To explore the of age of onset distribution for Perthes' disease of the hip, with particular reference to gender, laterality and conformity to the lognormal distribution.Patients and Methods
A total of 1082 patients were identified from the Liverpool Perthes' Disease Register between 1976 and 2010, of which 992 had the date of diagnosis recorded. In total, 682 patients came from the geographical area exclusively served by Alder Hey Hospital, of which 673 had a date of diagnosis. Age of onset curves were analysed, with respect to the predefined subgroups.Results
The age of onset demonstrated a positive skew with a median of 5.8 years (interquartile range 4.6 to 7.5). Disease onset was a mean five months earlier in girls (p = 0.01) and one year earlier in those who went on to develop bilateral disease (p < 0.001). There was no difference in the age of onset between geographical districts with differing incidence rates. The entire dataset (n = 992) conformed to a lognormal distribution graphically and with the chi-squared test of normality (p = 0.10), but not using the Shapiro-Wilk test (p = 0.01). The distribution for the predefined geographical subgroup (n = 673) conformed well to a lognormal distribution (chi-squared p = 0.16, Shapiro-Wilk p = 0.08). Given the observed lognormal distribution it was assumed that Perthes' disease followed on incubation period consistent with a point-source disease exposure. The incubation period was further examined using Hirayama's method, which suggested that the disease exposure may act in the prenatal period.Conclusion
The age of onset in Perthes' disease conforms to a lognormal distribution, which allows comparisons with infectious disease epidemiology. Earlier onset in girls and those who develop bilateral disease may offer clues to understanding the aetiological determinants of the disease. The analysis suggests that an antenatal aetiological determinant may be responsible for disease.Conclusion
Take home message: Perthes' disease age of onset conforms to a lognormal model, which is most typical of infectious diseases. The shape of the distribution suggests that an aetiological trigger in the pre-natal period may be an important determinant of disease.