Clinical predictive models in equine medicine: A systematic review

Area of Science:

• Veterinary Medicine
• Clinical Epidemiology

Background:

• Clinical predictive models offer potential for aiding equine clinical decision-making by forecasting patient outcomes using baseline data.
• The current landscape and quality of equine-specific clinical predictive models remain largely uncharacterized in scientific literature.

Approach:

• A systematic literature review was conducted using PubMed and Google Scholar to identify multivariable predictive models for equine patients.
• Included models were peer-reviewed, predicting clinical, laboratory, or imaging outcomes in individual horses or herds.
• Models were evaluated for patient populations, development methods, performance metrics, validation, risk of bias (using PROBAST), and applicability.

Key Points:

• Ninety predictive models and nine external validation studies were identified.
• Forty-one percent of models focused on colic-related outcomes, such as surgical necessity or survival.
• All reviewed models exhibited a high risk of bias, mainly due to analysis-related issues, necessitating cautious clinical application.

Conclusions:

• While all identified equine predictive models carry a high risk of bias, they are not necessarily unusable and warrant careful evaluation.
• Applicability concerns were generally low across the majority of models.
• This review aims to enhance veterinarian awareness of available predictive models and their performance evaluation in diverse equine populations.