The benefits and harms of breast cancer screening in Australia

Abstract

The introduction of screening mammography in Australia has been associated with an increase in the incidence of early-stage breast cancer. Concern is growing about the problems caused when women are diagnosed with breast cancer and undergo treatments that do not benefit them because their cancer would not progress, or would progress, but would not become symptomatic within the remaining lifetime of the patient. This is known as overdiagnosis. Screening mammography aims to decrease breast cancer morbidity and mortality by advancing in time the diagnosis and thereby increasing the chance of successful treatment. But overdiagnosis and the consequent overtreatment can cause serious lifelong harm and are therefore considered the major downsides of breast screening. Mounting evidence of the extent of overdiagnosis has led to the recognition that the benefits and harms of breast screening are finely balanced, and women need to know the magnitude of the trade-offs. The extent of overdiagnosis due to breast screening is contested, with published estimates ranging from 1% to 57%. There is a critical need for research investigating the harm to benefit ratio in Australia and quantification of the effects of screening mammography on the incidence of stage-specific breast cancer and overdiagnosis. Individual women require information about the impact of regularly attending screening mammography on breast cancer mortality and overdiagnosis to make informed decisions. A challenge to estimating this in a randomised controlled trial is nonadherence to the trial protocol. Previous systematic reviews have provided estimates of the effect of receiving an invitation to screening on the risk of dying due to breast cancer. Chapter 2 presents a meta-analysis of the screening mammography trials using a simple adjustment that estimates the probability of a reduction in breast cancer mortality and risk of overdiagnosis due to the effect of receiving screening by regularly participating in a breast screening program. Adjustment for nonadherence increased the size of the size of the effect by up to 50%. The prevented fraction of breast cancer mortality at 13-year follow-up increased from 0.22 to 0.30 with deattenuation. The percentage risk of overdiagnosis during the screening period in women invited to screening increased from 19.0% to 29.7%. From 2013 through 2017, the Australian national breast cancer screening programme gradually invited women aged 70 to 74 years to attend screening, following a policy decision to extend invitations to older women. Yet no formal evaluation of the effects of the change in policy on outcomes for women was undertaken. Building on my meta-analysis, in Chapter 3 I used a Markov model and applied the breast cancer mortality reduction and overdiagnosis estimates reported in Chapter 2 to Australian breast cancer incidence and mortality data to estimate the benefits and harms of the new package of biennial screening from age 50 to 74 compared with the previous programme of screening from age 50 to 69. I found that the extra five years of screening results in approximately seven more overdiagnosed cancers to avert one more breast cancer death. Thus extending screening mammography in Australia to older women results in a less favourable harm to benefit ratio than stopping at age 69. To identify temporal trends in stage-specific breast cancer in Australia, I used an observational study design to analyse data on women who received a diagnosis of breast cancer from 1972 to 2012 as recorded in the New South Wales Cancer Registry (Chapter 4). I explored trends in stage-specific incidence before screening and compared them to periods after screening began. I found that screening was not associated with lower incidence of late-stage breast cancer at diagnosis and incidence for all stages remained higher than prescreening levels. In women aged 50 to 69 years, the incidence of carcinoma in situ, localised and regional breast cancer has more than doubled compared to the prescreening era. The data presented in Chapter 4 indicate that excess detection of breast cancer is a problem in New South Wales. Thus I designed an ecological study to quantify overdiagnosis. I estimated the background trend of increasing incidence using two approaches, the first based on the prescreening trend in women of screening age (50 years and older), and the second based on the contemporary trend in women too young to be screened (40 to 44 years of age). From these trends, I estimated the expected age-standardised incidence of breast cancer, by stage at diagnosis, in the absence of screening, for women aged 50 years and over in the years since the national screening mammography program, BreastScreen, was introduced (1988 to 2012). I then calculated the difference in observed and expected incidence rates to determine the excess incidence of early-stage breast cancer and reduction in the incidence of late-stage disease. I found that screening mammography has resulted in overdiagnosis of early-stage and regional breast cancer. I estimate that around six additional cases of early and regional breast cancer are detected for every distant metastatic breast cancer prevented. Due to the substantial increase in detection of carcinoma in situ of the breast observed in Chapter 4, I explored sub-types and causes of this. Chapter 6 presents a descriptive analysis of temporal trends in the incidence of ductal carcinoma in situ (DCIS) and lobular carcinoma in situ (LCIS) in women who received a diagnosis from 1972 to 2012, recorded in the NSW Cancer Registry. Carcinoma in situ as a proportion of all breast cancer increased dramatically, and incidence of DCIS across all ages rose from 0.15 per 100,000 during 1972 to 1983 to 16.81 per 100,000 over 2006 to 2012, with the greatest increase seen among women in the target age group for screening (50 to 69 years). DCIS incidence has not stabilized despite screening being well established for over 20 years, and participation rates in the target age range remaining stable. Our observational estimate of overdiagnosis from Chapter 5 is different to those from meta-analyses of randomised controlled trials of screening mammography and some other observational studies. Thus the final paper in this thesis presents a Cochrane Protocol on Overdiagnosis due to screening mammography for women aged 40 years and older (Chapter 7). We present methodology to identify and evaluate all primary epidemiological studies that have quantified overdiagnosis resulting from screening mammography and provide estimates of its frequency (including randomised and observational studies). An approach to assessing the risk of bias due to lead time is also discussed. Trying to quantify the benefits and harms of screening mammography and present them to women using natural frequencies should better enable informed choice, and is consistent with the international shift towards promoting and supporting shared decision making for screening. This thesis provides valuable new evidence about the trade-offs of screening mammography in Australia. The finding that risk of harm is greater than the chance of benefit is consistent with international studies of breast screening and demonstrates the importance of continuing this work to better quantify overdiagnosis for women, clinicians, and policymakers.