When no diagnostic label is appliedBMJ 2010; 340 doi: http://dx.doi.org/10.1136/bmj.c2683 (Published 25 May 2010) Cite this as: BMJ 2010;340:c2683
- Roger Jones, Wolfson professor of general practice1,
- Kevin Barraclough, general practitioner2,
- Christopher Dowrick, professor of primary medical care3
- Correspondence to: R Jones
A 38 year old woman presented with “indigestion type” symptoms. She had a long history of frequent visits to her general practitioner (GP) and investigations for unexplained pelvic pain, dizziness, and headache. On this occasion she felt a “burning” sensation “in a band across my tummy” and felt as though the ground was unsteady when she walked. Her GP could reach no diagnosis and said he believed her symptoms were likely to resolve, advising follow-up in a few weeks.
A “diagnostic label” in general practice is generally a working diagnosis that functions as a decision node. It is different from the histologically or microbiologically proved diagnoses made in secondary care and tends to take one of three forms:
A working diagnosis on which treatment is based (such as “acute otitis media”)
A working diagnosis on which further investigations are planned (such as “bloody diarrhoea ? inflammatory bowel disease”)
A working diagnosis indicating the absence of serious disease (such as “calf pain, not DVT”).
The aim of this article is to understand why diagnostic labels are sometimes not applied in general practice consultations and to understand how this approach contributes to making decisions about diagnoses.
Why is a diagnosis sometimes missing?
Bruce Thomas, a general practitioner in Hampshire, UK, pointed out in the 1970s that in up to 40% of patients in general practice no diagnostic label could be attached.1 2 These patients did not require, and generally were not given, specific treatment, and most recovered spontaneously. This remains the case. In a 2005 review of consultation data in the UK, “symptoms, signs and ill-defined conditions” was the computer coded group of disorders presenting most frequently in general practice.3
Also in the 1970s, Howie described “the relative rarity of fully developed hospital illness” in general practice. He pointed out that using the same diagnostic terminology in general practice as was used in hospitals could sometimes be misleading because of the “different spectrum of disease” and the different level of diagnostic certainty in primary care.4 In primary care, diagnostic labels can give spurious and erroneous diagnostic precision when what is being managed is what Howie described as a symptom:sign complex (such as non-cardiac chest pain with mild localised chest wall tenderness).
When is management without a diagnostic label used?
For many patients in general practice, the diagnostic strategies described by Heneghan et al (figure⇓) are appropriate.5 When pattern recognition and iterative diagnosis6 do not give an answer, when symptoms are vague, or when a clear diagnosis does not “crystallise” from investigations or tests of time or of treatment, patients will need to be managed without having a diagnostic label. In not applying a diagnostic label the clinician may be “leaving the door open” for further investigations if symptoms persist. Alternatively, the lack of a label may be part of letting the patient know that, as far as can be ascertained, their symptoms do not add up to anything substantial or serious. Many patients to whom no diagnostic label can be attached can be managed without treatment or further investigation.
In the 1980s Thomas examined patient satisfaction in 200 consultations with patients who had symptoms but no abnormal physical signs. The patients were randomised to receive either a “positive” consultation (when they were given a firm diagnosis and told they would be better in a few days) or a “negative” consultation (in which no specific diagnosis or assurance was given), and also to receive either thiamine hydrochloride as a placebo or no treatment. Two weeks after the consultation 64% of those who were managed with a positive consultation reported being recovered as opposed to 39% of those given no assurances.7 The placebo drug treatment made little difference. This is consistent with a recent review of the placebo effect, which attributed the efficacy of the placebo to the overall content of the therapeutic interaction (doctor, patient, and context) rather than to any psychological effect that is specific to the (therapeutically inert) pill.8 Thomas’s study suggests that taking a positive attitude in the consultation may accelerate patients’ spontaneous recovery from a vague symptom complex. Many patients without an obvious diagnostic label could be managed in this way.
The remaining patients, those with persistent symptoms still requiring investigation, contain three subgroups:
Those with important diagnoses that have not yet been made, which may be rare or are easily missed,9 such as Addison’s disease, coeliac disease, and cancer of the ovary or pancreas
Those with new or emerging conditions, such as obstructive sleep apnoea or chronic cough from gastro-oesophageal reflux disease (many of whom were previously labelled as having “psychogenic cough”)
Those with persisting symptoms whose physical basis is unclear.
The last (and largest) group will often have been extensively investigated for poorly defined, often painful, conditions that are often associated with apparently disproportionate effects on mood and general wellbeing. These include patients with so called medically unexplained physical symptoms (MUPS), which comprise conditions such as dizziness, temporomandibular joint dysfunction, fibromyalgia, pelvic pain, “functional” abdominal pain, dyspepsia, and non-cardiac, non-reflux chest and thoracic pain. These patients present particular challenges for long term management, because they generally attend often and generate high healthcare costs, through over-investigation and inappropriate referral. From time to time clinicians may encounter patients with definite physical symptoms, such as haematuria or haemoptysis, in whom no definite diagnosis can be made. In a cohort study of patients with alarm symptoms in primary care, GPs did not make a diagnosis in over 60% of 37 000 patients.10
The following week, the 38 year old patient consulted her out of hours service and was given a prescription for a proton pump inhibitor for a presumed diagnosis of indigestion. Next morning she telephoned the surgery asking for a sick note as she was unable to get to work.
How does diagnosis go wrong?
Misdiagnosis can occur in two directions. Firstly, the doctor may conclude incorrectly that a patient does not have a condition to which a diagnostic label should be attached, or for which treatment is needed, and hence allow undiagnosed symptoms to persist, such as gastro-oesophageal reflux causing “non-cardiac” chest pain or chronic cough. Conversely, the doctor may confer a formal diagnosis on the patient’s symptoms, despite the absence of adequate evidence of a recognisable medical disorder, such as making a diagnosis of vertebrobasilar insufficiency in a patient with vague unsteadiness. A sense of pressure to make a specific diagnosis can come from patients, from fear of litigation if a disease is missed,11 or from the expected role of general practitioners as identifiers of disease.12
Wrong diagnoses in primary care can be caused by atypical and non-specific presentations, very rare conditions, and comorbidity, and also perceptual factors such as failure to observe subtle changes in colour or contour or abnormal swellings13 or gradual changes over time, as sometimes happens with patients with hypothyroidism or acromegaly. Other causes are shown in the box.
Causes of wrong diagnoses in primary care
Cognitive oversight—simply not thinking of the correct diagnosis14—for example, forgetting coeliac disease as a cause of iron deficiency anaemia
Failure to gather adequate data—for example, inadequate physical examination for lower bowel symptoms caused by colorectal cancer
Misinterpretation of data—for example, diagnosing gout on the basis of a raised serum urate concentration or excluding it on the basis of normal serum urate.
Anchoring—sticking to an initial diagnosis despite disconfirming evidence,15 such as treating fatigue as depression despite evidence of abnormal renal function
Inappropriate confirmation—selective use of evidence to confirm an incorrect diagnosis, such as attributing importance to minor abnormalities in laboratory tests as an explanation for fatigue in someone with depression
Premature closure—arriving at a conclusive diagnosis before collecting all the data,16 such as diagnosing intermittent (vascular) claudication in a patient with lumbar canal spinal stenosis
How can we improve?
Clinicians need to understand the various ways in which diagnostic labels should and should not be applied to patients in general practice, and they should be prepared to deal with the potential causes of diagnostic error listed in the box. This has important implications for medical education: performing clinical investigations and looking for symptoms and signs need to be taught with a probabilistic approach to differential diagnoses. Spuriously accurate clinical diagnoses need to be discouraged, keeping in mind that patients benefit from a certainty of approach.
Managing unexplained symptoms requires a high quality relationship between doctor and patient, characterised by a sympathetic and empathic inquiry into the psychosocial background of symptoms as well as the provision of considerable emotional support. Patients with medically unexplained symptoms are no more likely than those with explained symptoms to seek explanation, reassurance, or somatic interventions such as special investigations, drug treatment, or specialist referral—but they are much more likely to seek emotional support from their clinician.17 They are also more likely to be cautious about the extent to which clinicians will take their problems seriously.18 19 Clinicians may manage these patients more effectively if they explain the limited, and potentially negative, value of repeat investigations, and if they avoid premature closure by regularly reviewing previous diagnostic labels or apparent lack of such labels. Clinicians may avoid unfounded diagnoses and subsequent somatic interventions by getting patients to talk about their social and psychological problems.20 21 Brief educational interventions, including didactic sessions, small group work, and role play with video feedback can increase GPs’ awareness of medically unexplained symptoms.22
The general practitioner asked the woman to return for review. A careful history elicited a description of a superficial “burning sensation” across the lower chest and increasing difficulty in walking. A Romberg test was positive, and she had increased tone and hyper-reflexia in her legs. Admission and magnetic resonance imaging of her neck showed a central cervical disc prolapse causing a myelopathy.
Some patients who present in primary care cannot be given a definite diagnostic label
Not giving a diagnostic label may indicate to the patient that nothing serious is wrong or may leave the door open for further investigation and clarification of symptoms
A definite, final diagnosis cannot be made in a large proportion of patients presenting with physical symptoms
Patients falling into the spectrum of medically unexplained physical symptoms (MUPS) are likely to require particularly skilful management
Cite this as: BMJ 2010;340:c2683
This series aims to set out a diagnostic strategy and illustrate its application with a case. The series advisers are Kevin Barraclough, general practitioner, Painswick, and research fellow in community based medicine, University of Bristol; Paul Glasziou, professor of evidence based medicine, Department of Primary Health Care, University of Oxford; and Peter Rose, university lecturer, Department of Primary Health Care, University of Oxford.
Contributors: RJ wrote the first draft of the paper, which was extensively modified by CD and KB before being completed by all the authors. RJ is guarantor.
Competing interests: CD has two grants to study MUS in primary care with the Medical Research Council.
Provenance and peer review: Commissioned; externally peer reviewed.