Surveillance research in tobacco control should monitor and evaluate trends in tobacco use and related cancer risk factors, health services, and policy and environmental interventions to determine the influence of these factors on trends in cancer incidence, morbidity, mortality, and survival. INTRODUCTION A comprehensive and integrated program of surveillance research is an essential component of a complete tobacco control research portfolio, and must address a wide spectrum of planning, implementation, and evaluation needs. Researchers, policymakers, health care planners, and community organizers require data on the use of tobacco products, changes in tobacco use, and the factors that influence such behaviors in order to plan, develop, and implement more effective tobacco control interventions and policies. Ideally, a tobacco surveillance system should systematically monitor and analyze tobacco-use behaviors and changes in those behaviors in relation to a variety of social and environmental influences. A well-designed, integrated approach to data collection and analysis should go beyond conducting surveys about current tobacco use and how use changes over time. It also must assess how social and environmental factors interact with tobacco control initiatives and tobacco industry activities to influence tobacco use in various populations. Today's rapidly shifting societal attitudes toward tobacco use, along with new state and community legislative and policy initiatives, have created a unique opportunity to establish a tobacco surveillance system. Such a system should monitor the impact of these influences on tobacco use and provide the critical information tobacco control efforts need to succeed. RECOMMENDATIONS Research should be conducted on expanded surveillance systems to monitor tobacco use behaviors, the implementation and fidelity of tobacco-related interventions, and other factors that influence tobacco use. Tobacco surveillance currently consists primarily of (1) routinely measuring tobacco use (current, former, never) using questions added to existing national or state surveys, (2) intermittently administering comprehensive, specialized tobacco surveys, and (3) compiling tobacco sales data from federal or state excise tax receipts. Federal surveys that are conducted regularly measure current smoking status for a given population at a given point in time. While this information is important and should continue to be collected, national surveys should expand to follow individuals over time to assess the determinants of and changes in smoking behavior longitudinally. Such cohort surveys would enable researchers to assess population trends, and are critical for both evaluating interventions and understanding why changes in tobacco use did or did not occur and what factors are influencing trends at the national, state, or local level. Two national surveys (the Adult Use of Tobacco Survey and the Teenage Attitudes and Practices Survey) have provided detailed information about adult and teenage tobacco use behavior. However, they have not been conducted frequently enough to meet the needs of researchers, policymakers, and public health planners. New surveillance measures should delineate the biological, behavioral, and social influences underlying tobacco use. Special surveys and oversampling should be used, where needed, to collect data on teenagers, low-income adults, and other high-risk groups. Findings can be used to evaluate existing tobacco-control approaches and refine or design new ones where needed. For youths, monitoring and analyzing the social and environmental factors that lead to tobacco experimentation and subsequent addiction are needed to develop more effective tobacco-use prevention approaches. Data on attitudes and beliefs about low tar and nicotine tobacco products and dose-response effects and beliefs about nicotine addiction that result from using these products are needed to develop targeted behavioral and pharmacological approaches for all tobacco users. Additionally, expanded data on attitudes and beliefs about "smoke-free" accommodations and workplaces are needed to assess public opinions about tobacco policy initiatives and whether these policies reduce tobacco use in the population. As tobacco-control initiatives increasingly move to the state and local levels, we need to expand capabilities to monitor the dissemination, quality, and outcomes of those efforts. This includes continuing to track state and local tobacco control legislation and evaluating whether this effort needs to be expanded. The NCI should continue to fund Tobacco Use Supplements to the Current Population Survey (CPS), conducted by the Census Bureau, to track tobacco issues, such as prevalence, policies, and other measures, for the individual states. The CPS tobacco use questionnaire should be modified to incorporate new measures of tobacco-use behaviors and new supplements should be commissioned to cover new and emerging areas of importance. The NCI should build on its leadership in maintaining the State Cancer Legislative Database (SCLD) and continue to define and track measures of state-level tobacco policy implementation. Ultimately, the NCI should assess the most appropriate location for the SCLD within the NCI. Moreover, there should be discussion of how to integrate the NCI and CDC legislative databases for maximum efficiency and cost-effectiveness. The comprehensive and integrated program of surveillance research required to serve the needs of a broad range of organizations and researchers must ensure the consistency of data measured over time and completeness of measurement. Research collaboration is needed to develop a common set of basic core measures to allow comparisons of local, state, and national surveys. The collaboration should include researchers and activists working at these levels. Furthermore, standard definitions of what constitutes a "smoker" and the age range of teenagers for measurement purposes needs to be agreed upon and remain consistent over time in order to permit accurate tracking of trends in tobacco-use behaviors. The fidelity of tobacco-related interventions to their original design needs to be monitored when interventions are translated beyond their original research settings. Technical assistance to help modify interventions to adapt them to the new research settings should also be available. One of the greatest barriers to evaluating tobacco control interventions has been the absence of any ongoing monitoring of how these strategies are being applied (process evaluation). Hundreds of research papers have been published documenting the effectiveness (or lack thereof) of various interventions in controlled settings. However, when these interventions are moved from a research environment to the community, monitoring and evaluation often ceases or is insufficient. For example, while comprehensive school health programs that include tobacco prevention reduce youth smoking, it is commonplace to find that schools are not adhering to the program as it was originally designed and implemented only 3 to 5 years after a recommended curriculum is adopted. Comprehensive surveillance mechanisms are critical to evaluate both the processes and the outcomes of interventions directed at special populations. Cost-effectiveness of these programs should also be analyzed and data should be collected for that purpose. While it is not economically feasible, practical, or necessary to monitor all tobacco industry activities, the tobacco-control surveillance systems should be expanded to document and track industry counteractivities and their possible effect on program success. Newly released tobacco industry documents are expected to provide information to help guide this effort. Surveillance should track such tobacco industry strategies as new advertising campaigns, themes, promotions, and legislative lobbying efforts as well as changes in product packaging, design, pricing and positioning. OTHER SURVEILLANCE RECOMMENDATIONS It is important to regularly synthesize tobacco control research findings. For example, resources should be committed to summarizing what we have learned from interventions directed at low-income and minority smokers. Syntheses of research findings in major areas of tobacco control are an important part of surveillance and build the base of scientific knowledge that is needed to advance tobacco-control efforts. Such information is needed for both practitioners and researchers. This has rarely been done in the past. THE IMPACT OF SURVEILLANCE RESEARCH Surveillance research is a necessary and integral component of a complete tobacco research portfolio. Much of the research recommended in this report could not be carried out without a comprehensive surveillance system. By providing the critical information needed to understand tobacco-use behaviors and evaluate all aspects of tobacco-control interventions in community settings, a comprehensive, coordinated, and specialized tobacco-control surveillance research program would promote and enable development of a more comprehensive, stronger, lasting, and effective tobacco-control program.