Abstract
Domestic rats are the principal reservoir for urban leptospirosis. However, few studies have identified infestation markers in slums and evaluated their predictivity for leptospirosis risk. We compared households with leptospirosis cases in Salvador, Brazil between 2007 and 2009 and their neighbors using a case control design, surveying for rodent infestation signs and environmental characteristics. With the 2007-2008 data, a conditional logistic regression modeling identified the peridomiciliar presence of rodent burrows (OR, 3.30; 95% CI, 1.50-7.26), rat feces (2.86; 1.24-6.59), runs (2.57; 1.06-6.22), households bordering abandoned houses (2.48; 1.04-6.02), and unplastered walls (2.22; 1.02-6.02) as risk factors and developed a predictive score for leptospirosis. With an independent data set from 2009, a receiver operating characteristic (ROC) curve analysis evaluated the prediction score performance, with the area under the curve being 0.70 (95% CI, 0.64-0.76) for score development and 0.71 (0.65-0.79) for validation. Results indicate that high proportions of urban slum households are infested with R. norvegicus. The score performed well when identifying high-risk households within slums. These findings need confirmation in other urban centers, but suggest that community-based screening for rodent infestation can allow to target rodent and environmental control measures in populations at highest risk for leptospirosis.