An Overdiagnosis Epidemic Is Harming Patients' Mental Health

WIRED

05-03-2025

Mar 5, 2025 12:04 PM Diagnosing patients when there aren't effective treatments to give them can make their symptoms worse, argues neurologist Suzanne O'Sullivan. Photograph: Caiaimage/Martin Barraud
If you buy something using links in our stories, we may earn a commission. This helps support our journalism. Learn more. Please also consider subscribing to WIRED
Neurologist Suzanne O'Sullivan thinks that modern health care is overdiagnosing people but not necessarily making them healthier—and in fact, that it might be doing more harm than good.
In her new book, The Age of Diagnosis , she backs this assertion with some sobering facts. For instance, between 1998 and 2018, autism diagnoses jumped by 787 percent in the UK alone; Lyme disease has an estimated 85 percent overdiagnosis rate, including in countries where it's impossible to contract the disease; and there's still little evidence that many cancer screening programs actually reduce cancer-related death rates.
Ahead of her keynote speech at WIRED Health later this month, O'Sullivan spoke to WIRED to talk about the boundaries between illness and health, the nocebo effect, and the dangers of early detection. This interview has been edited for length and clarity.
WIRED: You've been a neurologist for nearly 35 years. When did you start seeing this new phenomenon of overdiagnosis?
Suzanne O'Sullivan: In the book I write about Darcy, a young lady who came to me with seizures. She's only 24 and she had ten other diagnoses. I'm an epilepsy specialist, and that should mean that I'm only seeing people with epilepsy, but unfortunately seizures are a thing that happen very frequently for psychosomatic reasons. When I started, people came to me with seizures that had a psychological cause. That's all they had.
WIRED Health showcases the most exciting and thought-provoking disruptors, scientists, and practitioners making a positive change in how we provide and access health care. Find out more.
But over the course of the last ten years, that particular group of patients started to gather a long list of diagnoses. What seems to be happening now is that if you go to different doctors with multiple symptoms, you will get a name for them all. These symptoms always existed, but the naming of them has been detrimental to patients. It's caused them to pay a lot of attention to their bodies and that makes the symptoms worse. That's the nocebo effect.
Every week now I see at least one Darcy, a 24-year old with different medical labels. Most of those labels have no treatment and aren't making them better. That's a very concerning trend for me.
Why are doctors over-labeling symptoms rather than trying to get to the root cause of the disease?
We did have an underdiagnosis problem in the past, particularly with learning or behavioral problems such as autism and ADHD. We didn't recognize people who needed help, so we've been trying to correct that. But we've overcorrected. We have been working on the assumption that the more we diagnose, the healthier you can make the population. That was probably true to a point, but the improvement isn't sustained when you get into the milder end of any disease spectrum.
Patients come to us wanting answers. People want to understand why they are the way they are. A satisfying consultation between a doctor and a patient is often one where the patient asks about a symptom and the doctor can explain what it is. I think it's a little bit of a collusion between what patients want and how doctors can satisfy them with labels.
The increase in the diagnosis rates for ADHD and autism, in particular, have exploded over the past decade. Is this the result of the shifting boundaries between what we now consider to be health and sickness?
That's my concern. Autism has risen from like one in 2,500 people to one in 36 children in the UK and one in 20 in Northern Ireland. That happened because people realized that there must be kids out there who've got milder forms of this, and if we help them, they'll get better.
However, what has happened now is that we have a massive increase in autism, and it is not having the downstream effect of making children better. We should be seeing a slightly happier population, but all we're seeing is worse mental health. We did something well-intentioned but there's no evidence that it's working.
The reason it's not working is because when you get to the very mild end of a spectrum of behavioral or learning problems, you have a balancing act between the benefit of being diagnosed along with the help you can get, and the drawbacks of being diagnosed, which is telling a child that they've got an abnormal brain. What does that do to a child's belief in themselves? How does it stigmatize them? How does it affect their identity formation? We thought it would be helpful to tell children this, but the statistics and the outcome is suggesting it isn't helpful.
You're also worried about another aspect of diagnostics, which is overdetection. One example you give in the book relates to modern cancer screening programs that detect the disease at earlier and milder stages. But so far there's little evidence that these are actually beneficial to patients.
Every cancer screening program will lead to some people getting treatment when they didn't need to be treated. That will always be the case. What we're desperately wrangling with is that we want to make sure we keep the number of overdiagnosed people down and the number of people who need the treatment up. However, the more sensitive you make those tests, the more overdiagnosed people you will have. I read in a Cochrane review that if you screen 2,000 women, you save one life, and you over treat somewhere between 10 or 20 women. You're always overtreating way more people than lives you're actually saving. So the suggestion that we should do even more of these tests before we've perfected the ones we have does not make sense to me.
I do multiple brain scans a week and so many of them show incidental findings. Even though I'm a neurologist and I see brain scans all the time, I don't know what to make of half of them. We just don't yet know how to properly interpret these scans. We need to pay more attention to detecting symptomatic disease early, rather than trying to detect asymptomatic diseases that may never progress.
In some cancers—prostate cancer, for instance—patients can opt for watchful waiting rather than treatment. Should this be the norm for early detection?
If you're going to go for screening—and I don't want people not to go for the suggested screenings—you do need to understand the uncertainties and realize you don't have to panic. Of course, the minute you hear there's some cancer cells, the panic kicks in, and you want it out and you want the maximum amount of treatment. But actually, in medicine, a lot of decisions can be made slowly. There are watchful waiting programs.
I want to suggest to people that, before you go for the screening, know these uncertainties exist, so that you can decide before the test comes back positive what you think you'd likely want to do, and then you can take time to think about it afterwards, and you can ask for a watchful waiting program.
I think one of the solutions would be to call these abnormal cells that we find on screening something other than 'cancer.' The moment you hear that word, people's immediate reaction is to get it out, because otherwise they think they will die of it. Watchful waiting is just something people find hard to do.
Hear Suzanne O'Sullivan speak at WIRED Health on March 18 at Kings Place, London. Get tickets at health.wired.com .