Statistical analysis of fever interval data.

Periodic fever syndromes usually do not show a clear fixed periodicity, and hence cannot be described by the mean and the standard deviation of the interval length. In those cases statistical tools must be used. Fever interval data can be described as a random point process; the renewal density function and its Fourier transform can be used to uncover hidden periodicities. In this report the estimation and evaluation of the renewal density function are described. The data of three patients with periodic fever were investigated. In one of these patients, suffering from periodic fever originating at the level of the hypothalamus, a 30-day periodicity was observed. In a female patient with hyperimmunoglobulinaemia D and periodic fever, menstruation often coincided with or was preceded by a fever episode. In the third patient with a Familial Hibernian Fever-like syndrome, an irregular fever pattern and no periodicity were found. The computerized analysis of the fever intervals may be useful in assessing various periodic fever syndromes.