Time series analysis of urethral obstruction in male cats in a veterinary teaching hospital in São paulo, Brazil.

Abstract

Time series analysis can be used to understand and forecast patterns in sequential data. This study evaluated three statistical models-ARIMA, Holt-Winters, and linear regression-on the time series of urethral obstruction (UO) cases in male cats treated at the Veterinary Teaching Hospital - São Paulo State University, Botucatu, Brazil. Among the 5,230 cats evaluated between 2010 and 2020, the prevalence of UO in male cats was 7.4% (95% CI: 6.7-8.1%), and the incidence among cats showing lower urinary tract signs was 36.0% (95% CI: 33.19-38.93%). Most affected cats were neutered (60.94%), with a mean body weight of 4.24 ± 1.11 kg and higher body condition scores. ARIMA closely followed historical data but was ineffective for future forecasting, showing a flat projection from 2021 to 2024 (rate: 0.64) despite past fluctuations. The Holt-Winters model projected a rise in UO cases, from 0.70 (95% CI: 0.43-0.97) in 2021 to 1.09 (95% CI: 0.38-1.79) in 2024, but its wide confidence intervals indicated potential overestimation. Meanwhile, linear regression revealed a significant annual increase of 2.6% in UO cases (p = 0.042), explaining 38% of the variance and offering a more accurate long-term forecast, and then, was considered the most suitable model, capturing trends without overestimating future rates. These findings support improved surveillance, clinical protocols, preventive strategies, and hospital resource planning for managing UO in male cats in a teaching veterinary hospital scenario.