【AASV】个体水平与窝水平上的流行率对PRRS的防控有何意义

仔猪个体水平上的PRRSV流行率与产房内窝水平上的流行率之间关系的线性表征

Onyekachukwu H. Osemeke¹兽医学博士; Eduardo de Freitas Costa²,兽医学博士;维尼修斯-韦德³兽医学学士，博士；Swaminathan Jayaraman¹医学学士，硕士；Gustavo S. Silva¹兽医学博士, MS, 博士; Daniel C. L. Linhares¹兽医学博士, MBA, 博士

1爱荷华州立大学兽医学院兽医诊断和生产动物医学专业，爱荷华州艾姆斯；2流行病学、生物信息学和动物模型部，瓦赫宁根生物兽医研究，莱利斯塔德，荷兰；3南里奥格兰德州联邦教育、科学和技术研究所，巴西，法鲁皮尔哈

简介

监测/监视仍然是猪繁殖与呼吸综合征病毒(Porcine reproductive and respiratory syndrome virus，PRRSV)控制和消灭方案的一个组成部分。监测/监视的取样通常以假设为指导，其中最重要的是假设疾病在被取样猪群中的流行率^[1-2]。随着最近美国繁育猪群PRRSV监测模式的转变，采样方式从基于动物个体的取样转换到整体取样^[3]，在这些综合样本中，估计流行率的单位从单只动物转移到提供样本的群体动物，例如，在样本为窝唾液（family oral fluids，FOF）的情况下，是一窝猪^[4]。据作者所知，没有任何已发表的研究文献描述了上述流行病类型的关系。因此，本研究的目的是描述产房中带毒仔猪的比例（viremic piglets in a farrowing room，PPV）、至少有一头PRRSV阳性仔猪的窝的比例（真实窝流行率：true litter prevalence，TLP）和通过FOF检测到的PRRSV阳性窝的比例（表观窝流行率：apparent litter prevalence，ALP）之间的关系。

材料和方法

参考研究的参数

基于Almeida等的研究^[5]，首先建立了一个预测模型，以描述窝内流行率（一窝猪中感染病毒的仔猪的比例）与该窝仔猪中FOF阳性样本的概率之间的关系。对参考研究中取样窝内PRRSV阳性猪的聚类程度（异质性）进行缩放和测量，并将中值（0.61）作为基线。本研究使用的聚类因子范围从0（意味着PRRSV阳性猪在各窝之间完全随机分布）到1（指PRRSV阳性猪会被聚类在尽可能少的窝内）。还得到了参考研究中所有取样的猪群中，被取样哺乳栏的窝规模的经验分布。

随机模型

产房模拟了有固定数量的窝（与产房中的限位栏数相对应），并使用仔猪水平的PRRSV流行率（PPV），范围从1%到50%。每窝仔猪的数量从参考研究中得到的离散经验分布中抽取。每窝中感染病毒的仔猪数从递归二项式模型中抽出，基线聚类因子为0.61。

每个迭代哺乳栏（iterated room）的真实窝水平流行率（TLP）是指至少有一头猪感染病毒的窝的比例，而表观窝水平流行率（ALP）是指迭代哺乳栏中通过FOF检测为PRRSV阳性的窝的预测比例。通过蒙特卡洛模拟（Monte Carlo simulatio）共获得5000个迭代哺乳栏，并获得了TLP和ALP的中位值。所有数据的分析都是利用R统计软件完成的^[6]。

结果

表1列出了56个哺乳栏的仔猪水平流行率和估测的窝水平流行率的匹配值。

讨论和结论

本研究显示了如何比较一个产房中的仔猪水平（PPV）、真实窝水平（TLP）和表观窝水平（ALP；通过FOF）的流行率，为养猪业的从业者提供了关于PRRSV流行率值的见解，可用于估计血清或FOF采样的可比样本量，以监测断奶期仔猪的PRRSV流行性。例如，PPV为5%，ALP为6.48%（约4个哺乳栏），对于一个有56个哺乳栏的产房，使用传统的样本量计算器，这意味着要通过FOF的RT-rtPCR检测，有≥95%的把握检测到至少一窝有阳性仔猪。

本研究的结果还提供了一个框架，根据FOF的检测结果（检测阳性率），估计产房内病毒感染仔猪的比例。

本研究对以前的一系列研究进行了补充，对传统的样本量概念进行了调整，以更好地适应美国传统猪舍的特殊性和PRRSV的生态学。

参考文献：略

In-silico characterization of the relationship between PRRSV prevalence at the individual piglet level and prevalence at the litter level in a farrowing room

Onyekachukwu H. Osemeke1, DVM, MS; Eduardo de Freitas Costa2, DVM, PhD; Vinicius Weide3, BS, PhD; Swaminathan Jayaraman1, BE, MS; Gustavo S. Silva1, DVM, MS, PhD; Daniel C. L. Linhares1, DVM, MBA, PhD

1Veterinary Diagnostic and Production Animal Medicine, College of Veterinary Medicine, Iowa State University, Ames, Iowa;2Department of Epidemiology, Bioinformatics and Animal Models, Wageningen Bioveterinary Research, Lelystad, The Netherlands;3Instituto Federal de Educação, Ciência e Tecnologia do Rio Grande do Sul, Farroupilha, RS, Brazil

Introduction

Monitoring/surveillance remains an integral component of PRRSV control and elimination programs. Sampling for monitoring/surveillance is generally guided by assumptions, paramount of which is the assumed prevalence of the disease in the population to be sampled.1,2With the recent paradigm shift in US breeding herds’ PRRSV surveillance, from individual-animal-based sampling to aggregate sampling,3the unit for which prevalence is estimated for these aggregate samples shifts from the individual animal to the animal group giving the sample, for example, the litter, in the case of family oral fluids (FOF).4To the best of the authors’ knowledge, no published work describes how the mentioned prevalence types relate. Therefore, the objective of this study was to characterize the relationship between the proportion of viremic piglets in a farrowing room (PPV), the proportion of litters with at least one PRRSV-positive piglet (true litter prevalence, TLP), and the proportion of PRRSV-positive litters likely to be detected by FOF (apparent litter prevalence, ALP).

Materials and methods

Parameters from the referenced study

Based on the study of Almeida et al[5],a predictive model was first built to characterize the relationship between within-litter prevalence (the proportion of viremic piglets within a litter) and the probability of a positive FOF sample from that litter. The degrees of clustering (heterogeneity) of PRRSV-positive pigs within sampled rooms from the reference study were scaled and measured, and the median value (0.61) was used as a baseline. The clustering factor used for this study ranged from 0 (meaning a completely random distribution of PRRSV-positive pigs between litters) to 1 (meaning PRRSV-positive pigs would be clustered within the fewest number of litters possible)). An empirical distribution of litter sizes from the sampled rooms across all the herds sampled in the reference study was also obtained.

Stochastic model

Farrowing rooms were simulated with a fixed number of litters (corresponding to the number of crates in the farrowing room) and using piglet-level PRRSV prevalence (PPV) ranging from 1 to 50%. The number of piglets within each litter was drawn from a discrete empirical distribution obtained from the referenced study. The number of viremic piglets within each litter was sampled from a recursive binomial model using a baseline clustering factor of 0.61.

The TLP per iterated room was obtained as the proportion of lit­ters with at least one viremic pig, and the apparent litter prevalence was obtained as the predicted proportion of the litters in an iterated room that will be PRRSV-positive by FOF testing. A total of 5000 iterated rooms were obtained by Monte Carlo simulation, and the median values of TLP and ALP were obtained. All analyses were done on R statistical software.6

Results

Table 1 presents matched values of piglet-level prevalence and es­timated litter-level prevalence for a 56-crate room.

Table 1: Relationship between the piglet-level PRRSV prevalence in a 56-crate farrowing room and the true and apparent (by family oral fluids [FOF]) litter-level prevalence assuming a clustering level of 0.61

Discussion and conclusion

The results of this study demonstrate how PPV, TLP, and ALP (by FOF) in a farrowing room compare. This study provides insight to swine practitioners on prevalence values to be used in estimating comparable sample sizes for either serum or FOF sampling for PRRSV surveillance in weaning-age pigs. For example, a PPV of 5% matches with an ALP of 6.48% (~4 crates), For a 56-crate room, using a conventional sample size calculator, this would mean sampling about 29 crates to have ≥ 95% confidence of detecting at least one positive litter by RT-rtPCR tests on FOF.

The results of this study also provide a framework for estimating the proportion of viremic piglets within a farrowing room, given the results of FOF testing (test positivity rate).

This study adds to the series of previous studies tailoring conven­tional sample size concepts to better fit peculiarities in typical US swine barns and the ecology of PRRSV.

References

1. Stevenson MA. Sample Size Estimation in Veterinary Epidemiologic Re­search. Front Vet Sci. 2021;7:539573.doi:10.3389/fvets.2020.539573

2. Fosgate GT. Practical sample size calculations for surveillance and diagnostic investigations. J Vet Diagnostic Investig. 2009;21(1):3-14. doi:10.1177/104063870902100102

3. Trevisan G, Linhares LCM, Crim B, et al. Macroepidemiological as­pects of porcine reproductive and respiratory syndrome virus detection by major United States veterinary diagnostic laboratories over time, age group, and specimen. Shaman J, ed. PLoS One. 2019;14(10):e0223544. doi:10.1371/journal.pone.0223544

4. Osemeke OH, de Freitas Costa E, Almeida MN, et al. Effect of pooling family oral fluids on the probability of PRRSV RNA detection by RT-rtP­CR. Prev Vet Med. 2022;206:105701.doi:10.1016/j.prevetmed.2022.105701

5. Almeida MN, Zhang M, Zimmerman JJ, Holtkamp DJ, Linhares DCL. Finding PRRSV in sow herds: Family oral fluids vs. serum samples from due-to-wean pigs. Prev Vet Med. 2021; 193: 105397. doi:10.1016/j. prevetmed.2021.105397

6. R Core Team. R: A Language and Environment for Statistical Comput­ing: R Foundation for Statistical Computing Vienna, Austria. 2019.