Investigating the spatial distribution of campylobacteriosis in New Zealand

Abstract

Background Infection with Campylobacter is thought to account for about 5% - 14% of all food and waterborne diarrhoea cases worldwide. By international standards, New Zealand has extremely high rates of campylobacteriosis which are thought to be the highest reported rates worldwide. The incidence has been steadily increasing since 1980 (when the disease became notifiable), reaching a peak of cases in 2003 (396/100,000). Although different surveillance systems complicate international comparisons, New Zealand's particularly high rate still lacks a conclusive explanation. Aims This study investigates the geographical distribution of campylobacteriosis in New Zealand and the relative importance of factors assumed to be affecting the distribution of this disease, including those related to climate, landuse, water and food. The approach aims to explain why certain areas might increase the probability of becoming infected. Methodology A Geographical Information System (GIS) is used to visualise the disease rate, investigate potential disease clustering and identify outliers. Hierarchical regression, including the analysis of residuals, is applied to analyse the variables in their complex interrelation and to investigate whether there is statistical evidence explaining the geographical variation in campylobacteriosis. This study is undertaken at the territorial local authority level, as all required data are available at this spatial scale and covers the period 1997 to 2005. Results and conclusion There is a large geographical variation in campylobacteriosis across New Zealand, ranging from an average annual rate of 97/100,000 to 526/100,000 per territorial local authority (TLA). Generally, there is statistical evidence for global and local clustering of the disease rate. There are upper and lower outliers of campylobacteriosis in New Zealand; however, higher rates primarily appear in the South Island. The hierarchical modelling confirms statistical significance for some of the environmental and sociodemographic variables. The final model explains about 58% of the variation in campylobacteriosis, and the residuals reflect this variation relatively accurately in approximately 75% of all TLAs. Although the evaluation of the results is confronted with a number of challenges, it is concluded that socioeconomic and demographic factors are crucial factors in explaining the observed spatial patterns in the notification data.