Diagnosis of Rare Diseases
Issues in Current Practice
Diagnostic error in medicine is common. For example, a study from an intensive care unit demonstrated nearly 20% discordance between the clinically-defined cause of death and findings at postmortem examination. Not surprisingly, therefore, the diagnosis of rare diseases is often delayed. Errors in diagnosis have been divided into three main categories.
'No fault errors' include misdiagnosing a rare condition that mimics a common one, and misdiagnosing a condition presenting in an atypical fashion.
'System errors', in which organisational deficiencies delay or negate the correct diagnosis, include factors such as faulty or inadequate equipment.
'Physician cognitive errors': clinicians make an initial diagnosis based on how closely their perception of available evidence (such as history and examination) matches typical presentations, for example, thunderclap headache suggests subarachnoid haemorrhage. The probability of this initial diagnosis is then modified in a Bayesian fashion based on test results. If the test results match the expectation, the diagnosis is typically confirmed and a search for alternative diagnoses will stop. However, at least 30 types of cognitive errors can impede this process and complicate diagnosis. Clinicians may commit an 'anchoring error' by focusing on a potential diagnostic clue prematurely, which may be compounded by 'confirmation bias', whereby the clinician looks for confirmatory evidence to confirm the diagnosis at the expense of searching for evidence to refute it. 'Availability bias' may play a role in this, with diagnoses appearing higher up the differential diagnosis if they come to mind more readily. This can, in turn, be compounded by a 'search satisfaction' error when the diagnostic pathway is prematurely terminated after an abnormal result is found. Although clinicians are educated to search for a unifying diagnosis according to the principle of 'Occam's razor', an abnormal result may lead to failure to diagnose a rare disease if there is, in fact, more than one pathology. All of these cognitive errors may be particularly problematic in the diagnosis of rare diseases.
Guidance
The General Medical Council framework does not offer specific advice to manage rare diseases, but their advice relates equally to rare and common diseases. With respect to diagnosis, clinicians should provide 'investigations…as necessary' and 'good use of the resources available to you'.
Practical Tips
There are several proposed strategies to reduce cognitive errors in the diagnostic process. An awareness of the common cognitive biases is important. If a provisional diagnosis is achieved, clinicians should routinely reflect on this: 'could this be anything else?'; 'is there anything which is atypical?'; and 'does the abnormal test result explain the clinical presentation or do I need to keep searching?' Allocating an appropriate amount of time for complex diagnostic pathways is still important in an increasingly time pressured clinical environment.
If conventional investigation by first principles fails to achieve a diagnosis, there are several options. The portal 'Orphanet' (http://www.orpha.net) allows the input of symptoms to search for the diagnosis. A search string, including 'deafness', 'ataxia' and 'dementia', raised nine potential diagnoses but, as a limitation, superficial siderosis was not on the list, probably because it is the consequence of disease, rather than a primary disease.
Many countries have a well-developed tertiary and quarternary referral process for neurology, which likely facilitates the diagnosis of rare diseases. In the UK, about half of reference centres are disease based (eg, mitochondrial disease centre), with the remainder focusing on specific treatments. European reference centres can be found using the 'Orphanet' portal, with search options allowing the input of the disease (or class of diseases), the aspect of management (eg, medical management) and the country of practice to search for the appropriate reference centre. If a genetic disorder is suspected, then review by the local clinical genetics team may facilitate further counselling and advice for both patient and family. In the USA, the National Institutes of Health has an Office of Rare Diseases Research which provides accessible information on rare diseases (including links to relevant websites and trials) and stimulates research on the prevention, diagnosis, or treatment of rare diseases or conditions. Among many other activities, it runs the Underdiagnosed Diseases Program, which adopts a cross-specialty approach to the diagnosis of diseases that have eluded treating physicians, in a selected population.
After diagnosis, patients need counselling with respect to diagnosis, clinical issues, prognosis and management. This may be limited by paucity of evidence. In particular, clinicians should be aware of the inherent selection bias when evaluating case reports and series, which may skew interpretation of the severity and natural history of the disease. A good case series will have explicit inclusion and exclusion criteria, with consecutive patient enrolment, and a high follow-up rate of prospectively collected clinically relevant outcomes. Available information may be provided to patients in a variety of ways. It is important that it is timely and comprehensible, with the ability to discuss, reiterate and clarify over time, if needed. Written information including patient information leaflets may help patients with sufficient literacy skills. Disease specific information leaflets may be provided by healthcare organisations such as 'NHS choices' (http://www.nhs.uk), disease specific charities and patient support groups, and these should be assessed for appropriateness. Patients may also benefit from accessing the website run by EURORDIS (Rare Diseases Europe), a non-governmental organisation, which provides links to various patient-driven groups. Patients can also search 'Orphanet' for information on a large number of rare diseases and information about their relevant tests.