Editor,—We read with interest the paper by Dr Hardie and colleagues.¹ We all agree now on the non-specificity of the case definition and on the heterogeneous nature of Reye’s syndrome.^2–4 Updated Reye’s syndrome is not a specific clinicopathological entity, but a descriptive term used to designate the condition of a child presenting with an unexplained non-inflammatory encephalopathy and signs of liver dysfunction (international workshop ‘Reye’s syndrome revisited’, Leuven, 3 May 1996).

However, as there is a wide spectrum of differential diagnoses in patients meeting the diagnostic criteria,⁵ classifying them now in two groups, the ‘Reye group’ and the ‘Reye-like inherited metabolic disease group’ is inaccurate: the Reye group again is heterogeneous, composed of patients with infections (and fever), and of patients with toxic and other diseases. This classification again enhances the risk of epidemiological biases^1–3.

Whether the scoring system devised by the authors is a valid predictor of ‘Reye’ (high score) versus ‘Reye-like’ (low score) has to be challenged as well. This scoring system does include exclusion criteria for Reye’s syndrome (for example a patient with positive cerebrospinal fluid can still be diagnosed as having Reye’s syndrome), and in several sections a not recorded feature does increase the score in such a way that an insufficiently documented patient already gets a score of 8 simply by not measured or not recorded features.

Moreover, when applying this ‘Reye score’ to our two patients reported earlier on, they would both be high scorers, namely 17 and 19 respectively.² Yet—by including careful analysis of the preadmission medication—we proved that their ‘Reye’s syndrome’ was a combination of infection (influenza A and cytomegalovirus respectively) together with extrapyramidal reactions after antiemetics.

Either one can conclude that the scoring system is not a valid one to differentiate ‘non-classical’ (low scorers) from ‘classical Reye’s syndrome’ (high scorers). Or one can also assume that high scorers defined by the authors as those with ‘onset in mid to late childhood; peak in winter months; influenza-like or varicella prodrome; profuse vomiting before change in conscious level’ are the ones who most likely have been given antiemetics and are indeed the patients with ‘classical Reye’s syndrome’. This would be a major argument for our hypothesis supported by the data from the Food and Drug Administration—as to the role of the side effects of antiemetics in the report of this syndrome in 1963 (temporal relation with the marketing of the phenothiazines-antiemetics) and for its boom in the seventies.^2–4 Later on, once the extrapyramidal reactions and the malignant neuroleptic syndrome got well known, they contributed to the decline of Reye’s syndrome as they were no longer misdiagnosed as ‘Reye’s syndrome’.³

The agreement on the heterogeneous nature of Reye’s syndrome is the crux of the whole discussion. Logically, it implies that the hypothesis of the epidemiological surveys suggesting a link between Reye’s syndrome and aspirin is not valid, as the case subjects of these surveys were not specific.⁴

As to the decline, may we draw attention to the erroneous referral to our publications: we never attributed this decline of Reye’s syndrome solely to a correct identification of ‘Reye-like’ inherited metabolic disorders. We mentioned clearly that the apparent decline of ‘Reye’s syndrome’ can be explained by ‘an increased recognition of metabolic, viral or toxic diseases’.³ The increase of a more precise diagnosis is also proved by the authors’ own data, which show a steady rise of revised diagnoses from 18.9% in 1982 to 43.8% in 1990.¹ The authors of this letter have no ongoing affiliation or financial involvement with any pharmaceutical company nor with any other entity with a financial interest in the subject matter.