JMIR Publications

ABSTRACT

Background:

Influenza is a leading global cause of death and contributes to heavy economic losses to individuals and communities. Early prediction of and interventions against influenza epidemics are essential to reduce mortality and morbidity. As in other countries, the Taiwan Centers for Disease Control and Prevention (TWCDC) has implemented influenza surveillance and reporting systems, which mainly rely on influenza-like illness (ILI) data reported by healthcare providers for early prediction of influenza epidemics. However, these surveillance and reporting systems have at least a 2-week delay, leaving room for improvement.

Objective:

We aimed to integrate the TWCDC’s ILI data with the electronic medical records (EMRs) of multiple Taiwan hospitals. Our overall goal in doing so was to develop a more accurate and efficient tool for national influenza trend prediction than that available currently.

Methods:

First, surveillance variables relevant to the prediction of influenza epidemics were identified by the influenza expertise team at the Taipei Medical University Healthcare System (TMUHcS). Second, we developed a framework for integrating hospital EMR data with the ILI data from the TWCDC website. Third, we used Pearson correlation coefficients to measure the strength of the linear relationship between the TMUHcS’s EMR data and TWCDC’s ILI data (including regional and national ILI data) for two weekly time series datasets. Lastly, we evaluated the predictive power of each surveillance variable for influenza epidemics using the Moving Epidemic Method (MEM) analyzes.

Results:

For the past three influenza seasons (October 2014 to September 2017), three surveillance variables (TMUHcS-ILI, TMUHcS-RITP, and TMUHcS-IMU, reflecting patients with ILI, those with positive results from rapid influenza diagnostic tests, and those treated with antiviral drugs, respectively) were identified from the EMRs of multiple hospitals and showed strong correlations with the TWCDC’s regional and national ILI data (r = 0.86–0.98). Two surveillance variables (TMUHcS-RITP and TMUHcS-IMU) showed predictive power for influenza epidemics 3–4 weeks prior to an increase in TWCDC ILI reports.

Conclusions:

This framework integrates surveillance data from multiple hospitals and the TWCDC website. It could be easily extended to other infectious diseases, mitigating the time and effort required for data collection and analysis. Furthermore, this approach could be developed as a cost-effective electronic surveillance tool for early and accurate prediction of epidemics of influenza as well as other infectious diseases in densely populated regions and nations.