Effect of pathogen-specific clinical mastitis on herd life in two New York State dairy herds Y.T. Gro ¨hn a, * , R.N. Gonza ´lez b , D.J. Wilson b , J.A. Hertl a , G. Bennett b , H. Schulte b , Y.H. Schukken b a Section of Epidemiology, Department of Population Medicine and Diagnostic Sciences, College of Veterinary Medicine, Cornell University, Ithaca, NY 14853, USA b Quality Milk Production Services, Cornell University, Ithaca, NY 14850, USA Received 2 September 2004; received in revised form 6 June 2005; accepted 9 June 2005 Abstract The objective of this study was to estimate the effects of clinical mastitis (CM) (both with and without specific pathogen identification) occurring in different stages of lactation on length of herd life in two New York State dairy farms. The 2697 cows in the study were followed for one lactation (the first-occurring one on or after 1 October 1999), until it ended because of a new lactation, culling, or end of study (31 March 2001 in one farm; 31 July 2001 in the other). A Cox proportional hazards model with time-dependent covariates, in SAS 1 , was used to measure, within a lactation, the effect of the first occurrence of CM (without specific pathogen identification) occurring 1–7, 8–66, 67–100, 101–225, or 226 days in milk (DIM), on how long cows remained in the herd. For the first occurrence of CM due to Streptococcus spp., Staphylococcus aureus, Staphylococcus spp., Escher- ichia coli, Klebsiella spp., and ‘no pathogen isolated’, the intervals were before and after the median DIM of first occurrence of each pathogen. There were too few cases due to Arcanobacterium pyogenes, and ‘other pathogens grouped together’ to split into intervals, so they were modeled as binary variables, i.e. as they occurred. CM was modeled using time-dependent covariates, to account for its differing effects throughout lactation on culling. Other variables controlled for were herd, parity, calving season, and other significant diseases. In the dataset, the lactational incidence risk of the first occurrence of CM was 18.2%; 20.0% of the cows did not survive the lactation that was studied. The overall annual culling percentage for both herds during the study period (including all cows, whether eligible for the study or not) was 35.6%. For cows with CM without pathogen identification, their highest hazard ratio (HR) of culling occurred from 67 to 100 DIM. All of the www.elsevier.com/locate/prevetmed Preventive Veterinary Medicine 71 (2005) 105–125 * Corresponding author. Tel.: +1 607 253 4394; fax: +1 607 253 3440. E-mail address: ytg1@cornell.edu (Y.T. Gro ¨hn). 0167-5877/$ – see front matter # 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.prevetmed.2005.06.002