U.S. Study Finds a Shrinking Disease-Free Share of Adults

A report dated October 13, 2026, described UTMB’s study “Sicker at all ages? Two decades of changing disease histories and mortality in the United States,” published in the International Journal of Epidemiology.

What was the study’s central finding?

The researchers found that the share of adults classified as disease-free declined by nearly 1% annually during the study period. Cardiometabolic and high-burden multimorbidity profiles accounted primarily for the shift toward greater illness burden. Their analysis also found that improvements in mortality within health profiles drove most of the overall mortality decline, while the changing mix of profiles partially offset those gains. For healthy-aging research, the central message is that improving survival within a health category and reducing the population’s burden of illness are distinct goals that need to be considered together.

Who was studied, and what did researchers measure?

Conducted in humans: This was an observational analysis of population survey responses linked to mortality records.

The researchers analyzed 510,372 U.S. adults aged 30 and older. They used National Health Interview Survey data collected from 1997 to 2018 and linked those responses to mortality records with follow-up through 2019. This connected information about reported health problems with subsequent deaths.

The analysis included reported cancer, diabetes and cardiovascular disease. It also covered mental-health problems, internal diseases, obesity and physical limitations. Rather than examine each condition separately, the researchers used latent-class analysis to identify groups with similar combinations of health problems.

That approach put combinations of conditions at the center of the analysis. The authors’ aim was to understand population health through both the prevalence of different health profiles and the mortality associated with them. Their framework therefore considered illness burden alongside survival.

How did the five health profiles differ?

The researchers identified five profiles: disease-free, cardiometabolic and moderate multimorbidity, plus mental-physical burden and high-burden multimorbidity. Multimorbidity refers here to combinations of health conditions rather than a single diagnosis considered in isolation.

Across the study population, 62.4% of participants belonged to the disease-free profile. The high-burden multimorbidity profile accounted for 5.9%. The remaining participants were classified within the other profiles according to their reported combinations of conditions.

The authors reported that the shrinking disease-free share was primarily replaced by cardiometabolic and high-burden profiles. They also found substantial increases in two conditions included in the analysis. From 1997 to 2018, obesity prevalence rose by almost 15 percentage points and diabetes prevalence rose by 6 to 8 percentage points.

The profile approach gives those individual-condition trends a broader context. Its focus is not simply whether one diagnosis became more common. It asks how the overall pattern of reported health problems changed across the population.

What happened to mortality within the profiles?

The study’s five-year death probabilities differed substantially between the disease-free and high-burden profiles. Among men, the estimates were 3.7% for the disease-free profile and 17.8% for high-burden multimorbidity. Among women, the corresponding estimates were 2.4% and 12.6%.

The researchers also examined how mortality changed over time within the profiles. Mortality generally declined through the mid-2000s. After that period, the authors reported stagnating or increasing mortality for most profiles. The disease-free profile was the exception and continued to show improvement.

This adds an important distinction to the headline finding. The study described both a shift toward more burdensome health profiles and changes in survival within those profiles. Those are separate parts of the population-health picture.

Why separate the population’s health mix from survival within each group?

The authors used a decomposition analysis to distinguish these contributions. Improvements in mortality within profiles were the dominant contributor to declining overall mortality. Changes in the prevalence of the profiles worked in the opposite direction and partially offset that progress.

In plain language, overall mortality reflects more than the death risk associated with a given health profile. It also reflects how many people belong to each profile. The authors’ analysis treated both components as relevant to understanding the national trend.

For longevity research, this makes the distinction between lifespan and healthspan central to interpreting the findings. A population-health assessment needs to ask about survival and the burden of illness experienced alongside it. Looking at mortality alone would leave out the changing distribution of reported health problems documented in this study.

The authors interpreted their results as support for stronger prevention and chronic-disease management. Their broader argument was that tracking combinations of conditions could help identify opportunities for intervention. This directs attention toward the health problems people live with, alongside the causes and timing of death.

What does younger-adult research add to the context?

A separate 2026 CDC report adds a younger-adult perspective. Among U.S. adults aged 18 to 29, frequent mental distress rose from 12.7% in 2012 to 21.3% in 2024. Physical inactivity increased from 15.9% to 16.8% over the same period.

These findings bring younger adulthood into the healthy-aging discussion. Read alongside the UTMB analysis, they raise a useful editorial question: how should researchers track health burdens across adulthood rather than focus only on illness near the end of life?

What should limit how these findings are used?

• The study did not test causes or treatments. Its observational design identifies changing patterns without establishing why they occurred. The authors’ prevention and chronic-disease management recommendations are research-informed priorities rather than trial results. Claims about a particular diet, supplement, drug or longevity program require evidence that tests that intervention directly.
• Health information was self-reported. The authors noted that more detailed information would have been useful. Changing awareness or stigma could affect reports of mental-health problems, while reported height and weight could underestimate obesity prevalence. Interpret the trends as changes in reported health profiles rather than a complete clinical record of every participant.
• The profiles do not reveal the sequence of diagnoses or biological age. They describe combinations of conditions rather than the order in which those conditions developed. Longitudinal health data would be needed to examine those sequences more deeply. Keep these categories separate from claims made about molecular measures of aging.
• The mortality analysis had a limited window and population coverage. It used five-year follow-up and excluded people aged 85 and older. The authors noted that community-dwelling respondents at those ages would be an unrepresentatively healthy subgroup. Exclusion of the institutionalized population also limits generalization, so the analysis should not stand in for a complete account of the oldest adults.
• Some mortality changes were imprecise. Confidence intervals overlapped across adjacent time periods for some profiles. That means smaller period-to-period changes deserve caution. The broad pattern is more useful than treating every movement between neighboring periods as an established change or assigning a precise turning point to each group.
• Disease profiles are not a direct measurement of healthspan. The researchers measured reported health conditions and mortality rather than the number of years lived free of disability. The disease-free label also refers to the conditions captured by this analysis. Use the findings to discuss population illness burden without translating them into years of healthy life gained or lost.
• Group mortality estimates are not personal forecasts. The reported probabilities summarize outcomes associated with profiles across the study population. They are not individualized risk assessments. Personal questions about existing conditions or future health belong in a discussion with qualified healthcare professionals rather than a self-assigned study category.
• The CDC report is adjacent context rather than a replication. It examined a younger age group and different measures from the UTMB study. Its findings should therefore be interpreted within that report’s own scope. The two analyses should not be combined into a single estimate of multimorbidity or treated as interchangeable confirmation.

What evidence should come next?

The next useful phase is longitudinal research that follows how conditions accumulate, alongside intervention trials testing whether prevention and multimorbidity management improve survival and time lived in good health.

How AgeAmaze helps

For a team’s health-research lead, interpreting multimorbidity headlines means checking the evidence behind longevity claims; AgeAmaze makes that science more readable. The publication puts disease-profile findings in context so research-minded adults can distinguish population trends from overstated claims about supplements and emerging therapies without treating early research as a health promise.

Read the research

Sources

1. Sicker at all ages? Two decades of changing disease histories and mortality in the United States
2. Mental Distress and Physical Inactivity Among Adults Aged 18–29 Years