Do variations in pertussis notifications reflect incidence or surveillance practices? A comparison of infant notification rates and hospitalisation data in NSW.

The incidence of pertussis cannot be directly measured; estimates are generally based on data sources such as notifications, hospitalisations and deaths. However, these data represent only a proportion of the total cases occurring in the community. The accuracy of notification data may vary between states and territories or over time due to different surveillance practices and as new diagnostic tests are introduced. These surveillance issues, and the typical three-to-five-year cycles of pertussis epidemics, make comparisons over time and between states and territories difficult. However, the comparisons should be more valid for hospitalisation data than for notification data, because methods of collecting hospitalisation data are likely to be more uniform. Some variation in incidence by geographical area is to be expected, depending on the timing of the last epidemic and also on geographical differences in past or present pertussis vaccination coverage. If geographical variations in pertussis notifications reflect real variations in pertussis incidence rather than reporting differences, areas with high infant (aged less than 12 months of age) notification rates would be expected to have high infant hospitalisation rates. This article describes and compares the notification and hospitalisation rates in infants for pertussis, by health area, over a period incorporating two pertussis epidemics in NSW.