Age- and sex-specific 5-year cumulative incidence of all visual field loss.
Frequency of causes of incident visual field loss in one or both eyes by age category. Numbers in parentheses are absolute numbers of persons with visual field defects in at least 1 eye. Relative rank may differ from Table 4 because frequency refers to individuals and not to eyes. The category of optic nerve head disease other than open-angle glaucoma also contains secondary glaucoma cases. Six cases had a combination of more than 1 of these 5 causes. The most plausible one was used here.
Age- and sex-specific 5-year cumulative incidence of glaucomatous visual field loss.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Skenduli-Bala E, de Voogd S, Wolfs RCW, et al. Causes of Incident Visual Field Loss in a General Elderly Population: The Rotterdam Study. Arch Ophthalmol. 2005;123(2):233–238. doi:10.1001/archopht.123.2.233
To determine the incidence and causes of visual field loss (VFL) in a general elderly population.
Central visual fields of both eyes were examined with suprathreshold perimetry in 3761 persons aged 55 years or older and free of VFL at baseline from the population-based Rotterdam Study. Goldmann perimetry was performed to confirm suprathreshold VFL. Causes of incident VFL (iVFL) were assessed based on all available ophthalmologic and neurological examination data and medical records.
After a mean follow-up time of 6.3 years, 175 persons developed VFL. The overall incidence rate of iVFL was 7.4 per 1000 person-years, increasing to 21.1 per 1000 person-years in those aged 80 years and older. Glaucoma was the leading cause of iVFL in all age categories. The overall incidence of glaucomatous VFL was 2.0 per 1000 person-years.
The incidence of all VFL increased 5-fold between 55 years and 80 years of age or older. After glaucoma, stroke was the second most common cause of iVFL in persons younger than 75 years, followed by age-related macular degeneration and retinal vascular occlusive disease.
The elderly population in the Western world is rapidly increasing. This applies also to the number of elderly who live independently. Besides low visual acuity, another limiting factor for independent living is visual field loss (VFL), which is found to be associated with impairment in daily functioning,1 with an emphasis on worse driving performance,2,3 frequent falls,4 and decreased quality of life.5 It is possible to prevent progression of VFL in some diseases by early detection of the related disease. Accurate estimates of incident VFL (iVFL) and its related causes are valuable to estimate the magnitude of the VFL problem, the likelihood of the underlying cause, and the optimal frequency of visual field (VF) examination either as screening or as part of an ophthalmologic examination, as well as for risk analyses of diseases leading to VFL, such as glaucoma. Although a few studies have provided prevalence estimates of VFL1,2,5,6 and more studies have focused on prevalence, incidence, and causes of visual impairment,7-13 no population-based study has yet reported on iVFL and its related causes.
The aim of this report from the Rotterdam Study is to provide estimates of the overall incidence of VFL and its related causes in a general, noninstitutionalized elderly population.
This study is part of the Rotterdam Study, a population-based, prospective cohort study of all residents aged 55 years and older in a district of the city of Rotterdam, the Netherlands. Details regarding the study design and identification of the eligible cohort have been described in previous reports.1,14 In brief, examinations of cardiovascular, locomotor, neurological, and ophthalmologic systems were conducted at the examination center after the medical ethics committee of Erasmus Medical Center had approved the study protocol and all participants had given written informed consent, all according to the Declaration of Helsinki.
After the baseline examinations in 1990-1993, a follow-up examination was performed in 1993-1994 in which we did not include perimetry because of the expected low number of iVFL, and a second follow-up was performed in 1997-1999 that did include perimetry. This study presents the results of the examinations at the second follow-up.
Of the initial eligible cohort of 10 275 individuals, 7983 persons (78%) agreed to participate in the Rotterdam Study. The ophthalmic part started after the pilot study, and therefore there were only 6780 participants in the ophthalmologic examination. At least 1 suprathreshold perimetry test was performed in 6347 of the 6396 noninstitutionalized persons at baseline. Baseline VF examination identified in this group 351 participants with prevalent VFL in at least 1 eye on Goldmann or suprathreshold testing, leaving a cohort of 5996 noninstitutionalized participants at risk for developing iVFL. This cohort provided the study population for this study.
The complete ophthalmologic examination and the protocol of VF examination have been described in detail in the baseline report.1 In brief, it included indirect ophthalmoscopy of the central and peripheral retina, stereophotography of the macular area (35° field), simultaneous stereophotography of the optic disc (20° field), and VF testing. The VF of each eye was separately examined using a 52-point suprathreshold test that covered the central 24° of the field, modified from a standard 76-point screening test (Humphrey Field Analyzer 640; Zeiss, Oberkochen, Germany). If the first test was unreliable or a reliable test showed VFL in at least 1 eye, a second suprathreshold test was performed on that eye on the same day. In some cases when the second test was not feasible on the same day either because the person was tired or because of a lack of time, this test was done during the participant’s second visit to the center, on average 3 weeks later. We considered VFL to be present if the participant did not respond to the light stimulus in at least 3 contiguous test points or in 4 contiguous test points when the blind spot was included in the defect. When VFL was still present on the second suprathreshold test or the test was unreliable again, Goldmann kinetic perimetry was performed on both eyes of these participants by 1 of 2 experienced full-time perimetrists, on average 3 months later. The perimetrist was requested to look only at the result of the last suprathreshold screening. The VF examination protocol and the equipment were the same at baseline and at follow-up. Care was taken that the same isopters tested at baseline were also tested at follow-up. Six researchers at baseline and 4 of them at follow-up independently graded all Goldmann VFs. In cases of disagreement, a consensus was reached among graders regarding the presence and type of defect.
For the assessment of causes of VFL, the same ophthalmologist (P.T.V.M.de J.) as the one at baseline plus another ophthalmologist (R.C.W.W.) independently determined whether and where VFL was to be expected based on the presence and localization of fundus signs. First they looked at macular and second at optic disc–centered stereoscopic transparencies, trying to predict if and where VFL would be found. To prevent bias, transparencies of both eyes of incident cases with VFL were examined, regardless of the presence or absence of VFL. The examiners were unaware of the outcome of the Goldmann perimetry at baseline or follow-up or the clinical data. Next, the examiners looked at the VF and checked if the location and cause they predicted agreed with the outcome of the perimetry. For cases in which fundus signs did not correspond with the VFL or no cause could be found, we used information from the ophthalmologic examination at the study center as well as ophthalmologic and neurological information from medical records from hospitals and general practitioners. All discordant outcomes were discussed between the 2 examiners to determine the definite cause after extra information, where possible, had been obtained. The same procedures were applied to determine the presence and causes of iVFL for those cases for which we only had results from suprathreshold perimetry.
Incident VFL was defined as the presence of a VF defect in at least 1 eye on Goldmann perimetry in a participant from the cohort at risk or the presence of a defect of at least 6 contiguous points on the last reliable suprathreshold perimetry performed at follow-up in those cases for which Goldmann perimetry was indicated but not performed. The flowchart for the classification of VF defects on Goldmann perimetry and the definitions of the type of defect used at follow-up were the same as at baseline.1,15 Glaucomatous VFL (GVFL) was defined as VFL compatible with glaucoma, thus excluding hemianopia, quadrantanopia, and an isolated central defect, after exclusion of all other ophthalmic or neuro-ophthalmic causes. Thus, optic disc characteristics were not taken into account when the diagnosis of GVFL was made. If the cause of iVFL was a set of multiple related disorders, the initiating process was taken as the cause. In some cases, iVFL was considered to be due to separate disorders. These cases were classified as having combined mechanisms of iVFL.
We used univariate analyses to compare baseline characteristics of participants and nonparticipants in the follow-up examination, adjusting for age and sex when appropriate. Sex differences in iVFL cases were analyzed using logistic regression modeling, adjusted for age and follow-up time.
The age-specific incidence rates (IRs) of iVFL were obtained per 5-year age category by dividing the number of incident cases by the number of person-years of the corresponding age category. The 95% confidence intervals (CIs) of the IRs were calculated using Poisson standard errors. The number of person-years was calculated by adding each individual’s contribution of follow-up time to the successive age categories. For those persons who had Goldmann perimetry, this date was used for determination of follow-up time. Similarly, the date of the suprathreshold screening was taken for persons without a Goldmann examination. For incident cases, we assumed that the VFL started at the midpoint of the follow-up time. Consequently, their contribution to the total person-years was half their individual follow-up time. The cumulative incidence (CumI) of VFL (risk of developing VFL) over a time period (t) of 5 years was derived from the corresponding IR using the following exponential formula16: CumI(t) = 1−e(−IR × t).
Incidence rates of GVFL, 95% CIs, and 5-year cumulative incidence for 3 age categories were calculated in the same way as the overall incidence of VFL. Overall, age-specific IRs for all-cause VFL and GVFL were given per 1000 person-years. The data on the type of defects and causes of VFL were given as number of eyes and percentage of total number of eyes. We used SPSS for Windows, version 11 (SPSS Inc, Chicago, Ill) for all statistical analyses.
From the study population of 5996 participants, during a mean ± SD follow-up time of 6.3 ± 0.76 years, 858 participants (14.3%) died and 1337 (22.3%) refused or were unable to participate in the 1997-1999 follow-up examination. Of 3801 persons (63.4%) who participated in the ophthalmologic examination at follow-up, 40 (1.1%) did not have suprathreshold screening and were unable to undergo any VF examination mainly because of their mental or physical disabilities, and 3761 (98.9%) had perimetry performed on at least 1 eye. General baseline characteristics of participants and nonparticipants at follow-up examination are presented in Table 1. Compared with the participants, nonparticipants were more often female, on average older, and had a higher prevalence of stroke, dementia, and visual acuity lower than 20/40. Of 3761 participants who had the first suprathreshold screening in at least 1 eye, 538 (14.3%) needed further perimetry because of either abnormal test results or unreliable perimetry. Of these, 423 (11.2%) took the second suprathreshold test, and 60 (1.6%) had Goldmann perimetry without a second suprathreshold screening mainly because this test was judged to be too difficult for them. After the second suprathreshold test, 166 participants (4.4%) needed confirmatory perimetry, and 137 (3.6%) had Goldmann examination. There were 55 persons (1.5%) who refused further perimetry after the first suprathreshold screening, and 29 (0.8%) refused Goldmann perimetry after they had 2 suprathreshold tests.
After 23 786 person-years of follow-up, 175 individuals (240 eyes) were identified as having iVFL, of which 140 persons were detected on Goldmann perimetry; 35 persons had iVFL on suprathreshold testing but refused further perimetry. Bilateral iVFL was present in 65 participants (1.7% of our study population), of whom 36 (55.4%) were in the age category 65 to 74 years. The overall IR of iVFL in at least 1 eye for persons aged 55 years and older was 7.4 per 1000 person-years, increasing from 3.7 per 1000 person-years at age 55 to 59 years, to 21.1 per 1000 person-years at age 80 and older. Table 2 also presents age-specific IRs, the corresponding 95% CIs, and the 5-year cumulative incidence of developing VFL for 5-year age categories. Figure 1 shows the 5-year cumulative incidence according to age and sex. Logistic regression showed a statistically significant difference in the IR of VFL between men and women when adjusted for age. Men overall were at a higher risk for developing iVFL compared with women (odds ratio, 1.5; 95% CI, 1.1-2.0). This was consistent for most ages.
For the cases with iVFL, we assessed the type of defect and causes of VFL per eye. Table 3 gives the type of defect(s) per eye in 140 cases, including 192 eyes with at least 1 defect as detected by Goldmann perimetry. The most frequent defects were paracentral scotomas, present in 32.3% of eyes with 1 defect and 39.7% of eyes with 2 defects, and arcuate scotomas, present in 19.4% of eyes with 1 defect and 32.8% of eyes with 2 defects.
Causes of iVFL per age category and per eye are shown in Table 4. Open-angle glaucoma (OAG) was the overall leading cause and was present in 58 eyes (24.2%), followed by stroke (n = 33; 13.8%) and age-related macular degeneration (AMD) (n = 28; 11.7%). Figure 2 presents the 5 most frequent causes of iVFL by age category. Among 65 cases with bilateral iVFL, 52 persons (80.0%) had the same cause of iVFL for both eyes. Stroke was the leading cause of bilateral iVFL and was present in 15 participants (28.8%), followed by OAG (n = 11; 21.2%) and AMD (n = 9; 17.3%). Other causes of bilateral iVFL comprised diabetic retinopathy (n = 4), myopia (n = 3), secondary glaucoma (n = 3), ocular trauma (n = 1), tilted disc (n = 1), cataract (n = 1), and unspecified retinal disorder (n = 4).
We also estimated the incidence of GVFL (Table 5). The overall IR for participants 55 years and older was 2.0 per 1000 person-years. It increased significantly from 0.9 per 1000 person-years in the age category 55 to 64 years, to 4.0 per 1000 person-years in the age category 75 years and older. Figure 3 shows the 5-year cumulative incidence of GVFL according to age and sex. The overall IR of GVFL seemed to be higher for men when adjusted for age (odds ratio, 1.7; 95% CI, 0.9-3.0).
Our findings demonstrate that the overall incidence of VFL for persons aged 55 years and older increases significantly with age and is higher for men than for women. The leading cause in all age groups was OAG followed by stroke and AMD. The estimated overall IR of incident GVFL for persons aged 55 years and older was 2.0 per 1000 person-years, and the risk of developing GVFL in the next 5 years increased 5-fold for persons aged 75 years and older compared with those aged 55 years. To the best of our knowledge, this is the first study providing data on the overall incidence of VFL and its causes in a general, noninstitutionalized elderly population. The relatively large study cohort ensures reliable incidence estimates for GVFL.
Before we accept our findings, we should consider the limitations and strengths of our study and analysis. Our findings relate to the Rotterdam Study cohort of participants aged 55 years and older, restricted only to noninstitutionalized persons. No perimetry was performed at baseline or follow-up on institutionalized participants because at the start of the baseline examination they frequently had unreliable perimetry, owing to physical and mental disabilities.1 The fact that institutionalized persons are in general older and have a higher rate of eye disorders related to cardiovascular disease than independently living individuals18 probably led to an underestimation of iVFL in this population. Furthermore, selection bias due to the large amount of nonparticipants either refusing or unable to participate (22.3%), who were on average older and more often had a history of stroke, might further underestimate actual iVFL.
The large number of deaths among participants at risk for iVFL during the follow-up time is due to the fact that our study cohort is composed of elderly persons. According to our findings for cause-specific iVFL, selective survival in both men and women can be an explanation for the fewer cases of iVFL related to stroke and retinal vascular occlusive disease in the age group 75 years and older. However, this was not the case for GVFL. We recently showed that there is no selective survival related to glaucoma in a general elderly population.19 Therefore, incidence estimates of GVFL seem less biased.
A major advantage of our study is that the screening perimetry was performed on both eyes of all participants, without previous selection. Also, most persons had 2 suprathreshold tests before Goldmann perimetry was performed. This means that most individuals had at least 4 perimetric examinations (1 at baseline and 3 at follow-up) to determine if they had iVFL. This is important because after each repeated test fewer participants retained their defect. This could be explained by a possible learning curve for the static perimetry or better comprehension of kinetic perimetry. The use of Goldmann perimetry as a confirmatory test could be considered as either a limitation or a strength. Goldmann kinetic perimetry was the gold standard when this study was designed, but recently it has been replaced with automated static perimetry. However, kinetic perimetry has more reliable results when performed on elderly people, especially those with neuro-ophthalmic disorders, because automated static perimetry can be more tiring and may lead to fixation loss. Another strength of our study is that our findings are based on a large general population cohort that was observed on average for 6.3 years. The same eye examination protocols were used at baseline and follow-up, which provided better monitoring of iVFL and its related causes and more accurate incidence estimates.
In the baseline report, we estimated the prevalence and causes of VFL. Comparing the frequency of causes of VFL at baseline with those at the second follow-up, we found a slightly different distribution of causes. The leading cause in all age categories was still OAG; stroke became an important cause at follow-up, overtaking AMD and optic nerve head diseases other than OAG, which were more frequent at baseline. The higher frequency of paracentral and arcuate scotomas among cases with iVFL corresponded to OAG as the leading cause of iVFL. Our cumulative incidence of incident GVFL was lower than the 4-year risk of incident OAG in the Barbados Study.20 This can be explained by racial influences and differences in procedures and diagnostic criteria.
Prevalence studies on VFL differed in their estimates of the overall and sex-specific prevalence, mainly owing to differences in either examination techniques or population sampling, but were consistent in showing the increase in prevalent VFL with an increase in age.1,2,5,6 The results of our study on iVFL confirm this age trend, but the paucity of other similar incidence studies prevents us from generalizing our finding that men have a higher risk of developing iVFL compared with women.
As to the implication of our findings, the estimates of the 5-year risk of developing iVFL and incident GVFL for persons aged 55 years and older in different age categories can be easily applied in everyday clinical practice, although this of course becomes less reliable when the numbers become smaller.
In conclusion, our findings from this population-based study demonstrate that the incidence of all VFL increased 5-fold between 55 and 80 years of age or older. Open-angle glaucoma was the leading cause in all age categories. The 5-year risk of developing incident GVFL increased 5-fold for elderly people aged 75 years and older compared with those 55 years of age. Stroke was the second most common cause of iVFL before age 75 years, followed by AMD and retinal vascular occlusive disease.
Correspondence: Paulus T. V. M. de Jong, Netherlands Ophthalmic Research Institute, KNAW, Meibergdreef 47, 1105 BA Amsterdam, the Netherlands (firstname.lastname@example.org).
Submitted for Publication: September 26, 2003; final revision received August 11, 2004; accepted August 11, 2004.
Financial Disclosure: None.
Funding/Support: This study was supported by grant 2200.0035 from the Netherlands Organization for Scientific Research (NWO), The Hague; Foundation Fondsenwerving Volksgezondheid, The Hague; Optimix Foundation, Amsterdam; the National Foundation for the Blind and Visually Impaired (LSBS), Utrecht; Topcon Europe BV, Capelle aan de IJssel; and Merck Sharp and Dohme, Haarlem, the Netherlands.
Additional Information: Drs Skenduli-Bala and de Voogd contributed equally to this article.
Create a personal account or sign in to: