Data were from the HRS database, a longitudinal panel study that surveys a nationally representative sample of Americans aged 50 years and older every two years, with follow-up continuing until death. The HRS uses a biennial, multistage area probability sample survey design, using a combination of in-person and telephone interviews to collect data. The survey focuses on community-dwelling individuals. This longitudinal design enables the tracking of participants’ health, economic, and social outcomes over time, providing a comprehensive understanding of aging and retirement in America. The HRS is supported by the National Institute on Aging (NIA U01AG009740) and the Social Security Administration. For this study, we used data from 14 waves of the HRS (1996‐2020).

The analytic subsample for this study comprises community-dwelling respondents aged 65 years and older from the 1996‐2020 HRS survey rounds, as the falls questionnaire was only administered to participants at or older than this age threshold, with all included variables having item nonresponse rates below 10%.

Since the HRS database is deidentified and publicly accessible, this study was exempt from the Mass General Brigham institutional review board approval and patient permission was waived. The study adhered to the tenets of the Declaration of Helsinki.

The primary outcome measures included self-reported falls, serious injury falls, and the number of falls. Participants were asked if they had fallen in the past two years. If they answered “yes,” they were then asked how many times they had fallen and whether any of the falls were serious. Serious injury falls were defined as those requiring medical treatment.

Ophthalmic parameters were acquired from the HRS survey assessments and included overall eyesight, distance vision, and near vision (see Multimedia Appendix 1), each rated on a scale from 1 to 5, with scores of 4 (fair) and 5 (poor) considered indicative of impaired vision. For overall eyesight, a score of 6 could be assigned for patients with legal blindness. The history of glaucoma was based on whether participants reported ever being treated for glaucoma.

The HRS also collected information on age, gender, race, level of education, marital status, self-reported diagnoses of hypertension, diabetes, stroke, heart conditions (including heart attack, coronary heart disease, angina, congestive heart failure, or other heart problems), arthritis or rheumatism, smoking and alcohol consumption status, psychiatric medication use, exercise activities, and household income.

The data were analyzed cross-sectionally and longitudinally to confirm the correlations between ophthalmic parameters and falls. Each participant observation was analyzed as individual encounters in both analyses. In the cross-sectional analysis, self-reported ophthalmic parameters and falls were ascertained from the same 2-year interview cycle. Each completed survey was counted as 1 entry. In the longitudinal analysis, the self-reported ophthalmic parameters were obtained during the survey and analyzed with the reported falls recorded in the patient’s survey 2 years later.

Data are presented as base-year weighted proportions, accounting for the Health and Retirement Study adjustments for attrition and sampling error. Descriptive statistics were used to summarize participant data on falls outcomes, including falls, no falls, serious falls, and nonserious falls. We used t tests for continuous variables and χ² tests for categorical variables to evaluate differences between outcomes. We used univariable and multivariable logistic regression models to determine the cross-sectional and longitudinal association between ophthalmic parameters and falls, adjusting for age and other potential confounders by calculating the odds ratio (OR) and their corresponding 95% CI. For the cross-sectional model, we assessed the association between vision status and reported falls in the same 2-year survey cycle, while for the longitudinal model, we considered falls reported in the subsequent 2-year survey cycle.

All variables with a P value of <.05 in the univariable analyses (including age, sex, Hispanic ethnicity, race, education, marital status, hypertension, diabetes, heart disease, stroke, arthritis, smoking, alcohol status, psychiatric medications, degree of exercise, and household income) were included in multivariable logistic regression models. Separate multivariable regression models were used to determine the relationship between each ophthalmic parameter and falls and serious falls. In addition, the association between the degree of vision impairment and the number of falls was assessed using a 1-way analysis of variance. All models were analyzed using generalized estimating equations for multiple observations from the same participant and multiple imputation techniques to replace the missing data. Data analysis was done with IBM SPSS Statistics 22.0 for Microsoft Windows and P values <.05 were considered statistically significant.

A total of 38,835 respondents contributed to 117,834 participant-observations (see Tables 1 and 2). The total number of observations in which participants reported a total of 40,477 falls and 13,471 serious falls. The weighted proportion of participants who reported any falls was OR 37.9% (95% CI 37.7%‐40.1%), while the proportion reporting serious falls was OR 30.9% (95% CI 30.7%‐31.1%). Furthermore, the weighted proportion of participants who experienced falls in the future was OR 36.5% (95% CI 35.8%‐37.2%), and OR 32.5% (95% CI 32.2%‐32.9%) for future serious falls. Falls increased with older age, and over half of those aged 90 years and older reported falling. The proportion of falls qualifying as serious also increased with age. White people had the highest fall rate, and a higher proportion of falls and serious falls was observed in females and individuals with lower levels of education. In total, 50% of patients with a history of stroke reported falling, with one-third of the falls being serious falls.

Higher rates of any type of falls and serious falls were observed with worse eyesight (see Tables 3 and 4). Approximately half of the participants who reported poor vision or blindness also reported falling, while 42.8% (95% CI 38.3%‐46.7%) of those with blindness had serious injuries associated with the fall. Of the participants with poor distant and near vision, 53.2% (95% CI 52.7%‐54.1%) and 50.2% (95% CI 49.9%‐50.6%) reported falls, respectively, and over a third of the falls were serious. In addition, falls were reported by 39.6% (95% CI 38.8%‐40.5%) of those who stated they had glaucoma.

Cross-sectionally, vision impairment was associated with higher odds of falls: impaired overall eyesight (aOR 1.36, 95% CI 1.20‐1.56; P<.001), impaired distance vision (aOR 1.37, 95% CI 1.32‐1.42; P<.001), and impaired near vision (aOR 1.33, 95% CI 1.27‐1.37; P<.001) all conferred significantly increased odds of falls after adjusting for confounding factors (see Table 5 and Figure 1). The odds of serious falls increased in impaired overall eyesight with adjusted ORs of 1.20 (95% CI 1.03‐1.44; P<.001), while having overall eyesight qualifying as legal blindness increased the odds of falls (aOR 1.32, 95% CI 1.26‐1.37; P<.001) and serious falls (aOR 1.16, 95% CI 1.03‐1.21; P<.001) compared to not having impaired vision. Self-reported glaucoma increased the adjusted odds of falls (aOR 1.15, 95% CI 1.07‐1.24; P<.001) and serious falls (aOR 1.15, 95% CI 1.05‐1.26; P<.001).

In the longitudinal analysis of the association between ophthalmic parameters with falls and serious falls, impaired overall eyesight, impaired distance vision, impaired near vision, and self-reported glaucoma all increased the odds of future falls with an OR of 1.23 (95% CI 1.16‐1.29; P<.001), 1.27 (95% CI 1.20‐1.38; P<.001), 1.23 (95% CI 1.19‐1.32; P<.001), and 1.25 (95% CI 1.13‐1.37; P<.001), respectively (see Table 6 and Figure 2). Impaired overall eyesight, distance vision, and near vision also increased the risk of serious falls in the subsequent 2 years with an OR of 1.18 (95% CI 1.05‐1.23; P<.001), 1.13 (95% CI 1.04‐1.21; P<.001), and 1.15 (95% CI 1.07‐1.22; P<.001), respectively. There was no association between self-reported glaucoma and the odds of future serious falls (P=.32).

The mean number of falls was associated with the degree of self-reported vision impairment (see Table 7). The level of eyesight reported in the questionnaire was associated with the reported number of falls in the same survey cycle as well as the subsequent cycle (P<.001 for both). Approximately 2.57 (SD 4.58) falls per person who fell were observed in the group with excellent current eyesight compared to 3.70 (SD 5.33) for those with poor reported vision. Similarly, self-reported excellent overall eyesight had an average of 2.78 (SD 4.17) falls per person while poor overall eyesight had approximately 3.65 (SD 5.40) falls per person in the subsequent cycle.

In this large and diverse longitudinal population-based study of older adults, poor vision, including both distance and near vision, increased the odds of falling, and more severe vision loss was related to the number of falls. We additionally found that the presence of self-reported glaucoma was associated with falls.

Many population-based studies have found associations between self-reported vision impairment and falls. The Survey of Health, Ageing, and Retirement in Europe (SHARE) study found that self-reported distance and near vision impairment were cross-sectionally and longitudinally associated with increased risk of falls [21]. The Beaver Dam Eye Study reported that people with worse binocular acuity had higher odds of 2 or more falls [12], while the EPIC-Norfolk eye study found significant associations between poor self-reported distance vision and falls independent of visual acuity [20]. Similarly, we demonstrate here that the presence of visual impairment, including overall eyesight, near vision, and distance vision, is all associated with significantly increased odds for falls and severe falls in both our cross-sectional and longitudinal analyses. Our results suggest that self-reported visual status is predictive of falls and serious falls.

Importantly, the degree of self-reported visual impairment was associated with the number of falls even after adjusting for potential confounders. When looking at the mean number of falls that occur per person for each level of reported vision, we found that though people with excellent vision had falls, the number increased significantly as vision worsened, with those with blindness having the highest risk of falls. This is true for both falls that occurred 2 years before and those that occurred in the following 2 years of the reported vision and does not take into account the fact that blindness likely reduces the number of steps taken.

This study also subclassified falls as serious based on whether a fall had any associated injuries requiring medical treatment. The incidence of serious falls increased with age and deterioration of eyesight, and we found the risk factors for falls and serious falls to be similar. Other studies, including a Finnish population-based study, have examined risk factors of major injurious falls and similarly found poor distant visual acuity to be a risk factor for major falls among disabled elder adults [22].

Visual input plays a critical role in coordinating and planning movement, as well as maintaining balance [23]. Individuals with poor vision are more likely to lose stability, alter their gait to avoid obstacles, and ultimately increase their risk of slips, trips, and falls [24]. Our study differs from previous research in this area, which has largely relied on data from previous falls and visual function assessments. By using self-reported data, our research offers a fresh perspective and highlights the potential of self-reported visual function data as a predictive tool for falls risk.

We also found that glaucoma is a risk factor for serious falls resulting in injury despite the fact that our definition of glaucoma was imperfect (based on self-report only). Any nondifferential misclassification in diagnosing glaucoma would have driven our associations toward the null, and this implies that the associations found were real and may have been underestimated. Previous studies have also reported glaucoma as an independent risk factor for falling [13]. Glaucoma severity has been reported as a risk factor for falls, with advanced glaucoma patients having the highest risk of falls compared to controls with no glaucoma and mild or moderate glaucoma [25]. While central vision and, therefore, visual acuity are often preserved in glaucoma until severe stages of disease, glaucoma patients experience changes in multiple other aspects of vision, including VF loss, decrease in contrast sensitivity, and loss of depth perception. All these visual measures have been shown to independently increase the risk of falling [13,14,26]. Indeed, the Salisbury Eye Evaluation found that loss in peripheral VF is a more important risk factor for falls than loss of central VF [14]. Patients with glaucoma may experience one or all visual deficits simultaneously, further increasing the risk of falling.

Studies have suggested various interventions to reduce the risk of falls, including mobility training, vision interventions to improve vision (eye examinations, cataract surgery, and prescription glasses) [27], risk factor assessment, and environmental modifications to make living environment less hazardous [28]. Visual impairment leads to physical impairment of mobility and balance control, increasing the risk of falls [29]. Glaucoma initially affects peripheral vision through retinal nerve fiber layer loss, which is associated with decreased postural stability; therefore, glaucoma patients may benefit from mobility training to reduce falling risks [30,31].

Our results provide evidence to support the development of multifaceted fall prevention strategies. Specifically, our findings suggest that self-reported vision screening could be a valuable tool in identifying individuals at risk of falls. This screening could be easily integrated into telemedicine platforms or incorporated into standard clinical guidelines. Furthermore, we propose the development of a simple, score-based questionnaire that can be used to predict falls risk in older adults. Those with higher scores could be targeted for earlier intervention, thereby reducing their risk of falls.

There were several limitations to our study. All data used in this study were from self-reported questionnaires administered every 2 years and are susceptible to recall bias and recall error. Specifically, participants may have difficulty recalling their visual function or falls history, which could lead to inaccurate reporting. In addition, while self-reported visual function has been shown to correlate closely with objective visual function, it is also understood as a multidimensional measure that reflects some, but not all components of vision [19]. Since no eye examinations were done and objective visual function was not measured, we cannot determine which visual components contributed to participants’ self-reported visual function. This study is also limited by the specific questions collected as part of the HRS study. No questions were asked about the other leading causes of blindness and visual impairment in this older adult population, such as age-related macular degeneration and other neurological conditions that could affect falls, such as Parkinson disease, dementia, or epilepsy. Finally, we could not explore other known vision-related risk factors for falls such as contrast sensitivity, VF deterioration, depth perception, and stereoacuity [26]. Nonvision factors such as cognitive ability, balance, motor control, and degree of mobility, which are known to be associated with falls, were not evaluated since they were not included in the questionnaire. The omission of these factors may limit the generalizability of our findings and highlights the need for future studies to incorporate a more comprehensive set of measures.

Our analysis of data from a large, longitudinal national survey of older adults demonstrated that vision-related risk factors, particularly the presence and degree of visual impairment and glaucoma, increase the rate of falls and serious falls. We additionally show that self-reported vision status correlates with the number of fall events both at the time of and occurring after the survey date. Our results support screening for the presence of visual impairment and signs of glaucoma in older adults to identify individuals at the highest risk of falling who might most benefit from fall prevention interventions.