Another twist in the controversial debate surrounding the value of mammography screening for breast cancer is presented in a fast-track research letter and on THE LANCET’s website this week.
The publication of a meta-analysis by Danish investigators Olsen and Gotzsche (Lancet 2000; 355: 129-34) and a follow-up analysis (Lancet 2001; 358:1340-42), concluded that there was no evidence that mammography screening saves lives (based on data from the two ‘reliable’ randomised trials done in Sweden and Canada). Publication of both papers resulted in extensive media coverage, notably in the USA where front-page news, editorial, and correspondence appeared in the New York Times in December 2001. Although there remains considerable public confusion over mammography screening, last week the US-based PDQ Screening and Prevention Editorial Board concluded that there was insufficient evidence to show that mammograms prevented breast-cancer deaths, and said it would rewrite its previous statements to reflect this view.
A new perspective on the mammography debate is detailed by Claudia Henschke and colleagues from CornellMedical Center, New York, USA. They reviewed the Swedish trial previously reviewed by Gotzsche and Olsen, and conclude that reduction in breast-cancer mortality is apparent after sufficient time has elapsed-around seven years-for the influence of successful screening and treatment to take effect. They state that the reduction in breast-cancer mortality is substantial (55%) for women aged 55 years or older at entry to the study who were followed up 8-11 years after screening; the corresponding risk reduction among women aged 45-54 years at study entry was estimated at 30%.
Claudia Henschke comments: "It is critical to understand two deceptively simple but fundamental facts which have been overlooked in both the original meta-analysis and subsequent commentary. Screening does not have an immediate effect; the deaths that get to be prevented by screening are in the future, years away. Thus the analysis of the deaths must focus on the relevant measure, case-fatality rate rather than the overall mortality rate and it must focus on this measure during the relevant time interval during which this benefit is achieved, that is, at some point sufficiently distant from the onset of screening. Therefore, the screening itself must continue for a sufficiently long time to realize its benefit, namely the reduction in case-fatality rate.” (Quote by e-mail; does not appear in published paper).
* A condensed version of the study by Henschke and colleagues appears as a research letter in this week’s issue of THE LANCET (p 404); a full version of the study is available from The Lancet website (http://www.thelancet.com).
** See also News (p 409) and Correspondence (pp 439-42).
Contact: Jonathan Weil, Department of Public Affairs, Weill Medical College of Cornell University, 1300 York Avenue, New York, NY 10021, USA; T) +1 212 821 0566; F) +1 212 821 0576; E) (Claudia Henschke) chensch@med.cornell.edu
Journal
The Lancet