Respiratory viruses cause significant global mortality and morbidity each year. Influenza virus is of particular public health concern due to its association with severe infections and deaths [1,2]. Disease surveillance provides useful information that helps monitoring trends and disease burden, planning, implementing, and evaluating appropriate prevention and control interventions as well as allocating resources [3]. At present, the World Health Organization (WHO) runs the largest human influenza virus surveillance network [4], which includes 136 institutions from 106 countries mainly focusing on the genetic and antigenic characteristics of prevailing strains. While the original purpose of WHO’s global influenza surveillance network was to recommend the content of the influenza vaccine for the subsequent influenza season [5], local and national prospective influenza and influenza-like illness surveillance systems also provide important information to policy makers and public health practitioners for situational awareness during periods of influenza activity.

Great strides have been made exploring and exploiting new and different sources of respiratory disease data, especially after the severe acute respiratory syndrome (SARS) epidemic in 2003 [6,7] and the 2009 influenza A (H1N1) pandemic. However, as different surveillance methods target only specific target groups of population possibly at different stages of disease progression, thus providing related but different information [8], using a single surveillance data stream to reflect overall disease activity may not be appropriate. It remains challenging to incorporate multiple surveillance data and fully utilize the available information from the data to summarize the overall situation.

On the other hand, there has been rather less research in the area of data dissemination, while the feedback of contextualized data to stakeholders is and must be a key aspect of any surveillance system if confidence with and enthusiasm for the system is to be maintained [9,10]. It is also important that the dissemination is not neglected. No matter how accurate and timely the upstream surveillance data collection and analysis of the surveillance system are, the return for the time and effort invested would be discounted if the means of health communication for data dissemination were suboptimal. Poor communications could lead to delayed information transfer, loss of information, or even misinterpretation of the surveillance results. Effort should also be taken for data dissemination to complete the surveillance-to-action loop [3].

Digital dashboards (also known as executive dashboards) describe computerized interactive tools typically used by managers to visually ascertain the status (or "health") of their business via key performance indicators. These tools emerged from the concepts of decision support systems in the 1970s, and, with the rapid growth in information technology through the 1990s, these have developed into standard tools in executives offices [11,12]. Their use has expanded to the field of medical science, for example, displaying bed occupancy, the availability of clinical staffs for hospital management [13,14], or monitoring the stock of different pharmaceutical items in the dispensary [15]. In a more general sense, digital dashboards simply describe systems for the visual presentation of key indicators in appropriate context, allowing rapid interpretation by an end user. A dynamic dashboard can be made interactive and user-friendly by allowing extraction of information in different perspectives and contexts. By investing in the design of data dissemination, the quality of presentation and communication of data by local health authorities as well as the general public could be much improved, thus maximizing the usefulness of available data of the surveillance systems.

In this research paper, we developed a generic framework for a digital dashboard and illustrate this framework by specific application to influenza surveillance. Specific features were designed and incorporated according to the requirements identified to facilitate efficient dissemination of surveillance data. The dashboard can reveal both the influenza activity in different population sectors and the overall situation at a glance, maximizing the use of available data for policy makers as well as the general public for appropriate actions.

The influenza surveillance dashboard is hosted on the server of the School of Public Health, the University of Hong Kong (http://sph.hku.hk/dashboard). Here we present the design patterns [19,20] describing the main features of the surveillance dashboard.

We developed an influenza surveillance dashboard that serves as a template to demonstrate efficient synthesization and dissemination of multiple-source surveillance data, which may also be applied to other diseases. Surveillance data from multiple sources can be disseminated efficiently using a dashboard design, which facilitates the translation of surveillance information to public health actions.

Public health surveillance is defined by Thacker as “the ongoing, systematic collection, analysis, and interpretation of health data essential to the planning, implementation, and evaluation of public health practice, closely integrated with the timely dissemination to those who need to know” [21]. Previously, much effort and many resources were required to improve the quality of surveillance data and the accuracy of diagnostic tests. Until the late 1990s, with the rapid development of information technology and the Internet, the timeliness of surveillance data transfer improved. Nevertheless, the methods and targets of data dissemination have been somewhat neglected, probably because communication and graphic design are apparently unrelated to the field of medical science. Indeed, timely and effective health communication is extremely valuable to transfer information to the relevant people who can take appropriate action. In particular, timeliness is always a critical factor for infectious disease containment and control. Comparatively speaking, fewer resources are needed to develop effective data dissemination methods than are required for data collection, quality control, diagnosis, and analysis. Researchers should also keep an eye on data dissemination to maximize the usefulness of the upstream surveillance effort.

A review of national surveillance websites [8] provided evidence that less attention has been paid to designing user-friendly and efficient dissemination of surveillance data. In particular, many websites have presented or highlighted data in report form, which made information more difficult to be identified or synthesized. Long reports with scattered information over pages also made it difficult to grasp the overall situation of disease activity. Also, functions for data retrieval were not commonly available for most websites. Such functions are greatly needed for efficient and timely data sharing.

The dashboard style can automatically summarize data and present figures of different types in the drill-down pages. Other detailed or supplementary information, such as past records of individual surveillance data streams or recommendations can be available in other drill-down pages. We also incorporated a data export function in standard format (eg, CSV and XML) to facilitate data sharing, which was shown to be important for epidemiological enquiry in the recent human influenza A (H1N1) pandemic.

Moreover, a primary motivation for publicizing surveillance data online should be to allow timely risk communication thereby facilitating disease prevention [22]. As public health awareness has been elevated since the SARS outbreak in 2003, it is expected that sources of surveillance data will only increase. In the influenza surveillance dashboard, we demonstrated how recommendations can be made corresponding to multiple surveillance data so that the collected information can be more efficiently transferred into public health actions.

While more surveillance data streams are being developed, there is a need for developing more sophisticated algorithms for aberration detection. Here, some challenges remain for the development of multivariable analytic algorithms as different surveillance data streams represent situations of different targets at different stages of disease progression. Furthermore, evidence is still lacking about the association between these new surveillance data streams and influenza activity. Whether these data provide more signal or noise is still in doubt due to their nonspecific nature. These data streams will still need to be evaluated and filtered to increase the signal-to-noise ratio before adding them as a routine surveillance practice; thus, they can only serve as supplementary information at the current stage. Other challenges for using combined surveillance data from different sources include the handling of missing data (eg, school absenteeism during school holidays), data provided in different resolutions (daily versus weekly resolutions), and unsynchronized data receiving protocols from different sources, which the algorithm should also be capable of handling.

Generally speaking, an ideal surveillance website should have comparable hardware and software and a standard user interface, data format, and coding for easy data sharing. The influenza surveillance dashboard demonstrates how different sources of surveillance data streams can be synthesized and efficiently displayed in a dashboard to provide an overview of disease activity as well as each individual source. The influenza surveillance dashboard serves as a template that can provide for efficient dissemination in a similar manner of information about other diseases with multiple surveillance data sources. With a back-end algorithm, indicators or statistics can be generated from individual or all data streams. Other more complicated indicators can also be generated that are tailored to specific disease type by modifying the back-end algorithm.