Approaches to canine health surveillance.

Effective canine health surveillance systems can be used to monitor disease in the general population, prioritise disorders for strategic control and focus clinical research, and to evaluate the success of these measures. The key attributes for optimal data collection systems that support canine disease surveillance are representativeness of the general population, validity of disorder data and sustainability. Limitations in these areas present as selection bias, misclassification bias and discontinuation of the system respectively. Canine health data sources are reviewed to identify their strengths and weaknesses for supporting effective canine health surveillance. Insurance data benefit from large and well-defined denominator populations but are limited by selection bias relating to the clinical events claimed and animals covered. Veterinary referral clinical data offer good reliability for diagnoses but are limited by referral bias for the disorders and animals included. Primary-care practice data have the advantage of excellent representation of the general dog population and recording at the point of care by veterinary professionals but may encounter misclassification problems and technical difficulties related to management and analysis of large datasets. Questionnaire surveys offer speed and low cost but may suffer from low response rates, poor data validation, recall bias and ill-defined denominator population information. Canine health scheme data benefit from well-characterised disorder and animal data but reflect selection bias during the voluntary submissions process. Formal UK passive surveillance systems are limited by chronic under-reporting and selection bias. It is concluded that active collection systems using secondary health data provide the optimal resource for canine health surveillance.