Prediction of foaling using mammary secretion constituents

AIMS: To test published models for predicting time of foaling using data derived from prepartum mammary secretions and to develop a new model based on concentrations of mammary secretion constituents that accurately predicts time of foaling in Thoroughbred mares. METHODS: Concentrations of sodium, potassium, calcium, citrate and lactose were measured in prepartum mammary secretions of 20 Thoroughbred mares in the 2 weeks before expected date of foaling. Models to predict time of foaling were fitted to data based on absolute concentration, change in concentration, and percentage change in concentration of mammary secretion constituents in relation to actual intervals to foaling. RESULTS: Concentrations of potassium, calcium, citrate and lactose increased, and concentration of sodium decreased as foaling approached but variation between mares was large. Models to predict time of foaling based on percentage change in electrolyte concentrations were less accurate than those based on absolute concentration and change in concentration. When data from this study were fitted to two previously published models, the statistical sensitivity, specificity, and positive and negative predictive values of both models were lower than those originally reported. CONCLUSIONS: The use of prepartum equine mammary secretion electrolyte concentrations for prediction of time of foaling is unreliable, due to large variation in both absolute concentrations and change in concentrations between mares. Models that use a combination of mammary secretion electrolytes and physical and behavioural factors may better predict foaling than those based on mammary secretions alone.