Development and validation of a hemangiosarcoma likelihood prediction model in dogs presenting with spontaneous hemoabdomen: The HeLP score

Objective To calculate a risk prediction model for hemangiosarcoma (HSA) diagnosis in dogs presenting with nontraumatic hemoabdomen. Design Retrospective multicenter observational cohort study enrolling dogs presented 2003-2016. Setting Five academic veterinary medical centers. Animals A total of 406 dogs with nontraumatic hemoabdomen as the presenting complaint that underwent surgical exploration or necropsy and received a histological diagnosis. Overall, 219 dogs from 3 centers provided the data for model construction, and 187 dogs from 2 centers provided the population for external validation. Interventions None. Measurements and Main Results The risk score was modeled on 4 predictors: bodyweight (P = 0.01), total plasma protein (P < 0.01), platelet count (P < 0.01), and thoracic radiograph findings (P = 0.02). The incidence of HSA diagnosis was 36%, 76%, and 96% in the low risk (<= 40), medium risk (41-55), and high risk (>55) score groups, respectively. The risk score AUROC was 0.85 (95% CI 0.79-0.90) on the construction population, and 0.77 (95% CI 0.70-0.84) on the validation population. Conclusions The risk of HSA diagnosis in dogs presenting with nontraumatic hemoabdomen could be predicted using a simple risk score, which could aid in identification and treatment of dogs at lower risk for this diagnosis.