In general, the spectrum of vestibular symptoms according to the Bárány Vestibular Symptoms grid was broader in RVS-NOS and VM than in MD. First, the average number of different vestibular symptoms per patient was higher in VM and RVS-NOS than in MD. A patient with VM reported 6.03 ± 4.02 (mean ± SD) different vestibular symptoms (median: 5), which was not statistically different from the 5.91 ± 2.81 symptoms (median: 5) experienced by an RVS-NOS patient (two-sided Fisher's exact test: p = 0.99). On the other hand, only 3.61 ± 2.80 symptoms (median: 2) were described on average per patient with MD, which was significantly less compared to both VM (p < 0.0001) and RVS-NOS (p = 0.0002). Second, the type of vestibular symptoms on the group level was very similar for patients with RVS-NOS and VM. None of the symptoms from the Bárány Vestibular Symptoms grid occurred at statistically different frequencies between these two groups (Figures 1, 2 and Table 1C). On the other hand, some symptoms were reported less frequently by patients with MD than those with RVS-NOS (Table 1A) and VM (Table 2A).

Regarding internal vertigo (part one of the Symptoms grid), positional, head-motion triggered and orthostatic vertigo were experienced more often by patients with RVS-NOS than those with MD (Table 1A and Figure 1). Non-spinning head-motion triggered vertigo and transient spinning positional vertigo were only reported by RVS-NOS (14 and 11% each), but not by MD patients. Spontaneous spinning vertigo was the most common vestibular symptom in all three disorders occurring at relative frequencies of 60% in RVS-NOS, 54% in MD and 47% in VM (Supplementary Material 2). There was, however, no significant difference between the groups (Chi-square test of three rows: p = 0.32).

The symptom dizziness (Symptoms grid, part 2) provided no additional information for the discrimination between RVS-NOS and MD (Figure 2A). External vertigo (i.e., the false sensation that the visual surround is spinning or flowing) was the only vestibulo-visual symptom (Symptoms grid, part 3) distinguishing RVS-NOS from MD, and the only vestibular symptom that occurred more often in MD (59%) than in RVS-NOS (26%) and VM (37%) (Figure 2B and Tables 1B, 2B). None of the postural symptoms (Symptoms grid, part 4) occurred with significantly different frequencies between the groups (Figure 2C).

As mentioned above, many symptoms of the Bárány Vestibular Symptoms grid, part 1–3, were reported at different frequencies by patients with VM compared to those with MD (Table 2 and Figures 1, 2). In line with the broader spectrum of symptoms in VM as compared to MD, almost all vestibular symptoms occurred more often in VM than in MD (Table 2A), apart from external vertigo, which was more common in MD (Table 2B). Of note, there were some symptoms of internal vertigo that were only described by patients with VM (non-spinning head-motion vertigo, positional vertigo and visually induced vertigo), but not by those with MD. Postural symptoms (Symptoms grid, part 4) were not significantly different between VM and MD (Figure 2C).

Regarding attack frequency, patients had to choose one answer in the “Vertigo PEVS” questionnaire (Supplementary Material 1 and section Rationale of the Present Study). For all three disorders, the most common attack frequency was “≥ 1/month” (i.e., at least one attack per month, but < 1 per week), reported by 27% (MD), 26% (VM), and 23% (RVS-NOS) of patients each (Figure 4A). While MD patients displayed a single peak for this attack frequency, the distribution was more even in VM, where another 26% chose “≥ 1/week” as most common attack frequency (i.e., at least one attack per week, but < 1 per day). A second peak at “≥ 1/year” (i.e., at least one attack per year, but < 1 attack within 6 months) was observed for 14% of RVS-NOS patients.

For attack duration, multiple answers were possible (see question 5.1 in Supplementary Material 1). On the group level, MD patients displayed a single peak (66% of patients) with an attack duration from one to 4 h (Figure 4B). This duration allowed to distinguish patients with MD from VM (OR = 4.4) and RVS-NOS (OR = 9) on the group level (see Tables 1B, 2B for further details). As observed for symptom quality above, the distribution of attack durations was more even across the spectrum of durations for VM and RVS-NOS than for MD. In VM, attack durations ranging from “1–5 min” to “up to 3 days” were evenly distributed, reported by around 30% of patients each. Two peaks were discernible for RVS-NOS (“ < 1 min”: 31% and “5–24 h”: 23%).

As more than one answer was possible for attack duration in the “Vertigo PEVS” questionnaire, we sought to identify whether the broad spectrum of attack durations observed for VM and RVS-NOS as groups was also present in the individual patients. On average, patients of all three groups reported between one and two different attack durations (VM: 1.66 ± 1.00; MD: 1.50 ± 0.62; RVS-NOS: 1.22 ± 0.48), and only 16% (VM), 6% (MD) and 3% (RVS-NOS) of patients each described more than two distinct attack durations. Thus, the majority of patients in all three groups have one or two dominant attack durations contrasting with the broad spectrum at the group level for VM and RVS-NOS.

Periods with many attacks (“clusters”) were more common in the MD (59%) than in the RVS-NOS group (34%; two-sided Fisher's exact test: p = 0.0125) (Table 1B). Clusters lasting for months (instead of weeks) were more often reported by patients with MD (33%) than those with VM (12%, two-sided Fisher's exact test: p = 0.004), while no significant difference was observed between patients with RVS-NOS (8%) and the other two disorders (Table 2B).

One of the most important findings of the present study is the broad spectrum of vestibular symptoms reported by patients with RVS-NOS, both on the group and on the patient level (Figures 1, 2). Spontaneous spinning vertigo was the most common vestibular symptom in RVS-NOS patients (60%). It was however not found in every patient, thus confirming the observation by Pan et al., where spontaneous vertigo was experienced by most (77,8%), but not all patients with BRV (17). Besides spontaneous vertigo, spontaneous dizziness was commonly reported by RVS-NOS patients in our study (34%) and the study by Pan et al. (15%).

While none of the symptoms from the Bárány Vestibular Symptoms grid occurred at different frequencies in RVS-NOS vs. VM in the present study, patients with RVS-NOS experienced certain vertigo types (head-motion, positional and orthostatic) more often than those with MD (Table 1A and Figure 1). In order to capture this broad spectrum of vestibular symptoms, inclusion criteria for RVS-NOS should comprise all kinds of vestibular symptoms in contrast to previous definitions of BRV and RV that were often restricted to spontaneous (spinning) vertigo (see section Rationale of the Present Study).

Of note, positional vertigo without nystagmus or persistent positional nystagmus not compatible with BPPV have been observed in patients with BRV before (11, 24), and Slater (8) described a period of positional vertigo following the core event of a BRV attack (8). These observations from previous studies might have been contaminated by patients with VM who may present with positional vertigo / nystagmus during an attack as well (26–28). It is possible that some RVS-NOS patients with positional vertigo might actually suffer from BPPV, as the pathognomonic BPPV-type nystagmus may not be visible on every examination (5). On the other hand, diagnosis had to be changed from BRV to BPPV in only 9% of patients after a follow-up of 2 to 8.5 years in two studies (11, 14). In line with these previous observations are the data from our study where the mean disease duration was not different for RVS-NOS patients with attack durations < 1 min (4.75 ± 7.73 years), between 1 and 5 min (5.77 ± 8.80 years) and longer than 5 min (6.32 ± 5.28 years). Not to make a BPPV diagnosis over such a long period would be highly unlikely.

The vestibular syndrome of RVS-NOS seems to be stable over time rather than a transitory condition converting into another vertigo syndrome. Previous studies reported that the diagnosis of patients with BRV/RV had to be changed to migraine in only 2 to 7.5% of cases and to MD in only 1 to 4% over median follow-up times between 31 and 63 months (12, 17, 29). This is in line with the results from our study where the mean duration of the vestibular syndrome before the patients' first visit in a neurotology clinic was not significantly different between patients with RVS-NOS (5.40 ± 6.10 years) as compared to VM (5.26 ± 6.59 years) or MD (7.63 ± 8.09 years) indicating a rather stable vestibular syndrome not changing over the years.

Accompanying auditory or migraine-type symptoms in ≥50% of attacks were observed in some RVS- NOS patients although all patients fulfilling the criteria for VM/pVM and MD were excluded from this group. 2.9% reported tinnitus, while hearing loss or aural fullness were each present in 5.7%. First, these low rates of accompanying auditory symptoms indicate that the “Vertigo PEVS” questionnaire was applied correctly by the investigators. Second, the fact that some of the patients with RVS-NOS did report auditory symptoms without fulfilling the criteria of MD raises the question whether exclusion of auditory symptoms should be part of the diagnostic criteria for RVS-NOS as proposed before (8, 9). Similar rates of accompanying auditory symptoms (3–14%) have been reported for BRV patients without migraine before. Furthermore, the relative frequencies of photophobia or phonophobia (5.7% each) and non-migrainous headache (14%) in our study were similar to values from the literature (13, 16, 17, 23).

RVS-NOS patients experienced nausea during ≥50% of attacks less frequently (34%) than patients with VM (61%) or MD (81%); vomiting was less common (11%) than in patients with MD (46%). Both findings indicate a relatively mild attack intensity in RVS-NOS as compared to the other two groups (for details see section Disease Severity). The low prevalence of vomiting in the present study is in accordance with the first description of BRV (8), where none of the patients reported vomiting associated with an attack. In general, frequencies for nausea and vomiting in the present study were lower for all three disorders compared to previous reports (11, 13). This discrepancy is most likely due to the fact that we counted only patients suffering from these symptoms during most of (i.e., ≥50%) the attacks.

No significant difference was observed between the proportion of RVS-NOS, VM and MD patients who experienced anxiety during the attacks (31, 47, and 34%, respectively). This is in line with the study by van Esch et al. (13), who found no difference in Hospital Anxiety and Depression Scores (HADS) between these patient groups (13).

Several factors in the present study advocate that disease severity was generally milder in patients with RVS-NOS as compared to those with MD or VM. First, the fraction of patients suffering mostly from mild attacks was higher for RVS-NOS (23%) than MD patients (3%). In addition, less patients with mostly mild or moderate attacks experienced also severe attacks in RVS-NOS (35%) than in VM (83%) or MD (78%). Second, clusters of attacks occurred less frequently in the RVS-NOS (34%) as compared to the MD group (59%). Finally, the relatively mild nature of the attacks was reflected by the lower prevalence of nausea and vomiting as compared to the other two disorders (see section Accompanying Symptoms).

At group level, RVS-NOS patients displayed a “two-peak” pattern for attack frequency and duration contrasting the “single-peak” pattern for MD and the “plateau-like” distribution for VM (Figure 4). The even distribution of attack durations on the group level for VM resembles the results from previous studies on the duration of headaches and vertigo attacks in VM (30, 31). For RVS-NOS, the second peak was always localized within the lower range of reported values in the present study, that is, attack duration < 1 min (31%) and more than one attack per year, but < 2 within 6 months (14%).

This observation has several implications. First, it confirms the notion that a subset of patients with RVS-NOS shows a relatively mild disease severity (see above). Second, the high proportion of attacks < 1 min (31%) indicates that there is a considerable number of patients with short-lived recurrent vestibular symptoms that cannot be classified as BPPV or VP, confirming observations by Pan et al. (duration < 5 min in 22.5%) and Lee et al. (duration < 10 min in 6%) (12, 17). Considering the stable course of the vestibular syndrome in these patients (see section Vestibular Symptoms), it is rather unlikely that they will convert into BPPV or VP on the long term. Third-window syndromes are also an unlikely differential diagnosis, as symptoms like sound- and pressure induced vertigo / dizziness were virtually absent in the RVS-NOS group of the present study (Figures 1, 2A).

Of note, a bimodal distribution of attack duration for BRV on the group level has been reported before by Lee et al. (“few minutes”: 38.9%; “few hours”: 51.4%) and Brantberg and Baloh (1–5 min: 20% and 1–4 h: 30%) (16, 24). In contrast to the broad spectrum of attack durations on the group level in the present study (Figure 4B), only 3% of RVS-NOS patients reported more than two different attack durations on the individual patient level. In summary, these distributions of attack duration on the group and individual level indicate that RVS-NOS is—at least currently—a heterogeneous subset of disorders that need to be further characterized in future studies. We propose that there should be no lower limit of attack duration in the definition of RVS-NOS in order to grasp the full spectrum of this multi-facetted disorder.

The different symptoms of the Bárány Vestibular Symptoms grid occurred with similar frequencies in patients with VM and RVS-NOS. In particular, none of these vestibular symptoms allowed to distinguish between the two disorders. While previous studies suggested a link between BRV/RV and migraine (8, 11, 22–25), several findings from the present study indicate that RVS-NOS is not just another migraine variant.

First, we observed a balanced gender distribution in patients with RVS-NOS (51% females) in contrast to the well-known female preponderance in VM of 85% in the present study (32–34), confirming the results of the study by van Esch (13). At this point, it should be noted that the female preponderance described in BRV/RV studies before 2012 (8, 23–25) might be due to inclusion of patients that would probably have been diagnosed with VM today (see also Section Rationale of the Present Study). This notion is supported by Brantberg and Baloh, who found a female preponderance only for those BRV patients with a positive migraine history (84% female), while those without migraine displayed a more balanced gender distribution (58%) (16).

Second, the temporal profile of vertigo attacks was different between RVS-NOS and VM in the present study. In summary, patients with RVS-NOS displayed two peaks in the distribution of attack frequencies and duration on the group level in contrast to the “plateau pattern” observed for patients with VM (Figure 4). Although the pathophysiological correlate for the two peaks in RVS-NOS is not clear to date (see Section Temporal Characteristics of Attacks), this observation might contribute to a better separation between the two disorders in clinical practice.

In general, the clinical presentation of MD patients in the present study was more stereotyped as compared to RVS-NOS: the spectrum of vestibular symptoms was narrower both on the group and on the individual patient level (Figures 1, 2). While a “double-peak” pattern was observed for attack frequency and duration on the RVS-NOS group level, MD patients displayed a single peak for both parameters (Figure 4). The marked stereotypic pattern of MD attacks might be due to a common underlying pathology in these patients, such as endolymphatic hydrops (35), which is present on inner ear hydrops MRI in almost all patients with MD (36, 37).

A further characteristic feature of MD in the present study was the relatively high proportion of patients experiencing attack clusters of several months (33%). This has an important implication for designing clinical studies with MD patients: if patients are recruited during a cluster of attacks and then return back to baseline, one may have the illusion of a treatment effect, in particular if the endpoint is defined several months or years after inclusion into the study.

It has been debated whether recurrent vertigo attacks without accompanying hearing loss are a subset of MD (“vestibular MD”) (38). In around 20% of cases, MD begins with isolated vestibular symptoms. While 80% of patients develop the full audiovestibular spectrum of symptoms within 5 years, patients with merely vestibular symptoms and without hearing loss over periods of 20 years and more have been described (36, 38). A closer look at the study by Paparella and Mancini reveals, however, that all but one of 51 patients diagnosed with vestibular MD reported aural fullness and 84% suffered from tinnitus. These accompanying symptoms were only encountered in 5.7 and 2.9% of RVS-NOS patients each in the present study. Therefore, it seems unlikely that RVS-NOS is just another subgroup of MD. Nevertheless, a possible conversion into MD cannot be excluded, in particular in those with a disease duration < 5 years (63% in the present study).

So, what are the mechanisms behind RVS-NOS? As this is a symptom-oriented study, it can only provide a tentative answer to this question.

We were not able to identify any pathognomonic vestibular symptoms or other operational criteria that clearly separated RVS-NOS from already known vestibular disorders. Therefore, it seems unlikely that RVS-NOS represents a single, so far unrecognized disease entity. The number and the quality of vestibular symptoms in patients with RVS-NOS (both on the group and on the patient level) suggest that part of these patients suffer from a mild form of VM, while the subgroup with auditory symptoms might represent a very mild form of MD. But it probably also includes some as yet unidentified entities, like recurrent spontaneous vertigo with interictal headshaking nystagmus which was only described after our data collection (39).

In summary, these findings indicate that RVS-NOS is a heterogenous group of different disorders with relatively mild clinical presentations—too mild to fulfill the current diagnostic criteria of, for example, VM or MD- and as yet unidentified diseases, one of which was only described after our data collection.