Diabetes surveillance data
National data on diabetes prevalence and incidence are available dating back to 2000. State-level estimates are available for all years beginning in 2000 to the current year of available data. County-level estimates, including age-adjusted rates and rankings, are available for all years beginning in 2004 to the current year of available data.
The surveys and databases used by the US Diabetes Surveillance System to examine trends in diagnosed diabetes and its complications do not currently distinguish between types of diabetes. Therefore, our estimates include data on all types of diabetes. Because type 2 diabetes accounts for about 90%‒95% of the cases of diagnosed diabetes in adults, trends in type 2 diabetes are likely to be similar to trends documented by the surveillance system. However, because type 1 diabetes accounts for approximately 5% of all diagnosed cases of diabetes among adults, trends documented in the surveillance system may not be reflective of trends in type 1 diabetes.
Unfortunately, we do not have surveillance data on insulin use by type of diabetes. National surveillance data are available on diabetes medication use among adults with diagnosed diabetes, though not specifically type 2 diabetes.
No. The data sources used for diabetes surveillance do not provide an adequate sample size of this population to produce accurate and reliable estimates. However, several states, including Minnesota, Montana, New Mexico, North Carolina, and Oklahoma, have conducted surveys with an oversample of American Indians. For more information, contact the Behavioral Risk Factor Surveillance System’s state coordinators. For the latest statistics about diagnosed diabetes in American Indians/Alaska Natives, see the National Diabetes Statistics Report.
Currently, the US Diabetes Surveillance System includes prevalence estimates of diagnosed diabetes for Hispanics and Asians overall but not for Hispanic subgroups, such as Puerto Ricans, Mexican Americans, and Cubans, or Asian subgroups, such as Chinese, Filipinos, or Asian Indians.
The number of people with diagnosed diabetes in the US Diabetes Surveillance System and the National Diabetes Statistics Report are close but not exactly the same because the data sources, methodologies, and time periods are different. In the surveillance system, we use National Health Interview Survey (NHIS) data to estimate the US population of all ages with diagnosed diabetes. In the statistics report, the estimate of the US population with diagnosed diabetes comes from the National Health and Nutrition Examination Survey (NHANES) for people aged 18 years or older and from the NHIS for people younger than 18 years. We use NHANES data in the statistics report to estimate the number of people with diagnosed and undiagnosed diabetes; the medical examination component in NHANES allows us to estimate the number of people with undiagnosed diabetes. From NHIS, we can only obtain estimates of diagnosed diabetes.
Data from the Behavioral Risk Factor Surveillance System and from the US Renal Data System were used to obtain state-level estimates of diagnosed diabetes, risk factors, preventive care practices, and complications. Data from the BRFSS and from the US Census Bureau’s Population Estimates Program were used to model the estimates of diagnosed diabetes and its risk factors at the county level.
In the US Diabetes Surveillance System application, state-level data are available for all 50 states, the District of Columbia, Guam, Puerto Rico, and the US Virgin Islands. Data are also available for counties and county equivalents (e.g., parishes, boroughs, municipalities) for the 50 states, the District of Columbia, and Puerto Rico.
The US Diabetes Surveillance System, both for state-level and county-level data, show uninterrupted trend data and indicates with a colored bar or an asterisk that the BRFSS survey methodology changed in 2011. BRFSS data are used in obtaining state-level estimates and county-level estimates. However, the changes in BRFSS survey methods do not require any changes in the methods to calculate county-level estimates, and thus have no effect on county-level estimates.
In the US Diabetes Surveillance System application, you can find county-level estimates of diagnosed diabetes, obesity, and leisure-time physical inactivity. Click on “County Data” and then select an “Indicator,” such as newly diagnosed diabetes.
Bayesian methods were used to obtain county-level estimates for more than 3,200 counties and county equivalents in the United States. The county-level estimates are modeled, not direct, estimates using a statistical model that “borrows strength” in making a single county estimate from surrounding counties. You can view trends in county-level data beginning in 2004 in the US Diabetes Surveillance System application.
CDC has updated the method used to calculate county-level estimates in the US Diabetes Surveillance System application. All county-level prevalence estimates since 2004 have been recalculated using the power prior log-weights (PLOW) small area estimate approach.
See the following figure for similarities and differences in the previous method and the power prior log-weights (PLOW) small area estimation method to calculate county-level estimates in the US Diabetes Surveillance System application.
In general, county-level estimates based on power prior log-weights (PLOW) small area estimation are lower than previously reported estimates. Previous estimates and PLOW estimates are not comparable. If you need data from prior years, download PLOW estimates from 2004–2019 from the USDSS site.
Internal and external validation suggest that estimates based on power prior log-weights (PLOW) small area estimation are more accurate and precise compared to previous 3-year averaged estimates. PLOW estimates are more sensitive to changes over time and have less bias and lower variance (i.e., narrower 95% confidence intervals). They are also more generalizable to the county level because survey sampling weights are included in the models.
More information is available in the following article: Xie H, Barker LE, Rolka DB. Incorporating design weights and historical data into model-based small-area estimation. J Data Sci. 2020;18(1):115–131.
The county-level estimates are modeled estimates that were obtained using Bayesian multilevel modeling techniques whereas other estimates seen elsewhere may be direct estimates using different methodology.
No, the US Diabetes Surveillance System uses data from CDC’s Behavioral Risk Factor Surveillance System (BRFSS), which is an ongoing, monthly, state-based telephone survey of the adult population aged 18 years or older in the United States.
No, data on diabetes incidence and prevalence, as well as data on prevalence of obesity and physical inactivity, are provided for counties or county equivalents within Indian reservations, but not for reservations. For a map of US counties and American Indian/Alaska Native areas, see the Census Brief. [PDF – 3 MB]
Incidence is the rate at which new events occur in a population. The numerator is the number of new events that occur in a defined period; the denominator is the population at risk of experiencing the event during this period.
Prevalence is the total number of all individuals who have an attribute or disease at a particular time (or during a particular period) divided by the population at risk of having the attribute or disease at this point in time (or midway through the period).
The crude rate is the raw or unadjusted estimate.
The age-adjusted rate is an estimate that helps us compare populations that have different age distributions.
Diabetes Surveillance Applications
In the US Diabetes Surveillance System application, after selecting the indicator (e.g., diagnosed diabetes), click the play (right arrow) button below the map to start the animation and view trends. Legend categories have been set to the 2005 natural breaks to see the change in prevalence or incidence across years. You may also view trends using quartiles. For quartiles, click the “Data Filters” button, look for “Data Classification,” and switch the radio button to “Quartiles.” Note, however, that in quartiles legend categories will change from year to year depending on the data distribution for that year.
Yes, the US Diabetes Surveillance System includes US maps and tables that help you compare states across the same data category. Pay close attention to the lower and upper limits of the confidence interval, which you will find in the table, before reaching conclusions about significant differences between states. If the confidence intervals don’t overlap, the differences are statistically significant. However, if they do overlap, the differences may or may not be statistically significant.
In the US Diabetes Surveillance System application, after selecting the indicator (e.g., diagnosed diabetes), look for the download icon on the top right of the US map, the heat map, or the line chart. Click the icon to print the data to a PowerPoint slide that you could then copy and paste into your presentation.
In the US Diabetes Surveillance System application, after selecting the indicator (e.g., diagnosed diabetes), look for the download icon on the top right of the US map, the heat map, or the line chart to download the CSV (Excel) files with the data that you see on the screen.
Maps showing trends in both diabetes and obesity at the county level are available. See presentation slides for Age-Adjusted Prevalence of Obesity and Diagnosed Diabetes Among Adults, by County, United States (2004, 2010, 2016)
A county’s rank is a reflection of its burden of the disease or condition relative to other counties. Ranks for county-level data of diagnosed diabetes (existing and new cases), obesity, and physical inactivity were based on age-adjusted rates. Confidence intervals around these ranks convey the uncertainty associated with a county’s rank and need to be considered before reaching conclusions based on ranks. Learn more about the methods to obtain county rank [PDF -158 KB].
In the US Diabetes Surveillance System application, after selecting “County Data” and the indicator of interest, click on “Data Filters.” Select “Rank” as the datatype to view the map of county ranks.
Exercise caution when making comparisons using county maps and county estimates. The maps are presented for displaying possible geographic patterns and stimulating further investigation but are not intended as formal representations of similarities and differences. Do not assume that counties mapped in different colors have significantly different estimates. County-level estimates are intended as individual point estimates that are grouped in categories by various methods to produce a state or national map. This grouping does not incorporate the standard deviation or confidence interval and does not imply any formal comparison between counties. Bayesian 95% confidence intervals and standard deviations are provided as precision indicators of the individual county-level point estimates and should be used in data analyses.