Elsevier

Preventive Medicine

Volume 53, Issue 3, 1 September 2011, Pages 131-133
Preventive Medicine

Overdiagnosis in breast cancer: Design and methods of estimation in observational studies

https://doi.org/10.1016/j.ypmed.2011.05.012Get rights and content

Abstract

Objectives

In recent years observational epidemiological studies have been used to estimate overdiagnosis in breast cancer screening. These estimates vary widely. In this paper we present some of the methodological issues which explain the large variability of the reported findings.

Methods

Different types of observational studies were identified according to study design, definition of the population, adjustment for breast cancer risk and adjustment for lead time.

Results

The majority of observational studies that have estimated breast cancer overdiagnosis have analyzed temporal trends or geographical differences in breast cancer incidence. Estimates of overdiagnosis in a dynamic population vary widely, from 4% to 52%. Only a few studies have used the cohort approach and they found estimates varying from 1% to 5%.

Conclusions

The cohort approach is preferable to the analysis of a dynamic population because it allows the follow-up of a group of women who have had the opportunity for screening and evaluates if there is sufficient follow-up after the last screen.

Section snippets

Background

Overdiagnosis of breast cancer is primarily an epidemiological construct since it can be assessed only at population level. On the other hand the woman's decision to attend or not to attend a screening program is an individual choice. Therefore, the benefits and harms of screening, known through epidemiological studies, should be communicated in such a way that each woman can decide based on valid and reliable information. Indeed, it is currently impossible to identify breast neoplasms that

Methodological framework for the evaluation

Estimates of overdiagnosis obtained from observational studies vary widely according to the methodology used. The following methodological issues should be considered:

Results

The majority of observational studies that have estimated breast cancer overdiagnosis have analyzed temporal trends or geographical differences in breast cancer incidence. Estimates of overdiagnosis obtained using the compensatory drop method in a dynamic population vary from the 4% reported by Duffy et al. (2010) to the 52% estimated by Jorgensen and Gotzsche (2009). Similarly, estimates from studies using adjustment by statistical method vary widely (Jonsson et al., 2005, Paci et al., 2006).

Discussion

The cohort approach is preferable to the analysis of a dynamic population because it allows the follow-up of a group of women who have had the opportunity for screening and evaluates if there is sufficient follow-up after the last screen. A deficit in incidence above the age limits for screening can only occur in cohorts that have been through the screening program. We can distinguish two types of cohorts: birth cohort and cohort by enrolment. One example of birth cohort is the cohort of women

Conclusions

The methodological requirements to correctly estimate overdiagnosis is achieved only by those randomized screening trials which did not offer screening to the control group and reported incidence for more than 5 years afterwards. However, these estimates refer to many years ago. In order to better understand how changes in screening intensity and technological advances modify the risk of overdiagnosis, it is advisable to monitor service screening programs through a cohort approach.

Such

Conflict of interest statement

The authors declare there is no conflict of interests.

References (13)

There are more references available in the full text version of this article.

Cited by (0)

View full text