The PSA Screening Test: Think very carefully before you act on the results

[Mark Rippetoe]

Screening with a PSA test has a small impact on prostate cancer deaths but leads to overdiagnosis, finds study

The CAP trial, published in the Journal of the American Medical Association (JAMA) and carried out by researchers from the universities of Bristol, Oxford and Cambridge, involved over 400,000 men aged 50-69. Just under half received a single invitation for a PSA test as part of the trial.

After following up for 15 years, there was a small difference in the number of men who died from prostate cancer between the two groups—nearly 7 men out of every 1,000 in the group invited for screening had died from prostate cancer, compared to nearly 8 men out of every 1,000 in the group who hadn't been invited for screening.

Professor Richard Martin, lead author and Cancer Research UK scientist at the University of Bristol, said, "Our studies have been measuring the effectiveness of the PSA test with hundreds of thousands of men for 15 years. The key takeaway is that the small reduction in prostate cancer deaths by using the test to screen healthy men does not outweigh the potential harms.

This has been widely known for years, but I thought I'd better show you this in case there remained any credulity in the Forum.

[Cascadian]

Just one of many problems with tests is that the average person - and often the physician - does not understand basic statistics.

This lack of understanding then leads to additional tests and procedures, all of which are prone to more errors, more misinterpretation, more costs, more risk, and more anxiety.

Let me give a simple example:

• Assume there is an infectious disease in the population.
• Assume that 1000 people are tested for the disease, using a test that has "90% accuracy."
• Assume you test positive for the disease.
• Assume the infection rate is 1%.

What are the chances you actually have the disease?

Most people would assume 100%. After all, they tested positive.

Some with slightly bigger brains might think 90%, given that the test has "90% accuracy."

The correct answer is: 8.33%

Why? Because Bayes' Theorem is a thing, and gender studies isn't.

- - - - - - - - - - -

For the mathematically curious, given the above assumptions:

9 = True positives
1 = False negatives
99 = False positives
891 = True negatives

Chances of having the disease if you test positive = (True positives) / (True positives + False positives) = 9 / 9 + 99 = 9 / 108 = 8.33%

(Again, if you don't know basic logic, basic statistics, and basic double-entry accounting, you're doomed to be an idiot.)

[Browndog]

Seems that you're not the only one to take notice of that study ... a couple of days later, but better late than never.

Assume The Position – small dead animals

[gadders]

We get them offered them once you are over a certain age here as part of our annual medical. As I'm symptom-less I turned mine down. I mentioned about it here before when discussing Colonoscopies.

Slightly off topic - I got offered a PSA test last week as part of the annual medical screening by my employer. Only due to my age, no symptoms.

After what I'd read on here, and listening to a podcast on screening I found this link

They use "number needed to treat" as their metric for deciding whether a test is worth having or not. The PSA test got a Black Rating which means "Therapies rated black have very clear harms without any benefit."

I think I'll skip it.