Dizziness is a nonspecific term that refers to various abnormal sensations of body orientation in space; patients often find these sensations difficult to describe. The prevalence of dizziness in the community ranges from 1.8% among young adults to over 30% among the elderly. Yearly consultation rates because of dizziness in primary care ranges from 2.5% among patients aged 25 to 44 years, to 8.3% among patients aged 65 years and older, and to 18.2% among patients aged 85 years and older. 1–5
Dizziness can be classified into four subtypes: vertigo, disequilibrium, presyncope and atypical dizziness. This classification is based on the study by Drachman and Hart. 6 There are few studies of the distribution of the specific diseases that can cause dizziness. The following data should, therefore, be interpreted with caution (Appendix 1, available at www.cmaj.ca/cgi/content/full/cmaj.080910/DC1). 5,7–11 In primary care patients, vertigo (prevalence 38%) is mainly caused by otolaryngical conditions; disequilibrium (prevalence 10%) is mainly caused by orthopedic, neurologic or sensory problems; presyncope (prevalence 10%) is mainly caused by cardiac or vasomotor conditions; and atypical dizziness (prevalence 17%) is mainly caused by psychiatric problems. In about 25% of patients, the type cannot be classified or the problem has multiple causes.
The prevalence of these diagnostic categories differs according to age. In younger patients, atypical dizziness and presyncope are most common. In middle-aged patients, vertigo is the most prominent. Presyncope and disequilibrium are more prevalent in elderly patients. 5,12,13 However, elderly patients’ symptoms cannot always be placed in one category because their symptoms often have multiple causes. 5,12,13
Practice guidelines 14–16 advocate the use of several diagnostic tests in the evaluation of dizziness, including history-taking, pulse measurement, carotid sinus massage, nystagmus tests and the Dix-Hallpike manoeuvre. However, these recommendations are based more on opinion than on evidence.
Many tests can only be performed in secondary and tertiary care settings, although most patients are first seen in primary care. The main problem for primary care physicians is to decide which patients need additional testing, which should be referred to secondary care, which require immediate therapy, and which should receive an explanation, reassurance, advice and a “wait and see” approach.
Life-threatening conditions requiring immediate treatment are rare in patients with dizziness (Appendix 2, available at www.cmaj.ca/cgi/content/full/cmaj.080910/DC1). 3,6,8,17 In these mainly acute conditions (mastoiditis, pacemaker failure, myocardial or brainstem/cerebellar infarction, hemorrhage and serious electrolyte disturbances), dizziness is almost never the only presenting symptom, and obtaining a careful patient history and performing a targeted physical examination is usually sufficient for triage. Further diagnostic testing should be done in a secondary care setting. Primary care physicians need to know the characteristics of the diagnostic tests that can be used as point-of-care tests for the diagnosis of the more common conditions.
We performed a systematic review of diagnostic tests that can be used to diagnose dizziness in patients in primary care. Because prevalence is important in the discriminative power of such tests, we provide epidemiologic information about target conditions in patients who present with dizziness. We have limited our review to the evaluation of these tests in patients with dizziness.
Methods
Literature search
We searched MEDLINE, EMBASE, PsycINFO, CINAHL, and GeroLit from database inception to May 2009 to identify studies of diagnostic tests for dizziness in primary care. The complete search strategy is shown in Appendix 3 (available at www.cmaj.ca/cgi/content/full/cmaj.080910/DC1). We used the search filters reported by Bachmann and colleagues. 18,19
Study selection
Two reviewers (J.D., O.M.) independently selected potentially relevant studies based on titles and abstracts. If disagreement occurred, consensus was reached with a third reviewer (H.vW. or H.vdH.). We excluded studies if the title did not include “dizziness,” “disequilibrium,” “(pre)syncope” or ”vertigo” or a word with the same meaning, or if it did not involve a disease that can cause dizziness (Appendix 3). We excluded articles if the abstract did not mention any diagnostic procedure, if the study population did not include patients with dizziness or if the test studied was too complex, time-consuming or expensive to be feasible in primary care (e.g., carotid sinus massage, tilt table testing, posturography and specialized imaging techniques). We included only studies published in English, French, German or Dutch.
Data extraction and quality assessment
The methodological quality of the studies was assessed by two pairs of reviewers (O.M. and H.vdH. or J.D. and H.vW.) using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS) checklist. 20 Within each pair, both reviewers independently scored all studies with the QUADAS checklist, extracted the data and constructed 2 × 2 tables. Consensus was reached for any discrepancies. If an article reported the accuracy of more than one index test, target condition or cut-off point, we report the data for these tests, target conditions and cut-off points separately.
Data synthesis and analysis
We calculated sensitivity and specificity from the 2 × 2 tables and plotted the results in forest plots and receiver operating characteristics plots. We pooled the results from studies involving the same tests and conditions by use of a bivariate regression model that takes into account the correlation between sensitivity and specificity. This method was recently recommended for meta-analysis of diagnostic tests. 21,22 If we could only extract the sensitivity of the test from the original data, we calculated the mean sensitivity and weighted it for the number of patients in the study. To assess clinical usefulness, we derived likelihood ratios and pre- and post-test probabilities of the target condition in the study population.
Because we were interested in the accuracy of the diagnostic tests in primary care but most of the studies were done in secondary or tertiary care settings, we calculated pre- and post-test probabilities for a hypothetical primary care situation, assuming that the sensitivity and specificity would not be different in primary care. We estimated and retrieved the pretest probabilities of the target conditions in primary care from pertinent prevalence studies in primary care (Appendix 1).
We used STATA/SE 10.1 (StataCorp, College Station, TX, USA) to perform the bivariate analyses.
Results
The study selection process is shown in Figure 1. We identified 290 primary articles that studied diagnostic tests that can be used in the primary care of patients with dizziness. The kappa statistic was 0.81 for agreement on study selection by the two reviewers (J.D., O.M.). Based on a review of the full text, we excluded 162 articles. We assessed 128 articles using QUADAS. 20 After this review, we excluded 102 articles (Figure 1). Thus, we included 26 articles in this review (Table 1). 23–48 We included one study of a psychiatric test (subtype atypical), and 25 studies of neuro-otologic tests (subtype vertigo). No studies about disequilibrium and presyncope were available. Five studies involved two or more tests, five studies reported on target conditions, and two studies evaluated two cut-off points of the same test (Table 2).
Table 1: Characteristics of 26 studies of diagnostic tests of dizziness feasible in primary care (part 1)
Table 1: Characteristics of 26 studies of diagnostic tests of dizziness feasible in primary care (part 2)
Table 1: Characteristics of 26 studies of diagnostic tests of dizziness feasible in primary care (part 3)
Table 2: Data from 26 studies that tested the accuracy of diagnostic tests for diagnosing dizziness in primary care (part 1)
Table 2: Data from 26 studies that tested the accuracy of diagnostic tests for diagnosing dizziness in primary care (part 2)
Table 2: Data from 26 studies that tested the accuracy of diagnostic tests for diagnosing dizziness in primary care (part 3)
Figure 1: Flow chart of the study selection process.
Methodologic quality
Quality assessment with the QUADAS tool is presented in Appendix 4 (available at www.cmaj.ca/cgi/content/full/cmaj.080910/DC1). Each study had at least two methodologic limitations that could introduce bias. In almost all studies, the spectrum of included patients was not representative of primary care patients.
Sensitivity, specificity and likelihood ratios
Table 2 shows the sensitivity, specificity and positive and negative likelihood ratios of the included studies. The pre-and post-test probabilities of the target condition in the study population and in a theoretical primary care population are shown in Appendix 5 (available at www.cmaj.ca/cgi/content/full/cmaj.080910/DC1).
Tests
The performance of all studied tests and the analyses of tests described in a single study are presented in Appendix 6 (available at www.cmaj.ca/cgi/content/full/cmaj.080910/DC1). Accurate evaluation of diagnostic tests should be based on the results of more than one study. Therefore, we describe four tests, all targeted for neuro-otologic conditions, that were evaluated in more than one study.
Dix-Hallpike manoeuvre
Both studies of the use of the Dix-Hallpike manoeuvre to diagnose benign paroxysmal positional vertigo used multiple vestibular tests as reference standards. Data from 114 patients were available, and we calculated a mean sensitivity of 80% (95% confidence interval [CI] 71%–87%).
Head-shaking nystagmus test
All nine studies of the use of the head-shaking nystagmus test used caloric measurement as part of the reference standard. Data from 4059 patients (eight studies) were available for pooled analysis (Figure 2 and Appendix 7, available at www.cmaj.ca/cgi/content/full/cmaj.080910/DC1). The pooled sensitivity was 45% (95% CI 30%–62%), the pooled specificity was 82% (95% CI 68%–90%), the pooled positive likelihood ratio (LR) was 2.47 (95% CI 1.48–4.14) and the pooled negative LR was 0.67 (95% CI 0.51–0.87). In these studies, the pooled probability of peripheral vestibular dysfunction after a positive head-shaking nystagmus test result increased from 27% to 48%, and the probability after a negative result decreased from 27% to 25%. Using an estimated prevalence of 33% for peripheral vestibular dysfunction in primary care patients with dizziness, 7,9–11 the post-test probability of a positive head-shaking nystagmus test result was 55% and the post-test probability of a negative result was 25%.
Figure 2: Sensitivity and specificity of the head-shaking nystagmus test for diagnosing unilateral vestibular dysfunction.
Head impulse test
The seven studies that evaluated the head impulse test for peripheral vestibular dysfunction used caloric measurement as a reference standard. Data were available from 869 patients (six studies) for pooled analysis (Figure 3 and Appendix 8, available at www.cmaj.ca/cgi/content/full/cmaj.080910/DC1). The pooled sensitivity was 63% (95% CI 40%–81%), and the pooled specificity was 93% (95% CI 83%–98%). The pooled positive LR was 9.29 (95% CI 4.28–20.16), and the pooled negative LR was 0.40 (95% CI 0.23–0.68). In these studies, the probability of peripheral vestibular dysfunction after a positive test result increased from 33% to 82%, and the probability after a negative result decreased from 33% to 17%. Because the estimated prevalence in primary care 7,9–11 matched the prevalence in the studied population (both 33%), the post-test probabilities were the same.
Figure 3: Sensitivity and specificity of the head impulse test for diagnosing unilateral vestibular dysfunction.
Two studies 28,40 evaluated the accuracy of the head impulse test for central vestibular dysfunction in patients with vertigo. These studies used cranial magnetic resonance imaging as a reference standard. Data from 125 patients were available for pooled analysis. A negative result of this test suggests a central lesion. The pooled sensitivity was 74% (95% CI 63%–83%), and the pooled specificity was 94% (95% CI 83%–99%). The probability of central vestibular dysfunction after a positive head impulse test result decreased from 62% to 31%, and the probability after a negative result increased from 62% to 95%.
Vibration-induced nystagmus test
In the two studies that evaluated vibration-induced nystagmus, data were available from 463 patients, giving a calculated sensitivity of 84% (95% CI 81%–88%). We calculated specificity based only on one study, with a positive result of the vibration-induced nystagmus test increasing the probability of peripheral vestibular dysfunction from 13% to 59%. Using the estimated prevalence of 33% for peripheral vestibular dysfunction in primary care patients with dizziness, 7,9–11 we estimated the post-test probability of a positive vibration-induced nystagmus test result to be 83%.
Discussion
The results of our review show that the empirical validation of commonly used diagnostic tests for dizziness in primary care is poor. We found few studies that focused on dizziness. It was only possible to perform a meta-analysis of two tests in the neuro-otological field. Although many studies were performed for tests used in diagnosing dizziness, most were not diagnostic studies, were methodologically inadequate or did not include patients with dizziness.
All of the included studies used some type of preselection of patients and were intended to diagnose specific conditions, such as Ménière disease, peripheral vestibular dysfunction, major depressive disorder or panic disorder. The main interest in primary care, however, is the differentiation between self-limiting conditions, those for which adequate treatment is available and those that are dangerous or progressive or both and require referral or immediate treatment.
Although many tests are used to diagnose dizziness in primary care, we could not find any studies that included consecutive patients with dizziness. Studies about the use of patient history, pulse measurement, heart auscultation and balance in the diagnosis of dizziness are lacking.
Although we looked for studies of tests that are feasible in primary care, all of the included studies were, at least, partially conducted in secondary or tertiary settings. As a result, the high prevalence of the target conditions in the included populations (6% for Ménière disease and 80% for vertigo) inflates the positive predictive values and decreases the negative predictive values. Furthermore, symptoms are probably more severe or pronounced in patients referred for secondary or tertiary care, which may inflate both the sensitivity and specificity. Although we used prevalence in a hypothetical primary care population to estimate likelihood ratios, we could not correct for this phenomenon. This limits the generalizability of our results to patients in primary care. An exception may be the head-shaking nystagmus test and the head impulse test. The prevalence of peripheral vestibular dysfunction in primary care patients with dizziness was similar to that in the studied populations.
Although the prevalence of dizziness increases with age and the need for valid diagnostic tests or protocols is evident, especially for use in elderly patients, none of the studies in our review included an elderly population. As age increases, both the prevalence of dizziness and the risk of more serious causes of dizziness (stroke or cardiovascular diseases) increase. Comorbid conditions causing dizziness (e.g., diabetes, Parkinson disease) also become more prevalent.
Most tests for the diagnosis of dizziness were analyzed in a single study. The exceptions were the head-shaking nystagmus test, the head impulse test, the vibration-induced nystagmus test and the Dix-Hallpike manoeuvre. Promising tests within single test studies were the Vestibular Vertigo Questionnaire for the diagnosis of BPPV (positive LR 10.56, 95% CI 3.54–31.5), the depression part of the Patient Health Questionnaire for the diagnosis of major depressive disorder (positive LR 12.51, 95% CI 4.96–31.53) and the panic disorder part of the Patient Health Questionnaire for the diagnosis of panic disorder (positive LR 25.1, 95% CI 8.22–76.65).
Both the head-shaking nystagmus test and the head impulse test seem to be better suited for ruling in than for ruling out peripheral vestibular dysfunction. However, for conditions that are most often self-limiting (e.g., peripheral vestibular dysfunction), an increased probability from 27% to 48% after a positive test result is of limited value. The head impulse test performs better for the diagnosis of peripheral vestibular dysfunction, with an increased probability from 33% to 82% after a positive result and, based on two studies, an increased probability of central vestibular dysfunction from 62% to 95% after a negative result.
Many tests have been evaluated in patients without dizziness (e.g., peripheral neuropathy tests, visual acuity) or are included in the definition of the condition (e.g., hypoglycemia or orthostatic hypotension tests). Before the results of these studies can be used in the assessment of a certain diagnosis in a patient with dizziness, these tests must be validated in the appropriate domain. This is especially true for dizziness, because the cause is often not sufficiently supported by scientific evidence. In elderly patients, who often have more than one condition that can cause dizziness, physicians must be cautious in not treating these patients on the basis of test results that are not consistent with the patient’s symptoms.
The appropriate patients to be included in studies of diagnostic tests for dizziness are those who present with dizziness as an isolated symptom and who are not at risk for the acutely life-threatening illnesses mentioned earlier. Prior selection of patients may affect test characteristics. The signs and symptoms must be described precisely, and the study design must comply with accepted methodologic criteria. 49,50 Additional research into the tests that can be used in primary care to diagnose dizziness or to determine which patients need further testing or treatment is highly warranted.
-
Studies on diagnosing dizziness have been conducted in highly selected homogeneous groups of patients only.
-
Evidence to support the diagnostic process in primary care is scarce.
-
An exception is the head impulse test, with a positive test result being diagnostic of peripheral vestibular dysfunction and a negative test result diagnostic of central peripheral dysfunction.
Key points
Additional information about the four tests studied in this article
-
Dix-Hallpike manoeuvre 29
www.dizziness-and-balance.com/disorders/bppv/dix%20hallpike.htm
-
Head-shaking nystagmus 34
www.dizziness-and-balance.com/research/hsn/HeadShakingNystagmus.htm
-
Head impulse test 38
http://www.dizziness-and-balance.com/practice/head-impulse.html
-
Vibration-induced nystagmus test 38
www.dizziness-and-balance.com/practice/nystagmus/vibration_test.htm
Footnotes
-
Previously published at www.cmaj.ca
This article has been peer reviewed.
Competing interests: None declared.
Contributors: All of the authors contributed to the conception and design of the study, the acquisition and interpretation of the data, and the drafting and revising of the manuscript. All of the authors approved the final version of the manuscript for publication.
Funding: The Netherlands Organisation for Health Research and Development provided funding for this review (grant no. 4200.0018). They had no role in the design of the study; data collection, analysis, or interpretation of the data; nor any role in the approval of publication of the finished manuscript.
REFERENCES
In this issue
Article tools
Jump to section
Related Articles
Cited By...
- No citing articles found.