Intended for healthcare professionals

Observations Yankee Doodling

How to decrease overtreatment in cancer

BMJ 2013; 347 doi: (Published 14 August 2013) Cite this as: BMJ 2013;347:f5071
  1. Douglas Kamerow, chief scientist, RTI International, and associate editor, BMJ
  1. dkamerow{at}

Screen more selectively, do different research, and change our vocabulary

In the US and in other “overdeveloped” nations, we excel at finding and treating medical problems, at least in some members of our populations. Among the results of these capabilities are overdiagnosis and overtreatment—finding diseases that don’t need to be treated and providing treatments that lead to more harm than good. I first understood this to be a generalized problem after reading Shannon Brownlee’s Overtreated in 2007,1 but the BMJ had published a theme issue on “Too Much Medicine” in 2002.2 The BMJ and others are now sponsoring a conference on this topic3 and have launched a campaign “to highlight the threat to human health posed by overdiagnosis and the waste of resources on unnecessary care.”4

As with threats to patient safety, we need to move beyond the simple recognition and documentation of overtreatment to find ways to prevent it. Cancer screening tests may provide a good opportunity. A group of experts convened by the US National Cancer Institute (NCI) recently published a commentary in which they presented a useful perspective on evaluating cancer screening tests for their overdiagnosis potential as well as suggestions for improvement.5

As the commentary points out, the goal of cancer screening is to increase the detection of consequential disease, which more than likely leads to metastasis and death, while minimizing the discovery of indolent cancers, which will not progress and threaten life. These outcomes involve three factors: the screening test used, our treatment capabilities, and the biological nature of the cancers. Cancers that are fast growing are rarely good for screening. Slow growing tumors that eventually metastasize can be successfully screened for, especially if they have diagnosable precancerous stages, such as in colon and cervical cancer. They are to be distinguished from slow growing, indolent tumors, which are unlikely to threaten life but can be diagnosed early, leading to overtreatment, preventable side effects, and no improvement in clinical outcome.

To document this, the NCI experts presented incidence and mortality data over time for cancers exemplifying three groups. The first group, in which screening has reduced both incidence and mortality, includes cervical and colorectal cancers, which are ideal for screening. The second group, including breast and prostate cancers, has a mix of consequential and indolent cell types; screening for them has resulted in a decrease in mortality but an increase in incidence, as more and more indolent tumors are diagnosed with the consequential ones. The third group, including thyroid cancers and melanoma, represent the worst of both worlds: screening has led to the identification of more indolent tumors (and thus increased incidence) without affecting the small number of lethal cancers (leading to no improvement in mortality).

What can we do about all this?

One conceptually simple recommendation from the expert panel, which got the most press attention, was to literally change the vocabulary of cancer screening and diagnosis: remove the word “cancer” from the names of indolent lesions. When a doctor or a patient sees a diagnosis of “ductal carcinoma in situ” (DCIS) from a breast biopsy it is very hard to resist immediate treatment, even though these cancers behave very differently from invasive tumors. Before routine mammography screening, DCIS was uncommonly reported. Now it a major driver behind the 25% increase in breast cancer incidence in the past 35 years. By renaming DCIS and other premalignant lesions with low potential for progression, doctors and patients might feel less compelled to treat them immediately and aggressively.

A second recommendation is to step up research to identify and validate tools to accurately diagnose indolent or low risk tumors so they can be segregated and treated differently. Cancer doctors have come up with the lovely almost-acronym of IDLE (indolent lesions of epithelial origin) for such tumors. Thirdly, we need extensive registers to be able to follow up all of these newly categorized tumors and their treatment and outcomes in order to inform future care.

All of this needs to be done in the context of improved clinical strategies to help mitigate overdiagnosis, by reducing screening test frequency, focusing on high risk populations, and expanding “watchful waiting” rather than recall and biopsy as the strategy of choice. Many of these changes require further research, but some can begin right away.

One new opportunity to test these principles has arisen with the publication of a major trial of low dose, computed tomography screening for lung cancer6 and new draft recommendations for such screening from the US Preventive Services Task Force.7 Their screening recommendation is clearly and exclusively targeted at the most high risk patients: healthy 55-79 year olds with ≥30 pack years of smoking history. Careful tracking of this potentially powerful but costly (in terms of dollars and false positive results) strategy should help assess its effectiveness while preventing overdiagnosis and overtreatment.

Overtreatment is about much more than cancer screening, but changing the way we define cancer and improving our understanding of and strategies for screening is one way we could make an important difference in clinical outcomes and costs.


Cite this as: BMJ 2013;347:f5071


  • Competing interest: DK is the author of Dissecting American Health Care.

  • Provenance and peer review: Commissioned, not externally peer reviewed.


View Abstract