Abstract
BACKGROUND: Leptospirosis is one of the most important zoonosis in the tropics. Currently, specific laboratory diagnostic test for leptospirosis such as polymerase chain reaction (PCR) or direct culture cannot be applied at the primary care setting especially in the resource- limited countries. Therefore, clinical presentation and laboratory examination are still the primary diagnostic tools for leptospirosis. OBJECTIVES: To detect clinical factors for predicting leptospirosis in suspected cases, and to create a clinical prediction score (THAI-LEPTO) that is practical and easy to use in general practice while awaiting laboratory results. MATERIALS AND METHODS: We performed a prospective multicenter study with a development and a validation cohort of patients presenting with clinical suspicion of leptospirosis as per the WHO clinical criteria. The development cohort was conducted at 11 centers in 8 provinces around Thailand. The validation cohort was conducted at 4 centers in 1 province from the Northeastern part of Thailand. Leptospirosis confirmed cases were defined if any one of the tests were positive: microscopic agglutination test, direct culture, or PCR technique. Multivariable logistic regression was used to identify predictors of leptospirosis. The clinical prediction score was derived from the regression coefficients (original) or from the odds ratio values (simplified). We used receiver operating characteristic (ROC) curve analysis to evaluate the diagnostic ability of our score and to find the optimal cutoff values of the score. We used a validation cohort to evaluate the accuracy of our methods. RESULTS: In the development cohort, we enrolled 221 leptospirosis suspected cases and analyzed 211. Among those, 105 (50%) were leptospirosis confirmed cases. In logistic regression adjusted for age, gender, day of fever, and one clinical factor at a time, leptospirosis group had more hypotension OR = 2.76 (95% CI 1.07-7.10), jaundice OR = 3.40 (95%CI 1.48-8.44), muscle pain OR = 2.12 (95%CI 1.06-4.26), acute kidney injury (AKI) OR = 2.90 (95%CI 1.31-6.15), low hemoglobin OR = 3.48 (95%CI 1.72-7.04), and hypokalemia with hyponatremia OR = 3.56 (95%CI 1.17-10.84) than non-leptospirosis group. The abovementioned factors along with neutrophilia and pulmonary opacity were used in the development of the score. The simplified score with 7 variables was the summation of the odds ratio values as follows: hypotension 3, jaundice 2, muscle pain 2, AKI 1.5, low hemoglobin 3, hypokalemia with hyponatremia 3, and neutrophilia 1. The score showed the highest discriminatory power with area under the curve (AUC) 0.82 (95%CI 0.67-0.97) on fever day 3-4. In the validation cohort we enrolled 96 leptospirosis suspected cases and analyzed 92. Of those, 69 (75%) were leptospirosis confirmed cases. The performance of the simplified score with 7 variables at a cutoff of 4 was AUC 0.78 (95%CI 0.68-0.89); sensitivity 73.5; specificity 73.7; positive predictive value 87.8; negative predictive value 58.3. CONCLUSIONS: THAI-LEPTO score is a newly developed diagnostic tool for early presumptive diagnosis of leptospirosis in patients presenting with severe clinical suspicion of the disease. The score can easily be applied at the point of care while awaiting confirmatory laboratory results. Each predictor used has been supported by evidence of clinical and pathophysiological correlation.