2020 Adult Asthma Data: Technical Information
BRFSS Questionnaire
BRFSS is an annual statewide sampling telephone surveillance system. In each area where interviews are conducted, respondents are randomly selected from noninstitutionalized civilian adults (18 years of age or older) living in households with a telephone or cellphone (The BRFSS Data User Guide, https://www.cdc.gov/brfss/data_documentation/pdf/UserguideJune2013.pdf). Asthma prevalence data is produced from the responses to two asthma questions on the year 2020 BRFSS Core Questionnaire.
Lifetime asthma: Question number 6.4.
Section 6: “Has a doctor, nurse, or other health professional EVER told you that you had any of the following?: 6.4 . (Ever told) you had asthma? (Variable name: ASTHMA3)
The calculated variable _LTASTH1 (based on question 6.4) from the public use file was used when creating the tables.
Current asthma: Question number 6.5
If the response to 6.4 was Yes, then question 6.5 was asked: “Do you still have asthma?” (Variable name: ASTHNOW). The calculated variable _CASTHM1 (based on questions 6.4 and 6.5) from the public use file was used when creating the tables.
Subgroup tables are based on data from the following questions:
| Survey Question | Variable name | Codes used |
|---|---|---|
| C8.11 Indicate sex of respondent | _SEX | 1 = Male 2 = Female |
| C8.14 Imputed Age value collapsed above 80 | _AGE80 | If age less than 80, _AGE80 = Age, If age 80 or older, collapsed If age missing, imputed missing age |
| C8.13 Reported age in two age groups calculated variable | _AGE65YR | Grouped into: 1 = 18 <= AGE <= 64 2 = Age 65 or older Notes: 65 <= AGE <= 99 3 = Don’t know/ Refused/Missing |
| C8.6 Hispanic, Latino/a, or Spanish origin calculated variable | _HISPANC | 1 = Hispanic, Latino/a, or Spanish origin 2 = Not of Hispanic, Latino/a, or Spanish origin |
| C8.4 Calculated non-Hispanic Race including multiracial | _MRACE1 | 1 = White 2 = Black or African American 3 = American Indian, Native Alaskan 4 = Asian 5 = Native Hawaiian, Pacific Islander 6 = Other race 7 = Multiracial |
| C8.9 Computed Five level race/ethnicity category. | _RACEGR3 | 1 = White, non-Hispanic 2 = Black, non-Hispanic 3 = Other, non-Hispanic 4 = Multiracial, non-Hispanic 5 = Hispanic |
| 8.7 What is the highest grade or year of school you completed? (EDUCA) | EDUCA | 1,2,3 = <HS graduate 4 = HS graduate 5 = Some college 6 = College graduate |
| 8.17 Is your annual household income from all sources: (INCOME2) | INCOME2 | 1,2 = <$15,000 3,4 = $15,000-$24,999 5,6 = $25,000-$49,999 7 = $50,000-$74,999 8 = $75,000+ |
BRFSS Sampling and Weighting Changes
The 2011 BRFSS data collection changed from a landline sample survey to a dual-mode survey, landline and cellphone. The 2020 BRFSS data continues to reflect the changes initially made in 2011 in weighting methodology (raking) and the addition of cell phone only respondents. The aggregate BRFSS combined landline and cell phone dataset is built from the landline and cell phone data submitted for 2020 and includes data for 50 states, the District of Columbia, Guam, and Puerto Rico.
For data analysis, the 2011 BRFSS data should be considered a baseline year and are not directly comparable to previous years of data because of the changes in weighting methodology and the addition of the cell phone only respondents . Please see the BRFSS Comparability of Data document for additional information: https://www.cdc.gov/brfss/annual_data/2020/pdf/compare-2020-508.pdf [PDF – 167 KB]
Survey Design and Sample Weights
The survey design is described in the technical pages document, entitled, “2020 Overview” at h https://www.cdc.gov/brfss/annual_data/2020/pdf/overview-2020-508.pdf [PDF – 120 KB]. Sample weights are assigned, according to BRFSS methodology, to the variable _LLCPWT as described in the BRFSS Weighting Formula.
Data Analysis
Data Used:
The BRFSS 2020 Survey Data file was used to calculate estimates for all states.
Software:
Prevalence and standard error estimates were calculated using SUDAAN Release 11.1 (Research Triangle Institute, P.O. Box 12194, Research Triangle Park, NC 27709).
Data Management:
Only respondents with values of the variables as described in the table above are included in each prevalence table. Responses of “don’t know/not sure,” “refused,” or missing values were excluded.
Confidence Intervals:
The 95% Confidence Intervals were calculated using the following formulae:
Lower 95% Confidence Interval = % prevalence – t (sep)
Upper 95% Confidence Interval = % prevalence + t (sep)
Where sep is the standard error of the prevalence percent and t equals the z distribution value of 1.96 for state level prevalence in Table 1. For other tables with smaller sample sizes due to the use of demographic subcategories, the exact value from the t distribution appropriate to the sample size for the percent prevalence was used instead of the approximate z-value of 1.96. Because of the use of the more exact t-value, 95% Confidence Intervals in the tables may differ slightly from those presented in other representations of these data.
Table Conventions:
The “US Total” line of each table excludes Puerto Rico and Guam. “US Total,” therefore, designates an estimate for the 50 states and the District of Columbia combined.
In the tables, states are listed in Federal Information Processing Standard (FIPS) code order.
Washington, D.C., is omitted from the maps. Its area is so small that the map colors and patterns were not discernible.
Ranges in all maps (both overall and for subgroups) are based on quintiles of the overall prevalence estimates from year 2011 data. These same ranges will be used in future years to facilitate year-to-year comparison of the maps.
Chart:
The chart presents the state and territory prevalence percent from Table 1 sorted from high to low. Confidence intervals are also represented graphically for each prevalence value. Overlapping confidence intervals for two states is a rough indication that state prevalence values are not significantly different from one another.
Small Sample Size
When sample sizes in cells are less than 50, as they are in some of the subgroup tables, the standard error may be large relative to the prevalence value, leading to a wide 95% confidence interval. Caution should be used in interpreting such “imprecise” estimates of prevalence.
Also, when one or more of the following situations occurred, the indicated solutions were applied:
| Situation | Solution |
|---|---|
| 1. The normal distribution approximation to the binomial distribution did not apply. | 1. Values for the standard error and the 95% confidence interval were not provided. |
| 2. If the normal distribution approximation does apply, it is appropriate to use values of the t-distribution to compute confidence intervals. | 2. Two-sided 95% confidence interval values from the t-distribution with (n-1) degrees of freedom (where n is the number of observations for the subgroup) were used. |
| 3. The lower limit of the 95% confidence interval was negative. | 3. The lower limit was set to zero. |
Content source: National Center for Environmental Health
