Year : 2017 | Volume
: 5 | Issue : 2 | Page : 55--59
Informatics enables public health surveillance
Scott J. N McNabb1, Paige Ryland2, Joy Sylvester2, Affan Shaikh2,
1 Department of Global Health, Emory University, Rollins School of Public Health, Atlanta, GA 30305; Public Health Practice, LLC, Brooklyn, NY, USA
2 Public Health Practice, LLC, Brooklyn, NY, USA
Scott J. N McNabb
Emory University, Rollins School of Public Health, 2575 Peachtree Road NE, Atlanta, GA 30305
Over the past decade, the world has radically changed. New advances in information and communication technologies (ICT) connect the world in ways never imagined. Public health informatics (PHI) leveraged for public health surveillance (PHS), can enable, enhance, and empower essential PHS functions (i.e., detection, reporting, confirmation, analyses, feedback, response). However, the tail doesn't wag the dog; as such, ICT cannot (should not) drive public health surveillance strengthening. Rather, ICT can serve PHS to more effectively empower core functions. In this review, we explore promising ICT trends for prevention, detection, and response, laboratory reporting, push notification, analytics, predictive surveillance, and using new data sources, while recognizing that it is the people, politics, and policies that most challenge progress for implementation of solutions.
|How to cite this article:|
McNabb SJ, Ryland P, Sylvester J, Shaikh A. Informatics enables public health surveillance.J Health Spec 2017;5:55-59
|How to cite this URL:|
McNabb SJ, Ryland P, Sylvester J, Shaikh A. Informatics enables public health surveillance. J Health Spec [serial online] 2017 [cited 2020 Nov 29 ];5:55-59
Available from: https://www.thejhs.org/text.asp?2017/5/2/55/205075
The world is different now. Informatics is transforming public health surveillance (PHS), and electronic surveillance (e-Surveillance™) is evolving. However, public health practice must adapt to reap the benefits because the tail does not wag the dog. This means that Information and Communication Technology (ICT) cannot (should not) drive PHS strengthening (PHSS™). However, essential PHS functions (e.g., detection, reporting, confirmation, analysis, feedback and response) can be enabled, enhanced and empowered through automated, electronic processes.
As the proliferation of global travel, urbanisation and mass gatherings accelerate the occasions of the spread of human disease, it is critical that PHS implements targeted prevention strategies and detects outbreaks more quickly to prevent epidemic spread. New technology provides tools for PHS to keep up with the pace of microbe evolution; however, a strategic approach is needed to make these tools widely available and usable by public health professionals. In this review, we explore promising trends in public health informatics (PHI) for prevention, detection and response.
Advances in E-Surveillance™
Detection and reporting
Provider-initiated manual reports have been the hallmark of conventional, passive PHS. However, manual, case-based reporting falls outside the traditional clinical workflow; it is an added burden for the healthcare worker. We must ask this question: “What is the impact of this added burden to the timeliness and completeness of PHS reports?” Advances in ICT automate case detection and reporting, minimising – at times even shifting – the burden from healthcare workers. The movement of data can be passive (provider-initiated) or active (epidemiologist-initiated), and advances in ICT enhance, enable and empower both.
Automated case reporting
Recent ICT progress focuses on shifting the case detection and reporting burden from manual, clinical and laboratory reporting to automated, artificial intelligence (machine supports man). Conventionally, active PHS involved epidemiologists looking for PHS data by hand from electronic medical records (EMRs) during case investigation and contact tracing. This activity was generally limited to high impact scenarios such as outbreaks, mass gatherings or disease elimination/eradication.
Cutting edge projects show that epidemiologists can now electronically “request” data from EMRs in real time. The Massachusetts Department of Public Health partnered with researchers from Harvard University to implement MDPHnet, a two-part artificial intelligence enabling public health to “request” data every 24 h from a network of EHRs covering 700 physicians at 30 practice sites. With access to data, researchers developed and validated case-finding algorithms. Studies validating case-finding algorithms found a sensitivity of 100% (95% confidence interval [CI] = 52–100) for tuberculosis and 99.1% (95% CI = 97–100) for hepatitis B and a positive predictive value of 91% (95% CI = 57–100) and 97.4% (95% CI = 94–100), respectively., Currently, MDPHnet identifies and reports acute hepatitis A, B and C; tuberculosis; chlamydia; giardiasis; gonorrhoea; Lyme disease; pelvic inflammatory disease; pertussis and syphilis.
Electronic laboratory reporting
Electronic laboratory reporting (ELR) of notifiable diseases has the potential to improve the completeness, timeliness and accuracy of case-based reports. ELR significantly improved timeliness in several reports.,,, In a comparison of automated ELR and immediate, paper-based reporting of notifiable conditions in Indianapolis, Indiana (USA), ELR identified 4.4 times as many cases as traditional reporting, and among the cases that overlapped, reporting them an average of 7.9 days earlier. In Oklahoma, for diseases requiring reporting within one business day or less, 91% of cases reported using ELR were on time versus 87% of conventionally reported cases. In North Carolina, a 2012 study found median case processing time of 20 days for ELR versus 25 days for conventional reports.
ELR has been shown to drastically improve the number of cases reported and therefore the coverage PHS., For example, both laboratories and providers in the New York City are required to report viral hepatitis B, C and enteric infections. A review of reported cases from 2007 to 2010 showed out of all cases of hepatitis C, 90% were reported by ELR only, 2% were reported by providers only and 8% were reported by both laboratories and providers. A random sample of the 2% that were reported by only providers found that only 79% were true hepatitis C cases, with the others reported by error or in duplicate. This led Stachel et al., to conclude that only 1% of hepatitis C cases would be missed if provider reporting was discontinued completely. Analysis of the same differences between hepatitis C and enteric diseases found similar, though not as extreme, results.
While ELR is useful for notifiable diseases requiring laboratory confirmation, it is limited in scope and can miss culture-negative diagnoses. In addition, there are challenges with completeness of data variables in laboratory reports, specifically with patient identifier information. Further, there is the challenge of matching laboratory reports with clinical ones; this must be mitigated by uniform use of case identifiers. Automated reporting of health indicators from EHRs can alleviate the clinical reporting burden.
In Kenya, researchers field-tested automated health indicator reporting from the Open Medical Record System (OpenMRS) to the national health information system run on the District Health Information System (DHIS2) platform and compared data received with monthly reports entered directly into a DHIS2 data entry module. They found that electronic reporting yielded 100% completeness and accuracy for all 45 data values tested as compared to completeness of 66.7%–100% and accuracy of 33.3%–95.6% in manual data entry and reporting.
They measured that manual data entry into DHIS2 took over ½ h of staff time across multiple people, compared to  Similarly, add-on applications have been developed to identify potential vaccine adverse events in real time, prompt a physician to evaluate the condition and then facilitate the automated generation and transmission of the mandated report.
Advances in Public Health Analytics
After data are collected from clinicians and laboratories, it is analysed for changes in trend. Analyses were historically performed by hand, requiring large inputs of skilled labour and time. Recently, epidemiologists rely on manually initiated analysis programs with the continued challenge of a time delay between receipt of data and aberration detection.
With advances in PHI, the speed of outbreak detection is improved, and, in some cases, the time and place of an outbreak can be predicted with varying degrees of accuracy enabling opportunities for prevention and control. Advances in analytics include the identification of statistical and visualisation applications, generation of algorithms to automatically alert users to aberrations in health events and leveraging high-performance computational resources for large data sets or complex analyses.
ICT innovations include real-time, environmental data, remote sensing systems, global positioning systems and geographic information systems. Combined with an ever-growing number of people accessing the Internet (and the crowd-sourcing and real-time information seeking data resulting from such use), ICT can lead to improved outbreak detection.
In the United States, CDC PulseNet uses DNA fingerprinting to improve food safety systems through a national laboratory network to detect food-borne illness outbreaks early and trace the pathogen source. PulseNet's network includes local, state and national laboratories relying on standardised protocols for molecular subtyping of case isolates to identify clusters of illness and standardised methods of communication for exchanging results and other information.
The speed at which PulseNet can detect outbreaks means a quick response (through product recalls, restaurant closures and others) breaking the chain of ongoing disease transmission. The CDC laboratory for bacterial pathogen analysis in China, a member of PulseNet International, is working to adapt the idea and methodology of PulseNet for all bacterial infectious disease.
Advances in analytics (and ICT generally) are sometimes not widely available although their benefits are universally necessary for global health security. For example, early warning systems (EWS) in Iraq, Macedonia, Morocco, Serbia and Sudan struggle with a lack of standardisation, automation and statistical methods for detecting changes in disease trends. In the effort to assist these countries to improve their EWS, the World Health Organization has worked to standardise data entry, analysis and reporting in a computerised system and provide training to healthcare staff and epidemiologists.
While these systems can improve, the simple advance of a computerised reporting system using country-identified thresholds for priority diseases showed improvements in monitoring trends and predicting outbreaks in a matter of weeks post-implementation.,, One of the greatest challenges still remaining for EWSs is, ironically, being early enough. In Lebanon, an assessment of the electronic surveillance of outbreaks based on the EWS for four diseases showed a time interval between the first cases and first abnormal signals to be on average 4 weeks in 2005 and 5 weeks in 2007.
Improving the accuracy and lead time of EWS can be improved using temporal modelling. Mostafavi et al., found that combining laboratory, weather, veterinary and national statistics data with temporal modelling produced a model predicting the number of cases of Crimean–Congo haemorrhagic fever in Iran 1 month in advance. Improvements to EWS like these stem from advances in predictive PHS.
Predictive public health surveillance
Predictive PHS offers early warning of disease outbreaks using computer algorithms validated by historic data and through analyses of real-time data. A standout of predictive analytics is its focus on finding and testing relationships in these data (especially between human and environmental health data) to anticipate risk as opposed to the one-dimensional view descriptive analytics provides. One example includes modelling vector-borne disease and weather conditions to predict potential disease outbreaks depending on the weather. Improved technologies allow for the creation of models that are realistic and testable.
The power of PHI advances in analytics relies on the quality of data and algorithms used. Modifications to algorithms affect the sensitivity and positive predictive value, and the balance between the two depends on local analysts making decisions based on the local context.
A test of four-time series algorithm modifications for improving sensitivity for detecting artificially added data in the US BioSense system showed that improved detection algorithms expand the usefulness of automated biosurveillance by increasing sensitivity 20%–40% without changes to the alert rate. This increased sensitivity is important for accurately and quickly recognising a potential outbreak, and models must continue to be modified for the best results.
At a provincial level, KFL&A Public Health in Ontario, Canada, uses data from the acute care enhanced surveillance (aces) system to communicate both seasonal influenza risk and a projection (by 1 week) of activity level in ACES. The predictions consider current outpatient visits to the emergency department, hospital admissions for pneumonia and influenza-like illness (ILI), and knowledge of the locally circulating strains of influenza and their epidemiology. This information is presented on an ILI website accessible by the public showing a heat map of seasonal influenza risk.
A three-tiered, novel approach to using predictive PHS comes from Sri Lanka where a social media system is used for dengue prevention. First, a computer simulation is used as a traditional piece of predictive PHS to warn of an outbreak. Second, a civic engagement component allows the public to use social media tools to share information on symptoms, mosquito bites, and breeding grounds with health officials. Finally, a health communication component provides health awareness information back to citizens based on the computer simulation and citizen-provided data.
The success of this project was largely the result of widespread mobile technology enabling public health officials and citizens to actively, effectively and proactively engage with one another on an important health issue.
PREDICT, a joint program led by the UC Davis One Health Institute working in over thirty countries, is the building systems for viral surveillance at the interface of human, animal and environmental health to proactively discover pathogens of pandemic potential at their source. Using broadly reactive consensus polymerase chain reaction in conjunction with high throughput sequencing, data are produced which allow for rapid detection of known and potential pathogens. Other data sources employed in the PREDICT project include land use, socioeconomic, agricultural changes with surveys of human behaviour, market value chains and livestock production. Viral testing data and outbreak scenario modelling using these varied data sources tell PREDICT which pathogens are most likely to become pandemic.
New Data Sources
With the advent of the Internet, PHS has benefited from rapid advancements in ICT and a number of novel approaches that complement more traditional PHS methods. As such, one of the major revisions in the 2005 International Health Regulations (2005) is the inclusion of event-based information from unofficial sources to detect rumours about possible outbreaks or cases of diseases considered public health emergencies.
Event-based PHS is an emerging scientific discipline that leverages the Internet's diverse streams of data for the benefit of human and or animal health. While conventional indicator-based surveillance systems utilise routine collection of structured data, newer event-based surveillance systems use unstructured data from the media and other non-traditional sources to detect anomalies that may indicate a new threat. Unofficial sources largely come down to whatever information is available on the Internet, whether disseminated by news websites or collected through crowdsourcing methods. Computational methods have also enabled data mining or the analysis of semantics and keywords scattered on the Internet to identify trends.
Event-based PHS can be classified into three main categories: news aggregators, automatic systems and moderated systems. News aggregators collect reports and articles from sources, screened by various geographic or languages, and offer a common access portal. A popular aggregator for PHS was the now-decommissioned Google Flu Trends and Google Dengue Trends.
Automatic systems are fairly more complex, involving a series of analysis steps that differ in the scope of information sources, language coverage, delivery and visualisation. Examples of automated systems include EpiSPIDER, HealthMap, EpiSimS, MedISys and GETWELL. Finally, moderated systems, such as ProMED-Mail, have some human component, where an analyst processes and interprets information.,,,,
PHS using the Internet, also known as web-based reporting and surveillance, first emerged as a viable source of reliable data as early as 1994, with the release of Program for Monitoring Emerging Diseases (ProMED-Mail). ProMED-Mail has the aim of rapidly disseminating disease-related information to a wide audience and allowing for informed discussion in real time.
Since 1999, the International Society for Infectious Diseases has operated ProMED-Mail. The system functions as a listserv that receives reports of public health events in humans, animals and plants from its subscribers who report information from many different sources, including personal observations, written or electronic reports, the news media and various other online sources. Submissions are moderated by a dedicated group of epidemiologists. After a vetting process, verified reports are accepted and disseminated with references and suspected aetiology. By its 10th year, ProMED-Mail was sending an average of seven reports each day to user-base of 35,000 listserv subscribers from around the world.
The second stage of development of the worldwide web was characterised by the change from static web pages to dynamic or user-generated content and the growth of social media. Coined Web 2.0, these changes brought more robust forms of PHS. This new generation of web-based PHS uses data from a wide range of sources including query data from online searches and social media and was typically developed and supported at a national and international level as complementary to more traditional PHS activities. Digital platforms such as HealthMap, Google Flu Trends and Flu Near You allow visualising epidemiologic scenarios around the world.
The future of PHS is extremely promising as continual advances in PHI work to enable, enhance and empower the core functions of PHS. However, these advances can only be harnessed if public health decision-makers and practitioners are prepared to adapt to a new landscape. The ICT improvements that support PHS are expensive, complicated and wrought with privacy, security and quality challenges, and the amount of available digital data and information grows each day. ICT must support existing business processes within a public health organisational entity; however, works such as digitising data transfer and automating analyses do impact business processes. People, politics and policies are the most challenging issues to navigate in the implementation of PHI solutions.
With a seemingly never-ending flow of new data and new technologies, only a skilled workforce, continuously trained on updated tools, standards and use requirements can take advantage of these advancements. Collaboration among all partners (e.g., government, academia and not-for-profit public health institutions) is vital to ensure that the best technologies often developed at academic institutions make it to primary health units and public health organisations where the work of PHS is being conducted. Best practices, which should accompany these technologies, should be passed to public health professionals.
In our world, the drawn boundaries separating countries are invisible to the world of public health. Data sharing across borders is necessary and allows for collective action to identify, analyse and react to health data that concerns us all. To accomplish this, politicians must be transparent in what data they share and how they do it, in addition to sharing human and material resources for PHS action. Strong top-down leadership and champions of change are required to move a country or region forward in a unified manner.
To ensure a wider network, which can seamlessly exchange the data necessary for global health security, health data and information standards must be adopted. These standards should include provisions for interoperability and privacy/security. The challenge of data quality that comes from large volumes of unchecked data can be lessened somewhat by innovations in PHI. Automatic systems can be set up to monitor data quality of incoming data at regular intervals and deliver alerts when an aberration is detected.
We would like to acknowledge Ms. Sonia Shahin and Dr. Sara Cheung.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
|1||McNabb SJ, Conde JM, Ferland L, MacWright W, Okutani S, Park M, et al. Transforming Public Health Surveillance: Proactive Measures for Prevention, Detection, and Response (Imprint). 1st ed. Amman, Jordan: Elsevier; 2015. p. 850.|
|2||Lazarus R, Klompas M, Campion FX, McNabb SJ, Hou X, Daniel J, et al. Electronic support for public health: Validated case finding and reporting for notifiable diseases using electronic medical data. J Am Med Inform Assoc 2009;16:18-24.|
|3||Klompas M, Haney G, Church D, Lazarus R, Hou X, Platt R. Automated identification of acute hepatitis B using electronic medical record data to facilitate public health surveillance. PLoS One 2008;3:e2626.|
|4||Calderwood MS, Platt R, Hou X, Malenfant J, Haney G, Kruskal B, et al. Real-time surveillance for tuberculosis using electronic health record data from an ambulatory practice in Eastern Massachusetts. Public Health Rep 2010;125:843-50.|
|5||Notifiable Diseases â ESP; 2017. Available from: http://www.esphealth.org/index.php/disease-surveillance/notifiable-diseases/. [Last accessed 2017 Feb].|
|6||Samoff E, Fangman MT, Fleischauer AT, Waller AE, Macdonald PD. Improvements in timeliness resulting from implementation of electronic laboratory reporting and an electronic disease surveillance system. Public Health Rep 2017;128:393-8.|
|7||Johnson MG, Williams J, Lee A, Bradley KK. Completeness and timeliness of electronic vs. Conventional laboratory reporting for communicable disease surveillance - Oklahoma, 2011. Public Health Rep 2017;129:261-6.|
|8||Overhage JM, Grannis S, McDonald CJ. A comparison of the completeness and timeliness of automated electronic laboratory reporting and spontaneous reporting of notifiable conditions. Am J Public Health 2008;98:344-50.|
|9||Stachel AG, Waechter H, Bornschlegel K, Reddy V, Hanson H, Wen T, et al. Reassessing provider reporting in the age of electronic surveillance. J Public Health Manag Pract 2017;20:240-5.|
|10||Gluskin RT, Mavinkurve M, Varma JK. Government leadership in addressing public health priorities: Strides and delays in electronic laboratory reporting in the United States. Am J Public Health 2014;104:e16-21.|
|11||Kariuki JM, Manders EJ, Richards J, Oluoch T, Kimanga D, Wanyee S, et al. Automating indicator data reporting from health facility EMR to a national aggregate data system in Kenya: An Interoperability field-test using OpenMRS and DHIS2. Online J Public Health Inform 2017;8:e188.|
|12||Hinrichsen VL, Kruskal B, O'Brien MA, Lieu TA, Platt R. Using electronic medical records to enhance detection and reporting of vaccine adverse events. J Am Med Inform Assoc 2007;14:731-5.|
|13||Buehler JW; Centers for Disease Control and Prevention. CDC's vision for public health surveillance in the 21st century. Introduction. MMWR Suppl 2012;61:1-2.|
|14||Boxrud D, Monson T, Stiles T, Besser J. The role, challenges, and support of pulsenet laboratories in detecting foodborne disease outbreaks. Public Health Rep 2010;125 Suppl 2:57-62.|
|15||Li W, Lu S, Cui Z, Cui J, Zhou H, Wang Y, et al. PulseNet China, a model for future laboratory-based bacterial infectious disease surveillance in China. Front Med 2012;6:366-75.|
|16||WHO,Early Warning Systems. WHO; 2015. Available from: http://www.who.int/csr/labepidemiology/projects/earlywarnsystem/en/. [Last accessed on 2015 Jul 24; 13:23:51].|
|17||Pinto A, Saeed M, El Sakka H, Rashford A, Colombo A, Valenciano M, et al. Setting up an early warning system for epidemic-prone diseases in Darfur: A participative approach. Disasters 2005;29:310-22.|
|18||Valenciano M, Bergeri I, Jankovic D, Milic N, Parlic M, Coulombier D. Strengthening early warning function of surveillance in the Republic of Serbia: Lessons learned after a year of implementation. Euro Surveill 2004;9:24-6.|
|19||Valenciano M, Coulombier D, Lopes Cardozo B, Colombo A, Alla MJ, Samson S, et al. Challenges for communicable disease surveillance and control in Southern Iraq, April-June 2003. JAMA 2003;290:654-8.|
|20||Ghosn N, Nasredine A, Baddour YM, Coulombier D, Nasserdine S. Electronic surveillance of outbreaks in Lebanon. BMC Proc 2008;2 Suppl 3:S2.|
|21||Mostafavi E, Chinikar S, Bokaei S, Haghdoost A. Temporal modeling of Crimean-Congo hemorrhagic fever in Eastern Iran. Int J Infect Dis 2013;17:e524-8.|
|22||Tokars JI, Burkom H, Xing J, English R, Bloom S, Cox K, et al. Enhancing time-series detection algorithms for automated biosurveillance. Emerg Infect Dis 2009;15:533-9.|
|23||Health KA. Ontario ILI Activity Trends. Available from: http://www.mapper.kflaphi.ca/ilimapper/. [Last cited on 2017 Feb].|
|24||Lwin MO, Vijaykumar S, Fernando ON, Cheong SA, Rathnayake VS, Lim G, et al. A 21st century approach to tackling dengue: Crowdsourced surveillance, predictive mapping and tailored communication. Acta Trop 2014;130:100-7.|
|25||What We Do,UC Davis School of Veterinary Medicine,One Health Institute; 2017. Available from: http://www.vetmed.ucdavis.edu/ohi/predict/predict-activities/index.cfm. [Last accessed on 2017 Feb].|
|26||Chan EH, Brewer TF, Madoff LC, Pollack MP, Sonricker AL, Keller M, et al. Global capacity for emerging infectious disease detection. Proc Natl Acad Sci U S A 2010;107:21701-6.|
|27||Christakis NA, Fowler JH. Social network sensors for early detection of contagious outbreaks. PLoS One 2010;5:e12948.|
|28||Eysenbach G. Infodemiology and infoveillance: Framework for an emerging set of public health informatics methods to analyze search, communication and publication behavior on the Internet. J Med Internet Res 2009;11:e11.|
|29||Choi J, Cho Y, Shim E, Woo H. Web-based infectious disease surveillance systems and public health perspectives: A systematic review. BMC Public Health 2016;16:1238.|
|30||Lyon A, Nunn M, Grossel G, Burgman M. Comparison of web-based biosecurity intelligence systems: BioCaster, EpiSPIDER and HealthMap. Transbound Emerg Dis 2012;59:223-32.|
|31||Khan AS, Fleischauer A, Casani J, Groseclose SL. The next public health revolution: Public health information fusion and social networks. Am J Public Health 2010;100:1237-42.|
|32||Shaman J, Karspeck A. Forecasting seasonal outbreaks of influenza. Proc Natl Acad Sci U S A 2012;109:20425-30.|
|33||Anderson NG, Pollack J, Williams D. The value of healthcare data in ophthalmology. Curr Opin Ophthalmol 2014;25:191-4.|
|34||Mykhalovskiy E, Weir L. The global public health intelligence network and early warning outbreak detection: A Canadian contribution to global public health. Can J Public Health 2006;97:42-4.|
|35||Polgreen PM, Chen Y, Pennock DM, Nelson FD. Using internet searches for influenza surveillance. Clin Infect Dis 2008;47:1443-8.|
|36||Hulth A, Rydevik G. GET WELL: An automated surveillance system for gaining new epidemiological knowledge. BMC Public Health 2011;11:252.|
|37||Wójcik OP, Brownstein JS, Chunara R, Johansson MA. Public health for the people: Participatory infectious disease surveillance in the digital age. Emerg Themes Epidemiol 2014;11:7.|