‘In clinical practice, to say a person has cancer gives as little information about the possible course of his disease as to say that he has an infection. There are dangerous infections that may be fatal and there are harmless infections that are self-limited or may disappear. The same is true of cancers. Cancer is not a single entity. It is a broad spectrum of diseases related to each other only in name.’ Cancer surgeon Dr George Crile – 1955 LIFE magazine.
In an opinion piece in JAMA Surgery, Dr Gilbert Welch (MPH), Centre for Surgery and Public Health, Brigham and Women’s Hospital, Boston, writes that Crile’s recognition of the heterogeneity of cancer growth back in 1955 presaged why early cancer detection might defy simple intuition.
He writes:
It is tempting to think that cancer screening can only help individuals and that all survivors of cancer detected by screening provide powerful evidence that it saves lives.
However, cancer screening is counter-intuitive. It turns out that the harms are more certain than the benefits; the survivors less likely to be evidence of its benefit and more likely to be evidence of its harms.
It was probably Crile who initially introduced the barnyard pen analogy – birds, rabbits and turtles – to describe the heterogeneity of cancer. The birds have already escaped the barnyard: they are the fastest growing and most aggressive cancers, those that have already spread by the time they are detectable. Screening cannot help with the birds; the question is whether treatment can. The rabbits are more slowly progressive cancers, and they can be caught early. They are the cancers with which screening can potentially help. Then there are the turtles: there’s no need catch them because they’re not going anywhere anyway.
Limited (or uncertain) benefit
The goal of cancer screening is to reduce cancer mortality. But the existence of birds necessarily limits its effectiveness. Screening tends to miss the fastest growing cancers because these have such a short time window during which they are detectable by screening, but they are not clinically evident. Furthermore, effective screening requires not only earlier detection, but also treatment initiated earlier is reliably better than treatment initiated later.
Although this is often true for acute disease (e.g. cardiovascular emergencies), it may not be for long-term disease. Detecting cancerous tumours that are a few millimetres smaller than those detected previously may make less difference than what was once believed, particularly as we learn that tumour biology is more relevant than its size.
These factors explain why the most favourable effects observed in randomised trials of screening tests result in a less than one-third reduction in cancer mortality.
Ironically, as cancer treatment improves, the benefit of screening decays. If clinically detected cancer can be routinely treated successfully, the utility of cancer screening naturally falls to zero. (For example, why do we not screen for pneumonia? The answer is because we can treat pneumonia.)
Poorly recognised (or hidden) harms
From an individual’s perspective, over-diagnosis is the most consequential harm of screening. Over-diagnosis reflects the existence of turtles: abnormalities that meet the pathologic criteria for cancer yet are not destined to cause either symptoms or death.
Screening is very good at finding turtles: it disproportionally detects slow-growing and small cancers – which are biologically favourable. However, we physicians are not very good at distinguishing turtles from rabbits. Consequently, we tend to treat everybody given a cancer diagnosis. Because there is nothing to fix, patients with over-diagnosed cancers cannot benefit from treatment, but they can be harmed by it.
Because over-diagnosis is so rarely confirmed in an individual (i.e. a patient with a cancer that is detected by screening but is not treated, never develops symptoms, and dies of some other cause), there was considerable debate about whether the problem really existed.
However, over-diagnosis can be easily confirmed at the population level. Thus, debates about the existence of over-diagnosis are now largely settled and have rightly moved to the question about its frequency – and how much it matters. In the case of breast, prostate, skin and thyroid cancer screening, patients are more likely to experience the harm of over-diagnosis than they are the benefit of screening – avoiding a cancer death.
Nevertheless, over-diagnosis remains, thankfully, a relatively rare event.
From the population perspective, the most consequential harms relate to the dynamics of screening: many individuals must be screened to potentially benefit a very few. For cancer screening in the general population (i.e. not targeting an extremely high-risk group, such as screening heavy smokers for lung cancer), roughly 1,000 people must be screened to avert 1 cancer death in 10 years. Thus, questions about what happens to the other 999 individuals become relevant.
False alarms affect many: there are as many as 600 false-positive results in a 10-year course of mammography. However, the bigger problem is that many people with false-positive test results are not told that the test was wrong, but rather that something is wrong with them. These people are told that they have a lesion with cellular atypia or dysplasia, or they themselves are more vulnerable to develop cancer and thus require more frequent testing.
Even the screening campaigns induce vulnerability – the fear of cancer – as a means to persuade people to get screened. Thus, cancer screening injects a sense of “dis”-ease into the population.
Misleading feedback, financial incentives, and distraction
These harms might be acceptable were they accompanied by substantial and certain benefit. Unfortunately, screening itself provides misleading feedback that always suggests it is more beneficial than it really is. After the onset of screening, clinicians note a shift in stage distribution: the proportion of cancers presenting in the late stage of development falls, even if this is simply an artifact of more early diagnosis and not fewer instances of late presentation. The proportion of late-stage cancers detected falls from 50% to 25%, despite no change in late-stage incidence.
Over time, five-year survival rises owing to the combined association of lead time and over-diagnosis bias, even if the age of death is unchanged. Survivor stories are particularly pernicious: the more over-diagnosis from screening, the more people there are who believe they owe their life to the test – and the more popular screening becomes.
Screening campaigns routinely make use of this misleading feedback; they point to higher survival rates and cancer survivors as evidence supporting screening.
Cancer screening is typically motivated by a genuine belief in its value, but it has also become an important revenue stream for volume-driven medical care systems in the US. Although multiple payers are a barrier to nationwide accounting, US expenditures related to screening are substantial: my back-of-the-envelope estimate is $40 to $80bn per year.
These expenditures represent revenues for the system, revenues not only from screening itself but also from the diagnostic and therapeutic services it triggers. The importance of this revenue stream was highlighted after the substantial decline in screening that occurred after the onset of the COVID-19 pandemic. Although other services and businesses remained closed, screening was rapidly restored within a few months.
Crile believed that medical care should be driven by patient needs, not surgeon needs (or now, system needs). He recognised there was a price to be paid for getting ahead of symptoms.
Although cancer screening may make sense in selected high-risk individuals, I believe general population screening, as currently practised in the US, has become a huge distraction to our core work. It distracts the system away from acutely ill and injured patients: as physician performance is measured in terms of how frequently they test the well and not how well they care for the sick.
General population screening distracts patients and policy makers away from the genuine determinants of human health. The tremendous resources involved in screening—in terms of money, people, and effort—would be better directed elsewhere.
JAMA Surgery article – Cancer screening – the good, the bad, and the ugly (Open access)
See more from MedicalBrief archives:
Blood test that screens for DNA fragments of tumour billed as ‘new frontier’ in cancer screening
SA medical schemes: Claims plummet and a dangerous lack of cancer screening
Breast cancer screening does not reduce mortality
UK review recommends earlier bowel cancer screening