Age-Adjusted BRFSS Percents

We are adding age-adjusted percents to the Vitalnet BRFSS module. The goal is allow users to quickly and reliably do BRFSS age-adjusted analyses, with: * No programming knowledge required. * User is assured correct, fully documented results. * Publication-ready results produced in about a minute.

For now, we are testing on an internal version of the software that produces text output only. When we are sure that everything is working correctly, we will add the age-adjusted percents to the public interfaces with HTML, chart, and map output.

Here are two tables showing the difference age-adjustment can make:

- Text table with weighted percents (not age-adjusted).
- Text table with weighted percents (age-adjusted).

If not age-adjusted, Hispanic (8.0%) is slightly higher than White (7.1%).
If age-adjusted (AA), Hispanic (11.4%) is much higher than White (6.4%).
Here is a summary table showing the differences:

Ever had diabetes (excluding pregnancy)?
Race	Weighted % Yes
Race	Not AA	Is AA
White	7.1%	6.4%
Black	12.1%	12.9%
Hispanic	8.0%	11.4%

To compare risk or prevalence between populations with different age structures, it is generally recommended to adjust for age. It's the same principle that explains why age-adjusted rates are used to compare death rates between populations. Healthy People 2010 and 2020 Objectives with BRFSS data.

To assist with testing for correctness, we are looking for existing "gold standard" age-adjusted analyses to compare the Vitalnet results with. After we have verified the Vitalnet BRFSS module is producing correct age-adjusted percents, we plan to add confidence intervals to the age-adjusted percents. We plan to use Jackknife replication to do the confidence intervals.

For the population standard, we currently use population distribution #10 from Klein and Schoenborn (2001).

Let us know any suggestions or comments you have. Or contact us if you are interested in licensing the professional Vitalnet solution to make better use of BRFSS data.