Limitations and missing data

While we strive for accuracy, some of the limitations include:

Limitations

Data Accuracy and Completeness

Data collected from sources like the CDC may have inaccuracies or incompleteness due to errors in reporting, variations in data collection methods, or missing data from certain regions or populations. This can lead to biases or distortions in the representation of disease prevalence.

Temporal Lag

There might be a lag between the occurrence of a disease and its reporting. This delay can affect the timeliness of the data presented on the health equity tracker, potentially hindering real-time decision-making.

Geographical Resolution

While choropleth maps provide a visual representation of data at the regional level (e.g., states, counties), the granularity of the data may not be sufficient for detecting localized outbreaks or disparities within smaller geographic areas.

Population Heterogeneity

Population characteristics can vary significantly across different regions and demographic groups. Aggregating data at the state or county level may obscure important variations within populations, such as differences in socioeconomic status, access to healthcare, or cultural factors that influence health outcomes.

Ecological Fallacy

This occurs when inferences about individuals are drawn from group-level data. Just because a certain area has a high disease prevalence does not mean that every individual within that area is affected similarly. Drawing conclusions about individuals based solely on aggregated data can lead to erroneous assumptions.

Data Interpretation

Visualizations like choropleth maps can simplify complex data, but they can also oversimplify the underlying patterns. Care must be taken to ensure that interpretations of the maps consider the context, potential confounding variables, and limitations of the data source.

External Factors

Other external factors, such as changes in healthcare policies, diagnostic criteria, or environmental conditions, can influence disease prevalence and distribution. These factors may not be adequately captured in the data or accounted for in the analysis.

What Data Are Missing

Missing and misidentified people

Currently, there are no required or standardized race and ethnicity categories for data collection across state and local jurisdictions. The most notable gaps exist for race and ethnic groups, physical and mental health status, and sex categories. Many states do not record data for American Indian, Alaska Native, Native Hawaiian and Pacific Islander racial categories, lumping these people into other groups. Individuals who identify as Hispanic/Latino may not be recorded in their respective race category. Neither disability nor mental health status is collected with most data sources, and in almost all cases sex is recorded only as female, male, or other.

Missing population data for Census Island Areas

Population data for Northern Mariana Islands, Guam, American Samoa, and the U.S. Virgin Islands are not reported in the ACS five year estimates; in these territories, for current and time-series based population figures back to 2016, we incorporate the 2020 Decennial Island Areas report. For time-series data from 2009-2015, we incorporate the 2010 release of the Decennial report. Note: The NH, or Non-Hispanic race groups are only provided by the Decennial report for VI but not the other Island Areas. As the overall number of Hispanic-identifying people is very low in these Island Areas (hence the Census not providing these race groups), we use the ethnicity-agnostic race groups (e.g. Black or African American) even though the condition data may use Non-Hispanic race groups (e.g. Black or African American (NH)).

Missing and suppressed COVID data

For COVID-19 related reports, this tracker uses disaggregated, individual case level data reported by states, territories, and other jurisdictions to the CDC. Many of these case records are insufficiently disaggregated, report an unknown hospitalization and/or death status, or otherwise fail to provide a complete picture of COVID-19 and its overall impact.

Missing COVID-19 vaccination data

• Population data: Because state-reported population categories do not always coincide with the categories reported by the census, we rely on the Kaiser Family Foundation population tabulations for state-reported population categories, which only include population numbers for Black, White, Asian, and Hispanic. 'Percent of vaccinated' metrics for Native Hawaiian and Pacific Islander, and American Indian and Alaska Native are shown with a population comparison metric from the ACS 5-year estimates, while Unrepresented race is shown without any population comparison metric.
• Demographic data: The CDC's county-level vaccine dataset only provides vaccination figures for the All group, but does not include any demographic disaggregation.

Missing data for women in legislative office

• The Center for American Women in Politics (CAWP) dataset uses unique race/ethnicity groupings that do not correspond directly with the categories used by the U.S. Census. For this reason, Middle Eastern & North African (Women), Asian American & Pacific Islander (Women), and Native American, Alaska Native, & Native Hawaiian (Women) are presented without corresponding population comparison metrics.
• We are currently unable to locate reliable data on state legislature totals, by state, by year prior to 1983. For that reason, we cannot calculate rates of representation historically before that year.
• Legislature seat totals (denominator): Annual seat counts for all 50 states and DC are scraped from CAWP's state information pages. Seat totals for territories (American Samoa, Guam, CNMI, Puerto Rico, and the U.S. Virgin Islands) are sourced from NCSL (National Conference of State Legislatures) and updated manually each year. Note: Puerto Rico's Legislative Assembly has a constitutional baseline of 78 seats (27 Senate + 51 House), but a minority-protection clause can temporarily expand it following certain election results; the current 20th Assembly (2025-2029) sits at 81 seats.

Missing data for HIV deaths, diagnoses, and prevalence

• County-level data is suppressed when the population denominator is less than 100, the total case count is between 1–4 cases, or when querying HIV or AIDS deaths.
• To protect personal privacy, prevent revealing information that might identify specific individuals, and ensure the reliability of statistical estimates, small data values may not be available in some circumstances.
• There isn't enough data to accurately calculate subpopulation rates by age, sex, and race/ethnicity for the Census Island Areas (US territories other than Puerto Rico). As a result, the analysis or report will not provide detailed information about these specific groups in those regions.
• The Asian category includes cases previously classified as "Asian/Pacific Islander" under the pre-1997 Office of Management and Budget (OMB) race/ethnicity classification system when querying HIV prevalence.

Missing and suppressed PrEP data

• The race and ethnicity of individuals prescribed PrEP are only available for less than 40% of all people prescribed PrEP and are limited to four categories: White, Black, Hispanic/Latino, and Other.
• State-level and county-level PrEP data are not available for race and ethnicity.
• PrEP coverage data are suppressed at any level if the number of persons prescribed PrEP is suppressed, the estimated number of persons with indications for PrEP (PrEP-eligible population) is suppressed, or if the number of persons prescribed PrEP is less than 40.

Medicare Administration Data

What demographic data are missing?

• Gender: The Medicare source files did not include information on gender. Gender is not presented in this data.
• Sexual Orientation: The Medicare source files did not include information on sexual orientation. Sexual orientation is not presented in this data.
• Disability: Although we can display rates for those who are eligible due to disability generally, we can not represent disparities associated with specific physical or mental disabilities.
• Social and Political Determinants of Health: Unfortunately, there are crucial data missing in the Medicare FFS (Fee-For-Service) data, such as the impacts of racism and discrimination on health outcomes and adherence to medicines.

Who is missing?

• Data Suppression: To protect patient privacy, all data representing 1-10 people were suppressed.

Missing America's Health Rankings data

• Population data: AHR publishes its measures as rates only, without the underlying case counts or population denominators. To display estimated totals and percent share metrics, we merge in American Community Survey (ACS) population estimates. Because most AHR measures describe adults only, we use the adult (18+) population as the comparison denominator for those measures, so that a group's share of cases and its share of the population are measured on the same population basis.
• Measures without a matching denominator: Two measures are reported over a narrower sub-population than we have population counts for. Voter participation is measured among citizens ages 18+, and preventable hospitalizations among Medicare enrollees ages 18+. Because we cannot break those sub-populations down by race, ethnicity, or sex, we cannot calculate a percent share for these measures. Instead we substitute the closest population we can break down: the adult (18+) population at the national and state levels, and the general all-ages population at the county level, where age-specific counts are not available. This stand-in is broader than the group the rate actually covers, so we label the difference on those tables and treat the comparison as context rather than an exact share.
• Age reports: AHR reports its adult measures against three broad age groups (18-44, 45-64, and 65+), but reports its all-ages measures against narrower groups such as 15-24 and 65-74. Because the narrower groups overlap the broad ones, and because 15-24 includes minors, the adult population share is shown only for the three broad groups, where it accounts for 100% of the adult population.

Missing WISQARS Data

• Legal intervention data: Data on deaths caused by legal intervention is limited. Therefore, we choose to show raw counts of deaths caused by legal intervention in states where the data is available. This approach ensures that the information presented is accurate and reflective of the available data without making potentially misleading calculations based on incomplete information. It is important to note that the number of deaths from legal intervention available from the CDC is widely considered to be underreported.
• Data suppression: WISQARS suppresses data where the value is between one to nine deaths. This suppression is done to protect the privacy of individuals and ensure the confidentiality of sensitive information.

Do you have information that belongs on the Health Equity Tracker? We would love to hear from you!