Absence of sustained breast cancer incidence inflation in a national mammography screening programme

Abstract

Objective

To investigate the impact of population mammography screening on breast cancer incidence trends in New Zealand.

Methods

Trends in age-specific rates of invasive breast cancer incidence (1994–2014) were assessed in relation to screening in women aged 50–64 from 1999 and 45–69 following the programme age extension in mid-2004.

Results

Breast cancer incidence increased significantly by 18% in women aged 50–64 compared with 1994–98 (p<0.0001), coinciding with the 1999 introduction of mammography screening, and remained elevated for four years, before declining to pre-screening levels. Increases over 1994–99 incidence occurred in the 45–49 (21%) and 65–69 (19%) age groups following the 2004 age extension (p<0.0001). Following establishment of screening (2006–10), elevated incidence in the screening target age groups was compensated for by lower incidence in the post-screening ⩾70 age groups than in 1994–98. Incidence in women aged ⩾45 was not significantly higher (+5%) after 2006 than in 1994–98. The cumulated risk of breast cancer in women aged 45–84 for 1994–98 was 10.7% compared with 10.8% in 2006–10.

Conclusions

Increases in breast cancer incidence following introduction of mammography screening in women aged 50–64 did not persist. Incidence inflation also occurred after introduction of screening for age groups 45–49 and 65–69. The cumulated incidence for women aged 45–84 over 2006–10 after screening was well established, compared with 1994–98 prior to screening, shows no increase in diagnosis. Over-diagnosis is not inevitable in population mammography screening programmes.

Keywords

Breast cancer service screening mammography incidence New Zealand

Introduction

The population of New Zealand (NZ) in 2013 was 4.2 million.¹ Breast cancer is the most common cancer in NZ women, accounting for 28% of all female cancers, and is the second most common cause of cancer death in women after lung cancer.² For women aged 25–64, breast cancer is the leading cause of cancer death.² Accurate breast cancer incidence from complete cancer registration is available in NZ from 1994.³

Reflecting trends in other developed countries,^4,5 the incidence of invasive breast cancer in NZ may have risen over the past half century from changes in risk factors such as obesity,^6,7 fertility patterns,^8,9 use of hormone replacement therapy,¹⁰ and the introduction of mammography screening.^4,5,11 A component of screening-related incidence may include over-diagnosed cancers, which is evident in some screening programmes,^5,12–14 but not others.^15,16 Immediately following the introduction of a mammography screening programme, there is generally a rise in recorded breast cancer incidence, known as the ‘prevalence peak’,^11,16,17 from diagnosis of prevalent cancers in the initial screening round, and earlier diagnosis of incident cases. Following this initial increase, the overall incidence is expected to return to pre-screening levels as incident cancers are redistributed to younger ages compensated for by lower incidence at older ages. This expected pattern has been observed in mammography screening programmes in Denmark and Spain,^16,18 but not in Australia,^5,11 the United Kingdom,^4,19 and Sweden,¹² where incidence has remained elevated over an extended period of time and raises the possibility of over-diagnosis, where cancers detected by screening would never have manifested clinically over a lifetime.

This investigation aimed to determine the extent to which the introduction of mammography screening in NZ has been associated with initial incidence inflation, and whether incidence has returned to background levels (as expected), or whether any incidence inflation has continued, suggesting over-diagnosis. The analyses here are of annual and biennial trends in breast cancer incidence in relation to screening, for age groups targeted and not targeted by service screening, to be directly comparable with other studies of breast cancer incidence and screening, including those conducted in New South Wales (NSW), Australia,⁵ and South Australia.²⁰

BreastScreen Aotearoa (BSA), the national, publicly funded mammography screening programme, commenced in 1999, initially targeting women aged 50–64. In July 2004, the target age group was extended to 45–49 and 65–69. BSA offers bilateral, two-view mammography screening to age-eligible women without direct charge. All mammograms are independently read by two radiologists, with arbitration of conflicting findings by a third. Recruitment is via self-initiated enrolment or referral from a medical practitioner, then invitation by the programme, with a target of 70% coverage for screening age groups.²¹ The latest biennial coverage for women aged 50–69 screened June 2015–May 2017 is 72%.²²

Methods

Annual female breast cancer registrations by five-year age group were obtained from the NZ Ministry of Health for 1994–2014, using data reported to the NZ Cancer Registry,^2,23 which collects all incident cases of cancer nationally with mandatory pathology notification since 1994. Only incident invasive breast cancers are analysed. Population estimates are based on successive quinquennial censuses (most recent 2013), with intervening annual totals estimated by interpolation, and by projection for post-2013 populations. Mammography screening rates for BSA are obtained from de-identified unit record data made available to the authors by BSA.²⁴ Coverage is expressed as eligible women with one or more mammograms over two years divided by the average annual population (biennial rate).

Annual age-specific breast cancer incidence rates (women aged ⩾45) are calculated by five-year age group. Incidence rates are aggregated by age groups 45–49, 50–64, and 65–69 and by screening regime. Incidence rates for women aged 50–64 are directly age standardized to the 2006 NZ Census female population by five-year age group.²⁵ Confidence intervals (CIs) (95%) of incidence rates are from the normal approximation of the binomial.²⁵Age-specific analysis of breast cancer incidence, aggregated over five years prior to the introduction of screening in 1999 (1994–98), and for five years following the first prevalent screening round after age extension in 2004 (2006–10), is used to assess evidence for over-diagnosis associated with screening mammography, as indicated by incidence inflation persisting in screening target age groups after the establishment of screening without compensating lower incidence in older post-screening age groups. Cumulative risks with 95% CI²⁶ are estimated for women aged 45–84 to compare the overall difference between the two incidence plots. Areas under these incidence curves are also estimated and compared directly, using a simple Riemann sum approximation,²⁷ and also from definite integrals of six-degree polynomials fitted to each plot, each estimated over the 45–⩾85 age range. Two-yearly incidence rates in target populations are used for comparison with corresponding biennial screening coverage in trend analyses. Annual incidence by age groups is compared between the periods prior to and after the introduction of service screening: 1994–98 compared with 1999–2003 when women aged 50–64 were targeted for screening, and 1994–98 compared with 2004–14 following age extension of screening to ages 45–49 and 65–69.

Ethics approval for this study was not required as it relied on de-identified screening data and publicly available aggregate cancer incidence and population data.

Results

In women aged 50–64 initially targeted for service screening during 1999–2003, age-standardized breast cancer incidence increased significantly, by 18%, over 1994–98 to 1999–2003 (p<0.0001) (Table 1, Figure 1), with rates declining thereafter and returning in 2004–14 to levels similar to 1994–98 (Figure 1). Incidence also increased in women aged 45–49 by 21% over 2005–14, after age extension of service screening and when screening participation among women aged 45–49 became extensive. In women aged 65–69, incidence increased by 19% (p<0.0001) in 2004–14 compared with 1994–98. In women aged 70–74, never targeted for screening, breast cancer incidence rates were not significantly different in 1999–2005 from 1994 to 1998, but in 2005–14 were 5% lower than rates in 1994–98.

Table 1.

Breast cancer incidence in relation to service mammography screening in New Zealand 1994–2014.

Age range	Annual breast cancer incidence per 100,000 (95% CI)
Age range	1994–98Before screening	After 50–64 yrs screening introduced (1999)	After 45–49 and 65–69 yrs screening introduced
50–64 yrsaTarget from 1999	261(252–270)	1999–2003	2004–14
		309	274
		(301–319)	(269–279)
		+18.4% > 1994–98	+5.0% > 1994–98
		p<0.0001	n.s. p=0.5848
45–49 yrsIncluded in screening target group from mid-2004	195(184–206)	1999–2004b	2005–14b
		182	236
		(172–191)	(229–244)
		−7% < 1994–98 n.s. p=0.0723	+21.0% > 1994–98p<0.0001
65–69 yrs^cIncluded in screening target group from mid-2004	297(279–306)	1999–2003^c	2004–14^c
		284	352
		(277–301)	(341–364)
		−4.4% < 1994–98 n.s. p=0.3173	+18.5% > 1994–98p<0.0001
⩾45 yrsd	270(264–275)	1999–2004280	2005–14e284
		(275–285)	(280–288)
		+3.7% > 1994–98	+5.2% > 1994–98
		n.s. p=0.0762	n.s. p=0.0662
70–74 yrsNever a target for screening	286(267–305)	1999–2006	2006–14f
		290	262
		(275–306)	(249–274)
		+1.4% > 1994–98 n.s. p=0.7094	−5.0% < 1994–98p=0.0334

CI: confidence interval.

^aAge standardized (direct) breast cancer incidence using three five-year age groups.

^bScreening participation in 45–49 year age group was low until 2005.

^cScreening participation in 65–69 year age group increased with age extension in mid-2004 because participants were already enrolled with the screening service.

^dAge standardized (direct) breast cancer incidence using eight five-year age groups and age group 85+.

^ePeriod following target age extension in 2004 and detection of predominantly prevalent breast cancers in previously un-screened women aged 45–49 and 65–69.

^fA proportion of these women would have received screening mammography previously when in target age group in 1999–2005.

Figure 1.

Incidence of breast cancer in New Zealand and biennial mammography screening coverage: 50–64 yrs, 45–49 yrs, and 65–69 yrs.

Biennial screening in women aged 50–64 rose steadily, from 54% in 1999–2000, to 69% by 2014 (Figure 1). Most of the screening-related breast cancer incidence increase occurred during 1999–2002, followed by a rapid return to pre-screening incidence levels by 2003–4.

With the extension to age 45–69 in mid-2004, screening participation in women aged 45–49 rose from 0.5% in 2003–4 to 36% in 2005–6, and exceeded 70% by 2013–14. Breast cancer incidence in these women correspondingly rose from 195/10 ⁵ in 1994–98 to over 250/10 ⁵ (+30%) by 2011–12 and 2013–14 (see Table 1). Breast cancer incidence in women aged 45–49 has not declined to the levels observed prior to the introduction of service screening for this age group. In women aged 65–69, screening coverage increased from 21% in 2003–4, exceeded 50% by 2005–6, and reached 72–73% by 2009–10 and 2011–12. Breast cancer incidence, correspondingly, increased from below 300/10 ⁵ before 2003–4, by +19% to over 350/10 ⁵ by 2013–14, also without decline to levels observed prior to the introduction of service screening for this age group.

From 2006, when screening was well established with sufficient screening participation and predominantly incident breast cancers detected, age-specific breast cancer incidence for 2006–10 was higher in the screening target age groups than in 1994–98 prior to the establishment of screening, and lower for women aged 70 and older compared with the pre-screening period (Figure 2). These trends show inflated screening-related incidence from cancers detected earlier through screening than they would otherwise have been being compensated for by lower incidence in older age groups, suggesting that little or no over-diagnosis has occurred. The differences in the areas under the incidence curves in women over ages 45 to ⩾85 indicate total incidence differences between populations offered and not offered screening. From the Riemann approximation, the area under the 2006–10 plot was 2335 per 10 ⁵ compared with 2347 per 10 ⁵ for 1994–98 (−0.49% incidence decrease). Using the definite integral of a six-degree polynomial fit to each plot, the corresponding areas were 2323 and 2327, a 0.17% increase. The cumulative risk of breast cancer for women aged 45–84 in 1994–98 was 10.7% (95% CI: 10.4–10.9%) compared with 10.8% (95% CI: 10.5–11.0%) in 2006–10, a small and non-significant increase (the 95% CIs cover the point estimates). Figure 3 shows the birth cohorts in the age group ⩾70 previously exposed to screening and provides empirical evidence for the association of screening with lower incidence in this age group compared with the incidence in the pre-screening period.

Figure 2.

Age-specific breast cancer incidence in New Zealand women 1994–98 and 2006–10.

Figure 3.

Screening coverage by (birth) cohort and period for screening age groups, and previous exposure to service screening in ⩾70 year women in 1994–98 and 2006–10, for the age-specific incidence comparison in Figure 2.

For all women aged 45 and older, breast cancer incidence (age standardized) was not significantly higher in 1999–2005, during the establishment of screening and before age extension, than prior to service screening (1994–98). In 2006–14, following the 2004 age extension, incidence in women aged ⩾45 was not significantly higher than in 1994–98. Despite higher incidence in women aged 45–49 and 65–69 extending for a longer period compared with women aged 50–64, significantly higher breast cancer incidence overall for women aged ⩾45 is not manifest. No sustained rise attributable to screening mammography was observed.

Discussion

The introduction of screening mammography in 1999 for women aged 50–64 coincided with a subsequent 1999–2002 increase in the incidence of invasive breast cancer. When screening became available to women aged 45–49 and 65–69 in 2004, an increase in breast cancer incidence also occurred in these women.

The subsequent fall in incidence rates by 2003 in women aged 50–64, to levels close to those observed in the period before service screening, represents evidence against any substantial over-diagnosis as a feature of organized screening in NZ for these women. This is highlighted by the comparison of age-specific incidence before and after the advent of service screening, showing the expected higher incidence in women aged 50–64, followed by compensatory lower incidence in the older ⩾70 age groups compared with the period before service screening (Figure 2). Even though women aged ⩾70 are not targeted for screening, by 2006–10 some of these women would have been screened when younger (see Figure 3). This represents a virtual textbook example of how screening should affect breast cancer incidence, by detecting cancers earlier than they would otherwise have been detected, but not producing a significant rise in overall incidence, which is confirmed by before-and-after comparisons of cumulative incidence, areas under the age-specific incidence curves, and by lack of difference in age-adjusted incidence in women aged ⩾45 in 2006–14 compared with 1994–98. Further, the lower incidence rates following the establishment of screening in those aged ⩾70, who experienced screening exposure at younger ages, illustrate the population effects of earlier diagnosis of breast cancer.

Evidence of over-diagnosis would manifest as increased incidence in the screening age group persisting across these age groups after the establishment of the screening programme, and also in elevated overall breast cancer incidence, as has occurred in some service screening programmes, such as in NSW Australia,⁵ Norway,¹⁴ and Sweden.¹² However, the quantification of the extent of over-diagnosis is difficult, and estimates can be sensitive to the approach used, especially so in the example of Norway.²⁸

The less immediate increase in breast cancer incidence in women aged 45–49 following the age extension of screening in mid-2004 may be attributed to the need for initial recruitment and enrolment with the screening service of these women, who had not previously been offered service screening. Incidence increased in 2005, coinciding with increased mammography coverage in women aged 45–49. In contrast, many women aged 65–69 had previously been enrolled in mammography screening when aged 60–64 and were already recruited and screened with consequent higher initial coverage than in women aged 45–49, thus manifesting a correspondingly more immediate and pronounced incidence increase in 2004. In NZ, unscreened women are recruited to screening through health providers and the general promotion of screening, while in other countries unscreened women may be recruited through personal invitations sent from the screening programme when they reach the screening target age range. In NZ, once women are recruited and on the BSA screening register, they are sent routine screening invitations and reminders.

Initial increases in incidence coinciding with the introduction of screening are expected, as prevalent cancers in asymptomatic women produce incidence inflation from earlier diagnosis.^11,16 This finding is consistent with incident increases seen in Sweden (45%) and Norway (50%) during the early years of screening.¹² In Spain, incidence increased by 20–30% in target age groups,¹⁸ and in NSW, Australia, the incidence increase was 33%.¹¹ In NZ, initial incidence rates in women aged 50–64 returned to levels slightly above rates before service screening after four years, but not beyond those expected as a result of screening detecting incident cancers before they become clinically manifest. It is expected that after most prevalent cases are detected, breast cancers detected by screening will continue to be diagnosed earlier in predominantly already-screened women, producing incidence rates somewhat above those in populations not offered screening, and that are expected to be compensated for by lower incidence in older women who age into groups no longer offered screening, as observed in NZ.

The observed NZ increases in breast cancer incidence in women aged 45–49 and 65–69 coinciding with screening have not yet declined to levels similar to those prior to the age extension in 2004 (see Figure 1). It is possible that, for women aged 45–49 and 65–69, higher proportions of previously unscreened women with prevalent cancers are being screened over a more extended period. Nevertheless, incidence in these age groups should not be expected to return to pre-screening levels, as cancers are being detected earlier. In the absence of over-diagnosis, raised screening-related incidence in women aged 45–49 would be compensated for by lower incidence in women aged ⩾50, as women aged 45–49 age through the screening target age groups. Similarly, raised incidence in women aged 65–69 should be compensated for by lower incidence in women aged ⩾70, who are out of the screening target age range, as has been observed.

Strengths of this study are that incidence data are sourced from a complete and accurate national cancer registry, and service screening participation rates originate from complete and accurate national aggregations of BSA data. Deficiencies include the lack of information on mammography in the private sector, which may include ad hoc mammography and de facto screening. Anecdotal information indicating that screening mammography is infrequent outside of government service screening is supported by the most recent programme screening participation rates of 72%. Screening participation in NZ is similar to that of the screening trials, and in Australia such screening rates are approached only if private mammography de facto screening is included.¹¹ There is no suggestion of breast cancer incidence increases occurring just prior to the actual introduction of service screening in NZ, coincident with publicity concerning the imminent introduction of screening, that could be attributable to increased private screening, as was documented in NSW, Australia.¹¹ On these grounds, evidence for substantial private mammographic screening in NZ is meagre, although there may have been some private mammography screening before 1999, prior to the introduction of service screening. Although lack of complete cancer registration before 1994 poses some issues of data completeness, baseline incidence was available for five years prior to the introduction of screening in women aged 50–64, where incidence varied between 250 and 275/10 ⁵, and for 10 years prior to the screening age extension to 45–49 and 65–69, for comparison with changes consequent to the introduction of service screening.

Reduction in overall breast cancer incidence following the prevalence peak has been observed in some other mammography screening programmes. The initial breast cancer incidence increase of 20–30% in Spain following the gradual introduction of mammography screening for women aged 45–64 and then 45–69 during the 1990s¹⁸ began to fall in all age groups ⩾45 within five years following achievement of high participation rates. Similarly, in Denmark a pronounced peak in breast cancer incidence in screened age groups after the introduction of screening around 1992 was followed by a decline, to levels compatible with those expected from this population in the absence of screening.¹⁶ In contrast, several countries have experienced a sustained breast cancer incidence increase following the introduction of service screening. Breast cancer incidence in NSW, Australia, rose by 33% between 1992 and 1994 in the mammography screening target age group, coincident with increased mammography coverage.¹¹ Australian data for 2009 indicate that incidence in women aged 50–69 has remained elevated well above levels prior to service screening,²⁹ and incidence in women aged ⩾70 (not a target for screening) has not fallen below the incidence in women aged ⩾70 in the period prior to service screening.⁵ Breast cancer incidence in Sweden rose by 45% during 1986–2000 following the introduction of screening, and in Norway by 54% during 1998–2001 following screening introduction in 1996.¹² No decline in Sweden was observed by 2004,³⁰ and by 2008, incidence in Norway remained elevated in all screening age groups, well above levels observed prior to service screening.³¹

In England, mammography screening was initiated between 1988 and 1993, and breast cancer incidence rates in the initial screening target age group of women aged 50–64 rose sharply over 1989–92, from 200 to nearly 300/10 ⁵,¹⁹ and over 1999–2013 remained at around 300/10 ⁵.³² Incidence in women aged 65–69 also showed a sharp increase in 2002–3, when invitation to screening was extended to this group,¹⁹ with rates remaining significantly elevated by 2013 at around 400/10 ⁵.³² In the Netherlands, screening was introduced between 1989 and 1997, and incidence increased by 7.8% annually between 1989 and 1993 in women aged 50–69. Despite incidence rates falling subsequently, by 2006 incidence remained substantially higher than prior to the introduction of screening.³³

Sustained increased breast cancer incidence following the introduction of population mammography screening suggests over-diagnosis. Estimation of over-diagnosis involves separating the increased incidence due to lead-time effects from earlier pre-clinical detection, from the effects of underlying secular trends in breast cancer risk factors. Estimates of rates of over-diagnosis in mammography screening programmes have varied widely, depending on the programme studied and the methodology used, from negligible to more than 50%.³⁴ A review of the UK screening programme in 2012 estimated over-diagnosis to be, on average, 19%,³⁵ although it may be inappropriate to produce average estimates of screening-related over-diagnosis when such over-diagnosis appears not inevitable, as exemplified by the present study and other reports.^34,36

A factor that may account for differences in over-diagnosis between different screening mammography programmes is the medico-legal environment. In NZ, the no-fault insurance system through the Accident Compensation Corporation may mean that there is less medico-legal pressure on histopathologists. In Australia, the medical compensatory system is fault based, subject to tort liability, and medical practitioners, including pathologists, are required to obtain (mandatory) medical indemnity insurance cover.³⁷ Histopathologists may tend to err on the side of caution to avoid misclassifying cancer as a benign entity and then be subject to legal consequences. There seems little doubt that mammography-related breast cancer over-diagnosis is evident in the Australian and several international service screening programmes. A possible, if unlikely, scenario is that over-diagnosis was already present in clinical practice in NZ prior to the advent of screening, in which case this study has shown no overall increase in breast cancer incidence and over-diagnosis associated with the advent of service screening mammography.

Breast cancer mortality reductions of the order of 30% have been shown to be associated with screening mammography in meta-analyses of randomized controlled trials^38–41 and reviews of service studies.^42,43 Breast cancer mortality reductions of −34% are evident for ever-screened NZ women,⁴⁴ and are 25–34% in aggregate analyses of NZ breast cancer mortality and −17% in incidence-based analyses before and after introduction of service screening.⁴⁵ The present study shows that expected breast cancer mortality reductions in NZ have occurred with the introduction of organized service screening mammography, without evidence of over-diagnosis. This study provides evidence that screening-related over-diagnosis of breast cancer is not inevitable, nor necessary, in successful population mammography screening programmes. Further, breast cancer over-diagnosis thus may be remedied by health service interventions.

Footnotes

Acknowledgement

The authors would like to acknowledge one of the anonymous peer reviewers of this paper whose suggestions improved it considerably.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

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

References

Statistics New Zealand. Quickstats about New Zealand’s population and dwellings. Wellington: Statistics New Zealand, 2016.

New Zealand Ministry of Health. Cancer: new registrations and deaths 2012. Wellington: New Zealand Ministry of Health, 2015.

New Zealand Ministry of Health. New Zealand Cancer Registry (NZCR) New Zealand (updated 2013), Wellington, New Zealand: New Zealand Ministry of Health, 2014.

Brown

SBF

Morrison

Cooke

TG.

Increasing incidence of breast cancer: distinguishing between the effects of birth cohort and a national breast screening programme. Breast Cancer Res Treat 2009; 116: 603–607.

Morrell

Barratt

Irwig

et al . Estimates of overdiagnosis of invasive breast cancer associated with screening mammography. Cancer Causes Control 2010; 21: 275–282.

Pathak

Whittemore

AS.

Combined effects of body size, parity, and menstrual events on breast cancer incidence in seven countries. Am J Epidemiol 1992; 135: 153–168.

Morimoto

White

Chen

et al . Obesity, body size, and risk of postmenopausal breast cancer: the Women’s Health Initiative (United States). Cancer Causes Control 2002; 13: 741–751.

Kelsey

Gammon

John

EM.

Reproductive factors and breast cancer. Epidemiol Rev 1993; 15: 36.

McPherson

Steel

Dixon

Breast cancer – epidemiology, risk factors, and genetics. BMJ 2000; 321: 624–628.

10.

Coombs

Taylor

Wilcken

et al . HRT and breast cancer: impact on population risk and incidence. Eur J Cancer 2005; 41: 1775–1781.

11.

Chiu

Morrell

Page

et al . Population-based mammography screening and breast cancer incidence in New South Wales, Australia. Cancer Causes Control 2006; 17: 153–160.

12.

Zahl

Strand

Maehlen

Incidence of breast cancer in Norway and Sweden during introduction of nationwide screening: prospective cohort study. BMJ 2004; 328: 921–924.

13.

Puliti

Miccinesi

Zappa

et al . Balancing harms and benefits of service mammography screening programs: a cohort study. Breast Cancer Res 2012; 14: R9.

14.

Kalager

Adami

Bretthauer

et al . Overdiagnosis of invasive breast cancer due to mammography screening: results from the Norwegian screening program. Ann Intern Med 2012; 156: 491–499.

15.

Paci

Warwick

Falini

et al . Overdiagnosis in screening: is the increase in breast cancer incidence rates a cause for concern? J Med Screen 2004; 11: 23–27.

16.

Svendsen

Olsen

von Euler-Chelpin

et al . Breast cancer incidence after the introduction of mammography screening: what should be expected? Cancer 2006; 106: 1883–1890.

17.

Jørgensen

Gøtzsche

PC.

Overdiagnosis in publicly organised mammography screening programmes: systematic review of incidence trends. BMJ 2009; 339: b2587.

18.

Pollán

Michelena

Ardanaz

et al . Breast cancer incidence in Spain before, during and after the implementation of screening programmes. Ann Oncol 2010; 21: iii97–iii102.

19.

Duffy

Tabar

Olsen

et al . Absolute numbers of lives saved and overdiagnosis in breast cancer screening, from a randomized trial and from the Breast Screening Programme in England. J Med Screen 2010; 17: 25–30.

20.

Beckmann

Duffy

Lynch

et al . Estimates of over-diagnosis of breast cancer due to population-based mammography screening in South Australia after adjustment for lead time effects. J Med Screen 2015; 22: 127–135.

21.

Group

BAIM.

BreastScreen Aotearoa Monitoring Report No. 1: women screened between January 1 and June 30, 1999. New Zealand: University of Otago, Medicine DoPaS, 2000.

22.

BreastScreen Aotearoa National Screening Unit. Screening participation rates. Data supplied to authors, Wellington, New Zealand: New Zealand Ministry of Health, 2017.

23.

New Zealand Ministry of Health. Cancer data and stats, Wellington, New Zealand: New Zealand Ministry of Health, 2016.

24.

Page

Morrell

Taylor

. Screening and assessment report of women attending BreastScreen Aotearoa: women screened July 2010–June 2012. Independent Monitoring Group Report. Wellington: BreastScreen Aotearoa, 2014.

25.

Armitage

Berry

Statistical methods in medical research. Oxford: Blackwell Scientific Publications, 1987.

26.

Day

Cancer incidence in five continents, vol. V, chap. Cumulative rates and cumulative risk. IARC Scientific Publications No. 88. Lyon: International Agency for Research on Cancer, 1987, pp.787–789.

27.

Anton

Calculus: a new horizon. 6th ed. New York: Wiley, 1999.

28.

Lund

Mode

Waaseth

et al . Overdiagnosis of breast cancer in the Norwegian Breast Cancer Screening Program estimated by the Norwegian Women and Cancer cohort study. BMC Cancer 2013; 13: 614.

29.

Australian Institute of Health and Welfare & Australasian Association of Cancer Registries. Cancer in Australia: an overview. Cancer series no. 74, cat. no. CAN 70. Canberra: AIHW, 2012.

30.

Haukka

Byrnes

Boniol

et al . Trends in breast cancer mortality in Sweden before and after implementation of mammography screening. PloS One 2011; 6: e22422.

31.

Hofvind

Sakshaug

Ursin

et al . Breast cancer incidence trends in Norway – explained by hormone therapy or mammographic screening? Int J Cancer 2012; 130: 2930–2938.

32.

Cancer Research UK. Breast cancer incidence (invasive) statistics, London UK: Cancer Research UK. London UK: Cancer Research UK, 2016.

33.

Otten

JDM

Broeders

MJM

Fracheboud

et al . Impressive time-related influence of the Dutch screening programme on breast cancer incidence and mortality, 1975–2006. Int J Cancer 2008; 123: 1929–1934.

34.

Marmot

Altman

Cameron

et al . The benefits and harms of breast cancer screening: an independent review. Br J Cancer 2013; 108: 2205–2240.

35.

Independent UK Panel on Breast Cancer Screening. The benefits and harms of breast cancer screening: an independent review. Lancet 2012; 380: 1778–1786.

36.

Duffy

Agbaje

Tabar

et al . Overdiagnosis and overtreatment of breast cancer: estimates of overdiagnosis from two trials of mammographic screening for breast cancer. Breast Cancer Res 2005; 7: 258–265.

37.

Armstrong

Tess

Fault versus no fault – reviewing the international evidence. Sydney: Institute of Actuaries of Australia, 2008.

38.

Nystrom

Larsson

Wall

et al . An overview of the Swedish randomised mammography trials: total mortality pattern and the representivity of the study cohorts. J Med Screen 1996; 3: 85–87.

39.

Kerlikowske

Grady

Rubin

et al . Efficacy of screening mammography. A meta-analysis. JAMA 1995; 273: 149–154.

40.

International Agency for Research on Cancer. IARC handbooks of cancer prevention, volume 15. Breast cancer screening. Lyon: IARC, 2016.

41.

Nelson

Cantor

et al . Effectiveness of breast cancer screening: systematic review and meta-analysis to update the 2009 U.S. preventive services task force recommendation. Ann Intern Med 2016; 164: 244–255.

42.

Gabe

Duffy

SW.

Evaluation of service screening mammography in practice: the impact on breast cancer mortality. Ann Oncol 2005; 16: ii153–ii162.

43.

Weedon-Fekjær

Romundstad

Vatten

LJ.

Modern mammography screening and breast cancer mortality: population study. BMJ 2014; 348: g3701.

44.

Morrell

Taylor

Roder

et al . Mammography service screening and breast cancer mortality in New Zealand: a National Cohort Study 1999–2011. Br J Cancer 2017; 116: 828–839.

45.

Accompanying Paper: Taylor et al. JMS-17-092.