According to various professional guidelines, if we’re the right age and gender, we’re supposed to have our breasts, lungs, prostate gland, cervix, colon, aorta, [1] liver, [2] pancreas, [3] heart and brain [4] routinely screened for various disorders. And, according to recommendations this year, our minds should be screened too; the United States Preventive Services Task Force recommends that everyone aged 12 years and older should be screened for depression.
But is all this screening saving lives as is so often claimed?
Not everyone is convinced. It turns out that the claim that lives are “saved” by screening programs are often based on limited science and misleading language.
This claim is widespread and occurs when disease-specific and all-cause mortality benefits are conflated. But the two are not the same. Indeed, the assumption that a reduction in disease-specific mortality automatically leads to net benefit (and not net harm) has led to a number of disasters and failed screening tests. Take neuroblastoma screening of infants, or screening for ventricular dysrhythmias (which resulted in a quadrupling of deaths, according to the CAST trial).
Using disease-specific mortality as a proxy for overall mortality assumes that the screening is without harms or that harms must be exceeded by benefit. When clinical trials that report disease-specific mortality were later paired with studies of all-cause mortality, the assumption frequently proved to be unsupported or wrong.
What most patients really want to know is: Will this screening test reduce my chance of dying early? Unless a screening test has been shown to reduce all-cause mortality, any claim that a test “saves lives” is unsupported at best and false at worst.
Screening advocates often assert that overall mortality benefits are simply not detected by screening studies because they are underpowered, simply too small to detect an all-cause mortality benefit.
However, in the case of prostate cancer screening, the two major trials involved over 310,000 men, and the failure to detect an overall mortality benefit was surrounded by very narrow confidence intervals. Those narrow confidence intervals mean it is possible that there is either a very small net decrease or increase in deaths from screening. It was little surprise that the United States Preventive Services Task Force gave prostate cancer screening a “D” (Don’t do it) rating.
Some screening advocates have said it would be impossible to conduct proper studies of overall mortality since randomized controlled trials of screening studies would require prohibitively large numbers of test subjects (enrolling millions of men in the case of prostate cancer screening) – something they say is far too expensive.
Yet a study design that could provide adequate information about disease-specific and all-cause mortality for a large-scale screening program impacting millions of people has been described. And large national observational registries could allow millions of participants to be studied; one registry-based randomized controlled trial was conducted for just $50 per participant, making the cost of a trial with 4 million participants comparable to the cost of current screening trials, and orders of magnitude cheaper than most drug trials.
Screening programs have the potential to cause enormous physical and psychological harms since they typically involve large populations. They can also cost billions of dollars in downstream expenditures to treat indolent cancers that didn’t need treatment, or to reduce a surrogate marker that proves not to change outcome, or to manage the side effects of both diagnostic and therapeutic interventions.
Before we launch such ambitious programs we can and must conduct proper studies.
Jeanne Lenzer is a medical investigative journalist who sadly has to leave fascinating tidbits about various stories on the cutting room floor. The Backstory is an attempt to archive some of those bits and to provide a bit of insight for the public about the “behind-the-scenes” aspect of investigative journalism.
Competing interests: None declared.
[1] Aortic aneurysm
[2] Hepatitis B and C
[3] “Pre-diabetes”
[4] Cognitive disorder