Methodological appraisal of studies included in the review
| Article | Selection | Measurement | Confounding | Generalisability to entire dog population |
|---|---|---|---|---|
| Messam et al31 USA/Jamaica | Selected from veterinary population, possibly selecting dogs that are better managed in the population | Bite data: owner recallDesexing data: owner recall | Multifactorial analysis controlling for measured confounders | Dog bites from population going to veterinary clinics |
| Messam et al32 USA/Jamaica | Selected from veterinary population, possibly selecting dogs that are better managed in the population | Bite data: owner recallDesexing data: owner recall | Multifactorial analysis controlling for measured confounders | Dog bites from population going to veterinary clinics |
| Gershman et al30 USA | 54.9% of eligible cases included, cases from notified dog bite report | Bite data: objective from medical records Control data: owner recall Desexing data: owner recall | Multifactorial analysis controlling for measured confounders | Severe dog bites towards non-household members |
| Shuler et al33 USA | Cases from notified dog report Missing data for non-licensed dogs, possible bias away from the null | Bite data: 19.8% of attacks on a family member (expected underreporting of bites)Desexing data: from licensed database and from bite report | Univariate analysis reported for desexing effect estimate | Dog bites from entire population |
| Guy et al28 Canada | Selected from veterinary population, possibly selecting dogs that are better managed in the population. Control include dogs that may have bitten strangers, expected bias towards the null | Case-control classification: possible misclassification of exposure and outcome from owner recall bias | Multifactorial analysis controlling for measured confounders | Dog bites from population going to veterinary clinics and certain type of dog bites |
| Guy et al34 Canada | Selected from veterinary population, possibly selecting dogs that are better managed in the population Case-control include dogs that may have bitten strangers, expected bias towards the null | Bite classification: known misclassification of exposure and outcome as found in the 2001 study, likely introduced bias | Multifactorial analysis controlling for measured confounders | Dog bites from population going to veterinary clinics and certain type of dog bites |