Sebastian Schmidt: Overdiagnosis in Lung Cancer (Part 2) – warning signs in real-world evidence

Sebastian Schmidt, the Head of Strategy and Medical Affairs Computed Tomography at Siemens Healthineers, shared on LinkedIn:

“Few days ago I explained how to identify overdiagnosis in controlled trials – if you plot cancer incidence over time and the incidence curve for the screened population stays sustainably higher than for the non-screened control group.

I also showed the misleading effect of short follow-up time. Earlier diagnosis can be mistaken as overdiagnosis, if follow-up is too short. And how the long-term data of the NLST participants shows that overdiagnosis in lung cancer is very low, different from the original analysis with short follow-up (likely negligible).

In real-world data like cancer registries, normally we neither have a control group nor follow-up data. What are signs of overdiagnosis in such data? You have to compare time before and after start of screening.

To illustrate this, I selected two examples: Start of wide-spread PSA screening for prostate cancer around 1991 compared with start of lung cancer screening.

The left chart (1) shows the start of PSA screening. You see a sharp increase in incidence, more than doubled, but no matching decrease on mortality. A relatively strong sign for overdiagnosis and lack of efficacy (doi:10.1056/NEJMcp1103642).

The middle and right chart (2) show data after introduction of #lungcancerscreening. The incidence (middle chart) continued to decrease, reflecting the long-term trend of less smoking. Mortality declined more steeply after many years of running parallel with incidence, as shown by a declining mortality/Incidence ration. The right chart shows the change in stage distribution: More early/local stage at diagnosis, less progressed/late stage. (doi.org/10.3322/caac.21708).

Conclusion: A typical sign of overdiagnosis in cancer screening is an increase in incidence that is not followed by a decrease in mortality and/or a decrease in late-stage diagnosis.

None of these warning signs were seen after start of lung cancer screening with CT in US. They were seen after start of PSA screening historically, which lead to a negative USPSTF recommendation later.”

Source: Sebastian Schmidt / LinkedIn

You may read also:
Sebastian Schmidt: Overdiagnosis in Lung Cancer (Part 1) – is it real? And is it a problem?