Statistisk analys av journalmaterial från två stuterier : en retrospektiv studie

Abstract: Retrospective data from two stud farms, in this study named stud farm A and B, including 742 mares, was statistically analysed. Parameters included in the analysis were type of insemination, date of the first insemination, number of inseminations, the age of the mares, if the mares had a foal or not, foaling date, result of pregnancy examination (if it was done), twin pregnancy and treatments given to the mares. The pregnancy results and the treatments where compiled into a number of frequency tables in which the mares where arranged according to age, month of first insemination, foal or not and type of insemination. There was a significant difference in pregnancy result between the years at both stud farms (2001 better than 2002). At stud farm A, in year 2001, month (at start of insemination) significantly influenced the pregnancy result. This was not the result in the year 2002 or at stud farm B. Frozen sperm was only used at stud farm A. The total number of insemination per mare was significant lower for mares inseminated with frozen sperm. A significant higher frequency of mares at stud farm A were treated with PG at the end of the season. HCG was used more frequently during the year 2002 compared with 2001. The age of the mares did not influence the pregnancy result probably due to a selection of mares for treatment. Both stud farms used antibiotics and Caslick operations to a small extent. Furthermore, Standardbred mares (trotters) that got pregnant in June/July had on average 5 days shorter period of gestation compared with mares getting pregnant in May. A statistic logistic regression analysis including all the parameters in this study, showed how the different parameters influenced the chance of a mare to get pregnant. As an example, the difference between the stud farms concerning pregnancy result disappeared. Conclusion: Statistic analysis of retrospective data from two stud farms showed that simple frequency tables are not sufficient to make an adequate comparison, a deeper statistic analysis that take into account all available data is needed.