Update on Public Health Surveillance in Emergency Departments
Shawn M. Varney, Lt Col, USAF, MCa,*, Jon Mark Hirshon, MD, MPHb,c
a59 MDW/MCED, 2200 Bergquist Drive, Suite 1, Lackland AFB, TX 78236-5500, USA
bDivision of Emergency Medicine, Department of Emergency Medicine and Department of Epidemiology and Preventive Medicine,
University of Maryland School of Medicine, Baltimore, MD, USA cThe Charles McC. Mathias, Jr. National Study Center for Trauma and EMS,
University of Maryland School of Medicine, 701 West Pratt Street, Fifth Floor, Baltimore, MD 21201, USA
The systematic collection and analysis of health data are important actions required to help understand the health needs of a population. When it is done to investigate a problem to contribute to generalizable knowledge, it is defined as research [1]. If these activities are done through the collection of health data in an ongoing manner to influence the health of the public, it can be considered public health surveillance. Considering that in 2003 there were an estimated 113.9 million emergency department (ED) visits nationwide [2], EDs are an ideal location to col- lect de-identified information on the acute health needs and patterns of the population of the United States. The systematic collection of data from multiple EDs can also serve as a barometer of the overall status of the US health system. While there are a number of logistical and in- frastructural barriers that can impede the development of surveillance systems, the potential benefits from these systems are significant. The ability to analyze data; distribute results; and influence policy, funding, and patients’ behavior are important outgrowths of public health surveil- lance in emergency departments.
* Corresponding author. Department of Emergency Medicine, 59 MDW/MCED, 2200 Bergquist Drive, Suite 1, Lackland AFB, TX 78236-5500.
E-mail address: shawn.varney@lackland.af.mil (S.M. Varney).
0733-8627/06/$ – see front matter ! 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.emc.2006.06.004 emed.theclinics.com
Emerg Med Clin N Am 24 (2006) 1035–1052
mailto:shawn.varney@lackland.af.mil
What is public health surveillance?
Definition of surveillance
The Centers for Disease Control and Prevention (CDC) has defined pub- lic health surveillance as ‘‘the ongoing systematic collection, analysis and interpretation of health data essential to the planning, implementation and evaluation of public health practices, closely integrated with the timely dissemination of these data to those who need to know. The final link in the surveillance chain is the application of these data to prevention and control’’ [3]. Surveillance systems are used to prepare, execute, and assess public health intervention programs and relay the acquired information to decision makers. In the present age of heightened security awareness and threats of bioterrorism, surveillance systems play an additional role in the early detec- tion of health use anomalies. Through the rapid recognition of multiple patients with similar symptoms suggestive of an atypical or biologic agent, alerts are triggered so that public health professionals are notified of a poten- tial threat.
Surveillance system components
Surveillance systems may range from rudimentary to complexdie, from manual collection and documentation on sheets of paper to automated real- time data delivery. The steps required for a public health surveillance system include data acquisition on a periodic and ongoing basis, timely data colla- tion and analysis, and the application of these data by the proper public health professionals. The basic components of a surveillance system include equipment, personnel, and the required resources for the personnel to analyze the data, communicate promptly and effectively, and maintain the system adequately.
The ability to amass and analyze large amounts of information has mark- edly improved with the advent of current computer technology. Therefore, essential equipment for an ED-based public health surveillance system now includes a robust computerized database system with appropriate Internet and networking capabilities, along with sophisticated software to analyze data for areas of interest. The potential applications of data and the require- ments for interoperability with collaborators, such as regional, state, or national systems, dictate the necessary degree of complexity.
Fundamental personnel consist of individuals responsible for (1) data col- lection, (2) information analysis, and (3) timely response to material col- lected. Thus, many partners are involved, including health care providers in physicians’ offices and EDs and public health professionals in local, state, and federal agencies, as well as laboratory workers, researchers, academi- cians, and information technology (IT) experts. The ability to maintain mul- tidirectional communication flow among these team members is critical for a functional system.
1036 VARNEY & HIRSHON
Required additional resources include financial, institutional, and IT (encompassing communication, data management, and data analysis). To be effective, surveillance system development requires full endorsement and involvement from interested public health, political, and private leaders in many fields. Data sources may include standardized clinical databases from hospitals, doctors’ offices, EDs, pharmacies, telephone health lines, and others. The integration of these databases into a cohesive system requires significant time and effort to garner support of critical partners and to make the system fully operational.
Definition of syndromic surveillance
Syndromic surveillance describes a dynamic process of collecting real-time or near real-time data on symptom clusters suggestive of a biological disease outbreak. Ideally, these diseases will be detected early in the processdbefore the definitive diagnosisdto enable a rapid response and mitigate adverse outcomes [4,5]. Syndromic surveillance systems have secondary objectives including determining the size, spread, and tempo of an outbreak, or even providing reassurance that an outbreak has not occurred [4].
Initially, syndromic surveillance systems were designed for the early detection of biological terrorism agents. The focus has evolved subsequent to the 9/11 World Trade Center and anthrax terrorist attacks of 2001. Pres- ent emphasis lies on the timely collection, assimilation, and analysis of health care data gathered from existing community systems to provide immediate feedback to decision makers about unexpected disease clusters or sentinel cases [4].
In contrast to the standard diagnosis-based disease surveillance (labs and cultures), syndromic surveillance is prediagnosticdie, it recognizes a cluster of symptoms, or the onset of a disease, before full-blown illness manifesta- tion. Identifying a peak of unusual symptoms above the background/steady state may allow a few extra days for further observation, evaluation, and treatment before the severe illness becomes apparent by conventional diag- nostic methods. Theoretically, early detection equates to earlier treatment and decreased morbidity and mortality.
Syndromic surveillance systems tend to derive their data from two sources: (1) clinical data from health care services (ED visits, clinic visits, or Emergency Medical Services [EMS] records), and (2) alternative sources (work or school absentee rates, pharmaceutical sales, calls to emergency or information hotlines, Internet-based illness reporting systems) [6]. Each data source has advantages. For example, clinical data sources provide the ability to follow patients and, in the case of a public health emergency, to contact infected individuals. These actions, however, would require significant efforts and high-level approvals to override existing privacy and confidenti- ality safeguards. In addition, clinical data encourages bidirectional commu- nication and fosters improved relationships between community providers
1037UPDATE ON PUBLIC HEALTH SURVEILLANCE IN THE ED
and public health staff, which is an important step in a functional public health system. Alternative data sources, such as pharmacy sales including over-the-counter products, may signal the occurrence of events before people seek formal health care and may represent a broader sample of the population at risk.
In one study of 3919 ED visits, Begier and colleagues [5] found good overall agreement (kappa ¼ 0.639) between chief complaint and discharge diagnosis, but substantial variability by specific syndromes. All ED patient encounters were coded via a mutually exclusive algorithm into one of eight syndromes: death, sepsis, rash, respiratory illness, gastrointestinal illness, unspecified infection, neurologic illness, and other. They observed lower agreement among sepsis, neurologic, and unspecified infection. Begier and colleagues concluded that although there is good agreement for most syndromes, the chief complaint better identifies illnesses with nonspecific symptoms (ie, fever), while discharge diagnoses detect illnesses requiring clinical evaluation (ie, sepsis and meningitis).
Another form of syndromic surveillance is ‘‘event-based’’ or ‘‘drop-in’’ surveillance, which lasts for a finite period or event. It relies on health care providers in EDs and large clinics to collect nonroutine data. Such a sys- tem was implemented and proved useful during the 2000 Democratic National Convention in California and the 2002 Winter Olympic Games in Utah [7,8].
Although syndromic surveillance may be able to play a key role in early recognition of disease outbreaks, it neither replaces traditional public health surveillance nor supplants the critical role of an astute physician reporting atypical diseases and events.
Why is surveillance important?
General rationales for ED-based public health surveillance
There are a number of rationales for the development of public health surveillance based on ED visits [9]. These include:
1. Improved communication between health departments and emergency departments for addressing ongoing local, regional, and state-level problems.
2. Improved public health response to rapidly developing public health emergencies.
3. Improved ability to correlate environmental events and visits. 4. Improved information on the scope and nature of ED visits for injuries
(both minor and major). 5. Improved documentation and evaluation of ED visits for infectious
diseases. 6. Improved hospital-based patient record systems.
1038 VARNEY & HIRSHON
7. Influence policy discussions and decisions through improved data.
These rationales can be conceptually divided into those designed to im- prove the health of the public and those designed to improve the security of the population.
Improving public health
Surveillance is an outcome-oriented science that provides information for action. Public health surveillance focuses on health-related issues or their preceding events. It plays a key role in protecting the public by devising ways to improve health and to mitigate morbidity and mortality. In the con- text of public health, Teutsch and Churchill [10] described multiple ways that surveillance data are useful: to estimate the magnitude of a health prob- lem; to understand the natural history of a disease or injury; to detect outbreaks or epidemics; to document the distribution and spread of a health event; to test hypotheses about etiology; to evaluate control strategies; to monitor changes in infectious agents; to monitor isolation activities; to detect changes in health practice; to identify research needs and facilitate epidemiologic and laboratory research; and to facilitate planning.
Surveillance allows for the monitoring and evaluation of the health of the public. However, it is critical that appropriate public health professionals then translate the information garnered from these efforts into action. A feedback loop is thus developed to produce positive effects within the monitored population. Timely and accurate health-related data, properly collected and analyzed, allow public health leaders, politicians, and others to act appropriately to mitigate disasters or epidemics through judicious allocation of suitable resources.
A current example of ongoing surveillance of a potential public health threat is the actions by national governments in Asia, the CDC, and the World Health Organization (WHO) to monitor the current status of avian influenza (bird flu), especially the influenza A (H5N1) virus [11]. While H5N1 primarily affects fowl, there is concern for the potential person- to-person transmission of the virus leading to a pandemic. Thus the CDC has recommended enhanced surveillance for this disease in the United States to promote its rapid diagnosis and to prevent its dissemination. If bird flu were discovered in a patient in the United States, the CDC could rapidly mobilize resources to limit the spread of infection and panic among the population.
Terrorism response/homeland security
According to the Advisory Panel to Assess Domestic Response Capabil- ities for Terrorism Involving Weapons of Mass Destruction, ‘‘a robust public health system is fundamental to a long-term solution for a variety of health issues, including terrorism’’ [12]. Public health surveillance
1039UPDATE ON PUBLIC HEALTH SURVEILLANCE IN THE ED
systems, such as those based on ED visits, are part of this solution. While it is difficult to assess the magnitude of the threat, there is no question that all societies are at risk from conventional explosives and, potentially, from weapons of mass destruction. Within this global context of increased insecu- rity, it is important to be able to detect unusual diseases and events. The ongoing, systematic collection of ED data to identify unusual diseases and patterns may help shorten the time required to respond to biological or chemical attacks and thus decrease the morbidity and mortality from these weapons. Recognition on the national level can be seen by the increased federal dollars allocated to public health, much of which has been used for increased disease surveillance and response. Additionally, a number of projects focusing on syndromic surveillance, such as the Electronic Surveil- lance System for the Early Notification of Community-Based Epidemics (ESSENCE), were developed or tested through funding from the Defense Advanced Research Project Agency (DARPA) and the Department of Defense [13].
Stakeholders in developing surveillance systems
Health care facilities
The ED plays a key role in the development and use of a public health surveillance system. Patients come into EDs 24 hours a day, 7 days a week, every day of the year, making it an appropriate place for data gath- ering and collation. Health care providers in the ED simultaneously see mul- tiple patients and often have high daily patient volumes. This enables the derivation of the relative prevalence of symptom clusters that may represent worrisome syndromes or epidemics. Outlying clinics frequently refer sicker patients to local EDs, facilitating collection of information on more cases. Emergency physicians are taught to have a high index of suspicion for uncommon diseases, leading to broad differential diagnoses and clinical acumen. They are the first physician contacts for patients in many situations and may detect aberrations in the usual incidence of disease. From these frontline positions, they need to be able to transmit their findings and con- cerns in a timely and accurate manner to the appropriate public health authority. As a primary participant in the disease recognition process, emergency physicians and other ED staff must be involved in surveillance system development.
The information collecting process should be simple, quick, and easy to implement with minimal or no impact on health care practitioners. Auto- matic classification of broad symptom categories for chief complaints can be included as a part of triage. Alternatively, a computer can be placed in a kiosk by the registration desk in the ED. Simple questions may identify symptom clusters that the computer can analyze at regular intervals and produce warnings or alerts to hospital personnel or public health agencies.
1040 VARNEY & HIRSHON
ED personnel end up participating in the surveillance process to some degree whether they realize it or not. Simply observing patients and assim- ilating and documenting information (gathering chief complaints, identify- ing trends, and so forth) contributes. Passing the data to the public health sector may mitigate morbidity and mortality. Automatic data entry from multiple hospitals into a centralized repository may facilitate disease recog- nition and coordinate findings citywide, similar to a well-run emergency medical services system. Ideally, a large funding source, such as state and federal governments, should support this initiative in the interest of the public’s health.
Public health agencies
Public health agencies and their staff play a pivotal role in monitoring and managing the public’s health, from scrutinizing for disease outbreaks to implementing quarantine measures. They function as the keystone of a public health surveillance system and their involvement in system develop- ment and use is crucial. While EDs and other data sources, such as labora- tory personnel and pharmacists, supply the input, public health professionals must accept the collected data, analyze it, and then return rec- ommendations and policy actions to appropriate officials. Timely reporting is critical to allow public health professionals to perform their jobs.
As part of this involvement, bidirectional communication is vital be- tween frontline providers, such as emergency physicians, and public health experts. While it is important that accurate information be sent to the health department in a timely manner, it is of equal significance that in- formed and authoritative health messages be disseminated to both health care professionals and to the public. The information received by emergency physicians and other practitioners influences the evaluation and treatment of patients. Public health messages can assist in the effective management of the behavior and responses of the community at large, especially in times of crisis.
Of additional consequence in this partnership between health care and public health is the understanding that system development requires the support, financial and otherwise, of health departments and public health professionals. An individual ED is not a surveillance system, although it may function as a monitoring station within one. A public health surveil- lance system based on ED visits, as well as other potential data sources, requires significant infrastructural support to receive large amounts of health-related data and then to rapidly analyze it for unusual patterns or increased disease frequency.
Information technology
With the increased ability to rapidly collect and analyze data from multiple sources, the involvement and support of experts in information
1041UPDATE ON PUBLIC HEALTH SURVEILLANCE IN THE ED
technology are important aspects of the team effort to develop a functional public health surveillance system. In general, data are not transmitted as a continuous stream, but rather at periodic intervals (eg, hourly, daily, weekly). Data can be collected and analyzed manually, but the greater the automation, the more rapid and accurate the results are likely to be. Automation can enhance the data collection and analysis process, mini- mizing delays and decreasing inaccuracy caused by the need to depend on human interactions. Through the use of software that automatically collects the number of visits (or other data parameter) by category, the amount of effort required by health care providers in data input can be significantly decreased. Advanced logic algorithms can help look for un- usual trends through analysis of the data from multiple sources and can be instructed to alert when specific patterns are noted. While these pro- cesses can decrease the daily effort required of health care and public health professionals, individuals knowledgeable about the appropriate software and hardware are required for a smoothly functioning, integrated system.
Surveillance implementation
Health data standards and timeliness
An ideal public health surveillance system would be interoperable, uni- versal, automated, real-time, economical, secure, sensitive, and specific. To date, information technology (IT) developers have not created a product to satisfy these parameters. To enhance interoperability between different current systems and between existing and future systems, certain informa- tion system standards have been identified. In addition, information systems supported by government funds must comply with federally mandated standards.
Broome and Loonsk [14] discussed three vital justifications for stan- dards-based system development: (1) electronic messaging (ie, Standard Health Level 7, or HL7, interface) provides the most effective and efficient way to collect real-time data from multiple sources; (2) specified standards provide public health departments greater control over previous invest- ments in their IT infrastructures; and (3) standard formats and electronic data delivery reduce the burden on individual providers’ reporting practices.
Multiple government agencies have identified important standards integral to improved information exchange between clinicians and health departments [14]. The CDC and its state and local delegates formed the Public Health Information Network that identified standards for data, technology, terminology, and confidentiality. This network named five major functional areas (detection and monitoring, data analysis, knowl- edge management, alerting, and response) and itemized specifications
1042 VARNEY & HIRSHON
for nine IT functions that form the basis for interoperable standards- based systems:
1. automated data exchange between public health partners; 2. use of electronic clinical data for event detection; 3. manual data entry for event detection and management; 4. specimen and laboratory result information management and exchange; 5. management of possible case, contacts, and threat data; 6. analysis and visualization; 7. directories of public health and clinical personnel; 8. public health information dissemination and alerting; and 9. IT security and critical infrastructure protection [15].
At the request of the CDC Information Council, the Gartner Group, an independent IT consulting firm, reviewed the Public Health Information Network’s specifications and functions and endorsed them as the ‘‘founda- tional road map’’ for systems integration in public health [16].
Timeliness related to surveillance systems impacts all aspects of the pro- cess from data collection, through data transfer and analysis, to returning treatment and policy recommendations. These criteria are ranked among the most important and most often described in published reports [17,18]. The ability to react quickly to public health emergencies depends on rapid recognition and response to possible or actual threats, which is the core issue of timeliness.
Data collection
Two prevailing data-gathering principles in public health surveillance are (1) collect information judiciously, and (2) gather and retain information as locally as possible [6]. Both principles facilitate compliance with the Health Insurance Portability and Accountability Act (HIPAA) of 1996 and also help limit the amount of labor involved in the data input phase. From a pragmatic perspective, it is important to limit the amount of data collec- tion effort required by frontline providers, to achieve high levels of compli- ance and data fidelity without impacting providers’ ability to care for patients.
While most hospitals do not have real-time or near-real-time surveillance systems, some have adapted current systems to achieve this objective. For example, in Hong Kong the hospital authority developed an ED computer system used across the region. For 2 years (1999–2000) they gathered data on common diseases, namely upper respiratory infections and gastro- intestinal illnesses, and followed trends and seasonal peaks. They tracked diagnoses, prescriptions, specialty information, and patient demographics monthly. When peaks exceeded two standard deviations of variance, a com- puter-generated report was sent to the ED director and hospital authority officials for appropriate intervention [19]. Noting the unexpected infectious
1043UPDATE ON PUBLIC HEALTH SURVEILLANCE IN THE ED
disease surges, the health authorities alerted the media and educated the public. This may have helped curtail disease transmission.
The simple step of compiling computerized ED records of patient volume, chief complaints, and diagnoses, along with applying a standard statistical program, forms the first step in disease surveillance. Providers, in general, do not prioritize nonclinical responsibilities during clinical hours. In a busy ED, an emergency physician will need to see direct patient care benefit from data collection, otherwise data acquisition will be inconsistent. Much of the data collation can occur in an environment away from the clin- ical area thereby limiting the impact on the health care providers.
Data transfer
As discussed above, standards are important to ensure timely and accu- rate transfer of data. Considering the current state of computer technology, electronic data transfer best meets these needs. In addition, data security, such as encryption, is of fundamental importance, especially when consider- ing the increased responsibility of covered entities to securely protect the confidentiality of personal health information. While HIPAA allows for exemptions concerning the use of data for public health purposes, surveil- lance systems and related stakeholders would have difficulty withstanding public scrutiny if data were mishandled or inappropriately released.
Data analysis
Rapid, accurate analysis of the data is important to develop appropriate and timely policy recommendation. It is the critical step in turning large amounts of seemingly unrelated data into coherent information, and subse- quently into action. Individuals may analyze data manually with statistical programs or using automated algorithmic processes. In most instances it will be a combination of both modalities. Automatic algorithms can greatly enhance the speed of analysis and produce predefined alerts, but will still require interpretation and monitoring by those with an in-depth knowledge of the surveillance system.
While there are many similarities between systems designed to collect health data, sources of information vary. Data analysis solutions require modifications for specific circumstances. For example, the detection of an abnormal increase in a disease is dependent on the definition of the baseline incidence of that disease, as seen by syndromic surveillance for flu-like illnesses. The number of cases that would be considered abnormal will be very different in the winter months during the ‘‘flu season,’’ as opposed to the summer months when influenza is unlikely. One solution to this problem is to use a progressive baseline derived from the number of flu-like cases in the previous 2 weeks. Thus, when influenza spreads through the community in late fall/early winter, the system would initially produce alerts based on
1044 VARNEY & HIRSHON
a predetermined variance from baseline but would quickly develop a modi- fied baseline that would be appropriate for a season of increased cases. This baseline would then decrease as the number of flu-like illnesses drop in late winter/early spring leading to an appropriate baseline for a low-incidence season.
Another aspect of data analysis involves investigating specific alerts. For example, signal investigation plays a key role in outbreak detection. Steiner- Sichel and colleagues [20] described their experience with the New York City Department of Health and Mental Hygiene (DOHMH), which has operated a syndromic surveillance system based on ED chief complaints since November 2001. The DOHMH conducted field investigations of suspected outbreaks when the surveillance systems signaled an unexpected increase/ excess above the expected rates for respiratory, fever, diarrhea, and vomit- ing syndromes. They sought to determine if the signals correlated with clinically significant disease outbreaks. In more than 40 signal investiga- tions, none definitively detected an infectious disease outbreak. They also found that none of the localized outbreaks investigated by the traditional methods revealed any syndromic surveillance signal. Steiner-Sichel and colleagues attributed this to the difficulty of proving causality and using a sensitive, but not specific, detection system. The advantage of early detection may be offset by the complexities of field investigation and epide- miologic data acquisition.
At the present time a number of issues need to be addressed to improve data analysis, particularly as it relates to syndromic surveillance. These issues include how to best analyze data from multiple data streams [21,22], improve the linkage of data from different data sources [23], and create flexible space-time shapes in the analysis of disease clusters [24]. While a great deal of energy and resources have been spent to improve public health surveillance, especially as it relates to syndromic surveillance, further work is clearly necessary.
Use and misuse of data
Although there are clear public health and public safety aspects to the use of aggregated health-related data, the potential misuse of data is of signifi- cant concern. Misuse and abuse may come in many forms. The ability to con- tact trace individuals in case of a highly transmissible and deadly infectious disease or a bioterrorism event is critical to decreasing the potential morbid- ity and mortality. On the other hand, sufficient safeguards must be in place to prevent the inadvertent or malicious release of personal information. HIPAA attempts to address many of the issues related to the use and sharing of individual health records, especially as it is collected from clinical encounters, and mandates the appropriate handling of this personal information.
One way to address the conflict between public health and patient privacy is by releasing only de-identified data to the public health agency collecting
1045UPDATE ON PUBLIC HEALTH SURVEILLANCE IN THE ED
the data, thus making inadvertent tracing much less likely. If specific infor- mation is needed to prevent a public health emergency, then the appropriate individuals with the proper legal authority could request the specific identifi- able information from the data-collecting site, such as the hospital. This multiple-step process may slow the evaluation and possible response by public health officials, but this must be balanced with the need to protect in- dividuals’ privacy. The complex interplay between the health needs of the general public and an individual’s rights and privacy is placed within an intricate legal setting and leads to one of the great dynamics of public health, namely balancing human rights and public safety.
Barriers to surveillance systems development
General barriers to ED-based public health surveillance
There are a number of barriers to the development of public health surveillance based on ED visits [9]. These include the following:
1. Costs for public health agencies 2. Costs for emergency departments and hospitals 3. Need to improve and standardize data collection 4. Security and confidentiality issues 5. Obtaining acceptance and support from emergency medicine leadership
and practitioners
These can be conceptually divided into funding issues, data-related issues, and the need to obtain acceptance and support from key partners.
Funding
Development of the public health surveillance infrastructure requires significant financial investment, especially by public health authorities. Whereas governments can mandate certain actions (especially on the part of large entities such as hospitals), institutions will resist actions that adversely impact their financial status. Since the end users of these systems are public health authorities and ultimately the public, it is the responsibility of the government to bear a significant burden of the cost.
Since the terrorist attacks of September 11, 2001, and the subsequent an- thrax letters, a significant amount of federal dollars has gone to public health agencies, especially at the local jurisdiction. Much of this money was designated for terrorism response activities, including improving sur- veillance and communications. Despite these large sums of money, there is considerable variance in current public health surveillance infrastructure throughout the United States. Additional funding is needed to continue to improve and standardize public health surveillance activitiesdespecially syndromic surveillance. These resources will need to be shared between
1046 VARNEY & HIRSHON
the data collection entities and those entities analyzing the data and producing public health responses.
Sharing of data
The data that routine surveillance systems collect differ from syndromic surveillance data in that the former are based on diagnostic or culture- positive diseases, whereas the latter are founded in prediagnostic, or clusters of symptoms suggesting potentially infectious disease outbreaks. Health information privacy rules such as HIPAA may apply differently to routine and syndromic surveillance data. The prevailing feeling among some physi- cians is that reporting and investigating patients with culture-positive diseases do not violate patient privacy, whereas inconclusive disease pro- cesses are not certain enough to warrant full disclosure of patient privacy information for further contact [25]. Data collection for syndromic surveil- lance requires the ability to identify and contact individual patients when a surge in unusual symptoms (signal) occurs.
In a survey sent to state epidemiologists and terrorism preparedness coordinators regarding the effects of HIPAA Privacy Rule requirements on syndromic surveillance system implementation, Drociuk and colleagues [25] found that more than half reported ‘‘some’’ or ‘‘substantial’’ problems. HIPAA’s ‘‘minimum necessary’’ stipulation thwarted disease surveillance activities. The ‘‘minimum necessary’’ standard states that health care pro- viders must take reasonable steps to limit the use or disclosure of protected health information (PHI) to the minimum necessary to accomplish the in- tended purpose [26]. Covered entities (ie, all health care organizations) have the flexibility to make their own assessment of what PHI is reasonably necessary for a particular purpose. Unfortunately, there is no broadly ac- cepted definition for ‘‘minimum necessary’’ in either routine or syndromic surveillance systems.
As noted above, a proper balance must exist between protecting personal health information and the need to protect the general public health. The HIPAA Privacy Rule permits PHI disclosures without individual authoriza- tion to public health agents and designees when intended to prevent or con- trol disease, injury, or disability, including public health surveillance, investigation, and intervention [27]. One solution to satisfy patient confiden- tiality concerns is collecting limited data sets, ie, information that is not directly identifiable. Specific data use agreements must establish who is per- mitted to use the data. The benefit is fewer problems with HIPAA and potentially better participation from surveillance institutions, but the draw- back includes delayed signal investigations. The delays may significantly counter the potential theoretical advantage of early outbreak detection by syndromic surveillance.
Regarding data transfer, 27/32 (87%) respondents reported no security concerns because of the secure transmission measures and off-system
1047UPDATE ON PUBLIC HEALTH SURVEILLANCE IN THE ED
data-archiving protocols [25]. Despite the problems with patient confidenti- ality and data transfer, physicians felt more secure and ready for potential terrorist attacks. Furthermore, the mere fact that surveillance systems exist may serve as deterrence against terrorist strikes since the community may appear poised and ready to act.
Buy-in from collaborators
There are two main groups that are important for a functional public health surveillance system based on ED data: the hospitals (ie, the data sour- ces) and the public health departments (ie, data analyzers and users) [9]. Both groups should be actively involved in the creation and deployment of the final system, since ongoing bidirectional communication and inter- agency cooperation are critical. The development of the relationship between the hospitals and the health department are as important as the final system structure, since public health professionals need the data to help make informed policy decisions and action recommendations, and medical professionals, such as emergency physicians, may need to imple- ment these recommendations.
Obtaining this buy-in requires a commitment from both sides. Coopera- tion is developed through working together and developing a shared vision. There needs to be mutual understanding of the goals and expectations for the system and the roles that each participating organization will play. There are a number of potential barriers to the creation of a relationship, including the costs involved for both the data collection and the data analysis. How- ever, a clearer understanding of the importance of the public health–medical collaborations has grown over the past 5 years with the increased awareness of the risks of disease spread, such as from avian influenza, and the potential for bioterrorism.
Criteria for evaluating a surveillance system
Buehler [28], from the CDC 2003 Working Group on Public Health Sur- veillance Systems, described a comprehensive framework of four categories for evaluating all public health surveillance systems: system description, out- break detection, experience, and conclusions and recommendations.
In summary, the system description should clearly state the system’s purpose, including indications for its use, duration, area of emphasis, and the desired sensitivity and specificity. It should identify the stakeholders, meaning those supplying the data and applying the information. Finally, it should provide a detailed description of all operational aspects including data flow, data sources, data processing before analysis, statistical analysis, and epidemiological analysis and interpretation [25].
The second category is outbreak detection and discusses factors affecting timely data gathering and processing, data validity, and data
1048 VARNEY & HIRSHON
aberrancy–detection methods. Timeliness describes a continuum from the onset of symptoms to public health intervention. Establishing the validity of a surveillance system to detect an outbreak requires epidemiological tools like outbreak and case definitions, statistical analysis, and assessment of the data quality.
The third area for system evaluation is documented experience with the system. Important factors for system experience include the following: use- fulness (impact of its application), flexibility (ability to adapt easily to changing needs and new technology), acceptability (willingness of parties to submit timely and complete data, widespread use), portability (ease of reproduction in other centers), stability (minimal downtime and maximal consistency), and cost (for software and support as well as for false alarms and failed detection).
The final system evaluation category is a summary of the conclusions and recommendations of the advantages and disadvantages of each system. A useful approach would include possible modifications of present systems to meet the increasing needs in public health surveillance.
Bravata and colleagues performed a systematic review on surveillance systems for the early detection of bioterrorism-related diseases [15,29]. After reviewing more than 17,000 article citations and 8000 web sites, they found 192 reports on 115 systems that gathered surveillance data, including nine syndromic surveillance systems. Bravata and colleagues evaluated the sys- tems for reports on nine qualities the CDC had defined previously: useful- ness, importance, timeliness, flexibility, sensitivity, representativeness, simplicity, acceptability, and specificity [30–32]. Only one article addressed all nine criteria [16]. Usefulness, importance, and timeliness were most com- monly described, whereas only three reports of three systems provided actual values for sensitivity and specificity [15]. Clearly, there is little scien- tific evidence supporting the use of surveillance systems. Future studies are needed to evaluate present and new systems for these nine characteristics of effective surveillance systems. Three of the most important criteria appear to be sensitivity and specificity, timeliness, and the ability of the system to impact decision making.
Sosin and DeThomasis [33], members of the CDC 2003 Working Group on Public Health Surveillance Systems, summarized the group’s findings by developing a task list of specific, goal-directed questions for early outbreak detection. Sosin and DeThomasis reviewed 99 abstracts presented at the 2003 National Syndromic Surveillance conference and found limited infor- mation on system evaluation. Because a detailed analysis of systems would likely be laborious and expensive, Sosin and DeThomasis proposed empha- sizing timeliness, validity, and usefulness to measure the success of detection methods.
Criteria for evaluating surveillance systems are more complex and diffi- cult to assess than originally conceived. Despite rigorous descriptions and defined criteria, few researchers have produced data following the CDC’s
1049UPDATE ON PUBLIC HEALTH SURVEILLANCE IN THE ED
recommended framework. Perhaps a simplified approach reviewing only timeliness, validity, and usefulness may show the impact and cost of detec- tion methods.
Summary
The development of public health surveillance systems based on ED visits, in conjunction with other health and nonhealth-related data, is an im- portant step to better understanding the health needs of the US population. There are multiple steps required to develop a functional organization, and these actions require the support and involvement of many different part- ners. In any given jurisdiction a number of obstacles to structure develop- ment may exist and will require teamwork to overcome. Yet, the information derived from these systems on the acute health needs and health care usage patterns of the US population can help both to improve the health of the public and to serve as an early warning system for a possible bioterrorism event. Whereas surveillance systems can serve many important functions, it is also critical to maintain the privacy and confidentiality of protected health information while these systems are created and used. Through the establishment of public health surveillance systems, bidirec- tional communication is developed, strengthening the relationship between clinical and public health practitioners. The ability to (1) analyze data; (2) distribute results; and (3) influence policy, funding, and patients’ behavior are important outgrowths of emergency department–based public health surveillance systems.
References
[1] 45 CFR 46.102(d). [2] McCaig LF, Burt CW. National Hospital Ambulatory Medical Care Survey: 2003
emergency department summary. Advance data from vital and health statistics; No. 358. Hyattsville, MD: National Center for Health Statistics. Available at: http://www.cdc.gov/ nchs/data/ad/ad358.pdf. Accessed July 16, 2006.
[3] Centers for Disease Control and Prevention. Comprehensive plan for epidemiologic surveil- lance. Atlanta, GA: CDC; 1986.
[4] Henning KJ. Overview of syndromic surveillance. What is syndromic surveillance?MMWR 2004;53(Suppl):5–11.
[5] Begier EM, Sockwell D, Branch LM, et al. The national capitol region’s emergency depart- ment syndromic surveillance system: do chief complaint and discharge diagnosis yield differ- ent results? Emerg Infect Dis [serial online] 2003 Mar. Available at: http://www.cdc.gov/ ncidod/EID/vol9no3/02-0363.htm. Accessed March 7, 2005.
[6] Buehler JW. Review of the 2003 National Syndromic Surveillance Conferencedlessons learned and questions to be answered. MMWR 2004;53(Suppl):18–22.
[7] County of Los Angeles, Department of Health Services, Acute Communicable Disease Control. Special studies report 2000: Democratic National Conventiondbioterrorism syndromic surveillance. Los Angeles: County of Los Angeles Department of Health Services, 2000. Available at: http://www.lapublichealth.org/acd/reports/spclrpts/spcrpt00/ demonatconvtn00.pdf. Accessed July 16, 2006.
1050 VARNEY & HIRSHON
http://www.lapublichealth.org/acd/reports/spclrpts/spcrpt00/demonatconvtn00.pdf
http://www.cdc.gov/nchs/data/ad/ad358.pdf
http://www.cdc.gov/nchs/data/ad/ad358.pdf
http://www.cdc.gov/ncidod/EID/vol9no3/02%130363.htm
http://www.cdc.gov/ncidod/EID/vol9no3/02%130363.htm
http://www.lapublichealth.org/acd/reports/spclrpts/spcrpt00/demonatconvtn00.pdf
[8] Gesteland PH, Wagner MM, Chapman WW, et al. Rapid deployment of an electronic disease surveillance system in the state of Utah for the 2002 Olympic Winter Games. Proc AMIA Symp 2002;285–9.
[9] Hirshon JM. The rationale for developing public health surveillance systems based on emergency department data. Acad Emerg Med 2001;7:1428–32.
[10] Teutsch SM, Churchill RE. Principles and practice of public health surveillance. 2nd ed. Oxford, NY: Oxford University Press; 2000.
[11] Centers for Disease Control and Prevention: Key Facts About Avian Influenza (Bird Flu) and Avian Influenza A (H5N1) Virus. Available at: http://www.cdc.gov/flu/avian/gen-info/ facts.htm. Accessed August 28, 2005.
[12] Third annual report to the president and the Congress of the advisory panel to assess domes- tic response capabilities for terrorism involving weapons of mass destruction. Dec 15, 2001. p. 25. Available at: http://www.rand.org/nsrd/terrpanel/terror3-screen.pdf. Accessed July 16, 2006.
[13] BurkomHS, Elbert Y, Feldman A, Lin J. Role of data aggregation in biosurveillance detec- tion strategies with applications from ESSENCE. MMWR 2004;53(Suppl):67–73.
[14] Broome CV, Loonsk J. Public health information networkdimproving early detection by using a standards-based approach to connecting public health and clinical medicine. MMWR 2004;53(Suppl):199–202.
[15] CDC. PublicHealth InformationNetwork standards, specifications, and functions. Atlanta, GA: US Department of Health and Human Services, CDC, 2003. Available at: http:// www.cdc.gov/phin/architecture/index.htm. Accessed.
[16] Gartner, Inc. Technical review of issues related to version 1 of the Public Health Information Network functions and specifications. Stamford, CT: Gartner, Inc.; 2003. Available at: http://wwwcdc.gov/phin/conference_presentations/05-13-03/. Accessed August 29, 2006.
[17] Bravata DM,McDonald KM, SmithWM, et al. Systematic review: surveillance systems for early detection of bioterrorism-related diseases. Ann Intern Med 2004;140:910–22.
[18] Takahashi H, Fujii H, ShindoN, et al. Evaluation of the Japanese school health surveillance system for influenza. Jpn J Infect Dis 2001;54:27–30.
[19] Chan JTK, Cameron PA. A pragmatic approach to timely disease surveillance in the emer- gency department. Emerg Med J 2003;20(Iss. 5):443–8.
[20] Steiner-Sichel L, Greenko J, Heffernan R, et al. Field investigations of emergency depart- ment syndromic surveillance signalsdNew York City. MMWR 2004;53:184–9.
[21] Burkom HS, Murphy S, Coberly J, et al. Public health monitoring tools for multiple data streams. In: Syndromic surveillance: reports from a national conference, 2004. MMWR 2005;54(Suppl):55–62.
[22] Wong W-K, Cooper G, Dash D, et al. Use of multiple data streams to conduct Bayesian biologic surveillance. In: Syndromic Surveillance: Reports from a National Conference, 2004. MMWR 2005;54(Suppl):63–9.
[23] Magruder S,Henry J, SnyderM. Linked analysis for definition of nurse advice line syndrome groups, and comparison to encounters. In: Syndromic surveillance: reports from a national conference, 2004. MMWR 2005;54(Suppl):93–7.
[24] Iyengar VS. Space-time clusters with flexible shapes. In: Syndromic surveillance: reports from a national conference, 2004. MMWR 2005;54(Suppl):71–6.
[25] Drociuk D, Gibson J, Hodge J. Health information privacy and syndromic surveillance systems. MMWR 2004;53(Suppl):221–5.
[26] United States Department of Health and Human Services. Office of Civil RightsdHIPAA. Last revised 16 Sep 04.Available at: http://www.hhs.gov/ocr/hipaa/. AccessedMarch 20, 2005.
[27] CDC. HIPAAPrivacy Rule and public health: guidance fromCDC and the USDepartment of Health and Human Services. MMWR 2003;52(Suppl):1–20.
[28] Buehler JW. Framework for evaluating public health surveillance systems for early detection of outbreaks: recommendations from the CDC Working Group. MMWR Recomm Rep 2004;53(RR-5):1–11.
1051UPDATE ON PUBLIC HEALTH SURVEILLANCE IN THE ED
http://www.hhs.gov/ocr/hipaa/
http://www.cdc.gov/flu/avian/gen-info/facts.htm
http://www.cdc.gov/flu/avian/gen-info/facts.htm
http://www.rand.org/nsrd/terrpanel/terror3-screen.pdf
http://www.cdc.gov/phin/architecture/index.htm
http://www.cdc.gov/phin/architecture/index.htm
http://www.cdc.gov/phin/conference_presentations/05-13-03/
[29] Bravata DM, Sundaram V, McDonald KM, et al. Evaluating detection and diagnostic decision support systems for bioterrorism response. Emerg Infect Dis 2004;10(1):100–8.
[30] CDC. Guidelines for evaluating surveillance systems. MMWR 1988;37(No. S-5):1–18. [31] Sosin DM.Draft framework for evaluating syndromic surveillance systems. J UrbanHealth
2003;80(2, Suppl 1):i8–13. [32] CDC. Updated guidelines for evaluating public health surveillance systems. Recommenda-
tions from the Guidelines Working Group. MMWR 2001;50(RR13):1–35. [33] Sosin DM, DeThomasis J. Evaluation challenges for syndromic surveillance: making
incremental progress. MMWR 2004;53(Suppl):125–9.
1052 VARNEY & HIRSHON
Update on Public Health Surveillance in Emergency Departments
What is public health surveillance?
Definition of surveillance
Surveillance system components
Definition of syndromic surveillance
Why is surveillance important?
General rationales for ED-based public health surveillance
Improving public health
Terrorism response/homeland security
Stakeholders in developing surveillance systems
Health care facilities
Public health agencies
Information technology
Surveillance implementation
Health data standards and timeliness
Data collection
Data transfer
Data analysis
Use and misuse of data
Barriers to surveillance systems development
General barriers to ED-based public health surveillance
Funding
Sharing of data
Buy-in from collaborators
Criteria for evaluating a surveillance system
Summary
References