Reye's syndrome

EDITOR, - Maria Casteels-Van Daele and Ephrem Eggermont propose that Reye's syndrome is associated not with aspirin but with antiemetics given for the profuse vomiting which characterises this disorder and that antiemetic toxicity is often misclassified as Reye's syndrome.1 As evidence they cite a paper describing two patients with classic extrapyramidal symptoms of antiemetic toxicity. Since these have been known for decades, it is unlikely that “better recognition” of these side effects could explain the decline of reported Reye's syndrome in the United States and the United Kingdom.

The authors criticise the North American case-control studies for recording only the drugs given before the onset of vomiting. They question whether this reflects the onset of Reye's syndrome itself. The onset is indeed hard to time precisely, but there is histological and ultrastructural evidence that Reye's syndrome is well established shortly after the onset of vomiting, before consciousness becomes impaired.2

Ironically, the early American case-control studies were accused of temporal precedence bias caused by analysing medications given both before and after the onset of vomiting.3 In later studies, including one funded by the aspirin industry, a detailed definition of day of onset was used; it included both vomiting and neurological symptoms. Temporal precedence bias was avoided by analysing only onset exposures before onset.4 No further criticisms of the timing of exposure to drugs have been published.

Casteels-Van Daele and Eggermont state that only drugs given before vomiting were registered in the second year of the Ohio study. Nevertheless data on phenothiazine ingestion throughout the illness were published. Although there was a significant excess among cases, only 19% had received an antiemetic compared with 100% exposed to aspirin. In the British study the corresponding figures were 14% and 59%. Thus if antiemetics do have a role in the pathogenesis of Reye's syndrome it must be substantially smaller than that of aspirin.

The authors further argue that the subjects in the case-control studies had a heterogeneity of conditions, including inherited metabolic disorders. Children with those inherited metabolic disorders that cause a Reye-like illness usually present in the first two years of life with an encephalopathy precipitated by a variety of viral infections. By contrast, Reye's syndrome in the United States in the 1970s and 80s occurred epidemically in association with influenza and endemically in association with chickenpox; patients had a median age of 8-9 years.

The problem of heterogeneity was addressed in the British study, which used a clinicopathological scoring system. Patients who scored highly satisfied all the criteria of the case definition and resembled clinically those recruited to the American studies. A significant correlation was found between increasing score and aspirin exposure, suggesting that there was indeed a “subset” effect.

Therefore, though Casteels-Van Daele and Eggermont correctly state that Reye's syndrome is a heterogeneous condition, there is evidence for the existence of a subset which is clinically and epidemiologically distinct and nearly always associated with exposure to aspirin. Not only have the numbers of reported cases in the United States and the British Isles declined since the aspirin warnings, but also the epidemiological and clinical pattern has changed. The median age has fallen dramatically; furthermore, the virologically confirmed cases associated with influenza and varicella have selectively declined in the British series since 1986.5

Better diagnosis of inherited metabolic disorders has probably also contributed to the fall in the number of cases of Reye's syndrome reported, although there is still a need for better recognition of those disorders that cause Reye-like illnesses. There is no evidence for an aetiological role for antiemetics.