Monday, March 7, 2011

The Problems with Evaluating Detection Technologies with Disease Caught in the First Round of Screening

My research is focused on breast cancer screening methods, however, the ideas presented here apply to the detection of most diseases.

Ideally, when we want to evaluate how good a given method is for detecting breast cancer, we monitor a population over a long period of time with the given detection methods. Typically, evaluation involves comparing a particular patient outcome after years of being involved in a clinical trial. The most accepted outcome for a disease that kills people is demonstrating a reduction in the number of deaths in the group being monitored (ie. is there a lower mortality rate?). Other potential methods of evaluation include looking at the size and progression of the disease at the time it is detected.

Typically a study is performed and all of the tumours detected by one method are compared with those detected by another method. However, there is a distinction between detecting a disease in the first round of screening and detecting a disease in a subsequent round (ie. an interval cancer). In the case of the interval cancer, we know that the detection methods employed did not catch the cancer in the previous round of screening. However, when we catch a tumour in the first round of screening, we do not know if screening could have caught that disease earlier had the detection mechanism been introduced earlier from the patient's perspective.

Ideally, comparing screening methods would involve looking only at interval cancers. This is a challenge because screening clinical trials tend to catch relatively few interval cancers compared with those caught in the first round. It requires a much longer and inevitably more expensive trial to accumulate many examples of interval cancers.

Incidentally, exciting results have recently been published indicating that adding breast MRI based detection to a screening program results in very few women succumbing to breast cancer. This conclusion is reached while still including cancers caught in the first round of screening.

Jacob Levman, PhD