A clinical prediction rule for American cutaneous leishmaniasis in Colombia.

Neither parasitological nor molecular diagnosis of leishmaniasis is widely available in clinical settings where American cutaneous leishmaniasis (ACL) is endemic. Therefore four clinical prediction rules for ACL were developed which incorporated physical examination findings (clinical rule), physical examination and leishmanin skin test (LST) (clinical-LST rule), physical examination and historical information (clinical-historical rule), or physical examination, historical information and LST (clinical-historical-LST rule). One hundred parasitologically diagnosed ACL cases and 38 cases of chronic skin lesions of other aetiologies comprised the derivation set. The validation set consisted of 124 ACL cases and 35 patients with lesions of other aetiologies. Components of each rule were selected by bivariate analysis, then step-wise logistic regression. Sensitivity, specificity and efficiency were calculated for each score threshold; the threshold achieving greatest efficiency was selected for each rule. When these rules were applied to the validity set the sensitivity, specificity and efficiency were respectively: clinical 93%, 31%, 79%; clinical-LST 90%, 73%, 85.9%; clinical-historical 97%, 51%, 87%; clinical-historical-LST 92%, 70%, 87%. Inclusion of LST skin test consistently improved the specificity of the rules. Should a given clinical setting warrant optimizing either sensitivity or specificity alone, the rule thresholds can be adjusted. These and other prediction rules, once evaluated in other settings, should be incorporated into leishmaniasis control programmes.