Taiwan has very high population density (Taiwan, 632.23 persons/km2; Taipei, 9662.53 persons/km2), which increases the spread of many human-to-human infectious diseases . This makes the surveillance of infectious diseases very important for this island. Taiwan's ED-SSS is the first syndromic surveillance system to be implemented in Asia. It also represents the first time that Taiwan's public health agencies have attempted active nationwide surveillance. Its automatic data collection mechanism is capable of capturing comprehensive population-based health information and providing important details on current disease epidemics at the community level. The information it provides can also be used as community baseline data for further infectious disease modeling and can also improve the detection of emerging infectious diseases.
In addition to the information that the ED-SSS can provide for disease control, it can open avenues for further investigation. For example, in addition to the neurological syndrome, asthma syndrome, and syndrome for severe symptoms, there were clear and consistent weekend and holiday increases in the visits of other nine syndrome groups. Because cost of ED visits in Taiwan is not as expensive as it is in other countries, especially the United States, it is very likely that many patients seek ED medical care when local clinics are closed on weekends and holidays. It is also possible that the gathering of people on the holidays would increase the transmission of certain pathogens, particularly on cold days in closed spaces where respiratory viruses including influenza virus are easily transmitted. Therefore, future research might want to investigate the effect of holidays on the aberration detection of outbreaks and prediction of number of cases for certain infectious diseases using Taiwan's ED-SSS.
We also found differences in seasonal trends in visits due to symptoms/signs related to respiratory, influenza-like illness and asthma syndromes. Our ED-SSS found a summer peak in visits for cases with influenza-like illnesses in 2004. This has seldom been found by the previous passive surveillance systems used in Taiwan. These summer cases of influenza-like illness occurred before annual vaccinations usually done in October or November. Therefore, a further longitudinal analysis of influenza-like syndrome patterns is needed to formulate the best vaccination policy on human influenza.
Another epidemiological finding from our ED-SSS was an increasing trend in visits due to gastrointestinal syndrome starting late autumn 2004. Such trends have not been detected by other infectious disease surveillance systems in Taiwan. There are two possible explanations for this finding. One reason for the increases might be related to the increased activity of certain pathogens, including rotavirus or norovirus, during winter season, as was found during the winter of 2006 in both Japan and Taiwan [20–27]. Another reason might be the social habits of Taiwanese who like to dip raw meats and seafood into boiling water fondues and eat from the chafing pot during the winter season. This would increase the change that inexperienced or careless diners would consume undercook seafood or use chopsticks contaminated by raw seafood.
The findings of our ED-SSS, the first time in Taiwan to use daily rather than weekly data, suggest further directions for research into GI syndrome and many other diseases of significant interest to public health. For example, during the 1998, 2000 and 2001 enterovirus 71 epidemics, children aged 3 years and younger who were at higher risk of severe or fatal cases of the disease were identified for more effective prevention only after the occurrence of several cases of sudden deaths from weekly sentinel physician surveillance and later retrospective epidemiological data analysis on those cases when sample size became larger. Therefore, prospective monitoring of daily ongoing data of EVI syndrome in this high risk age group and early ED-SSS detection of enterovirus activities by local public health personnel might help minimize social panic among parents. Furthermore, the results on seasonal pattern of enterovirus-like infection in our ED-SSS was consistent with the previous epidemic patterns in Taiwan, again demonstrating the usefulness of ED-SSS to avoid future large-scale or severe epidemics caused by enteroviruses . In summary, these initial findings suggest that it is necessary to develop algorithms capable of detecting aberrations for different syndrome groups from patients in different geographical areas of Taiwan, taking into account variations in the levels of medical care and the effect of weekends and holidays on ER visit.
The ED-SSS did not, however, reveal obvious trends in all syndrome groups. For example, it was hard to find seasonal patterns or secular trends in cases of coma/death, skin rash, or neurological symptoms – the three syndrome groups that might be useful in the detection of severe outbreaks caused by bioterrorism, e.g., anthrax, during the study period without bioterrorism attacks [28, 29]. Certainly, continuous monitoring for these syndrome groups at both local and national levels will be very helpful in detecting possible bioterrorism or EIDs in future years. Using those trends in coma/death and other syndrome groups of clinical severity or unexpected symptoms/signs, our ED-SSS data have provided directions for further research in the areas of pathogen detection, epidemiological clues, and improvement in public health policies. Therefore, future investigations have to control the weekend and holiday effects of ED visits for better aberration detection even during long holidays.
In daily public health practice to monitor the data of ED-SSS, careful verification and systematic management is needed once the aberration signals are detected. The server needs an automatic error feedback system function instead of the original use of engineers to double check for data errors would increase the efficiency and completeness of surveillance. Future efforts require closer collaboration between computer-science professionals and medical informatics personnel at the Taiwan-CDC to establish a system with the standard operating procedures (SOP) for database maintenance and to provide more continuous on-job training for both hospital users and local and central public health agencies .
The major difficulty in developing our ED-SSS was diverse formats for different types of data, including categories of chief complaints, the ways to fill out ICD-9-CM codes, and even the different number of digits used in home zip codes in different participated hospitals. For example, most Taiwan hospital ED physicians/nurses EDs only write down one chief complaint, which is very different from the ED reports made by most U.S. hospitals which list all possible complaints (with text format) in English. Several participating hospitals only had a paper system for recording triage chief complaint data. A standard format for select syndromes and variables needs to be established and continuously reevaluated to improve data quality and stability of data transmission. There are needs to have more research into the chief complaints with Chinese styles, the suitability of chief complaints vs. ICD-9 codes, how to combine symptoms/signs and link data to improve sensitivity.