Over-diagnosis and breast cancer screening: a case study

Women need to be made aware of over-diagnosis and given enough information to make up their own minds about screening. Johan/Flickr

OVER-DIAGNOSIS EPIDEMIC – Today Robin Bell and Robert Burton examine breast cancer to evaluate the role of population-wide screening in over-diagnosis.

Since the national screening mammography program (Breastscreen) began in Australia in 1991, mortality from breast cancer has declined by 28%. We wanted to know how much of this improvement was due to Breastscreen and how much was due to advances in breast cancer treatment.

This is an important question to ask in light of the recent passionate debate about the benefits and harms of mammographic screening in medical literature as well as in the lay press. We published an analysis that addressed this issue using three different approaches.

Who avoids breast cancer death?

We started by comparing the relative reductions in breast cancer-specific mortality between 1991 and 2007 (the latest data year available) in women across different age groups. We know women aged 50 to 69 years were invited to screen, so their uptake of mammographic screening should have been higher than other age groups. So if screening was important, this is where we would expect to see the most impact.

But what we found was that the greatest relative reduction in breast cancer mortality (44%) occurred in the youngest age group. These women (aged 40 to 49 years) are not invited for screening. In contrast, women aged 60 to 69 years, who are invited to screen, had the smallest relative reduction in mortality (19%).

Given that three times as many women aged 60 to 69 (about 60%) participated in Breastscreen (compared to 20% of women aged 40 to 49 years), our finding is not consistent with screening having a major impact on the reduction in breast cancer mortality since 1991.

When was mortality reduced?

We then considered when Breastscreen could have had an impact on breast cancer mortality between 1991 and 2007.

The proportion of invited women who attended for screening didn’t exceed 50% until 1996. And we know from randomised trials of screening that the impact of screening on mortality was not seen for four to six years after the commencement of the trials (they had an even higher level of participation of 70% and more).

So we compared the reductions in mortality that occurred in two time periods – (1) between 1991 and 1999 when we wouldn’t have expected Breastscreen to have an impact and (2) between 2000 and 2007 when the impact of Breastscreen would’ve been expected.

For women invited for screening (50 to 69 years), most of the reduction in mortality that occurred between 1991 and 2007 took place prior 2000. The reduction was before the impact of screening could be expected.

Only about a third of the total 28% reduction in breast cancer mortality seen in this age-group over the full time period occurred after 1999 (9%). This finding is also inconsistent with mammographic screening having a major impact on the observed reduction in breast cancer mortality since 1991.

Improved treatment

Finally, we used a number of different data sources to estimate the likely impact on breast cancer mortality of changes in the use of adjuvant chemo - and endocrine therapy over the same time period. We obtained published data from the Early Breast Cancer Trialists’ Collaborative Group (EBCTCG) to provide the estimated impact of regimens of chemo and endocrine therapy on women in different age groups with early breast cancer.

We used this data, together with data from a population-based survey of women treated for early breast cancer in Victoria in 1999, to calculate the impact the use of adjuvant chemo- and endocrine therapy could have had on breast cancer mortality in these women. We found that the adjuvant therapy Victorian women received in 1999 could have produced a mortality reduction of about 26% in women in the age group invited to screen.

Diagnosis comes with its own worries and stresses. Ronny-André Bendiksen

Indeed, improvements in the treatment of breast cancer could explain almost all of the reduction in breast cancer mortality that has occurred in women invited to screen since 1991.

We estimate mammographic screening is responsible for a minority of the reduction in breast cancer-specific mortality seen in Australia since 1991 and that advances in adjuvant therapies adequately explain the majority of the observed mortality reduction.

Breast cancer and over-diagnosis

A modest reduction in mortality as a result of mammographic screening would seem acceptable if it were not for the issue of over-diagnosis. Over-diagnosis is the diagnosis and treatment of breast cancer that would have never have made a woman ill in her lifetime – she would never have known about it if she had not been screened.

The extent of the problem of breast cancer over-diagnosis has been estimated in the Cochrane review of screening mammography at 30%, based on the randomised trials. In Australia, based on increases in new case rates, over-diagnosis was estimated at 30% to 40% in the women aged 50 to 69 years invited to screen in New South Wales in 2001 and 2002.

Putting their estimates of the relative mortality reduction in breast cancer (15%) and over-diagnosis (30%) together to calculate the balance of benefits versus harms, the Cochrane review summarised:

“… for every 2000 women invited for screening throughout 10 years, one will have her life prolonged. In addition, 10 healthy women, who would not have been diagnosed if there had not been screening, will be diagnosed as breast cancer patients and will be treated unnecessarily.”

Our estimate of the likely breast cancer-specific mortality reduction from screening in women aged 50 to 69 years (9%) and the New South Wales estimates of 30% to 40% over-diagnosis in this age group are similar to the Cochrane estimates.

Based on these Australian estimates and the incidence and mortality of breast cancer in women aged 50 to 69 years in 2007, we know that for every death from breast cancer prevented, 15 healthy women, who wouldn’t have been diagnosed if they hadn’t been screened, will be diagnosed as having invasive breast cancer and will be treated unnecessarily.

Although there’s considerable argument about the exact extent of over-diagnosis, most estimates find the balance of lives saved to cases over-diagnosed and treated unnecessarily unfavourable, with more women over-diagnosed than saved.

We believe that the decision to invite women for screening should be reviewed with the aim of ceasing the standard invitation for screening. As women aged 50 to 69 years have been invited for screening for 20 years now, a change in policy may take some time to work through.

In the meantime, women need to be made aware of the issue of over-diagnosis and the information provided to those invited for mammographic screening needs to be better balanced between benefits and harms so they can make up their own minds about whether to be screened.

Have you or someone you know been over-diagnosed? Share your story below or email the series editor.

This is part two of our series on over-diagnosis, click on the links below to read other articles:

Part one: Preventing over-diagnosis: how to stop harming the healthy

Part three: The perils of pre-diseases: forgetfulness, mild cognitive impairment and pre-dementia

Part four: How genetic testing is swelling the ranks of the ‘worried well’

Part five: PSA screening and prostate cancer over-diagnosis

Part six: Over-diagnosis: the view from inside primary care

Part seven: Moving the diagnostic goalposts: medicalising ADHD

Part eight: The ethics of over-diagnosis: risk and responsibility in medicine

Part nine: Ending over-diagnosis: how to help without harming