Ask a primary care doctor how often an ordinary test turns up something nobody was looking for, and you’ll get a tired laugh.
In 2019, a research team surveyed a national panel of practicing American internists and published the results in JAMA Network Open. Of the physicians who answered, 99.4% said they had lived through a cascade: a chain of extra tests, scans and referrals set off by an incidental finding. An incidental finding is something a test spots by accident, unrelated to the reason the test was ordered. More than nine in ten said they’d been through at least one cascade that ended with nothing worth treating.
Those dead-end cascades are not free. In that same survey, 68.4% of physicians said a cascade had caused their patient psychological harm, 57.5% said it had caused financial strain, and 15.6% said it had caused physical harm. Half of the doctors said the experience left them frustrated. About one in three said the test that started the whole thing may not have been appropriate in the first place.
This is the half of testing that almost never gets explained before you roll up your sleeve. Tests find disease, and that saves lives. Tests also generate noise, and the noise lands on you.
Overdiagnosis is the hardest idea here, and the most important
Start with the concept most people have never heard, because almost everything else follows from it.
Overdiagnosis means a test correctly finds a real abnormality that was never going to hurt you. The pathology is genuine. The cells under the microscope are genuine. What’s missing is the future in which that abnormality would have grown, spread, caused a symptom or shortened your life. You get the diagnosis, the treatment and the side effects, and you receive none of the benefit, because there was no harm coming.
That is not the same thing as misdiagnosis, and mixing the two up is the most common error readers make. A misdiagnosis is a mistake. The lab was wrong, the radiologist read the wrong shadow, the doctor named the wrong disease. Overdiagnosis involves no mistake at all. Everyone did their job correctly. The problem is that medicine found something true and unimportant, and then treated it as though it were true and urgent.
This is the part that makes it so slippery. For any one person, overdiagnosis is invisible. If your small cancer was removed and you lived another 30 years, you cannot know whether the surgery saved you or whether that tumor would have sat there forever. You feel like a survivor either way. That’s why overdiagnosis can only be measured across large groups, by comparing how many people get diagnosed in a screened population versus an unscreened one.
How large is it
A 2022 review published in the Journal of the National Cancer Center, written in part by authors from the National Cancer Institute’s Division of Cancer Prevention, laid out the logic plainly. A screening test that works should do more than raise the count of early cancers found. It also has to bring down how many late-stage and spreading cancers show up, and after that, how many people die of the disease. When early diagnoses climb and advanced disease doesn’t fall, the review argues, screening is probably turning up a pool of slow-growing or non-growing abnormalities that would never have announced themselves in a person’s lifetime. The authors point to prostate, thyroid, lung, melanoma, breast and neuroblastoma as areas where this pattern has shown up.
A 2023 study in Annals of Internal Medicine put numbers on one slice of it. Researchers linked Medicare claims to the federal SEER cancer registry and followed 54,635 women aged 70 and older for up to 15 years, comparing those who kept getting screening mammograms with those who stopped. Among women aged 70 to 74, roughly 31% of the breast cancers found in the screened group were potentially overdiagnosed. For women 75 to 84 the estimate was about 47%. For women 85 and older it reached as high as 54%. The researchers did not see a statistically significant drop in breast cancer deaths in the screened group at those ages.
Read that carefully, because it is easy to over-read. The study was designed to estimate overdiagnosis in older women specifically, not to settle whether mammography works. The authors say so themselves, and they name confounding factors that could shift the numbers. Screening earlier in life sits on much stronger ground. What the study does establish is that overdiagnosis is a measurable quantity, not a theory, and that it grows as the chance of dying from something else grows.
The cascade: one finding, six more appointments
The second harm is mechanical. Modern scanners are extraordinarily sensitive, and a sensitive scanner pointed at a healthy adult will find things.
A radiologist writing in the British Journal of Radiology in 2021 described the daily version of this problem. Abdominal CT, MRI and ultrasound all capture organs that have nothing to do with why the scan was ordered. Most of those extra findings are harmless, and the author argues that some of them, like small simple cysts, arguably shouldn’t be reported at all. The trouble starts with lesions that look probably benign but not certainly benign. Somebody has to decide whether to repeat the scan in six months, order a more definitive one, or take a tissue sample.
That decision rarely has a clear rulebook. In the JAMA Network Open survey, when physicians described their most recent cascade, 53.2% said that no guidelines for the follow-up existed as far as they knew. Another 8.1% said guidelines existed and weren’t followed. Asked what would limit the damage, 62.8% picked easier access to guidelines, the most popular answer on offer.
So you get the phone call. Then the repeat scan. Then the specialist who orders a different scan. Each step is individually reasonable. Together they can run for a year.
Radiation, money and the worry nobody bills for
The physical harms are real but worth stating precisely, without inflation.
The Food and Drug Administration’s page on computed tomography is blunt about the trade-off. CT uses ionizing radiation, and the FDA says radiation exposure may slightly raise a person’s lifetime cancer risk. For a scan that’s medically necessary, the agency considers that risk very small next to the value of getting the diagnosis right. Children are more vulnerable, both because their risk per unit of dose is higher and because they have more years ahead for an effect to appear, which is why pediatric settings matter.
Two points on that same FDA page deserve more attention than they get. The agency advises providers to check a patient’s imaging history specifically to avoid duplicate exams. And the FDA states it sees no benefit to whole-body CT scanning of people without symptoms, which is exactly the product sold in shopping-mall screening clinics.
Money is the harm patients notice first. Consider one small, well-documented example. A 2014 study in Osteoporosis International used complete Medicare claims to look at bone density scans in women over 66. About one in ten scans was repeated sooner than the two-year interval that rheumatologists had flagged as excessive. Those short-interval scans alone cost Medicare roughly $16 million in 2011. Rates varied enormously by region, which is a hint that the driver was local habit rather than patient need.
Then there’s anxiety, which insurers don’t code and studies struggle to capture. A 2024 review in the Journal of Breast Imaging, written by radiologists who support screening, acknowledges that the debate over mammography turns on three things: the effect on deaths, overdiagnosis, and false positives, including the distress a callback causes. That’s worth noticing. Even clinicians arguing for more screening treat the psychological cost of a false alarm as a real item on the ledger, not a complaint to be waved off.
Why the test gets ordered anyway
If the harms are this well documented, why does overtesting persist? Four reasons, and only one of them is anybody’s fault.
Fear of being wrong. Physicians sometimes order extra tests to protect themselves legally, a practice researchers call assurance behavior. A 2020 study in the American Journal of Clinical Pathology surveyed 160 dermatopathologists nationally and found that 95% reported using at least one such behavior, like ordering extra stains or seeking a second review, because of malpractice concerns. The interesting finding is the comparison. For nearly every behavior, a higher share of doctors attributed it to patient safety concerns than to legal ones. So defensive medicine is real, but it tangles with genuine caution in ways that are hard to pull apart.
What patients ask for. Many people arrive wanting a test, and a scan feels like being taken seriously. Saying no costs a clinician time they do not have in a 15-minute slot.
How the money moves. The bone density study offers a quiet clue here. When Medicare payment changed, the rate of short-interval scans moved with it, while appropriately spaced scans held steady. Necessary care tends not to flex with reimbursement. Marginal care does.
Uncertainty is genuinely hard to sit with. This is the honest one. A doctor who tells you “I don’t think we need to image this, let’s see how it looks in three weeks” is accepting risk on your behalf and on their own. Ordering the scan is emotionally cheaper for both of you.
Choosing Wisely, in the past tense
The biggest organized attempt to name overused tests in American medicine was Choosing Wisely, a campaign run by the ABIM Foundation. It is important to get the tense right, because plenty of articles still describe it as live.
Choosing Wisely launched in 2012 with nine national specialty societies representing about 375,000 clinicians, who between them named 45 widely used tests and treatments that the evidence did not really back up. Over the following 11 years, more than 80 specialty societies contributed. The campaign ended up with over 700 recommendations and spread to more than 30 countries. The ABIM Foundation has since shifted its focus to a different effort, its Building Trust initiative, which looks at the role of trust in care quality. The campaign’s individual recommendation lists are no longer maintained on the Choosing Wisely website, and specialty societies have been encouraged to publish their own.
What survives is the body of work and the habit of mind. Those 700-plus recommendations were written by the specialists who perform the procedures, which is the opposite of an insurer deciding what you don’t need. If you want to know whether a specific test is considered overused, the relevant specialty society is the place to ask.
Choosing Wisely also published a wallet card with five questions for patients, in English and Spanish. The card covered the basics: need, risk, simpler alternatives, the option of waiting, and price. Useful, and a reasonable starting point. The questions below are a different set, built around an idea that card doesn’t quite reach.
Five questions that change the answer
The most powerful question in medicine is not “do I need this?” It’s a question about consequences. Ask these in order.
1. What will we do differently depending on the result? If the honest answer is “probably nothing,” you’ve found a test worth skipping. If the answer is “a normal result means we stop worrying about X,” that’s a test earning its place. This one question resolves more unnecessary testing than any other.
2. What are the odds this finds something that turns out not to matter? You’re asking about false positives and incidental findings. A good clinician will have a rough sense for the test in question and will tell you.
3. What’s the plan if it comes back borderline? Decide this before you get in the scanner, not after. If nobody can tell you what a gray-zone result triggers, you’re walking into an unplanned cascade.
4. Does anything change if we wait three weeks? For a lot of complaints, time is a free diagnostic test. Many things resolve. The ones that don’t have declared themselves more clearly.
5. Who calls me with the result, and by when? Loop closure is where real harm happens. A test nobody follows up on is worse than no test, because it generated risk and cost without producing information.
If you don’t get satisfying answers, you are not obligated to decline. You can ask for the reasoning to be written in your visit note, which tends to produce a more careful answer than any amount of arguing.
Now the other half, because undertesting is real too
An article about too much testing owes you the opposite case, stated just as plainly.
The Agency for Healthcare Research and Quality, the federal agency that funds research on care quality, estimates that each year 795,000 Americans die or end up permanently disabled because their disease was diagnosed wrong. AHRQ has funded research on diagnostic safety since 2007 and runs a federal interagency workgroup on the problem. That number dwarfs anything in the overdiagnosis literature.
The pattern in malpractice data makes the stakes concrete. A 2017 analysis in the Joint Commission Journal on Quality and Patient Safety reviewed 3,407 closed malpractice claims in outpatient general medicine. Cardiovascular cases were far more likely than other cases to involve an allegation of diagnostic error, 75% versus 47%, and far more likely to end in the patient’s death, 75% versus 27%. Among the cardiovascular cases where other conditions were recorded, 63% of patients already had at least one standard cardiac risk factor on the chart. What tripped things up were initial diagnoses of nonspecific chest pain and look-alikes such as esophageal problems.
Preventive care has the same gap in the other direction. The Centers for Disease Control and Prevention notes that the U.S. Preventive Services Task Force recommends colorectal cancer screening for adults aged 45 to 75, with several options including a yearly at-home stool test. Most insurance covers it without a copay. Plenty of eligible people still skip it.
The USPSTF grading system is a useful frame here, because it is explicitly about balance rather than volume. A grade A is assigned when the Task Force is highly confident the service does substantially more good than harm. A grade D is a recommendation against the service, given when the panel is reasonably sure the good does not exceed the harm. An I statement means the evidence isn’t strong enough to tell either way. Notably, the Task Force does not consider cost when it assigns these grades. The question is always benefit against harm, never fewer tests for their own sake.
When to stop reading and make the call
None of this applies to a symptom that’s new, severe or getting worse. If your doctor says a test is time-sensitive, that is not the moment to deploy question four. Sudden chest pressure, one-sided weakness, trouble speaking, a severe headache unlike any you’ve had, trouble breathing, or bleeding that won’t stop are emergencies, and they belong in an emergency department, not in a conversation about appropriateness. The same goes for a screening test you’re already due for under a current recommendation.
The target is precision, not scarcity.
The one thing to take away
Walk in knowing what each answer would change.
That single habit covers nearly everything this article describes. It filters out the scan that leads nowhere, because nothing would be done differently. It catches the panel added out of reflex. It also protects you from the opposite mistake, because a test that would genuinely redirect your care is one you want, and the question makes that obvious too.
Write it on the back of your appointment card if you have to. “If this is normal, what do we do? If it’s abnormal, what do we do?” If both answers are the same, you’ve learned something more valuable than the test result.
Sources
- ABIM Foundation (Choosing Wisely). “Patient Resources.”
- Choosing Wisely and Consumer Reports. “5 Questions to Ask Your Doctor Before You Get Any Test, Treatment, or Procedure.” choosingwisely.org
- U.S. Food and Drug Administration. “Computed Tomography (CT).” fda.gov
- U.S. Preventive Services Task Force. “Grade Definitions.” uspreventiveservicestaskforce.org
- U.S. Preventive Services Task Force. “About the USPSTF.” uspreventiveservicestaskforce.org
- Agency for Healthcare Research and Quality. “Diagnostic Safety and Quality.” ahrq.gov
- Centers for Disease Control and Prevention. “Screening for Colorectal Cancer.” cdc.gov
- MedlinePlus, U.S. National Library of Medicine. “How to Understand Your Lab Results.” medlineplus.gov
- Ganguli I, et al. JAMA Network Open, 2019. PubMed. DOI: 10.1001/jamanetworkopen.2019.13325. “Cascades of Care After Incidental Findings in a US National Survey of Physicians.” doi.org
- Richman IB, et al. Annals of Internal Medicine, 2023. PubMed. DOI: 10.7326/M23-0133. “Estimating Breast Cancer Overdiagnosis After Screening Mammography Among Older Women in the United States.” doi.org
- Dunn BK, et al. Journal of the National Cancer Center, 2022. PubMed. DOI: 10.1016/j.jncc.2022.08.005. “Cancer overdiagnosis: a challenge in the era of screening.” doi.org
- Morden NE, et al. Osteoporosis International, 2014. PubMed. DOI: 10.1007/s00198-014-2725-2. “Overuse of short-interval bone densitometry: assessing rates of low-value care.” doi.org
- Quinn GR, et al. Joint Commission Journal on Quality and Patient Safety, 2017. PubMed. DOI: 10.1016/j.jcjq.2017.05.001. “Missed Diagnosis of Cardiovascular Disease in Outpatient General Medicine: Insights from Malpractice Claims Data.” doi.org
- Reisch LM, et al. American Journal of Clinical Pathology, 2020. PubMed. DOI: 10.1093/ajcp/aqaa088. “Malpractice and Patient Safety Concerns.” doi.org
- Pickhardt PJ. British Journal of Radiology, 2021. PubMed. DOI: 10.1259/bjr.20211167. “Incidentalomas at abdominal imaging.” doi.org
- Choe AI, et al. Journal of Breast Imaging, 2024. PubMed. DOI: 10.1093/jbi/wbae031. “Tips for Addressing Screening Concerns: Harms of Screening.” doi.org