About the diabetes data
- What do these data tell us?
- How can we use these data?
- What can these data not tell us?
- What is the source of the data?
- What are the limitations of these data?
In depth about the BRFSS survey:
- How does the survey work?
- How were people with diabetes identified?
- Why can't I compare data from 2010 and before to data from 2011 and after?
- Where can I get more technical information about the data?
What do these data tell us?
- The number of people living with diabetes in Minnesota is higher than ever, due to an increase in the 1990s and 2000s.
- Age is a risk factor for diabetes.
- Diabetes is more common for adults age 18-64 living in households that earn less than $35,000 per year.
- Among adults age 18-64, diabetes is more common among people who are out of work unable to work, or retired.
How can we use these data?
- To describe the number of people living with diabetes in Minnesota, and the current trends.
- To provide context for efforts to address diabetes and prediabetes.
- To open a dialogue about diabetes and health equity, starting with income data.
What can these data not tell us?
- Prevalence of diabetes by Minnesota county.
- Prevalence of diabetes in children in Minnesota.
- For income and employment, these data cannot tell us which came first: income and employment status, or diabetes diagnosis. The data are a snapshot in time - they cannot tell us if one preceded or caused the other.
What is the source of the data?
The Behavioral Risk Factor Surveillance System (BRFSS) is a telephone survey given each year by the Minnesota Center for Health Statistics in partnership with the national Centers for Disease Control and Prevention (CDC). To learn more about BRFSS, including how it works and questions are asked, visit CDC BRFSS.
What are the limitations of the data?
- The BRFSS administers questionnaires to a sample of Minnesota adults in households. It does not represent all Minnesota adults, such as those in long-term care facilities, nursing homes, the military, or correctional institutions. BRFSS may underestimate diabtes prevalence because people in some of the groups that are not represented, such as older adulst in long-term care facilities, have been shown to have high prevalence of diabetes.
- National estimates suggest 25-30% of all people with diabetes are undiagnosed. This means the true prevalence of diabetes, diagnosed and undiagnosed cases, is higher than the estimates for diagnosed diabetes presented here.
- About 15% of adults refused to provide information about their household income. Missing data can sometimes bias results. An analysis was conducted to determine the potential impact of missing data on the income chart. The analysis showed that the chart's pattern holds: adults with low income have higher prevalence of diabetes regardless of the treatment of missing data.
- BRFSS captures a range of income for each individual's household (for example, $0-15,000 per year), but does not ask for exact income or the number of people in the household supported by the income. Asking about exact income and the number of people in the household would be a stronger way to assess income. This information is used to calculate adjusted household income, which can be compared to Federal Poverty Levels.
In depth about the BRFSS survey
How does it work?
- For those more familiar with survey methods: BRFSS uses a weighted stratified sampling design in its telephone-based survey. Recent surveys have included both adults who have landline telephones and cell phone users, and have had between 10,000 and 15,000 participants. All percentages shown in graphs have been weighted to account for non-response and for participant characteristics so that weighted results should capture the experience of the entire state.
- For those interested in learning about survey methods: The BRFSS invites a large number of Minnesota adults to participate in a telephone survey. The type of survey is a weighted random-sample. This means that adults are selected randomly and invited to answer questions included in the survey. In the last few years, between 10,000 and 15,000 people agreed to take part each year– roughly a group the size of the city of Grand Rapids. The answers that these people give to the survey questions are used to estimate the answers we would expect if we asked every one of the 4.1 million adults in Minnesota the same questions. The answers we would expect if everyone in the state responded are called weighted estimates in survey terms.
How were people with diabetes identified?
People who took part in BRFSS were asked the question, "Has a doctor, nurse, or other health professional ever told you that you had diabetes?" If they said "yes," they were counted as having diabetes. If they were female and said, "Yes, but only during pregnancy," they were counted as not having diabetes. Gestational diabetes or diabetes during pregnancy is different than having a diabetes diagnosis. Those who responded "No" or "No, prediabetes or borderline-diabetes" were not counted as having diabetes.
Why can't I compare data from 2010 and before to data from 2011 and after?
- For those more familiar with survey methods: Beginning in 2011, BRFSS included cell phones in its sampling scheme. Prior to 2011, people who were cell phone only users were excluded from the survey. In addition, a new method for weighting the sample was introduced called iterative proportional fitting or raking. This methodology better allows data for population subgroups that are underrepresented in the sample to be more accurately represented in weighted estimates. The raking method allows more demographic variables to be used in creating sample weights leading to more accurate estimates.
- For those interested in learning about survey methods: BRFSS survey methodology was improved beginning in 2011. Earlier versions of the survey only contacted people with landline telephones. In 2011, the new survey included cell phones, no longer leaving out people who only used cell phones. Other changes were made to improve the process of taking data from the approximately 10,000 to 15,000 people who answer BRFSS questions and estimating what the results would be if we asked everyone in the entire state. These new methods allow us to take into account more characteristics of the people who are participating in the survey as we estimate what their answers mean for the state of Minnesota.