An obligatory post

Like it or not, I am required to comment on the study in the new NEJM on screening mammography in Norway. It is difficult to accurately estimate the effect of a large-scale effort like population screening because the before and after comparison is unavoidably "contaminated," as the statisticians say, by other changes over time. Specifically, it is very tricky to decide whether changes in breast cancer mortality are associated with screening, or better outcomes due to continual improvements in treatment.

Norway provided a natural experiment in that population based screening was rolled out region by region, creating side-by-side comparison groups that were temporally matched. It also has universal health care, including universal access to high quality, multidisciplinary breast cancer treatment, and a 100% breast cancer registry, so Bob's your uncle.

What the investigators found is that screening did indeed reduce breast cancer mortality for women age 50 to 69. But the difference, after more than 8 years of follow up, was 2.4 deaths per 100,000 person years. Most of the reduction in the death rate that occurred during the study period was the result of improved treatment. Indeed, women older than 69, who were not offered screening at all, had a comparable reduction in deaths from breast cancer.

In a companion essay, H. Gilbert Welch explains that if screening mammography results in a 10% reduction in the risk of death from breast cancer -- an optimistic assumption -- the difference between offering universal screening and no screening is .4 deaths from breast cancer per 1,000 women. Put another way, you would need to screen 2,500 women for 10 years to prevent one death. (Of course that woman would still have died from something else eventually, perhaps not much later.) Among the other 2,499 women, more than 1,000 would have a false positive result, and somewhere between 5 and 15 would be treated unnecessarily.

Is that worth it? Maybe, it's a matter of opinion. But it is not a matter of opinion that there would be even greater benefit in offering universal access to high quality medical care.