By Marcello Cherchi, MD PhD
When you are not sure
The following are metacognitive comments on how to approach diagnostic uncertainty. They are not unique to otoneurology. While many of these will probably appear obvious to the reader, breaking down the causes of uncertainty may make it easier to recognize which type of problem holds in a given situation.
What leads to diagnostic uncertainty?
The practice of clinical medicine frequently requires working with incomplete information, and this can introduce uncertainty in various ways.
Examples of misdiagnoses fuel grand rounds, and morbidity and mortality conferences, and could fill libraries. The specifically otoneurological examples described below are chosen simply to illustrate broader concepts.
Information is unavailable
This can happen in a variety of circumstances, such as:
- Previous records are unavailable.
- The current clinical setting is not outfitted to undertake an adequate workup (e.g., diagnostic equipment is lacking).
- The current clinical setting does not facilitate a workup (e.g., telemedicine limiting physical examination).
Strategy: change the setting if possible; or refer.
Information is available, but is incorrect
Basically, a wrong test result, such as:
- Example: unilaterally absent vestibular evoked myogenic potentials (VEMP) test that is due to technical failure (burned-out ear but inserts) rather than to actual pathology (such as vestibular weakness).
- Example: a rotatory chair study in which step velocity testing shows low gain but normal time constants due to a calibration error.
Strategy: Everything in medicine is fallible. If a test result is glaringly discrepant with aspects of a patient’s history, examination and other workup, then it is sensible to repeat the test.
Information is available, and is correct, but is irrelevant
- Example: In an 80-year-old with episodic, positionally-triggered spinning disequilibrium, symmetrically absent ocular vestibular evoked myogenic potentials (oVEMP) are probably not informative since VEMP responses symmetrically decline with age.
- Example: A patient with orthostatic hypotension shows an aphysiologic pattern on computerized dynamic posturography (CDP); but this test is not designed to detect orthostatic hypotension, so the “aphysiologic” result here is meaningless.
Strategy: Beware of red herrings. Not all “abnormalities” are clinically meaningful.
Information is available, and is correct, but is insufficient.
- Example: Outside video head impulse testing (vHIT) was normal, so vestibular neuritis (VN) was “ruled out.” More complete workup (videonystagmography, vestibular evoked myogenic potentials) showed lateralizing weakness compatible with vestibular neuritis (VN).
- Example: Outside cervical vestibular evoked myogenic potentials (cVEMP) was unilaterally absent, suggesting vestibular weakness. Subsequent audiogram showed substantial conductive hearing loss on that side, suggesting that the absent cVEMP was due to middle ear pathology rather than to true vestibular disease.
Strategy: Undertake a reasonably thorough, appropriately targeted workup, including instrumented testing if necessary.
Information is available and is correct, but was interpreted incorrectly.
- Example: Outside VNG showed “bilateral weakness,” though the report failed to note that air calorics were used. (Water caloric testing is a weak stimulus and results may appear falsely positive for bilateral vestibular weakness; air caloric stimulation is an even weaker stimulus, and thus even more prone to be falsely positive.)
- Example: A patient suffers brief (seconds-long) episodes of disequilibrium triggered by sneezing or coughing. Outside cervical vestibular evoked myogenic potentials (cVEMP) showed lateralizing “weakness” that was diagnosed as “vestibular neuritis (VN)” (even though that does not fit the clinical history) and the patient was not responding to physical therapy for VN, but actually the other side showed a larger than normal response. Repeat testing also showed low thresholds on the side with the larger response, suggesting superior semicircular canal dehiscence.
Strategy: Review original data yourself, when available.
Information from the history is changing: History is not clear because a gradually evolving disease has not yet manifested with all its symptoms.
- Example: A patient has a first episode of vertigo lasting hours, without any aural symptoms. If subsequent vertiginous episodes are accompanied by aural symptoms (hearing loss, tinnitus, fullness), then Meniere’s may be suspected.
- Example: Gradually progressive unilateral hearing loss without disequilibrium. If subsequently the patient also develops symptoms of gradual unsteadiness, then a structural lesion, such as a vestibular schwannoma, should be considered.
Strategy: Consider interval follow-up to ascertain whether the history changes.
Information from physical examination or testing is changing: Physical examination or test results are not yet clear because a gradually evolving disease has not yet manifested all its clinical signs.
- Example: a parkinsonian syndrome initially resembles idiopathic Parkinson’s disease, and ocular motor findings only showed mild convergence insufficiency. When ocular motor testing was repeated 18 months later, vertical saccades exhibited clearly reduced velocity, suggesting progressive supranuclear palsy (PSP).
- Example: an older man suffers a small cerebellar stroke resulting in mild unsteadiness, and initial examination only shows modest upper extremity dysmetria ipsilateral to the stroke. Over the next few months he gradually develops constant though fluctuating oscillopsia, and hears a constant clicking sound; exam shows pendular nystagmus synchronous with rhythmic oscillation of the soft palate; he has developed oculopalatal tremor.
Strategy: Consider interval follow-up to detect if physical examination or instrumented test results change.
The practitioner’s knowledge is insufficient to recognize a particular disease’s profile.
A common aphorism in medicine is, “You find what you look for, and you look for what you know” (Molini et al. 2010) and similar quotes (Harkless and Dennis 1987) — in other words, you will only ever find what you know to look for in the first place. Thus, if one’s lack of knowledge is the principal limiting factor to identifying a diagnosis, then the solution should be to expand one’s knowledge.
Strategy: Keep learning. Artificial intelligence is beginning to help with this.
Absent information: Disease has not yet been characterized.
All diagnoses start somewhere and at some time. At risk of appearing more pedantic than this section already is, consider:
- List of diseases discovered by year
Strategy: Keep learning, stay abreast of new findings in the field.
Symptomatic management
The likelihood of managing a patient’s symptoms appropriately will increase if the diagnosis (underlying cause of symptoms) is known, so in our view the process of securing a diagnosis should initially have a very high priority.
But given the limitations enumerated above, there will be cases in which you, the practitioner, are uncertain of the diagnosis, and this complicates selecting a management strategy. In such a circumstance, a clinician may offer “symptomatic management” which – though less than ideal – is an attempt to alleviate symptoms, even in the absence of a clear diagnosis.
There are some obvious benefits to symptomatic management. First, if symptoms improve, then a patient will get some relief. Second, there are some situations in which response (improvement) to a treatment may allow one to infer a specific diagnosis (sometimes referred to by the Latin phrase, ex juvantibus, meaning “from that which helps”). An example would be parkinsonian symptoms that promptly and significantly improve on levodopa-carbidopa, which strongly suggests idiopathic Parkinson’s disease (as to one of the other parkinsonian disorders).
There are also, of course, risks in treating when the target disease is unclear. Given the diagnostic uncertainty, it is advisable to select “symptomatic management” strategies that are as low-risk as possible. For example, if a patient’s history is strongly suggestive of BPPV but their ocular motor examination is normal, or if a patient’s history is strongly suggestive of VN but their examination and tests are normal, then a trial of vestibular rehabilitation therapy (VRT) is reasonable and low-risk.
Offering symptomatic management does not mean that one must give up on the goal of securing a diagnosis. Follow-up is often helpful in this regard, in case symptoms evolve, or physical examination findings change.
References
Harkless LB, Dennis KJ (1987) You see what you look for and recognize what you know. Clin Podiatr Med Surg 4: 331-9.
Molini L, Ciortan E, Bianchi S (2010) Bilateral elastofibroma dorsi: A case report. J Ultrasound 13: 199-201. doi: 10.1016/j.jus.2010.10.011
