Main findings and interpretation
Media items in NZ’s major newspapers were examined to assess the nature of the media discourse and determine how the release of journal articles, reports and scholarly blogs have influenced subsequent media discourse on nitrates in drinking water..
The temporal pattern of media items suggests that events of national importance (ie, the 2017 NZ general election) and the release of certain studies and scholarly blogs were associated with the publication of subsequent media items. There was a large increase in the number of media items published directly prior to the 2017 NZ general election, for which water quality was a major campaign issue. There also appeared to be an increase in the number of media items following the publication of the meta-analysis by Temkin et al. , which observed a statistically significant positive association for nitrate exposure from drinking water and adverse birth outcomes and colorectal cancer risk. Increases in the frequency of media items were also observed on several occasions in 2021 (Fig. 1), especially after the NZ media became aware of the preliminary results of a study by Richards et al. , which estimated that 3.26% (95%CI: 0.84, 5.57) of colorectal cancer cases in NZ were attributable to nitrates, resulting in 100 cases (95%CI: 25.7, 171.3) and 41 deaths (95%CI: 10.5, 69.7) annually.
There were also increases in media items that coincided with the publication of (i) a Letter to the Editor by Chambers et al. , which outlined concerns related to the meta-analysis by Hosseini et al.  and its conclusions on the quantification of the link between nitrates in drinking water and colorectal cancer; (ii) the release of a Public Health Expert (PHE) scholarly blog  published in response to a report co-funded by the major dairy company Fonterra and the NZ Ministry of Business Innovation and Employment (MBIE), which concluded that it is highly unlikely that nitrates in drinking water or diet present an increased risk of cancer ; and (iii) the publication of the study by Richards et al.  in Environmental Research.
The fewest media items were identified in 2020, likely to be partly due to the predominance of Covid-19 pandemic coverage that year. However, the percentage of media items with a health focus steadily increased over time, and there became a stronger focus on the contamination of drinking water with nitrates, particularly in the Canterbury Region, which has experienced rapid intensification of dairy farming and which typically uses extracted groundwater for drinking water.. The suggestive temporal association between the frequency of media items and the timing of key publications (ie, journal articles, reports, and scholarly blogs) presenting new research related to nitrates and human health could benefit from further investigation, possibly through a time series analysis of media items over a longer time period.
Generally, there seems to have been useful input from researchers (especially university-affiliated researchers) into the media discourse on nitrates in water. However, there has been relatively limited input from health officials, especially those at the national level, who often downplayed any potential risks (Table 5). Currently, the official Ministry of Health website on nitrate in drinking water only mentions methemoglobinemia as a potential health hazard . Likewise, public statements from the Associate Minister of Health have focused on existing literature linking nitrate and colorectal cancer as not proving causation .
Pro-industry (agricultural) opinion pieces appear to be adequately balanced by opinion pieces by those concerned with public health, ecological health, and the values of recreational water users (eg, swimmers, anglers). However, there are multiple examples of the lines between agricultural spokespeople and researchers becoming blurred. For example, a report co-funded by Fonterra (the largest dairy company in NZ) and the Government concluded it was highly unlikely nitrate could contribute to bowel cancer . However, this was based on an analysis with important limitations [29, 30]. Additionally, university researchers who were advising key health-related non-governmental organizations (NGOs) and acting on their behalf, have also contributed to a trade magazine for Ravensdown (the largest fertilizer company in NZ) . Other researchers who have contributed to the media discourse while presenting themselves as independent, also sit on the boards of major agricultural companies and have failed to provide appropriate conflict of interest statements .
The media occasionally gave voice to poorly informed views (eg, stakeholders blaming improbable sources for water quality issues), and it is unclear whether there is sufficient skepticism by the media on pro-industry statements given the inherent competing commercial interests. Of particular concern, Māori perspectives were often neglected in health-focused media items. In addition, many media items tended to be one dimensional in their presentation of issues related to nitrates in water. For example, many health focused articles did not discuss the other dimensions of environment, recreational use, and cultural aspects of water quality protection. This may obscure the co-benefits of protecting freshwater from nitrate contamination.
Study strengths and limitations
This study presents the first media discourse analyses of the issue of nitrates in water outside of Europe (to the best of our knowledge). Two previous studies, both conducted in Germany, have presented the results of discourse network analyses on the case of nitrate pollution of groundwater. Vogeler et al.  reported that clear sectoral divisions were visible in a region of Germany with intensive livestock farming, with an agrarian coalition that included individual farmers, farmers’ interest groups, municipalities that relied heavily on the agricultural sector, and certain political parties; and an environmental coalition that included environmental organizations and initiatives at the local level, as well as administrations responsible for environmental protection, and certain political parties (eg, Green Party). Schaub et al.  analyzed newspaper articles and press releases from 2010 to 2020 in Germany, and the results also indicated that two opposing stakeholder coalitions had formed over the issue of fertilizer regulation, and that the coalition in favor of tighter fertilizer regulation had highlighted the potential public health risks associated with contaminated drinking water. In this present study, strongly polarized views were also observed in the media discourse in NZ, a high-income nation with intensive dairy farming and other industrial agricultural production.
Limitations of this study include potential issues with content analysis coding, a relatively short time period, and focusing only on media items in major newspapers. Specifically, while there was good inter-observer reliability, it is still possible that there may have been some aspects of the discourse that were sub-optimally coded. Additionally, this study only examined media items for the five-year period from 17 December 2016 to 20 December 2021. Therefore, only three different central government administrations in NZ were captured, albeit two led by the same political party. Furthermore, we only searched for media items in major newspapers and may have missed important discourse that took place on radio, television, or social media platforms, or Māori and agricultural sector media. Further research that includes these other media formats would help to provide a more robust analysis of the discourse.
A further potential limitation of this study is that it has been conducted by a group of researchers who have actively contributed to the science outputs described here that have impacted the media discourse around nitrates, water, and human health in NZ. We do not consider this perspective has reduced the validity of our findings. It is a descriptive study, the methods used have been well described, and the analysis could be easily repeated by others.
Potential implications for further research and more informed public discourse
As noted above, further analysis of different types of media for a longer time period (eg, 10 years), along with interviews with journalists and key spokespeople would help to provide a more robust view of the media discourse in NZ. Additionally, further investigation of the potential link between the release of new research and the timing of media items is warranted, possibly through a time series analysis. Investigation of how media coverage of public health issues influences decision making is also needed, although it was beyond the scope of this study. Such work could be justified given the importance of risk communication in communities that could be impacted by the contamination of drinking water with nitrates.
These issues are also relevant for the handling of other environmental health risks. Some environmental hazards result in unequivocal public health harm, such as enteric pathogen contamination of drinking water. However, the implications of other environmental contaminants, such as nitrates in drinking water, that are likely to have negative long-term effects, are more contested. This is where concepts such as the ‘precautionary principle’ are often proposed to prioritize the protection of public health in situations where the evidence is uncertain and management of risk can only be over a long period of time. An effort to discuss the co-benefits of protecting freshwater from nitrate contamination in the media could encourage the adoption of the precautionary principle by regulators and policy makers.
There are potential implications for policy making. While a diverse range of opinions were evident in this discourse analysis, often the views presented were very divisive (eg, farmers vs environmentalists) and it is possible that domains of consensus may have been overlooked in the media. Consequently, policy makers may be given the impression that they need to balance deeply polarized views that are not necessarily representative of public opinion. Therefore, as part of the policy making process it may be useful to pair a period of media discourse with a short period of a Polis-like process, as is used in Taiwan . Polis has been directly used as part of policy making at both the national and local levels . This type of process uses a pro-social media platform which builds domains of consensus so that policy makers can see normalized positions upon which to subsequently build policy solutions. For example, if such a process was used to help set new regulatory levels for nitrates in NZ, it might more clearly show politicians where New Zealanders stand on the issue of water quality. Nevertheless, previous surveys have suggested that the NZ public are generally very united in their views on freshwater and both urban and rural New Zealanders are equally concerned and identify agriculture as the primary cause of water pollution , despite media suggestions of a deep divide between urban and rural NZ.