Evaluation of a contract breeding management program in selected Ohio dairy herds with event-time analysis I. Cox proportional hazards models.
An observational study was conducted in order to assess the impact of a contract breeding program on the reproductive performance in a selected group of Ohio dairies using event-time analysis. The contract breeding program was offered by a breeding co-operative and featured tail chalking and daily evaluation of cows for insemination by co-operative technicians. Dairy employees no longer handled estrus detection activities. Between early 2002 and mid-2004, test-day records related to production and reproduction were obtained for 16,453 lactations representing 11,398 cows in a non-random sample of 31 dairies identified as well-managed client herds of the breeding co-operative. Of the 31 herds, 15 were using the contract breeding at the start of the data acquisition period, having started in the previous 2 years. The remaining 16 herds managed their own breeding program and used the co-operative for semen purchase. Cox proportional hazards modeling techniques were used to estimate the association of the contract breeding, as well as the effect of other significant predictors, with the hazard of pregnancy. Two separate Cox models were developed and compared: one that only considered fixed covariates and a second that included both fixed and time-varying covariates. Estimates of effects were expressed as the hazard ratio (HR) for pregnancy. Results of the fixed covariates model indicated that, controlling for breed, herd size, use of ovulation synchronization protocols in the herd, whether somatic cell score exceeded 4.5 prior to pregnancy or censoring, parity, calving season, and maximum test-day milk prior to pregnancy or censoring, the contract breeding program was associated with an increased hazard of pregnancy (HR=1.315; 95% CI 1.261-1.371). The results of the time-varying covariates model, which controlled for breed, herd size, use of ovulation synchronization protocols, somatic cell score above 4.5, parity, calving season, and testing season also found that the program was associated with an increased hazard of pregnancy (HR=1.387; 95% CI 1.327-1.451). The fixed and time-varying covariates models both found similar sets of predictors when analyzing the association of the contract breeding program with hazard of pregnancy. Both models identified a 30% or greater increase in hazard of pregnancy associated with use of the contract breeding program, suggesting that herds subscribing to the program achieved pregnancies in a more timely fashion.