Accessibility Skip to Global Navigation Skip to Local Navigation Skip to Content Skip to Search Skip to Site Map Menu

Drinking-water quality and gastrointestinal disease in New Zealand

Andrew Ball, PhD 2010


This study sought to (1) determine the relationship between drinking-water quality and gastrointestinal disease (GID) in primary school-aged children and (2) to estimate the burden of waterborne GID in New Zealand.

A prospective cohort study was conducted using children from 28 primary schools across the country. The exposed group comprised 465/2,492 children from 14 schools supplied with drinking-water of generally poor microbiological quality whereas the unexposed group comprised 850/6,562 children from 14 schools that consistently complied with the New Zealand Drinking Water Standards (DWS). Exposure and illness data were collected using a daily diary that was collected from each participant after each of the six to seven weeks of the study period. Two drinking-water samples were collected each week from each of the participating schools and analysed for E. coli and Campylobacter spp. The health measure used was highly credible gastrointestinal disease (HCGI).

A total of 47,330 diary days were obtained from the 1,315 participants, a return of 84.4%. The incidence of HCGI in the study population was 1.63 cases of HCGI per person-year (Ie = 1.84 and I0 = 1.50). Logistic regression analysis adjusted for clustering by schools failed to demonstrate a significant association between HCGI and the water quality group (OR = 1.34; CI95% 0.88 - 2.02). However, an association was found between HCGI and the consumption of any water of poor microbiological quality during the study period (OR = 1.62; CI95% 1.15 - 2.26). Longitudinal analysis was also conducted using logistic analysis after segregating the diary days into the seven days immediately prior to HCGI onset (exposure period) and seven-day blocks of symptom-free comparison periods. The association between HCGI and the proportion of days on which any poor quality water was consumed in the week preceding the onset of symptoms was borderline (OR = 1.39, CI95% 1.00 - 1.93). The association between the mean daily intake (in 250 mL glasses) of poor quality water during the week prior to onset of symptoms and HCGI was slightly more significant (OR = 1.11; CI95% 1.02 - 1.20).

This study uses two different approaches to estimate the annual burden of waterborne GID. The bottom-up approach provided an estimate of 239,000 GID cases per year using data from the longitudinal cohort study about primary school-aged children and extrapolating to the entire population. The top-down approach was derived from three estimates of the proportion of notified cases being waterborne from a detailed examination of the 2001 and 2002 notified diseases case records, and the New Zealand AGI Study estimate of 0.4% notification of cases of acute gastrointestinal disease made by Lake et al. (2009). A conservative estimate of 175,000 waterborne GID cases per year was made using 4.1% drinking-water attribution. The upper (33%) and mid-range (16%) estimates of the proportion of GID cases attributed to drinking-water translate to 1,400,000 and 682,000 waterborne GID cases per year respectively. However, untested assumptions were used in all of these estimates and further investigations are required to improve the estimate of national burden of waterborne GID. To this end, a method to attribute the source of infection to water, food, infected people or animals etc has been developed using a modified GID case questionnaire, with the risk factors ranked by an expert panel using a Delphi process. This method is presently being trialled on campylobacteriosis cases in Canterbury.

Supervisors: Cheryl Brunton, Patrick Graham