Occasional Paper Number 15 September 2008 Improving the detection of correctable low vision in older people This publication summarises findings from research funded by Thomas Pocklington Trust and carried out at the Institute of Optometry by Zahra Jessa and Prof Bruce Evans in collaboration with Prof David Thomson, Prof Gillian Rowlands, Mitesh Amin, Hannah Sawyerr and John Cooper. Summary The research sought to investigate new vision screening instruments as methods of improving the take-up of primary eyecare services by older people in South London. In Screening Study 1, a computerised vision screener (CVS1) was used to test 180 older people (aged 65+) in South London. All participants also received a ‘gold standard’ eye examination, which found that 32% of participants had significant cataract and 39% needed new spectacles. Overall, 58% of participants had at least one of these problems and the computerised vision screener detected about 80% of these cases. This first study determined the usefulness of each subtest in the screener and two tests were subsequently dropped from the screener. In Screening Study 2 the revised computerised vision screener (CVS2) and a rapid flipchart vision screener (FVS) were used to test a new sample of 200 older people. The gold 1 standard eye examination found that 31% had cataracts and 30% correctable refractive errors; 51% had at least one of these conditions. Both screening tools detected about 80% of cases of visual loss due to these problems or to age-related macular degeneration (AMD). The tools also correctly classified about 70% of those with normal visual function. Further analyses revealed the best combination of tests in the screeners and showed, as expected, that simple vision screening is not good at detecting glaucoma, which is known to be difficult to diagnose. Participants reported a significant improvement in quality of life, most marked in those who were prescribed new spectacles. Vision screening does not replace the need for professional eyecare, but can help to identify older people with correctable visual problems. All people who are screened should be reminded of the need for regular eyecare. Introduction A recent systematic review reveals that 20-50% of older people 1 have undetected reduced vision. The majority of these people have correctable visual problems (refractive errors or cataract). It is particularly startling that, in a ‘developed’ country, between 7% and 34% of older people have visual impairments that could be cured simply by appropriate spectacles. This has been attributed, at least in part, to inadequate levels of attendance at the community optometrist. This undetected, yet correctable, reduced vision is associated with impaired quality of life and ability to carry out activities of daily living, depression, and falls and other accidents. Those with low vision are about twice as likely to have falls than fully sighted people. This correctable reduced vision is likely to be particularly prevalent amongst people who suffer from the effects of poverty and/or are from ethnic minorities. Two different (but not mutually exclusive) approaches to improving the detection of visual problems in older people are better to publicise the need for regular optometric eye 2 examinations and to screen for visual problems. If an older person attended a vision screening programme and was told personally that their visual problem had a high likelihood of being treatable, or at the least might benefit from low vision services, then the research team believes that they would be very likely to follow up on the referral that would result from the screening. It is believed that the likelihood of their seeking care is much greater than a person who has only seen publicity advertising NHS sight tests. In view of this, it is surprising that there has been relatively little research on vision screening in older people. As part of this project, the research team carried out a detailed and systematic literature review of research on vision screening in older people. This literature review was published in the peer-reviewed journal of the College of Optometrists, Ophthalmic & Physiological 2 Optics. Only the main findings will be summarised here. There is a lack of rigorous research investigating whether vision screening is effective at detecting correctable low vision in older people. Visual function in older people is not fully described by high contrast visual acuity (e.g., the conventional letter chart test), or by self-reports of visual difficulties. Other tests that may be relevant include visual field testing, low contrast visual acuity, contrast sensitivity, and stereoacuity. 1 Evans BJW, Rowlands G (2004) Review article. Correctable visual impairment in older people: a major unmet need. Ophthal.Physiol.Opt. 2004;24:161-80. 2 Jessa Z, Evans B, Thomson D, Rowlands G (2007) Vision screening of older people. Ophthal. Physiol. Opt. 2007;27:52746. 3 The review paper noted that there is still uncertainty over the battery of vision tests that are most appropriate. This, and optimum venues for screening, requires further research before it can be determined conclusively whether vision screening of older people is worthwhile. The review noted that if a vision screening programme using a battery of vision tests, perhaps computerised, can be established then this should be tested to determine how well the screening can detect the target conditions. The literature suggests that quite basic tests (described below) will be able to detect uncorrected refractive errors and cataract. These are the main target conditions in this research because: (a) they have a very high prevalence, (b) they can readily be cured, and (c) their treatment is of direct and immediate benefit to the public through correcting visual loss and improving quality of life. The computerised vision screener and flipchart screener that were developed for this research also include a test of visual acuity, which will detect visual impairment from another common cause in older people: age-related macular degeneration (AMD). Although this condition is not readily correctable in most cases, it is helpful to detect cases so that they can be referred when appropriate for ophthalmological investigation and for further support and low vision aids (e.g., magnifiers). There have also been recent advances in the treatment of some cases of AMD and it is likely that further technological developments will lead to a higher proportion of cases being treated. In view of these developments, it was decided to evaluate the performance of the screening tools at detecting significant AMD and the cases which were most at risk of deteriorating to the sight-threatening wet AMD. The only common visual anomaly that it was anticipated would be difficult to detect with vision screening is glaucoma. Although visual loss from glaucoma cannot be treated, it can be prevented through early detection. It is difficult to screen for glaucoma, since all three commonly used glaucoma tests have a poor ability to diagnose glaucoma when taken in isolation and using all three tests in screening by non-healthcare professionals is impractical. 4 A visual field test, which is likely to be of some use for detecting glaucoma, was included in the computerised vision screener. Nonetheless, it was accepted from the outset that such a test is unlikely to match the accuracy of a full eye examination for detecting glaucoma. This research followed the tenets of the Helsinki declaration for research involving human subjects and conformed to the Department of Health’s Research Governance Framework. The research received approval from appropriate ethical and NHS committees. All participants were given full information about the research, both verbally and in writing, and it was explained that participation was optional and that refusal to participate would not in any way influence their continued medical, optometric or social care. Particular care was taken not to alarm or confuse older people. Preliminary study on the provision of NHS eyecare & eyewear in South London Introduction All people over the age of 60 are entitled to eye examinations in the community funded by the NHS, and people on low incomes or with unusually high optical prescriptions are entitled to an NHS Optical Voucher. This is a voucher that can be used to pay for or reduce the cost of spectacles. This preliminary study was carried out to determine which optical practices in the South London area have a reasonable range of spectacles whose cost is fully covered by the NHS Optical Voucher (referred to below as Voucher Value Spectacles: VVS). Methods A questionnaire was developed for sending to optical practices. This was designed to determine the availability of NHS-funded eyecare and eyewear. 5 Results In South London, 65 questionnaires were sent out and there were 53 responses (response rate 82%). Nearly a third of the practices that responded to the survey do not provide VVS. It has been suggested that optometric practices are deterred from providing VVS because the values of the vouchers are uneconomic. Compelling practices to provide VVS could be counter-productive if the values of the vouchers remain uneconomic, because it could force practices to withdraw from providing NHS services altogether. It seems likely that the limited availability of spectacles fully funded by the NHS is one of the reasons why so many older people in the UK have poor vision simply through lack of appropriate spectacles. The list of practices where VVS spectacles are available was printed and issued in Screening Study 1 and Screening Study 2 to participants who were eligible for VVS. Screening Study 1 Introduction The main goal of Screening Study 1 was to determine the most appropriate screening tests for inclusion in the final version computerised vision screener and flipchart screener. Some basic epidemiological data were also obtained. Most participants were seen at the Institute of Optometry but some community based venues were also used. The Institute of Optometry is situated in Lambeth, which has the eighth highest ethnic minority representation nationally. Over half of the local population is part of an ethnic minority group. The required number of participants was calculated using sample size calculations. The target sample size in Screening Study 1 was 180. 6 Methods Participants were people aged 65 or older who consented to participate in the study. People of all ethnic backgrounds and with or without physical or mental disabilities were welcomed. Computerised vision screener (CVS1) The CVS1 was a software program that ran on a laptop computer, the display of which had been calibrated. The following seven tests were included: symptoms & history near visual acuity (binocular) visual field test fixation disparity (a test of eye coordination) stereoacuity (depth perception) high contrast visual acuity low contrast visual acuity. In addition to the CVS1, all participants underwent a full ‘gold standard’ eye examination by an optometrist. The optometrist and the lay person carrying out the screening were unaware of each other’s results. The gold standard eye examination included the following tests: detailed symptoms & history checking the prescription in the current glasses determination of refractive error by both objective and subjective tests six tests of binocular coordination high and low contrast distance visual acuity and near visual acuity visual fields four tests of ocular health (tonometry, pupil reactions, anterior segment biomicroscopy, dilated fundoscopy). 7 Based on the results of the gold standard eye examination, standardised methods were used to diagnose undercorrected refractive error (an improvement of two lines of a letter chart with new spectacles), cataracts (the Lens Opacities Classification System III), AMD (Clinical Age Related Maculopathy Staging System). Participants with or at risk of glaucoma were identified according to conventional clinical criteria. The performance of the screening test was assessed by determining the sensitivity (the ability of a test to identify correctly people who have the target conditions) and specificity (the ability of a test to identify correctly people who do not have the target conditions). The optimum test cut-off values (pass/fail criteria) were determined using a standard method, involving plotting graphs that show the trade-off between sensitivity and specificity (receiver-operator characteristic (ROC) curves). Results The average age of the 180 participants was 77 years (range 67-99 years), 46% were male and 12% were seen in the community. Overall, 58.3% had correctable visual loss (31.7% cataract, 39.4% correctable refractive error; some both) and 28.9% had significant AMD. All participants were able to carry out the CVS1 and none complained that it was difficult to do. Extensive analyses were carried out to determine the value of each component subtest of the CVS1 for detecting undercorrected refractive error, cataract, AMD, glaucoma, and a history of falls. Two of the tests were found to be of little value: the fixation disparity and the stereoacuity tests. These were dropped from the screener for the second study. The most valuable test was high contrast visual acuity, which detected nearly 80% of cases of under-corrected refractive error or cataract and correctly classified nearly 60% of participants who were visually normal. 8 A supplementary study was included in Screening Study 1 to evaluate whether cataracts could be reliably graded using a portable version of the table top slit lamp biomicroscope that conventionally is used to grade cataracts. This study showed that cataracts can be reliably graded in this way, which allowed Screening Study 2 to be based in the community to a greater extent. Screening Study 2 Introduction Screening Study 2 evaluated the revised computerised vision screener (CVS2) and a flipchart vision screener (FVS) on a new group of 200 people aged 65+. Many of the participants were seen in the community, either at a community centre or a GP practice. Methods CVS2 was similar to CVS1, except that the fixation disparity and stereoacuity tests were not included. The FVS included three tests, presenting binocular near visual acuity, and presenting monocular distance high contrast and low contrast visual acuities. The gold standard eye examination was the same as in Screening Study 1. 9 Since Screening Study 2 investigated the ‘final versions’ of the screening tools, greater emphasis was placed on follow3 up. The Quality of Life (QoL) questionnaire, a brief 4 standardised test (one side of A4 paper, taking approximately 5 minutes) was applied at the time of the eye examination and up to three months after any intervention. Results As in Screening Study 1, ROC curves were used extensively to determine the best compromise between sensitivity and specificity. An additional variable was calculated: the positive predictive value (PPV). This is the probability that the condition is present when the test is positive (failed). The average age of the 200 participants in Screening Study 2 was 77 years (range 65-94 years); 31% were male and 32% were seen in the community. Overall, 51% had correctable visual loss (30.7% cataract, 30% correctable refractive error; some both) and 22.5% had significant macular degeneration. All participants were able to carry out the CVS2 and none complained that it was difficult to do. This screening tool was found to be appropriate for use in community settings. Some of the staff at a community venue were trained in the use of the CVS2 and they found it easy to use. 3 Wolffsohn JS, Cochrane AL (2000) Design of the low vision quality of life questionnaire (LVQOL) and measuring the outcome of low vision rehabilitation. Am. J. Ophthalmol. 2000;130:793-802. 4 de Boer MR, Moll AC, de Vet HC, Terwee CB, VolkerDieben HJ, van Rens GH (2004) Psychometric properties of vision-related quality of life questionnaires: a systematic review. Ophthal. Physiol. Opt. 2004;24:257-73. 10 Revised Computerised Vision Screener (CVS2) Monocular data were used to determine the best cut-off (pass/fail point) for each test for detecting the target conditions. These results agreed with those of Study 1. To investigate the overall performance of the CVS2, significant eye disorder (excluding glaucoma) was defined as the presence, in one or both eyes, of significant uncorrected refractive error, cataract, or AMD. High contrast visual acuity gives a sensitivity, specificity, and PPV of 75%, 73%, and 83%. The equivalent figures for low contrast visual acuity are 71%, 80%, and 86%. The next stage in the statistical analysis was to combine tests from the screener. The challenge when combining tests is as follows. If a person is said to ‘fail’ the screening if they fail any one of a number of subtests in the screener, then a great many people will fail. The sensitivity will be high (most people with a visual disorder will fail) but the specificity will be low (many people without a visual disorder will also fail). To achieve good sensitivity without also having poor specificity requires the judicious selection of the tests that should be used, and the results from 37 different test combinations were analysed to determine this. For example, if people are said to fail the CVS2 if they fail at least one of high contrast acuity, low contrast acuity, symptoms, or near acuity then the test fails 82% of people with significant eye disorders (defined above), but only passes 60% of people who do not have these disorders. If people are failed solely on the basis of high contrast or near visual acuity, then CVS2 correctly identifies 79.5% of people with the disorders and correctly passes 68% of those who are normal. 11 The last stage of the analysis of the test results from Screening Study 2 involved the inclusion of glaucoma in the diseases to be detected. The visual field test was included specifically to provide some ability to detect glaucoma, and this test did detect 15 of the 19 participants with or at risk of glaucoma. If glaucoma was added to the ‘significant eye diseases’ defined above, then the combination of high contrast acuity and visual field test would detect 88% of participants with these target conditions. However, the inclusion of this criterion reduces the overall test specificity to about 25%. This means that 75% of visually normal people would fail the screener. Flipchart Vision Screener (FVS) All participants in Screening Study 2 could be tested with the FVS and none complained that it was difficult to do. This screening tool was found to be appropriate for use in community settings. Some of the staff at a community venue were trained in the use of the FVS and they found it easy to use. The FVS results were analysed in the same way as for the CVS2 and the data were comparable to those for the CVS2. The results show that if one test alone had to be chosen from the rapid flipchart tool, it ought to be low contrast VA. Quality of Life (QoL) A response rate of 97% was obtained for the QoL follow-up. In no case did the QoL decrease following an eye examination and, across all participants, there was a significant improvement in QoL (Wilcoxon signed ranks test, Z=10.5, p<0.001). In participants who did not receive an intervention, the improvement in QoL appeared to be attributable to an improvement in the response to the question ‘How well has your eye condition been explained to you?’ 12 The largest QoL gain was in participants who were prescribed new spectacles. In this group, the degree of gain in QoL was proportional to the degree of improvement in visual acuity with new spectacles. Discussion Screening Study 2 was more community based than Screening Study 1 and found a similar high prevalence of correctable visual problems. The simple distance visual acuity test in CVS2 detected 75% of cases of uncorrected refractive error or cataract or AMD. By combining tests in the screener a sensitivity of over 80% can be achieved, although the specificity drops. As anticipated, the detection of glaucoma is more problematic. The screener was able to detect 15 out of 19 patients with or at risk of glaucoma but this criterion reduced the overall specificity of the CVS2 to about 25%, meaning that 75% of visually normal people would have failed the test. The FVS performed similarly to the CVS2. An interesting exception relates to high and low contrast distance visual acuity testing. If it was asked which subtest by itself would be most useful, then with the CVS2 this would be high contrast visual acuity but with FVS this would be low contrast visual acuity. It is tempting to attribute this to properties of the computer monitor, but this had been carefully calibrated. The research team feel that this is most likely to be a chance finding. Screening Study 2 demonstrated that QoL can be improved by refractive correction, and this is in agreement with the literature. It is believed that this study is the first to demonstrate a correlation between the degree of acuity gain and the improvement in QoL. Although it was possible to evaluate the effect of spectacle intervention on QoL, this was not possible for cataract surgery. This is because of hospital waiting times and because very few ophthalmologists provide feedback to the referring optometrist. 13 The results also indicate that there was an increase in the QoL even in those patients who did not require intervention. It is thought that the reason for this may simply be that the practitioner took the time to answer questions and reassure patients about their ocular health. Although this improvement in QoL was smaller than in those who had a spectacle intervention, this important benefit should not be underestimated. Older people are often concerned about their eyesight and even when it cannot be improved then information from a caring eyecare practitioner is still helpful. Summary of results The preliminary study revealed that nearly a third of optical practices in South London do not provide any spectacles whose cost is fully covered by the NHS Optical Voucher. This indicates that older people with low incomes may find it difficult to obtain spectacles that they can afford. This finding is most likely explained by the fact that the NHS sight test fee is uneconomic, covering less than half the cost of providing a modern eye examination. This low fee is subsidised by spectacles, and yet the NHS Optical Voucher value is also uneconomic. In other words, the current system of funding community eyecare in England requires spectacle wearers to subsidise the uneconomic cost of the NHS sight test. People with low incomes are unable to make this subsidy comfortably, and it seems likely that many therefore avoid seeking eyecare. The solution to this problem is not to force practices to provide a range of VVS, because if they are compelled to participate in uneconomic activities practices are likely to withdraw from NHS services. A better option would be to fund NHS eyecare in England properly, mirroring recent developments in Scotland and Wales. Screening Study 1 demonstrated that the CVS1 has reasonable sensitivity and specificity for detecting the target conditions 14 (cataract and under-corrected refractive error) and is also effective at detecting AMD. This study showed that the fixation disparity and stereoacuity tests in CVS1 did not provide very useful data and could be dropped in CVS2. The supplementary study on cataract grading showed that a portable slit lamp biomicroscope can be used to grade cataracts using the LOCSIII system. This allowed Screening Study 2 to be based in the community to a greater extent. Considering the results from both screening studies, the gold standard eye examinations revealed a high prevalence (over 50%) of the target conditions in a South London population of people over the age of 65 years. About 30% have significant cataract, 30-40% under-corrected refractive error, and 20-30% significant AMD. These findings are particularly noteworthy since the research was carried out in an area of London where poverty is prevalent and where over half of the population belongs to an ethnic minority group. Most of these cases of correctable visual problems were detected by the screening instruments. In Screening Study 1, high contrast visual acuity alone detected about 80% of cases of correctable visual loss. This sensitivity was good, but the specificity of just 60% meant that the test would have ‘failed’ about 40% of people who were in fact visually normal. The sensitivity and specificity were explored further in Screening Study 2 where, using the final version of the computerised vision screener (CVS2), a sensitivity of about 80% could be achieved for detecting cataract, correctable refractive error, and significant AMD. A specificity of about 70% was possible and the best test combination was obtained by selecting participants who failed the high contrast visual acuity or near visual acuity tests. 15 The FVS was found to be very simple to use and performed similarly to CVS2. For detecting cataract, correctable refractive error, and significant AMD a sensitivity of about 80% could be achieved with a specificity approaching 70%. These values were obtained by selecting people who failed either the high contrast or low contrast distance visual acuity tests, and 79% of people who failed one of these tests had one of these conditions. Even in a gold standard eye examination where three or more glaucoma tests are carried out, the detection of glaucoma is challenging. It was anticipated from the outset that the screening tools would not have a high sensitivity or specificity for detecting glaucoma and this was found to be the case. The visual field test on CVS2 detected 15 of 19 cases who had or were at risk of glaucoma, but this greatly reduced the specificity of the screener – causing it to ‘fail’ about three-quarters of those who were visually normal. The data from both screening instruments partly support the hypothesis raised in the introduction that vision screening should include more tests of visual function in addition to high contrast visual acuity. However, high contrast visual acuity was found to be one of the best predictors of the presence of the target conditions, although it was also found to be useful to include other tests (e.g. near visual acuity, low contrast acuity, visual fields). The QoL data show that a consultation with a caring eyecare practitioner brings about an improvement in QoL even when no intervention is required, simply because the older person has received an explanation of their visual status. When a new refractive correction is required this is associated with a marked improvement in QoL. The greater the improvement in vision from the refractive correction the more substantial is the improvement in the QoL. 16 This research supports previous findings that there is a high prevalence of correctable visual loss in older people. The notion that older people with visual problems will fully engage in eyecare services is clearly nothing more than an ideal and this underlines the need for methods that will encourage older people to seek regular eyecare. The screening instruments that have been investigated here are reasonably efficient at detecting people with visual problems and it was found that when participants were told in person that they would be likely to benefit from an intervention then most were keen to follow this advice. One important caveat needs to be stressed. Any type of screening can be dangerous if it has less than 100% sensitivity and if participants are so reassured by the screening that they fail to participate in healthcare services. It is very important that people who undergo vision screening are given verbal and written information stressing that a ‘pass’ in the vision screening test does not diminish the need for regular professional eyecare. Both the computerised and flipchart screening instruments were found to be easy to use and appropriate for use by lay people in community settings. Participants did not find the testing difficult or unpleasant. The time taken to test a participant was about 810 minutes for CVS2 and 5 minutes for FVS. The research team’s experience is that, although GPs were generally supportive of the research and pleased to publicise the studies, most GP practices are not likely to be suitable as venues for vision screening as set up in this study. Surgeries are very busy and generally do not have unused areas suitable for vision testing. The two community venues where the screening instruments were found to be most appropriate were an intermediary care centre and a community centre. The FVS is easy to use, highly portable, and inexpensive to produce. One possibility might be for a community nurse to carry it so that older people could be tested at home, for 17 example if they were seen by a community nurse when recovering from a fall. The time taken to test a participant was about 8-10 minutes for CVS2 and 5 minutes for FVS. Authors Professor Bruce Evans and Miss Zahra Jessa Institute of Optometry 56-62 Newington Causeway London SE1 6DS References de Boer MR, Moll AC, de Vet HC, Terwee CB, Volker-Dieben HJ, van Rens GH (2004) Psychometric properties of visionrelated quality of life questionnaires: a systematic review. Ophthal. Physiol. Opt. 2004;24:257-73. Evans BJW, Rowlands G (2004) Review article. Correctable visual impairment in older people: a major unmet need. Ophthal.Physiol.Opt. 2004;24:161-80. Jessa Z, Evans B, Thomson D, Rowlands G (2007) Vision screening of older people. Ophthal. Physiol. Opt. 2007;27:52746. Wolffsohn JS, Cochrane AL (2000) Design of the low vision quality of life questionnaire (LVQOL) and measuring the outcome of low vision rehabilitation. Am.J.Ophthalmol. 2000;130:793-802. 18 How to obtain further information A summary report, in the form of a ‘Research Findings’, entitled Improving the detection of correctable low vision in older people, by Professor Bruce Evans and Zahra Jessa, is available from: Thomas Pocklington Trust 5 Castle Row, Horticultural Place London W4 4JQ Telephone: 020 8995 0880 Email: info@pocklington-trust.org.uk Web: www.pocklington-trust.org.uk Copies of this report in large print, audio tape or CD, Braille and electronic format are available from Thomas Pocklington Trust. Background on Pocklington Thomas Pocklington Trust is the leading provider of housing, care and support services for people with sight loss in the UK. Each year we also commit around £700,000 to fund social and public health research and development projects. Pocklington’s operations offer a range of sheltered and supported housing, residential care, respite care, day services, home care services, resource centres and community based support services. A Positive about Disability and an Investor in People organisation, we are adopting quality assurance systems for all our services to ensure we not only maintain our quality standards, but also seek continuous improvement in line with the changing needs and expectations of our current and future service users. We are working in partnership with local authorities, registered social landlords and other voluntary organisations to expand our range of services. 19 Our research and development programme aims to identify practical ways to improve the lives of people with sight loss, by improving social inclusion, independence and quality of life, improving and developing service outcomes as well as focusing on public health issues. We are also applying our research findings by way of pilot service developments to test new service models and develop best practice. In this publication, the terms ‘visually impaired people’, ‘blind and partially sighted people’ and ‘people with sight loss’ all refer to people who are blind or who have partial sight. Published by Thomas Pocklington Trust 5 Castle Row, Horticultural Place Chiswick London W4 4JQ Tel: 020 8995 0880 Email: info@pocklington-trust.org.uk Website: www.pocklington-trust.org.uk Registered Charity No. 1113729 Company Registered No. 5359336 ISBN 978-1-906464-08-0 20