Evaluación de tres modelos y factores de riesgo asociados a la mortalidad de lechones al nacimiento en el trópico de México

Abstract

Study objectives were to analyze stillborn piglet data using three logistic regression models and to determine the effect of some risk factors on piglet mortality at birth. Data were collected from January 2003 to December 2004 on a commercial, farrowto-finish farm in the State of Yucatan, Mexico. Response variables were: a) litters with zero or at least one stillborn piglet (nmc, n=2561 litters), and b) dead or live piglet (nm, n=27,108 piglets). The risk factors were: farrowing year, farrowing season (dry, rainy and north wind), parity number (1, 2-4, >4), litter size ( 12), litter with at least one mummified fetus (no, yes), and type of artificial insemination. Data were analyzed with standard logistic regression (SLR, nmc and nm variables) and random binomial logistic regression (RLR, nm variable) models. The RLR model fit best the data. Not using the appropriate model produced changes in risk factor significance levels. According to the RLR model, risk of mortality was greater for piglets born in the north wind season versus the dry and rainy seasons (P=0.023). Risk of piglet mortality was also higher in first parity sows and in sows with 5 or more parities (P=0.001). The risk of mortality was 1.41 times higher in litters with at least one mummified fetus compared to litters without mummified fetuses. Litters with six or less piglets had a higher risk of mortality.