Epidemiology of vestibular vertigo

Dizziness ranks among the most common complaints in medicine, affecting approximately 20% to 30% of the general population.^1–3 However, the term dizziness encompasses a variety of different sensations that point in distinct diagnostic directions: rotational vertigo or other illusory sensations of motion indicate vestibular symptoms, whereas a sensation of light-headedness, giddiness, unsteadiness, drowsiness, or impending faint implies dizziness of nonvestibular origin. However, even though vertigo and dizziness may reflect a distinction between vestibular and nonvestibular symptoms, patients often use these terms interchangeably, and few physicians other than otoneurologists clearly differentiate between the two. This may be one of the reasons why the epidemiology of vestibular vertigo has been studied only in selected patient groups but not in the general population. The aim of this neurotologic survey was 1) to assess the prevalence and incidence of vestibular vertigo in the general adult population in Germany, 2) to describe clinical characteristics of vestibular vertigo in an unselected sample, and 3) to identify sociodemographic characteristics and comorbid conditions associated with vestibular vertigo.

Methods.

We conducted a cross-sectional neurotologic survey nested within the German National Telephone Health Interview Survey 2003 (GNT-HIS). The GNT-HIS is a nationally representative health survey of the noninstitutionalized adult population in Germany with computer-assisted telephone interviews (n = 8,318). To achieve a representative sample of the population, a two-stage sampling design was used. Households with fixed-line telephones were selected using a modified random digit dialing sampling design.⁴ Interviewers asked to speak with the member of the household aged 18 years or older who had the next birthday (random next-birthday method). The overall response rate was 59.2% (52.3% completed interviews, and 7% completed short questionnaires). The survey comprised a range of questions covering demographic and socioeconomic characteristics, physical health, lifestyle behaviors, health-related quality of life, and health care utilization. The following question screened for the occurrence of dizziness: Did you ever experience moderate or severe dizziness or vertigo?

Out of a simple random sample of 4,869 GNT-HIS participants, 1,157 fulfilled the inclusion criteria for the neurotologic survey (consent for further interviews, a history of moderate or severe dizziness/vertigo, and still valid telephone number), and 1,003 (87%) completed the neurotologic interview (table 1). The interview was developed by the authors through piloting and validation in a specialized dizziness clinic. It contained both open-ended and standardized questions and had a core part designed to differentiate between vestibular vertigo and nonvestibular dizziness, followed by detailed questions on four specific syndromes (benign paroxysmal positional vertigo, migrainous vertigo, Ménière disease, and orthostatic dizziness). The methods and results regarding the specific syndromes are reported separately.

The description of the dizziness/vertigo was recorded first in the participants' own words to avoid suggestive questioning and was then followed by a semistructured interview. Participants were asked about the type of dizziness, duration, provoking factors, impact of dizziness, health care utilization, and previous diagnoses. In addition, all participants were explicitly asked whether they had experienced weakness, numbness, speech abnormalities, or other symptoms with the vertigo, how often these symptoms had occurred, and how long they had persisted. All participants with vestibular vertigo were asked a series of semistructured questions about headaches to diagnose migraine according to the criteria of the International Headache Society (IHS).⁵

Diagnostic criteria with an emphasis on specificity were developed for differentiation of vestibular and nonvestibular vertigo and for the specific syndromes. Vestibular vertigo was diagnosed when one of the following applied: 1) rotational vertigo, 2) positional vertigo, or 3) recurrent dizziness with nausea and either oscillopsia or episodic imbalance. Rotational vertigo was defined as an illusion of self-motion or object motion, and positional vertigo was defined as vertigo or dizziness precipitated by changes of head position, such as lying down or turning in bed. If participants reported both vestibular vertigo and nonvestibular dizziness, only vestibular vertigo was inquired about in more detail.

The interviews were conducted by two medical student interviewers (F.L. and M.F.) who were extensively trained in a neurologic dizziness clinic during a period of 1 year. They discussed each interview with one of four experienced neurotologists (H.K.N., M.v.B., A.R., and T.L.), who also supervised 10% of all interviews in person. Inconclusive cases were discussed in case conferences with all four neurotologists involved in the study. If necessary, additional information was obtained by a complementary interview.

Results.

The lifetime prevalence of moderate or severe dizziness or vertigo in the adult residential population in Germany, based on data of all 8,318 participants of the GNT-HIS, was 29.5% (women 36.2%, men 22.4%). In the neurotologic survey, which comprised only persons with a history of moderate or severe dizziness or vertigo, 243 of 1,003 participants reported vestibular vertigo (24%), 742 (74%) reported nonvestibular dizziness only, and 18 (2%) reported dizziness or vertigo of undetermined origin. This distribution did not significantly differ by interviewer. The proportion of vestibular vertigo among participants with moderate-to-severe dizziness or vertigo increased with age: 14% in the age group 18 to 39 years, 28% in the age group 40 to 59 years, and 37% in the age group 60 years and older. Twenty-nine percent of those who consulted a doctor for their dizziness had vestibular vertigo.

The lifetime prevalence of vestibular vertigo was 7.4%, the 1-year prevalence was 4.9%, and the 1-year incidence was 1.4%. The age- and sex-adjusted population prevalence and incidence rates of vestibular vertigo are shown in table 3. Of all participants with vestibular vertigo, 70% had consulted a doctor for their vertigo, and 80% reported severe vestibular vertigo (defined as vertigo leading to a medical consultation, interruption of daily activities, or sick leave). Vestibular vertigo was recurrent (at least two attacks) in 89%. Two thirds had experienced vestibular vertigo in the past 12 months. Prevalence and incidence rates of vestibular vertigo were consistently higher in women than in men.

The majority (83%) of respondents had rotational vertigo, which was spontaneous in approximately half of the cases and position triggered in the other half. In addition, 17% had vestibular dizziness according to the study criteria (recurrent dizziness with nausea and either oscillopsia or episodic imbalance). The attacks were shorter than 1 minute in approximately half of the participants, lasted between 1 and 60 minutes in a quarter of the participants, and lasted an hour to several days in the remaining quarter of participants. The clinical characteristics of the vertigo are summarized in table 4. There was no significant difference in the distribution of the clinical characteristics between men and women except for a history of headache and IHS migraine, which were more frequent in women (headache 84.7% vs 67.2% and IHS migraine 52.2% vs 26.6%; p < 0.005).

Six participants reported neurologic symptoms with their vertigo. Of these, three had migrainous vertigo with associated visual, sensory, and motor aura symptoms, and one had multiple sclerosis (MS) and reported vertigo lasting several weeks accompanied by motor weakness and speech difficulties. Furthermore, a 28-year-old and a 40-year-old participant had recurrent tingling and numbness with the vertigo lasting between 30 minutes and 1 hour, but an ischemic origin seemed unlikely, considering their overall clinical history.

We examined the association of vestibular vertigo with sociodemographic factors and comorbid conditions, comparing the 163 participants with vertigo in the past 12 months to 2,816 vertigo- and dizziness-free participants from the initial simple random sample of GNT-HIS participants. The associations tested were prespecified in the protocol: age, sex, educational level, depression, tinnitus, hypertension, increased blood lipids, diabetes, coronary heart disease, stroke, body mass index, smoking, menopause in women aged 35 years and older, and intake of hormone replacement therapy or oral contraception in women older than 45 years. In the univariate analysis, vestibular vertigo in the past year was significantly associated with all of the above factors except for hormone replacement therapy or use of oral contraception. However, after adjustment for age and sex, there was no longer any evidence for an association of vestibular vertigo with diabetes, stroke, smoking, or menopause (table 5). In a multivariate stepwise backward logistic regression model including all remaining factors, only sex, self-reported depression in the past year, hypertension, increased blood lipids, and bothersome tinnitus in the past 7 days had an independent effect on vestibular vertigo in the past 12 months.

Discussion.

Our study provides estimates of the prevalence and incidence of vestibular vertigo in the general adult population, which have not been investigated so far. These estimates are based on validated neurotologic interviews and explicit diagnostic criteria. The GNT-HIS has confirmed that moderate or severe dizziness is a very frequent complaint in the general population, with a lifetime prevalence of 30%, and the neurotologic survey has shown that vestibular vertigo accounts for a quarter of these reports of dizziness. Vestibular vertigo is not only very frequent at the population level, with a lifetime prevalence of 7.8%, a 1-year prevalence of 5.2%, and an incidence of 1.5% per year, but also has a considerable individual and public health impact because 80% of affected individuals sought medical consultation, interrupted their daily activities, or were absent from work.

Thus far, the prevalence of vestibular vertigo has been determined only in specialized care settings (for a summary, see Sloane et al.⁷ and Tinetti et al.⁸) and not at the population level. Results of these studies cannot be extrapolated to the general population because of selection bias and other types of biases. A few population-based studies have been conducted on dizziness, without specifying for vestibular vertigo and mostly focusing on the geriatric population. Prevalences of 20% to 40% in the elderly^1,9–11 and 23% in working-age individuals have been reported,^1,2 but these studies differ by the time interval to which prevalence rates refer. In general, because dizziness and vertigo are subjective phenomena that are not easy to describe, considerable misclassification may result from noninteractive questionnaires or interviews conducted by lay interviewers.

The strengths of our study are its nationwide general population setting with a large sample size and wide age range, the use of explicit diagnostic criteria, and the use of validated interactive neurotologic interviews with high specificity and sensitivity. In addition, the high fixed-line telephone coverage (94.5%) and low cell phone–only (4.2%) user prevalence in Germany (according to the National Statistical Office in 2003) allow for telephone sampling of the general population.

However, as in any observational study, the potential for misclassification and bias and its magnitude must be discussed. We attempted to minimize misclassification by asking for moderate or severe dizziness in the screening question (mild dizziness being more difficult to classify) and by putting an emphasis on specificity rather than sensitivity in the diagnostic criteria of vestibular vertigo as confirmed by the validation study. Furthermore, the nonresponder analyses, which we have conducted at all levels where nonresponse occurred, are reassuring with respect to selection bias (see tables 1 and 2). In particular, the willingness to participate in a further interview was not associated with the presence of dizziness and vertigo. In addition, the demographic features and selected health indicators of these GNT-HIS responders who consented to further interviews were generally similar to the distribution known from national statistics (see table 2). As an exception to this, participants with a lower secondary school education were underrepresented in the GNT-HIS, whereas those with a middle and higher educational level were overrepresented. Because the prevalence of vertigo was inversely associated with educational level—similarly to most diseases and health conditions¹²—our findings may even underestimate the true prevalence of vestibular vertigo in the general population.

The relatively small difference between the 1-year and the lifetime prevalence rates of vestibular vertigo may be surprising at first glance. It may be an indication of recall bias, i.e., that more remote episodes of dizziness or vertigo tend to be more easily forgotten. In this case, however, the presented prevalence and incidence rates can be regarded as conservative estimates.

The duration of vestibular vertigo attacks, which is an important diagnostic feature in clinical practice, was shorter than 1 minute in more than half of the respondents with vestibular vertigo. This coincides with findings from neurotologic dizziness clinics, in which the most frequent cause of vestibular vertigo is benign paroxysmal positional vertigo (BPPV), which typically lasts less than 30 seconds. Accordingly, almost 80% of participants with short-lasting vestibular vertigo reported positional vertigo, which also points to a high proportion of BPPV.

Our study also investigated sociodemographic factors and comorbid conditions associated with vestibular vertigo. However, the study design allows only the analysis of associations but not causal inference.

In previous epidemiologic studies, unspecified dizziness—referring to both nonvestibular dizziness and vestibular vertigo of mostly older persons—was associated with increasing age, female sex, depression, anxiety, multiple sensory deficits, coronary heart disease, cerebrovascular disease, medication intake, and lower socioeconomic status.^{1,2,8,9,11,13} In contrast, far less is known about risk factors for vestibular vertigo. We have found an association of vestibular vertigo with female sex, which has been previously reported for specific vestibular disorders including BPPV,¹⁴ Ménière disease,¹⁵ and migrainous vertigo.¹⁶ Case series have suggested that premenstrual or oral contraceptive–related hormonal changes may increase the risk of vestibular disorders,^17,18 but this was not consistently confirmed in a large cohort study.¹⁹ Similarly, we could not find an association between use of oral contraceptives or hormone replacement therapy and occurrence of vertigo in the past 12 months among women older than 45 years, and there was no significant association with menopause after adjustment for age and sex.

One of the factors contributing to the higher prevalence of vestibular vertigo among women may be migraine. Migraine is associated both with female sex and specific vestibular diagnoses such as migrainous vertigo,¹⁶ Ménière disease,¹⁵ and BPPV²⁰ but also unclassified recurrent vestibulopathy.^21,22 In our study, the prevalence of migraine according to the criteria of the IHS was at least three times higher in participants with vestibular vertigo than in previous population-based studies²³ and was approximately two times higher in women than in men. However, because a migraine history was not taken in the GNT-HIS, migraine status was unknown in the control group and could not be included in the logistic regression model.

We found the strongest association of vestibular vertigo with tinnitus, reflecting the frequent otogenic origin of vestibular vertigo. In addition, there was an almost equally strong association of vestibular vertigo with self-reported depression, which is in line with two previous specialized care studies^24,25 and complements the well-documented association of dizziness with both depression and anxiety.^1,2,8

The association of vertigo with cardiovascular risk factors is more complex, with a few studies suggesting a link but with insufficient evidence to support an independent association in unselected individuals after taking into account potential confounders. Impairment of inner ear function in diabetes mellitus has been suggested by studies showing decreased hearing acuity in diabetic patients compared with controls,^26,27 by findings of pathologic changes in the saccule and utricle in experimental diabetes,²⁸ and by an increased prevalence of both peripheral and central electronystagmographic abnormalities in diabetic patients compared with controls.^29,30 However, these studies did not show an association of diabetes with subjective vestibular symptoms. In accord with these findings, we have found an association of vestibular vertigo with diabetes in the crude analysis, which, however, became nonsignificant after adjustment for age and sex and remained nonsignificant in the multivariable analysis.

A link between hyperlipidemia and vestibular disorders has been likewise discussed but relies on older, uncontrolled observations, which are prone to multiple biases.^17,31,32 We were surprised to find a significant association between self-reported increased blood lipids and vestibular vertigo in the multivariable analysis. However, an OR of less than 2 with a 95% CI lower limit, which comes close to unity, can easily be due to unmeasured confounding by other factors.

We would advocate the same cautious interpretation for the finding of a significant comorbidity of hypertension and vestibular vertigo. Vestibular vertigo is not a symptom of hypertension itself. Hypertension may induce small vessel disease of the brainstem and labyrinth leading to vertiginous TIAs and strokes, but this accounts for only a small proportion of patients, even in specialized dizziness clinics.^33,34 The frequency of vertigo in stroke is as low as 2%,³⁵ which means that at the population level, the comorbidity of vertigo and stroke should be very low. In accord with this, in none of the seven participants with a history of stroke had the stroke presented with vertigo. It is also very unlikely that there were many undetected strokes presenting with vertigo among our participants because those with vertigo and accompanying neurologic symptoms were questioned in detail (with negative results), and vertebrobasilar TIA presenting with isolated vertigo is very rare, both in dizziness clinics³⁴ and stroke registry case series.³⁶

Our risk factor analysis is limited by the cross-sectional design and by the assessment of certain associated factors by self-report rather than measurement. However, its strengths include the rigorous assessment of vestibular symptoms, the sampling of controls from the same population as the cases, and the use of identical methods for assessing risk factors in cases and controls. Overall, none of the observed associations gives an easy answer to the causal pathways leading to vestibular vertigo. This is not unexpected, because vestibular vertigo encompasses a variety of heterogeneous disorders, and often, the exact pathophysiology cannot be determined with certainty.

More importantly, however, our study shows that vestibular vertigo is a frequent condition at the population level and involves medical consultations in the majority of cases. The frequency and health care impact of vestibular symptoms at the population level have been underestimated so far. Vestibular vertigo should therefore not be regarded as a specialized domain of neurotologists only, but as a primary care problem.