Model Selection for Examining the Association Between Dog Intake Quantity and Socioeconomic and Geographic Conditions

Introduction: Dog intake to government shelters reflects both community conditions and local animal management practices, yet predictors of intake in Japan have not been evaluated at a national scale. The objective of this study was to identify community-level socioeconomic and geographic predictors of dog intake to government shelters in Japan, stratified by intake type (owner-unknown vs. owner-relinquished) and administrative setting (urban vs. non-urban). Methods: Publicly available municipal and prefectural data were used to model dog intake counts for 2022. Separate generalised linear models were developed for owner-unknown and owner-relinquished intake in urban areas (designated and core cities; n = 82) and non-urban areas (n = 47). Socioeconomic-related explanatory variables tested included the proportions of low-income households, older adults’ households, owner-occupied households, and individuals with higher education levels. Geographically related explanatory variables included the proportion of vacant houses, cultivated land area, the highest temperature, and the lowest temperature. Results: Intake rates varied widely across regions. In urban areas, higher owner-unknown intake was associated with lower education level (incidence rate ratio [IRR] = 0.28, 85% confidence interval [CI]: 0.18–0.43) and higher cultivated land area (IRR = 1.34, 85% CI: 1.03–1.73). In non-urban areas, owner-unknown intake was higher in areas with more vacant housing (IRR = 14.64, 85% CI: 7.54–28.40), more cultivated land (IRR = 2.27, 85% CI: 1.58–3.25), and higher lowest temperature (IRR = 1.12, 85% CI: 1.04–1.20). Urban owner-relinquished intake was associated with owner-occupied housing (IRR = 1.64, 85% CI: 1.22–2.20). Non-urban owner-relinquished intake was higher in areas with a greater proportion of low-income households (IRR = 1.73, 85% CI: 1.03–2.89). Conclusion: Supported predictors differed by intake type and administrative setting, suggesting that intake prevention strategies may be more effective when tailored to local context and dominant intake pathway. The associations observed in this study can serve as a basis for informing future policy development and further research.