Two screening test stories – both of which discuss the work of the US Preventive Services Task Force – that deserve much more analysis than I can afford to give right now. This is a casualty of our current funding plight. As it is, in the first four hours that I’ve been at the computer today, I’ve received at least five requests from readers to review/analyze/comment on things they’ve sent me. I can’t get to it all. *
McPherson told the BMJ, Screen detected cancers are necessarily slower on average than non-screening detected cancers because they thus stand a better chance of being detected at any one mammogram in their (longer) pre-clinical period. That is precisely why screening is evaluated with randomised controlled trials, being the only way to eliminate these biases in the relevant comparison. These [failure] analyses are subject to serious unknown biasin spite of their obvious appeal.
Karsten Jørgensen, a researcher at the Nordic Cochrane Centre in Copenhagen, told the BMJ, We already know that breast cancer deaths are more common in non-attenders to breast screening. We also know that much of this difference is due to selection bias and are not an effect of screeninga factor this study does not explore. Health-concerned women in high socio-economic groups with a strong network are more compliant with treatment and are also more likely to attend screening. This is what makes case-control studies problematic in breast screening.
He added, This study also disregards that overdiagnosis of invasive breast cancers in screened women will substantially affect their analyses. The largest reductions in breast cancer mortality, 30-50%, have been seen among women too young to be screened and are a major achievement of modern breast cancer treatment. This has substantially reduced the potential absolute benefit of breast screening for younger women, a group where older randomised trials already showed a small effect, probably because these women often have fast growing tumours, which are more difficult to catch with screening (length bias).
* In the past, with foundation funding, we would have been all over these stories with independent, expert analysis. I could have turned to a terrific team of reviewers. Currently I am keeping this site going by myself and cannot devote the time to the kind of analysis that’s required. Rather than post a half-hearted analysis, I simply bring these to your attention and await the day when we can return to our former robust reporting on such issues.
Follow us on Twitter: