Evaluation of the NHS Breast Screening Programme: Results from the London pilot study


Year:

Session type:

Nathalie Massat1, Amanda Dibden1, Dharmishta Parmar1
1Queen Mary, University of London, London, UK

Background

Breast cancer is the most commonly diagnosed cancer among women in the UK, and the second most common cause of cancer death. In England, a national breast screening programme, the NHSBSP, has been in place since 1988, and assessment of the impact of the programme using observational studies is needed.

The Policy Research Unit in Cancer Awareness, Screening and Early Diagnosis has a remit to address the effects of the current programme on mortality from breast cancer, and the case-control design of the cervical screening programme evaluation was highly effective in informing policy (Sasieni et al., 2009).

Method

We are currently implementing a retrospective case-control evaluation of the NHSBSP to cover the whole of England. The major principle of the case-control evaluation is that, if the breast screening is preventing mortality from breast cancer, women who have died of primary breast cancer (cases) will be characterised by a lesser screening history than women who have not (controls). We therefore select cases and age- and area- matched general population controls and compare their screening histories prior to the cases' date of first diagnosis. A pilot phase will initially be conducted for the 2008-2009 period in London which will include a minimum of 900 cases and 1800 controls.

Results

Assessing the impact of various measures of screening exposure will be of interest, in particular time since last screen, and whether that time span fell within the past three years, as this corresponds to the NHSBSP protocol and approximates the estimated preclinical screen-detectable period. Pilot data illustrating some of the principles will be presented.

Conclusion

Biases will be addressed at the analysis stage, in particular self-selection bias (Duffy et al., 2002) and exposure (screening) opportunity bias (Walter, 2003, Duffy et al., 2008). Other potential biases and methodological issues will be discussed.