27 Bovine brucellosis, caused by Brucella abortus, is a highly infectious and zoonotic disease responsible for economic losses in livestock worldwide. While it is largely eradicated in many developed countries, it is endemic in most parts of the world, including India. The serological evidence of infection has been reported from various states of the country (Trangadia et al. 2010, Trangadia et al. 2012, Jagapur et al. 2013, Singh et al. 2014). Isolation and identification of Brucella spp. is performed as the gold standard for diagnosis of brucellosis (OIE 2008). In the absence of the bacterial isolation, other methods such as serological test or combination of tests with known sensitivity and specificity estimates can be used to define the status of animals. There are several serological techniques available, the most common being Rose Bengal test (RBT), complement fixation test (CFT), fluorescent polarization assay (FPA) and indirect and competitive enzyme-linked immunosorbent assay (ELISA). Among these tests, ELISA offers several advantages such as ease, Indian Journal of Animal Sciences 85 (1): 27–31, January 2015/Article Evaluation of ELISA kits for brucellosis in naturally infected Indian cattle population by latent class analysis B J TRANGADIA 1 , N TOFT 2 , K NAGAMANI 3 , S K RANA 4 and V A SRINIVASAN 5 National Dairy Development Board, Gachibowli, Hyderabad, Andhra Pradesh 500 032 India Received: 20 June 2014; Accepted: 1 October 2014 Key words: Brucellosis, Competitive ELISA, Indirect ABSTRACT Brucellosis is endemic in India and the disease is confirmed by various laboratory tests. Among the tests available, ELISA offers several advantages such as ease, cost effectiveness and possibilities of automation for screening large number of serum samples at any given period of time. This study was performed to evaluate 5 commercially available indirect ELISA kits and 1 competitive ELISA kit in naturally infected cattle population in India. Six organized cattle herds located in Andhra Pradesh, Gujarat and Odisha were selected. Sensitivity (Se) and specificity (Sp) of the tests were estimated in absence of gold standard using latent class analysis. Nearly perfect Sp was achieved by the CEDI test 99.8% (95%; credibility posterior interval (CPI) = 98.9-100%), Chekit test 99.4% (95%; CPI = 98.3-99.9%) and Svanova test 99.2% (95%; CPI = 99.2-99.9%). Se of the 5 indirect ELISA was ranged between 75.0 and 77.3%. However, a Se of 56.1% (95%; CPI= 36.1-75.1%) and Sp of 91.7% (95%; CPI = 88.7- 94.2%) were recorded for cELISA. Key words: Brucellosis, Competitive ELISA, Indirect ELISA Present address: 1 Scientist-I, 3 Executive 4 Sr. Scientist R&D Centre, (drbjt77@gmail.com, nagamani@indimmune.com, ranask_1960@yahoo.co.in,). 5 Advisor (srinivasanva1948 @gmail.com), National Dairy Development Board, Gachibowli, Hyderabad. 2 Professor (ntoft@vet.dtu.dk), National Veterinary Institute, Technical University of Denmark, 1870 Frederiksberg C, Denmark. cost effectiveness and possibilities of automation for screening large number of serum samples at any given period of time. Usually the performance of a diagnostic test is evaluated by the sensitivity (the proportion of true positives among infected) and the specificity (the proportion of true negatives among non-infected). Thus, to evaluate the diagnostic performance of several ELISAs for brucellosis in naturally infected cattle it is preferable to avoid the assumption of a perfect reference test. Muma et al. (2007) evaluated the performance of serological tests for brucellosis using latent class analysis (LCA) (Hui and Walter 1980) which under certain assumptions permit evaluation of performance of the diagnostic test in the absence of a gold standard. Therefore, the objective of this study is to evaluate the diagnostic performance of commercially available 5 indirect and 1 competitive ELISA for diagnosis of brucellosis in naturally infected cattle in absence of ‘gold standard’. MATERIALS AND METHODS Animals and serum samples: Blood samples (404) were randomly collected from indigenous and crossbred cattle housed in 6 organized farms located in Gujarat (3 farms), Andhra Pradesh (2 farms) and Odisha (1 farm) during 2007- 08. These animals were not vaccinated against brucellosis. Approximately 5 ml blood was collected from the jugular vein and allowed to clot. At the laboratory, the samples