Mississippi Behavioral Risk Factor Surveillance
It is generally agreed among health care professionals that certain conditions and behavior patterns are associated with disease, injury and death. Some examples are cigarette smoking, physical inactivity, obesity, alcohol consumption and risky sexual behavior. The Behavioral Risk Factor Surveillance System (BRFSS) is a telephone survey designed to estimate the prevalence of these and other health risk factors in all states in the United States. The results provide a tool for evaluating health trends, assessing the impact of chronic disease, along with measuring the effectiveness of policies, programs, intervention strategies and awareness campaigns.
The BRFSS is a cooperative agreement between the Centers for Disease Control and Prevention (CDC) and the Mississippi State Department of Health (MSDH). The first survey was completed in 1984 when the data was collected at one given point in time. The survey was repeated in 1988 using the same methodology. Beginning in 1990 there has been an annual survey with the data being collected monthly.
The BRFSS survey contains a set of core questions provided by the CDC to gather comprehensive standard information nationwide. The questions are related to health status, access to health care, health awareness, lifestyle, preventive health and chronic health conditions. Individual states are allowed to include optional module questions addressing specific issues that may be of particular interest to that state.
A. Sampling design
The Mississippi BRFSS is a random sample telephone survey. Utilizing a disproportionate stratified sample (DSS) design with random digit dialing and the Computer Assisted Telephone Interviewing (CATI) system, the survey has the potential to represent all households in Mississippi that have telephones which according to the United States Census Bureau, Housing and Household Economic Statistics Division is approximately 94.5 percent. A sample size of between approximately 4,000 and 5,000 persons is selected each year to give a 95% confidence interval of ±3% or less on risk factor prevalence estimates of the adult population. Prevalence estimates by individual demographic variables, comprising smaller sample sizes, do not achieve the same level of accuracy as the total sample.
Until the 2011 survey, the BRFSS had relied exclusively on interviews of households with only land line phones. But the number of households with only cell phones increased by more than 700 percent between 2003 and 2009. Approximately three in ten American homes now have only cellular telephones; in Mississippi the rate is 35.1 percent. This trend has been especially strong among younger adults and those in social and ethnic minority groups. The 2014 Mississippi BRFSS had approximately 62 percent land line and 38 percent cell phone households in the survey and the proportion of cell phones in the survey is being increased each year to account for the increasing prevalence of cell phone only households.
For land line surveys, interviewers, contracted by the MSDH, contact the residences during weekdays between 9:00 a.m. and 9:00 p.m. and Saturdays between 10:00 a.m. and 4:30 p.m. After a residence has been contacted, one adult (18 years of age or older) is randomly selected to be interviewed from all adults residing in the household. The majority of interviews are collected over a two-week period each month of the survey year.
For cell phone surveys, the same protocol is followed except that the interviewer establishes that the person answering the phone is at least 18 years old, that it is safe for the respondent to be interviewed and that the person uses the cell phone for at least 90 percent of their telephone service.
The questionnaire, designed through cooperative agreements with the CDC, contains questions from three categories. The first category contains questions on health status, access to health care, health awareness, health risk behaviors, and preventive health; the second contains demographic information (age, sex, race/ethnicity, education, annual household income, and employment status); and the third contains questions addressing specific issues that are of particular interest to the state conducting the survey.
C. Data Analysis
The other significant change that has been introduced to the BRFSS is a new weighting method called iterative proportional fitting, also known as "raking." The procedure, while not new, has been made feasible through the development of ultra-fast computer processors. The new weighting methodology began with the 2011 survey.
In addition to the standard age, gender, race and ethnicity variables, the use of raking allows for consideration of demographic variables such as education level, marital status, renter or owner status, and phone source. Inclusion of these additional variables in the weighting process allows the survey to more accurately reflect Mississippi's adult population. The data collected by the MSDH Office of Public Health Statistics are edited and weighted by the CDC. Weighted counts are based on the most recent available Mississippi population estimates to accurately reflect the population demographics.
Therefore, the estimated prevalence of any risk factor from the survey represents the total population of Mississippi residents 18 and older very well. The reader should be aware that the numbers presented in the tables of the reports reflect the actual, non-weighted observations for each cell while the percentages in each cell represent the weighted prevalence.
Reports on this site present the weighted percentage of high-risk behaviors, conditions and certain chronic diseases by gender, age group, race, education level, annual household income, and employment status. Respondents who either refused to answer or did not know the answer to the questions on demographics were excluded from the tables. For this reason the total for each of the demographic sections may not be equal to the total for the entire table.
D. Limitations of the Data
All data collection systems are subject to error, and records may be incomplete or contain inaccurate information. All information in this survey is self-reported; people may not remember essential information, a question may not mean the same thing to different respondents, and some individuals may not respond at all. It is not always possible to measure the magnitude of these errors or their impact on the data. The user must be the final arbiter in evaluating the data.
E. Sample Size
Sample size in the BRFSS survey from year to year is largely determined by available funding. The reader should note that sample sizes by question and response category may vary because of non-response and skip patterns within the survey instrument. Overall estimates generally have relatively small sampling errors, but estimates for certain population subgroups may be based on small numbers and have relatively large sampling errors. Interpreting estimates that are based on small numbers can mislead the reader into believing that a given finding is more precise than it actually is. When the number of events is small and the probability of such an event is small, considerable caution should be observed in interpreting the estimates or differences among groups. The BRFSS recommends not interpreting percentages where the denominator is based upon fewer than 50 non-weighted respondents. In the tables of reports on this site, such results are marked with an asterisk that indicates a sample size less than 50 for the particular cell in the table.