Abstract:  Objective  To apply moving percentile method on surveillance and early-warning on visceral leishmaniasis in Kashgar Region and evaluate its effect.  Methods  Incidence data of visceral leishmaniasis in Kashgar Region were collected from the National Web-based Infectious Diseases Report System. Monthly detection was carried out by using moving percentile method. The 50th percentile（P50）, 70th percentile（P70） and 90th percentile（P90） of historical baseline data were calculated for drawing a control chart, and P70 was adopted as the warning threshold to determine whether an epidemic would appear. If the number of cases in one month is higher than the corresponding P70 of historical baseline data, the warning signal will be generated. The sensitivity, specificity and positive predictive value were calculated for the evaluation of early-warning effect.  Results  During the study period, 61.0% cases were reported in the year of 2008 and 2009, the incidence peak was from September to December, accounting for 51.9%, and infants under 3 years old were the population most threatened by visceral leishmaniasis, accounting for 62.7%. A total of 58 detections were performed, and 17 warning signals were generated by the threshold on P70. Among them, the numbers of cases in 9 detections were higher than the corresponding P90 of historical-baseline data. Based on the actually epidemic status of visceral leishmaniasis in study period, according to the threshold on P70, a total of 10 warning signals of 11 epidemics were detected, and the sensitivity of the warning model was 90.9% （10/11）. 7 wrongly signals of 47 non-epidemics were detected, and the specificity was 85.1%（40/47）. 10 of 17 signals were proved to be correct, and the positive predictive value was 58.8%（10/17）.  Conclusion  The moving percentile method can effectively perform surveillance and early-warning on visceral leishmaniasis in Kashgar Region.

Key words: Moving percentile method, Visceral leishmaniasis, Kashgar Region, Surveillance, Warning