Massey Documents by Type
Permanent URI for this communityhttps://mro.massey.ac.nz/handle/10179/294
Browse
79 results
Search Results
Item Spatial and temporal epidemiology of FMD in Bhutan (2011–2019)(BioMed Central Ltd, 2025-08-19) Letho S; Compton CBackground: Foot and Mouth Disease (FMD) is an important disease in livestock in Bhutan due to its significant social and economic impacts to the farmers as well as to the government. FMD outbreaks continue to occur every year with greater frequency in some parts of the country despite the implementation of control measures. It is imperative to understand the current patterns of the disease for planning effective control programs. Therefore, this study aimed to investigate the spatial and temporal patterns of FMD in Bhutan since the most recent national survey. Methodology: Nine year’s (2011–2019) of national FMD outbreak data was used for this study. An investigation of global spatial autocorrelation was undertaken using the K difference function statistic and local spatial clusters of FMD outbreaks using Kulldorff’s spatial scan statistic. Retrospective spatio-temporal analysis was conducted using a Bernoulli probability model with monthly aggregation of data. A non-stationary cosinor model was used to examine seasonality and trend of outbreaks. Results: The K function statistic detected significant global spatial autocorrelation of FMD outbreaks (p < 0.02) and Kulldorff’s spatial scan statistic identified two outbreak clusters in west Bhutan- the first primary cluster (p < 0.002) with relative risk (RR) of 5.22 and radius of 19.77 km in Paro and Thimphu districts and the second primary cluster (p < 0.006, RR = 8.44 radius: 8.98 km) in Punakha and Wangdue Phodrang districts. The spatio-temporal scan test detected a single significant (P < 0.001) space–time cluster of 22 FMD outbreaks centred in south-west Bhutan with a radius of 60 km over an 8-month period in 2018—2019. The temporal analysis indicated that, on an average there were 0.5 (95% CI: 0.2—0.8) additional outbreaks per month in the seasonal winter peaks at 1.9 months (95% CI: 12.55 -3.64) compared with the overall monthly average. Conclusion: The western and southern regions of Bhutan have experienced the greatest overall incidence and significant spatial and spatio-temporal clusters of outbreaks of FMD during the period 2011—2019, These findings in districts in the western medium FMD risk surveillance zone threaten progress to control of FMD in Bhutan.Item Is early life antibiotic-use a risk factor for the development of Type 1 diabetes? : a longitudinal data linkage study : a thesis with publications presented in partial fulfillment to the requirements for the degree of Doctor of Philosophy in Public Health (Epidemiology), Massey University, Wellington, New Zealand(Massey University, 2025-09-01) Ram, SharanIntroduction: Early-life antibiotic-use may disrupt the gut microbiota, potentially increasing the risk of Type 1 diabetes (T1D). This thesis assessed associations between prenatal and early childhood antibiotic exposure and T1D. Methods: A meta-analysis of 14 studies (2006-2020), encompassing 3,066,063 participants, assessed associations between early-life antibiotic-use and T1D. Subsequently, a longitudinal nation-wide linkage study examined this association in 315,789 New Zealand children born between 2005-2010 and followed until 2021, using Cox proportional hazards regression controlled for potential confounders. Patterns of antibiotic-use during pregnancy and early childhood were also analysed across demographic characteristics. Associations between both pre and post natal antibiotic exposure were assessed, including analyses by antibiotic class and spectrum, and stratification by delivery mode. T1D was identified using insulin dispensing and hospitalisation records. Results: The meta-analysis, showed a pooled risk estimates for prenatal exposure of 1.10 (95% CI: 1.00–1.21); for postnatal antibiotic-use a pooled risk estimate of 1.11 (95%CI 1.04–1.18) was found, with ≥5 courses resulting in a pooled estimate of 1.36 (95%CI 1.15–1.61). Broad-spectrum antibiotics were associated with higher risk (HR: 1.13, 95% CI: 1.03–1.23). In New Zealand, 30% of pregnant women received antibiotics, predominantly penicillin (73.7%). Higher usage was observed among Pacific (38.7%) and Māori (35.7%) women, those most deprived (i.e. those from the lowest socio-economic group) (39.5%). Those who had caesarean deliveries had higher rates of antibiotic use, with incidence rate ratios (IRRs) of 1.27 for elective and 1.09 for emergency procedures. By age five, 96% of children had received antibiotics, with similar subgroup patterns as observed for pregnant women. Prenatal antibiotic-use was associated with an increased T1D risk in a dose-dependent fashion (≥3 courses, HR 1.86; 95%CI:1.44–2.39), with the highest risk for broad-spectrum antibiotics (HR: 1.30; 95%CI: 1.12–1.57). Postnatal antibiotic-use was associated, also in a dose-dependent way, with T1D (≥13 courses, HR 1.93; 95%CI 1.18–3.17; broad-spectrum antibiotics, HR 1.74; 95%CI 1.10-2.78). Stratified analyses by delivery mode resulted in mixed results across different analyses. Conclusion: These findings show high antibiotic-use in New Zealand and among specific ethnic and socio-economic subgroups. It also showed clear associations with the development of childhood T1D, which is consistent with international studies as shown in the meta-analysis, underscoring the need for judicious antibiotic stewardshipItem Epidemiology of non-communicable diseases : a collection of published papers presented in application for the degree of Doctor of Science at Massey University(Massey University, 2002) Pearce, NeilThe scientific publications contained in this thesis represent more than 20 years of work in the epidemiology of non-communicable disease. Chapter 1 includes methodological papers that are relevant to non-communicable diseases and epidemiology in general. This work falls into four main areas. Firstly, several papers discuss the analysis of epidemiologic studies. These particularly involve work that I did in the 1980s on the analysis of cohort studies, particularly involving time-related factors. At that time, there were well-established methods for analysing case-control studies, but the analysis of cohort studies was less developed, and there were few programmes or analytical approaches available for taking time-related factors into account. More recently, I have published on epidemiologic concepts of interaction and the implications for approaches to data analysis. Secondly, two papers from the 1980s discuss the epistemological basis of epidemiologic research. At that time, the leading textbook of epidemiology advocated a Popperian approach to the philosophy of science. These papers discussed the limitations of the Popperian approach, and showed that the actual practice of epidemiologists was more consistent with alternative approaches to the philosophy of science. Thirdly, several papers discuss the principles of epidemiologic study design. My contributions particularly involve the theory and practice and case-control studies, and the demonstration that these do not involve "backwards causality" or a radically different study design to that used for cohort studies. Finally, in recent years I have played a major role in an ongoing debate on the future of epidemiology, and published a series of papers on the need for epidemiology to rediscover the population perspective and to use theories and methods that take the population context into account. Chapters 2-5 then include substantive studies of these particular public health issues, as well as more specific methodological papers arising from these studies. My early interest was in occupational and environmental health, with a particular emphasis on cancer, and this work is included in chapter 2. This work falls into three major areas. Firstly, my methodological work particularly includes the textbook of occupational epidemiology published by Oxford University Press in 1989, which I co-authored. This also produced a series of papers on the theory, practice, and teaching of occupational epidemiology. Once again, it particularly involved the analysis of cohort studies, and the design of case-control studies for the specific situation of occupationalhealth research. Secondly, my occupational cancer work has particularly included studies of cancer in farmers and related occupations. These studies were the first to show that meat workers have an increased risk of some types of cancer. This work has also included the firstoccupational cancer cohort study in New Zealand, which examined cancer risk in nuclear test veterans. Finally, I have also studied the particular issues of occupational cancer in developing countries, and the more general issues of environmental change and human health. Chapter 3 includes papers on socio-economic determinants of health, with a particular emphasis on non-communicable disease. In the 1980s I conducted the first studies of socio-economic differences in health and life expectancy in New Zealand. Since the first publication in 1983, at a time when there was little interest in this field of research, this has grown to become a major area of research in New Zealand. My work in Māori health has involved assisting the work of the late Professor Eru Pomare and others in documenting and discussing the causes of differences in health between Māori and non-Māori in New Zealand. This included the first studies to show that the high mortality rates in Māori were not solely to due to socio-economic factors, but also represented problems of access to health care for Māori. Chapter 4 covers studies of asthma mortality, particularly the studies of the role of fenoterol in the New Zealand asthma mortality epidemic of the 1970s and 1980s. These studies were the first to identify that the beta agonist fenoterol was responsible for an epidemic of asthma deaths in New Zealand. These studies were highly controversial, and were strongly criticised, but eventually the New Zealand Ministry of Health restricted the availability of fenoterol in New Zealand and the asthma death rate immediately fell by one half. Unlike the other chapters in this thesis, the work described in this chapter was more clearly done as part of a group, rather than by myself as an individual. The group included an epidemiologist (myself), a clinical research, respiratory physician and a pharmacologist. Inevitably with such a multidisciplinary approach the authorship was variable, although I played the major role in the design and analysis of the epidemiological studies. I have included several review papers since they give an overview of research in this area, and the relative contribution of my own studies to the development of this debate. Finally, chapter 5 includes studies on the prevalence, causes and management of asthma, as well as related methodological work. This work falls into four main areas. Firstly, I have produced a textbook of asthma epidemiology, published by Oxford University Press. This has involved a substantial amount of work in developing the theory and practice of epidemiology in the context of asthma. This is a relatively new field, since asthma is a non-fatal chronic disease that is difficult to diagnose, and asthma studies usually involve prevalence rather than incidence or mortality. This work has led to a number of review papers on asthma epidemiology methods. Secondly, I have been heavily involved in the development of the International Study of Asthma and Allergies in Childhood (ISAAC). Phase I of the study involved more than 700,000 children in 155 centres in 60 countries, and Phase III is currently in progress. Thirdly, in more recent years my research has focussed on the primary causes of asthma, and particularly on the role of non-allergic mechanisms. In addition to specific studies on environmental and occupational causes of asthma, I have published a series of reviews and commentaries that question the importance of allergic mechanisms for asthma. This series of papers is becoming increasingly influential in re-orienting asthma research towards a greater interest in non-allergic mechanisms. My current research interests focus on studies of these mechanisms.Item Mpox: A case study for a one health approach to infectious disease prevention(Elsevier B V, Amsterdam, 2025-06) Hayman DTS; Koopmans MPG; Cunningham AA; Bukachi SA; Masirika LM; Markotter W; Mettenleiter TCMpox has been declared a global health emergency twice by the World Health Organization due to its impacts within and beyond Africa. Enzootic in Central and West African wildlife, mpox outbreaks have resulted from zoonotic spillover, with recent events revealing increased human-to-human transmission. Factors like population growth and environmental disruption, alongside reduced smallpox immunity, increase emergence risk. In addition, the emergence in South Kivu of a distinct subclade of mpox virus points at a currently understudied aspect of mpox virus lineages and their dynamics in reservoir hosts. A One Health approach—integrating human, animal, and environmental science—is essential for reducing the risk of mpox emergence. This approach should encompass ecological studies to understand putative reservoir population dynamics and the potential for interventions, reducing activities that increase human-animal contacts, respectful community engagement to reduce spillover risk from cultural practices (such as hunting multiple species of wildlife for consumption), and socially acceptable and equitable access to medical and non-medical countermeasures to prevent or control ongoing human-to-human transmission. Politically supported collaborative efforts across disciplines with involvement of stakeholders are critical to promote and strengthen socially and environmentally sustainable practices to mitigate future outbreaks.Item Development of evidence-based strategies to control Brucella spp. in dairy herds in Henan Province, China : a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Veterinary Epidemiology at Massey University, Manawatu, New Zealand(Massey University, 2025-03-25) Wang, YuBrucella spp. remains a significant challenge in China, affecting the dairy industry and public health despite decades of voluntary control measures. This thesis aims to address key knowledge gaps in diagnosis, epidemiology, and disease impact of Brucella spp. in dairy herds to inform evidence-based control strategies in Henan Province, China. A scoping review of 61 studies (2004-2022) characterized the epidemiological landscape, revealing that B. abortus biovar 3 predominated 85.8% of Brucella spp. isolates recovered from dairy cattle in China. Considerable heterogeneity was observed in prevalence estimates and Brucella spp. isolation across provinces. These findings guided subsequent investigations into diagnostic accuracy, biosecurity practices, disease impact, and financial analyses. To enhance diagnostic accuracy, a cross-sectional study evaluated the diagnostic performance of four serological tests in Henan dairy herds. Using a novel Bayesian latent class model, optimal cut-off values were established for fluorescence polarization assay and competitive ELISA, estimating test sensitivity (69.7%-89.9%) and specificity (97.1%-99.6%). These findings provide a foundation for improving brucellosis diagnostic strategies. Beyond diagnosis, an assessment of farm biosecurity practices and stakeholder motivations was conducted using a structured questionnaire. Two distinct clusters of dairy herds were identified. Cluster 1, characterized by medium (400-1000) to large (>1000) herd sizes and higher educational levels, showed better adherence to proper biosecurity practices than Cluster 2. Stakeholders prioritized disease impacts and economic losses, highlighting the need for integrating these into the brucellosis control programs. To address this knowledge gap about disease impact, a longitudinal study estimated the effects of Brucella seroconversion on key production indicators. The annual incidence of seroconversion of brucellosis was 13.1% (95% CI: 10.9, 15.6) at the cow level. Seroconverted cows exhibited a reduction in daily milk yield (3.2 kg/day, 95% CI: 2.4, 4.0), elevated somatic cell counts, and increased pregnancy loss (relative risk: 4.26, 95% CI: 3.17, 5.73), compared to that of consistently negative cows. These findings provided essential epidemiological insights into the seroconversion of Brucella spp. and its implications on three essential dairy productivity outcomes. Building on these epidemiological insights, a financial analysis estimated the direct annual costs of Brucella infection at 78.9 Chinese Yuan (CNY) per animal and 4,019 CNY per infected cow. Among the three evaluated interventions (vaccination, test-and-culling, and test-and-culling plus vaccination), vaccination was the most cost-effective, yielding a benefit-cost ratio of 5.84 (95% CI: 4.34 – 7.42) and a net present value of 408.2 thousand CNY over ten years. Overall, this thesis integrates epidemiological, diagnostic, biosecurity, and financial analyses to inform evidence-based brucellosis control in Henan dairy herds. These findings could support farm stakeholders and policymakers in improving the control of Brucella spp. in Henan dairy herds, contributing to broader brucellosis control efforts in China.Item Changing epidemiology of Leptospirosis in New Zealand, with a focus on the novel strain of Leptospira borgpetersenii : a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Veterinary Science at Massey University, Palmerston North, New Zealand(Massey University, 2024-12-25) Sokolova, MarynaIn New Zealand, leptospirosis has been a common disease in dairy cattle since the 1940s. Six pathogenic Leptospira serovars from two species have been identified as endemic to New Zealand: Leptospira borgpetersenii serovars (sv.) Hardjobovis (Hardjo), Tarassovi, Ballum, Balcanica, and Leptospira interrogans sv. Pomona and Copenhageni. From these, sv. Pomona and Hardjo are the most commonly reported in cattle, and sv. Ballum, Tarassovi, and Copenhageni are less common. The estimated 99% of the national dairy herd is vaccinated against leptospirosis by vaccines containing antigens to Pomona and Hardjobovis and sometimes vaccines also include Copenhageni antigen. Vaccines for protecting dairy cows against Tarassovi were unavailable in New Zealand before December 2023. Historically, leptospirosis due to Tarassovi infection in New Zealand cattle was considered accidental and clinically unimportant. Serosurveys of apparently healthy cattle in New Zealand showed that over the past fifty years, Tarassovi seroprevalence increased from 6% (50/300) at MAT ≥ 50 to 18% (698/3878) at MAT ≥ 48. More recently, a serology and urine shedding study from 2015- 2016 evaluated the status of 4,000 dairy cows from 200 randomly selected farms, stratified by New Zealand's geographical location and herd size. The study found that on the animal level, 17% of the study animals showed evidence of past infection with Tarassovi, as defined by at least one sample reacting at MAT ≥ 48 for the study's purposes. Moreover, 96% (90/94) of PCR-positive urine samples were sequenced, and 68% (54/80) of those were found to have a novel allele in the sequenced region at the glmU loci. Thus, the use of molecular diagnostic tools, specifically molecular typing targeting a partial region of the glmU gene, allowed New Zealand researchers to identify a novel L. borgpetersenii strain, informally called strain (str.) Pacifica, in the urine of these cows. The 2015-2016 survey reported that dairy cows with Tarassovi titres were associated with urinary shedding, as determined by microscopic agglutination test (MAT) and polymerase chain reaction (PCR) testing, respectively. Because of this association, str. Pacifica is thought to belong to the serogroup Tarassovi. Additionally, the DNA of str. Pacifica has been retrospectively detected in cattle and deer samples dating back as early as 2007. Moreover, the 19-year (1999-2007) average annual incidence of notified human cases of Tarassovi leptospirosis was estimated at 12.59/100,000 in dairy farmers, compared to an overall annual average incidence of 2.01/100,000. These coincidental findings raised public health concerns. Therefore, at least some cases of Tarassovi seropositivity, as identified by MAT and reported before 2021, could partially be attributed to str. Pacifica. To confirm str. Pacifica's serogroup, isolation by culture and complete genetic characterisation of an isolate are required. Since str. Pacifica was only recently detected, its epidemiology, morphology, maintenance, and pathogenicity in the host population, as well as its impact on animal and human health, were not well understood. In this study, we investigated the possibility of isolating str. Pacifica from cow's urine by running a series of laboratory experiments where laboratory-adapted strains were used as a proxy for L. borgpetersenii str. Pacifica in the absence of an isolate to better understand its growth requirements. Laboratory-adapted Leptospira borgpetersenii strains were seeded into different types of media, and Leptospira growth rates were evaluated (Chapter 3). As a result, we ruled out unsuitable media and growth conditions, and this work helped to select the best media and growth conditions for a follow-up field investigation, where freshly collected cow's urine was seeded into selected media. Str. Pacifica was isolated from the urine of a shedding cow using HAN medium at 37°C and 5% CO2. However, this medium failed to sustain str. Pacifica and the culture was lost (Chapter 4). In addition, over the 2020-2021-2022 milking seasons, we collected and tested blood and urine samples from dairy farms, identified as str. Pacifica positive from the 2016 survey. Our results revealed that str. Pacifica was still maintained in the same dairy herds six years after initial detection. Moreover, we reported an estimated prevalence ratio (PR) of 7, indicating that the prevalence of shedders was seven times as high at the beginning than at the end of lactation in primiparous cows (Chapter 4). These findings provide evidence that str. Pacifica is adapted to dairy cows in New Zealand, and the peak shedding in primiparous heifers occurs in early lactation. Since the highest levels of str. Pacifica shedding were detected at the start of the milking season during peak milk production, which also coincides with a relatively short 3-month mating period, the associations between str. Pacifica and milk production and reproductive performance of milking cows were also investigated using herd test data and serological and PCR test results of the 2016 survey. Statistical models, including linear, logistic, and generalised mixed models with fixed and random effects, as well as a shared frailty survival model, were used to evaluate the associations between str. Pacifica positivity and reproduction (Chapter 5) and milk reproduction (Chapter 6) in dairy cows. Results of the statistical analysis of the association between str. Pacifica positivity and reproduction (Chapter 5) of dairy herds showed that str. Pacifica delayed the time from calving to conception (HR = 0.84; 95%: CI 0.74-0.96), although there was no effect on the pregnancy rate (Chapter 5). An analysis of milk production data did not reveal any associations with str. Pacifica at either animal or herd level (Chapter 6). The absence of clinical signs and the lack of association with milk production and reproduction at both the animal and herd levels provides further evidence towards str. Pacifica being well adapted to dairy cows. Therefore, evidence from this thesis suggests that dairy cattle are the maintenance host for str. Pacifica in New Zealand. It is also important to note that str. Pacifica shedders can infect people, especially dairy farmers, milkers, and farm workers who are in regular contact with str. Pacifica-shedding animals. Therefore, the prevention of str. Pacifica transmission via vaccination or the use of appropriate personal protective gear should be prioritised.Item Environmental Health Intelligence New Zealand (EHINZ): intelligence for public health action.(Springer Nature, 2022-09-21) Borman B; Haenfling C; Kowalik-Tait A; Hipgrave PThe New Zealand health system is data-rich, information-poor, and intelligence meagre. However, there is widespread confusion about the definitions of these terms, so they are often used synonymously. Like many jurisdictions, we continue to collect and collate vast quantities of data at an increasing rate. Many tools are available to “analyse” the data deluge with the false expectation that “intelligence” will be produced. Naively, such a data-driven, machine-analysed paradigm is often thought to produce the “evidence” for decision-making and policy development. Continuing such a blinded approach poses potential health risks to New Zealanders and remains a major impediment to improving our health status Creating intelligence from information involves humans (perhaps in concert with AI) utilising their interpretative abilities, asking the “so what, “what does it mean” questions, and employing their communication skills to disseminate the answers. To move from information to intelligence requires agencies to employ, develop and maintain a sufficiently skilled workforce over a long period, rather than the quick and easy investment in more and faster machines and software. Only through a human-driven evaluation of intelligence-based decisions and policies will our knowledge about the environmental health system increase and ultimately yield better health outcomes. Environmental Health Intelligence NZ (EHINZ) provides intelligence as evidence for decision-making and policy development in environmental health. It is based on the interpretation, communication, and dissemination of information from the surveillance more than seventy environmental health indicators (EHIs) across twelve domains (e.g., air and water quality, climate change), exposure to hazardous substances, and social vulnerability indicators to environmental hazards (e.g., flooding, climate change, sea-level rise, wildfires, heat waves). The paper details our approach, with two case studies, in providing the NZ health system with “intelligence for environmental health decisions.”Item Infection thresholds for two interacting pathogens in a wild animal population.(Elsevier B.V., 2024-07-17) Roberts MGWe present a model for the dynamics of two interacting pathogen variants in a wild animal host population. Using the next-generation matrix approach we define the invasion threshold for one pathogen variant when the other is already established and at steady state. We then provide explicit criteria for the special cases where: i) the two pathogen variants exclude each other; ii) one variant excludes the other; iii) the population dynamics of hosts infected with both variants are independent of the order of infection; iv) there is no interaction between the variants; and v) one variant enhances transmission of the other.Item A scoping review on the epidemiology and public significance of Brucella abortus in Chinese dairy cattle and humans(Elsevier B.V., 2024-01-31) Wang Y; Vallée E; Heuer C; Wang Y; Guo A; Zhang Z; Compton CBrucellosis, caused by Brucella spp., is a re-emerging One Health disease with increased prevalence and incidence in Chinese dairy cattle and humans, severely affecting animal productivity and public health. In dairy cattle, B. abortus is the primary causative agent although infections with other Brucella species occur occasionally. However, the epidemiological and comparative importance of B. abortus in dairy cattle and humans remains inadequately understood throughout China due to the heterogeneity in locations, quality, and study methods. This scoping review aims to describe the changing status of B. abortus infection in dairy cattle and humans, investigate the circulating Brucella species and biovars, and identify factors driving the disease transmission by retrieving publicly accessible literature from four databases. After passing the prespecified inclusion criteria, 60 original articles were included in the final synthesis. Although the reported animal-level and farm-level prevalence of brucellosis in dairy cattle was lower compared to other endemic countries (e.g. Iran and India), it has been reported to increase over the last decade. The incidence of brucellosis in humans displayed seasonal increases. The Rose Bengal Test and Serum Agglutination Test, interpreted in series, were the most used serological test to diagnose Brucella spp. in dairy cattle and humans. B. abortus biovar 3 was the predominant species (81.9%) and biovar (70.3%) in dairy cattle, and B. melitensis biovar 3 was identified as the most commonly detected strain in human brucellosis cases. These strains were mainly clustered in Inner Mongolia and Shannxi Province (75.7%), limiting the generalizability of the results to other provinces. Live cattle movement or trade was identified as the key factor driving brucellosis transmission, but its transmission pattern remains unknown within the Chinese dairy sector. These knowledge gaps require a more effective One Health approach to be bridged. A coordinated and evidence-based research program is essential to inform regional or national control strategies that are both feasible and economical in the Chinese context.Item Creating symptom-based criteria for diagnostic testing: a case study based on a multivariate analysis of data collected during the first wave of the COVID-19 pandemic in New Zealand(BioMed Central Ltd, 2021-12) French N; Jones G; Heuer C; Hope V; Jefferies S; Muellner P; McNeill A; Haslett S; Priest PBACKGROUND: Diagnostic testing using PCR is a fundamental component of COVID-19 pandemic control. Criteria for determining who should be tested by PCR vary between countries, and ultimately depend on resource constraints and public health objectives. Decisions are often based on sets of symptoms in individuals presenting to health services, as well as demographic variables, such as age, and travel history. The objective of this study was to determine the sensitivity and specificity of sets of symptoms used for triaging individuals for confirmatory testing, with the aim of optimising public health decision making under different scenarios. METHODS: Data from the first wave of COVID-19 in New Zealand were analysed; comprising 1153 PCR-confirmed and 4750 symptomatic PCR negative individuals. Data were analysed using Multiple Correspondence Analysis (MCA), automated search algorithms, Bayesian Latent Class Analysis, Decision Tree Analysis and Random Forest (RF) machine learning. RESULTS: Clinical criteria used to guide who should be tested by PCR were based on a set of mostly respiratory symptoms: a new or worsening cough, sore throat, shortness of breath, coryza, anosmia, with or without fever. This set has relatively high sensitivity (> 90%) but low specificity (< 10%), using PCR as a quasi-gold standard. In contrast, a group of mostly non-respiratory symptoms, including weakness, muscle pain, joint pain, headache, anosmia and ageusia, explained more variance in the MCA and were associated with higher specificity, at the cost of reduced sensitivity. Using RF models, the incorporation of 15 common symptoms, age, sex and prioritised ethnicity provided algorithms that were both sensitive and specific (> 85% for both) for predicting PCR outcomes. CONCLUSIONS: If predominantly respiratory symptoms are used for test-triaging, a large proportion of the individuals being tested may not have COVID-19. This could overwhelm testing capacity and hinder attempts to trace and eliminate infection. Specificity can be increased using alternative rules based on sets of symptoms informed by multivariate analysis and automated search algorithms, albeit at the cost of sensitivity. Both sensitivity and specificity can be improved through machine learning algorithms, incorporating symptom and demographic data, and hence may provide an alternative approach to test-triaging that can be optimised according to prevailing conditions.