The application of advanced statistical methods to investigate the epidemiology and improve the surveillance of companion animal poisonings using data from a North American poison control call centre

美国防止虐待动物协会（ASPCA）动物中毒控制中心（APCC）目前运营着一个毒理学呼叫中心，每年收到来自加拿大和美国各地的10万多个电话。APCC维护着一个大型数据库，涉及各种物种（包括猫和狗等伴侣动物）接触杀虫剂、药物、植物、金属和其他毒素的动物病例。 他们的数据库包括有关动物特征（年龄，品种和物种），呼叫者类型（兽医或公众），临床体征，可能涉及的器官系统，疑似毒素以及呼叫的地理位置和时间的信息。中毒事件的爆发，如与受污染的宠物食品有关的中毒事件，已被记录在案，并可能利用该数据库进行检测，以便采用先进的统计和流行病学方法进行监测。同样，这些数据可以提供大量有关中毒事件性质和这些事件报告的区域和季节差异的流行病学信息。 该研究计划将在未来五年内使用这些数据进行以下分析研究，以调查猫和狗中毒事件的流行病学，以及这些数据用于监测目的的潜力：*1。根据公众和兽医诊所报告的数据，描述涉及猫和狗的特定毒物、临床体征和器官系统的呼叫率的时间和空间分布。* 2.使用旨在评估不同组织级别的风险因素的多层次统计模型，以评估宠物特征、地理区域、社会经济因素、季节和数据来源等因素如何与特定毒物、临床体征和器官系统相关的呼叫率相关。3.使用空间、时间和时空聚类检测方法来确定中毒病例的爆发以及特定毒素和临床体征或临床体征组的高风险区域和时间（即，综合征）。 我们还将在运行这些分析之前评估对已知区域和时间因素进行统计控制对数据的影响，以了解消除系统性“噪声”对这些时空聚类检测方法性能的影响。4.调查机器学习技术和多元统计方法的能力，以识别可能在毒素来源不清楚时预测与呼叫相关的毒药/毒素的综合征。这些研究将提高我们对影响伴侣动物中毒的因素的理解，确定社区和时间，以针对特定信息来预防中毒事件，并确定使用这些数据和其他数据进行疾病监测的最佳方法。

The American Society for the Prevention of Cruelty to Animals' (ASPCA) Animal Poison Control Center (APCC) operates currently a toxicology call centre that receives over 100,000 calls each year from across Canada and the United States. The APCC maintains a large database concerning animal cases involving pesticide, drug, plant, metal and other toxin exposures in a variety of species including companion animals like cats and dogs. Their database includes information concerning the characteristics of the animal involved (age, breed, and species), the type of caller (veterinarian or general public), clinical signs, probable organ systems involved, suspected toxin, and the geographical location and time of the call. Outbreaks of poisonings, such as those associated with contaminated pet food, have been documented and might be detected using this database for surveillance purposes with advanced statistical and epidemiological methods. Similarly, these data could offer a great deal of epidemiological information concerning regional and seasonal differences in the nature of poisoning events and the reporting of these events. This research program will conduct the following analytical studies using these data over the next five years to investigate the epidemiology of these poisoning events in cats and dogs, and the potential of these data for surveillance purposes:***1. Describe the temporal and spatial distribution of the rate of calls involving specific poisons, clinical signs, and organ systems for cats and dogs based on data reported from the general public and veterinary clinics.***2. Use multilevel statistical models designed to evaluate risk factors at different organizational levels to assess how factors including pet characteristics, geographical region, socio-economic factors, season, and source of data are associated with the rates of calls concerning specific poisons, clinical signs, and organ systems.***3. Use space, time, and space-time cluster detection methods to identify outbreaks of poisoning cases and high risk areas and times for specific toxins and for clinical signs or groups of clinical signs (i.e., syndromes). We will also assess the impact of statistically controlling for known regional and temporal factors on the data prior to running these analyses to understand the impact of removing systematic "noise" on the performance of these space-time cluster detection methods.***4. Investigate the ability of machine learning techniques and multivariate statistical approaches to identify syndromes that might predict the poison/toxin associated with a call when the source of the toxin is unclear.***These studies will improve our understanding of factors that influence poisonings in companion animals, identify communities and times to target specific information to prevent poisoning events, and identify the best approaches to use these and other data for disease surveillance.**