Prevalence and Causes of Visual Field Loss in the Elderly and Associations With Impairment in Daily Functioning: The Rotterdam Study

Objectives  To determine the prevalence and causes of visual field loss (VFL) and the association between VFL and indicators of impairment in daily functioning.

Design  Population-based cohort study.

Setting  Suburb of Rotterdam, the Netherlands.

Participants  Community-dwelling elderly residents (n = 6250).

Main Outcome Measure  Visual field loss on suprathreshold static, Goldmann kinetic perimetry, or both.

Methods  Suprathreshold testing of the central visual field was performed on both eyes and repeated if results were abnormal or unreliable. Goldmann perimetry was performed to confirm defects. Causes were determined using ophthalmologic and neurologic examination data and medical records. Impairment was assessed using data from interviews and medical records on disability in daily life, falling, and fractures.

Results  The overall prevalence of VFL was 5.6% (3.0% in those aged 55-64 years to 17.0% in those ≥85 years); glaucoma was the leading cause in all age groups. Before age 75 years, other optic nerve diseases and stroke ranked second and third, respectively, as did age-related macular degeneration and retinal vascular occlusive disease, respectively, after this age. Also, after adjustment for visual acuity, VFL was associated with disability, diminished enjoyment of reading and watching television, and a higher risk of incident falling. Risk of incident hip fracture was not increased.

Conclusions  Visual field loss is present in 1 of every 20 community-dwelling elderly people and is associated with impaired daily functioning. Glaucoma is the leading cause in all age groups. Other high-ranking causes, some of which are partly preventable, vary by age.

DECLINE IN visual function with age is well known and is a source of major concern in the elderly population. Besides visual acuity, visual function also largely depends on the intactness of the visual field (VF).¹ The prevalence of visual acuity loss in the elderly population has been studied,² but data on the prevalence of VF loss (VFL) are sparse. Prevalences of VFL ranging from 13% to 17% have been reported in adults 40 years and older³ and in elderly driver's license applicants.⁴ These studies did not investigate the causes of VFL.

The presence of VFL may jeopardize the ability of community-dwelling elderly persons to maintain their independence. Impairment in daily functioning is associated with decreased visual acuity,⁵ but results of recent studies^6,7 indicate that VFL may also cause impairment. Visual field loss is associated with frequent falls and decreased quality of life.^6,7 Persons with VFL perceive more difficulty in pursuing daily activities even in the presence of good visual acuity.⁸ These studies were cross-sectional and clinic based^7,8 or did not exclude the effect of low visual acuity.⁶ To plan research and develop strategies for diagnosis of, prevention of, and therapy for disabling eye disease, accurate data on causes of VFL are needed. We set out to determine the prevalence and causes of VF defects and the association of VFL with daily functioning in community-dwelling white individuals, 55 years and older, in a large population-based study.

This study was conducted as part of the Rotterdam Study,⁹ a prospective, population-based study of determinants and prognosis of chronic diseases in patients 55 years and older living in Ommoorel, an urban district of Rotterdam, the Netherlands. The study has been approved by the medical ethics committee of Erasmus University Medical School, Rotterdam, and was conducted in accordance with the Declaration of Helsinki. Baseline investigations(a home interview and ophthalmologic and neurologic examinations at the research center within this district) were conducted between January 1, 1990, and January 1, 1993. Between September 6, 1993, and December 29, 1995, participants were interviewed again at home a mean of 3 years after their last visit. The present study is confined to community-dwelling individuals (ie, it excludes those in nursing homes).

Details of the eye examination have been described previously.² The examination included indirect ophthalmoscopy of the central and peripheral retinas, photography of the macular area (35 field), simultaneous stereophotography of the optic disc (20 field), and VF testing. The VF of each eye was screened using a 52-point suprathreshold test that covered the central part of the field with a 24° radius. The test was modified from a standard 76-point screening test (Humphrey Field Analyzer; Zeiss, Oberkochen, Germany) and tests the same locations as the Glaucoma Hemifield Test. If fixation losses or false-negative or false-positive results exceeded 20%, testing was halted, and the participant was instructed again before retaking this first test. If the percentage of false-positive or false-negative results or fixation losses exceeded 33%, despite restarting, the first test was considered to be unreliable. A second suprathreshold test was performed 2 weeks later if the first test was unreliable in at least 1 eye or if the first test was reliable but showed VFL in at least 1 eye. Visual field loss according to the screening protocol was present if the participant did not respond to the light stimulus in at least 3 contiguous test points or in 4 contiguous points if the VFL included the blind spot. In case VFL was present on this second suprathreshold test in at least 1 eye or this second test was unreliable unreliable, Goldmann kinetic perimetry was performed within a month by an experienced perimetrist on both eyes according to a standard protocol.¹⁰ This perimetrist saw the results of the second suprathreshold test in all cases before performing Goldmann perimetry. If the second suprathreshold test or Goldmann perimetry was not performed, although available reliable suprathreshold field testing or perimetry data from the participant's own ophthalmologist indicated VFL, the presence of VFL was scored using available suprathreshold or previously made VFs. To minimize false-positive scoring and for the purpose of this study, suprathreshold VFL was considered only in the absence of threshold VF tests to be present if it comprised at least 6 contiguous test points. The quality of the Goldmann perimetry was estimated during follow-up using 50 Goldmann fields and 50 automated static threshold perimetry tests (Humphrey 24-II, Humphrey Field Analyzer; Zeiss) of 50 consecutive eyes with defects on suprathreshold testing. Field defects were detected by Goldmann perimetry with 94% true-positive and 84% true-negative rates, taking Humphrey 24-II perimetry as the gold standard.

Six researchers (3 senior ophthalmologists [S.P.-J., J.B.J., and P.T.V.M.dJ.], 2 residents [R.S.R. and R.C.W.W.], and 1 perimetrist [D.B.]) who had no information other than refractive error data independently graded all Goldmann fields. Each field was graded on the presence or absence of VFL and on the type of defect if present (Figure 1). If there was disagreement on the grading of a Goldmann VF, which occurred in 34% of all fields, consensus was reached. Long-term intragrader agreement, determined by regrading 40 field charts after 2 years by 2 graders (R.S.R. and R.C.W.W.), was 75% and 75%, while intergrader agreement was 80%. For all eyes with VFL on Goldmann perimetry, a senior ophthalmologist (P.T.V.M.dJ.) unfamiliar with the outcome of the Goldmann perimetry determined whether and where VFL was to be expected based on the presence and localization of fundus abnormalities using macular and optic disc–centered transparencies. Control transparencies from individuals without defects on Goldmann perimetry were mixed with other transparencies to prevent bias. If VFL did not correspond with a funduscopic lesion, the cause of VFL was determined using data from ophthalmologic examination, home interview, and neurologic assessment together with information obtained from medical records from general practitioners and ophthalmologists. Hemianopia, quadrantanopia, and isolated central defects were not regarded as glaucomatous VFL, whereas other types of defects were(Figure 1). Subsequently, glaucomatous VFL was defined as any other VFL in the absence of any other ophthalmologic or neurologic cause of VFL.¹¹ Severe VFL was defined as constriction of the V4 Goldmann isopter to less than 20° from the fixation point in at least 2 quadrants. Sometimes multiple related causes of VFL were present, eg, diabetes mellitus–related proliferative retinopathy requiring laser coagulation. In these instances, the initiating process was recorded as the primary cause (diabetes mellitus). Causes of VFL on suprathreshold VFs were determined using the same procedures. Criteria for open-angle glaucoma(OAG) have been specified previously, and, for the purpose of this study, definite and probable OAG based on VFL were pooled.¹¹

Low vision was defined as best-corrected visual acuity of less than 0.3 (20/60) in the better eye,¹² and age-related macular degeneration (AMD) was defined as geographic atrophy or neovascular macular degeneration according to international standards.¹³ Diseases and abnormalities primarily affecting the prechiasmal optic nerve, excluding OAG, were called optic nerve–related disease and included tilted disc, myelinated nerve fibers, secondary or angle-closure glaucoma, ischemic optic neuropathy, and Leber hereditary optic atrophy. Parapapillary atrophy causing a scotoma of at least 6 points on suprathreshold perimetry was also included in this disease group. Retinal vascular occlusive disease comprised central or branch arterial and venous occlusions. Because medical records from general practitioners and ophthalmologists often did not specify whether occlusions were arterial or venous, they were combined. A history of stroke was assessed using the question, "Did you ever have a stroke that was diagnosed by a physician?" If findings from neurologic assessment, consisting of testing of brachial and gastrocnemius tendon reflexes, testing of motor tone of the limbs, and testing for hemiparesis, were incompatible with the stroke history, general practitioner records were checked.

Disability in daily activities was measured in 8 components (dressing, rising, reaching, hygiene, eating, walking, gripping, and activity), as described previously.¹⁴ Moderate or severe disability was present when participants perceived difficulties in at least 4 of 8 components. In addition, participants were asked the following: "Are you homebound because of health problems?," "Do you use a walking aid?," "Do you fall more than once a month?," and "Have you ever visited an ophthalmologist because of trauma, diabetes mellitus, glaucoma, or otherwise?" At the second interview, on average 3 years after the baseline examination, participants were also asked the following questions: "Do you enjoy reading books or watching television?," "Did you fall more than 4 times in the past 2 years?," and "If you have fallen, were any bones fractured?" If participants had moved into a nursing home, they were also interviewed. Nonvertebral fractures occurring after baseline examinations, up to March 1, 1996, were reported by general practitioners in the study district and were verified by staff physicians using medical records and hospital discharge records. Mean follow-up was 3.8 years, and complete follow-up was available for 5186 (83%) of 6250 participants.

All ophthalmologically examined participants with VF testing and ophthalmoscopic data on at least 1 eye were included in the analyses. Characteristics of included and excluded individuals and those refusing the complete eye examination were compared using multiple logistic regression analysis for categorical variables, with adjustment for age and sex. For continuous variables, means and differences across the 3 groups were calculated using analysis of covariance, with adjustment for age and sex. Prevalences are reported as percentages and 95% confidence intervals. Associations of VFL with indicators of impairment in daily life measured at baseline and at the second visit were analyzed using logistic regression and were adjusted for age, sex, and the presence of moderate or severe disability as an indicator of general health status. Associations of VFL with the occurrence of fractures after baseline examination were analyzed using the Cox proportional hazards model, with adjustment for age, sex, and the presence of moderate or severe disability. The effect of visual acuity on the associations was determined by excluding persons with low vision and by adjusting for logMAR visual acuity. Differences were considered statistically significant at P≤.05.

Of 9161 eligible community-dwelling individuals, 7086 (77%) consented to a home interview. Participation (>80% among younger individuals) declined with age but was not sex dependent. Of the 7086 individuals, 6304 underwent ophthalmologic examination at the research center and 6250, 68% of the eligible population, were finally included (Figure 2). General characteristics of the included and excluded individuals and those who refused the eye examination are presented in Table 1. Compared with included individuals, those who declined examination were substantially older, were more often homebound because of health problems, and had visited the ophthalmologist more often because of diabetes mellitus but not glaucoma.

A total of 347 (6%) of 6250 individuals had VFL in at least 1 eye: 226 were identified using Goldmann perimetry and 121 were identified using suprathreshold perimetry only. The prevalence of VFL increased from 3% (77/2555) in those aged 55 to 64 years to 19% (28/146) in those 85 years and older. Bilateral VFL was present in 2% (109/6250) of the participants.

In 69% (238/347) of all cases, the cause of VFL was present in only 1 eye or was the same for both eyes. However, in 16% of all cases, 51% of all bilateral cases, fellow eyes differed in cause. Therefore, numbers of VF defects per eye and prevalences of various causes are better presented as percentages of eyes rather than percentages of participants (Table 2). Table 3 gives the causes of VFL by age category. At all ages, OAG was the leading cause of VFL. In participants aged 55 to 74 years, other optic nerve head diseases and stroke were the second and third most frequent causes, respectively. In participants 75 years and older, AMD and retinal vascular occlusive disease ranked second and third, respectively. Of 109 cases with bilateral VFL, 28% were brought about by OAG and 21% by AMD. The cause of VFL in 23 eyes could not be determined, either because medical records could not be traced or causes were ambiguously recorded. The VFL in these eyes should be called glaucomatous VFL according to previously defined criteria,¹¹ but in the cause-specific subdivision in Table 3, they are reported separately as unknown. Using macular and optic disc transparencies and ophthalmoscopic data, definite OAG, AMD, myopic macular changes, diabetic retinopathy, and retinal detachment could be excluded as causes. Concentric constriction of the VF to the central 10° diameter was present in 3 eyes of 3 participants: 2 eyes were affected by OAG and 1 by myopic degeneration. Two hundred seventy-six participants had a history of stroke, and only 4.7% of these had VFL.

Figure 3 shows the absolute prevalences of the 5 major causes of VFL by sex. Relatively fewer women had VFL after adjustment for age (odds ratio, 0.63; 95% confidence interval, 0.38-0.79). There were no significant sex differences in the prevalence of VFL due to AMD, retinal vascular occlusive disease, stroke, and optic nerve head disease. Glaucomatous loss, was less frequent in women (odds ratio, 0.52; 95% confidence interval, 0.38-0.81).

Of 347 participants with VFL, 97 (28%) reported that they had never visited an ophthalmologist except for the prescription of glasses. These participants were not distinguishable on the basis of age, sex, education, and income from those with VFL who reported having visited their ophthalmologist for ophthalmic disease, but they less often had low visual acuity (odds ratio, 0.22; 95% confidence interval, 0.07-0.65).

Associations between unilateral, bilateral, and severe VFL and indicators of impairment in daily life are given in Table 4. Because severe VFL in 1 or both eyes was present in only 33 individuals, those with severe unilateral (n = 28) and bilateral (n = 5) VFL were pooled; even then, these associations were not significant. Associations with indicators of impairment tended to be stronger for bilateral than for unilateral or absent VFL. Satisfaction experienced from reading books or watching television was lower in persons with bilateral VFL, even if visual acuity was 0.3 or greater. Their chance of experiencing frequent falls was 6 times higher than that of participants without VFL. Incidences of wrist (n = 66) or hip (n = 56) fractures, both most commonly associated with falling, were not higher in participants with bilateral or severe VFL. Yet, individuals with unilateral VFL more often experienced falls in the 2 years after the eye examination and wrist fractures during follow-up than those without VFL.

We present cause-specific prevalences of VFL in a general population of community-dwelling elderly individuals and demonstrate that the prevalence of VFL increased from 3% in those aged 55 to 64 years to 19% in those 85 years and older; OAG was the leading cause at all ages. Before age 75 years, other optic disc diseases and stroke were the second and third most common causes, respectively. After this age, AMD and retinal vascular occlusive disease ranked second and third, respectively.

Before these findings can be accepted, some methodological issues have to be addressed. A potential limitation of this study is its restriction to noninstitutionalized individuals. Too often, physical or mental disabilities of institutionalized individuals prevented reliable VF testing. Eye diseases associated with cardiovascular disease¹⁶ or bilateral blindness¹⁷ will be more frequent in these persons. From an earlier study² we calculated that the percentage of visually impaired people according to World Health Organization criteria was 0.8% in the independently living and 11.0% in the nursing home population. For blind persons, these percentages were 0.2% and 4.7%, respectively; as a cause for blindness, AMD was 7 times higher than primary OAG. Associations of VFL with measures of impairment in daily life will therefore be underestimated because of the exclusion of institutionalized individuals. However, restriction to noninstitutionalized individuals is necessary to estimate the impact of VFL on the ability to maintain independence in community-dwelling elderly persons. Furthermore, the general validity of our results is a reflection of the examination techniques and the process of determining the cause of the VFL. Small- or shallow-field scotomas may have remained undetected, as we required a minimum size for defects and used suprathreshold field testing for screening. Therefore, clinically more relevant defects were found. Also, our prevalence may be underestimated because we screened participants initially on the central 48° of the VF. It is unlikely that associations with disability will be substantially affected by this underestimation because central VFL is clearly more disabling in daily life than are defects in the far peripheral VF.

An advantage of this study is its large size, which allowed more accurate estimation of the relative contribution of different causes, in particular rare ones, to the prevalence. The number of participants 80 years and older(n = 535) was considerable compared with other studies.^3,8 The nonparticipation in this age group, however, may be another reason for underestimating the prevalence of VFL in the highest age groups. Also, we cannot exclude the possibility that nonparticipation rates for patients with OAG or AMD are different owing to differences in disability related to visual acuity loss. However, interview data on 193 (25%) of the 782 nonexamined eligible individuals indicated that consultation with an eye specialist was not more frequent, and neither was the self-reported presence of glaucoma. This suggests that cause-related bias was not substantial.

To our knowledge, 3 surveys^3,4,8 have addressed the frequency of VFL in elderly persons. A population-based study³ of patients 40 years and older identified VFL in 17% of all eyes. This 3-fold higher prevalence compared with our study can be explained by less strict criteria for size of defects, use of threshold testing, and the absence of confirmatory repeated testing. A 78-point field screening test with a single-intensity light stimulus in elderly drivers demonstrated that the prevalence of VFL in 55 to 59 year olds was 3% and increased to 13% in those older than 65 years.⁴ Another study⁸ using a single-intensity stimulus found that the number of points missed in the central 60° tripled with increasing age. Although all surveys found increasing prevalence with age, differences in measuring techniques and population sampling prevent reliable comparison.

Previous studies¹⁸ showed that elderly individuals progressively rely on visual feedback to maintain balance with increasing age because proprioceptive feedback, musculoskeletal strength, and often vestibular function decline with age. It is well known that low visual acuity is associated with a greater risk of falling, hip fractures, and mortality in the elderly population.^19,20 In contrast, only 1 study⁶ has focused on the association between VFL and falling: missing at least 5 points on suprathreshold testing doubled the chance of having fallen before. Because this research started as a cross-sectional study of community-dwelling elderly persons, we cannot exclude the possibility that persons with VFL who experience multiple falls were more likely to be nonparticipants at baseline because they resided in nursing homes. This is likely because 3.1% of all nonsyncopal falls in community-dwelling elderly persons 60 years and older lead to fractures.²¹ Seven percent of these are hip fractures that lead to institutionalization in 32% of patients or death in 16%.²² This nonparticipation bias may explain why individuals with bilateral VFL in our study less frequently had a history of hip fractures than those without VFL. This underestimation from history data underscores the importance of prospective data when associations with VFL are investigated and explains why we prospectively found a higher risk for falling than did a previous study.⁶ We found that unilateral and bilateral VFLs were associated with a 6-fold risk of frequent falling in the 2 years after the eye examination. This association remained after adjustments for visual acuity and disability(eg, walking problems). This makes it more likely that not only visual acuity but also VFL are causally related to fall risk. Paradoxically, the incidence of hip fractures was low in persons with VFL. It has been pointed out that fall characteristics, such as fall direction²³ and outdoor mobility,²⁴ may shift fracture location from the hip to other bones. This could explain our findings but remains speculative because the number of individuals with fractures was low.

In conclusion, this population-based study showed that unilateral VFL was present in 1 of 20 and bilateral VFL was present in 1 of 50 elderly persons. Regardless of visual acuity, VFL was associated with diminished enjoyment of reading and watching television, moderate restriction of daily activities, and, more important, fall accidents.

Accepted for publication August 1, 2001.

This study was supported by grants from the NESTOR Stimulation Program for Geriatric Research in the Netherlands (Ministries of Health and Education, Rijswijk); Topcon Europe BV, Capelle a/d IJssel (Dr de Jong); Landelijke Stichting voor Blinden en Slechtzienden, Utrecht (Dr de Jong); Netherlands Society for the Prevention of Blindness, Amsterdam (Dr de Jong); Optimix Foundation, Amsterdam(Dr de Jong); Stichting Fondsenwervingsacties Volksgezondheid, The Hague (Dr de Jong); Stichting Blindenpenning, Amsterdam (Dr de Jong); Stichting OOG, The Hague (Dr de Jong); Stichting Blindenhulp, The Hague (Dr de Jong); Rotterdamse Verenging voor Blindenbelangen, Rotterdam (Dr de Jong); and Stichting ROOS Rotterdam (Dr de Jong), the Netherlands.

Corresponding author and reprints: Paulus T. V. M. de Jong, MD, PhD, FRCOphth, the Netherlands Ophthalmic Research Institute, Meibergdreef 47, 1105 BA Amsterdam, the Netherlands (e-mail: p.dejong@ioi.knaw.nl).