Using BRFSS Data from SubState Regions in South Dakota

1
Using BRFSS Datafrom Sub-State Regions in South
Dakota

• Analysis Team
• Roland Loudenburg, M.P.H., ABD
• Gregg Drube, M.A.
• Gary Leonardson, Ph.D.
• State Coordinator
• Robin Erz, Deputy Director
• South Dakota Division of Alcohol and Drug Abuse

2
Geographic Unit of Analysis

• Our goal is to be able to provide epidemiological
  substance abuse data useful in supporting local
  substance abuse prevention efforts
• SD has 66 counties
• In 2006 13 of 66 counties have more than 50
  surveys
• Of the 13 counties, only 4 counties had more than
  300 surveys which limits ability to look at
  sub-groups (example age, ethnicity)
• Use of County level data has limited utility for
  consistent prevention approach statewide

3
The South Dakota Unified Judicial System (UJS)
has divided the state into 7 regions

• Each region is a collection of counties that are
  geographically and socially similar
• The UJS regions are similar to prevention and
  treatment provider catchment/service areas
• Additional state data is available on the UJS
  Region level
• Requested NSDUH to adjust their sub-state
  geographic areas to match the SD UJS regions
  (future data from NSDUH will be available by
  regions)
• The following map provides a summary of regions
  and illustration of the distribution of the
  population

4
(No Transcript)
5
Number of Years

• Used just one year to establish a baseline
• Rationale
• Allowed for exploration of utility of using BRFSS
  data on a regional level
• Combining more than 1 year would not allow for
  trends and by using Regions as apposed to
  counties will allow for trends in the future
• Links to our strategy of similar epidemiological
  prevention data available at sub-state level for
  prevention policy, planning and programming

6
Methods

• The US rate and the South Dakota rate and
  confidence interval were obtained using the CDC's
  BRFSS Prevalence data application
  http//apps.nccd.cdc.gov/brfss. The US rate is
  the median of 51 State and DC rates.
• The rates and standard errors used for
  calculating the confidence intervals for the
  seven South Dakota regions were calculated using
  the R survey package (http//faculty.washington.ed
  u/tlumley/survey/).

7
Weighting and Sample Size Requirements

• Compared the BRFSS weighting included in the
  state file with weighting calculated based on age
  and gender distribution for each region based on
  2006 census estimates
• Negligible difference found between the BRFSS
  weighting and the regional weighting
• Decided to retain the BRFSS rules for not
  reporting data
• Use caution in interpreting cell sizes less than
  50.
• N/A Not available if the unweighted sample size
  for the denominator was lt 50 or the CI half width
  was gt 10 for any cell

8
Example Heavy drinkers (adult men having more
than two drinks per day and adult women having
more than one drink per day)