Highlights
-
•
Visual impairment predicted higher all-cause mortality in NHANES 2001-2008.
-
•
Excess risk concentrated in non-cardiovascular, non-cancer deaths.
-
•
Cardiovascular mortality showed modest association; cancer mortality showed none.
-
•
Associations were consistent across thresholds and sensitivity analyses.
Objective
To determine whether the excess mortality associated with visual impairment (VI) in U.S. adults is driven primarily by non-cardiovascular, non-cancer causes of death rather than cardiovascular or malignant causes.
Design
Population-based cohort study.
Subjects, Participants, and/or Controls
Adults aged 40 years or older participating in the National Health and Nutrition Examination Survey (NHANES) 2001-2008 with measured presenting visual acuity in both eyes and eligibility for linkage to the National Death Index.
Methods, Intervention, or Testing
Presenting visual impairment was defined as visual acuity worse than 20/40 in the better-seeing eye using standardized examination protocols. Mortality outcomes were ascertained through December 31, 2019. Survey-weighted Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs), adjusting for demographic, socioeconomic, and clinical covariates.
Main Outcome Measures
All-cause mortality. Secondary outcomes included non-cardiovascular, non-cancer mortality, cardiovascular mortality, and cancer mortality.
Results
The analytic cohort included 11,938 participants, of whom 1179 (9.9%) had visual impairment at baseline. Over a mean follow-up of 13.2 years, visual impairment was most strongly associated with non-cardiovascular, non-cancer mortality (adjusted HR, 1.55; 95% CI, 1.30-1.84). Visual impairment was also associated with increased all-cause mortality (HR, 1.36; 95% CI, 1.17-1.57) and cardiovascular mortality (HR, 1.33; 95% CI, 1.02-1.72), but not cancer mortality (HR, 0.98; 95% CI, 0.69-1.40). Associations were robust across sensitivity analyses, including exclusion of participants with diabetes, exclusion of early deaths, and alternative visual impairment thresholds.
Conclusions
In this nationally representative cohort, visual impairment was independently associated with increased mortality, driven primarily by non-cardiovascular, non-cancer causes of death. These findings suggest that vision loss may serve as a marker of systemic vulnerability and support integrating vision assessment into chronic disease management and preventive care strategies.
Visual impairment (VI) affects millions of adults in the United States and represents a growing public health challenge as the population ages. In 2015, an estimated 1.0 million U.S. adults were blind and more than 3.2 million lived with VI, with projections suggesting that this burden will nearly double by 2050. Common causes of vision loss in older adults– including uncorrected refractive error, cataract, glaucoma, and diabetic retinopathy– remain highly prevalent despite the availability of effective preventive and therapeutic interventions.
Although VI is often conceptualized as a localized sensory deficit, accumulating evidence suggests that vision loss has far-reaching consequences for health and functional well-being. Prior studies have linked VI to increased risk of falls, reduced mobility, cognitive decline, psychosocial distress, and greater healthcare utilization. ,,,,,, Older adults with VI also frequently experience multimorbidity, including diabetes, hypertension, arthritis, heart failure, and other chronic conditions, underscoring the intersection between vision loss and systemic vulnerability. ,,,
Effective management of chronic disease depends on patients’ ability to engage in health self-management tasks such as medication adherence, symptom recognition, appointment attendance, and navigation of complex health systems. Vision loss may disrupt these processes by limiting the ability to read medication labels, use medical devices accurately, maintain health literacy, and access outpatient care. Consequently, VI may function not only as a marker of ocular disease but also as an indicator of diminished physiologic reserve and impaired capacity for health system engagement.
Consistent with this broader framing, multiple population-based studies have reported an association between VI and increased all-cause mortality. ,, However, large contemporary analyses have primarily emphasized cardiovascular or heart disease-specific mortality, often treating non-cardiovascular deaths as competing events rather than as outcomes of interest. As a result, whether excess mortality associated with VI is concentrated in non-cardiovascular, non-cancer causes remains insufficiently characterized.
We evaluated whether VI is disproportionately associated with non-cardiovascular, non-cancer mortality compared with cardiovascular or malignant causes of death.
METHODS
Data Source and Study Population
We conducted a population-based cohort study using data from the NHANES linked to the National Death Index. NHANES uses a complex multistage sampling design to generate nationally representative estimates of the noninstitutionalized U.S. population. We pooled 4 NHANES cycles (2001-2008) with standardized visual acuity (VA) exams and mortality linkage. Mortality follow-up was available through December 31, 2019.
Adults aged ≥ 40 years with presenting VA in both eyes and mortality linkage were included in the analysis. Participants with missing VA measurements or who were ineligible for mortality linkage were excluded. The final analytic sample included participants meeting all inclusion criteria after pooling across survey cycles.
Exposure: Visual Impairment
The primary exposure was presenting VI, defined using examination-based VA measured with participants’ usual correction. VA was assessed monocularly and recorded as Snellen denominators; the better-seeing eye was the eye with the lower denominator. VI was defined a priori as presenting VA worse than 20/40 in the better-seeing eye, consistent with commonly used clinical and public health thresholds. Presenting VI reflects functional vision in daily life and may result from heterogeneous etiologies, including uncorrected refractive error or irreversible ocular disease. VI was modeled as a binary variable. Sensitivity analyses evaluated alternative thresholds for VI. For all-cause mortality, we evaluated >20/30, >20/50, and >20/60. For cause-specific outcomes, we evaluated >20/50 and ≥20/200 (Supplementary Tables S1 and S2).
Outcomes
The primary outcome was all-cause mortality. Secondary outcomes included cause-specific mortality, categorized as cardiovascular mortality, cancer mortality, and non-cardiovascular, non-cancer mortality. Cause-specific analyses were conducted to characterize the contribution of cardiovascular, cancer, and non-cardiovascular, non-cancer causes to all-cause mortality. Cardiovascular mortality included heart disease or cerebrovascular deaths, cancer mortality included malignant neoplasms, and non-cardiovascular, non-cancer mortality included all remaining causes. Exploratory analyses further disaggregated this category into leading subcauses. Time-to-event was defined as months from the NHANES examination to death or censoring and was converted to years for analysis. In sensitivity analyses, mortality was alternatively classified as cardiovascular-kidney-metabolic (CKM) vs non-CKM mortality, consistent with the American Heart Association CKM framework.
Covariates
Covariates were selected a priori based on clinical relevance and prior literature and included age (continuous), sex, race and ethnicity, educational attainment, and poverty-income ratio. , Clinical and behavioral covariates included hypertension, body mass index, smoking status, and diabetes. Hypertension was defined as a mean systolic blood pressure of 140 mm Hg or greater or a mean diastolic blood pressure of 90 mm Hg or greater based on examination measurements. Smoking status was defined using serum cotinine levels, with levels of 10 ng/mL or greater indicating current smoking. Serum cotinine was selected to capture objective current tobacco exposure and minimize recall bias. In sensitivity analyses, smoking was alternatively modeled using a questionnaire-derived 3-category smoking history variable (never, former, or current smoker). Diabetes was defined as self-reported diabetes or hemoglobin A1c (HbA1c) of 6.5% or greater. Self-reported cataract and glaucoma were evaluated in sensitivity analyses but were not included in primary adjustment models. In sensitivity analyses, binary hypertension and diabetes variables were replaced with continuous measures of cardiometabolic severity, including mean arterial pressure (MAP) and HbA1c.
Survey Design and Weighting
All analyses incorporated NHANES examination sampling weights, primary sampling units, and strata to account for the complex survey design. To pool four 2-year survey cycles, 8-year examination weights were constructed by dividing the 2-year examination weights by 4, in accordance with NHANES analytic guidelines.
Statistical Analysis
Baseline characteristics were summarized by VI status using survey-weighted means and proportions. Differences between groups were assessed using survey-weighted t tests for continuous variables and Rao-Scott adjusted chi-square tests for categorical variables. Associations between VI and mortality outcomes were evaluated using survey-weighted Cox proportional hazards models, with cause-specific mortality models treating deaths from other causes as censored at the time of death. Sequential models were fit, including an unadjusted model, a demographics-adjusted model including age, sex, and race and ethnicity, and a fully adjusted model that additionally included education, poverty-income ratio, hypertension, body mass index, smoking status, and diabetes. Results were reported as hazard ratios with 95% CIs.
Kaplan-Meier survival curves were generated, with differences assessed using log-rank tests. Incremental prognostic value was evaluated by comparing nested survey-weighted Cox models with and without VI using design-based Wald tests. Model discrimination was summarized using Harrell’s C index.
Effect Modification and Sensitivity Analyses
Prespecified effect modification analyses examined whether associations differed by age, sex, socioeconomic status, educational attainment, and clinical risk burden. Stratified survey-weighted Cox models were fit within subgroups, and interaction terms were tested using joint design-based Wald tests.
Multiple sensitivity analyses were performed, including exclusion of deaths occurring within 1 and 2 years of follow-up to address potential reverse causation, alternative definitions of VI, stratification by diabetes status, and exclusion of participants with cataract or glaucoma.
Software
All analyses were conducted using R statistical software (R Foundation for Statistical Computing, version 4.5.2), with survey-weighted analyses performed using the survey and survival packages.
Ethics Approval
The study was deemed exempt from institutional review board review under 45 CFR §46.102(d) because it used publicly available, deidentified data. The study was conducted in accordance with the tenets of the Declaration of Helsinki, and informed consent was not required. All data use complied with applicable U.S. federal regulations, including the Health Insurance Portability and Accountability Act (HIPAA).
RESULTS
Study Population and Baseline Characteristics
The analytic cohort included a total of 11,938 participants aged 40 years or older, of whom 1179 (9.9%) had presenting VI at baseline. The mean (SD) age of the cohort was 56.6 (12.2) years, with participants with VI being significantly older than those without VI (63.5 vs 56.0 years; p <.001) ( Table 1 ).
Table 1
Baseline Characteristics of the Study Population By Visual Impairment Status
| Characteristic | Overall (N = 11,938) | No Visual Impairment (n = 10,759) | Visual Impairment (n = 1179) | P value |
|---|---|---|---|---|
| Age, years , mean (SD) | 56.6 (12.2) | 56.0 (11.8) | 63.5 (15.0) | <.001 |
| Body mass index , kg/m², mean (SD) | 28.9 (6.4) | 28.9 (6.3) | 28.4 (6.4) | .033 |
| Systolic blood pressure , mm Hg, mean (SD) | 127.6 (19.5) | 127.2 (19.0) | 133.6 (23.8) | <.001 |
| Diastolic blood pressure , mm Hg, mean (SD) | 72.4 (13.2) | 72.7 (12.8) | 68.6 (17.7) | <.001 |
| Hemoglobin A1c , %, mean (SD) | 5.7 (0.9) | 5.6 (0.9) | 5.9 (1.4) | <.001 |
| Age group , % | <.001 | |||
| 40-49 y | 35.6 | 36.2 | 28.4 | |
| 50-59 y | 28.5 | 29.5 | 14.6 | |
| 60-69 y | 17.8 | 18.0 | 15.2 | |
| 70-79 y | 12.0 | 11.4 | 18.7 | |
| ≥80 y | 6.1 | 4.9 | 23.0 | |
| Female sex , % | 52.4 | 52.2 | 54.7 | .23 |
| Race and ethnicity , % | <.001 | |||
| Mexican American | 5.0 | 4.9 | 7.1 | |
| Non-Hispanic Black | 9.8 | 9.6 | 12.9 | |
| Non-Hispanic White | 77.2 | 77.8 | 68.9 | |
| Other Hispanic | 3.5 | 3.3 | 5.7 | |
| Other or multiracial | 4.5 | 4.4 | 5.5 | |
| Educational attainment , % | <.001 | |||
| Less than 9th grade | 7.4 | 6.6 | 18.4 | |
| 9 th-11th grade | 11.2 | 10.9 | 15.0 | |
| High school/GED | 26.1 | 26.1 | 26.3 | |
| Some college/associate degree | 29.1 | 29.4 | 25.9 | |
| College graduate or higher | 26.2 | 27.0 | 14.4 | |
| Poverty-income ratio , % | <.001 | |||
| ≥3.0 | 56.2 | 57.7 | 35.0 | |
| 2.0-2.99 | 15.7 | 15.6 | 16.1 | |
| 1.0-1.99 | 18.7 | 17.9 | 29.5 | |
| Below poverty | 9.5 | 8.8 | 19.4 | |
| Hypertension , % | 24.9 | 24.4 | 32.8 | <.001 |
| Current smoker , % | 24.6 | 24.5 | 24.8 | .88 |
| Diabetes , % | 14.0 | 13.2 | 24.6 | <.001 |
| Cataract , % | 3.8 | 3.5 | 8.1 | <.001 |
| Glaucoma , % | 7.7 | 7.7 | 7.8 | .98 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
