Two of the 8 pilot indicators in the community setting — clean indoor air laws and smoking in the home — have readily available data sources (Table 2).

The legislative database in the State Tobacco Activities Tracking and Evaluation (STATE) system summarizes state tobacco legislation, including smoke-free indoor air ordinances for restaurants, day care centers, and public places (4,5). The Office on Smoking and Health at the CDC maintains the database, based on a quarterly search of the LexisNexis legal database (4,5). The database can be used to monitor the presence or absence of state policies and the content of those policies (e.g., restrictions, penalties, enforcement). The legislative database, however, does not capture municipal ordinances that might be enacted in the absence of state policies. Beginning in 1998, the optional Tobacco Indicators module of the annual Behavioral Risk Factor Surveillance System (BRFSS) asked respondents if anyone smoked anywhere in their homes. In 2001, this was changed to ask if smoking was allowed in their homes (6). The Tobacco Indicators module was used by 25 states in 2002.

Data sources also are available that partially measure 2 other community indicators: highway funding of transportation alternatives and the number of farmers' markets. The National Transportation Enhancements Clearinghouse maintains a database of transportation enhancements funds allocated and spent by each state under the Transportation Equity Act for the 21st Century (TEA-21). This searchable, on-line database is updated annually (7). Funds for transportation alternatives under TEA-21, however, do not represent the entire state budget for transportation alternatives, and the database does not include the total amount of the state transportation budget. The research teams found no additional data sources that provide relevant details on highway spending at the state or local level.

The USDA maintains a list of farmers' markets searchable on-line by state (8). The database depends on reports from individual state departments of agriculture. Because the definition of a farmers' market varies by state, the data might be inconsistent or incomplete across states. For example, at the time of this study, the South Carolina listing included only 3 state-run, year-round farmers' markets. The list was recently updated to include smaller local markets that operate on a seasonal basis.

Although regional milk production figures are available, no state data were found on milk production or sales. The research teams also noted that this indicator is not a measure of environment or policy but a community-level indicator of purchasing behavior.

Ten pilot indicators for heart disease and stroke prevention were identified in the school setting (Table 3). Seven indicators that refer to state policies on physical education requirements, student physical education assessments, food availability, certifications for food service staff and physical and health education teachers, and health education curriculum have readily available data sources.

All 7 of these indicators can be assessed using data from the School Health Policies and Programs Study (SHPPS), which is conducted every 6 years. The study surveys all state departments of education and a nationally representative sample of districts and schools (11). The state survey includes questions related to each of the 7 school indicators. These indicators assume that such policies are enacted at the state level; however, in states like South Carolina and Alabama, school policies are under the authority of school districts or the schools themselves.

The School Health Education Profile (SHEP) collects data that provide partial measurement of school health councils and tobacco-free schools. SHEP is a survey completed every 2 years by a sample of school principals and lead health educators in public schools containing classrooms at the sixth-grade level or higher (12). Because no similar data source is available for elementary schools, SHEP can only partially measure these indicators. In addition, the survey does not currently include questions that lead to the assessment of all components of the tobacco-free school policies recommended by the CDC.

Only one of the 8 pilot work site indicators — clean indoor air laws for work sites — has a readily available data source (Table 4). Neither research team found any data sources for other work site indicators.

The STATE system contains information that measures state clean air laws that apply to work sites (4,5). This indicator is subject to the same sensitivity concerns previously noted for other clean indoor air laws — it notes only the presence or absence of state policies. The BRFSS optional Tobacco Indicators module collects information from individuals about their work site tobacco policies, but it does not measure state indoor air laws. Data on work site policies collected by the optional module would provide an estimate of the percentage of employed adults protected by a work site smoking policy.

Questions from the National Worksite Health Promotion Survey could be used to assess on-site physical activity programs and nutrition or weight management programs (13). This survey collects and provides national data for Healthy People 2010 (14). The sample is too small, however, to draw conclusions by state. Other measurement tools assess policies and environmental characteristics related to heart disease and stroke prevention within work sites, including Heart Check (15) and the Checklist of Health Promotion Environments at Worksites (16). However, these instruments are not commonly used across the country and are not designed to be used as surveillance tools.

Among the 5 pilot indicators identified in the health care setting, only one has a readily available data source: smoking cessation advice delivered by health care professionals (Table 5). The proportion of smokers who received advice to quit smoking in the past year has been included in the optional Tobacco Indicators module of the BRFSS since 2000.

Given the overall lack of data in most settings assessed in this study, consideration should be given to designing and implementing new data collection processes. Vehicles for new data collection efforts are likely to be surveillance efforts now supported by the CDC (e.g., BRFSS, Youth Risk Behavior Surveillance System, SHPPS, SHEP). The SHPPS and SHEP are designed to collect policy data and are updated regularly to include more complete information. For example, SHEP 2002 included questions related to 2 school indicators: the percent of schools that provide health education instruction that includes the physical education topics listed in CDC's School Health Index and the proportion of schools that have adopted tobacco-free policies that meet CDC recommendations (20,21). Although the BRFSS is an individual-level surveillance tool, the optional Tobacco Indicators module already allows states to collect data to measure 2 indicators indirectly (smoking in the home and receiving advice to quit). Because this module is optional, the data are not available in all states. The availability and variability of relevant data across states can have important implications for achieving consistency within a national surveillance system for heart disease and stroke prevention. This study, however, did not explore a sufficient number of states to determine the extent of this variability.

Systems similar to the legislative database of the STATE system could be developed to monitor other state policies. In fact, in late 2003, the CDC Division of Nutrition and Physical Activity launched an on-line searchable database containing bill information related to physical activity and nutrition from all 50 states (22). Few existing national surveillance efforts, however, gather information from local governments, work sites, and health insurers. Important issues of cost — in terms of time, personnel, financial resources, and participant burden — must be considered when developing new data collection efforts or revising existing systems.

Although the research teams made extensive efforts to consult with a wide range of organizations, other data sources might exist. The research teams restricted their exploration to data that are collected either nationally or within their states. While this project did not complete an exhaustive review of data sources in other states, it did identify some noteworthy examples, such as New York's Heart Check (15). Additional surveys developed by other states (e.g., Montana, North Carolina) can be found on the Cardiovascular Health Council of the Chronic Disease Directors Web site: http://www.chronicdisease.org/cvh_council/Key%20Elements/ State%20Survey/CVH_state_survey.htm. The mechanisms illustrated at this site can serve as models for other states.

An additional challenge of data collection is assessing the impact of policy and environmental changes on behavior and health. Policy and environmental indicators provide only one part of the equation. For example, assessing the impact of school policies on children's behavior presents challenges in obtaining informed consent from the children, school administration, and/or parents.

To be useful to state programs, indicators for heart disease and stroke prevention examined in this study need to be refined to improve specificity and sensitivity. Including clear definitions would improve the specificity of the indicator and the accuracy and consistency of data collected. Sensitivity for many indicators could be enhanced by establishing criteria for evaluating policies and laws beyond consideration of their presence or absence at the state level. Some data sources like STATE and SHPPS already collect detailed information that could be used to evaluate the content and quality of policies in addition to tracking their presence or absence.

While it may be sufficient to look at states' policies for national surveillance, state programs might need additional surveillance data that show progress in meeting prevention goals within their own states. In some cases, particularly within school and community settings, it might be more relevant — albeit more costly — to assess the percentage of local jurisdictions (counties, municipalities, school districts) that implement a given policy.

The health care indicators provide the greatest challenge for surveillance. As worded, the indicators look at the percentage of insurers that provide a specific type of coverage. Knowing this information might not reflect the percentage of the population covered by those companies. For example, South Carolina currently has only 5 health maintenance organizations, which cover less than 10% of the state's population (23). Even if data indicated that all of these organizations followed the recommended guidelines, the data would not include 90% of the South Carolinians who might or might not have coverage under some other type of health care plan. In addition, insurance companies tend to negotiate with individual employers about the content of health insurance plans rather than having standard plans. Nationally, employers provide coverage for 58% of the population (23). If employer surveys are developed for other work site indicators, these surveys could include questions about health insurance provided by the employers.

The results of this investigation support the need for more attention, resources, and research to provide a consistent, documentable system for measuring indicators for heart disease and stroke prevention. It also will be important to improve the sensitivity and specificity of each indicator and to evaluate how each indicator corresponds to risk factors and health outcomes. These recommendations are consistent with the new Public Health Action Plan to Prevent Heart Disease and Stroke, which recommends enhancing data sources and systems to monitor key indicators for heart disease and stroke and "to systematically evaluate policy and program interventions" (24). Currently, the CDC is funding other projects to refine and validate these and other potential indicators for heart disease and stroke. With the evolving importance of policy and environmental factors influencing primary and secondary prevention efforts in public health, it is vital that a system be developed that will provide national, state, and possibly local data on indicators for heart disease and stroke. During the next decade, these indicators could provide valuable measurements to determine how environmental and policy changes are affecting heart disease and stroke prevention in this nation.