Alaska Behavioral Risk Factors Surveillance System (BRFSS) Health Profiles
The Alaska Behavioral Risk Factor Surveillance System (BRFSS) assesses the prevalence of diseases and risk factors in adults (individuals of 18 years of age and older) statewide through an on-going telephone survey. The Alaska BRFSS has been in place since 1991, interviewing over 64 thousand adult Alaskans through 2011. The Alaska BRFSS is part of the Centers for Disease Control and Prevention’s BRFSS to track health conditions and risk behaviors in the United States, District of Columbia, and territories. The content of the BRFSS includes demographics, chronic diseases, health risk factors, access to care, health screenings, and perceptions. The current versions of the health profile maps may contain up to 172 health variables with the number of available indicators dependent upon the geographic system depicted and the number of respondents within each area for the time period.
Crude versus Age-Adjusted Rates
The prevalence for each risk factor is available as both a percentage of the Alaskan adult population at the time of the survey (termed “crude” rate as there are no adjustments) and according to a standardized age-adjusted rate which removes the impact of the age composition of a region or time period when making comparisons. Age-adjusted rates are based upon the US 2000 Standard Population and provide a consistent basis for comparison. The two rates have separate objectives. The “crude” rate can be used to determine the number or proportion of the population which is affected by a risk factor. An age-adjusted rate is a measure that controls for the effects of age differences on health event rates. Age-adjustment also controls for age effects when comparing across several years of data, as the age distribution changes over time. Generally when evaluating the impact of a health intervention or progress toward objectives such as Healthy Alaskans 2010 or Healthy People 2010, age-adjusted rates are used.
Post-Stratification and Raking Weights
Until 2007, the BRFSS data were weighted according to a process known as post-stratification. Post-stratification is a simple weighting process that divides the population of an age grouping within sex within a region by the number of responses matching those characteristics. If there are fewer than 10 responses within the sex and age class (i.e., 18-24, 25-34, 35-44, 45-54, 55-64, 65+), then the age ranges are collapsed (i.e., added together) within that sex. Post-stratification is a simple weighting process that is appropriate when the random sample reflects the overall population distribution. Raking on the other hand is a mathematically intense procedure in which population margins are used to adjust the weight assigned to each respondent. Eleven or more margins based upon combinations of population by sex, age, region, race, ethnicity, marital status, education, home ownership, and cell phone usage are used to weight the survey sample to approximate the actual distribution. Raking allows the incorporation of those using cell phones into the statistically drawn sample. A discussion of changes in BRFSS survey methods and the need to adapt to the increasing use of cell phones can be found in the publication Changes to the Alaska Behavioral Risk Factor Surveillance System Methodology: Rationale and Application in Alaska .
Data Suppression and Statistical Unreliability
BRFSS data suppression rules require at least 50 respondents within the area and time period. To provide the greatest access to health information, a note of the possible statistical unreliability of the data appears in the “Alert” column of the presentations along with the prevalence. Three data quality criteria are used in evaluating the BRFSS prevalence estimates which are based upon the Joint Policy on Variance Estimation and Statistical Reporting Standards on NHANES III and CSFII Reports: HNIS/NCHS Analytic Working Group Recommendations:
- Inadequate sample size for normal approximation. For means and proportions based on commonly occurring events (where 0.25 < P < 0.75), an estimate is flagged if it is based on a cell size of less than 30 times a "broadly calculated average design effect."
- Large coefficient of variation. Estimates are flagged if the coefficient of variation (ratio of the standard error to the mean expressed as a percent) is greater than 30.
- Inadequate sample size for uncommon or very common events. For proportions below 0.25 or above 0.75, the criteria for statistical reliability is that the cell size be sufficiently large that the minimum of nP and n(1-P) be greater than or equal to 8 times a broadly calculated average design effect, where n is the cell size and P is the estimated proportion. (I.e., an estimate is flagged when n<8 * (avg. design effect)/min(P,1-P)).) The coefficient of variation is not used in these cases.
The end user can take these reliability alerts into consideration when using the data.
The Alaska BRFSS health data have been summarized using four geographic systems with differing levels of specificity: Alaska Public Health Regions with 6 areas, Boroughs/Census Areas with 29 areas, Metropolitan/Micropolitan Statistical Areas with 7 areas, and Tribal Health Regions with 12 areas. The geographic composition of each area is provided on the Geography webpage.
BRFSS Themes and Indicators
The BRFSS themes and indicators (or variables) are organized into 5 major health topics of Demography, Quality of Life, Chronic Disease, Risk Factors, and Prevention. The health topics are subdivided in 35 themes containing as many as 172 indicators (or variables) with multiple response categories. Within each InstantAtlas presentation, data are presented by response category by area by year. As not all BRFSS questions are asked every year, it is necessary to aggregate years to meet the data suppression criterion of 50 or more respondents within an area and time period. The six Alaska Public Health Regions are able to use single year data with the largest number of health indicators. This smallest scale (i.e., fewest divisions of the Alaska) presentation contains 114,972 data points for year, area, and response levels. Single year averages are also used for the 6 boroughs representing the Metropolitan/Micropolitan Statistical Areas and the remaining rural unit (123,816 data points). The Boroughs/Census Areas with 29 geographic divisions requires the use of three-year averages (240,480 data points). Also requiring the use of three-year averages is the analysis of the 12 Alaska Tribal Health Organization Regions (126,869 data points).
The Health Profile Maps will be available in the following templates: a Single Map template with an emphasis on the geographical dispersion of risk factors, Double Map template to explore the interaction of risk factors within an area, and an Area Profile template that lists all risk factors within an area. A non-Flash Player option, and visual alternative, is available in the web-accessible HTML Profile template which provides access to the same health data in a tabular format that can be used by assistive devices. The summarized data are available in comma-separated values (CSV) and Excel XLSX formats. A general User Guide and resource-specific Help Menu explain the features and capabilities of each template.