Shouldn’t we want to know early if we have a disease? It just seems obvious that we would want to screen for everything and treat it before it got bad. You have probably heard about big companies that each year give all of their executives an amazingly thorough checkup including lots of ultrasounds and even whole body MRI and extensive blood and urine analysis. The latest idea is to get your whole genome analysed. You can do it for $795 to $3000. Many more population screening tests are being promoted.
Population Screening is giving a test to everyone, even if they have no symptoms, to see if they have the earliest stages of a disease. But screening makes sense only if:
- The test is accurate.
- The test does not trigger harm.
- There is a treatment that works.
- The treatment is better if it is given early
- The treatment has a cost that is reasonable
So let’s look at these issues.
It is surprising to many that lots of population screening tests are not really that accurate. Of course it depends on what you mean by accurate.
The traditional way of defining accurate is sensitivity and specificity. Sensitivity is the proportion of true positives and specificity is the proportion of true negatives that the test correctly identifies. You might have seen a screening blood test for Alzheimer’s Disease mentioned on the news as a breakthrough discovery from an international group led out of Kings College in London England. They correctly identified 85% of people with dementia (sensitivity) and 88% of people who did not have dementia (specificity). Not only the newspaper but also reputable associations touted this test as a promising way to screen for Alzheimer’s. That is all pretty impressive for a blood test. But is it true? Look at this tree plot diagram from a blog by Professor David Colquhoun.
He showed that given that about 10% of us will develop Alzheimer’s, out of 1000 people tested, there would be 193 with a positive test result but 108 would be false positives. That means the 56% of people would be told in error they were developing Alzheimer’s. This is called the False Discovery Rate. They would be told they had a disease that they did not have. They would likely worry about normal lapses in memory. They might make life decisions such as taking early retirement because of the false information. They might get depressed because they thought they were going to get Alzheimer’s. So thank goodness, the government is not going to fund this test.
One way screening tests do harm is they trigger follow up tests. Follow up tests cost money and have risks of harm. Some harm is relatively minor such as having to come into the hospital for a test. Some may cause some pain. And some follow up tests have small but serious risks. People (not many) have had serious effects or even died from tests. Also, with so many tests being done, diagnostic tests for people with symptoms may be delayed
With Alzheimer’s, there is no test to confirm if you have the early stages of the disease. So more than half the people who tested positive using that test would be stuck with a wrong test with no way of finding out if it were wrong.
But what about the true positives? Might they not benefit from screening? Some would say, that is not so bad if there were a good treatment for Alzheimer’s. Everyone could take it. But there is no good treatment for Alzheimer’s disease. The treatments for Alzheimer’s are expensive and convey very little benefit. On average they may slow symptoms by a few weeks or months. As well, there is no evidence that these drugs are more effective if taken early. So there is no way of benefiting from a true positive result.
I read a very interesting article that summarized studies of attempts to add years to life by screening in diseases in adults that cause death. This systematic review found scant evidence that any screening program in adults delays death at all. Moreover, there have been recent challenges to screening for breast cancer and for prostate cancer. Needless to say, these ideas are controversial.
That is not to say that all screening programs are bad. Many of the screening programs that we do in babies where we detect serious illnesses that can be treated have value. For example, screening for PKU. Without screening there is almost certain very serious results and early death. With screening and early intervention there is a good quality of life.
As well, these arguments do not apply to diagnostic tests. Diagnostic tests are given to people when there is suspicion of disease. They can be helpful and form the basis of much of our medical care. However, a recent systematic review of 140 studies of new diagnostic tests showed only minimal effects on patient outcomes even if they changed treatment. So we are not getting that much better with new diagnostic tests.
There are two other uses of screening tests that can be valid. First, to screen high risk people for a specific disease when there is a good treatment that can be usefully given earlier. Second, to screen for inclusions into randomized trials. A major problem in research occurs when people with different diseases are included in a trial. The results may be a mess. So refining who goes into a trial can help immensely in figuring out if a treatment works. This use of screening will eventually help patients if a better treatment is found. It is very different from population screening of people who have no symptoms.
So, as you can see, I am skeptical about population screening. It may do more harm than good. We should look carefully at all population screening programs.
~Dr. Patrick McGrath