artigo da base de dados dada.pdf.crdownload

OBJECTIVES AND DESIGN OF THE BEHAVIORAL RISK FACTOR SURVEILLANCE SYSTEM David E. Nelson, Deborah Holtzman, Mike Waller, and Craig Leutzinger for the Behavioral Surveillance Branch, Centers for Disease Prevention and Control (CDC); Ken Condon, Texas Department of Health...

OBJECTIVES AND DESIGN OF THE BEHAVIORAL RISK FACTOR SURVEILLANCE SYSTEM David E. Nelson, Deborah Holtzman, Mike Waller, and Craig Leutzinger for the Behavioral Surveillance Branch, Centers for Disease Prevention and Control (CDC); Ken Condon, Texas Department of Health David E. Nelson, CDC Key Words: Population surveillance, health surveys, risk factors L INTRODUCTION By the early 1980s, it was evident from scientific encouraged by CDC to use cluster designs based on the research that personal health behaviors played a major role Waksberg method; however, even for the initial 29 point- in premature morbidity and mortality, but no data were in-time surveys, there was state variability; nine of these available on a state-specific basis. This was viewed as a states used simple random samples. Currently, the vast critical deficiency, since state health agencies have the majority of states use disproportionate stratified sampling primary role of targeting resources to reduce behavioral (DSS). risks and their consequent illnesses, and national data may One important characteristic of the BRFSS is its not be appropriate for any given state. flexibility. It permits states to add questions of their own At about the same time, telephone surveys design, but is uniform enough to allow state-to-state emerged as an acceptable method for determining the comparisons for certain questions. Participating states use prevalence of many health risk behaviors. Telephone these data for many purposes. Among these are to identify stu-veys were especially desirable at the state and local level, demographic variations in health-related behaviors, target where the necessary expertise and resources for conducting services, address emergent and critical health issues, area probability sampling for in-person household propose legislation for health initiatives, and to measure interviews were not likely to be available. As a result, progress towards state and national health objectives. The surveys were developed to monitor state-level prevalence of system's broad network for information gathering also the behavioral risks associated with premature morbidity enables states to evaluate their disease prevention and and mortality among adults. Behavioral.data were thought health promotion efforts. to be especially useful for planning, initiating, supporting, and evaluating health promotion and disease prevention Keyfeatures of the BRFSS: programs. Initial point-in-time state surveys were conducted with small resources (federal awards to states in 29 states from 1981-1983. In 1984, the Behavioral Risk average about $60,000), data are continuously Factor Surveillance System (BRFSS) was established by collected in all 50 states, DC, and one territory CDC, with 15 states participating in monthly data data remain available to states collection. Although designed to collect state-level data, a flexible system that has avoided becoming number of states from the outset stratified their samples to tradition-bound and more rigid with growth allow them to estimate prevalence for regions within their timely and relevant data respective states. By 1994, all 50 states and the District of Columbia were participating; as of 1998, Puerto Rico also II. QUESTIONNAIRE was collecting monthly data and the Virgin Islands and Guam were conducting point-in-time surveys. Criteria for including items in the BRFSS A standard questionnaire was developed at CDC questionnaire: for states to use to provide data that could be compared Relationship of the variable to personal behaviors across states. The initial survey primarily included existing linked to promoting health, preventing disease, questions from national surveys such as the National Health and/or reducing health risks Interview Survey. The basic questionnaire was designed to Suitability of the question for telephone last no more than ten minutes so that states could add their interviewing own questions. In general, the BRFSS gathers information - Pertinence of the variable to national health on health behaviors related to the leading preventable objectives or other priority health issues causes of death, including physical inactivity, injury, weight Need to measure the variable over time control, alcohol consumption, tobacco use, and HIV-AIDS. Need to have state-specific data It also collects data on preventive health practices such as Degree to which alternative data sources are mammography use. unsatisfactory For data collection, states were historically Degree to which the prevalence of the variable 214  will be adequate for planned analyses optional modules remain unchanged throughout the year. Relationship of the variable to other questionnaire However, the flexibility of state-added questions does topics permit additions, changes, and deletions at any time during Validity of questions the year for these questions. Financial and/or technical resources available for The core component is asked first, optional support of the question modules are asked next, and state-added questions last. If Effect on questionnaire length, considering both a significant portion of the state population does not speak the total number of questions and the proportion English, states have the option of translating the of respondents to be queried questionnaire into other languages; presently,.CDC provides a Spanish language version of the questionnaire. The questionnaire has three parts: 1) the core CDC provides states with CATI programming, edit component, consisting of the fixed, rotating, and emerging programming, and annual data tables for questions on the core; 2) optional modules; and 3) state-added questions. coreand for selected optional modules. Core component. Thefixed core is a standard set Before 1993 the content of the core and optional of questions asked by all states. It includes queries about components of the questionnaire was determined by CDC current behaviors that affect health (e.g., tobacco use, and the states one year at a time. Each year, changes were alcohol consumption) and questions on demographic proposed, debated, and agreed upon. A long-term plan for characteristics. The rotating core is comprised of two the content of the questionnaire was adopted in 1992 and distinct sets of questions, each asked in alternating years by implemented in 1993. The plan divided the core into a all states, that address different topics. The emerging core fixed and a rotating section. The topics included on the is a set of up to five questions that are added to the fixed and fixed and rotating cores are shown in Table 1. rotating cores. Emerging core questions typically focus on In years that rotating topics are not used in the issues of a "late breaking" nature and do not necessarily core, they are supported as optional modules. At least once receive the same scrutiny that other questions receive prior a year staff within the Behavioral Surveillance Branch to being added to the instrument. These questions are part convenes a meeting of state and CDC program staff (the of the core for one year and are evaluated during or soon BRFSS Working Group) to discuss the questionnaire and after the year concludes to determine their potential value in other issues related to the BRFSS. The long-term future surveys. questionnaire plan, the content of the core, and the optional Optional CDC modules. These are sets of modules are reviewed annually by CDC and the Working questions on specific topics (e.g., smokeless tobacco, Group. Based on recommendations from the BRFSS arthritis) that states elect to use on their questionnaires. Working Group, a draft questionnaire and proposed State-added questions. These are questions optional modules are prepared for review and discussion by developed or acquired by participating states and added to all BRFSS coordinators at the annual BRFSS conference. their questionnaires. CDC reviews this input, consults with the BRFSS Working Group, arranges for cognitive testing of new or substantially Each year, states and CDC agree on the content of revised existing questions, field tests a prototype the core components and optional modules. For questionnaire, and releases to the states the final draft of the comparability, many.of the questions are taken from core and optional modules by September of each year. established national surveys. This allows the BRFSS to take advantage of questions that may have been tested and IH. STATE COORDINATORS AND allows states to compare their data with those from other INTERVIEWERS surveys. BRFSS protocol specifies that all states ask the Historically, as new states joined the BRFSS core component questions without modification; they may system, start-up training was provided by a CDC Project choose to add any or none of the optional modules; and Officer with considerable experience with the BRFSS. states may add question(s) of their choosing at the end of the Once start-up training was provided, the state assumed questionnaire. New questions on optional modules and on responsibility for training new coordinators and other staff. the fixed, rotating, or emerging core are required to have CDC project officers supplement state-provided training for undergone cognitive testing. new coordinators with briefings during site visits. Although CDC supports about 18 optional Interviewer retention is very high among states modules annually, it is not feasible for a state to use them that conduct the survey using state health department staff. all. States are selective with their choices of modules and When a new interviewer is hired, the Coordinator or a state-specific questions to keep the questionnaire at a supervisor usually conducts the training. This material reasonable length (total number of questions used in states covers seven basic areas: overview of the BRFSS, role ranges from 90 to 160). New questionnaires are descriptions for staff involved in the interviewing process, implemented in January, and the core components and the questionnaire, sampling, codes and dispositions, survey 215 TABLE 1: BRFSS TOPICS FROM 1993-2000 Fixed Core Rotating Core I Rotating Core II (1993, 1995, 1997, 1999) (1994, 1996, 1998, 2000) Topic No. Questions Topic No. Questions Topic No. Questions Health status 4 Hypertension 3 Physical activity 10 Health insurance 3 Injury 5 Fruits & vegetables 6 Routine check-up 1 Alcohol 5 Weight control 6 Diabetes 1 Immunizations 2 Smoking 5 Colorectal screening 4 Pregnancy 1 Cholesterol 3 Women's health 10 HIV/AIDS 14 Demographics 14 Total: Women 53 Total: 22 Total: 22 Men 42 follow-up, and practice sessions. rely heavily on students are subject to greater turnover. Contractors typically use interviewers who have experience conducting telephone surveys. These IV. BRFSS SAMPLE DESIGN interviewers are given additional training on the BRFSS Population. The target population for the BRFSS questionnaire and procedures before they are certified to is the non-institutionalized civilian population age 18 years work on BRFSS. When a state has a new contractor who and older with telephones in each participating state or has not had prior experience with BRFSS, a CDC Project territory. Officer participates in the initial training of the contractor's Coverage. Telephone coverage is known to be at interviewing staff. least 95% in the U.S., but is lower for some groups, The format of ongoing training for interviewers including minorities and those with lower socioeconomic varies, and is determined by states. Some states meet with status. No direct method of compensating for non- their interviewers to discuss procedures at the beginning of telephone coverage is employed by the BRFSS. Post- every monthly interviewing period; others have a formal stratification weights by age/race/sex categories are used in quarterly meeting. CDC does not specify skill requirements the BRFSS, and this may partially correct for any bias for state coordinators. About 60% of coordinators have caused by non-telephone coverage. either masters or doctoral degrees. BRFSS Samples. The BRFSS surveys in each States are required to conduct call. back state and territory employ random digit dialing (RDD) verifications on a 5% sample of completed interviews. In methods of sampling. Specific sampling methods vary addition, they are expected to monitor interviews. Thirty among states as indicated in Table 2. An important tenet states currently have systems connected to the telephones of BRFSS sampling methods is that each residential that enable monitoring of the respondent and the telephone number should have a known, non-zero interviewer; ability to practice electronic monitoring varies probability of being in the sample. The BRFSS uses a because of state laws. Contractors typically conduct rigorous, standardized protocol that requires up to 15 call somewhat more systematic monitoring. Each interviewer is attempts to reach randomly selected adults. monitored a certain amount of time every month. The Interviews are conducted each month of the year, monitoring is usually documented and shared with the usually during a two week period. Each state has a target interviewer. States that conduct in-house interviewing are number of interviews, currently ranging from about 125- usually not as systematicl BRFSS coordinators typically 405 per month, yielding annual state samples ranging from listen to interviewers at random and provide feedback if a about 1500-4860, with a total annual sample of problem is identified. Turnover among BRFSS approximately 125,000 across all states. States that are interviewers is minimal compared to the turnover of interested in making estimates for sub-state areas may interviewers in general. University-based contractors that sample at different rates in particular stratum to ensure a 216 minimum sample size per stratum. As shown in Table 2, 3 characteristics of the data collection facility, problems of the 52 areas (mostly states) are using paper observed during monitoring of interviews, verification questionnaires; in the others, interviewing is done using procedures, interviewer/supervisor training in the past 6 computer assisted telephone interviewing (CATI). Of the months, number of callbacks to a single number before final 52 project areas participating in the 1998 BRFSS, 30 disposition, handling of refusals; use of data, and plans for employed contractors that are either universities or follow-up. commercial survey research groups. CDC also produces a number of quality measures that are shared with states, including surs'ey efficiency, the number of states withelectronic monitoring (in 1997, 30 Table 2. 1998 BRFSS Sampling Design Characteristics states used either audio or video monitoring), discrepancy between percent of sample that was female and census Total states/territories" 52 estimates (median for 1997:4.8 percentage points; range: Sample Designs: 0 to 11.0), and item nonresponse for demographic and other Waksberg cluster 11 selected variables. To date, values of these quality Disp. Strat. Sample 38 assurance data items have almost uniformly been high. SRS/Strat. random 3 Finally, many methodologic studies have examined State Interviews/month 125-405 the reliability and validity of the BRFSS. The Behavioral State Interviews/year 500-4860 Surveillance Branch maintains a bibliography that Total interviews (est. for all states) 125000 summarizes the more than 30 BRFSS methodologic studies Data collection method: (list available from the author); the vast majority CATI 49 demonstrate very high reliability and validity for BRFSS Paper 3 data. Personnel used for interviewing: State health department staff 22 VII. DATA DISSEMINATION Contractors 30 The Behavioral Surveillance Branch has a policy to provide timely access to BRFSS data to as diverse an array "includes District of Columbia and Puerto Rico of public health professionals as possible, while, ensuring data quality and respecting the needs of participating states. Specifically, the Branch provides BRFSS data that have been edited and are ready for statistical analysis with Response Rates. As found in other telephone weights and uniform variable formats. The first priority is surveys, response rates vary by state, and rates have been to provide states access to their own data as rapidly as declining in recent years. Median BRFSS CASRO possible. response rates were 71% from 1991 to 1993 but declined After all states have had an opportunity to review their to70% in 1994, 68% in 1995, and 63% in 1996, and 62% own data, the BRFSS data for all states are made available in 1997 (CASRO rate in 1997 ranged from 37% to 89%). for internal CDC use, usually 6 months after the end of each Participation rates (the percentage of persons reached by calendar year. BRFSS data are then made publicly telephone who agree to be interviewed) declined from 84% available through publication of an annual summary report, in 1991 to 78% in 1997. posting data tables on the Internet, and releasing data on CD-ROM products. VI. QUALITY ASSURANCE Oversight by CDC of state BRFSS activities is VIII. SUMMARY done through a combination of visits by CDC staff to states, The BRFSS is a unique data system developed development of written policies on data quality standards collaboratively by CDC and states to provide ongoing and monitoring compliance, and calculating and publishing surveilllance of adult health risk factors and other health- quality assurance measures. related factors on a monthly basis. The BRFSS has a strong State site visits are conducted by CDC project history of producing valuable and valid, high quality data. officers approximately every 12-18 months. Typically, With increased emphasis and need for data at the state and project officers are public health advisors who have had even the local level, the BRFSS has become an even more experience as state public health program managers. valuable resource. Increased efforts are being made by During these visits, project officers use the following list of CDC to provide ready access to BRFSS and to encourage items to assess state BRFSS operations: response rates, more analysis and use of these data. mode of data collection, sample size and strategy, review of optional modules and state-added questions, interviewing, 217  REFERENCES Marks JS, Hogelin GC, Gentry EM, Jones JT, Gaines KL, Forman MR. The behavioral risk factor surveys: I. State- specific prevalence estimates of behavioral risk factors. Am J Prev Med 1985; 1:1-8. Remington PL, Smith MY, Williamson DF, Anda RF, Gentry EM, Hogelin GC. Design, characteristics, and usefulness of state-based behavioral risk factor surveillance 1981-1986. Public Health Rep 1988;103:366-375. Powell-Griner E, Anderson JE, Murphy W. State- and sex- specific prevalence of selected characteristics--Behavioral Risk Factor Surveillance System, 1994 and 1995. In: CDC Surveilla!a.ce Summaries, August 1, 1997. MMWR 1997;46(No. SS-3): 1-31. Centers for Disease Control and Prevention. Behavioral Risk Factor Surveillance System Guide. Atlanta, GA: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, 1998. Centers for Disease Control and Prevention. Behavioral Risk Factor Surveillance System 1997 Summary Prevalence Report. Atlanta, GA: Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promtion,.Division of Adult and Community Health, Behavioral Surveillance Branch, 1998. 218