Background
Emerging public health threats often originate in countries that lack many public health resources and infrastructure [
1]. Because of the rapidity with which these diseases can spread, particularly with international air travel, early detection of disease outbreaks is extraordinarily important because it can provide for a quicker response and potentially limit morbidity, mortality, and the spread of the outbreak. In recognition of this fact, the World Health Organization issued revised International Health Regulations in 2005 (IHR 2005) that took effect in 2007. The purpose of IHR 2005 is to enhance global cooperation and protect populations from emerging health threats [
2] by requiring participating countries to “develop core capacities for surveillance, detection, reporting and response.” These core capacities include legislation and financing, national and international communication, preparedness, human resources, and laboratory resources. The 195 member nations of the World Health Organization had until the summer of 2012 to comply with IHR 2005, which requires establishing capabilities for detecting, reporting, and assessing public health events involving a disease that would be a public health emergency of international concern. [
3]. These requirements may be challenging for resource-limited countries, but there are ways in which technology may help. In 2005, Fraser et al. [
4] described a practical guide for implementing electronic medical record systems using open standards and open source software, based on pilot projects in six developing countries. Recently, Were et al. [
5] described a scalable open source electronic health record (EHR) implementation model that relies upon a national technical expertise center for external support and maintains electronic health records at multiple sites in resource-limited settings. Dennehy et al. [
6] described a partnership model for electronic health records in resource limited primary care settings. Ashar et al. [
7] described a variety of information and communications technologies that can be used for electronic health data capture and assessed their use in resource-limited settings. In 2010, Hartley et al. [
8] described the Global Health Security Initiative and discussed how electronic biosurveillance systems complement traditional public health surveillance to provide early warning and international awareness of disease outbreaks.
Syndromic surveillance systems typically use electronic, non-traditional, pre-diagnostic health indicators as surrogates for disease incidence to detect potential outbreaks in populations [
9]. These indicators may include a wide variety of data sources [
10], such as over-the-counter and prescription drug sales data, emergency department visit chief complaint data, physician office visit insurance claims data, nurse hotline data, etc. Syndromic surveillance systems complement traditional public health surveillance by providing non-specific yet early pre-diagnostic indications of potential disease outbreaks [
11]. The Electronic Surveillance System for the Early Notification of Community-based Epidemics (ESSENCE) is one example of an automated syndromic surveillance system. ESSENCE is a Java-based application used to monitor the health of populations and to detect disease outbreaks early and help prevent their spread [
12]. The fully-functional web-enabled version of ESSENCE (called Enterprise ESSENCE) is used by local and regional public health departments in different areas of the US and by the US Department of Defense and Veteran’s Administration [
13]. Enterprise ESSENCE is capable of collecting and analyzing a variety of data types and sources, and uses multiple anomaly detection algorithms to flag unusually high counts of disease indicators or alerts that are difficult for a human observer to see due to the volume and rate of change. System users can view, parse, plot, and map results, and share selected information with other users. Biosurveillance systems for resource-rich environments, such as Enterprise ESSENCE, are designed to use automated electronic data feeds and are best suited to areas with stable internet access.
Many global disease threats, like the 2003 Sudden Acute Respiratory Syndrome outbreak, first appear in resource-limited areas where electronic health data feeds and internet access are relatively unavailable or unreliable. Therefore, two additional versions of ESSENCE have been developed [
14]: ESSENCE Desktop Edition (EDE) and OpenESSENCE (OE). Unlike Enterprise ESSENCE which uses proprietary commercial products, EDE and OE utilize freely available open source software that provides several advantages in managing health data and performing medical surveillance in a variety of global communities. These advantages include wide scrutiny for quality assurance, low cost for acquisition and maintenance, and extensive user input on requirements, usage, and adaptability [
15]. Examples of the utility of open source software in health records and biosurveillance include: the public health data and information exchange methodology developed by the US Centers for Disease Control and Prevention (CDC) [
16]; an open source electronic medical record for implementation in developing countries described by Mamlin et al. [
17]; implementation of the Shibboleth information exchange for biosurveillance described by Lambert and Leonhardt [
18]; and an open source cyber-environment especially for disease surveillance described by Edwards et al. [
19].
Working closely with local public health departments, as well as the US military in their role of collaborating with host country military partners participating in support of IHR 2005 [
20] in resource-limited countries in Asia, Africa, and elsewhere, the Johns Hopkins University Applied Physics Laboratory (JHU/APL) obtained potential user input to determine a set of requirements for OE and EDE. These health departments desire an open source software system that would place a minimum burden on data providers, easily allow the user to tailor the system to their needs, and provide sustainability by allowing local jurisdiction to control their data, minimize costs, and easily maintain the system. Table
1 compares the features of Enterprise ESSENCE, EDE, and OE based upon these user requirements.
Table 1
Comparison of features of Enterprise ESSENCE, OpenESSENCE (OE), and ESSENCE Desktop Edition (EDE)
Uses proprietary 3rd party software | X | | |
Deployable using only open source software | | X | X |
Supports multiple types of data | X | X | X |
Supports a variety of data sources | X | X | X |
Plug-in API* for detection algorithms | X | X | X |
Security and encryption built into the core system | X | | X |
Customizable, dynamic, flexible design (property file driven) | X | X | X |
Designed for internet access | X | | X |
Supports language internationalization | | | X |
Supports font internationalization | | | X |
Configurable security/access authentication | X | | X |
EDE is a desktop version of ESSENCE that was developed for resource-limited and disaster settings with little or no internet access. EDE runs on a stand-alone computer as a self-contained application designed to deploy easily and function in diverse settings. EDE supports open development and extensibility. Although it does not have a built-in data collection capability, EDE easily reads a variety of data file formats, and allows the user to rename and characterize input variables. In the Philippines and Asia, EDE data have been collected by personnel entering data directly into electronic clinic records and via simple short message service (SMS) text messaging. Other countries plan to use smart phone data forms to collect health data and transmit it via SMS.
As the name implies, OpenESSENCE (OE) is an open source application that includes key features of Enterprise ESSENCE and can be used either with the internet or as a stand-alone system [
14,
21]. OE does not require the automatic secure internet data feeds that are used in Enterprise ESSENCE. Because of the distinct differences in Enterprise ESSENCE and its user community, converting all instances of Enterprise ESSENCE to OE is not planned at this time. Like EDE, OE provides for open development and extensibility, however, unlike EDE, it also contains a built-in data entry module. Data can be entered directly into the OE server, or via the web by multiple, geographically distributed users.
It should be emphasized that the primary goal of these efforts is to build capacity by giving resource-limited countries the independent ability to collect and analyze their own data. Because of this, JHU/APL does not have access to their data nor is that a goal. Based on more than a decade of experience working with local public health officials within the US, JHU/APL interacts closely with local public health officials in these countries in order to address quickly their concerns about best utilizing and maintaining their new system. The emphasis is on rapidly providing value as recognized by the user so that they begin using these systems as much as possible.
Results
The OE system was only recently deployed in 2011 and results are not yet available, but the EDE system has been in use long enough for preliminary results to be presented. The initial deployment of EDE was as an add-in module attached to the national disease surveillance system in the Philippines called the Philippines Integrated Disease Surveillance and Response (PIDSR) program. The goal of PIDSR is to reduce morbidity and mortality through a nationwide system that integrates facility-based information systems. EDE is now integrated with PIDSR and is used to monitor the temporal trends of diseases that are officially notifiable in the Philippines. Epidemiologists and computer scientists from JHU/APL visited the Philippines to solicit input from the stakeholders about the system architecture, types of usage, means of inputting data, and training. In 2009, a pilot study was begun to evaluate a simple fever surveillance protocol using SMS text messages to send daily, patient-level data from several local health clinics to the city health office in Cebu City [
28]. These data were included in a single SMS text message for each patient who presented with fever at the local clinics. Family and address codes, age, sex, date of onset, and presenting signs and symptoms were recorded for each patient as per the usual protocol. A formatted SMS text message about each eligible patient was sent each day to a receiver phone connected to a computer at the city health office. Standardized abbreviations were adopted for specific signs and symptoms (e.g., ha = headache). An SQL application was used to download the SMS data from the phone to an Epi Info database, which was analyzed using EDE by the city epidemiologist. Recently, the local health department has expanded SMS data inputs to all of Cebu City.
Before implementation of the SMS system, there had been a minimum two weeks delay between case presentation and case reporting to the city health office. SMS text messaging was the only practical alternative for this particular location because none of the local health clinics had an operable computer or internet connection. SMS was inexpensive and commonly used in this region. Using SMS also allowed the currently used logbook format to be maintained. Within a month of implementation of fever case data collection, 30 % of the local health clinics were using SMS texting to send daily fever case reports to the city health office. Within two months of implementation, this usage expanded to 90 % of the local health clinics and so far remains at about that level. The primary obstacle to increasing usage has been that local health department personnel have been accustomed to monitoring data post-event or at weekly or longer intervals based on previous data availability. Now that they have the ability to do daily monitoring, it can take some time for the users to become accustomed to doing so.
Discussion and conclusions
Based on information obtained from biosurveillance stakeholders in resource-limited countries, JHU/APL learned the following lessons: 1) the ability to maintain control of one’s own data is very important to the users, 2) open-source software is particularly desirable, 3) the system should place a minimum burden on those providing data, 4) system acceptance results from enabling the user to easily tailor the system to local needs, and 5) sustainability results from local ownership and working within existing needs and capabilities. This list of points is based upon informal qualitative discussions with local users regarding their needs, as our primary focus has been on rapid technology insertion, refinement, and increasing usage capacity. In addition to these lessons, our experience has revealed the importance of interaction with the appropriate levels of the local and national governments and the identification of key individuals who can serve as champions of the project, including those involved in local policy and financial matters as well as the actual users.
Therefore, the OE and EDE systems were created by JHU/APL to offer self-contained disease surveillance tools that can be deployed efficiently at a variety of resource-limited locations, as well as disaster locations. Both these systems are easily upgradeable and extendable. While skilled information technology professionals may be difficult for public health departments to find and retain, the OE and EDE systems have been simple enough to operate that this has not been a significant limitation. EDE was designed as a stand-alone desktop application, and OE can be used as a desktop or web-based application. Each provides similar functionality to the current web deployment of Enterprise ESSENCE [
12]. Both systems are based on a modular, component-based application design. This design allows for improved testability of components and isolation of problems. Repurposing and reuse is also easier and more likely with a modular design because it mitigates the need to redevelop or rebuild an application to incorporate adjustments or enhancements. The system structure is data driven to allow for dynamic extension and reconfiguration.
These biosurveillance systems were developed to support a wide variety of user needs in different settings. Both systems will provide user-defined preferences and mechanisms for data input from several types of databases. Users can configure the system specifically for the variables included in their database, thereby easing common data ingestion problems, especially the difficulties in trying to get disparate data formats to fit a specific type of data ingestion. Users can design their own queries and use different detection algorithms to analyze their data, including temporal and spatial analysis.
EDE utilizes an Eclipse RCP framework, while OE uses the Spring framework and Groovy Java language extensions. While both systems can operate on a single computer without a network connection, the OE system is designed to utilize the benefits of network connectivity. The OE system can also be used to share actionable information via the internet among multiple users and across different jurisdictions. Both systems provide results in a format similar to Enterprise ESSENCE [
12], including graphs, charts, detailed data on individual illness reports, and geographic maps of locations of individual illness reports. As open source stand-alone desktop applications, they are easily deployable, upgradeable and extendable. These features make EDE and OE ideal for rapid deployment in resource-limited environments where the infrastructure for a full-scale web-based biosurveillance system is not immediately feasible. Future efforts are to continue to improve the software features available to the user, while keeping the interfaces and components as simple as possible so that maintenance and sustainability do not require someone with a high level of information technology expertise. Such biosurveillance systems may then improve the timeliness of data collection and enhance the early detection of disease outbreaks, thereby allowing time for mitigation of the effects of these outbreaks.
Acknowledgements
We would like to acknowledge the collaboration with and financial support of the U.S. Armed Forces Health Surveillance Center, Division of GEIS Operations and the invaluable assistance from our colleagues, including: CDR David L. Blazes, AFHSC-GEIS; Dr. InKyu Yoon from the US Armed Forces Research Institute of Medical Sciences in Bangkok, Thailand; Dr. John Mark S. Velasco, Ms. Agnes D. Tomayano, Mr. Danny C. Obidas, Mr. Jewernest C. Casquejo, and Dr. Maria Theresa P. Alera from the US Armed Forces Research Institute of Medical Sciences Virology Research Unit, Republic of the Philippines; Dr. Fe A. Cabugao, Dr. Ilya A. Tac-an, and Ms. Durinda R. Macasocol from the Cebu City Health Office, Republic of the Philippines; and Dr. Joan M. Neyra, Dr. Delphis M. Vera, Dr. Ricardo A. Hora, and Dr. Joel Montgomery at the US Naval Medical Research Center Detachment in Lima, Peru. We thank Ms. Shradda Patel and Dr. Brian Feighner for their assistance on this manuscript. We are deeply indebted to Mr. Wayne A. Loschen, the developer of the Enterprise ESSENCE system, for his sage guidance on these projects.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
TCC was the technical lead and developer for the OE application, and helped develop EDE. CJH was the technical lead and developer for the EDE application and helped develop later versions of OE. SMB wrote the manuscript, selected figures, and obtained references, based on his work on the ESSENCE project. AMP was the technical lead for telephone and SMS data ingestion design and development and was a software developer for OE. RAW was the chief software engineer for all versions of ESSENCE. JFS was the original software technical lead and software developer for OE. JSC was the chief epidemiologist on the project and obtained user input and usage data on the application. ZM developed several key components on the EDE project. CJH, TCC, RAW, JSF, JSC, and SHL helped develop user requirements for these applications. All authors read and approved the final manuscript.