The effect of different criteria on the number of patients blind from open-angle glaucoma

Open-angle glaucoma is the second most common cause of blindness globally [1]. It is important that we have knowledge about the prevalence of visual impairment caused by glaucoma in order to enable correct allocation of resources for glaucoma care, and also to allow evaluation of the effects of treatment and potential benefits of population screening.

Evaluation of the prevalence of blindness and visual impairment from glaucoma is influenced by the criteria used to define these entities, which differ between countries and regions. The definition of visual impairment stipulated by the World Health Organisation (WHO) can be considered the current gold standard in this context. However, many countries use their own criteria for blindness. An example of this is the definition used in the United States here designated the US criteria, which is also widely accepted and used in other countries. The number of people identified as blind from glaucoma should be higher when using the US definition than with the WHO definition, since the latter includes stricter criteria for blindness both by visual acuity (VA) and visual field loss.

In some cases, the prevalence of visual impairment (blindness and low vision) in glaucoma is studied and reported based on VA alone, i.e., not including visual field status [2‐21]. It seems obvious that this approach will yield falsely low estimates of the prevalence, because visual impairment will not be recognised in patients who have end-stage glaucomatous visual field loss with preserved central VA.

The US Social Security Administration (SSA) recently endorsed use of a new criterion for disability determinations, the perimetric mean deviation (MD). This ruling [22] states that an MD of -22 dB on a 30-2 Humphrey threshold visual field corresponds approximately to a constriction of the visual field to less than 20° of fixation, and recommends an MD of ≤ -22 dB as a visual field criterion to define disability.

The objective of the present study was to ascertain the extent to which different criteria of blindness and visual impairment influence estimates of the number of patients classified as blind or visually impaired by glaucoma. Our primary aim was to investigate how omitting visual field status from criteria affects the rate of visual impairment. We also wanted to compare the number of patients identified as blind from glaucoma by the US and the WHO glaucoma blindness definitions, respectively, and to estimate the number classified as disabled by the US SSA visual field criterion compared to the WHO criteria.

A total of 914 eligible patients were evaluated. Their mean age was 79 years (range 30-100 years). Nearly all the patients were Caucasians.

Table 2 presents the total numbers of patients identified as blind or visually impaired by the WHO and US criteria with and without taking visual field status into account. The number of blind patients was 30% higher according to the US criteria than the WHO criteria. Using the WHO criteria, the number of patients classified as blind was 40 when visual field was included but only 14 when visual field status was omitted, which represents a 65% reduction (Figure 2). Similarly, the WHO criteria identified visual impairment in only 37 patients when visual field status was ignored as compared to 58 patients when field status was included, a 36% reduction. Using the US criterion, the number of blind patients was only 24 when visual fields were omitted compared to 52 when field status was included, a reduction of 54%.

Eighty-five patients (9.3%) met the SSA criterion of bilateral visual impairment caused by glaucoma in both eyes with measured or assigned MD values ≤ -22 dB. Among those, 52 were visually impaired also by the WHO definition. No patients who were impaired according to the WHO criteria had MD values better than -22 dB.

Fifty-four patients had visual fields tested in one eye only, and 36 had no field tests at all. In 54 (60%) of those cases, the reason tests were not performed was that the patients already had very low VA and totally cupped discs at the initial visit, and in 28 cases it was because the patients' visual impairment was due to a cause other than glaucoma. For nine patients, perimetric results were not available for the study period or had been obtained only by testing on the Competer perimeter. Three patients were considered unable to do undergo perimetric testing.

The most important finding of the current study is that a very large proportion of the visually impaired glaucoma patients were not identified as being visually impaired when only VA criteria were used to define such low vision. This can be compared with an investigation performed by Hattenhauer et al [24], in which the risk of bilateral blindness was calculated for patients 20 years after a glaucoma diagnosis had been made using the US definition of blindness. Those authors found that the risk was 9% based on both visual field status and VA together but was 5% when based on VA alone, which is similar to our results. The visual field data in the study by Hattenhauer and colleagues came mainly from kinetic Goldmann perimetry. The Baltimore Eye Survey [25] using similar visual field testing, on the other hand suggested that the number of patients classified as having bilateral blindness would increase by at most 25% when combining visual field data and VA compared to if evaluation were based on VA alone. The visual field criteria for blindness and for visual impairment are quite restrictive. In glaucoma the remaining visual field is seldom circular, and many eyes are not classified as blind despite very small remaining field areas. A narrow segment of the field with test points with sensitivity values of 10 dB or better, that extends outside 10° from fixation may be enough for an eye not to fulfill any of the two blindness criteria (Figure 3). It is, therefore, obvious that impairment caused by glaucomatous field loss requires loss of a very considerable proportion of normal function and has large consequences if bilateral, and that field loss is a very important part of the visual burden caused by glaucoma.

Our results indicate that it is very likely that glaucoma impairment and blindness have been considerably underestimated in papers reporting rates of blindness determined using VA data, but omitting visual field status [3, 4, 15, 18]. Furthermore, glaucoma must have been substantially underestimated as a cause of blindness in numerous population studies which have relied solely on VA [2, 5‐14, 16, 17, 19‐21]. Relying on visual acuity alone results in under-estimation of impairment caused by other disease, e.g., in onchocerciasis [26].

The number of patients classified as blind from glaucoma differs between investigations using the US criteria and those applying the WHO criteria. To our knowledge only two studies including visual field status has used both the WHO and the US definition to evaluate glaucoma blindness [27, 28]. However, very few patients in the cited investigations were blind from glaucoma, and thus the results of that evaluation do not allow any conclusions to be drawn about the difference in the numbers of blind identified using these two approaches.

We found it interesting to study results obtained using the US SSA criteria for visual impairment. To our knowledge no other investigators have published visual impairment data based on MD values. It would no doubt be easier to use this MD-based criterion in surveys comparing visual impairment than performing actual measurements of visual field constriction, and that would also facilitate comparison of observations made in different studies.

This study had several strengths. Automated HFA perimetry is standard practice in our clinical setting, and other types of perimetry are rarely used. Ophthalmic practice in Sweden is somewhat unusual in that glaucoma care is delivered predominantly by the public health services. Our hospital provides primary glaucoma care for approximately three quarters of all patients with this disease in the catchment area. The current study was cross-sectional in design and included a rather large number of visually impaired patients. Therefore, our results should be reasonably representative of clinic-based populations. One may speculate on whether the proportions of individuals classified as blind from glaucoma using visual acuity alone or taking both visual acuity and fields into account are the same in clinic-based cohorts and in the population. If so the true prevalences of glaucoma blindness could be more than twice as large as those reported from many countries, and a "correction/multiplication factor" of 2.2 (US) or 2.9 (WHO) could be used to help provide at least a rough estimate of the total number of blind glaucoma patients in areas where such blindness has been estimated using visual acuity alone. Similar multiplication factors have been used by WHO to estimate low vision in areas where only blindness and not low vision have been assessed [29]. Reduced visual acuity seems to influence measurable reduction of quality of life earlier than field loss, and measurable utility may not be affected until end stage visual field status in the better eye [30, 31]. It is, therefore, even possible that the proportion of undiagnosed individuals with glaucoma in the population who are blind according to visual field criteria, but not because of poor visual acuity, are even more common in the population than in clinic-based cohorts.

A relative weakness of our investigation is that it was retrospective, and therefore some data were incomplete or missing. However, missing data constitute a problem in prospective studies as well. Visual field testing can also pose a problem in very old patients or in elderly with very low vision, who might have to be followed without such testing during their last few years of life. Also, considering prospective studies, some institutionalised patients are lost to follow-up, and some continue to come for examinations but no subjective measurements of visual function can be performed. The great majority of our patients were subjected to regular follow-up that included visual field testing, and there was little occurrence of incomplete or missing data.

In conclusion, we found that the number of patients with bilateral blindness from glaucoma was greatly underestimated when the evaluation was based solely on VA testing. The number of bilaterally blind was considerably higher when using the US definition of blindness compared to the WHO definition. The number of impaired patients was 60% higher by the US SSA impairment criterion than by the WHO criteria. Visual field testing is very important to achieve correct assessment of visual impairment caused by glaucoma.