Infectious Disease Early Warning Of Automatic Early Warning Information System Effect Factors

ObjectiveThis paper was conducted to study the influence factors of the automatic early-warning information system of infectious diseases.Research objects and methods1 Research objectsThe data was mainly obtained from China information system for disease control and prevention.We selected the counties where events of research diseases were reported in China in the report and management system of emergent public health events and regarded it as research counties,according to the standard of rejecting.2 MethodsWe selected bacillary dysentery and measles as research diseases.When the cases of research diseases reached the standards of information report and management of emergent public health events,we regarded the disease as an outbreak.If the automatic early-warning information system of infectious diseases can produce early-warning signal after the cases were reported and before the event was reported, we regarded it as a successful early-warning.We selected fixed period and mean incubation period as the number of days for observed periods(W),and 1 or 2 periods as the contemporaneous historical reciprocating periodicity(N).The referential historic data came from 2004 to 2006.Alert threshold was P80.The evaluation indicators of alert effect were sensitivity,positive predictive value(PPV) and lead time.Excel 2003 was used to arrange data and calculate the moving percentile.SAS 9.0 was used to do single factor analysis and Logistic regression analysis.Results1 Alert effect under different alert parameters conditionsWhen the alert parameters of bacillary dysentery were W=7,N=1;W=7,N=2;W=3, N=1and W=3,N=2,the sensitivity were 30.8%,15.4%,46.2%and 26.9%,the PPV were 10.0%,18.8%,6.0%and 10.0%respectively,and the median of lead time were all zero day.When the alert parameters of measles were W=7,N=1;W=7,N=2;W=12,N=1 and W=12,N=2,the sensitivity were 59.2%,55.1%,44.9%and 30.6%,the PPV were 13.0%,13.8%,12.8%and 21.5%,and the median of lead time were 3 days,2 days,zero day and zero day respectively.2 Influence factors(Alert parameter was W=7,N=2.Alert threshold was P80)To bacillary dysentery,single factor analysis showed that the influence factors were x1 and x3.Logistic regression analysis didn't find the influence factorsTo measles,single factor analysis showed that the influence factors were x3,x4,x7 and x8.Logistic regression analysis showed that x4 was the influence factor.Logistic regression analysis showed that x3 and x8 can influence the alert effect of bacillary dysentery and measles.3 Results of the analysis of the control chartsBetween the successful alert counties and failed alert counties,the occurrence rate of the peak of control chart had significant difference.Conclusions1 Different parameter setting had different alert effect.2 x3 and x8 can influence the alert effect of bacillary dysentery and measles.3 In order to improve the alert effect,we must improve the reporting quality of infectious disease.4 We suggested that the alert algorithm should be improved and new data sources should be added in the automatic early-warning information system of infectious diseases.