Dizziness accounts for more than seven million clinic visits a year. Identifying the cause can be difficult, and treatment is frequently unsatisfactory. Dizziness typically is a benign and self-limiting condition; however, it can be a sign of a serious underlying cardiac or neurologic problem. Kroenke and colleagues searched the literature to identify the incidence and common causes of dizziness. Such information is of use in guiding the initial assessment and in ordering diagnostic tests.

A MEDLINE search of articles from 1966 to 1996 was conducted to identify studies that addressed vertigo or dizziness. The principal inclusion criterion was a presenting complaint of dizziness, regardless of symptom character. Twelve studies involving more than 4,500 patients reported on the etiology of dizziness in consecutive patients presenting to emergency departments, primary care clinics or subspecialists. The methodology of two studies was rated as very high. Seven studies had intermediate methodology scores, and three studies had low scores.

Peripheral vestibular conditions were the most common cause of dizziness, accounting for 44 percent of cases. Of these conditions, benign positional vertigo (16 percent), labyrinthitis (9 percent) and Meniere's disease (5 percent) were most common. Nonspecific vestibulopathy and drug-induced ototoxicity occurred occasionally. Dizziness also was attributed to psychiatric conditions, including hyperventilation (16 percent); cerebrovascular disease, principally stroke and transient ischemic attack (6 percent); and cardiac irregularities (1.5 percent). Brain tumor was an uncommon finding, accounting for less than 1 percent of cases. No definitive etiology could be established in 13 percent of patients.

The authors conclude that a thorough history and physical examination remain the prime diagnostic strategies in identifying causes of dizziness. Serious treatable causes of dizziness are uncommon, and most cases can be attributed to vestibular causes. Vestibular function tests and screening for depression and anxiety disorders may be necessary, but the yield of other diagnostic testing is likely to be low in unselected patients.