The Psephology of Early Diagnosis in Britain’s NHS

flying cadeuciiConsider this equation.

Early Diagnosis = Early Diagnosis + Overdiagnosis (1.1)

This sort of unequal algebra will fail high school mathematics. A new NHS initiative is arithmetic defying as well. Patients who think they have symptoms of cancer will be allowed to book medical imaging directly, without seeing their GP. This is to catch cancer early. The logic is impenetrable: early diagnosis of cancer saves lives.

Here is the problem. Cancer does not unequivocally announce its arrival. Early cancer presents with non-specific symptoms, such as an uncomfortable niggle in the back.


Let’s take Tom, who has advanced pancreatic cancer. He recalls that three years earlier he noticed a dull pain in his back during a misguided drinking binge. He would be correct in thinking that had he attended the emergency department and had a CAT scan of his abdomen, the cancer would have been smaller and would not have spread to other organs. He is right in contending that had the cancer been removed then, he would have a longer survival than presently.

The rationale has implications if extrapolated to everyone. To understand the consequences of extrapolation let’s visit a logical fallacy.

1) All Mr Smiths are over six feet tall.

2) He is above six feet tall so he must be Mr Smith.

This is affirming the consequent. Not all men above six feet in height are Mr Smiths. In fact, most are not.

3) Early pancreatic cancer presents with back pain.

4) All patients with back pain have early pancreatic cancer.

Similarly, (3) doesn’t imply (4).

Cancer often presents with non-specific symptoms, such as a vague discomfort, early on. But the majority of people with a vague discomfort do not have cancer. That is, the chance that someone with pancreatic cancer has dull back pain should not be confused with the chances that someone with dull back pain has pancreatic cancer. The chances of the latter are much lower than the former.

To pick up Tom with back pain who has pancreatic cancer, we will image many Toms with back pain who don’t have pancreatic cancer. Because we don’t know which Tom with back pain has cancer and which Tom does not have cancer.


In the search for a Tom with back pain with pancreatic cancer, we will pick up a Dick with back pain who does not have pancreatic cancer. Dick has a small kidney tumour. The tumour was minding its business and would not have caused Dick any issues in his life. But we will now remove Dick’s kidney to deal with an incidental finding, which has nothing to do with his symptoms. Thus, in order to save Tom from pancreatic cancer, Dick loses a kidney. This is collateral damage. Tom’s early diagnosis leads to Dick’s overtreatment.


We will also pick up a Harry with back pain and a small cyst in his pancreas, which may or may not be cancer. We will then scan Harry repeatedly, perhaps every six months, to make sure that it really is a harmless cyst. It probably is, but you can never be too careful.

To save Tom from pancreatic cancer, we end up wasting Harry’s time by CAT scanning him every six months, driving his anxiety off the scale. Because when the doctor said “probably not cancer,” Harry heard “cancer.” Harry is a false positive. He won’t know of the “false” bit for a very long time.

The equation can be modified to:

Early diagnosis = Early Diagnosis + Overdiagnosis + False Positives (1.2)

Tom Jones

A flamboyant Tom Jones also has back pain. His CAT scan fails to reveal the cancer that presents later. CAT scans are good. Very good. But they are not perfect. Tom Jones is very peeved because he believed the “early diagnosis saves life” mantra. As far as he is concerned, he did his bit. He complied.

But we lied. Tom Jones is a false negative.

The equation becomes:

Early Diagnosis = Early Diagnosis + Overdiagnosis + False Positive + False Negative (1.3)

The right side of the equation has costs. Opportunity costs. There will be less money for other services, such as emergency medicine. 

Tom, Dick, Harry, and Tom Jones

Tom, Dick, Harry, and Tom Jones will Google “back pain + cancer.” The search will yield “pancreatic cancer.” After a while this will become a self-fulfilling prophecy, thanks to Google’s ranking algorithms, which do not distinguish between the incidence of and anxiety from cancer.

Del Trotter

When Tom et al can’t get a CAT scan within 48 hours, they will throng the emergency department because of anxiety. This will affect the care of Del Trotter from underserved Peckham, London. Del is having a heart attack. Del will die if his left anterior descending artery is not promptly opened. Del’s doctors are being distracted by the demanding Toms.

Britain’s NHS has problems—structural problems, through systematic under investment. Emergency departments and intensive care units are in crisis. Is the pursuit of early diagnosis of cancer really its highest priority? Why are politicians promising cake when there is not enough bread?

Actually, this pursuit is rewarding. Why? Because the public will be grateful. Tom is happy that his pancreatic cancer was detected early. The lucky Toms, the true negatives, are even happier that they don’t have cancer. Dick is ecstatic that his kidney tumour was taken out. He does not know the whole exercise was an utter waste of his time and an innocent kidney was sent to the gallows (surgical pathology). He does not know that he was overdiagnosed and overtreated.

Harry, the false positive, is relieved. After several surveillance CAT scans, his doctors broke the news that the pancreatic lesion was a false alarm. He is not only relieved but prostrating with gratitude. He is suffering from a medical variant of the Stockholm syndrome. Del Trotter from Peckham, the opportunity cost, has other things on his mind. This leaves Tom Jones. He is annoyed. But a root cause analysis, with considerable assistance of hindsight, will have found an element missing from the diagnostic pathway. His anger has been redirected to the imperfect medical profession, which refuses to adequately self-regulate.

We can show:

Early Diagnosis = Early Diagnosis + Overdiagnosis + False Positives + False Negative + Opportunity Costs + Votes ++++ (1.4)

Quod Erat Demonstrandum

If you were a politician, would you honestly see a downside to this?

Saurabh Jha, MD is a radiologist practicing in Pennsylvania.   (This article originally appeared in BMJ Blogs)


Categories: Uncategorized

17 replies »

  1. Thanks for the comment Tracy. I do believe ADHD is a real condition but there decisions to be made by individuals and society about how we address it. I think that the ADHD mind/body set was likely very helpful at some point in our evolution. It is just against the grain of our culture now, at least in America.

  2. I fell for it!

    Why? Because NHS England is like a deer in headlights. There is no coherent strategy. Emergency medicine is in a crisis. This is a useful deflection.

  3. I Believe that both Dr Palmer and Dr. Jha have excellent arguments, but from the clinicians point of view I see more validity in the search for knowledge that Dr. Palmer has spoken of. Internists are taught to seek out large quantities of information and put the data together generally to find one unifying disease entity. They don’t just look at one symptom and try to attach a disease to it. Of course data can be quite troubling because the more data the more potential decisions that may be required unless the data only presents a singular diagnostic entity. All that knowledge on each case adds to the knowledge base and helps in the diagnosis and treatment in future cases and can very well help in the treatment of the present patient..

    Dr Jha points out that the complexities can cause the physician to go down to many dead ends that may cause good or harm. To prevent harm is one of the difficulties faced by physicians every time a patient enters his examination room. But that is what physicians are supposed to do, make hard calls. On the other hand the clinician should not bite off more than he can chew.

  4. There are the “Frequentist” and “Bayesian” camps, broadly. Count me firmly in the latter. Base rates matter.

    I am also a Chebychev-ist, at the margins.

  5. Will bypassing the GP reduce the backlog for GPs? I’m not sure that is empirically correct.

    When there is an explosion of incidental findings, findings of uncertain significance, and even real cancer diagnoses, it is the GPs who will have to deal with the resurgence.

    In the NHS GPs do not send patients for CAT scans. There they have something that is anathema to us: a budget.

    Nor am I sure about the efficiency bit. It costs more to scan 30 people than 8 people. Someone has to physically put the patient on the scanner and take the images. It’s not a drive through machine. Yet.

    (Your response could have been entirely satirical in which case the joke is on me)

  6. Well done. The reason they encourage patients to self refer is to eliminate some of the backlog in the GPs office. He would also most lilkely refer them for a CAT scan, as well. It does not cost more to run the CT Machine constantly, it is there,the staff is there. l It is much more efficient to have everyone self refer. CT Scan is ‘MAGIC”

  7. Good analogy, homeland security and “never miss a cancer” brigade.

    The “shoe bomber”, Richard Reid, lowered the base rate even further.

    It’s amazing how many physicians don’t understand Bayes. I have a feeling that often the ignorance is willful.

  8. I’m unsure what point you are making.

    Are you saying false positives don’t exist or don’t matter?

    Information is not perfect. With imperfect information there is a trade off between false positive and false negative. If you want fewer false negatives you will get more false positives. The trade off is not one to one.

    “My point is that more information can never be bad, if we learn not to act rotely upon it, because there will be pearls within.”

    CT shows a kidney tumor. At that point one does not know that the tumor will or will not progress to invasive cancer. In some it will. In many it won’t. If you take it out in all you’ll have false positives (overdiagnosis) but no false negatives. If you leave it in all you will have no false positives but false negatives.

    This is a trade off. As uncomfortable as it may be to shed binary approach to life (“always”, “never”, “good vs evil”, “you are either with us or the terrorists”), gray does not become black-white just because we wish for it to be so.

  9. How can information about a patient be bad? I’m not arguing that the process of getting the information might not be harmful or costly (e.g. to much radiation.) But information by itself?…can it be intrinsically bad?

    You are saying that some of that information may lead one to inconsequential or misleading diagnoses, false positives, etc., and that these can be costly and harmful to pursue. But isn’t the cure for this false pursuit more knowledge?, acquired by more information?…i.e. knowing that the .5cm shadow on the CT scan of the lung is more likely to be non-malignant and not pursuing it? This is knowledge that we had to acquire by first chasing down some of these shadows and pursuing them. We don’t have to go after every diagnosis that our information thinks is a possibility. We learn from more information not to choose to go after organizing pneumonia, e.g., or sarcoid, or a patch of hypersensitivity pneumonia in a bird lover. Plenty of information, some of which is spurious, is not bad. It teaches us to ignore that which we learn is probably spurious.

    After all, lots of images, chemistries, hematologic testing et al, are just enlarging the “history” in the patient. It’s the history of their inner environment. How can a history that is elaborate be bad or wrong? We are taught that the detailed history is good. At first we learned that to gather information about color blindness history of a patient was trivial because we could not correlate color blindness with any illness. But we did not know this at first. How did we learn this?

    By only seeking information that we believe is going to yield useful results, we encourage false negatives. I.e. we falsely believe the patient does not have something, which they actually have. And we stop gathering more knowledge, pari passu.

    My point is that more information can never be bad, if we learn not to act rotely upon it, because there will be pearls within. And we learn not to act by gathering more information. So we grow if we always seek maximum information. And this helps us not to act or to act appropriately if we learn.

    It is just like we ignore much of the patient’s history. But we like to have a complete one, nevertheless.

  10. Thanks for reading, Perry.

    And in both countries the politicians are not honest with the electorate leaving physicians to deal with imbalanced desires versus reality.

  11. You can take a people with a history of always maintaining that “stiff upper lip” and convert them to a free system and it will work well for awhile because their collective character delays the weaknesses of the new system from coming to light. But you can never ever take a people used to free and get them to even consider having a “stiff upper lip.” The initial success, which had nothing to do with the system and everything to do with the character of society, fades completely as that system changes society’s character.

  12. “One thing we’ve learnt in the NHS over the years is the more we provide, the more people want. Waiting lists come down, activity goes up,” she said.

    Sills did stress, however, that the delivery of care for free at the point of need was still a core principle of the NHS, something that does not appear to be up for discussion in the current political climate.

    This is quoted from an article in Medscape about the NHS and its current problems. You notice the word “free”. If politicians just use that one word, they are golden. The problem is, someone is paying for it, and as you have pointed out, there are always unintended consequences from trying to give everyone everything.