Table 3.

Advantages and limitations to data sources used for health surveillance in dogs

Data source	Advantages	Limitations	Applications
Pet insurance databases	Large size	Difficult to validate	Agria Pet Insurance data analysis in Sweden [15]
	Defined denominator	Questionable representativeness of the general population	Pet Protect insurance data analysis in UK [58, 69]
	High reliability for breed and sex	Loss of data on low-cost or excluded disorders
	Coded diagnoses
Referral practice clinical records	Good diagnostic reliability?	Referral bias	Veterinary Medical Data Base (VMDB) [75]
	Coded diagnoses?	Poorly defined denominator
	Large databases	Poorly representative
Primary-care practice clinical records	Large databases	Diagnostic reliability?	Banfield Pet Hospital [104]
	Highly representative?	Technical complexities	NCAS [5]
	Coded diagnoses	Only events with veterinary care	NCASP [116]
	Defined denominator		VetCompass [47]
	Generalisability		SAVSNET [129]
			CEVM [130]
Veterinary cancer registries	Human registries common	Referral bias	Veterinary Medical Data Base (VMDB) [75]
	Good diagnostic reliability	Poorly defined denominator	Danish Veterinary Cancer Registry [137].
		Poorly representative
Questionnaire-based data collection	Relatively inexpensive	Response rate	The KC/BSAVA UK health survey of purebred dogs [148].
	Flexible	Difficult to validate
	Can nest within other study designs	Loss of information on temporality
Canine health schemes	Large databases	Poorly representative	BVA/KC hip dysplasia and elbow dysplasia scheme [194]
	Diagnostic reliability	Selection bias	BVA/KC elbow dysplasia scheme [195]
	Linkage to KC pedigree data		The BVA/KC/ISDS eye scheme [172].
	Permanent animal identification
Other companion animal surveillance systems in the UK	Relatively inexpensive	Under-reporting	SARSS [176]
		Poorly defined denominator
		Selection bias	DACTARI [186]
		Poor generalisability	CICADA [189]