Every patient who presents with a medically unexplained symptom (MUS) will eventually die, and many of them will consult a doctor with symptoms of their final illness. This sobering thought is the reason for this chapter, which aims to highlight particular problems and pitfalls when managing functional symptoms. A long history of MUS, particularly when combined with frequent attendance, can sometimes distract clinicians from one of their core tasks—diagnosing serious illness.

The chapter aims to answer three questions: how commonly does the diagnoses of MUS need to be revised; what are the factors associated with practitioner delay in diagnosing cancer; and what are the commonest diagnostic errors made by doctors.

This chapter does not list specific sets of red flags–they are described in individual chapters—but several themes are consistent across symptoms and body systems. Bleeding is never a symptom of MUS; similarly unintentional weight loss and night sweats need investigation—sometimes extensive investigation—to look for disease.

Symptom-specific recommendations for investigations are also included in the relevant chapters. However, as a rule of thumb, most non-trivial new symptoms in a patient who has not had recent investigations warrant basic blood tests: full blood count, renal, liver, thyroid and bone chemistry and inflammatory marker—with more added as clinically indicated. There is little evidence that deferring investigations is better or worse than carrying them out on the first occasion the patient presents with potentially significant symptoms.

Surprisingly few studies have reported this. One small UK study found that in primary care, 10% of symptoms that have been present for several months and were thought to be MUS turned out to be due to organic disease. In secondary care the proportion is smaller, especially when the specialist concludes that there is a functional disorder rather than the diagnosis remaining ambiguous. A diagnosis of functional symptoms from a neurologist turns out to be wrong in only 2–3% of cases and similar proportions are probably seen by specialists in other disciplines.

New symptoms that are accompanied by anxiety are especially challenging, particularly when the patient has a past history of anxiety or panic disorder. Anxiety is one of a range of factors that may raise the practitioner’s threshold of suspicion regarding new symptoms and which may inhibit timely recognition, diagnosis and referral. This kind of parallel presentation does not mean that recognition and treatment of the psychological disorder is unimportant, rather it acts as a reminder that the two can coexist.

Practitioner delay has been studied most thoroughly in relation to cancer diagnosis and the evidence for this has recently been exhaustively reviewed. The effect of patients’ sociodemographic characteristics has a variable effect on practitioner delay.

Patient characteristics

Patient age is a factor in delayed cancer diagnosis, particularly for gastrointestinal cancers. Younger patients are at greater risk of diagnostic delay. Although this is perhaps understandable—the probability that a new disorder is functional is higher in younger patients—it is a salutary reminder of the need to consider the possibility of organic disease. Practitioners need to be alert to the possibility of patients presenting outside ‘typical’ age ranges—the young patient presenting with a familial colorectal cancer is a classic example. Diagnostic delay of urological, gynaecological and lung cancer is associated with lower educational attainment in patients, perhaps because of lower health literacy or because of greater reluctance to challenge the doctor’s (incorrect) opinion. Recent evidence based on audit of cases of cancer referrals suggests that patients who are housebound may experience longer delays; multiple comorbidity may also lead to longer diagnostic intervals. In general the more complex the ‘background’ level of symptoms, the more likely it is that a diagnosis might be delayed.

Patient healthcare behaviour

Frequent healthcare seeking and seeing multiple providers—as is the case for some patients with MUS—are associated with greater delay in diagnosis of gynaecological and colorectal cancers. It is important to remember that patients with MUS have the same risk of serious illness as those without MUS. Practitioners need to be vigilant and monitor the pattern of presentation, looking particularly for subtle changes that might signal an emerging organic illness.

Practitioner response

Diagnostic delay due to practitioner response is associated with errors of judgement, including incorrect diagnosis, or symptomatic treatment without a clear diagnosis. It is also associated with errors of procedure such as inadequate examination, failure to organise tests and failure to ensure adequate follow-up of patients or tests. Importantly, it appears that diagnostic delay is reduced—at least in gastrointestinal cancer—by following referral guidelines.

Health system factors

Factors such as short consultation times and lack of access to diagnostic investigations can also lead to prolonged diagnostic intervals. In primary care we typically place great store in continuity of care—that is, seeing the same doctor on a regular basis. Although the benefits of continuity of care have been well described, there is at least anecdotal evidence that sometimes a ‘fresh pair of eyes’ can shed a different light on a difficult diagnosis. There is probably a case for encouraging long standing MUS patients with complex symptoms to see more than one practitioner over prolonged periods. The gatekeeper role of primary care is also widely supported yet we should keep an open mind about whether it might itself lead to delays in diagnosis; indeed there is some evidence that countries with strong gatekeeper systems have longer intervals to a diagnosis of cancer.

Apart from the work on cancer, there has been relatively little research on diagnostic errors specific to primary care. However, more general work on errors has been carried out, especially in the USA. Although the relative incidence of errors may not be transferrable to UK primary care it is nonetheless worthwhile examining the common errors.

The commonest error in several series is failure to consider the diagnosis. There are several possible mechanisms for this and the cognitive processing errors that underpin these are described below.

Other common causes of diagnostic error include failure to order tests (either by not ordering or through logistical error) and difficulties with interpretation of results (including false negative results). Less common, although still important, are errors in history taking (failure to elicit the critical piece of information) and examination (omitting the critical element). Errors of judgement between two diagnoses occur but are not among the most common errors reported by doctors. Strikingly, in this and other studies of medical error, lack of knowledge is rarely the main problem.

Misdiagnosis is the most common factor in medical litigation cases in primary care. It is rare for such cases to identify significant knowledge deficits among practitioners; more typically misdiagnosis is found to be associated with poor communication, procedural errors, and failure to consider more serious diagnoses in the background of multiple, vague, or atypical symptom presentations.

Practitioners typically use a hypothetico-deductive model in reaching diagnoses. This model relies on selective enquiry as various avenues of diagnosis are explored until the practitioner is satisfied he/she has reached a conclusion that matches the presentation. Of course, this relies on quite complex cognitive processes and many errors appear to be underpinned by problems in the way clinicians process information. These are human characteristics that have been classified as cognitive processing errors. Awareness of these errors may help clinicians recognise when they are in danger of making them.

Premature closure

This underpins the common diagnostic error of failing to consider the diagnosis. It relates to the point at which the clinician switches from searching for possible diagnoses to deciding that there is sufficient evidence to proceed with the best candidate and stop searching for more information. Interestingly age and experience have little effect on premature closure and it appears to be a characteristic of some doctors’ problem-solving style.

Availability bias

People tend to overestimate the frequency of easily remembered events and underestimate the frequency of ordinary or uninteresting events. Unusual clinical cases are more memorable than routine ones and so may lead doctors to overlook the ordinary and unremarkable diagnoses. Availability bias is one of the reasons doctors are repeatedly taught that ‘canaries’ are usually just ‘sparrows’.

Representativeness bias

Clinicians naturally try to fit cases to the most typical condition. Although this seems like an efficient pattern-matching approach, it often operates independently of rules of probability. This has two implications: first if the best-fitting diagnosis is a rare condition and a nearly fitting diagnosis is common, then the nearly fitting common diagnosis is more likely, but representativeness bias will argue the other way. Second, when one feature (for instance a red flag symptom) is strongly indicative of a serious condition but nothing else quite fits, the doctor may ignore it when the remaining symptoms fit better with an alternative diagnosis.

Anchoring and conservatism

As clinicians build up the evidence in order to solve a diagnostic problem, the natural tendency is to stick to the first hypothesis and test information against this. This ‘anchoring’ on the first hypothesis leads to conservatism as new information is gathered. In turn, new information that fits the anchor is more likely to be retained whereas that which points to another diagnosis will be ignored or discarded.