Medical diagnosis is the process of determining which disease or condition explains a person’s symptoms and signs. The diagnostic challenge that drives work in teaching hospitals is certainly one of the appealing aspects of medicine and no doubt that challenge can be very rewarding, and especially so if effective treatments are available. However, to apply the diagnosis and treatment model to everyday primary care inevitably leads to ‘over-diagnosis’. For it has to be said that many (perhaps most?) patients who come to GPs bring problems and concerns that don’t easily conform to this labeling process. They may be sad or distressed people mired in insoluble predicaments, people with social problems who have nowhere else to turn, or very often people who find it difficult to cope day-to-day with longterm mental or physical conditions. These are not patients who fit convenient diagnostic boxes. But lacking the time and skills to support and advise these needy folk, it may in the short term be easier just to bamboozle ourselves with a ‘diagnosis’ and to offer the sop of a ‘treatment.
But if, as Dr Freud told us, ‘what is not resolved must be repeated’, patients like these will inevitably keep returning. Another way of making work for ourselves is by diagnosing ‘disease’ that will never cause symptoms or death. Although routine screening for early disease can save lives (eg blood pressure measurement), crucially, not every preclinical disease will become clinical disease. And some forms of screening are frankly ineffective and serve only to turn people unnecessarily into patients and/or to their being prescribed treatments that do no good, and may do harm. Ironically the pursuit of longer life through ‘better’ diagnosis leads to increasing numbers of people being labeled as diseased. And arguably, in countries such as America, the healthcare funding system itself encourages unnecessary investigations and unwarranted interventions; something that is borne out by research in the US suggesting that virtually every family in the country has been subjected to some form of over-testing and overtreatment (Gawande 2015). Yet over-diagnosis is not necessarily a new phenomenon: in the 1970s in Middlebury, in the state of Vermont in the US, 7% of children under the age of 16 had their tonsils removed, and in Stowe, also in Vermont, 70% of children had the operation, despite the two communities being demographically similar (Brownlee 2007). At the time of course tonsillectomy was almost universally agreed to be a necessary and effective ‘treatment’ for children subject to recurrent upper respiratory illness. Nowadays of course we know the procedure is almost always unnecessary and ineffective. H Gilbert Welch, a Dartmouth Medical School professor, is an expert on over-diagnosis, and in his book, Less Medicine, More Health, he explains that we have assumed, for instance, that cancers are like rabbits which you want to catch before they escape the barnyard pen. But some cancers are more like birds – the most aggressive have already taken flight before you can discover them, which is why many people still die from cancer, despite early detection (Welch 2015). And other forms are more like turtles: they aren’t going anywhere, so removing them won’t make any difference. We have had to learn these lessons the hard way over the past two decades. In that time the number of thyroid cancers detected and removed in the US has trebled, yet the death rate from thyroid cancer remains unchanged. In South Korea, widespread ultrasound screening has led to a fifteen-fold increase in detection of small thyroid cancers, making it the most often diagnosed and treated malignancy in that country. But, as Welch points out, the death rate has not dropped. Meanwhile, the number of people with permanent complications from thyroid surgery has skyrocketed. We’re just catching the turtles, says Welch. The whole exercise is just over-diagnosis. The media have a very powerful influence on patient opinions and expectations. Recently a ‘celebrity’ on primetime television told how horrified he was when, having suffered prostate cancer, his son was refused a PSA test when ‘most prostate cancer is hereditary and his uncle was also suffering from symptoms of prostate cancer’! He then quoted the price of a PSA blood test which resulted in the host of the programme describing the decision not to test the son as ‘shocking’. The programme triggered a spike in demand for random PSA testing even though prostate cancer is not hereditary, and the rate of false positives is rather high, and evidence suggests that harms of PSA testing for prostate cancer in asymptomatic men can outweigh its benefits (ECC 2013). In an ideal world, GPs would have ample time to discuss all this in detail, underlining the pros and cons of prostate screening with patients who, fully obedient to the new terms of engagement in family medicine, had only ever presented with one problem during a single ten-minute consultation. In reality of course, GPs often run out of time for this and so give in to a patient demanding testing without fully considering the consequences with them. There are many drivers of over-diagnosis in a dysfunctional and industrialised system: the use of increasingly sensitive diagnostic tools like MRI scanning, financial incentives to diagnose, or the commercial interests of large pharmaceutical companies. Doctors’ fear of litigation if they miss a diagnosis leads to more defensive medical practice, especially when they are ever more urged on to follow rigid diagnostic/treatment guidelines. But it is well known that the diagnostic thresholds shift as panels of health professionals (who sometimes have financial ties to companies that benefit directly from any expansion of the patient pool) change the criteria; a practice termed ‘disease mongering’ (Moynihan 2011). The potential for dysfunction in the NHS can be judged by two recent policies which are in direct contradiction: first it was proposed to name and shame GPs who miss cancer diagnoses (Pulse 2014) but this was closely followed by another Whitehall pronouncement offering GP practices incentives to cut the number of referrals they make including suspected cancer referrals (Pulse 2015)! The soaring increases in the numbers of patients diagnosed as having high cholesterol is widely debated, because there is scant evidence that prescribing statins en masse reduces all-cause mortality. Meta-analysis of the many existing statin trials found no benefit (Ray KK et al 2010). A parallel rise in the number of people diagnosed with depression may have been driven more by the availability of ‘more effective’ antidepressants than by NICE guidelines and payment for diagnosis (as has been the case for statins). Undoubtedly the growing temptation for overstretched GPs to provide a prescription rather than a listening ear has made them favour the pharmaceutical quick fix. GPs need time to explain the benefits and downsides of drugs and screening, so that testing can be used effectively and treatment options explored in patients’ best interests. The resulting double whammy could lessen the burden on patients’ who would otherwise have to live with a questionable ‘diagnosis’, while also avoiding unnecessary tests and reducing the likelihood of potential harm from unwarranted treatments. But to achieve these worthy aims and so free up NHS resources for people who really do need treatment, more GP time would have to be available. The increased demands on our NHS have led to radical ways of managing demands on the system. The algorithm-driven 111 system is possibly one of the least efficient examples of what happens when you apply an industrial flow-line model to triage and diagnosis. It is inefficient because it is designed so that non-medical staff who lack clinical experience can assess the risk of callers’ symptoms. Consequently far too many who ring 111 for advice are told to see a GP or frequently get sent to A&E. For instance I saw two patients who had been triaged by 111 and sent (in an ambulance) to A&E with pains in the chest. They were discharged by A&E many hours later and told to see their GP. Happily, they managed to get same day appointments with me. Both were young and otherwise healthy and one of them – who had been lifting heavy objects – was describing classical musculoskeletal pain. All the same, once on the A&E conveyor belt the staff had to run an ECG on him, take a chest X-ray and do serial troponin blood tests – which would have been part of a strict triage guideline for anyone with acute chest pain even if their ECG showed no sign of heart damage. This of course meant a lot of waiting around in A&E because only a second blood test at least three hours later can completely rule out a myocardial infarction (heart attack). The other lady went through the same drawn out process. When I’d finished seeing and reassuring the young man I asked, ‘Is there anything troubling you?’ and he, understandably responded ‘Yes. I’m concerned they kept me in hospital for so long’! On the other hand and in contrast to this investigatory overkill, this morning I was asked if an extra could be added to my list at the end of morning surgery. In came a 67-year-old man who had collapsed. Though they had called 111 his symptoms had clearly not triggered the ‘chest pain’ or another carved in stone algorithm so they were advised to see a GP within a few hours. When I saw him he could barely walk, he had a temperature of 35 degrees, crepitations at both lung bases and subsequently a white cell count of 20,000. A very ill man indeed, and someone who certainly did need urgent treatment. Speaking at a fringe session at the Conservative Party Conference this year, Health Secretary Jeremy Hunt informed delegates that Silicon Valley tech ‘gurus’ were confident that diagnosis by humans would be obsolete within two decades. This absurd statement illustrates the sheer lack of understanding that is driving the dysfunctional industrialisation of our health system. The future is uncertain and uncertainty is a fundamental aspect of life and medicine. If it has to be managed by computers and guidelines, then the vast grey areas of possibilities circling around what is actually causing a patient’s symptoms will all too often lead to over-investigation and over-diagnosis. Wise, holistic and insightful diagnosis (and sometimes masterly inactivity) depends on the sorts of pattern recognition and complexity that only humans and human relationship can provide. Good medicine transcends algorithms, computers and guideline systems. If we want a humane and cost-effective NHS, then we must defend their place in it.
- Brownlee S (2007) Overtreated. New York: Bloomsbury
- Davis J (2014) GPs to be ‘named and shamed’ after missing cancer diagnoses. Pulse, 30 June
- European Cancer Congress (2013) Organised screening for prostate cancer using the prostate-specific antigen test, does more harm than good. Press Release: Moynihan R (2011) A new deal on disease definition. BMJ; 342:d2548.
- Gawande A 2015) Overkill: An avalanche of unnecessary medical care is harming patients physically and financially. What can we do about it? The New Yorker, May 11.
- Ray KK et al (2010) Statins and all-cause mortality in high-risk primary prevention: a meta-analysis of 11 randomized controlled trials involving 65,229 participants. Arch Intern Med, June.
- Welch HG (2015) Less medicine, more health; 7 assumptions that drive too much medical care
- Matthews-King A (2015) A moral dilemma: GP practices offered incentives to cut urgent referrals. Pulse. 1 October.