The previous chapter examined existing sources of surveillance information. This chapter explores the different levels and uses of surveillance information and emerging sources of surveillance data.
The value of any surveillance system is ultimately realized through the timely and relevant information it offers and the decisions it informs and influences. To be successful, a surveillance strategy and system must engage potential decision makers and be relevant to the specific issues they confront—locally, nationally, and increasingly, globally. Information and knowledge needs vary by perspective, and resources are rarely available to support all needs. A nationwide surveillance system will involve consideration of a range of user groups. Effective patient care, health services management, public health, and health policy making is dependent upon reliable information about the individuals and populations being served and upon scientific knowledge about the health conditions being addressed and their effective delivery. This alignment of information has become increasingly critical with the implementation of healthcare reform and also with the growing global influence on national and local health.
The uses and impacts of systematic surveillance on decision making are a function of position in the overall health system. For example, a primary care or specialist clinician concerned with patient care will be influenced by clinical guidance from professional organizations and public agencies that is derived from longitudinal population analyses which are often focused on condition-defined subpopulations (e.g., patients with heart failure). A public health expert will look to surveillance data to understand disease burdens and their trends, to identify risk factors to intervene upon, and to monitor improvements in population health status.
A manager in a healthcare delivery system who must consider a range of possible clinical initiatives and allocate resources will look to surveillance data reflective of the population served to set priorities for staffing by various health professionals, development of patient and community educational programs, investment in services and facilities, and budgeting. To plan for clinical care, a manager of a health plan or Accountable Care Organization (ACO) will need information about the prevalence of CVD or COPD rates and risk factor rates of enrollees and the extent to which they correspond with rates in their communities. This information can be useful not only in allocating resources for clinical care; it will also inform the planning of educational and behavioral interventions that have potential to reduce costs or to improve health.
Researchers require data to make inferences about relationships between key variables (e.g., between race/ethnicity and treatment and outcomes). A key product of the proposed surveillance framework will be an increased ability to match research priorities to the needs of the population for health improvement.
Finally, a policy maker, such as a state or federal health officer, will seek perspective and support for public health programs and services, as well as new regulations or legislation based on demonstrated population needs, often involving multiple geographies. For example, a great deal of national, state, and local policy is based on national surveillance data on obesity, diet, and physical activity (including the impact of those health behaviors on obesity rates). First Lady Michelle Obama has led a national campaign to reduce obesity, encourage children and adults to eat more fresh produce, and increase physical activity. Her national campaign included highly visible public education events to promote home and community gardens, as well as efforts to mobilize voluntary health agencies and food manufacturing corporations to reduce the fat content and calories of their products. Public health leaders at the state and local levels have led similar public education and policy efforts as well as efforts to change food and beverage policies of school districts.
The types of information and level of detail required will vary among users of surveillance data. For example, physicians must apply available clinical guidance in the context of the individual needs of each patient. Such personalization may be best served by the ability to relate any patient to surveillance information collected and analyzed at a highly disaggregated level. Health system managers will typically need less granular information, although the degree to which surveillance information can be directly related to the population for which care is being provided will strongly influence its use. The researcher needs information on changing and contemporary trends in disease magnitude, management, predisposing factors, and acute and more long-term outcomes, including the utilization of healthcare services in order to identify topics worthy of further investigation for prioritizing research efforts, and also to generate hypothesis or to answer questions that may be addressed with surveillance data. Policy makers must balance multiple and often complex needs and perspectives, further testing the ability of surveillance systems to provide both applicable and actionable data. Data for policy development and advocacy must be relevant to the health issue that policy will target, the factors that are believed relevant to addressing the health issue, and the geographic location and demographic characteristics of the population involved.
These user examples (not an exhaustive list), the physician, the health systems manager, the researcher, the public health expert, and the policy maker may be found at the micro, meso, and macro levels (Table 6-1). Similar and often overlapping vertical hierarchies of role and professional decision-making data needs exist within other health-related perspectives, such as physician practice organization and governance, or among employer purchasers of health care. All these decision makers share the need for timely, relevant, and robust information about common populations of individuals, but they differ substantially in the granularity of their data needs and in how they process and leverage available data and information into action through the decisions they influence and control.
The different types of users of surveillance data access and use the data in different ways, as well as for different purposes. Researchers—whether they are clinical researchers or epidemiologists, behavioral scientists or health services researchers—are typically comfortable downloading micro data sets and analyzing them with appropriate data management and statistical software. Physicians and other clinicians benefit from the results of research studies that are transformed into guidelines by professional organizations and public agencies and then communicated widely to practitioners through peer-reviewed journals and other professional literature and conferences.
Policy makers and advocates as well as the news media similarly benefit from studies conducted with the data, using published findings to develop policies to address the health problem and the factors identified and supported by the studies. But policy audiences also benefit from being able to conduct descriptive analyses of the data to answer their own questions about an issue and to provide such descriptive evidence for the specific population—whether defined by demographic characteristics or geographic boundaries such as their district or county—for which they have some responsibility. Online data tools, such as CDC’s WONDER,1 which can be used to query many CDC data sources, and the California Health Interview Survey’s very user-friendly and flexible AskCHIS2 query tool, are valuable ways to provide access to surveillance data for policy audiences as well as for others.
|Place and Roles||Place||Type||Who||What||Implementation Levers||Linkage to 2010 Reforms|
• Business coalitions
• Benefit associations
• National employer
• Federal government organizations
• Medical society
Priority setting for
• Research and development
• Objectives/ targets (e.g., Healthy People 2020)
• Institutions (e.g., National Institutes of Health)
• Comparative effectiveness
• Meaningful use
• Regional/state employer
• Small business
• State board
• Medical society
• Multispecialty medical group
• Hospital medical staff
• Public health workers
• Local advocates
• Programs and initiatives
• Business planning and development
• Performance reporting
• Institutions and departments
• Communications incentives
ACOs HIT funding
• Beacon sites
• Meaningful use
• HIEe Chronic care Prevention
• “Mom and Pop”
• Medical practice
• Care and action plans
• Programs and initiatives
• Payment or coverage
• Free prevention services
• Payment reform
• Pay for performance
• ACOs; medical home
a PPACA = Patient Protection and Affordable Care Act.
b ACO stands for Accountable Care Organization. According to the Medicare Payment Advisory Commission, “The defining characteristic of ACOs is that a set of physicians and hospitals accept joint responsibility for the quality of care and the cost of care received by the ACO’s panel of patients” (MedPac, 2009).
c ARRA = American Recovery and Reinvestment Act of 2009.
d HIT stands for health information technology. ONC is the Office of the National Coordinator for Health Information Technology.
e HIE stands for health information exchange.
Surveillance design will require explicit trade-offs in what is included and which user needs are addressed as resources are constrained by time, funding, data accessibility, and acceptability of use. For example, cost constraints may result in sampling rather than assessment of an entire population or force a trade-off between detailed biological examinations versus self-reported information. In order to protect the confidentiality of individual patient data, sample size thresholds may be required for reporting.
Existing surveillance capabilities at best serve only a portion of the potential user continuum. Systems have generally been developed to inform policy (at the macro level) and guide priorities (at the macro and meso levels) while technical, confidentiality, and financial constraints have often limited the use of surveillance resources to more localized applications (including guidance of patient care and community policy making). Even then, effective surveillance research has influenced both the topics and specific recommendations of many clinical practice guidelines.
Strategies for improving surveillance, especially when coupled with constrained resources, must be dynamic and malleable to anticipate likely further progress in data collection, availability, use, and governance. In particular, electronic health information systems are rapidly expanding the diversity and accessibility of health data and are breaking down historical barriers to data collection and analysis. Insights into management and protection of individual patient data are allowing new uses of data collected for surveillance purposes, potentially providing timely and relevant decision support for an increasingly diverse group of users.
The roles and relationships among decision makers are also dynamic. Major changes in key structures and relationships within the health system (e.g., the recent health reform legislation) can impact who the decision makers are, what decisions they make, and how decisions are translated into action. ACOs, integrated health care delivery organizations promoted by federal healthcare reform, will have responsibility for the health of populations, as Kaiser Permanente, Group Health Cooperative, and other integrated managed care systems have had. They will need data on the health and risk factors of populations they serve to plan for the clinical, educational, and other resources needed to address their broad health needs, including for preventing and managing chronic diseases. State Health Benefit Exchanges, a key component of healthcare reform, will need good data on the health of populations, including their chronic diseases, that will seek health insurance coverage through them. The key overall design principle for a surveillance system is that it is a means to an end, and the dominant desired effect is to improve decision making by the system’s many different users.
As discussed in the previous chapter, surveillance data come from population-based surveys, investigational cohorts, registries, vital statistics and claims, other administrative data, and test results produced as a consequence of a healthcare encounter. Emerging experience with use of health information technologies (HITs) by both patients and providers suggests that in addition to current sources of information there will be expanding and potentially more efficient approaches to generating data for surveillance. Further, availability of new types of data such as patient care experiences and personal care preferences will expand the scope of what can be systematically monitored and examined. Expanded amounts of data with widening scope are increasingly being developed and generated by healthcare providers using HIT. Of particular interest is the potential, via the electronic health record (EHR), to economically and completely capture care events and processes and efficiently organize them into robust population- and condition-based registries.
In parallel, data are being recorded directly by patients, with or without initiation or direct support from providers and organized care systems. This direct patient involvement is being facilitated and promoted by a range of online personal health records. Systematic collection of information from patients in the form of individualized health assessments has been promoted by health systems, disease and care management entities, commercial wellness companies, and employers. Increasingly, this health assessment information is being commingled with other health data within large electronic data stores and used for population surveillance as well as performance assessment, predictive modeling, and care management.
The generation and sharing of personal health data by patients themselves is a growing health data phenomenon with potential implications for timely, robust, and relevant surveillance. This trend has its roots in the emergence of the Internet and has been amplified by the development and promotion of a range of new technical capabilities and appliances, commercial entities, and social relationships supporting the creation and wide sharing of highly personalized health data. A very recent contribution to surveillance of influenza is Google Flu (http://www.google.org/flutrends), which tracks Internet searches on flu-related topics. Comparing historical trends of the volume of such searches with CDC-produced trends of influenza cases shows a close correspondence; however, the volume of flu-related searches demonstrates the trend earlier than CDC data (Ginsburg, 2009). While a study by Ortiz et al. (2010) found that the estimates provided by Google Flu are not as reliable as the CDC national surveillance program, Valdivia and colleagues (2010) concluded that Google “could be a valuable tool for syndromic surveillance.”
While the more traditional surveillance data resources have matured through use over decades, these newer candidate sources generally lack a comparable experience and evidence base. As use of HIT grows in coming
years, the potential is high that some if not all of these approaches may complement and extend the data foundations that presently exist.
The paper-based medical record of healthcare providers historically has been an uneven and often an inaccessible source of surveillance data. To access data, permission was needed from both the individual patient and the individual provider. Data collection required manual abstraction from paper records, and even when resources were available to abstract paper records, observations in the medical record lacked standardization and often legibility. Data useful for surveillance programs were frequently not recorded in standard formats or were not recorded at all, and population coverage was incomplete. Performing serial observations of patients and populations generally required repeat of intensive manual extraction steps.
The healthcare reform goal of universal coverage, along with broad promotion of health information technologies (especially the electronic medical record), may markedly increase the value of the medical record for disease surveillance. Near universal coverage of the population will create data that are closer to a virtual population census than is current available at any particular point in time. If constructed with population health management goals in mind, the electronic record can provide a more timely, standardized, and relatively inexpensive source of surveillance data.
While large registries provide useful data, they are potentially costly if not linked closely to care delivery, and furthermore, there are challenges identifying the denominators that they represent. When an electronic medical record is suitably designed, the same population analyses can be performed without duplicative data generation and handling. For example, Yeh and colleagues (2010) recently published incidence and case-fatality rate trends for acute myocardial infarction among the population covered by Kaiser Permanente of Northern California. Electronically available demographic data and electronic medical records essentially enabled a virtual population registry and made this analysis possible.
Identifying Patients for Registries
EHR data can be used to identify potential registry patients rather than relying on healthcare providers to recognize and enroll eligible individuals. Investigators can use the EHR to generate lists and prospectively register patients, or to identify potentially eligible patients during healthcare visits. This provides a reminder to clinicians to assess eligibility status and enroll eligible patients. To ensure that patients with a wide range of disease types and severity are included, the algorithms for prospective enrollees must identify individuals with any indication of CVD or COPD in their records. For example, one might include all patients with the diagnosis of angina, even knowing that it is wrong much of the time. Similarly, one could include all patients being prescribed an inhaled COPD medication, knowing that these are frequently used in patients with acute reactive airways disease (such as asthma) as well as COPD. The healthcare provider can then select from this group and enroll those patients who have the diseases of interest.
Stand-alone registries are useful tools for capturing patient-specific data for individuals with selected conditions; however, they have several shortcomings. These shortcomings include possible bias related to which patients are enrolled, missed subsequent data on patients, limited ability to investigate secondary questions, and reluctance of personnel to register patients—particularly on busy days. Envisioning dynamic linkage between EHRs and registries could overcome many of these concerns. Fewer biases in enrollment will exist because most practices and hospitals with EHRs use them for all patients. Investigators can indentify and utilize data for individuals with a condition(s) of interest, provided that reliable data are contained in the EHR. Moreover, data collection is likely to be more complete because the clinicians and the clinical systems collecting and reporting data are doing so for their own practices, not for an extraneous registry.
Challenges in Using EHR Data to Form Registries for Surveillance
Several challenges will need to be addressed in order to more fully use the EHR and associated registries for surveillance. First, EHRs are currently used in only a minority of U.S. hospitals and practices. The American Recovery and Reinvestment Act of 2009 (ARRA)3 and the Health Information Technology for Economic and Clinical Health Act (HITECH)4 are providing financial incentives to promote the adoption and meaningful use of EHRs. The proportion of physician practices and hospitals using EHRs is expected to increase in the next decade, creating a growing opportunity for surveillance of chronic disease.
Patients with significant barriers to care will likely be underrepresented in EHRs. This problem may be resolved if national and state-level healthcare reform provides more Americans with health insurance and if other access barriers (such as health professions shortages and cultural barriers) are addressed. Healthcare providers only collect and record data needed to deliver care. This may not include data necessary for effective surveillance of cardiovascular disease (CVD) or chronic obstructive pulmonary disease (COPD). For example, although vital signs (e.g., blood pressure, heart rate, weight) are usually measured at all visits, capture of symptoms—especially the explicit absence of key symptoms—may be variable, with data preferentially recorded for patients who are symptomatic. Similarly, laboratory tests may be restricted to patients who are sick, have certain conditions, or are taking certain medications.
Sicker patients are likely to be overrepresented in EHRs. Since EHRs record healthcare delivery, there will likely be more data (more visits and more data per visit) for patients who are more acutely and chronically ill. Further, diagnoses are often coded and captured inexactly in EHRs. Some important data will be missing or difficult to analyze in most EHRs. For CVD and COPD, important descriptive and outcome data include symptoms and health-related quality of life which are not routinely recorded in EHRs by clinicians. When recorded, they are usually available only as non-standard free text. Free text (mostly in the form of dictated visit notes and letters) often contains selected, variable details in nonstandard formats that may not be quantifiable or amenable to combining and comparing data between providers or practices.
Despite these problems, EHRs can have an important role in CVD and COPD surveillance. EHRs reflect clinicians’ interpretations and the real-world care that patients receive. They can be useful in CVD and COPD surveillance in two ways: identifying patients for registries, and providing outcome data. The expected growth of EHRs necessitates their inclusion in planning and development of any chronic disease surveillance system.
Implications of Multiple Providers and Multiple EHRs: The Role of Health Information Exchange
For a registry to be reliable and credible for surveillance, it will need to fully reflect the care received by a population. However, CVD and COPD registry patients will often present for cogent outcomes (e.g., heart attacks or exacerbations of COPD or sudden death) to multiple facilities and healthcare providers, each requiring linkage to the registry. EHRs from nearby healthcare facilities can be regularly queried to update cogent outcome data. According to Jha and colleagues (2009), a comprehensive electronic records system can be found in only 1.5 percent of hospitals in the United States. The authors reported that an additional 7.6 percent have a basic system (i.e., present in at least one clinical unit). Therefore, most hospitals and practices lack a comprehensive EHR that could automatically support outcome surveillance, which would also be hampered by lack of a universal identifier to link patients’ data across healthcare institutions.
However, as noted above, the HITECH Act5 is accelerating the speed at which practices and hospitals are implementing EHRs. To take advantage of the propagation of EHRs, various stakeholders from federal and state governments to local provider networks are creating and growing health information exchanges.
There is no movement in the United States to implement a single national EHR. Therefore, the federal Office of the National Coordinator for Health Information Technology (ONCHIT) is supporting the development of Regional Health Information Organizations (RHIOs). The Agency for Healthcare Research and Quality (AHRQ)
3 See http://frwebgate.access.gpo.gov/cgi-bin/getdoc.cgi?dbname=111_cong_bills&docid=f:h1enr.pdf (accessed August 2, 2011).
4 See http://www.hipaasurvivalguide.com/hitech-act-text.php (accessed August 2, 2011).
has funded several statewide health information exchanges as examples of RHIOs and how they can be established and used to facilitate healthcare delivery by enhancing communication between providers (AHRQ, 2010). Adler-Milstein and colleagues (2011) found that most RHIOs that were in the planning phase in 2007 failed to become operational. Fifty-five RHIOs were operational, but data exchange was limited primarily to exchanging test results.
Combining data from multiple EHRs will require an EHR to embrace standards for patient identification (i.e., a core set of descriptive and demographic data that uniquely identifies a patient in multiple EHRs), content (i.e., the minimum set of core data an EHR must contain), coding (i.e., the terms it uses for conditions, tests, treatments, etc.), and messaging (i.e., the format in which patient data are exported from an EHR). Once established, health information exchanges and RHIOs could become rich resources for providing data for enrolling, describing, and following patients in registries.
Overcoming Barriers to the Use of the Electronic Medical Record as a Surveillance Tool: Protecting Patient Confidentiality and Data Sharing
In Essential Features of a Surveillance System to Support the Prevention and Management of Heart Disease and Stroke, Goff and colleagues (2007) recommended the development of mechanisms to “enable linkage between healthcare data systems, including the national surveillance programs (e.g., NAMCS, NHDS, and National Death Index), and electronic health records” (p. 3). They emphasized the critical importance of interoperability of national surveillance programs and electronic health records and the utilization of harmonized data standards. However, a formidable barrier to this goal is the lack of linkable unique health identifiers for individuals. Creative strategies are needed to facilitate this linkage while ensuring confidentiality (Goff et al., 2007).
In 1997, Lillard and Farmer described the advantages and challenges of linking Medicare and national survey data, and they concluded that the linkage of administrative and survey data could provide valuable information on health status, healthcare utilization, and socioeconomic characteristics. Currently, Medicare enrollment and claims data are linked with NCHS surveys including NHIS, NHANES, the Second Longitudinal Study of Aging, and the 2004 National Nursing Home Survey. Data are available for survey respondents who provided personal identification data that were successfully matched with Medicare administrative records. This effort was developed to “maximize the scientific value of the Center’s population-based surveys” and “provides the opportunity to study changes in health status, healthcare utilization and expenditures in the elderly and disabled U.S. population” (NCHS, 2011). CMS data provided to NCHS includes Medicare benefit claims data (1991–2007), Medicare Part D data (2006 and 2007), and Chronic Condition Warehouse data (2005–2007). The linked NCHS-CMS Medicare data are restricted-use files that can be accessed by submitting an application to the NCHS Research Data Center.
Virtual Data Warehouse (VDW) Patient confidentiality must be ensured, and incentives for data owners (e.g., medical groups, health plans) to share their data must be created to facilitate use of the medical care data for surveillance. The HMO Research Network (HMORN)6 is a collaborative network of 15 organizations that cover 11–15 million lives at any one time. The HMORN is addressing these confidentiality and data-sharing issues with the development of a Virtual Data Warehouse (VDW). The VDW was created as a mechanism to produce comparable data across sites for purposes of proposing and/or conducting research. The VDW is “virtual” in the sense that the raw (“real”) data remain at the local sites; the VDW is not a multisite physical database at a centralized data coordinating center. At the core of the VDW are a series of standardized file definitions. Content areas and data elements that are commonly required for research studies are identified, and data dictionaries are created for each of the content areas. A common format for each of the elements—variable name, variable label, extended definition, code values, and value labels—is specified.
The Cardiovascular Research Network (CVRN), a network within the HMO Research Network (see Appendix A), conducts some studies with a different model in which de-identified individual level data are transferred from one or more sites to a central site for analysis. In a current effort funded by an NHLBI-sponsored Grand Opportunity grant to create a cardiovascular surveillance network within the CVRN, all 15 sites of the CVRN
are sending up to 10 years of extensive data (2000–2009) from all the content areas of the VDW to a central site (Kaiser Permanente Northern California) for members who have been diagnosed with acute myocardial infarction, stroke, or heart failure. These data will be used to address research questions in the areas of comparative effectiveness and health disparities.
Local site programmers have mapped the data elements from their own legacy data systems onto the standardized set of variable definitions, names, and codes, as well as onto standardized SAS file formats. This common structure of the VDW files enables a SAS analyst at one site to write a program to extract and/or analyze data at all participating sites. The program from one site is emailed to programmers at other sites to run against their own VDW files. The resultant de-identified data are transferred to the analytic site via a secure encrypted website. Because the VDW maintains a history on past members, the number of people in the data base is much larger than the current membership. For example, HealthPartners in Minnesota actively covers about 700,000 lives, but it has 3 million individuals in the VDW. As of 2010, the standardized content areas developed include enrollment, demographics, utilization, diagnosis, procedures, tumor, pharmacy, census, provider specialty, vital signs, deaths, and laboratory data. Another example of a VDW is the development of the California Virtual Laboratory for Population-Level Analytics, which will integrate population health data (specifically, the California Health Interview Survey) with EHR data from participating healthcare delivery systems through a federated data-sharing system that pulls in key variables for specific analyses but allows the majority of data to remain at its source rather than being transferred en masse to a physical repository.
Veterans Health Administration (VHA) The VHA is the largest integrated healthcare system in the United States, comprising 153 VA hospitals, more than 750 community-based outpatient clinics, and 260 Vet Centers. The VHA-wide electronic health record is a notable data resource for disease surveillance, and a current initiative involves migration of their Computerized Patient Record System (CPRS) to a modern, web-based electronic health record (Department of Veterans Affairs, Strategic Plan Refresh: FY 2011–2015). Healthcare system data from the VHA has been cited for its potential usefulness as part of a national chronic disease surveillance system (Saran et al., 2010), and the public domain software developed and used by VHA to establish EHRs has been recognized as a model for promoting the use of EHRs in the ambulatory and impatient setting (Bufalino et al., 2011).
The VHA conducts numerous chronic disease surveillance activities, including those focused on diabetes, COPD, and CVD. The VHA collects, analyzes, and reports data on individuals with diabetes, including diagnoses, comorbidities, medications, healthcare utilization, and clinical outcomes. For COPD, outcome measures such as admissions and ICU stays, risk-adjusted standardized mortality ratios, risk-adjusted length of stay, and 30-day readmission rates are collected. The VHA has also been a leader in remote pacemaker monitoring, and established a National Implantable Cardioverter Defibrillator (ICD) Center in 2003 (Varosy, 2010). Another VHA initiative, The Clinical Assessment, Reporting, and Tracking (CART) system, is described as an example of national, proactive point-of-care device surveillance. The CART system is “a clinical application embedded in the electronic health record that enables clinicians to document any unexpected problems with devices used in cardiac procedures as part of regular care documentation and is linked to longitudinal outcomes data” (Rumsfeld and Peterson, 2010). CART involves standardized data capture across all the VA cardiac catheterization laboratories, is integrated within the CPRS, and the core data elements conform to the American College of Cardiology’s National Cardiovascular Data Registry. This system enables integration of data collection into the transaction of care, patient safety monitoring, device surveillance, and health services research (Varosy, 2010). The surveillance activities coordinated by the VHA may provide valuable information and lessons in the development of a framework for national cardiovascular and chronic lung disease surveillance.
Vaccine Safety Datalink (VSD) The VSD also uses standardized medical record data for surveillance (http://www.cdc.gov/vaccinesafety/activities/vsd.html). The VSD is a collaboration among the Centers for Disease Control and Prevention, America’s Health Insurance Plans, and eight HMOs to monitor and assess the safety of childhood and adult vaccines. The eight participating HMOs have created identical data sets of all vaccine exposures and all medically treated illnesses, as well as demographic information (birth, gender, race, residence, primary language, and need for interpreter). The VSD supported nearly 30 projects in 2009, including the following:
- H1N1 Vaccine Safety in Pregnant Women
- Safety of the Yellow Fever Vaccine Among Children and Adults
- Henoch Schonlein Purpura and Meningococcal Vaccine
- Rapid Cycle Assessment of Adolescent Tetanus, Diphtheria & Pertussis (TDaP)
- Influenza Vaccine Safety in Pregnant Women
- Rapid Cycle Analysis of Meningococcal Conjugate Vaccine Safety
- Wheezing and Lower Respiratory Disease (WLRD) Multisite Study
- Rapid Cycle Analysis of Pentavalent Rotavirus (RotaTeq) Vaccine Safety
- Assessment of the Burden of Rotavirus Disease and Impact of Rotavirus Vaccination Among Children < 5 Years of Age
- Safety of TIV in Children Aged 24 to 59 Months
- Injections Site and Local Reactions to the Fifth Diphtheria, Tetanus & Pertussis (DTaP) Vaccination
- Does Influenza Vaccination in Children with Sickle Cell Disease Result in an Increased Risk for Fever and/or Pain Crises?
Approximately 80 percent of Americans use the Internet, more than half of adults regularly go online, and a substantial part of this activity is health related according to data from the Internet and American Life Project of the Pew Research Center.7 Between 2006 and 2008, the proportion of individuals who responded that they or someone they know had been helped by medical advice found online grew from one-quarter to nearly one-half. In addition to seeking and consuming information, a growing proportion of these users are also sharing and contributing data and information. Over one-third of Internet users share images such as photos, and about 20 percent of cancer patients use social networking sites to share and obtain health information.
This pattern of patient information seeking and sharing is also reflected by infrastructure design and implementation. For example, among regional and state initiatives to create RHIOs, a rapidly growing proportion are developing capabilities for patients to both view their personal health information directly and to contribute information on their health status (eHealth Initiative, 2010).
Health Risk Appraisals
Efforts to systematically collect information from patients in the form of individualized health assessments have been promoted by many organized care delivery organizations, including health systems, disease and care management entities, and commercial wellness companies. Data from the 2004 National Worksite Health Promotion Survey demonstrated that 45.8 percent of work sites with more than 750 employees used health risk appraisals (Linnan et al., 2008). Among firms that offer health benefits to their workers, 11 percent give their employees the option of completing an HRA, about a fifth of which also offer financial incentives to encourage workers to complete them (Kaiser Family Foundation and Health Research & Educational Trust, 2011).
The term health risk assessment is sometimes used interchangeably with health risk appraisal. However, Anderson and Staufacker (1996) state that the term health risk appraisal (HRA) “formally refers to the instrument whereas health risk assessment refers to the overall process (e.g., orientation, screening, interpretation, counseling) in which the HRA instrument is used.” HRAs are used to develop health profiles, estimate future risks of adverse health outcomes, and provide information to reduce risks. Employers seeking to understand and address the health needs of their workforce have contributed to the use of proactive health surveys and personalized health appraisals. Furthermore, these employers have used survey tools as a mechanism to better create awareness and engagement among their employees. Health appraisal information is increasingly being commingled with other
7 See http://www.pewinternet.org/Trend-Data/Whos-Online.aspx (accessed August 2, 2011); http://www.pewinternet.org/Trend-Data/Whos-Online.aspx (accessed August 2, 2011).
health data within large electronic data stores and used for prioritization of population-based interventions as well as performance assessment, predictive modeling, and care management.
HRAs are widely used in workforce wellness programs and have the potential to provide important information for chronic disease surveillance. According to the CDC, however, how HRAs impact health risk behavior or related health indicators, such as body composition and cholesterol levels, is not well understood (CDC, 2011c).
Personal Health Records and Patient Access to the EHR
With federal incentives for the expanded use of HIT in clinical practices and hospitals as part of the American Recovery and Reinvestment Act of 2009, there is also a growing use of patient-facing aspects of these primarily clinically directed technologies. In addition to support for providers having complete and reliable access to information about their patients, priority is being directed at patient empowerment by HIT so that patients can take a more active role in managing their health. The initial phases of the Meaningful Use HIT incentive program included specific inducements for both direct provision of health information and health record access by providers to patients. It also promoted the development of capabilities to capture patient-identified preferences, experiences, and survey responses as a regular component of routine HIT supported care delivery.
Recording of data by patients in health IT systems is being further facilitated by a range of online personal health records. These may be provided by health insurers, integrated delivery systems, commercial providers of health information tools and support, and freestanding personal health records. Examples of this latter group include HealthVault from Microsoft,8 GoogleHealth,9 and Dossia.10 Patients are increasingly sharing their personal care experiences as well as impressions of individual physicians and other aspects of care delivery through online consumer reviews of individual physician practices.
Sharing of Data Among Patients
Timely access to personally relevant information has been a driving force for patients to form, join, and share experiences and data within a range of organizations independent from historically defined public health, healthcare delivery, and health research entities. These associations often arise among individuals with a specific health condition or disease and have a wide range of organizational structures. They range from formally organized not-for-profit and even commercial corporations imbued with substantial information and knowledge resources to ad hoc and spontaneous patient networks for communication and experience sharing that coalesce via newer Internet social networking tools such as Facebook and Twitter. A few examples of a rapidly expanding array of patient-oriented communities and companies include:
- The Association of Cancer Online Resources (ACOR). “ACOR offers access to 159 mailing lists that provide support, information, and community to everyone affected by cancer and related disorders” (http://www.acor.org/, accessed December 1, 2010). Through supported communities and networks, patients share care-related information such as treatment responses as well as drug adverse effects that they have encountered.
- The Chordoma Foundation (http://www.chordomafoundation.org/) links patients, families, clinicians, and researchers involved in the treatment of this rare cancer, providing pooling of clinical data, individual treatment responses, and researcher interests.
- PatientsLikeMe (http://www.patientslikeme.com/) is a privately owned online company supporting extensive communities of individuals with conditions like amyotrophic lateral sclerosis (ALS) and multiple sclerosis (MS) by providing online self-report tools and population-based reporting to monitor disease course and treatment responses.
Social Networking Registries
A new tool that has the potential to modify the future of surveillance and population-based research is the development of registries that integrate social networking (i.e., Facebook or similar sites) to recruit and retain subjects. Two forms of registries that are currently being used are population-based registries of a general population and disease-specific registries. Examples of population-based registries are ones recruiting in Kentucky and Illinois. As of December 2010, the Illinois Women’s Health Registry (https://whr.northwestern.edu/) had more than 5,000 subjects, and the Kentucky Women’s Health Registry had over 13,000 subjects (https://www.mc.uky.edu/kyhealthregistry/).
The purpose of the Illinois Women’s Health Registry is to provide a research and education tool that advances scientific knowledge of sex- and gender-based differences in health and disease. It is a confidential 30-minute health and lifestyle survey for female residents of Illinois over age 18 and includes questions regarding health, environment, health-related behaviors, symptoms, and illnesses or conditions that a participant may have now or has had in the past. By enrolling in the registry, women throughout the state are provided with information and access to clinical research studies that they may be eligible for based on their self-reported health information. The registry not only serves as a platform for recruitment into pivotal research studies but also represents the beginning of a statewide database that enables researchers to examine the collective de-identified health information provided by women living in Illinois (Bristol-Gould et al., 2010).
The Kentucky Women’s Health Registry is similar with regard to its mission and scope, and its data have already appeared in a peer-reviewed publication (Coker et al., 2009). Both of these registries have cross-sectional as well as longitudinal components and can be used to provide data for analytic studies and study subjects for more in-depth studies.
An advantage of registries linked to social networking capabilities is that one has the potential to follow people easily as they move around the country and even internationally. For those registries linked to social networking systems, their voluntary and non-randomized participation makes generalizing the data obtained from them challenging. For example, participants in the Kentucky Women’s Health Registry are more highly educated, with a much lower smoking prevalence than among the general population of women in Kentucky (Coker et al., 2009).
Privacy concerns are raised by such registries, particularly those linked to social networking systems or commercial enterprises. As social networking makes tracking individuals easier for research, there are concerns about the potential to identify individuals who have shared medical information with an expectation of privacy. Similarly, commercial organizations, such as health promotion firms that contract with employers or health insurers, are expected to protect the privacy of individuals and not share it with sponsoring entities.
Implications of Patient-Generated Data: Potential, But Uncertainty
The generation and sharing of patient information via the Internet and associated social media tools is increasingly common, despite substantial concerns about the protection of patient and provider confidentiality and the ability to reliably use and leverage these data as more than chaotic collections of independent anecdotes. As noted, patients are forming, refining, and increasingly relying on these sources for individual and family guidance. Patients are extensively sharing personal clinical findings, care experiences, and perceived impacts and outcomes of the care they are receiving. As these data are evolving in form, scope, and quality, potential integration into surveillance activities systems can be a focus of experimentation and learning, with one of the largest challenges being the lack of confidence that patterns of health among people who share medical information is similar to those who choose not to share. The growing abundance of data at a highly personal level provides opportunity for further exploration and development as part of a robust surveillance framework.
The current picture of CVD and COPD surveillance in the United States presents a wide range of disparate data courses, often driven by different needs. The creation of a surveillance system built upon current data collection approaches will need to balance a number of challenges, not least of all the tension between cost and granularity, and the differing needs of the different user constituencies of data. The growth of electronic records, as well as emerging data capture, mining, and search technologies, also pose major opportunities and challenges.
All-Payer Claims Databases
A number of states have begun developing all-payer claims databases (APCDs), which combine data from all the payers within a state. These APCDs may have a wide variety of claims data (medical, dental, pharmacy) from both public and private payers. Such databases are established by state legislative mandate, although some states are pursuing the creation of such databases on a voluntary basis. An article by Love and colleagues (2010) reported that 12 states had passed APCD legislation at the time of the article and that there were 11 existing APCDs, with an additional 2 expected by the end of 2010. There is significant variation in polices regarding release of data, with regulations established by each individual state with a legislative mandate. Because APCDs are based on claims data, they are subject to the same limitations as discussed earlier.
Public Health Information Network
The Public Health Information Network (PHIN) is an initiative undertaken by the CDC that is designed to improve public health capacity to use and exchange information electronically. The PHIN, first funded in 2004, originally focused on information systems for improving public health preparedness and response. Today, however, the PHIN strategic plan describes the mission of the PHIN as developing “shared policies, standards, practices, and services that facilitate efficient public health information access, exchange, use, and collaboration, among public health agencies and with their clinical and other partners” (CDC, 2011e).
According to its strategic plan, the PHIN has faced significant challenges, including lack of clear direction, disjointed program planning, alienation of state and local users, costs, and lack of necessary technical capability in many public health settings. To address these issues, the PHIN updated its vision, mission, and goals. The new goals are:
- Provide leadership in the selection and implementation of shared policies, standards, practices, and services for nationwide public health information exchange.
- Define, advocate for, and support public health needs and roles in national health information technology and exchange initiatives.
- Perform key public health information exchange and standards management roles.
The PHIN “will harmonize with and become integral to, the Nationwide Health Information Network, creating the easy-to-find ‘on- and off-ramps’ that enable public health information management systems to use the Nationwide Health Information Network superhighway” (CDC, 2011a). Among PHIN strategies for achieving its goals and objectives are those focused on:
- Establishing well-functioning governance structures and processes.
- Defining and maintaining an architectural framework for public health information exchange.
- Fostering development of information-sharing processes and agreements.
- Harmonizing PHIN as a component of the Nationwide Health Information Network.
- Developing, publishing, and maintaining public health information exchange specifications.
- Establishing PHIN certification for public health information technologies.
- Participating in national standards and implementation processes.
- Providing “data hub” services for national data sets.
- Providing technical services aimed at assisting public health agencies collaborate in standardization and interoperability processes.
BioSense One component of the PHIN is BioSense, which facilitates “the sharing of automated detection and visualization algorithms and approaches by promoting national standards and specifications developed by such
initiatives as the PHIN” (Loonsk, 2004). BioSense was established by the CDC in response to the Public Health Security and Bioterrorism Preparedness and Response Act of 2002, which mandated development of a national public health surveillance system to detect potential bioterrorism-related illness. In 2010, BioSense was redesigned in order to “provide nationwide and regional situational awareness for all-hazard health-related threats (beyond bioterrorism) and to support national, state, and local responses to those threats” (CDC, 2011b).
BioSense is national in scope and focuses on obtaining, analyzing, and reporting data on bioterrorism-related illness, as well as information on situational awareness, and public health response. According to CDC, there are over 800 registered users and the system connects with over 500 hospitals. The system receives data from over 1,000 Department of Defense and Veterans Affairs hospitals and healthcare facilities as well as laboratory data from LabCorp and Relay Health (CDC, 2011d).
National Electronic Disease Surveillance System (NEDSS) The National Electronic Disease Surveillance System is another component of the Public Health Information Network. It is designed to promote the use of data and information systems standards to advance the development of efficient, integrated, and interoperable surveillance systems at the federal, state, and local levels (NEDSS Working Group, 2001). NEDSS is a web-based system designed to enable the secure transfer of public health, laboratory, and clinical data from healthcare providers to public health departments. The broad initiative is intended to facilitate the rapid detection of outbreaks, facilitate electronic transfer of information, reduce provider burden in the provision of information, and enhance the timeliness and quality of information provided (CDC, 2011e).
The vision of NEDSS is “to have integrated surveillance systems that can transfer appropriate public health, laboratory, and clinical data efficiently and securely over the Internet. Once implemented, NEDSS is expected to improve the nation’s ability to identify and track emerging infectious diseases and potential bioterrorism attacks as well as to investigate outbreaks and monitor disease trends” (CDC, 2011e). The mission of NEDSS is to serve the following needs at the local, state, and national levels:
- Monitor and assess disease trends
- Guide prevention and intervention programs
- Inform public health policy and policy makers
- Identify issues needing public health research
- Provide information for community and program planning
- Protect confidentiality while providing information to those who need to know
The principles of the NEDSS design are based on utilization of industry standards, reliance on off-the-shelf software, Internet-based secure transmission of data, a common look and feel of systems, common reporting requirements, and no requirement to use specific software. NEDSS is intended to integrate and replace several current CDC surveillance systems, which are limited by various issues, such as the use of multiple incompatible disease specific systems, incomplete and delayed data, and lack of state-of-the-art technology (CDC, 2011e).
Results from a 2007 assessment of the use of various electronic surveillance systems showed that public health agencies in 16 states (32 percent) reported using the NEDSS Base System as their general communicable disease electronic surveillance system. The remaining 34 states (68 percent) reported using some combination of commercial, CDC, or state-developed electronic surveillance systems to meet their needs. Among the 50 states, 39 (78 percent) reported that at least one aspect of their surveillance system was under development or planned (CDC, 2009). These results demonstrated substantial variation in state electronic disease surveillance systems, although there was a strong commitment to achieving interoperability among systems within states. However, “as interoperability becomes the standard for electronic data sharing, more states will face customization costs and increasing demand for IT personnel in the workforce” (CDC, 2009).
Currently, PHIN, BioSense, and NEDSS are in various stages of development. As is the case with using EHRs for surveillance, a major challenge relates to the relatively small number of public health institutions that have effective, efficient, and interoperable health information technologies. Furthermore, it is likely that much of the data collected by the PHIN and BioSense are not relevant to CVD and COPD surveillance and much of the
data that are relevant are not likely to be collected in these systems. However, PHIN, BioSense, and NEDSS are interesting models for information exchange that could provide lessons in many issues related to the development of a nationwide surveillance system for cardiovascular and chronic lung disease. Such lessons could include those related to technical issues, challenges of integrating multiple stakeholder interests and systems, and collecting and providing information to users at multiple levels.
In response to the passage of the Food and Drug Administration Amendments Act (FDAAA), which mandated that the U.S. Food and Drug Administration (FDA) enhance their ability to monitor the safety of drugs after they reach the market, the FDA launched its Sentinel Initiative in May 2008. The goal of the initiative is to create a national, integrated, electronic system for monitoring medical product safety that will complement existing systems that track reports of adverse events linked to the use of regulated products. The Sentinel System, which will involve collaboration with a wide array of organizations (e.g., academic medical centers, healthcare systems, and health insurance companies), will be developed and implemented in stages and will draw on the capabilities of existing data systems such as electronic health record systems and medical claims databases. The electronic data used in this process will remain in existing, secure environments as a distributed system rather than being consolidated in one database. Within the distributed system, a coordinating center will receive and process FDA-generated safety questions (FDA, 2010).
The Sentinel System vision involves two main components: active surveillance via a distributed system, and expansion of FDA’s current safety surveillance capabilities.
The active surveillance environment will prioritize safety questions that emerge from premarket or postmarket safety data sources such as clinical trial data and spontaneous adverse event reports. The questions will be submitted to the coordinating center for evaluation where the data partners will securely access their databases to evaluate the question and compile HIPAA-compliant results that will ultimately be forwarded to FDA.
Two pilot programs, Mini-Sentinel pilot and the Federal Partners’ Collaboration, are helping shape the Sentinel System. Launched at the end of 2009, the Mini-Sentinel will enable FDA to query privately held electronic healthcare data (including administrative claims and clinical data) representing approximately 60 million patients. The Federal Partners’ Collaboration, which includes the Centers for Medicare & Medicaid Services (CMS), the Veterans Health Administration at the Department of Veterans Affairs (VA), and the Department of Defense (DOD), will enable FDA to query federally held electronic healthcare data, including administrative and claims data and data from electronic health record systems. These pilot projects will provide information about the complex needs of an active surveillance system and will encourage a design that addresses technological, methodological, legal, and operational challenges of the Sentinel System (FDA, 2010).
The emerging FDA Sentinel System provides a rich source of information for those charged with developing a national surveillance system for cardiovascular and chronic lung diseases. While the FDA system is dependent on emerging health information technology which, as yet, is not widespread among healthcare institutions, the challenges faced and solutions developed will be of great use in creating a surveillance system that provides necessary information on prevention, treatment, and outcomes for CVD and COPD.
Adler-Milstein, J., D. W. Bates, and A. K. Jha. 2011. A survey of health information exchange organizations in the United States: Implications for meaningful use. Annals of Internal Medicine 154(10):666-671.
AHRQ (Agency for Healthcare Research and Quality). 2010. Health information technology: State and regional demonstration projects. http://www.ahrq.gov/research/hitdemoproj.htm (accessed March 10, 2011).
Anderson, D., and M. Staufacker. 1996. The Impact of worksite-based health risk appraisal on health-related outcomes: A review of the literature. American Journal of Health Promotion 10(6):499-508.
Bristol-Gould, S., M. Desjardins, and T. K. Woodruff. 2010. The Illinois women’s health registry: Advancing women’s health research and education in Illinois, USA. Women’s Health (London England) 6(2):183-196.
Bufalino, V. J., F. A. Masoudi, S. K. Stranne, K. Horton, N. M. Albert, C. Beam, R. O. Bonow, R. L. Davenport, M. Girgus, G. C. Fonarow, H. M. Krumholz, M. W. Legnini, W. R. Lewis, G. Nichol, E. D. Peterson, W. Rosamond, J. S. Rumsfeld, L. H. Schwamm, D. M. Shahian, J. A. Spertus, P. K. Woodard, and C. W. Yancy. 2011. The American Heart Association’s recommendations for expanding the applications of existing and future clinical registries: A policy statement from the american heart association. Circulation 123(19):2167-2179.
CDC (Centers for Disease Control and Prevention). 2009. Status of state electronic disease surveillance systems—United States, 2007. Morbidity Mortality Weekly Report 58(29):804-807.
CDC. 2011a. About biosense. http://www.cdc.gov/biosense/ (accessed April 18, 2011).
CDC. 2011b. CDC division of informatics solutions and operations. http://www.cdc.gov/osels/ph_informatics_technology/diso/biosense.html (accessed April 20, 2011).
CDC. 2011c. Health risk appraisals. http://www.cdc.gov/nccdphp/dnpao/hwi/programdesign/health_risk_appraisals.htm (accessed June 14, 2011).
CDC. 2011d. National electronic disease surveillance system. http://www.cdc.gov/nedss/About/overview.html (accessed April 19, 2011).
CDC. 2011e. Public health information network (PHIN) strategic plan: Strategies to facilitate standards-based public health information exchange. Centers for Disease Control and Prevention.
Coker, A. L., C. Hopenhayn, C. P. DeSimone, H. M. Bush, and L. Crofford. 2009. Violence against women raises risk of cervical cancer. Journal of Women’s Health 18(8):1179-1185.
eHealth Initiative. 2010. eHealth initiative’s national survey on health information exchange (HIE) shows progress on cost savings and patient access. http://www.ehealthinitiative.org/2010-07-22.html (accessed March 10, 2011).
FDA (U.S. Food and Drug Administration). 2010. The Sentinel Initiative: Access to healthcare data for more than 25 million lives. http://www.fda.gov/downloads/Safety/FDAsSentinelInitiative/UCM233360.pdf (accessed April 19, 2011).
Ginsberg, J., M. H. Mohebbi, R. S. Patel, L. Brammer, M. S. Smolinski, and L. Brilliant. 2009. Detecting influenza epidemics using search engine query data. Nature 457(7232):1012-1014.
Goff, D. C., Jr, L. Brass, L. T. Braun, J. B. Croft, J. D. Flesch, F. G. R. Fowkes, Y. Hong, V. Howard, S. Huston, S. F. Jencks, R. Luepker, T. Manolio, C. O’Donnell, R. Marie Robertson, W. Rosamond, J. Rumsfeld, S. Sidney, and Z. J. Zheng. 2007. Essential features of a surveillance system to support the prevention and management of heart disease and stroke: A scientific statement from the American Heart Association councils on epidemiology and prevention, stroke, and cardiovascular nursing and the interdisciplinary working groups on quality of care and outcomes research and atherosclerotic peripheral vascular disease. Circulation 115(1):127-155.
Jha, A. K., C. M. DesRoches, E. G. Campbell, K. Donelan, S. R. Rao, T. G. Ferris, A. Shields, S. Rosenbaum, and D. Blumenthal. 2009. Use of electronic health records in U.S. hospitals. New England Journal of Medicine 360(16):1628-1638.
Kaiser Family Foundation and Health Research & Educational Trust. 2011. Employer health benefits 2010 annual survey. http://ehbs.kff.org/?page=abstract&id=1 (accessed March 26, 2011).
Lillard, L. A., and M. M. Farmer. 1997. Linking Medicare and national survey data. Annals of Internal Medicine 127(2):691-695.
Linnan, L., M. Bowling, et al. 2008. Results of the 2004 National Worksite Health Promotion Survey. American Journal of Public Health 98(8):1503-1509.
Loonsk, J. W. 2004. Biosense—a national initiative for early detection and quantification of public health emergencies. Morbidity Mortality Weekly Report 53 (Suppl):53-55.
Love, D., W. Custer, and P. Miller. 2010. All-payer claims databases: State initiatives to improve health care transparency. Issue Brief (Commonwealth Fund) 99:1-14.
MedPac. 2009. Improving incentives in the Medicare program. Washington, DC: Medicare Payment Advisory Commission.
NCHS (National Center for Health Statistics). 2011. NCHS data linked to CMS medicare enrollment and claims files. http://www.cdc.gov/nchs/data_access/data_linkage/cms.htm (accessed August 2, 2011).
NEDSS Working Group (National Electronic Disease Surveillance System Working Group). 2001. National electronic disease surveillance system (NEDSS): A standards-based approach to connect public health and clinical medicine. Journal of Public Health Management Practices 7(6):43-50.
Ortiz, J., H. Zhou, D. Shay, K. Neuzil, and C. Goss. 2010. Does Google influenza tracking correlate with laboratory tests positive for influenza? Paper presented at American Thoracic Society International Conference.
Rumsfeld, J. S., and E. D. Peterson. 2010. Achieving meaningful device surveillance: From reaction to proaction. The Journal of the American Medical Association 304(18):2065-2066.
Saran, R., E. Hedgeman, L. Plantinga, N. R. Burrows, B. W. Gillespie, E. W. Young, J. Coresh, M. Pavkov, D. Williams, and N. R. Powe. 2010. Establishing a national chronic kidney disease surveillance system for the United States. Clinical Journal of the American Society of Nephrology 5(1):152-161.
Valdivia, A., J. Lopez-Alcalde, M. Vicente, M. Pichiule, M. Ruiz, and M. Ordobas. 2010. Monitoring influenza activity in Europe with Google Flu Trends: Comparison with the findings of sentinel physician networks—results for 2009-10. Eurosurveillance 15(29).
Varosy, P. D. 2010. Veterans Health Administration: Surveillance of cardiovascular disease. http://www.iom.edu/~/media/Files/Activity%20Files/Disease/Epilepsy/Panel%203-3%20Varosy.pdf (accessed August 2, 2011).
Yeh, R. W., S. Sidney, M. Chandra, M. Sorel, J. V. Selby, and A. S. Go. 2010. Population trends in the incidence and outcomes of acute myocardial infarction. New England Journal of Medicine 362(23):2155-2165.