Hawaiʻi’s Longevity Is Predictable: Why Population Health Should Not Be Mistaken for Healthcare Performance
Hawaiʻi Life Expectancy and Healthcare Rankings | Task Force Report { "@context": "https://schema.org", "@type": "Report", "headline": "Hawaiʻi’s Longevity Is Predictable", "alternativeHeadline": "Why
HHTF Staff · 2026-09-23
Hawaiʻi’s Longevity Is Predictable
Why Population Health Should Not Be Mistaken for Healthcare Performance Hawaiʻi Healthcare Task Force Research Report · Analysis of 2022 state data
Executive Summary
Hawaiʻi is frequently cited as one of the highest-performing states in national healthcare rankings. U.S. News & World Report ranked Hawaiʻi first for healthcare in 2025, and the Commonwealth Fund ranked Hawaiʻi second in its 2025 State Health System Performance Scorecard.
These rankings combine measures of healthcare delivery with measures of insurance coverage and population health. That creates an interpretive problem because favorable population outcomes can improve a state’s healthcare ranking even when those outcomes are associated with socioeconomic and demographic conditions whose contributions cannot be separated from healthcare delivery in the ranking itself.
Using 2022 data from all 50 states, we examined how much variation in state life expectancy could be accounted for before introducing any direct measure of healthcare delivery. Median household income and poverty rate alone accounted for approximately 74% of the state-to-state variation in this dataset. A combined socioeconomic and demographic model accounted for approximately 86%, although that fit requires caution because of the number of predictors relative to the 50 observations and the absence of external validation.
Hawaiʻi’s observed life expectancy was 80.0 years, while an exploratory estimate based on population composition was approximately 80 years. The income and poverty model estimated approximately 79.1 years, and the combined model approximately 79.8 years. The demographic calculation relies on imperfect national categories and assumptions for populations not adequately represented in available life tables; it is not a finding about the expected lifespan of any racial or ethnic group.
The results do not establish that healthcare has little effect on longevity or that the associations are causal. They show that models using population characteristics can reproduce much of the observed pattern without measuring healthcare delivery directly, limiting what longevity alone can establish about healthcare-system performance.
Background
The Hawaiʻi Healthcare Task Force previously examined the distinction between insurance coverage and healthcare access in Insurance Is Not Access . Insurance coverage establishes a financing mechanism, but it does not establish whether clinicians, specialists, hospital services, or timely appointments are available.
A similar distinction applies to longevity. Life expectancy reflects the cumulative effects of socioeconomic conditions, demographic composition, health behaviors, environmental exposures, public health, migration, and medical care. When mortality and longevity measures are incorporated into healthcare rankings, part of the resulting score may reflect characteristics of the population rather than performance of the healthcare delivery system.
In 2022, Hawaiʻi had the highest life expectancy at birth in the United States at 80.0 years. West Virginia had the lowest at 72.2 years. The 7.8-year difference is substantial, but the extent to which it reflects healthcare delivery is not apparent from the life-expectancy statistic itself.
Research Question
The analysis asked how much of the variation in state life expectancy could be accounted for by income, poverty, and population composition before introducing measures of physician supply, specialist availability, hospital capacity, travel burden, network adequacy, appointment waiting time, or other features of healthcare delivery. Its purpose was to examine the fit of population-based models, not to construct a complete causal explanation of longevity or estimate healthcare’s independent contribution.
Methods
State life expectancy at birth was obtained from CDC/NCHS State Life Tables for 2022. Median household income, poverty rate, and race and ethnicity composition were obtained from the 2022 American Community Survey.
The primary regression model included median household income and poverty rate. A separate exploratory demographic estimate used each state’s race and ethnicity composition and available national life-expectancy estimates by race and Hispanic origin. A third model combined socioeconomic and demographic variables.
The demographic analysis requires caution. Race and ethnicity are population descriptors associated with differences in migration, income, education, occupation, discrimination, neighborhood conditions, public policy, environmental exposure, and other lifetime influences. They are not treated here as biological determinants of longevity.
National life tables do not provide directly comparable life-expectancy estimates for every population represented in Hawaiʻi. Multiracial populations and Native Hawaiian and Other Pacific Islander populations are particularly difficult to map to available national estimates. The demographic calculation therefore relies on assumptions and should not be interpreted as a finding about the expected lifespan of any racial or ethnic group.
Table 1. Analytic Approach Analysis Variables Purpose
Socioeconomic model Median household income and poverty rate Estimate the association between economic conditions and state longevity
Exploratory demographic estimate State race and ethnicity composition weighted by available national life-expectancy estimates Examine the relationship between population composition and observed longevity
Combined model Income, poverty, and demographic composition Examine how closely state longevity can be fitted before healthcare delivery is measured
These are state-level, cross-sectional comparisons. Associations among states cannot establish individual-level relationships, causal effects, or how much longevity is attributable to medical care. Income and demographic variables may also capture differences in healthcare access indirectly, even though no direct delivery measures enter the models.
The demographic calculation is an approximation rather than a reconstructed state life table. Weighting group life expectancies by population shares does not, in general, produce the life expectancy that would be calculated from the population’s age-specific mortality rates. The model results should therefore be read as exploratory associations and in-sample fits.
Findings
Socioeconomic Conditions
Median household income and poverty rate alone accounted for approximately 74% of the state-to-state variation in life expectancy in the fitted dataset. This is a measure of statistical fit, not the proportion of longevity caused by income and poverty.
The association is also visible in a simple comparison of states by income. The 10 states with the lowest median household incomes had an average life expectancy of approximately 73.85 years. The 10 states with the highest median household incomes averaged approximately 79.05 years, a difference of about 5.2 years.
Demographic Composition
The exploratory demographic estimate accounted for approximately 22% of state-to-state variation. Population composition contributed information, but it was a substantially weaker national predictor than income and poverty.
When socioeconomic and demographic variables were combined, the fitted model accounted for approximately 86% of state-to-state variation. Because this model contains several predictors and only 50 observations, that result should be interpreted as an in-sample association rather than evidence of validated predictive performance.
Table 2. Model Results Model Variation accounted for
Income and poverty 74%
Exploratory demographic estimate 22%
Combined socioeconomic and demographic model 86%
State-Level Comparison
The combined model follows much of the national distribution of reported life expectancy despite containing no direct measure of healthcare delivery. The comparison is included to show the fit across the entire country rather than only at the highest- and lowest-life-expectancy states.
These are in-sample fitted values from the same states used to construct the model. The differences between modeled and observed values should not be treated as validated estimates of healthcare performance, and several states show substantial residual differences.
Table 3. Reported 2022 Life Expectancy Compared With the Combined Model, in Years State Reported Fitted estimate Observed minus estimate
Hawaiʻi 80.0 79.8 +0.2
Massachusetts 79.8 79.8 0.0
New Jersey 79.6 80.2 −0.6
New York 79.5 78.3 +1.2
Connecticut 79.4 78.8 +0.6
California 79.3 79.5 −0.2
Minnesota 79.3 78.4 +0.9
Rhode Island 79.2 78.3 +0.9
Utah 79.0 79.1 −0.1
New Hampshire 78.7 79.1 −0.4
Colorado 78.5 78.2 +0.3
Idaho 78.4 77.5 +0.9
Washington 78.4 78.8 −0.4
Nebraska 78.3 77.2 +1.1
Vermont 78.3 77.4 +0.9
Wisconsin 78.1 77.4 +0.7
Florida 77.9 77.1 +0.8
Iowa 77.9 77.3 +0.6
North Dakota 77.9 77.1 +0.8
Maryland 77.8 78.0 −0.2
Oregon 77.7 77.3 +0.4
Illinois 77.5 77.1 +0.4
Virginia 77.5 78.1 −0.6
Montana 77.3 77.3 0.0
Pennsylvania 77.3 77.2 +0.1
South Dakota 77.3 76.2 +1.1
Texas 77.1 76.7 +0.4
Michigan 76.8 76.0 +0.8
Wyoming 76.8 77.9 −1.1
Arizona 76.7 77.3 −0.6
Maine 76.6 77.3 −0.7
Delaware 76.5 77.6 −1.1
Kansas 76.5 76.8 −0.3
Nevada 76.4 77.8 −1.4
Georgia 75.9 75.9 0.0
North Carolina 75.9 76.1 −0.2
Alaska 75.8 76.6 −0.8
Ohio 75.6 75.9 −0.3
Indiana 75.4 76.7 −1.3
Missouri 75.2 75.8 −0.6
South Carolina 75.1 75.5 −0.4
New Mexico 74.5 74.8 −0.3
Arkansas 73.9 73.4 +0.5
Alabama 73.8 74.1 −0.3
Louisiana 73.8 72.7 +1.1
Oklahoma 73.8 73.3 +0.5
Tennessee 73.8 75.7 −1.9
Kentucky 73.6 74.3 −0.7
Mississippi 72.6 72.0 +0.6
West Virginia 72.2 73.8 −1.6
Values are displayed to one decimal place for comparison, not as an indication of predictive precision. Positive residuals indicate observed life expectancy above the fitted estimate; negative residuals indicate observed life expectancy below it.
The model reproduces much of the national longevity pattern without directly incorporating healthcare capacity or access. Tennessee and West Virginia illustrate its limitations, with observed life expectancy substantially below the fitted estimates. Such differences may reflect omitted influences, measurement limitations, or model specification as well as potentially relevant differences in healthcare.
Hawaiʻi
Hawaiʻi’s results differ from the national pattern in one respect. Demographic composition alone was a modest predictor nationally, but the exploratory demographic estimate for Hawaiʻi was close to the state’s observed life expectancy.
Hawaiʻi’s observed life expectancy in 2022 was 80.0 years. The exploratory demographic estimate was approximately 80 years, the income and poverty model estimated approximately 79.1 years, and the combined model estimated approximately 79.8 years.
Table 4. Hawaiʻi Observed and Estimated Life Expectancy Measure Life expectancy
Observed Hawaiʻi life expectancy 80.0 years
Exploratory demographic composition estimate About 80 years
Income and poverty model 79.1 years
Combined socioeconomic and demographic model 79.8 years
The proximity of these estimates does not establish the contribution of healthcare to longevity. It shows that Hawaiʻi’s observed value is close to the fitted values from these population-based models. Because the models do not isolate healthcare’s effects, a small residual cannot establish either superior or inferior healthcare delivery.
Demographic Data Limitations
Hawaiʻi has an unusually large multiracial population, and standard national race and ethnicity categories describe the state poorly. More than one in five Hawaiʻi residents in the 2022 ACS dataset identified as non-Hispanic multiracial, while roughly one-third were classified as non-Hispanic Asian alone.
Many residents classified as multiracial have Asian, Native Hawaiian or Pacific Islander, White, or other ancestry in combinations that are not represented in national life tables. National non-Hispanic Asian life expectancy is substantially higher than the national average, but assigning that estimate to multiracial Hawaiʻi residents would not be methodologically justified.
The exploratory demographic estimate should therefore remain limited in interpretation. Its proximity to Hawaiʻi’s observed life expectancy is sensitive to the assumptions used, and the available data do not support stronger conclusions about the contribution of any specific racial or ethnic group.
Residual Life Expectancy as a Research Question
The difference between expected and observed life expectancy may be useful for health-system research. A state that consistently performs above or below a reasonable population-based expectation could be examined for factors that are not captured by socioeconomic and demographic variables.
Those factors could include healthcare quality, primary care access, specialist supply, hospital capacity, public health programs, health behaviors, environmental conditions, migration, violence, substance use, or other social conditions. The residual itself cannot be labeled a healthcare-quality measure because the model does not distinguish among those influences.
A stronger next analysis would test whether residual life expectancy is associated with direct measures of healthcare delivery, including clinician supply, appointment waiting time, specialist availability, geographic access, preventable hospitalization, network adequacy, continuity of care, and local service capacity. It would al