Gasqui Patrick, Coulon Jean-Baptiste, Pons Odile
Unité d'Epidémiologie Animale, INRA, 63122 Saint-Genès-Champanelle, France.
Vet Res. 2003 Jan-Feb;34(1):85-104. doi: 10.1051/vetres:2002058.
Clinical mastitis in dairy cows has for many years been the subject of numerous epidemiological surveys to determine the main risk factors. In most cases this data has been analysed using a standard Poisson model without taking into consideration possible dependence between consecutive pathological events. These analyses have brought to light a great many potential risk factors without making it possible to clarify a certain amount of confusion surrounding the effects. The extension of an individual within a lactation model, considering dependence between clinical cases of mastitis within lactation so as to take into account inter-lactation dependence (which has already been published) is presented in the form of mixed distributions within the same survival model framework. By introducing new parameters, infection rate at calving and the identification of a higher exogenous infection rate indoors than at pasture, it is possible to take into consideration what had previously appeared to be a lactation stage factor, a calving month factor or even part of a parity factor. By considering these two types of dependence within the same model, it appears to be possible to obtain a simpler model in terms of the factors to be taken into account, and one that is based on generally acknowledged and easily understandable biological considerations. Lastly, a possible way of extending the model is to consider the dry period before calving and this is presented. This would make it possible to envisage developing a complete model of the animal's lifetime in the not-too-distant future. It is still necessary, however, to determine the farming system factors in the general sense of the term, which specifically affect one or the other of the different model parameters, before one can draw conclusions as to the potential extension of this type of model. A national survey is currently being carried out on approximately 600 French breeding farms that will help meet this last objective.
多年来,奶牛临床型乳腺炎一直是众多流行病学调查的主题,旨在确定主要风险因素。在大多数情况下,这些数据是使用标准泊松模型进行分析的,没有考虑连续病理事件之间可能存在的相关性。这些分析揭示了许多潜在风险因素,但却无法厘清围绕这些因素影响的一些困惑。本文以同一生存模型框架内的混合分布形式,呈现了个体泌乳期模型的扩展情况,该模型考虑了泌乳期内乳腺炎临床病例之间的相关性,从而兼顾了不同泌乳期之间的相关性(已有相关发表)。通过引入新参数——产犊时的感染率,并确定室内外源性感染率高于牧场,就有可能考虑到此前看似是泌乳阶段因素、产犊月份因素甚至部分胎次因素的影响。通过在同一模型中考虑这两种相关性,似乎有可能得到一个在需考虑因素方面更简单的模型,且该模型基于普遍认可且易于理解的生物学考量。最后,本文还提出了一种扩展该模型的可能方法,即考虑产犊前的干奶期。这将使得在不久的将来有可能设想建立一个完整的动物一生模型。然而,在就此类模型的潜在扩展得出结论之前,仍有必要确定从广义上讲具体影响不同模型参数中某一个或另一个的养殖系统因素。目前正在对约600个法国养殖场进行全国性调查,这将有助于实现这一最终目标。