200 Chapter 9 efficient. This approach holds potential as a tool for pandemic surveillance in future outbreaks. In the first part of this thesis, the use of different data sources of routinely collected health care data for disease surveillance was studied. Traditional infectious disease surveillance systems have several limitations, as listed in the introduction of this thesis (Chapter 1). These systems are typically expensive and suffer from a timelag, because of the labour-intensive process of collecting data. One possible solution to this is to utilise routinely collected data. One outbreak (hepatitis in children) and one disease syndrome that needs continuous surveillance (Severe Acute Respiratory Infections) were analysed. Moreover, these studies were linked to two early steps in the data process for disease surveillance: data collection and data aggregation. Data collection can be a time-consuming part of infectious disease surveillance. If the information is collected directly from the patient, for example by asking about symptoms or taking a temperature measurement, data collection would be as accurate as possible. If the information is collected on a case-by-case basis from electronic health records (EHRs), it is very likely that the correct information is gathered, as long as it is present in the EHR. The obvious downside of both of these approaches is the amount of time it takes to collect this data, the potential for transcription errors when recording data from EHR into a separate database, and the use of skilled staff who also have other tasks related to patient care. As an alternative, software that allows automated data extraction from EHRs can be used. Sometimes, extracting data that reflects the exact case definition is difficult, as is the case with SARI. The disease definition of SARI consists of several elements which complicates rapid mining of structured data, which is currently the easiest method to extract data from EHRs. In those cases, substituting the case definition for a proxy can facilitate data extraction. In Chapter 2, automatically collected EHR data was used to evaluate different outcome measures for SARI surveillance, using different proxies for SARI. A downside of utilising proxies is that they are indirect indicators of SARI. While this approach is unlikely to detect the true absolute number of cases, it can detect changing trends over time, which is essential for syndromic disease surveillance6. Granular data in place and time is essential for effective disease surveillance. After local data collection, rapid sharing of this information between institutions can improve quality of surveillance. However, privacy concerns can complicate data sharing. A possible solution for this problem is to locally aggregate data to anonymised summary data that cannot be traced back to an individual. This anonymised and summarised data can then easily be shared between institutions and can be further aggregated, to show disease patterns on a local, regional, national and even international scale. In Chapter 3 this method was tested for the global severe hepatitis of unknown aetiology outbreak in children in 2022. As demonstrated in Chapter 3, it is feasible to collect large amounts of summary data, with relatively minimal efforts. With repeated data extractions over time, it would even be possible to set up a surveillance system with (near) real-time data from
RkJQdWJsaXNoZXIy MTk4NDMw