Data analysis:
Once the data from the three sources are received and cleaned. MBDS team and BIOPHICS from Mahidol University will work closely to conduct the data analysis. Data analysis will be performed following the steps:
1. Determine temporal pattern of 5-year historical data: This step will provide an overall description of disease trend, which can be used to identify disease outbreaks.
2. Determine the disease outbreak timelines measures (defined below) for each country and calculate the median times for each timeliness measures. Cox proportional hazards regression analysis will be applied using methods described in Chan et al (2010) manuscript.
3. Determine the lag time of disease outbreak reporting among different disease reporting systems. Mean/median of the lag time over the 5 years will be determined by disease and different border regions.
4. Determine validity of data from different systems. Number of cases during outbreaks reported to the three systems will be calculated and compared to determine the magnitude of over- or under-estimate (expansion factor) of cases from different systems.
5. Assessment of impact of reducing lag time of disease outbreak detection among different disease reporting systems so that percentage of early detection of outbreaks will be improved and made more effective. Changes to systems during 5 year period will be examined to determine the impact on timeliness of detection and other measures.
Responsibility of BIOPHICS are seen in Annex C.