Background
Bovine Tuberculosis (BTB) is a widespread and endemic disease of cattle in Ethiopia posing a significant threat to public health. Regular surveillance by skin test, bacteriology and molecular methods is not feasible due to lack of resource. Thus, routine abattoir (RA) inspection will continue to play a key role for national surveillance. We evaluated efficiency of RA inspection for diagnosis of Mycobacterium bovis infection and discussed its public health implications in light of a high risk of human exposure.

Methods
The study was conducted in five abattoirs: Addis Ababa, Adama, Hawassa, Yabello and Melge-Wondo abattoirs. The efficiency of routine abattoir (RA) inspection was validated in comparison to detailed abattoir (DA) inspection, followed by culture and microscopy (CM) and region of difference (RD) deletion analysis. Diagnostic accuracies (with corresponding measures of statistical uncertainty) were determined by computing test property statistics (sensitivity and specificity) and likelihood estimations using web-based SISA diagnostic statistics software. Post-test probability of detecting TB infected carcasses was estimated using nomograms. Agreement between RA and DA inspections was measured using kappa statistics. The study was conducted and reported in accordance with standards for reporting of diagnostic accuracy (STARD) requirements. Both routine and detailed meat inspection protocols were performed on a subpopulation of 3322 cattle selected randomly from among 78,269 cattle slaughtered during the study period. Three hundred thirty seven carcasses identified through detailed meat inspection protocols were subjected to culture and microscopy; of the 337, a subset of 105 specimens for culture and microscopy were subjected to further molecular testing.

Results
There was a substantial agreement between RA and DA inspections in Addis Ababa (Kappa = 0.7) and Melge-Wondo abattoirs (Kappa = 0.67). In Adama, Hawassa and Yabello abattoirs, the agreement was however poor (Kappa ≤ 0.2). RA inspection was able to detect only 117 of the total 3322 carcasses inspected (3.5%). The sensitivity (Sn) and specificity (Sp) of RA inspection were 28.2% (95/337) [95%CI: 23.4-33.0] and 99.3% (2963/2985) [95%CI: 99.0-99.6], respectively, when DA inspection was considered as reference test. When culture and microscopy (CM) was considered as reference test, the Sn and Sp of RA were 55.2% (58/105) [95%CI: 45.7-64.7] and 84.1% (195/232) [95%CI: 79.3-88.8]. RA inspection failed to detect 71.8% (242/337) and 44.8% (47/105) of TB infected carcasses as judged by DA inspection and CM, respectively. On the other hand, a much higher sensitivity of DA was obtained when CM and RD deletion analysis were considered as reference tests (96.3% (105/109) and 100.0% (24/24), respectively).