0.59).

Results of the matched logistic regression evaluating the association between screening attendance and breast cancer mortality.

a Self-selection correction performed using our first method (variant of Duffy et al. 15 ), with the OR of 0.78 associated with participation in cervical screening.

b Self-selection correction performed using our second method (Duffy et al. 15 ).

In order to analyse changes of the effect of screening over time, we excluded women diagnosed before year 2000 (13% of the total records), which led to a corrected OR of 0.56 (95% CI 0.51–0.63) for the effect of ever attending mammographic screening on breast cancer mortality. Women diagnosed from year 2003 onwards had an even larger benefit from being screened (OR corrected by first method = 0.53, 95% CI 0.47–0.59). The estimated effect continued to increase as we restricted the year of diagnosis/pseudodiagnosis further in time (Supplementary Fig.  1 ).

Table  3 shows how the effect of screening varies depending on how much time has passed between a woman’s last screen and her diagnosis/pseudodiagnosis. Screen-detected cancers (assumed to be cancers diagnosed within three months of screening) showed a positive association with breast cancer fatality, after self-selection bias correction by our first method (OR = 1.93, 95% CI 1.68–2.22), while women screened in any other time interval were at reduced risk of dying from breast cancer. This was lowest for women screened in the last year (OR corrected by our first method = 0.19, 95% CI 0.17–0.23) and gradually increased, while still conferring a beneficial effect to screening, for women screened further back in time with respect to their date of diagnosis/pseudodiagnosis. Results using our alternative correction for self-selection were very similar (Table  3 ). Note that the time is from screening to diagnosis, not to death. The Table shows risk of subsequently dying of breast cancer increasing by the time between the screen and diagnosis/pseudodiagnosis.

Results of the matched logistic regressions evaluating the association between time since last screening attendance and breast cancer mortality.

A similar analysis is shown in Table  4 and Fig.  2 for different time intervals after stratifying for three different age categories at diagnosis/pseudodiagnosis (younger than 60 years, between 60 and 64 years, and 65 years or older). The results show that the protective effect of a screen is greater and lasts longer in the oldest group. The benefit of attending screening in the three years prior to diagnosis/pseudodiagnosis, the recommended interval for screening in the NHS BSP, is shown in the final row of Table  4 , and shows close to a halving of risk with screening within the recommended interval, following self-selection correction by our first method (OR = 0.51, 95% CI 0.46–0.57). Results using our second method of correction were very similar to those using the first (Supplementary Table  1 ). The estimated effect of invitation to screening within the last 36 months using our second method was a 33% reduction in breast cancer mortality (OR = 0.67, 95% CI 0.61–0.73).

Results of the matched logistic regressions evaluating the association between time since last screening attendance and breast cancer mortality, stratified by age at diagnosis/pseudodiagnosis.

a Self-selection correction performed using our first method (variant of Duffy et al 15 ), with the OR of 0.78 associated with participation in cervical screening.

Note: the coordinates on the x -axis are the midpoints of the time intervals: 0–3, 3–6, 6–18, 18–36, 36–54 and 54–72 months.

Despite the many improvements in treatments, diagnostic procedures and technologies over the last thirty years, and changes in baseline rate of breast cancer mortality, our data showed an overall reduction in the risk of dying from breast cancer of ~38% for women attending at least one mammography screen, after adjusting for self-selection bias. This is in line with the results obtained from the pilot phase of the study, 10 in which a mortality reduction of 39% was seen for women attending screening in London (deaths occurring in 2008–2009). Using the same calculation method as in the review by the Independent UK Panel on Breast Cancer Screening UK Independent Review, 17 this would correspond to approximately nine breast cancer deaths prevented for every 1,000 women attending screening at ages 50–69 years, larger than but in the same general scale as the six deaths estimated from the UK Independent review.

It should be noted that there is a wide range of estimates of the absolute mortality benefit of mammography screening 18 – 21 some finding considerably smaller benefits than above. The size of the estimated effect depends on sources used and assumptions made. However, it has been shown to depend more crucially on whether the effect pertains to screening per se or to invitation to screening only, and on the timescale envisaged. 22 Screening prevents deaths not this year or next, but 5, 10, 15 or 20 years from now. Considering the effect of screening on 10-year mortality will considerably underestimate the absolute benefit. Nevertheless, it should be acknowledged that while the body of evidence, randomised and observational, points to a substantial reduction in breast cancer mortality with screening, there is sufficient variation that different views are still possible.

Our first method of correction for self-selection caused a decrease of about 25% in the estimated protective effect of screening for women having at least one mammogram. The second method yielded similar results. This is a greater correction than the one estimated in the pilot phase, 10 where self-selection only played a minor role, despite the fact that the final risk reduction is very similar. London has a lower coverage than the rest of England for both breast and cervical screening, which is largely explained by factors like deprivation and ethnicity. 23 Such variations in coverage might be one of the causes for the different impact of self-selection between the two phases of the study. For example, a larger population of non-participants, such as in London, may be less different in health status than a smaller population. In the Swedish two-County trial, 24 where only 15% of the population were non-participants, the rate of death from breast cancer in this population was very high. It is also worth noting that, during the early 21st century, breast screening attendance was rapidly increasing in London, and the socioeconomic gradient in attendance was reducing with time nationally. 25 , 26

Case-control studies tend to give higher estimates of benefit than other evaluations, largely because they assess the effect of actually being screened rather than simply being invited to screening. 19 , 27 It should be noted that with our second correction for self-selection bias, we were able to estimate the effect of invitation, giving a 26% breast cancer mortality reduction, similar to the effect observed in the randomised trials in this age group and to the prospectively estimated effect of a 25% reduction in the Copenhagen screening programme. 28 As a comparison, in the review by the Independent UK Panel on Breast Cancer Screening, 17 a meta-analysis of 11 RCTs found that the relative risk reduction of breast cancer mortality for women invited to screening was 20%. Furthermore, in the same report, the panel stated that the case-control studies that they had analysed seemed to inflate the benefit of screening compared to the trials and postulated that this may have been caused by some residual bias unaccounted for by the authors. We believe that our adjustments for self-selection bias has largely accounted for this and that the greater effect of screening in this study is due to technical improvements in mammography since the RCTs were carried out, accompanied by improved treatment and strong quality assurance measures in the NHS BSP. 11

The greater benefit of screening observed for women diagnosed after year 2000 was similar to the pilot study, 10 but here we were able to restrict the analysis to later years of diagnosis and see the benefit getting larger (data not shown). We could conjecture that this improvement was due to the introduction of better procedures in the NHS BSP, such as two-view mammography at every attendance in year 2000 4 ; however, there may be a bias in comparing different times since diagnosis as we only have data on deaths in years 2010–2011. In the first place, cases diagnosed before 2000 have a long survival by definition, and there might therefore be an over-representation of screen-detected cancers. In other words, it is more likely that a case diagnosed before year 2000, for example, who had a breast cancer for more than 10–11 years before dying from it, had a screen-detected cancer rather than a symptomatic one. This confers a bias against screening in the analysis of cancers diagnosed prior to the year 2000. In the second place, there will be a bias in favour of screening if the analysis is restricted to cancers diagnosed within a short time before death, i.e. if we only consider women (pseudo)diagnosed a few years before 2010–2011. We are therefore unable to make any definitive conclusions on the impact of any improvements in the NHS BSP over time.

As shown in RCTs of breast screening, 24 measures of the benefit of screening are largely influenced by the consequent reduction in mortality from symptomatic cancers. This is due to the fact that screen-detected cancers (defined as the ones diagnosed within three months of a screen), despite being less fatal overall, represent a larger proportion of the cancer-related deaths in the immediate period after a screen as it can be seen from the spike in excess mortality in Fig.  2 .

The duration of the benefit of attending screening appears to be greater in older women (Table  4 and Fig.  2 ). Women aged 65 or more see the greatest and longer lasting benefit, which might suggest that they could be screened less often than younger women. This result is in agreement with the impact of ageing on breast cancer biology 29 and is also potentially important in light of the recent incident in the NHS BSP, where a number of women aged 69 and 70 years did not receive the scheduled invitation to their last screening appointment. 30 The exact number affected has been debated but an Independent Review concluded that 5000 women were not invited as scheduled, and that a further 62,000 could be interpreted as having missed their final invitation as defined in the service specification. 30 Our findings suggest that the effect of a delayed screen in older women has a lesser consequence for increased risk of breast cancer mortality than it would have had in younger women. While three years is a longer interval than other programmes in Europe and North America, and further slippage of the interval should be avoided if at all possible, these results could also be used as guidelines for screening units at times of capacity constraints, with the provision that all women receive an opportunity for a final screen around or shortly after age 70. There is interest in stratified screening and these results may inform further thinking on this subject.

A limitation of the study is the retrospective design and the potential for self-selection bias. We have corrected for this in two different ways and for one of these, an effect of invitation to screening was derived which was consistent with trials results and prospective studies for this age group. However, it must be acknowledged that there remains some uncertainty about the extent of self-selection bias. Furthermore, case-control studies for cancer screening programmes are subject to an inherent type of anti-screening bias known as screening opportunity bias. 27 As most of the controls do not have a breast cancer diagnosis, the only way they can be exposed to screening is if they attended a mammography appointment in the past. Cases, on the other hand, may have had a screen in the past, but some of them will also have an additional screen for when their cancer was diagnosed. This induces an artificially higher retrospective probability of screening exposure among cases. Screening opportunity bias was corrected for in the pilot study, 10 where a 10–15% increase in mortality reduction was seen following this, but here we preferred to keep a conservative approach and not adjust for it. To minimise biases with respect to age and opportunity to be screened, we matched very closely for age. This meant that in 1107 cases out of 9550, we could only find one control.

Although the effect of the NHS BSP in preventing breast cancer mortality has been assessed several times, 31 – 34 we are aware of only one other case-control study conducted using national data. 34 The latter relies on data up to year 2005 (diagnoses and deaths took place between 1991 and 2005), while ours uses more recent data up to year 2012, arguably more in the epoch of effective adjuvant systemic therapies. It is of interest that our more recent case base shows similar results in terms of the reduction in risk of breast cancer death with screening. In any case, we suggest that it would be of interest to repeat this type of analysis for years thereafter, to ensure that the programme continues to deliver its aims even with the introduction of new diagnostic technologies (e.g. digital mammography). Before the establishment of the NHS BSP in 1987, it was suggested that a routine case-control assessment could and should be part of an ongoing evaluation of a mass screening programme. 35 For this reason, we believe that this exercise should be held on a two-yearly basis.

The results of further national case-control studies (1) evaluating the effect of the NHS BSP on breast cancer incidence and incidence of late stage disease, (2) estimating overdiagnosis, and (3) analysing the interplay of early detection, pathology and treatment on fatality of breast cancer will be published shortly.

To conclude, this study showed that the breast screening programme in England continues to play an important role in the control of breast cancer. The effect of screening within the NHS BSP in England is stronger and longer lasting in women aged 65 or over, but it remains highly relevant for younger women.

Supplementary information

Acknowledgements.

Data for this study is based on information collected and quality assured by the PHE National Cancer Registration and Analysis Service. Access to the data was facilitated by the PHE Office for Data Release. We would like to thank Rachael Brannan from the PHE Office for Data Release and David Graham from NHS Digital for their help with the data selection and matching of cases and controls. This work uses data provided by patients and collected by the NHS as part of their care and support.

Author contributions

R.M. oversaw the first draft of the manuscript draft and submission. N.J.M., J.C., P.D.S. and S.W.D. designed the study. D.P. contributed to data collection and cleaning. A.D. assisted with data interpretation. R.M. and S.W.D. analysed the data and produced the figures. All the authors critically reviewed the paper.

Ethics approval and consent to participate

The study protocol was reviewed and approved by the Department of Health. Ethical approval was obtained from the London Research Ethics Committee of the National Research Ethics Service (reference: 12/LO/1041), and by the National Information Governance Board Ethics and Confidentiality Committee (reference: ECC 6–05 (e)/2012). The ethics committee agreed that informed consent to participate for the study subjects was not necessary. The study was performed in accordance with the Declaration of Helsinki.

Data availability

Competing interests.

P.D.S. reports personal fees from GRAIL Bio outside the submitted work. J.C. and S.W.D. are members of the editorial board of the British Journal of Cancer. The remaining authors declare no competing interests.

Funding information

his research is funded by the National Institute for Health Research (NIHR) Policy Research Programme, conducted through the Policy Research Unit (PRU) in Cancer Awareness, Screening and Early Diagnosis, PR-PRU-1217-21601. The PRU is a collaboration between researchers from seven institutions (Queen Mary University of London, University College London, King’s College London, London School of Hygiene and Tropical Medicine, Hull York Medical School, Durham University, and Peninsula Medical School). The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care. The funding body was not involved in design, data collection, analysis or interpretation. The funding body had sight of the paper prior to publication but has not had input to its content.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

These authors contributed equally: Roberta Maroni, Nathalie J Massat

These authors jointly supervised this work: Peter D Sasieni, Stephen W Duffy

Supplementary information is available for this paper at 10.1038/s41416-020-01163-2.