EpiCentre, School of Veterinary Science, Massey University, Palmerston North, New Zealand.
Department of Infectious Disease and Public Health, Jockey Club College of Veterinary Medicine and Life Sciences, City University of Hong Kong, Kowloon, Hong Kong.
Transbound Emerg Dis. 2021 May;68(3):1586-1600. doi: 10.1111/tbed.13830. Epub 2020 Sep 22.
This study aimed at estimating parameters representing between-farm transmission of Salmonid Rickettsial Septicaemia (SRS) in Chile, and developing and validating simulation models to predict weekly spread of SRS between farms in Los Lagos (Region 10), using InterSpread Plus. The model parameters were estimated by analyses of the historical SRS outbreak data. The models incorporated time and distance-dependent transmission kernels, representing the probabilities of waterborne spread of SRS between farms. Seven candidate transmission kernels were estimated, with varying maximum distance of between-farm SRS spread (15-60 km). Farms were categorized by size (small; medium; large) and species (Coho salmon; Atlantic salmon; rainbow trout). The time that it took a farm to recover from infection was parameterized to be shortest for small Coho farms (median: 7 weeks), followed by medium and large Coho farms (median: 25 weeks), Atlantic salmon farms (median: 42 weeks, any size) and rainbow trout farms (median: 43 weeks, any size). The relative infectiousness parameters of rainbow trout farms were 1.5-6.3 times that of Coho or Atlantic salmon, or those of large farms was 1.3-4.2 times that of small or medium farms. The models predicted SRS prevalence in Region 10 between 2013 and 2015 (79 weeks) with 76.5%-93.0% overall accuracy. The model with a transmission kernel of <20 km (P20) achieved a maximum overall accuracy (93.0%). Within each neighbourhood, the accuracy of P20 varied between 32.4% and 88.1%; 13/20 neighbourhoods had a reasonable temporal agreement between the simulated and actual dynamics of SRS (within 5th-95th percentiles), but 5/20 neighbourhoods underestimated and 2/20 overestimated the SRS spread. The model could be used for evaluation of semi-global control policies in Region 10, while addition of other factors such as seasonality, ocean currents, and movement of infected fish may improve the model performance at a finer scale.
本研究旨在估计智利鲑鱼立克次体败血症（SRS）在农场间传播的参数，并使用 InterSpread Plus 开发和验证模拟模型，以预测拉戈斯（第 10 区）农场间 SRS 的每周传播情况。模型参数通过对历史 SRS 暴发数据的分析进行估计。这些模型包含时间和距离相关的传播核，代表了 SRS 在农场之间通过水传播的概率。估计了七个候选传播核，它们之间的最大农场间 SRS 传播距离不同（15-60 公里）。农场按规模（小、中、大）和物种（银鲑、大西洋鲑、虹鳟）进行分类。农场从感染中恢复所需的时间参数最短的是小银鲑农场（中位数：7 周），其次是中大和大银鲑农场（中位数：25 周）、大西洋鲑农场（中位数：42 周，任何规模）和虹鳟农场（中位数：43 周，任何规模）。虹鳟农场的相对传染性参数是银鲑或大西洋鲑的 1.5-6.3 倍，或者大型农场的 1.3-4.2 倍。这些模型预测了 2013 年至 2015 年第 10 区的 SRS 流行率（79 周），总体准确率为 76.5%-93.0%。传播核小于 20 公里（P20）的模型达到了最大的总体准确率（93.0%）。在每个邻里内，P20 的准确率在 32.4%至 88.1%之间变化；20 个邻里中有 13 个邻里的 SRS 模拟和实际动态之间存在合理的时间一致性（在第 5 到 95 百分位数内），但 5 个邻里低估了 SRS 的传播，2 个邻里高估了 SRS 的传播。该模型可用于评估第 10 区的半全球控制政策，而添加季节性、洋流和感染鱼类的移动等其他因素可能会提高更精细尺度的模型性能。
