Fig. 2

Results of cross-correlation function analysis for time series between the monthly incidence of brucellosis in human and cattle herd. This figure shows the cross-correlation function test for the residuals of the first-order-differenced data of brucellosis in human and cattle herd during the period of (a) Phase 1 (January 2004 to March 2005), (b) Phase 2 (June 2005 to June 2006), (c) Phase 3 (July 2006 to June 2007), and (d) Phase 4 (July 2007 to December 2014). The bars that cross the dashed horizontal lines (95% confidence interval) indicate considerable correlation between two-time series. a There was no significant cross-correlation during the initial stage of surveillance on beef cattle. b Significant cross-correlation without lag was observed after extending the surveillance to bulls prior to trade and to beef slaughter houses (β = 0.57, p = 0.04). c Significant positive cross-correlation was observed with a 1-month-lag after extending the surveillance to all beef farms with 10 or more heads (β = 0.65, p = 0.03). d Significant cross correlation with various month-lags is observed after adding the polymerase chain reaction diagnostic method on the definition of human case (a positive 3-month-lag: β = 0.35, p = 0.01; 0-month-lag: β = 0.24, p = 0.03; a negative 1-month-lag: β = 0.23, p = 0.03; negative 7-month-lag (β = − 0.30, p = 0.02; negative 8-month-lag: β = 0.23, p = 0.03). ^†Autocorrelation function (ACF)