Extending the benefits of breast cancer screening

Ever since the implementation of the NHS breast screening programme in 1988 two important questions have been consistently raised. Should the age range of women invited be extended from the current range of 50-64 years, and should the screening interval be reduced from the current three years? If we are to believe the cost effectiveness analysis by Boer et al in this week’s issue (p 376),¹ an increase in age to 69 and a two year interval would each generate substantial benefits in life years saved and deaths averted—but, needless to say, at a substantial cost. Moreover, the authors’ conclusion that extending the age range is expected to prevent more deaths, whereas shortening the screening interval would save more life years leaves policymakers with a—not unfamiliar—ethical dilemma.

It is obviously desirable to improve life expectancy in those women already eligible for screening. Reducing the interval cancer rate, which is particularly high in the third year of the screening interval, may prove crucial.²,³ Shortening the screening interval from three to two years will decrease the interval cancer rate as a proportion of the underlying incidence by 30%.³ Extending age range to 69 years, however, exploits the fact that age is by far the most important risk factor for breast cancer.⁴ Would it be equitable to deny older women the benefits of more effective routine breast screening for the sake of increasing the life expectancy of younger women (in whom breast cancer is much less common)? This is a hard choice to make. Unfortunately, the authors have not modelled the effect of the combined implementation of both strategies, which would have provided an assessment of how necessary it really is to pursue only one of these two options.

The epidemiological model applied here depends to a large extent on the comparison of tumour stage distributions before and after the introduction of a specified screening policy.¹,⁵ This may have led the authors into overestimating the cost effectiveness of the suggested alternatives by modelling each policy—that is, existing policy and suggested alternative—against a baseline in which no screening has been introduced into a population and then subtracting both results from each other. The breast screening programme has in fact been running for over 10 years and the tumour stage distribution in the population has changed as a result,⁶ so that using prescreening distributions may not be entirely valid when evaluating policy changes at this stage. To ascertain true marginal costs, we would need a validated modelling exercise which uses the current situation in the United Kingdom as a baseline.

The efficiency of extending the age range of breast screening will crucially depend on the acceptance rate among women aged 65 and above. Rubin et al’s preliminary report shows that more than 70% of these women took up their invitation (p 388).⁷ This is a much better uptake than reported from previous studies in Nijmegen⁸ and London,⁹ which achieved participation rates of 55% and 37% respectively. In London, however, this participation rate reflected overall attendance rates, and it was concluded that older women would potentially have attendance rates at routine screening similar to younger women if they were invited in the same way. If the results of Rubin et al are representative they might show how (old) peoples’ attitudes have changed as interventions like breast screening have become more accepted in such communities since the Forrest recommendations.¹⁰

Rubin et al do, however, report much higher cancer detection rates than expected. They suggest that the particularly high cancer detection rate in women aged 68 and 69 reflects both advancing age and not having been screened for 6 years.⁷ This fact cannot fully explain the result, however, since cancer detection rates were unexpectedly high in all age groups. Also, the Nijmegen study reported a detection rate of only 5.6/1000 in all women aged 65-69. This may suggest a high proportion of false positive screens, and further data on assessment and biopsies are required.

Both studies make an important contribution to the discussion about extending age range or shortening screening interval in the NHS breast screening programme, thereby departing from the recommendations of the Forrest report.¹⁰ Nevertheless, their findings may be of only limited validity because the first study does not use the current UK situation as its baseline whereas the second study, despite its encouraging result, may have identified a quality problem. Before either of these changes are implemented, resource implications and potential opportunity costs warrant much further discussion and analysis. Breast cancer screening is by no means the best way of obtaining health benefit per billion pounds. Indeed, current programmes have yet to show any unequivocal benefit¹¹ in terms of either mortality or of life years. But it is still early days.

Papers pp 376, 388