ReviewEffects of study methods and biases on estimates of invasive breast cancer overdetection with mammography screening: a systematic review
Introduction
Randomised trials of mammography screening (figure 1) have shown reductions in mortality from breast cancer,1 but there are downsides to the process, including overdetection of breast cancer. Overdetection is the detection of cancer that would not have presented clinically during a woman's lifetime (and therefore would not be diagnosed in the absence of screening). Several studies have tried to quantify overdetection of invasive breast cancer with mammography screening and estimates vary widely. We need valid and precise estimates of the magnitude of overdetection attributable to mammography screening to inform public-health policy, clinical practice, and decision-making for women participating in screening. Although overdetection of both invasive cancer and ductal carcinoma in situ are possible, in this paper we focus on invasive breast cancer.
We did a systematic review of all papers with estimates of overdetection of invasive breast cancer in mammography screening to identify how, in what direction, and to what extent different study methods and biases might affect estimates of overdetection.
Section snippets
Methods to estimate overdetection of invasive breast cancer by mammography screening
There are three broad approaches to the estimation of overdetection: the incidence-rate method, the cumulative-incidence method, and modelling approaches.
The first two designs can be conceptualised as follows. Imagine a group of women who start screening and continue being screened biennially for 20 years, and a second group of women of the same age who are not screened over the same time period. Ideally, these groups would be formed by randomisation to ensure similarity of breast-cancer risk.
Search strategy and selection criteria
Primary research and review articles that attempted to estimate overdetection of invasive breast cancer by mammography screening published in English in peer-reviewed journals were eligible for inclusion in the systematic review. We restricted the review to papers that provided estimates of breast cancer incidence in screened and unscreened groups and from which we were able to assess whether, and to what extent, the results might have been affected by bias.
We searched MEDLINE from 1980 to Dec
Results
We included two reviews and six primary studies in our systematic review (figure 4). The two reviews provided estimates of overdetection using data from all of the eight3 or six of the eight4 published randomised controlled trials of mammography screening (table 2). Of the six primary studies, two studies reported two estimates each based on different data sources9 or different time periods.10 The remaining four primary studies reported one estimate of overdetection each, so eight estimates of
Discussion
We have identified and described four important biases that might affect estimates of overdetection of invasive breast cancer substantially. These are underlying breast cancer risk being different in screened and unscreened populations, less than 100% participation in screening in the screened population and occurrence of screening in the unscreened population, screening in the control group after the end of the intervention (screening) period, and inadequate allowance for lead-time (table 1).
References (17)
IARC Handbook of Cancer Prevention
(2002)- et al.
Estimation of mean sojourn time in breast cancer screening using a Markov chain model of both entry to and exit from the preclinical detectable phase
Stat Med
(1995) Overdiagnosis and overtreatment of breast cancer: overdiagnosis in randomised controlled trials of breast cancer screening
Breast Cancer Res
(2005)On the benefits and harms of screening for breast cancer
Int J Epidemiol
(2004)- et al.
Rate of over-diagnosis of breast cancer 15 years after end of Malmo mammographic screening trial: follow-up study
BMJ
(2006) - et al.
Evaluation of overdiagnosis of breast cancer in screening with mammography: results of the Nijmegen programme
Int J Epidemiol
(1989) - et al.
Overdiagnosis in screening: is the increase in breast cancer incidence rates a cause for concern?
J Med Screen
(2004) - et al.
Estimate of overdiagnosis of breast cancer due to mammography after adjustment for lead time: a service screening study in Italy
Breast Cancer Res
(2006)
Cited by (157)
A novel methodological framework was described for detecting and quantifying overdiagnosis
2022, Journal of Clinical EpidemiologyCitation Excerpt :We may consider these people overdiagnosed as per a lack of clinical utility provided by diagnosis (Overdiagnosis: Utility Box in Fig. 1) [12]. Methods developed to quantify cancer overdiagnosis from screening of asymptomatic people aim to estimate the excess cumulative cancer incidence in a screened group compared to a nonscreened group with long-term follow-up to capture subsequent symptomatic disease [13–16]. These “excess incidence” methods assume that in the absence of overdiagnosis, the incidence of cancer in the screened and nonscreened groups will be the same, once the effect of lead-time is accounted for (i.e., an earlier increase in incidence in the screened group relative to the nonscreened group is offset by a later decrease in incidence-representing cancers that would have presented clinically but were diagnosed earlier because of screening).
Addressing harms of screening – A review of outcomes in Cochrane reviews and suggestions for next steps
2021, Journal of Clinical EpidemiologyCNR performance of semiconductor materials for X-ray imaging of breast calcifications
2023, Journal of Instrumentation