A study found that nearly 20% of dementia cases among older adults in the United States may be attributable to vision impairment.
In a cross-sectional analysis, published in JAMA Ophthalmology investigators used data from the 2021 National Health and Aging Trends Study to analyze the outcomes of 2,767 community-dwelling adults aged 71 years and older. The survey-weighted prevalence of vision impairment among the participants was 32.2% (95% confidence interval [CI] = 29.7%–34.6%). Among those with at least one vision impairment, near acuity impairment (65.3%, 95% CI = 60.9%–69.7%) and contrast sensitivity impairment (63.1%, 95% CI = 59.0%–67.3%) were twice as prevalent as distance acuity impairment (30.1%, 95% CI = 26.0%–34.3%).
The study defined vision impairments as:
- Near and distance visual acuity impairments: > 0.30 logMAR (Snellen equivalent, > 20/40)
- Contrast sensitivity impairment: < 1.55 logCS
Dementia was defined using a standardized algorithmic diagnosis, incorporating cognitive test scores, self- or proxy-reported dementia diagnosis, and the Ascertain Dementia–8 Dementia Screening Interview Score.
The investigators used Poisson regression models with robust variance to estimate prevalence ratios of dementia, adjusting for age, self-reported sex, race, and chronic disease multimorbidity. They calculated PAFs using the formula: PAF = exposure prevalence among cases × (PR − 1)/PR, where PR is the prevalence ratio.
Among the key findings were:
- The population attributable fraction (PAF) of dementia from at least one vision impairment was 19.0% (95% CI = 8.2%–29.7%).
- Contrast sensitivity impairment yielded the strongest PAF at 15.0% (95% CI = 6.6%–23.6%) followed by near acuity impairment at 9.7% (95% CI = 2.6%–17.0%) and distance acuity impairment at 4.9% (95% CI = 0.1%–9.9%).
- PAFs from at least one vision impairment were highest among participants aged 71 to 79 years (24.3%, 95% CI = 6.6%–41.8%), female participants (26.8%, 95% CI = 12.2%–39.9%), and non-Hispanic White participants (22.3%, 95% CI = 9.6%–34.5%).
Additional results included:
- PAFs for at least one vision impairment decreased with age, from 24% at ages 71 to 79 years to 8% at 90 years and older.
- The PAF from at least one vision impairment was fourfold greater among female participants (27%) compared with male participants (7%).
- PAFs from contrast sensitivity impairment were threefold greater among non-Hispanic White participants (20%) compared with non-Hispanic Black participants (6%).
- PAFs were largely similar across educational levels for at least one vision impairment and contrast sensitivity impairment.
- The prevalence of at least one vision impairment was higher among those with dementia (52.6%, 95% CI = 46.6%–58.6%) compared with the overall sample (32.2%, 95% CI = 29.7%–34.6%).
- The prevalence ratio of dementia for those with at least one vision impairment was 1.57 (95% CI = 1.24–1.98).
Sensitivity analyses revealed:
- PAFs for at least one vision impairment and contrast sensitivity were stronger among the participants without diabetes. For at least one vision impairment, the PAF was nearly threefold greater among those without diabetes (25.0%, 95% CI = 11.4%–38.0%) compared with those with diabetes (8.4%, 95% CI = −10.2% to 27.3%).
- Restricting the dementia definition to only probable dementia cases resulted in larger PAFs (at least one vision impairment: 24.9%, 95% CI = 9.1%–40.2%).
- Defining contrast sensitivity impairment based on standard deviations from the study sample mean lowered PAF estimates by at least 72%. However, parameterizing by quartiles indicated that 30% of dementia cases could be attributed to logCS values below the median, relative to the highest quartile.
The study's strengths included its use of a nationally representative sample, collection of multiple presenting binocular measures of objective vision impairment, and oversampling of adults aged 90 years and older. Limitations included the cross-sectional design, which precluded establishing temporality, and limited sample sizes for certain racial and ethnic groups.
Conflict of interest disclosures can be found in the study.