Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
Data visualization can make differences in diabetes prevalence across income groups easier to see, compare, and investigate. In a CDC 2021 comparison, diagnosed diabetes was reported by 16.4% of adults in the household-income category below $35,000, 11.0% of those earning $35,000 to under $75,000, and 7.7% of those earning $75,000 or more. That is a substantial descriptive gradient—not proof that income alone causes diabetes. A sound analysis defines both measures, shows uncertainty, and distinguishes individual-level findings from patterns across places.
What the income gradient shows—and what it does not
The CDC Diabetes State Burden Toolkit reports 2021 diagnosed-diabetes prevalence estimates for U.S. adults by household-income category:
| Household-income category | Diagnosed diabetes prevalence | 95% confidence interval |
|---|---|---|
| Below $35,000 | 16.4% | 15.8–16.9% |
| $35,000 to under $75,000 | 11.0% | 10.6–11.5% |
| $75,000 or more | 7.7% | 7.3–8.0% |
In this specific 2021 comparison, the difference between the lowest and highest income categories is 8.7 percentage points. Dividing 16.4 by 7.7 gives a prevalence ratio of about 2.13: diagnosed-diabetes prevalence in the lowest-income category was approximately 2.1 times the estimate for the highest-income category. Both calculations describe these estimates; neither is a causal effect or a forecast of what would happen if a person’s income changed. CDC State Burden Toolkit health-burden results
The estimates concern diagnosed diabetes among adults, not lifetime risk or every form of diabetes. The broad income bands are categories, not evenly spaced points on a continuous scale. They also cannot account, by themselves, for differences in age, health-care access, education, race and ethnicity, geography, or other conditions related to both income and health.
#1 Best Overall
Define diabetes and income before plotting them
Choose the diabetes outcome
“Diabetes rate” is ambiguous. Prevalence is the proportion of a population living with a condition at a specified time or period; incidence counts new cases over a defined period. Hospitalizations, complications, mortality, and medical costs describe other outcomes and should not be substituted for prevalence.
Many widely available U.S. surveillance measures capture diagnosed diabetes: respondents report whether a doctor or other health professional has ever told them they have diabetes. The CDC BRFSS prevalence visualization uses that question. It is a measure of self-reported diagnosed diabetes, not a clinical count of all diabetes; undiagnosed cases are not captured by that question. Broad adult surveillance measures may also not separate type 1 from type 2. CDC BRFSS prevalence dataset
Choose the income measure
Household income brackets, individual earnings, family income, poverty rate, income-to-poverty ratio, and area median household income measure different things. Household income brackets make group comparisons straightforward, but hide variation within a bracket. Area median income and poverty rates are useful for comparing places, but do not tell you the income of each resident. Income is also not the same as wealth or disposable resources.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errorsKeep the unit and definition visible in the chart title or subtitle. For example, “household income” should not silently become “personal income,” and a county poverty rate should not be described as the income of individual people. If dollar amounts from different years are compared, account for inflation and disclose the income year.
Decide whether the records describe people or places
- Individual-level data pair a person’s diabetes status with that person’s income or poverty measure. They allow direct group comparisons and adjustment for other characteristics, but survey answers can be inaccurate, income may be missing or imputed, and results depend on the sample and survey design.
- Area-level data summarize a county, state, or other geography. They suit maps and place-based comparisons, but an area-level association does not establish what is true of each person there. That mistaken leap is the ecological fallacy.
A county with high median income can still contain residents with low incomes, and a low-income county can contain high-income residents. CDC county diabetes estimates may be modeled, indirect estimates that borrow information across counties; they are not necessarily equally precise direct measurements for every county. CDC county diabetes estimates and methodology
Choose a chart that matches the question
| Chart | Best use | Main safeguard |
|---|---|---|
| Dot plot with confidence intervals | Comparing prevalence estimates across income groups or places | Label the estimate, interval, year, and population; include sample size when available. |
| Grouped bar chart | Showing a small number of income categories to a general audience | Use a clear zero baseline when bars encode magnitude and show uncertainty. |
| Scatterplot | Comparing area-level prevalence with continuous income or poverty measures | Make clear that every point is a place, not a person; do not imply the fitted line proves causation. |
| Choropleth map | Showing where prevalence or income varies geographically | Map rates rather than raw counts for place comparisons and state how estimates were produced. |
| Small multiples | Comparing the pattern by age, race and ethnicity, sex, region, or rurality | Keep scales and definitions consistent so panels remain comparable. |
| Time series | Checking whether a gap changes over time | Use comparable measures and mark survey or methodological breaks. |
Use intervals, not rankings alone
A dot plot with 95% confidence intervals shows both the point estimate and its uncertainty. Intervals that overlap substantially are a warning against declaring one group or county definitively higher based only on a visual ranking; formal comparisons require an appropriate statistical method. Small populations often yield less precise estimates, and some values may be suppressed or unstable. Do not rank areas as “worst” on tiny differences their uncertainty does not support.
Rank #3
Map carefully
A pair of maps—one for diabetes prevalence and one for poverty or income—can reveal geographic overlap, but similar colors on two maps do not demonstrate a statistical relationship. Pair maps with a scatterplot or another direct comparison. Large geographic areas can dominate a map visually even when they have fewer residents; use rates for comparing risk or prevalence and counts only when the question is total service demand or burden. CDC mapping shows substantial geographic variation, including concentration of many high diagnosed-diabetes county estimates in the Southeast, but location alone does not explain that pattern. CDC diagnosed-diabetes geographic data
Use an ordered, color-blind-accessible scale with a clear legend. Red-to-green colors can imply a moral judgment or a binary good-versus-bad message. If estimates are age-adjusted, say so; do not mix crude and age-adjusted estimates in one comparison without clear labeling.
Use small multiples to expose variation
An overall income gradient can conceal different patterns by age, race and ethnicity, sex, region, or rurality. A grid of consistently scaled charts can show where the pattern is steeper, weaker, or reversed. This is often easier to read than putting every subgroup into one crowded legend. Stratification can reveal variation, but it does not by itself explain why groups differ.
Find compatible U.S. data
- CDC U.S. Diabetes Surveillance System: national, state, and county indicators, with measures and breakdowns that vary by indicator. Check the selected indicator, year, age group, denominator, and estimation method before comparing locations. Open the CDC surveillance dataset.
- CDC BRFSS prevalence data: state-based adult survey estimates for self-reported diagnosed diabetes, with a series beginning in 2011. Useful for state and subgroup comparisons over time, subject to survey design, self-report, nonresponse, and access-to-diagnosis limitations. Open the BRFSS dataset.
- CDC Diabetes State Burden Toolkit: state-level health, economic, and mortality information. Its income-stratified 2021 estimates provide the example above; underlying health data can be accessed for analysis or visualization tools. Explore the toolkit and health-burden dataset.
- U.S. Census Bureau American Community Survey: a source for area-level income, poverty, education, insurance, and related characteristics. Select the relevant table, release, and geography, then align its year as closely as possible with the health estimates. Find ACS data.
Do not assume data products share a common year simply because they appear in the same toolkit. The toolkit’s health-burden data draw on 2020–2021 BRFSS sources, while its economic-burden measures use different source years, including 2014, 2021, and 2022. Present those modules as different-period measures, not as synchronized observations. CDC economic-burden dataset
Build the analysis in a reproducible sequence
- Write a measurable question. For example: “Among U.S. adults, how does diagnosed-diabetes prevalence differ by household-income category, and how does the pattern vary by age and rurality?” Specify population, outcome, income measure, geography, and period.
- Select data suited to the question. Use the toolkit’s published income comparison for a clear descriptive example; use individual survey data for individual-level modeling or compatible county estimates plus Census measures for an area-level analysis.
- Read the documentation. Record age range, diabetes definition, income definition, geography, observation year, crude or age-adjusted status, modeling method, confidence intervals, weighting, missing-data treatment, and suppression rules. CDC’s technical documentation describes the toolkit’s BRFSS-based estimates and subgroup variables. Read the technical documentation.
- Harmonize before joining. Match geographic identifiers and boundaries, denominators, years, and income definitions. Convert dollar values to a common year when needed. Keep confidence limits attached to their estimates and document exclusions; never treat household income as equivalent to individual income.
- Start with descriptive charts. Show the overall outcome, compare income groups, add uncertainty, then examine geography or subgroups. Make the definition, year, population, and source visible in the caption.
- Quantify the contrast. Report percentage-point differences alongside the underlying prevalence estimates. A prevalence ratio can add context, but label it descriptive unless an appropriate adjusted analysis supports a different interpretation.
- Check subgroup patterns and confounding. Examine age, sex, race and ethnicity, education, region, and rurality where data support it. A more advanced individual-level analysis may use survey-weighted logistic regression, survey-weighted Poisson regression with robust variance for prevalence ratios, or age-standardized comparisons. Explain model choices and uncertainty rather than presenting an adjusted result as self-explanatory.
- Publish enough to reproduce the figure. Provide source links, data years, variable definitions, methods, exclusions, and downloadable data or code when feasible.
Formal inequality analysis can go beyond comparing two endpoints: a CDC-linked analysis, for example, uses education and family poverty-to-income ratio as socioeconomic measures and applies metrics such as the slope index of inequality. Such metrics answer a different, more model-dependent question than a simple chart of income bands. CDC-linked diabetes inequality analysis
Interpret the pattern in context
Income may be connected to diabetes through conditions that affect health, diagnosis, and management: affordability and availability of food, housing stability, work schedules, transportation, preventive care, insurance, medication and monitoring costs, chronic stress, neighborhood safety, opportunities for physical activity, and access to diabetes education. These are plausible pathways to investigate, not explanations proved by a descriptive chart.
Best Value
Diagnosis itself complicates interpretation. A diagnosed-diabetes survey measure can reflect underlying disease prevalence as well as whether people can access screening and care. Limited access may leave some cases undiagnosed. The chart therefore describes diagnosed prevalence in its defined population, not necessarily all diabetes or the full underlying disease burden.
Likewise, age structure can affect comparisons because diabetes prevalence varies with age. Race and ethnicity, education, geography, rurality, insurance, employment, and health-care access may also be related to both income and the outcome. Consider these factors as potential confounders or sources of variation; do not imply that race is a biological explanation for socioeconomic differences.
- Say “prevalence was higher among adults in the lower-income category” rather than “low income causes diabetes.”
- Say “counties with higher poverty rates had higher estimated prevalence” for an area-level result, not “poor people had higher prevalence” unless the data directly measure individuals.
- Call a finding an association or observed pattern unless a causal design and assumptions justify causal language.
- Show outliers and uncertainty rather than letting an average gradient imply that every person or place follows it.
Design a dashboard without hiding the important details
An interactive dashboard can support exploration when readers need to compare states or counties, years, income groups, or demographic subgroups. Useful filters might include geography, year, age, sex, race and ethnicity, rurality, and diabetes measure. Put the active filters and data year in view, provide definitions and source links, and offer a downloadable table.
Recommended Free Tools
Every view should state whether estimates are crude, age-adjusted, modeled, or survey-based. Warn users when filters would combine incompatible measures or years. A dashboard should begin with a clear finding and then offer exploration; a wall of controls can make the main result harder to understand. Use public data responsibly, and do not upload identifiable health records to a public visualization service.
Example caption for an income comparison
Percentage of U.S. adults with diagnosed diabetes by household-income category, 2021. Estimates are from the CDC Diabetes State Burden Toolkit and are shown with 95% confidence intervals. This descriptive comparison does not establish that income causes diabetes. Source and definitions.
Conclusion
Visualization is valuable because it makes disparities, geographic patterns, subgroup differences, and uncertainty visible. In CDC’s 2021 income comparison, diagnosed-diabetes prevalence was higher in the lower-income categories. Treat that gradient as a starting point: define the measure and unit, align years, show uncertainty, investigate confounding, and keep place-level findings separate from claims about individuals. A chart can clarify an association; on its own, it cannot show that income alone caused it.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
