Report of the Tracking Network Workgroups Download as PDF [199 Kb]
Report of the Tracking Network Workgroups
- Executive Summary
- Background, Vision and Process
- Workgroup Recommendations
- Attachment A: Member Agencies and Organizations of the Tracking Workgroups
- Attachment B: Final Report (Workgroup 2)
- Attachment C: Final Report (Workgroup 3)
- Attachment D: Partnership for Integrated State Information Systems
The charge to this workgroup was to identify and describe existing tracking systems at the national, state, and local levels; determine priorities for integrating existing tracking systems; and identify and prioritize the development of new systems.
The Environmental Public Health Tracking Network (EPHTN) should be a system, or network of systems, that:
- Provides baseline and trend information about chronic diseases and
conditions of current and emerging interest and concern at the
national, state, and local levels
- Reintegrates public health and environmental practice and helps
elucidate relations between environmental exposures and human diseases
- Is accessible and useful to local communities and to public health and environmental protection officials
Inventory and Description of Existing Health Tracking and Surveillance Systems
To focus its work, the workgroup prioritized health endpoints for the EPHTN. These health endpoints reflect the priorities articulated by the Pew Environmental Health Commission together with the recommendations of the experienced public health professionals who participated in the workgroup. They share common attributes: with the exception of "acute chemical poisoning," each is a chronic disease or condition of complex, multifactorial etiology that is little understood but in which environmental factors have been strongly implicated. The priority health endpoints are:
- Birth defects
- Developmental disabilities (including cerebral palsy, autism, and mental retardation)
- Asthma and other chronic respiratory diseases
- Cancer, especially childhood cancers
- Neurologic diseases (including Alzheimer's disease, Parkinson's disease, and multiple sclerosis)
- Adverse reproductive outcomes
- Endocrine effects (including early menarche, hypospadias, and low sperm count)
- Autoimmune disease, especially lupus
- Acute chemical poisoning
Drawing on the expertise of its members and an extensive search of state, federal, and health-related websites, the workgroup developed an inventory of chronic and environmental health tracking systems and a set of criteria for assessing these against the goals of an EPHTN.
The workgroup took a broadly inclusive approach (1) to the identification and description of disease tracking and surveillance systems. In addition to the many CDC and ATSDR sponsored federal and state health surveys and registries, the workgroup also included in its inventory specific National Institutes of Health (NIH) and Health Resources and Services Administration (HRSA) projects. Although epidemiologic work is only one component of these projects, the workgroup thought they nevertheless could represent good strategic leverage points for the EPHTN. Finally, a sampling of state disease tracking programs, large health-care administrative data sets and a number of unique commercial health data collection and linkage efforts were included. These are briefly described below.
To help assess the utility of each system in the inventory to the epidemiologic work of the EPHTN, the workgroup characterized the systems according to the following variables:
- Scope (e.g., national, statewide, multistate, city, multisite, demonstration project)
- Standard definitions (e.g., ICD-9/ 10, data-dictionary, serologic finding)
- Level at which data are statistically meaningful (e.g., individual, block, ZIP code, city, state)
- Type of measurement (e.g., incidence, prevalence, lifetime occurrence)
- Population (e.g., sample, cases, population, enrollees)
- Study design (e.g., longitudinal or cross-sectional)
- Method of data collection (e.g., questionnaire, telephone, interview, vital records, administrative records)
Other variables include whether the data source is linked to demographic information; whether it includes children; whether it includes or can be linked to exposure information; and, for local and demonstration programs, whether it can be replicated.
1Although extensive, the workgroup's search was by no means exhaustive. The workgroup doubtless missed many important examples of state-level disease tracking initiatives, and we were able to include only a few examples of local projects. Because a fundamental premise of the EPHTN is that existing tracking systems do not adequately capture information about most of the priority health endpoints (diseases or conditions), we were most interested in casting a wide net to learn what other sources of this information might exist. The resulting inventory provided the opportunity to think more creatively and expansively about sources of epidemiologic data that bear on the EPHTN.
Public Health Data Sources
The primary tools for disease tracking include legal and administrative record systems such as vital statistics and hospital discharge data; disease registries such as cancer, birth defects, immunization, and hazardous exposure registries; population surveys such as the Behavioral Risk Factor Surveillance System (BRFSS), the National Health Interview Survey (NHIS), the National Health and Nutrition Examination Survey, and the State and Local Area Integrated Telephone Survey (SLAITS); and provider surveys such as the National Ambulatory Medical Care Survey (NAMCS), the National Hospital Medical Care Survey (NHMCS), the National Nursing Home Survey (NNHS), and the Medical Expenditure Panel Survey (MEPS). CDC and the states use many of these instruments, and most of these instruments have state modules that enable state health departments to include questions of specific local interest. Some are standardized at the national level, others are not. Some states use all of these instruments to develop a picture of priority conditions that is as comprehensive as possible (e.g., Connecticut's Asthma Report), while others use one or two, and may or may not integrate analyses.
The strengths and weaknesses of this public health "toolbox" for tracking and monitoring chronic and environmental health conditions have been summarized by The Pew Environmental Health Commission and others. As potential building blocks for the EPHTN, many of these systems are challenged by poor geographic specificity, lack of state-to-state comparability, inconsistent case definitions, few links to exposure measurement, and/ or a focus on specific diseases rather than syndromes of potentially common etiology. To these assessments, the workgroup would add that tracking many EPHTN priority diseases and conditions will require a much greater emphasis on ambulatory care data than the current public health toolbox affords. Only a small fraction of conditions such as asthma and chronic respiratory illness or endocrine-related conditions will turn up in hospitals in any given year, and this fraction diminishes as the health system changes and evolves new means and settings to manage chronic health conditions and health conditions that are not life threatening. Traditional telephone surveys and self-reported questionnaires are imprecise alternatives for capturing information about these conditions consistently and meaningfully.
Other Health Data Sources
In addition to the major public health surveillance tools in regular use, the tracking system inventory includes the substantial health data systems under the Medicare and Medicaid programs. Given the nature of these programs, their data sets represent potentially rich sources of information about populations of special interest to public and environmental health (low income women and children, people with disabilities, and the elderly). The inventory also includes Public Health Service (PHS) programs such as Community Health Centers, Federally Qualified Health Centers, and Maternal and Child Health programs that also provide services to populations that may be of special interest or concern and that collect and maintain extensive diagnostic, service, and demographic information in uniform databases.
Also contained in the inventory are several NIH-supported activities (Centers of Excellence and longitudinal studies) focused on EPHTN priority diseases or conditions. Though designed to serve research needs rather than public health applications, these programs were included because they represent important opportunities to capture environmentally related risk factors and/or because they can provide springboards for collaborative epidemiologic research. One potentially applicable model for bridging the gap between NIH-sponsored academic research centers and state health departments is the Prevention Research Centers program.
The inventory also identifies important surveys and longitudinal studies conducted by other federal departments. For example, the Department of Education collects extensive data under the Individuals with Disabilities Education Act that could support tracking of learning and developmental disabilities. These are priority conditions for the EPHTN.
In addition to federally sponsored health data and tracking systems, the inventory includes several exemplary surveys, registries, and databases developed by individual states that could contribute important information to disease tracking efforts (e.g., the Wisconsin Family Health Study). Some of these state-based initiatives were originally developed for surveillance purposes; others were established to capture health expenditure information or to support local or statewide planning and budgeting activities. Nevertheless, many contain useful health status, demographic, diagnostic, and utilization information.
Finally, the inventory contains information about a sampling of private sector health plans, data warehouses, and electronic data interchange (EDI) contractors, as well as physician practice research networks funded by a variety of interests that offer novel opportunities for broad population-based disease tracking.
Inventory and Description of Existing Environmental Tracking and Monitoring Systems
Following the work of The Pew Environmental Health Commission, the workgroup identified priority environmental hazards for inclusion in the EPHTN. These are:
- Persistent organic pollutants, including polychlorinated biphenyl (PCB) and dioxin
- Heavy metals, including mercury and lead
- Pesticides, including organophosphates and carbamates
- Air contaminants, including toluene and fine particles
- Drinking water contaminants, including pathogens
The workgroup also developed descriptive criteria to assess the utility of existing environmental monitoring systems to EPHTN applications. These criteria are:
- Type of exposure estimator (e.g., emissions, production, contamination, residue, personal monitor)
- Media (e.g., indoor or outdoor air; surface, drinking, or ground water; soil; food; human tissue; bulk chemicals)
- Geographic unit (e.g., national, regional, state, region of state, county, city, ZIP code, census, facility)
- Sample frequency (yearly, quarterly, monthly, weekly, daily, irregular, mixed)
- Sample locator (e.g., latitude and longitude; Universal Translocator System; metropolitan statistical area, city or municipality, county or parish, ZIP code, street address, Hydraulic Unit Codes, River Reach Number, other)
- Substances covered (e.g., acids, bases, PCBs, VOCs, metals, pesticides, criteria pollutants, aeroallergens, particulates, pathogens)
Data and information about priority environmental hazards are collected by a variety of state and federal agencies. However, few environmental monitoring systems were established with public health practice in mind, and as a result, they are not widely used by public health practitioners to characterize exposures that could potentially effect human health. Workgroup members were sensitive to complaints about the general quality of environmental data that currently are collected. A review of the major data sets in the inventory (Attachment) demonstrates four general issues:
- What is measured: Data collection systems vary, but in most cases
they exist to support EPA's ability to assess compliance with
environmental regulations or the success of regulatory programs to
reduce pollution. Thus, the monitors that capture environmental
contaminants are generally placed at the source of known or suspected
emissions and discharges (e.g., air monitors are placed in areas of
high pollution, and water monitors are placed at the end of the
discharge pipes), rather than distributed across areas where people
are likely to come into contact with the contaminants. As a result,
health investigators often must apply a still evolving complex of
modeling techniques to assess "upwind" or "downstream" exposures,
further complicating efforts to assess actual exposure effects.
- How often it is measured: Some systems, such as the Air Quality
Information System, take sample measurements every day and, for some
pollutants, every hour; others, such as the Safe Drinking Water
Information System, are required to report data only every quarter;
and still others record information annually, or on an ad hoc basis in
response to concerns about a particular area or event, such as an
unintentional release of hazardous material. The fewer the number of
measurements for any particular exposure, the more likely it is that
the effect of that exposure, if one exists, will be underestimated.
- How the measurements are defined: Because most measurements are
taken to support environmental regulations, the threshold values
captured can change with each new rulemaking.
- How the information is reported to and aggregated at the national level.
Fortunately, many states may collect far more environmental monitoring information than they report to EPA, although, as the workgroup members noted, the quality of that information also can vary considerably from state to state. Generally, the quality and completeness of state environmental data are driven by the needs of state public health and environmental protection officials to address popular, legislative, and gubernatorial concerns. That fact may be useful for identifying states where pilot tracking projects are likely to be enthusiastically supported.
The workgroup noted that while the mixed quality of
environmental hazard and exposure information has limited health
research and epidemiologic studies for many years, many researchers have
nevertheless found ways to work with existing data (e.g., the 6-cities
study and New Jersey drinking water study) (2).
In addition, EPA is committed to incorporating public health indicators
into its program planning and evaluation efforts, which will help propel
the development of better monitoring and exposure data. Finally,
although there have been many
calls and recommendations from the environmental health community for "better environmental exposure data," such calls have not often been accompanied by specific requests in the right venues for adjustments or enhancements to environmental monitoring systems. The EPHTN will create a strong venue for sharing practices and recommending refinements.
2Bove F, Environmental Health Perspectives; Cator KP, Drinking water and cancer. Cancer Causes and Control 1997; 8: 292-308; Massachusetts Department of Public Health, Woburn Childhood Leukemia Follow-Up Study, Bureau of Environmental Health Assessment, Massachusetts Department of Public Health, 1997; New Jersey Department of Health and Senior Services, Case-control Study of Childhood Cancers in Dover Township (Ocean County), New Jersey, Division of Epidemiology, Environmental and Occupational Health, 2001; Cohn P, Klotz J, Bove F, Berkowitz M, Fagliano J, Drinking water contamination and the incidence of leukemia and non Hodgkin's lymphoma, Environ Health Perspect 1994; 102: 556-61; Croen LA, Todoroff K, Shaw GM, Maternal exposure to nitrate from drinking water and diet and risk for neural tube defects, Am J Epidemiol 2001; 153: 325-31; Munger R, Isacson P, Hu S, Burns T, Hanson J, Lynch CF, et al, Intrauterine growth retardation in Iowa communities with herbicide-contaminated drinking wager supplies, Environ Health Perspect 1997; 105: 308-14.
Level I Recommendations: Link Existing Systems
As an initial step, the EPHTN should support states to create or build on links across health and related data sources for priority chronic disease and environmental health endpoints. Such links may be as simple as the copresentation of indicators for diseases and conditions of interest in a common medium (e.g., narrative report or geographic display), or as elaborate as fully integrated software systems, depending upon the state's current level of sophistication. Health data links should address the following objectives:
A. Characterize the disease or condition, and triangulate data sources
In theory, national and state-specific data sets can be used in combination in studies of the prevalence of a disease or condition; the morbidity, disability, and mortality associated with it; its distribution (e.g., geographic and ethnic variation); its associated cost burden; and its evolution (i.e., natural history) in a population. However, in practice, multiple data sources are infrequently linked because relevant indicators (e.g., for asthma, doctor visits and missed school days) may be collected by different federal and state departments, may not be captured in common space and time (e.g., bi-annual statewide telephone sample surveys versus annual hospital service area discharge statistics), may not be available in a timely manner, and may use different definitions (e.g., physicians' diagnoses versus self reported asthma cases). For these very reasons, however, linkage efforts are so often valuable. As part of the effort, researchers learn how well or how poorly comparable data elements from multiple systems confirm one another and what those findings suggest about the strengths and weaknesses of individual systems for public health tracking.
Best practice examples of data linking (that do not depend upon integrated data systems per se) include Connecticut and Nebraska reports on childhood asthma. Each of these states has compiled asthma statistics from a variety of public health, administrative, and school-based tracking systems (e.g., hospital discharges, emergency department (ED) visits, BRFSS, missed schooldays, pharmacy records) to present a comprehensive picture of asthma prevalence and disease burden within its borders. Through such triangulated data (the copresentation of statistics from different systems), the states' public health officials have compensated, in some measure, for the imperfections of the individual data sources and assembled a baseline series of indicators to track asthma. (3)
3Even so, the vast majority of asthma cases present and are diagnosed in primary care physicians' offices, rather than in hospitals, so existing public health data systems that rely heavily on a combination of respondent self reports and hospital admissions, discharge, and ED data will fail to accurately capture true disease prevalence and thus produce poor estimates of associated burden. The workgroup underscored that state and local public health agencies must have much better information about the prevalence of diseases and conditions that present only or mostly in the non hospital ambulatory setting.
How does this relate to the Requests for Proposals (RFP)?
Applicants should describe how they propose to create multidimensional baseline indicator reports for three chronic environmental health conditions. At least two of the three conditions should come from the priority list. Baseline indicators should include prevalence, morbidity, disability, mortality, ethnic variation, cost burden, and geographic and temporal trends.
In planning and compiling the reports, state health officials are encouraged to leverage resources beyond the traditional purview of health departments, including, as appropriate, schools, laboratories, Centers of Excellence, the aging network, Medicaid and Medicare data marts, health plans, and pharmacy sales. States should report to CDC and ATSDR on the concordance of these sources with other public health data sets (where they exist) for case definitions, key variables, and other characteristics relevant to their use as epidemiologic tools. The degree of actual data and software integration (versus copresentation in narrative or other reports) should be based on a realistic assessment of states' current data systems, with the goal of gradually enhancing sophistication and data integration capability.
Baseline data are needed for all EPHTN indicators. In addition, much learning, practice, and problem solving happens during the development of a data based report. State public health personnel will gain first-hand, applied knowledge of the availability of relevant data and data systems across programs and agencies. CDC and ATSDR will gain valuable insight into the potential contributions of nontraditional sources of public health information, as well as the obstacles to their integration with current CDC-sponsored systems. (4)
4These insights should be communicated in a timely and thorough manner to the Public Health Data Standards Consortium (PHDSC), the National Committee for Vital and Health Statistics (NCVHS), and the HHS Health Data Council, which oversees development of national health data standards.
B. View and assess the "health" of local communities with respect to chronic disease indicators
One goal of an environmental public health tracking system is to generate information about the co-occurrence of diseases or conditions (e.g., adverse reproductive outcomes, chronic respiratory diseases or neurologic disorders) that may share some etiologic agent(s), affected group(s), or other attribute(s) of interest and to compare this information across populations. However, as noted above, data on priority diseases and conditions, if they are collected at all, are collected and reported by different systems, registries, or surveys (e.g., cancer registries, birth defects surveillance systems, National Health Interview Surveys), making characterization of the health of a community with respect to any combination of these indicators exceedingly difficult. Geographic information systems (GIS) provide one means of presenting such information in common space and time (i.e., geography is the common denominating variable), thus enabling a more complete picture of community health status with the potential to unfold patterns of public health significance on key indicators.
How does this relate to the RFP?
Applicants should describe an approach to co-presenting data on at least three EPHTN priority health endpoints at the substate (ideally community) level. At a minimum, these data should include incidence or prevalence and basic demographic information for each health endpoint. The applicant should describe how underlying data would be made available to researchers, the public, or others. CDC and ATSDR should provide technical assistance to system administrators on privacy and small area analysis issues.
Where appropriate, proposals should include a description of existing state GIS capacity, its platform, current uses, degree of public access, and whether and how it can incorporate priority environmental health indicators.
This is a core objective.
C. Assess specific populations
The workgroup also noted several state-based data linkage
projects designed to better exploit the research and prevention
potential of combined databases for specific populations. Many of these
efforts were kindled under the Maternal and Child Health program and the
Robert Wood Johnson Foundation Information for State Health Policymakers
(InfoSHiP) grant programs, and thus focus on linking data sets that
particularly describe the vital statistics, health status, risk factors,
nutrition, and service utilization of pregnant women and children. Most
projects also incorporate birth defects and tumor registries, and
several include newborn metabolic and endocrine screening data. Because
children and pregnant women are populations of interest for so many
environmental exposures and conditions, these state based linkage projects could serve as platforms for more ambitious environmental public health data collection (including biomonitoring) and integration efforts.
How does this relate to the RFP?
Applicants should describe how they would integrate existing population-specific data linkage projects into their public environmental health tracking network. Applicants also should be encouraged to consider how routinely collected newborn blood spots and maternal cord blood samples could be used for environmental exposure measurement (biomonitoring).
Infants, children, and pregnant women are of great interest and concern to environmental health scientists, representing "sentinel" populations for many environmental exposures. The Maternal and Child Health/ InfoSHiP projects capture important contextual variables (e.g., maternal health, nutrition status, limited blood analysis) to enable a rich assessment of potential contributors and confounders to specific health outcomes. In addition, because newborns and children under age 6 years are excluded from the NHANES biomonitoring protocol, environmental exposure data for this population have been lacking, but blood spots that could be used for this purpose are routinely collected as part of the MCH newborn screening program. Finally, many of the existing state data linkage projects are funded, not through CDC or ATSDR, but through other federal agencies, providing a leveraging opportunity for limited health tracking funds.
D. Leverage NIH support, resources, and collaboration
NIH conducts a wide range of epidemiologic investigations pertinent to the EPHTN, and opportunities exist for coordination and collaboration with CDC and state public health officials. Examples of NIH projects that are relevant to the EPHTN include the National Institute of Environmental Health Sciences' (NIEHS) Environmental Genome Project, which currently is conducting population-based epidemiologic studies of Parkinson's disease, susceptibility to pesticides, and prostate cancer. Similarly, the National Cancer Institute (NCI) conducts large population-based studies on gene-environment interactions to explore the interplay between inherited susceptibility to cancer and environmental risk factors. Finally, NIH sponsors Centers of Excellence on several EPHTN priority health conditions, including Morris K. Udall Parkinson's Centers, Childrens' Environmental Health Centers, and Centers of Excellence for Autism and Developmental Disabilities. Where there is local concern about conditions within Centers' portfolios, public health officials should be able to leverage the extraordinary resources of these Centers to help support their own epidemiologic work. State health department collaboration with the Centers also could help ensure that new findings related to diseases and conditions of interest can be quickly assessed for their relevance to public health monitoring and assessment functions for both hazards and health effects. A model for such academic and public health coordination might be the Prevention Research Centers.
How does this relate to the RFP?
Applicants should be encouraged to link with NIH-supported population-based epidemiologic research on priority EPHTN health indicators.
Expand the resources available to environmental public health and the expeditious dissemination of new knowledge to the field.
A demonstrated relationship between state departments of health and state environmental protection agencies (or their appropriate analogues) that will facilitate data linkage, interpretation, and development should be a prerequisite for health tracking support. Opportunities for connectivity and leverage include shared GIS platforms, interdepartmental liaisons, ATSDR cooperative agreements, and state public health and environmental laboratories.
Fundamental to the EPHTN is the expectation that public health practitioners can characterize environmental hazards within their jurisdictions that could potentially affect human health. They should be able to identify the different types of hazards or pollutants; media and systems that can be affected (e.g., food; land; indoor and outdoor air; and lakes, rivers, pools, coastal areas; and public and private drinking water supplies); the likely sources of contaminants; the appropriate indicator organisms, trace elements, and chemicals; the manner in which humans are exposed to the contaminants; and the implications of exposures to peoples' health over time. In addition, health practitioners should know where chemicals are used and stored after manufacture and how the public is exposed. Finally, health officials should have the means to interpret environmental hazard and exposure data, including models necessary to convert raw data to relevant health information.
As noted above, however, although a wealth of environmental data exists, most of it has been developed for specific environmental regulatory purposes, making accurate interpretation and application of the data difficult for those outside the regulatory framework. An accessible resource with knowledge of the relevant questions, the ability to identify processes and linkages, and the ability to place the data in an appropriate context would make the data more meaningful to health officials. There are several models:
- The HHS Assistant Secretary for Planning and Evaluation (ASPE)
maintains a database of state health data integration efforts (current
as of 1998), and among the projects described are several that link
health and environmental information. As expected, these examples turn
up most frequently in states where departments of public health and
environmental protection remain closely tied. One example is Colorado,
which has instituted a GIS to serve both public health and
environmental protection programs and has developed a cooperative
Center for Health and Environment Statistics. A second example is
Indiana, which also has developed a GIS that includes demographic,
infrastructure, and environmental data to develop programs and track
health outcomes at the community level, supplemented with a public and
private Health Data Center that pulls information from public health,
Medicaid, hospitals, workers' compensation, and other sources.
Indiana's GIS was developed through the joint efforts of GIS users in
many state agencies who had been using GIS to study trends within
their agencies. Ongoing development of the system fits within a
broader overall strategy to integrate data and empower local areas to
understand their local health and environmental issues. A new example
is New York State's Cancer Surveillance Improvement Initiative, which
includes a cancer registry that allows for the analysis of cancer
incidence and its relation to environmental and other factors.
- Using a different model, Alaska designates liaison personnel to
manage information sharing between its public health and environment
protection departments. The interdepartmental liaisons translate data,
programs, and practices, and discern the right place in the other
organization to pose questions or request assistance.
- ATSDR's state cooperative agreements could naturally bridge public
health and environmental protection functions. At the national level,
ATSDR works closely with both CDC and EPA in responding to hazard
releases. At the state level, ATSDR supports multidisciplinary teams
of health professionals who characterize past and present chemical
releases and assess health threats posed to communities living around
approximately 500 National Priority List (NPL) hazardous waste sites.
This broad expertise in toxicology, environmental assessment, exposure
investigation, and community education is a tremendous resource for
state environmental public health programs.
- Finally, an important opportunity exists to expand and coordinate state laboratory capacity for environmental sampling and media monitoring with state and regional public health laboratories through ATSDR, CDC Bioterrorism Cooperative Agreements, and the Federal Emergency Management Agency.
How does this relate to the RFP?
Applicants should provide documentation of an agreement between the directors of the relevant public health and environmental protection agencies to collaborate on the development and interpretation of environmental data necessary to support the EPHTN. Applicants should explain how they intend to assess the capacity of existing environmental hazard data to provide information about patterns of exposure. They should select two health endpoints from the priority list, or one health endpoint from the priority list and one hazard of local concern (which need not be related to the health endpoint) with which to demonstrate links across health and environmental information sources. States should be encouraged to propose efforts to further develop these links using tools that serve common health and environmental monitoring purposes, including GIS, biomonitoring, ATSDR cooperative agreements, and environmental health laboratories. Applicants should describe a process by which "lessons learned" and recommendations to enhance and or refine health and environmental data to better serve the goals of the EPHTN will be communicated to the leadership of the relevant agencies and to CDC and ATSDR.
The use of data to reintegrate health and environmental practice is a core objective of the EPHTN. However, many public health professionals are not sufficiently familiar with environmental databases to use them comfortably. Many state environmental protection officials, too, have expressed frustration that media monitoring requirements, driven by the regulatory enforcement system, do not generate the data to answer the public health questions that the public is asking of them. This recommendation will require a detailed assessment of the public health utility of state-level environmental data, and create a feedback loop to state officials and system architects to help them refine their instruments to better meet public health needs.
CDC, ATSDR, and EPA should establish a regular forum for intensive, hands-on (applied) exchange between state and federal public health and environmental data developers, statisticians, and other users.
CDC, ATSDR, and EPA should provide national or regional training workshops for state public health and environmental protection officials to bring their own data sets together to approach and solve actual problems. The purpose of these workshops is to create a shared knowledge and facility across the two agencies' systems.
How does this relate to the RFP?
Applicants should agree to contribute to and actively participate in the workshops.
Peer-to-peer exchange on problem solving techniques and best practices will help data systems and practices evolve toward greater state-to-state consistency. This is a key goal of the EPHTN. Another important outcome of the workshops will be a feedback loop to state and federal data system architects about strengths and weaknesses of the systems for environmental health applications and the meaningful tracking of relevant indicators.
CDC and ATSDR should facilitate evaluation of GIS for application to the EPHTN
More and more states are using GIS to begin to assess public health questions. However, GIS systems vary considerably from one another in the sophistication of their underlying statistical software, their ability to manage large amounts of multilevel information, their methods for small area analysis, and other characteristics important to epidemiologic and environmental health applications.
Similarly, states that actively use GIS have very different practices for managing, sharing, and presenting data. For example, New Mexico maintains a data repository that was codeveloped by the New Mexico Department of Health, the New Mexico Health Policy Commission, the New Mexico Tumor Registry, and the University of New Mexico. Clinicians and communities throughout the state can use the data repository to assess community health status. The software platform is ArcIMS, a commercial Internet-based platform requiring no preexisting knowledge of GIS to use the data. Washington developed Epi-QMS, which allows users to carry out exploratory geographic and statistical analysis and is available to users at three levels: 1) the general public, 2) public health and medical practitioners, and 3) epidemiologists. Its database uses accumulated records, processed off line and uploaded, rather than record-level information. A key feature of the system is the availability of small-number adjustments that include nearest neighbor and empirical Bayesian smoothed rates. EpiQMS does not require purchase of mapping software but uses scaleable vector graphics (SVG), a new Internet graphics standard that speeds up map production. Other states have evolved their own customized GIS platforms, applications, and access requirements.
Finally, EPA and ATSDR have built extensive mapping platforms that include U. S. Geologic Survey hydrology data, National Oceanic and Atmospheric Administration atmospheric data, and other information that supports environmental modeling. Building environmental public health datasets on top of one of these systems might be advantageous (provided they represent state-of-the-art and state-of-the-science technology and statistical packages).
CDC and ATSDR should convene a meeting of GIS data developers, users, and stakeholders to assess all of these options and decide "best-platforms" and "best practices" for GIS public environmental health applications. CDC, ATSDR, and EPA may need to sponsor development of a customized system that includes all relevant functions and can be regularly upgraded and made available to state and local public health departments for one license fee. Because state-to-state consistency is a chief objective of the EPHTN, one goal should be to encourage proliferation of compatible systems.
How does this relate to the RFP?
Not applicable; this recommendation is directed to the federal agencies.
CDC and ATSDR should enhance core chronic and environmental health surveillance systems to better capture information about environmental exposures and conduct state pilots as part of the EPHTN.
Birth Defects Surveillance: In FY 2002, CDC had cooperative agreements with 35 states to report on the incidence of major birth defects, particularly neural tube defects. Agreements in seven states supported Centers for Birth Defects Research and Prevention, which collect cases and controls for the National Birth Defects Prevention Study. The remaining 28 states received support for surveillance and use of data for public health purposes. Congress has provided funds explicitly to expand and enhance the registries. Tracking funds should support linkages of birth defects data with 1) related conditions of interest, 2) diagnostic centers, and 3) data derived from hazard and exposure systems, which potentially could result in new information about possible causes, risk factors, and opportunities for prevention. CDC and ATSDR should work with the centers and state birth defects surveillance projects to develop the key elements, formats, and variables to enable these linkages.
National Program of Cancer Registries: In FY 2002, CDC provided grants to 45 states, three territories, and Washington, D. C., to improve cancer registries or develop new ones. However, many states still cannot effectively respond to reports of possible clusters of cancer cases. States need the ability to capture all the cancers in their states, determine whether a cancer cluster represents an excess cancer risk, and link information about environmental contamination with cancer registry data. Workgroup members noted that more effectively capturing data from physicians' offices is important because more and more detection, pathology, and treatment services appear to be occurring there. The workgroup noted that NCI's Surveillance, Epidemiology, and End Results program (SEER) is attempting to link registry data with medical records.
Behavioral Risk Factor Surveillance System: The BRFSS is a state-based telephone survey active in all 50 states, and a primary source of information about national and state trends in health risk behaviors. Adding an environmental health module to obtain information about environmentally related conditions and exposures could enhance the survey. However, because it is an interview survey and respondents are unlikely to have a detailed knowledge of local environmental factors or conditions to which they are exposed, the survey must be linked to environmental hazard or exposure data sources. Moreover, if the BRFSS is to become a platform for environmental health tracking, it must dramatically increase within state sample size, require standardized questions related to environmental health endpoints in ways that ensure state-to-state consistency and minimize bias (e.g., "Has a doctor diagnosed ______?" "Are you currently being treated for ____"), and better address the issue of households without telephones.
Hazardous Substances Emergency Events Surveillance System: Maintained by ATSDR, this system describes the public health consequences, including morbidity and mortality and associated risk factors, of acute releases of hazardous substances. Sixteen states currently implement this system. CDC and ATSDR might leverage bioterrorism funds to expand this program to the remaining states for dual use.
Concurrently improve the ability of existing systems to capture priority health endpoints.
The workgroup noted that many existing health data collection instruments do not precisely characterize priority environmental health endpoints, presenting challenges to accurate data collection, research, and interpretation, but also limiting opportunities to improve and expand surveillance capacity. Opportunities to improve the ability of existing systems to characterize and track health endpoints of interest reside in case definitions, coding and classifications systems, survey questions, sampling frames, and laboratory reporting practices.
CDC and ATSDR should accelerate work on standardizing case definitions for environmental public health tracking in concert with other federal and state agencies that use and report health statistics. Where possible, CDC and ATSDR should move to adopt case definitions that have achieved broad acceptance, with the understanding that retrospective adjustments may be needed.
Case Definitions: Developmental Disabilities
One of the chief obstacles to tracking childhood
disabilities has been the lack of a standard case definitions.
Capitalizing on recent conceptual and methodologic developments in the
demographic, social, and biomedical study of disability, the National
Institute of Child Health and Human Development is supporting the
development of concise measures of childhood disability, with a focus on
mental health and learning disabilities. The project relies on the 1994
and 1995 disability supplements to the NHIS, the 1997 NHIS, and the
1992 and 1993 Survey of Income and Program Participation (SIPP). The measures under development are specifically intended for use in population surveys and survey-based surveillance systems to monitor the prevalence of childhood disability.
Case Definitions: Mental Disorders and Cognitive Impairments
ASPE is reviewing survey elements for mental health and cognitive impairments that have been used in population-based national surveys. The project aims to review existing measures of mental disorders and cognitive or mental impairments, distinguishing diagnosis from impairment and symptoms in children, adolescents, working-age adults, and the elderly, and to document the validity and reliability of the measures to calculate prevalence. Additionally, coding and classification systems are not optimal for assessing many environmentally related health endpoints:
International Classification of Diseases
Most monitoring programs use the World Health Organization International Classification of Disease (ICD) codes, which is not particularly detailed. Biologically related conditions that nevertheless differ by etiology and pathogenesis might share the same ICD code; relatively few syndromes have a specific code; and many anomalies and conditions that are regarded generally as "minor" (although, as in the case of teratogen-induced conditions, they may be of diagnostic importance) cannot be easily coded. The birth defects system addressed this issue by appending modifiers to existing ICD-9 codes. The workgroup was uncertain to what degree similar issues inhere in other ambulatory coding schemes, HL-7 electronic data transaction standards, and the LOINC nomenclature that may confound efforts to track specific health endpoints.
CDC and ATSDR should assess coding schemes and transaction standards for their ability to capture with precision health effects that scientific evidence suggests may be linked to environmental factors and, where necessary, develop and advocate for necessary refinements.
National Health Interview Survey
The NHIS is one of the most important and often-used instruments in the public health toolbox. Because it serves as the sampling frame for many other federal surveys, it provides one of the few platforms for data linkages. However, a recent regrouping and recoding of health outcomes within the NHIS creates new and significant limitations on its ability to capture diseases and conditions considered priority for the EPHTN (5):
Lung and respiratory diseases: Many of the NHIS recode conditions correlate singly and specifically to specific health outcomes that have an environmental etiology (e.g., asthma) or groups of outcomes with known environmental etiologies (e.g., pneumoconiosis and asbestosis).
Neurodegenerative and neurotoxic disorders: With the exception of "migraine," numerous neurotoxic disorders are not included in the NHIS recodes, resulting in a great potential for underestimation of prevalence of environmentally related neurologic health outcomes.
Reproductive disorders: Neither male nor female infertility was included in the NHIS recodes, and thus prevalence rates can not be ascertained. One recode that does contain outcomes with possible environmental etiology is "delivery and other conditions of pregnancy and puerperium." However, this grouping contains complications ranging from pregnancy and fetal abnormalities affecting the mother to those of legally and illegally induced abortions and conditions arising from perinatal infections.
Surveillance Among Older Adults
The sample frames of household surveys (e.g., NHIS and
BRFSS) consist of noninstitutionalized persons. Approximately 5 % of
people aged 65 and over, and 20 percent of those aged 85 and over, are
nursing home residents and therefore are not included in the sampling
frame of household surveys. As a result, estimates of chronic diseases
that are drawn from household surveys, including respiratory illness and
neurodegenerative disorders such as Parkinson's disease, will
underestimate the occurrence of these conditions. The NHIS and the
Longitudinal Study on Aging used mixed modes of data collection and
reliance on proxy reporters to improve sample coverage. However, more
and better information about this population will require sampling
frames composed substantially of older people (6).
The Medicare Beneficiary surveys and claims files, the Nursing Home and Home Health provider surveys, and data maintained by organizations that focus on the aging population (including state agencies on aging) are valuable linkages and offer opportunity to coordinate information gathering across public health agencies. These surveys should be reviewed for their ability to capture health endpoints of interest and, if necessary, modified.
Laboratory Reporting Practices
Improved laboratory reporting practices facilitated by a developing common electronic reporting format and standardized data elements could offer numerous opportunities to contribute substantial information to the EPHTN.
For example, in New York, and probably in many other states, some cancer care is moving out of the hospital and into private physicians' offices. When diagnostic tests are conducted or ordered by an office-based provider from a free-standing laboratory, and the results are returned to that provider, the tumor registrar in the hospital is not alerted to the cases, and thus they are missed by the common channel of reporting. New York is moving to bring all cancer related laboratory tests from all laboratories under the state's reporting system to supplement the information provided by tumor registrars.
In addition, substantial opportunity exists to capture exposure and disease data lost through underreporting and threshold reporting. Many incidents of disease potentially related to environmental exposures remain largely unreported, and data necessary for good analysis are lost. For example, if readings of all blood lead levels (and other toxins) were made available, instead of only those surpassing thresholds, more realistic exposure patterns could be discerned.
With improved laboratory capacity and data standards, not only could the EPHTN achieve more complete reporting of health conditions, it could in many cases move toward more real time reporting. Thus, such systems can serve the needs of both chronic disease reporting and an early warning system.
CDC and ATSDR should actively and routinely survey the federal government's planned and ongoing studies for appropriate opportunities to integrate environmental health questions.
Level II Recommendations: Aiming Toward the Future
The workgroup noted that HHS has a departmentwide effort, led by ASPE and the HHS Data Council and advised by NCVHS, to integrate data and data systems to better support program planning and accountability. This effort includes an initiative to integrate existing HHS survey instruments. At the same time, the major administrative simplification requirements under the Health Insurance Portability and Accountability Act of 19996 (HIPAA) are effective this year, meaning that health-care transaction data, in both the public and commercial sectors, must conform to specified standards. Finally, despite early resistance by some in the medical community, electronic medical record systems are gradually coming on line. These developments, taken together, represent great potential for public health; they will allow the combining and sharing of data on a scale never before possible and will provide a font of clinical detail as yet unobtainable without painstaking and expensive manual medical record review.
All of these activities will take time to coalesce, even as they continue to evolve; however, the workgroup believed strongly that it is not too soon for public health practitioners to begin doing some early hands-on assessment. In particular, the workgroup believed that the current early standardization requirements for administrative and clinical encounter data, both public and private, offer an opportunity for federal and state public health officials not only to gain valuable experience with health plan data systems– systems that will, because of their breadth and standardization, eventually become extraordinarily powerful epidemiologic resources– but to influence the ongoing evolution of data standards to more precisely meet public health needs. Taken together, Medicare, Medicaid, and health plan and insurer data will be able to provide detailed information for about 85%-90% of a state's population at any point in time. That observation alone suggest that this evolution could overcome many of the shortcomings of existing tracking systems. Moreover, data derived from these sources can be grouped and analyzed in small geographic units. They can yield information about patient demographics; complaints, diagnoses, comorbidities and risk-factors; inpatient, outpatient, and professional service utilization; and laboratory and pharmacy use. Finally, they can capture incidence and prevalence of ambulatory conditions and those for which incidence is too low to be well characterized even by statewide sample surveys.
Tracking programs should focus on the increasing importance of ambulatory settings as sources of data because conditions of public health importance are increasingly being managed in these settings.
The ability to track many priority chronic and environmental health endpoints requires a much greater emphasis on ambulatory care data than is afforded by current public health tools. Only a small fraction of conditions, such as asthma and chronic respiratory illness or endocrine-related conditions, will turn up in hospitals in any given year, and that fraction diminishes as the health system changes and evolves new means and settings to manage nonlife-threatening or chronic conditions. According to some estimates, over 80% of patient visits for asthma occur in the physician's office, as do 60% of all visits related to multiple sclerosis.
Telephone surveys and self-reported questionnaires are imprecise tools for consistently and meaningfully capturing information about these and other conditions. Fortunately, many other sources of ambulatory care data exist, and with the proper input from public health professionals, the movement toward data standardization should facilitate their integration into the EPHTN. These sources include Medicare and Medicaid claims files, provider surveys such as NAMCS and NHMCS, provider-based data warehouses such as those initiated by the National Medical Group Association, Ambulatory Practice Research Networks, EDI of provider claims to state health departments or warehouses, health plan data-sharing agreements, and large commercial data vendors.
The EPHTN should support pilot projects to explore the trade-offs among different approaches to capturing epidemiologic data from the health services domain. For example, state public health officials should be encouraged to develop relationships with health plans or other sources of encounter data for the populations within their jurisdictions.
CDC has already recognized the opportunities in private sector provider and health plan data. It maintains a working relationship with a cluster of the nation's largest, most research-oriented health plans for both epidemiologic and intervention studies on a wide range of health status, clinical, and performance issues.
The workgroup reviewed information about several state efforts to tap health plan and provider information for public health purposes. In Minnesota, for example, a statutory mandate requires data sharing between the state's health plans and the Minnesota Department of Health (MDOH), and as a result of relationships built in the course of implementing the mandate, the state's health plans are now also active participants in community health planning meetings and routine partners in primary and secondary prevention campaigns. One expressly articulated purpose of the data-sharing mandate is "information to assess the prevalence of chronic diseases and other emerging health issues." MDOH has collected fee-for-service and managed-care claims data from Blue Cross and three of the largest HMOs in the state; public program claims data from HHS; and fee-for-service Medicare data from the regional CMS Data Mart. One major challenge noted by the MDOH is the reconciliation of different Third Party Administrator (TPA) practices. Taking a slightly different tack, Wisconsin requires information to be submitted to the state health department directly by providers, thereby avoiding the TPA issue but also the opportunity to enlist plans as partners in prevention. In Rhode Island, relationships and data sharing are evolving slowly under a Health Plan Performance Evaluation mandate and a widely embraced Medicaid managed care/ SCHIP quality assurance program. There are many other examples.
For tracking purposes, the advantages of working with such data include:
- The ability to track the prevalence, costs, and natural history of
conditions that no public health surveillance systems currently
captures, including morbidities of current and emergent interest in
environmental health (e.g., asthma, neurologic impairments, autoimmune
diseases, endocrine effects, and certain adverse reproductive
- Broad population coverage (Medicare + Medicaid + insurers =
85%-90% of the population)
- Refined levels of geographic information
- The ability to access diagnostic data from multiple sources,
including physician offices and clinics, and ability to match patient
- The ability to track comorbidities and other indicators
- The ability to make entities that largely determine health
services organization and delivery a partner in prevention
- The ability to do specific, local studies on conditions of
interest concern, for example by customizing data elements and
- The cost of collecting the data is not borne by the state health department
Additional sources of private sector data include third-party administrators and large data warehouses that support pharmaceutical research and development. For example, the eHealth Initiative (eHI) represents the largest vendors of provider software in the United States, along with a handful of smaller companies. eHI already has approached CDC with a proposal to use this software to support bioterrorism surveillance activities. Similarly, Quintiles/ Synergy is the world's largest data warehouse, capturing claims data from some 4,400 hospitals and 660,000 providers nationwide. It, too, has seen market potential in bioterrorism surveillance and public health applications. Representatives of both systems have testified publicly that their products could be integrated with the National Electronic Disease Surveillance System (NEDSS) architecture. The advantages of these systems are a large denominator and therefore presumed stability in the underlying population and near real time information. However, a national warehouse is less likely to be easily tailored to specific state concerns, to develop relationships with state health department officials, or to have a stake in what its findings suggest are opportunities for prevention and intervention. CDC and ATSDR should explore these trade-offs through state pilot programs under the auspices of the EPHTN.
How does this relate to the RFP?
Applicants should describe how they intend to initiate or expand relationships with the health services sector (e.g., health plans, providers, and data vendors) to access epidemiologic data for populations within their jurisdictions. At a minimum, they should anticipate the ability to collect and assess incidence and prevalence, location, demographic, and other variables related to health conditions of interest, their cost, and secular trends. Applicants should describe how they will develop, in consultation with state environmental protection officials, criteria for queries of such entities and entity reportable conditions, selecting at least two health conditions from the priority list. They should also describe how they will validate preliminary results of such queries against established public health data sources (e.g., registries, surveys, school reports) and agree to provide CDC and ATSDR with detailed reports on the strengths and limitations of such data sources.
The EHTN can help build the next-generation public health tool box by accessing the sources that build on HIPPA and e-Health Initiative momentum. Public health activities need to be positioned to help define and then meet the crest of this wave. Programs in the states should be encouraged to become engaged with and to use these tools early in their development. State programs should also establish a feedback loop to the architects of these tools.
CDC and ATSDR should provide technical assistance to states in developing data-sharing agreements based on lessons learned by other public and private data sharing partnerships in the health services domain (e.g., state Medicaid agencies and their contracted managed-care plans and CDC's own collaboration with the American Association of Health Plans and the HMO group). CDC and ATSDR also should ensure regular feedback from state experiences to HHS data standards groups (including the HHS Health Data Council, the National Center for Vital and Health Statistics, the Center for Medicare and Medicaid Strategies, the Public Health Data Standards Consortium, and HL-7). Finally, on the basis of states' experiences, CDC and ATSDR should advocate within these data standards groups for relevant variables, metrics, and coding practices.
How does this relate to the RFP?
See recommendation 8. Recommendation 9 is directed to the federal agencies.
Level III Recommendations: Develop State Survey and Exposure Measurement Capacity
CDC and ATSDR should pilot a modified "State National Health and Nutrition Examination Survey" with a smaller questionnaire and much larger sample with target oversampling.
Adding geographic information into disease-specific data can help identify where prevention efforts need to be intensified. However, better quality of exposure information is needed to get below ecologic assessments of health and environment interactions. In fact, exposure measurement and assessment is the missing link in efforts to evaluate environmental health risks; without it, public health practitioners are hard-pressed to answer fundamental community concerns.
With the exception of childhood blood lead screening, few systematic efforts have tracked individual levels of exposure to any hazardous substance. Although CDC and EPA have developed the methodologies for biologic monitoring of a wide range of substances, their application and availability have been limited. A recent report of the U. S. General Accounting Office called for a long-term coordinated strategy to measure exposures to pollutants. It specifically called for coordination between the biomonitoring component of CDC's NHANES and EPA's National Human Exposure Assessment Survey.
For these reasons, the workgroup recommended that CDC and ATSDR also consider tracking proposals that would strengthen primarily biomonitoring systems for EPHTN priority hazards.
How Does this Relate to the RFP?
Level III (biomonitoring) applications should address the development of analytic techniques and design and pilot a sampling strategy that would reflect the exposure status for the population of the state or a target subpopulation (e.g., children). The value of the proposal should be assessed based on the breath of impact and potential application to other states. Highest priority should be given to states that propose to link biomonitoring data with health data systems and/ or environmental data bases.
CDC, ATSDR, and EPA should provide technical assistance to states in selecting hazards and health effects, appropriate metrics, survey design, biomonitoring protocols, and interpretation of results. In addition, CDC and ATSDR should assist states in planning for dual use of public health laboratory enhancements supported through bioterrorism funds for population-based screening.