Prevalence of vestibular migraine in an otolaryngologic clinic: Preliminary clinical diagnosis versus diagnosis according to the strictly applied Bárány criteria

1 Background

The concept of episodic vertigo as a manifestation of migraine has evolved over the past decades [8]. This type of vertigo had been termed as migraine-associated vertigo/ dizziness, migraine-related vestibulopathy, and migrainous vertigo. Lack of universally accepted definition had led the Bárány Society to develop the diagnostic criteria for vestibular migraine (VM) and probable VM [8].

When a history of vertigo and migraine coexist, VM is likely to be at the top list of differential diagnoses. However, it is difficult to diagnose VM by checking the criteria one by one in all cases during the outpatient clinic. Bárány Society suggested four criteria for VM, from A to D. Criterion A identifies four characteristics of vertigo, including the vestibular symptoms according to the definition of the committee, number of episodes, intensity, and duration. Criterion B confirms a history of migraine according to International Classification of Headache Disorders (ICHD). Criterion C identifies seven migraine features during vertigo in at least 50% of episodes, and criterion D states that it must not be better accounted by another vestibular or ICHD diagnosis. ICHD definition for migraine without aura consists of eight criteria, including number of attacks, duration, four migraine features, and two accompanying symptoms [1]. ICHD definition for migraine with aura lists eight criteria, including aura symptoms and aura characteristics [1]. Therefore, a total of 27 items should be reviewed for the diagnosis of VM as defined by the Bárány Society. For this reason, even experienced specialists can make errors in making diagnosis in a busy outpatient setting. Even if the medical records are thoroughly reviewed, some items for the criteria could be missing due to insufficient interview and the complexity of the criteria may cause confusion in making precise diagnosis.

Many authors claim that VM is the most frequent entity of episodic vertigo and the second most common cause of dizziness [4, 5], although, in the authors’ experience, the prevalence seems not so high when the Bárány criteria are precisely applied.

Existing studies on the prevalence of VM were mainly conducted for European or North American patients complaining of migraine or vertigo. Many of them applied the previous Neuhauser criteria [10, 15]. Fewer studies have been reported on the prevalence of VM according to the Bárány criteria in East Asians.

Therefore, this study aimed to investigate the prevalence of VM according to strictly applied Bárány criteria in new patients complaining of dizziness who visited an otolaryngology department of a university hospital in Korea. The diagnosis was compared with preliminary clinical diagnosis of VM assessed by the otology specialists during outpatient clinic.

2 Methods

This study is a retrospective analysis of dizzy patients who visited the otolaryngology department from December 1, 2018, to November 30, 2020. Our institute has been using a vertigo questionnaire designed to determine whether the symptoms meet the criteria of VM as defined by the Bárány Society (Supplemental material). For new patients complaining of dizziness, the questionnaire was completed through interview conducted by an otolaryngologist and recorded in the electronic medical record (EMR). Medical records of patients with dizziness- or headache-related disease codes from the Korean Standard Classification of Disease (KCD8) were retrieved from the Clinical research Data Warehouse (CDW), an in-hospital big data system. Among those, only new patients who completed the questionnaire were analyzed.

Using Structured Query Language (SQL), demographic characteristics, contents of questionnaire, and preliminary clinical diagnoses were collected. Records of multiple-choice response were stored as numeric values and descriptive entries were stored as text values in a comma-separated value (CSV) file.

For example, characteristics of dizziness was encoded as “1” for spinning, “2” for imbalance, “3” for light-headedness, “4” for oscillopsia, and “5” for other. The frequency of dizziness was encoded as “0” for less than five times and “1” for more than five times. The duration of dizziness was initially stored in seconds, minutes, hours, and days, converted as “1” if between 5 minutes and 72 hours, and “0” for otherwise. The intensity of dizziness was encoded as “1” for mild, “2” for moderate, and “3” for severe cases.

If there was a current or previous history of headache, the questionnaire checked if the headache met the ICHD-3 criteria of migraine [1]. Headache attacks was encoded as “0” for less than five attacks and “1” for more than five. The duration of headache was encoded as “1” for less than 4 hours, “2” for 4–72 hours, and “3” for more than 72 hours. Headache characteristics such as unilateral location, pulsating quality, moderate or severe pain intensity, and aggravation by or causing avoidance of walking or climbing stairs were checked and each item was encoded as “1” if the symptoms were present. Nausea/ vomiting or photo-/ phonophobia during headache was also encoded as “1” if present. Visual aura lasting 5 to 60 minutes before the onset of headache were encoded as “1”.

Migraine features during vestibular episodes were checked and each item was encoded as “1” if more than half of dizziness episodes were accompanied by migrainous headache, photo-/phonophobia, and visual aura.

3 Results

A total of 955 new patients completed vertigo questionnaire through medical interview conducted by an otolaryngologist.

According to the medical charts, multiple differential diagnoses were frequently made for dizzy patients. BPPV (205 cases confirmed via Dix-Hallpike test and head roll test, 140 cases of suspicious BPPV) was the most common diagnosis followed by benign recurrent vertigo (134 cases) and VM (111 cases). Other diagnoses were Menière’s disease, non-specific dizziness, vestibular neuritis, benign paroxysmal vertigo of childhood, and persistent postural-perceptual dizziness in order of frequency. Although not classified as preliminary clinical diagnoses of VM, there were 35 cases of dizziness with headache and 28 cases of dizziness with alleged history of migraine.

In detail, 356 patients complained of accompanying or previous headache, which was 37.3% of the total dizzy patients. Of these, 68 patients (19.1%) met criterion A, 39 (11.0%) met criterion B, and 113 (31.7%) met criterion C. Out of all new patients with dizziness, only 2.9% were diagnosed with VM according to the Bárány classification, with six patients with VM (0.6%) and 22 patients with probable VM (2.3%). The average age of the patients was 43.7 years, and the male to female ratio was 1:3.

A preliminary clinical diagnosis of VM was made in 111 patients (11.6%). Among those, 32 patients (28.8%) met criterion A, 32 (28.8%) met criterion B, and 62 (55.9%) met criterion C, with only 17 patients (15.3%) meeting the criteria of Bárány Society.

Vestibular migraine was not the preliminary diagnosis for 11 of 28 patients who met the criteria of the Bárány society. In those cases, the initial diagnoses were persistent postural-perceptual dizziness (PPPD) in two cases, benign paroxysmal vertigo of childhood in a case, rule out resolved BPPV in a case, and rule out Menière’s disease in a case. The remaining six cases were assessed as recurrent orthostatic/ positional vertigo.

4 Discussion

This study was conducted on new patients who visited the outpatient otolaryngology department at a university hospital in Korea for dizziness. It was shown that 37.3% of the patients with dizziness had a history of headache. According to medical records, VM was the third most common preliminary diagnosis accounting for 11.6% of dizzy patients, following BPPV and benign recurrent vertigo. Although not assessed as VM, another 6.6% of dizzy patients were assessed as dizziness with headache or dizziness with alleged history of migraine.

Strict application of the Bárány Society criteria significantly reduced the prevalence of VM to 2.9%. This is very low compared to the previous studies which reported prevalence of 9.9 –41% in dizzy patients [12]. The prevalence in our study is somewhat close to the annual incidence of VM in the general U.S. adult population, which was 2.7% [5]. A Belgian study examining the prevalence of VM in an otolaryngology clinic reported a prevalence of 16% [17]. It included VM and probable VM, as well as atypical VM not included in the Bárány classification. Exclusion of atypical VM would reduce the prevalence to 9.9%. As a study on the prevalence of VM in East Asians, a Korean multi-center study using the Bárány criteria found the prevalence to be 12.8% in migraine patients [3]. However, the prevalence was 5.7% in whole headache patient group, suggesting that there may be significant differences in the prevalence depending on the study population.

Racial or regional differences may be the reason of the low prevalence in our study. In large population studies, 1-year prevalence of definite migraine was 6.0% in Korea [7], 6.0% in Japan [14], and 9.3% in China [18], while it was 11.7% in the United States [9] and 22.2% in Europe [13] to show some differences. VM may show larger racial or regional differences in prevalence than migraine itself. Although the race-based prevalence of VM has been rarely reported, a population-based study conducted in the United States found the prevalence of the VM to be 3.13% of African descendant, 2.64% of European descendant, and 1.07% of Asian descendant, suggesting there may be racial differences of VM [5].

Nevertheless, racial or regional differences alone cannot explain the exceptionally low prevalence in this study. We assume that the main differences arose from how strictly the diagnostic criteria were applied. The prevalence would increase if less stringent Neuhauser criteria were used. In some studies, an alleged history of migraine was accepted as migraine without confirming ICHD criteria, which might affect the prevalence of VM [5, 17].

The inclusion of dizziness other than vestibular symptoms as defined by the Bárány Society may increase the prevalence too.

The duration of vertigo is specified by the Bárány Society from 5 minutes to 72 hours. However, the comment notes describe that the duration of episodes is highly variable, lasting only seconds in 10% of cases, lasting longer than several days in 30%, and in some cases it may take up to 4 weeks to fully recover [1]. For recurrent attacks lasting several seconds, the duration of the episode is defined as the sum of total period. In these cases, the diagnosis of VM cannot be made without a thorough medical history. Given these unusual cases, there may be debates about how strictly the criteria for the duration should be applied for the exact diagnosis.

Presenting symptoms of VM may change over time. Symptoms of initial period might not meet the criteria at the first visit when the questionnaire was filled out. During the study period, patients diagnosed with or suspected of Menière’s disease visited the clinic 6.6 times on average, but patients diagnosed with VM according to the Bárány classification visited 4.1 times and patients with preliminary clinical diagnoses of VM visited 3.7 times. Vertigo in VM is usually milder than in Menière’s disease and the fear about the cause may be greater than the dizziness itself. When the fear is reduced after counseling, patients may visit the clinic less often, which reduces the likelihood of developing symptoms that meet the criteria. If we loosen the criteria, including dizziness other than vestibular symptoms as defined by Bárány Society, less than 5 episodes of dizziness, and dizziness lasting for only seconds or for several days, the prevalence of VM in this study reaches 13.0%.

If the vertigo questionnaire was used at every visit, patients who did not meet the criteria at the first visit were more likely to be diagnosed with VM at next visits.

We excluded patients with more than one vestibular diagnosis, which may lead to an underestimation of the VM prevalence as there are well-known associations of BPPV and migraine, Menières disease and migraine, and PPD and migraine. Similarly, exclusion of patients with a previous history of head trauma may decrease the measured prevalence as migraine and probably vestibular migraine may be activated by head trauma. We reviewed the records of patients diagnosed with BPPV, Meniere’s disease, PPPD, or with a history of previous head trauma among patients with a history of headache. In a follow-up of these 89 patients, three later developed symptoms suggestive of VM.

We also reviewed the records of other dizzy patients with a history of headache who were not diagnosed with any of the above-mentioned vestibular disorders (BPPV, Menière’s disease, PPPD, or head trauma history) or who were not diagnosed with preliminary or definitive VM. Eighty-six patients had follow-up records, of which 13 later developed symptoms suggestive of VM. Extrapolating our finding that only 15.3% of preliminary clinical diagnoses of VM met the Bárány criteria would add two more patients to the diagnosis of VM according to strictly applied criteria, which increases the prevalence to 3.1%.

In the absence of biomarkers, there may be controversy over the sensitivity and specificity of consensus criteria by expert committees. Previous Neuhauser criteria did not limit the frequency and duration of vertigo, and a diagnosis of probable vestibular migraine could be made if moderate to severe vertigo was induced by a migraine-specific vertigo precipitants or was responsive to antimigraine drugs. However, these criteria were excluded due to lack of randomized controlled trials or insufficient study results [8]. The criterion of “photophobia and phonophobia” is often changed to “photophobia or phonophobia,” assuming that each by itself is a predictor of migraine. The prevalence may be slightly higher if a looser criterion of “photophobia or phonophobia” is accepted. For research purposes, the Bárány Society recommends strict application of criteria for vestibular symptoms. A strict application of the Bárány criteria would result in high specificity and low sensitivity, so the prevalence would be lower than previously claimed by many experts.

Prevalence studies of VM should consider the influence of referral bias depending on whether patients were referred to otolaryngology or neurology. Otological diseases such as otitis media, cholesteatoma, and superior semicircular canal dehiscence syndrome may cause dizziness. Therefore, it is conceivable that referral to an otolaryngology clinic may reduce the prevalence of VM by selection of patients favoring otological over neurological diagnosis. However, patients with the above otological diseases often complain of hearing loss, otorrhea, and aural fullness rather than dizziness. In this study, one case of cholesteatoma and one case of chronic suppurative otitis media were confirmed among 955 patients with dizziness as the primary complaint, and there were no patients with other otological diseases, so the referral bias is negligible.

Of the patients with a preliminary diagnosis of VM, 84.7% did not meet the Bárány criteria when strictly applied. In the medical records of these patients, the differential diagnoses list other than VM included recurrent vestibulopathy, Meniere’s disease, and orthostatic dizziness. Although the Bárány Society has not yet formulated diagnostic criteria, many experts diagnose benign recurrent vertigo for recurrent episodic vertigo not diagnosed with other well-defined vestibular disorder. It was the second most common preliminary clinical diagnosis in our study population. Van Leeuwen et al considered benign recurrent vertigo to be closely linked to VM and Menières disease [16]. Baloh classified it as migraine equivalent [2]. Patients with a preliminary clinical diagnosis of VM that do not meet the Bárány criteria may fall into benign recurrent vertigo or a spectrum category between VM and benign recurrent vertigo.

A limitation of this study is that it is a retrospective study. Another disadvantage is that the questionnaire used in this study had limitations in diagnosing migraine with aura. According to ICHD-3, the diagnostic criteria for migraine with aura should identify 6 types and 6 characteristics of aura symptoms [1]. However, the questionnaire in the study checked the presence of visual aura only, and confirmed the duration of the aura and time interval between the aura and onset of headache. The reason for not checking other types of aura or characteristics was to prevent the questionnaire too complicated. Since visual aura is the most common type that occurs in almost all Korean patients with migraine with aura [6], it could be said that the chances of missing out on other types of migraine with aura are very low.

The advantages of this study are the high reliability of the collected data retrieved from EMR big data system and the use of the vertigo questionnaire. Although it is a retrospective study, the prevalence of VM is quite reliable because it included a large number of dizzy patients, and the questionnaire was designed to confirm the criteria of the Bárány Society.

5 Conclusion

Vestibular migraine was the third most common preliminary clinical diagnosis among dizzy patients visiting the otolaryngology department, with a prevalence of 11.6%. However, when the diagnostic criteria of the Bárány Society were strictly applied, only 2.9% could be diagnosed as VM and probable VM. A structured questionnaire was helpful to accurately diagnose VM according to the criteria of the Bárány classification. Given the undeniable effect of migraine on vertigo, it is expected that this gap should be filled in future revision of diagnostic criteria.