Testing for a change‐point in registry data with an example on hypospadias

Abstract

The Poisson distribution is often suitable for analysing non‐infectious disorders with low prevalence rates. The numbers of babies born with hypospadias in the Liverpool Congenital Malformations Registry during 1960 to 1982 do not fit a Poisson distribution with constant prevalence rates, but a change‐point Poisson process does fit. A significance test is described and its power is considered in two simulation studies. A small discrepancy is noted that does not diminish appreciably when Bartlett's correction is applied.