School of Epidemiology & Public Health, University of Ottawa, 600 Peter Morand Cres, Ottawa, ON, Canada.
Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, ON, Canada.
BMC Infect Dis. 2018 Mar 27;18(1):141. doi: 10.1186/s12879-018-3052-6.
West Nile Virus (WNV) is a mosquito-borne pathogen that has become established in North America. Risk for human infection varies geographically in accordance with climate and population factors. Though often asymptomatic, human WNV infection can cause febrile illness or, rarely, neurologic disease. WNV has become a public health concern in Canada since its introduction in 2001.
To identify predictors of human WNV incidence at the public health unit (PHU) level in Ontario, Canada, we combined data on environmental and population characteristics of PHUs with historical mosquito and human surveillance records from 2002 to 2013. We examined the associations between annual WNV incidence and monthly climate indices (e.g. minimum and maximum temperature, average precipitation), land cover (e.g. deciduous forest, water), population structure (e.g. age and sex composition) and the annual percentage of WNV-positive mosquito pools from 2002 to 2013. We then developed a generalized linear mixed model with a Poisson distribution adjusting for spatial autocorrelation and repeat measures. Further to this, to examine potential 'early season' predictors of WNV incidence in a given year, we developed a model based on winter and spring monthly climate indices.
Several climate indices, including mean minimum temperature ( C) in February (RR = 1.58, CI: [1.42, 1.75]), and the annual percentage of WNV-positive mosquito pools (RR = 1.07, CI: [1.04, 1.11]) were significantly associated with human WNV incidence at the PHU level. Higher winter minimum temperatures were also strongly associated with annual WNV incidence in the 'early season' model (e.g. February minimum temperature (RR = 1.91, CI: [1.73, 2.12]).
Our study demonstrates that early season temperature and precipitation indices, in addition to the percentage of WNV-positive mosquito pools in a given area, may assist in predicting the likelihood of a more severe human WNV season in southern regions of Ontario, where WNV epidemics occur sporadically.
西尼罗河病毒（WNV）是一种通过蚊子传播的病原体，已在北美立足。人类感染的风险因气候和人口因素在地理上有所不同。尽管通常无症状，但人类 WNV 感染可引起发热性疾病，或罕见地引起神经疾病。自 2001 年引入加拿大以来，WNV 已成为加拿大的公共卫生关注点。
为了确定加拿大安大略省公共卫生单位（PHU）层面人类 WNV 发病率的预测因素，我们将 PHU 的环境和人口特征数据与 2002 年至 2013 年的蚊子和人类监测记录相结合。我们检查了每年 WNV 发病率与每月气候指数（例如，最低和最高温度，平均降水量）、土地覆盖（例如，落叶林，水）、人口结构（例如，年龄和性别构成）以及 2002 年至 2013 年每年 WNV 阳性蚊子池的百分比之间的关联。然后，我们使用具有泊松分布的广义线性混合模型进行调整，以适应空间自相关和重复测量。在此基础上，为了检查给定年份 WNV 发病率的潜在“早期季节”预测因素，我们基于冬季和春季每月气候指数开发了一个模型。
几个气候指数，包括 2 月的平均最低温度（C）（RR=1.58，CI：[1.42，1.75]）和当年 WNV 阳性蚊子池的百分比（RR=1.07，CI：[1.04，1.11]）与 PHU 层面的人类 WNV 发病率显著相关。较高的冬季最低温度也与“早期季节”模型中的年度 WNV 发病率密切相关（例如，2 月最低温度（RR=1.91，CI：[1.73，2.12]）。
我们的研究表明，早期季节的温度和降水指数，以及特定地区 WNV 阳性蚊子池的百分比，可能有助于预测安大略省南部地区更严重的人类 WNV 季节的可能性，WNV 流行在那里零星发生。
