Age adjustment

Age-adjustment is a statistical process applied to rates of disease, death, or other health outcomes that correlate with an individual's age. Adjusting for age allows for fairer comparison between populations, where age might be a confounding risk factor and the studied groups have different distributions of individuals per age group. By normalizing for age, we can paint a more accurate picture of undue burden of disease and death between populations.

We have decided to present age-adjusted ratios when possible in order to show a more accurate and equitable view of the impact on non-White communities in the United States.

As of October 2026, we are able to calculate these age-adjusted ratios for HIV deaths, COVID-19 deaths and COVID-19 hospitalizations, and we present the findings in a distinct, age-adjusted table. All of the other data shown on the tracker, including visualizations across all topics, are not age-adjusted, or 'crude rates'. Showing non-adjusted data can mask disparities, and we are working to expand our analysis to provide a more equitable view of the impact to racial and ethnic minorities.

We use a direct standardization method, with the internal standard population for each state being that state's total population. Finally, the ratios we present for each race group is that race's age-adjusted count, divided by the age-adjusted count for White, non-Hispanic individuals in the same location. Thus, our age-adjusted ratios can only be used to compare race groups within each state, and not to compare race groups between states. For COVID-19 reports, we source the standard population numbers from the 2019 population numbers from County Population by Characteristics. For HIV reports, the population data is provided along with the condition rates from the same CDC Atlas data tables.

Data Sourcing

In order to do an age-adjustment, we needed the following pieces of information:

1. Condition counts broken down by both age and race: For COVID-19, we use the CDC Case Surveillance Restricted Access Detailed Data for this. It can break down by race and age to ten-year buckets. The age buckets are: 0-9, 10-19, 20-29, 30-39, 40-49, 50-59, 60-69, 70-79, 80+. For HIV, we use the CDC Atlas data tables.
2. Population counts broken down by both race and age: For COVID-19, the most reliable population source we could find with these particular age and race groupings were the County Population by Characteristics numbers provided by the census. For HIV, the CDC Atlas provides population counts in the same tables as the condition counts.

Algorithm

In order to generate the age-adjusted ratios, we do the following:

1. For each race/age combination, calculate the age-specific rate. Age-Specific Rate = (# cases for Race A, Age group 1 ÷ Pop. of Race A, Age group 1) × 100
2. For each age group, calculate the standard population. Std. Pop. Age group 1 = Pop. of Race A, Age group 1 + Pop. of Race B, Age group 1
3. Calculate the expected count for each race/age combination: To do this we multiply the age-specific rate by the location's total population for that age group. The expected condition counts are the number of people of the race group who would have been expected to have this condition if the race group had the same age breakdown as the population as a whole. Expected Condition Count = Age-Specific Rate × Std. Pop. for corresponding Age group
4. Calculate the total expected condition count for each race group: For each race group, sum together the expected condition counts for each of that race's age groups.
5. Calculate the age-adjusted condition ratios: For each non-White NH race, divide the total expected condition counts for that race by the expected White (NH) condition counts.
6. Edge cases: If a ratio ends up being less than 0.1, we report it on the tracker as Insufficient Data to prevent sharing potentially unreliable data.

Age-Adjustment Example: HIV Deaths

Here is an example of a single state with two races, Race A and Race B, with three age breakdowns: 0-29, 30-59, and 60+. Race A will be the race we divide against to obtain our ratios (like White, Non-Hispanic), and Race B is any other race group.

1) Calculate the age-specific HIV death rates, which will be each race/age group's death count divided by its population.

2) Get the standard population per age group, which will be the summed population of all race/age groups within that age group.

3) Calculate the expected deaths for each age/race group.

As noted above, the formula for each row is:

Formula: Expected deaths for each Age/Race group = (HIV Deaths ÷ Pop.) × Std. Pop. for corresponding Age group

4) For each race, sum the expected HIV deaths from each of its age groups to calculate the total expected HIV deaths for that race.

5) Calculate the age-adjusted death ratio.

Explore age-adjusted ratios on the tracker ↗

Key Terms

Age-adjusted ratios

Health equity significance: Adjusting for age allows for fairer comparison between populations, where age might be a confounding risk factor and the studied groups have different distributions of individuals per age group. By normalizing for age, we can paint a more accurate picture of undue burden of disease and death between populations.

Measurement definition: A statistical process applied to rates of disease, death, or other health outcomes that correlate with an individual's age.

Age-specific rate

Health equity significance: This rate is significant for examining health disparities within these groups while accounting for age-related variations in health outcomes.

Measurement definition: Age-specific rates are calculated for specific age groups.

Condition counts broken down by both age and race

Health equity significance: Such data is essential for understanding how health conditions affect different demographic groups, allowing for targeted interventions.

Measurement definition: This refers to data that provides a detailed breakdown of health condition counts, considering both age and race.

Direct standardization method

Health equity significance: It is significant for comparing health outcomes while removing the influence of differences in age distribution.

Measurement definition: This method in health equity data visualization involves recalculating age-specific rates for each subgroup using the same standard population.

Edge cases

Health equity significance: Handling these cases is significant to avoid misinterpretations of health disparities.

Measurement definition: In the context of health equity data visualization, "edge cases" may refer to outliers or unique situations where traditional methods of age-adjustment may not apply as effectively.

Expected condition counts

Health equity significance: These counts help assess whether observed counts deviate from what would be expected based on age distribution.

Measurement definition: Expected condition counts represent the number of cases that would be expected in a population if it had the same age-specific rates as a standard population.

Internal standard population

Health equity significance: An internal standard population is a reference population used in age-adjusted ratios to facilitate comparisons between different subgroups.

Measurement definition: A population with known age-specific rates and serves as the basis for calculating age-adjusted rates.

Population counts broken down by both age and race

Health equity significance: This helps identify disparities in population distribution and informing policies aimed at addressing health inequities.

Measurement definition: These counts provide a demographic profile that considers both age and race.

Standard population

Health equity significance: It is typically chosen to represent a population with a known age distribution, allowing for comparisons of health outcomes across different populations or time periods.

Measurement definition: A standard population serves as a reference for age-adjusted ratios.

• County Population by Characteristics
• CDC Atlas data tables