This is the first study in NZ to analyze longitudinal data on childhood gastroenteritis caused by non-viral pathogens. The trend towards a decline in both notifications and hospitalizations for most non-viral pathogens included in this study is reassuring. Most of these pathogens are predominantly foodborne, and the observed decline may have been contributed to by changes in food safety, such as food industry standards relating to food handling and general hygiene, as well as increased consumer awareness of food safety . The decline may also have been contributed to by changes in the food production sector, including farming, such as increased mechanization and more stringent monitoring of food safety indicators.
Consistent with several other international studies [25,26,27], Campylobacter was identified as the pathogen associated with the majority of notifications and hospitalizations. Hospitalization rates increased and decreased in parallel with notification rates, thus it is unlikely these results are artefactual. Our results are consistent with trends observed in the adult population in NZ, in which notification rates for Campylobacter peaked in 2006 (> 380 per 100,000) and were lowest in 2008 (161.5 per 100,000) .
Between 2004 and 2008, and most notably between 2006 and 2008, there was a reduction in Campylobacter notifications that was consistent across all age groups. This reduction coincided with the introduction of a range of voluntary and regulatory interventions to reduce Campylobacter contamination in poultry . In late 2006, the NZ Food Safety Authority (formerly part of the Ministry of Agriculture and Fisheries - now the Ministry for Primary Industries) released guidance aimed at reducing the incidence of poultry-associated foodborne Campylobacter disease. This was in response to research that indicated poultry contributed disproportionately to the incidence of Campylobacter disease in NZ [28, 29]. Interventions included the development and implementation of microbiological surveillance activities and increased reporting. The observed reduction in Campylobacter incidence in children seen in this study provides evidence of the positive effects of a food safety measure on child health. Whilst these findings are encouraging, there is some suggestion from our analysis that Campylobacter notification rates may have plateaued since 2008. This may, in part, be attributable to recent outbreaks, which have been water, rather than food-borne, or other changes such as the increasing popularity of the consumption of raw milk. This observation warrants purposeful surveillance.
Nontyphoidal salmonellae are also common bacterial pathogens associated with acute gastroenteritis in children [26, 30] and the leading cause of childhood bacterial enterocolitis requiring hospitalisation . Although the food safety interventions mentioned above were targeted predominantly at Campylobacter, they may also have served to reduce the incidence of disease due to nontyphoidal Salmonella, which is also commonly transmitted via poultry. The observed reduction in the incidence of nontyphoidal Salmonella between 1997 and 2016 (as for Campylobacter) supports this hypothesis.
Unlike for nontyphoidal Salmonella, the rates of notification and hospitalization for Salmonella typhi were very similar. This finding is not surprising given that infection with Salmonella typhi is more likely to present with severe disease requiring hospitalization.
The number of notifications of ETEC infection increased steadily after 1997, reaching a peak of 16 cases per 100,000 in the population across all age groups in 2016. This increase may, in part, be due to changes in laboratory testing practices over that period, with increasingly sensitive assays and algorithms used for the detection of Escherichia coli (E. coli)  It could also have been contributed to by improvements in surveillance due to increased public health concern . However, the increased rates of notification and hospitalization in the under one and 1–4 year age groups compared to those > 5 years are notable. Jaros et al. (2014) reported that the highest rate of E. coli infections in their national prospective case-control study of confirmed cases of Shiga toxin-producing E. coli in NZ occurred in the 1–4 year age group. The risk to children was increased further if there was a family member in contact with animals other than household pets. That study provides some support for the contention that the higher rates of disease in babies and children under 5 years of age found in our study is unlikely to be accounted for by changes in laboratory testing practices in those ≤4 years of age.
Exposures to farming environments have been reported as risk factors of sporadic ETEC infections, particularly in young children [33, 34]. Sheep and, in particular, cattle, are considered an important reservoir for ETEC and are the primary source of foodborne and environmental outbreaks in humans . Our study observed higher rates of ETEC infections in DHB regions with more intensive dairy farming compared to those DHB regions with less intensive dairy farming (Additional file 1: Figure S1).
Māori children had lower notification rates for disease when compared to non-Māori children. This is an unexpected finding. A previous study has reported that risk for morbidity due to infectious disease is higher in Māori compared to non-Māori, and also that infectious disease incidence in Māori has risen considerably over time, resulting in pronounced gradients of disease inequality . However, a lower incidence of Campylobacter in Māori adults has been reported previously [8, 36]. Despite Māori people being more likely to live in farming regions , their incidence of notified Campylobacter disease does not correspond with the higher rates seen overall in these areas. Disparities in Māori access to health care are well documented and may be contributing to the lower rates of notification in this population . In order for notification to occur, children must present to a health care provider and be tested. Hence, the results observed could be due to lower rates of both presentation and subsequent testing.
Strengths and limitations
NZ has a robust disease notification system regulated under the Health Act 1956. Notification can be initiated on the basis of ‘clinical suspicion’ by a medical practitioner, but is usually based on the isolation of a pathogen from a clinical sample. All admissions to public hospitals are recorded and data on hospital admissions collated and held nationally. This study has used data from the two most comprehensive national datasets in our analysis.
It is widely accepted that the incidence of gastroenteritis in NZ is underreported. Lake et al. (2010) reported that for every notified case of acute gastroenteritis in NZ, there were an estimated 222 cases in the community that were not reported . It is estimated that 0.4% of community cases of acute gastroenteritis are notified to the national surveillance system . Recent changes to reporting methods, alongside improvements to ETEC testing and detection, raise the possibility that there could have been more systematic underreporting and subsequent underestimation of disease earlier on in the study period.
Although considerable evidence has been published suggesting that interventions implemented by the food industry may have contributed to the decline in the incidence of Campylobacter gastroenteritis and disease due to nontyphoidal Salmonella in NZ, several alternative explanations should be considered. These include the possibility of surveillance artifact, a decline in the relative proportion of food versus waterborne disease, and changes in consumer behavior, such as increased awareness of food hygiene.
Our observations in Māori children regarding the low rates of notification are concerning and warrant further investigation. Of note, almost one fifth of cases were missing ethnicity data in ESR EpiSurv records prior to 2009, although the proportion of cases with missing ethnicity data decreased markedly after that. As such, differences in rates by ethnicity prior to 2009 should be interpreted with caution. Misclassification of ethnicity is also of concern and may have contributed to underestimation of disease incidence in Māori children.