Sir—Seiji Nishino and colleagues (Jan 1, p 39) 1 showed that patients with narcolepsy have reduced concentrations of hypocretin (orexin) in their cerebrospinal fluid. Because hypocretins stimulate food intake in animal models, Siegel hypothesised2 that patients with narcolepsy should be lean, rather than obese. However, in an earlier report Honda and colleagues3 suggest that in patients with narcolepsy the frequency of obesity and of non-insulin dependent diabetes is increased compared with a non-healthy control group of psychiatric patients. To gather more reliable information, we retrospectively identified all patients of 14–70 years who had narcolepsy, and who were treated in the Max-Planck-Institute of Psychiatry, Munich, Germany, between 1988 and 1999 (n= 45). We selected all patients with: excessive daytime sleepiness; cataplectic attacks; multiple sleep-onset REM (rapid eye movement) periods (SOREMs); and who were HLA-DR2 positive. In these 35 patients (24 men, 11 women) we recorded the body-mass index (BMI). At the time of height and weight measurements 18 patients (mean age 36· 3 [SD 17· 3] years) had never been treated pharmacologically for the disease. 17 patients (mean age 48· 5 [12· 6] years) had previously received tricyclic antidepressants, psychostimulants, or combinations of these drugs before.

The BMIs were plotted into BMI-percentiles representative of the German male and female population, respectively (figure). 4 The distribution of the BMIs differed significantly from the rectangular distribution on the 0–100 interval (Kolmogorov-Smirnov test). This was apparent both in male (mean BMI percentile 75· 6 [23· 0]) and female (61· 2 [32· 5]) patients. The mean BMI-percentile of those patients who had never been treated for narcolepsy (74· 8 [25· 0]) was in the same range.