The Influence of Climate and Livestock Reservoirs on Human Cases of Giardiasis

Enteric illness caused by the intestinal parasite Giardia duodenalis (Gd) is reportable in Canada. The parasite is the third-leading cause of enteric disease in the province of Ontario with an average provincial incidence of 9.9 cases per 100 000 person-years reported between 2011 and 2015 (Public Health Ontario 2017). Many patients remain asymptomatic or are only transiently afflicted causing underreporting of the true prevalence. Symptomatic patients contribute to the estimated 89.5 million disability-adjusted-life-years lost world-wide due to diarrheal disease (Conlan and Lal 2015; Murray et al. 2012). However, correlations between socioeconomic factors such as lack of plumbing and an increased risk of infection may explain why giardiasis has a higher prevalence in lower-income countries (Silva et al. 2009). Symptomatic cases of giardiasis may exhibit mild abdominal discomfort, diarrhoea, vomiting and weight loss, however, most infections are self-resolving.

Gd is composed of eight assemblages (A–H), which are genetic groupings with varying degrees of host-specific pathogenicity. Assemblages A and B primarily cause disease in humans (Yaoyu and Xiao 2011) but have zoonotic potential as they have been isolated from cattle, swine, dogs and cats, as well as some wild mammals (Cacciò and Ryan 2008; Farzan et al. 2011; Yaoyu and Xiao 2011). Assemblage types C–H have limited host ranges of domestic and wild animals, but rare occurrences in people have been reported (Yaoyu and Xiao 2011). Zoonotic assemblages in livestock contribute to human disease through direct exposure or through manure contaminated water sources used for recreation or drinking water (Hunter and Thompson 2005; Farzan et al. 2011; Abdel-Moein and Saeed 2016). Ontario dairy cattle and pigs have been shown to harbour zoonotic subtypes of Gd (Coklin et al. 2007; Dixon et al. 2011; Farzan et al. 2011) and cysts shed in host faeces are immediately infective in the environment. The low specific gravity of cysts along with high precipitation enhances transport from soil into water sources (Erickson and Ortega 2006; Wilkes et al. 2011). Cysts are well adapted for wet and humid conditions, remaining infective in water with a temperature of 0–4 °C for 84 days (Yaoyu and Xiao 2011). Warmer air temperatures lead to increased recreational use of water sources by people and hydrological characteristics of watersheds such as water flow and level can affect the suspension and sedimentation rate of cysts in surface waters (Erickson and Ortega 2006) facilitating Giardia cyst ingestion by people (Thomas et al. 2006; Peng et al. 2008). The most common risk factors for human infection in Ontario include ingestion of contaminated water (e.g. drinking or recreational water), direct transmission via the faecal-oral route from infected people or animals, or consumption of contaminated food such as unwashed vegetables (Public Health Ontario 2017). Although outbreaks of giardiasis have been reported globally, they are uncommon in Ontario with a recent descriptive study suggesting that 99% of Giardia cases are not linked epidemiologically (Keegan et al. 2009).

The seasonal pattern of human Giardia incidence has been reported with annual peaks in the late summer and early autumn (Addiss et al. 1992; Greig et al. 2001; Naumova et al. 2007; Keegan et al. 2009; Lal et al. 2012). This seasonal fluctuation, combined with the ability of Giardia to persist in the environment, is highly suggestive of a transmission cycle that is sensitive to climate and hydrological factors (Fisman 2007; Greer et al. 2008; Ng et al. 2008). Therefore, our a priori hypothesis is that environmental and hydrological conditions are associated with the occurrence of sporadic human Giardia cases within southern Ontario. Understanding associations between environment, livestock reservoirs, and human disease may enhance prevention and control strategies. In this respect, the use of a one health approach is appropriate for the study of Giardia to avoid simplifying complex relationships (Bouley et al. 2014). The one health concept is a framework that recognizes the interconnectedness between the health of humans, animals and the environment (Bouley et al. 2014). The objectives of this study were to identify temporal associations between environmental conditions, livestock Giardia reservoirs and human giardiasis incidence using an integrated time series (2006–2013) from the Waterloo Health Region (WHR) in Ontario, Canada, and to compare the results obtained using two different methods that account for confounding due to seasonality and annual trends.

This study contributes to our understanding of the relationship between environmental factors, livestock pathogen reservoirs, and the risk of human giardiasis. Our findings of a late summer peak in incidence are in agreement with previous studies examining temporal trends in human giardiasis (Addiss et al. 1992; Naumova et al. 2007; Keegan et al. 2009; Lal et al. 2012). The consistency of a seasonal pattern, along with the persistence of Giardia as a waterborne pathogen (Yaoyu and Xiao 2011), suggests human giardiasis is sensitive to climate and hydrology factors.

Associations between human case occurrence and hydrological variables were identified in both the multivariable and case crossover analysis. Inverse associations between aggregated human case counts and river water levels have been previously reported for another waterborne pathogen, Legionella, in multivariable models (Ng et al. 2008). In contrast to the findings of our multivariable model, the case crossover analysis found that the odds of human cases increased with the water level and flow of the Grand River occurring 1 week before cases. These findings differ from a case crossover study on the enteric protozoa, Cryptosporidium, which identified increased water flow and water levels as having a protective effect on human case risk and attributed this to dilution of pathogen concentrations in the watershed (Brankston et al. 2018). The discrepancy between these results may be related to external control of river flow and level exercised by the Grand River Conservation Authority via a system of dams and water reservoirs which minimize flooding risk and maintain water quality for discharging sewage plants. The Grand River at Doon, Ontario, where hydrology data for this study was collected, is heavily augmented during the late summer dry period. Up to 90% of the total river water volume at Doon in August can come from the reservoir (Grand River Conservation Authority 2014). While the effects of river level manipulation may be lost in monthly averaged exposure data, reservoir release seen as an acute exposure could influence associations being found with human case counts using the case crossover methodology. Alternately, resuspension of enteric pathogens from stream sediments during periods of high water flow, such as stormy weather, has been proposed as an explanation for measurements of increased concentrations of E. coli occurring with high stream flow rates (Dorner et al. 2006). While an explanatory hypothesis for a positive association between human giardiasis cases and water levels and flow rates in relation to increased river cyst concentration is unclear in the present study, cyst survival in river sediments and potential for resuspension is an area for future research. When the presence of livestock reservoirs was added to the model (Objective 2), an increase in the odds of human cases was seen with a 1-week lag period. Surveillance studies have confirmed the presence of zoonotic Giardia assemblages in Ontario cattle and swine herds (Coklin et al. 2007; Dixon et al. 2011; Farzan et al. 2011), providing initial support for the hypothesis of spill-over of zoonotic Gd from livestock reservoirs. Our results would benefit from molecular typing of Giardia cysts obtained from the Grand River to confirm if potentially zoonotic cysts from livestock can survive effluent transport mechanisms into the watershed.

Statistically significant associations between mean temperature and human case incidence from the case crossover analysis resulted in inverse effects at the lowest tertile of mean temperature with a lag period of 3 weeks. In both the full and truncated time series, the lowest mean air temperatures occurring three weeks prior to human cases decreased the risk of giardiasis.

Previous studies have reported associations between time-lagged high temperatures and increased risk of waterborne enteric disease (Thomas et al. 2006; Naumova et al. 2007; Brankston et al. 2018), possibly reflecting changes in human behaviour and greater opportunities for exposure to contaminated surface waters (Thomas et al. 2006; Peng et al. 2008; Brankston et al. 2018). While the results found in this study do not directly support this hypothesis, it is possible that on days with cooler than average temperatures a protective effect is observed as people may be less inclined to recreate in natural water bodies.

In the truncated time series, within-stratum highest precipitation levels occurring 4 weeks prior to human cases were inversely related to giardiasis occurrence (Table 2). This result contradicts two studies that found that high rainfall events preceded increased human enteric case incidence (Thomas et al. 2006; Escobedo et al. 2015), however, one study was based in Cuba where average rainfall is higher than what is observed in Ontario (Escobedo et al. 2015). The second study, based in Canada, used a longer hazard period of 6 weeks and stratified precipitation to assess only extreme rainfall events (Thomas et al. 2006). The difference in results may be related to stratification choice, length of hazard period, or pathogen biology.

Some limitations should be noted. Aggregating cases into monthly counts could mask underlying patterns. For example, environmental variables averaged by month may not reflect the true exposures influencing transmission, potentially leading to imprecise inferences. Case crossover analysis aims to avoid this potential bias by examining associations between acute exposures and daily case occurrence using a self-matched study design. However, the testing of multiple hypotheses in the case crossover method may increase the probability of a Type 1 error resulting in spurious associations. In this study, type 1 errors were minimized by grouping exposures to minimize the number of hypotheses tested.

The livestock sampling data for Gd was constrained by inconsistent monthly sampling and sporadic genotyping results, leading to assemblage type reporting for just over 50% of positive samples. We acknowledge that the Giardia-positive farm counts might not accurately reflect the true number of Giardia-positive farms in the region. Nonetheless, the positive counts taken from surveillance data represent a subset and likely underrepresent the true number of cases, influencing the bias towards the null. Assemblage typing data provided confirmation that potentially zoonotic Gd assemblages were present in livestock reservoirs in the region, supporting the possibility of a zoonotic transmission route. Molecular studies of isolates and genetic assemblage typing of samples would strengthen the spill-over hypothesis. Effect modification between positive-testing farms and environmental variables was not investigated due to limited farm data, however including this as a future direction could contribute to our understanding of livestock disease reservoirs.

Symptom onset dates were reported for 75% (Objective I dataset) and 73% (Objective II dataset) of human case counts used in this study. The remaining cases were included in the study to counteract underreporting which is known to occur however, it is possible that the inclusion of these cases resulted in incorrect estimates of associations and exposure lag periods.

Ecological systems are changing at an unprecedented rate (Walther et al. 2002; Post et al. 2009). With climatic conditions expected to change in the coming years, it is necessary to understand the environmental drivers of climate-sensitive diseases (Bouley et al. 2014; Machalaba et al. 2015). While the design of this study precludes conclusions being drawn regarding cause and effect, the relationships described provide an understanding of the associations between human giardiasis, livestock disease reservoirs, and environmental variables using a one health approach.