Early identification of strangulating obstruction (SO) in horses with colic improves outcomes, yet early diagnosis of horses requiring surgery for SO often remains challenging. To compare blood and peritoneal fluid l-lactate concentrations, peritoneal:blood l-lactate ratio, peritoneal minus blood (peritoneal-blood) l-lactate concentration and other clinical variables for predicting SO and SO in horses with small intestinal lesions (SO-SI) and then to develop a multivariable model to predict SO and SO-SI. Retrospective cohort. A total of 197 equids admitted to a referral institution for colic between 2016 and 2019 that had peritoneal fluid analysis performed at admission were included. Twenty-three admission variables were evaluated individually for the prediction of a SO or SO-SI and then using multivariable logistic regression. Odds ratios (ORs) with 95% confidence intervals (CI) and area under the curve of the receiver operator characteristic (AUC ROC) were calculated. All variables performed better in the model than individually. The final multivariable model for predicting SO included marked abdominal pain (OR 5.31, CI 1.40-20.18), rectal temperature (OR 0.30, CI 0.14-0.64), serosanguineous peritoneal fluid (OR 35.34, CI 10.10-122.94), peritoneal-blood l-lactate (OR 1.77, CI 1.25-2.51), and peritoneal:blood l-lactate ratio (OR 0.36, CI 0.18-0.72). The AUC ROC was 0.91. The final multivariable model for predicting SO-SI included reflux volume (OR 0.69, CI 0.56-0.86), blood l-lactate concentration (OR 0.43, CI 0.22-0.87), serosanguineous peritoneal fluid (OR 4.99, CI 1.26-19.74), and peritoneal l-lactate concentration (OR 3.77, CI 1.82-7.81). Retrospective, single-hospital study design. Blood and peritoneal fluid l-lactate concentrations should be interpreted in conjunction with other clinical variables. The relationship between peritoneal and blood l-lactate concentration for predicting SO or SO-SI was complex when included in a multivariable model. Models to predict SO probably vary based on lesion location.
Read full abstract