Colour vision testing for diabetic retinopathy: a systematic review of diagnostic accuracy and economic evaluation

Farnsworth–Munsell 100 hue test

The most comprehensive colour arrangement test, the Farnsworth–Munsell 100 hue test (FM-100), consists of four trays containing a total of 85 reference caps spanning the visible colour spectrum. The test is intended to evaluate hue discrimination ability (or colour vision aptitude). Hue discrimination ability is ascertained from the total error score, and the type of colour vision deficiency is established by interpreting a graphical illustration of the results.

Farnsworth D-15 and Lanthony desaturated D-15 tests

The Farnsworth D-15 test is an abridged version of the FM-100, consisting of 15 loose coloured caps and a single reference cap. Rather than measure overall hue discrimination ability, the original D-15 was developed to detect moderate and severe colour deficiencies and separate these from normal colour vision or more slight deficiencies. The Lanthony desaturated D-15 test is an arrangement test that is similar to the original D-15 but it has a Munsell value of 8 and chroma of 2 and should be presented under high levels of illumination (> 500 lux). 27

Lanthony New Colour Test

The Lanthony New Colour Test (NCT) contains 70 Munsell samples – four series of 15 colours with Munsell value 6 and chroma of 2, 4, 6, and 8, and 10 grey caps representing a lightness scale. For each series participants must first separate the coloured caps from the grey caps before arranging the coloured caps in colour order and the grey caps in lightness scale. As with other arrangement tests the results are plotted graphically and an error score is calculated. The NCT is intended to distinguish between slight, moderate and severe colour deficiency. 27

Mollon–Reffin Minimalist Test

Initially, participants must identify an orange ‘demonstration’ chip from among five grey chips of varying lightness. If successful, the participant must select a probe chip from the middle of a protan, deutan or tritan series. If successful, the participant is presented with a less saturated probe chip. If unsuccessful, a more saturated probe chip is presented. Participants are scored on the number of reliably identified coloured chips for each confusion line.

Farnsworth–Munsell 100 hue test

A total of 10 studies36–45 evaluated FM-100 CVT in diabetes patients.

Five36,38,39,42,44 of these studies compared mean error scores on the FM-100. The quality of reporting among these studies was generally low. Only one39 of the five papers had a representative spectrum of patients and reported the reference standard and index test in detail, and none of the papers reported the participant selection criteria that were used (Table 2 and Figure 2).

FIGURE 2.

Proportion of studies rated as yes, no or unclear for each of the QUADAS items for all tests evaluated in five or more studies. D-15, Lanthony desaturated D-15 test; FM-100, Farnsworth–Munsell 100 hue test.

The remaining five studies provided diagnostic accuracy data on the FM-100 test. QUADAS assessment indicated that the quality of reporting among these studies was generally poor (Table 2). The spectrum of patients included in each study was not representative of the general population in practice. In four studies37,41,43,45 patients were excluded if they did not have good visual acuity, whereas one study excluded patients with soft exudates. The English National Screening Programme states that all diabetes patients should be offered retinopathy screening regardless of how good their sight is, therefore these studies are not representative of the entire population who will be offered screening in practice. Only the paper by Trick et al. 45 described the patient selection criteria clearly.

Lanthony desaturated D-15 test

Six studies33,46–50 evaluated the Lanthony desaturated D-15 test in patients with diabetes. QUADAS assessment indicated that the quality of reporting among these studies was generally poor (Table 2 and Figure 2). None of the studies provided any information on blinding of outcome assessors, and only one indicated whether the clinical data available during the interpretation of test results reflected that which would be available in practice. Therefore the potential for test, diagnostic and clinical review biases among this group of studies cannot be ruled out. In addition, the participant inclusion criteria applied to these studies means that their results cannot necessarily be generalised to a diabetic screening population. In two studies46,47 participants were predominantly children and younger adults.

Lanthony New Colour Test

Two studies49,51 evaluated the NCT in diabetes patients. The quality of reporting of both studies was poor (Table 2). It was not possible to distinguish poor reporting of methods from poor methodological quality, but it is likely that there were limitations in both the internal and external validity of these studies as neither adequately reported the patient spectrum or any attempts to avoid review biases.

Mollon–Reffin Minimalist Test

One study32 evaluated the Mollon–Reffin Minimalist Test. The study was generally well reported, although no information on blinding of outcome assessors was given. In addition, only patients less than 50 years of age with type 1 diabetes were included, thereby limiting the generalisability of the study’s findings to a screening population (Table 2).

Farnsworth–Munsell 100 hue test

Five studies36,38,39,42,44 compared mean error scores on the FM-100 as opposed to investigating diagnostic accuracy (Table 4). One of these studies44 reported mean FM-100 scores and standard deviations for six grades of retinopathy, reporting an overall trend towards deterioration of colour vision (i.e. higher mean error scores) with increasing retinopathy grade (p < 0.05). However, FM-100 scores did not differ significantly among the less severe grades. A second study42 similarly found that FM-100 scores had some value in discriminating advanced retinopathy from no retinopathy but could not be used to detect early retinopathy. Two papers38,39 came from the ETDRS, one39 of which reported a significant correlation between FM-100 error score and the following factors: presence of clinically significant macular oedema (CSMO) involving the centre of the macula (p = 0.0001); presence of new vessels (p = 0.0001); presence of fluorescein leakage in centre of the macula (p = 0.0001); presence of cystoid changes in the centre of the macula (p = 0.003); and presence of focal leakage (p = 0.002). The second study38 briefly reported that there was no difference in mean FM-100 scores by level of retinopathy.

Five studies provided diagnostic accuracy data on the FM-100 test (Table 4). Figure 4 shows results from these studies plotted in ROC space. Patient spectrums were not representative of the general population who would be screened in practice; four studies37,41,43,45 excluded patients who did not have good visual acuity and one study excluded patients with soft exudates. 40

FIGURE 4.

FM-100 studies reporting 2 × 2 data plotted in receiver operating characteristic (ROC) space (see section on quality of included studies for methodological limitations of plotted studies). B-Y, blue-yellow partial error score; CV, colour vision; R-G, red–green partial error score; TES, total error score.

Thresholds to define a positive FM-100 score varied between studies. Sensitivity ranged from 11% to 74% and specificity ranged from 45% to 100%. The study reporting a specificity of 100% reported a sensitivity of only 24%. 41 The reference standard in all of the diagnostic accuracy studies was used to distinguish between patients with retinopathy and those without. One40 of these studies made the additional distinction between those with no or background retinopathy and those with more serious retinopathy. Three studies40,41,43 used the reference standard to establish the grade of retinopathy in line with the Airlie House classification system. The study40 distinguishing between those with no or background retinopathy and those with more serious retinopathy reported a sensitivity of 65% (95% CI 51% to 76%) and a specificity of 73% (95% CI 66% to 79%). One study40 reported a LR+ of 2.38 (95% CI 1.75 to 3.24). The remaining comparisons all reported even smaller LR+, with confidence intervals incorporating 1. Therefore, the available evidence does not suggest that FM-100 testing alone could be used to rule in or rule out retinopathy in diabetes patients, or to discriminate between no or background retinopathy and more serious disease.

Lanthony desaturated D-15 test

Six studies32,46–50 evaluated the Lanthony desaturated D-15 test in patients with diabetes, all of which provided diagnostic accuracy data (Table 5).

Stand-alone reference standards included ophthalmoscopy47,48 and fluorescein angiography. 50 Combined reference standards included ophthalmoscopy with fluorescein angiography49 and combined biomicroscopy/photography/angiography. 32 Only one study46 did not specify a reference standard.

Although they applied different measures, most studies used the reference standard to distinguish between diabetes patients with and without retinopathy. Retinopathy, when defined, was typically characterised by the presence of microaneurysms, haemorrhages and hard exudates. One study32 specifically evaluated the accuracy of the desaturated D-15 test in detecting ‘clinically significant macular oedema’ as opposed to any presence of retinopathy. Colour vision deficiency was generally defined as a ‘pathological’ or ‘abnormal’ desaturated D-15 score, although the specific threshold – when reported – was not consistent between studies (Table 5).

The studies investigating the presence or absence of retinopathy in adults with diabetes48–50 reported sensitivities between 79% and 87% and specificities between 33% and 47% (Figure 5). The study evaluating the D-15 test for the detection of CSMO in adults32 reported a sensitivity of 36% and a specificity of 88%.

FIGURE 5.

D-15 studies reporting 2 × 2 data plotted in receiver operating characteristic (ROC) space (see section on quality of included studies for methodological limitations of plotted studies). Grades 1–3, all ‘pathology’ on angiography; grade 1, angiography grade 1 (‘dry retinopathy’) only; 6-years, retinopathy assessed 6 years after baseline colour vision measurement.

Of the two studies investigating DR in younger patients, one47 reported a specificity of 70% (95% CI 51% to 85%) in patients with a mean age of 17 years, but provided insufficient data to calculate a sensitivity value. The second study46 administered the D-15 test in participants with a mean age of 14 years and assessed their retinopathy status 6 years later, reporting a sensitivity of 4% (95% CI 0% to 22%) and a specificity of 100% (95% CI 89% to 100%), which equates to a LR+ of 4.0 (95% CI 0.17 to 94.0) and a LR– of 0.95 (95% CI 0.85 to 1.07). On this basis there is little evidence to suggest that the D-15 test could be used to detect or predict retinopathy in young people with diabetes.

Likelihood ratios among the cross-sectional studies in adults were poor, with LR+ ranging from 1.29 to 3.87 and LR– ranging from 0.67 to 0.41, suggesting that there is little evidence to support the use of the desaturated D-15 test for detecting retinopathy in adults with diabetes. LRs were not calculable for the study investigating CSMO.

Lanthony New Colour Test

Two studies49,51 evaluated the NCT in diabetes patients (Table 6).

One study49 used a reference standard of combined ophthalmoscopy with fluorescein angiography, whereas the second51 did not specify a reference standard. Both studies aimed to differentiate between patients with and without DR. Mecca et al. 49 defined retinopathy as the presence of at least 10 microaneurysms and small haemorrhages, whereas Matsuo et al. 51 did not define retinopathy.

Both studies collected total error scores on the NCT, with Mecca et al. 49 dichotomising participants as having either ‘normal colour vision’ (no errors) or ‘altered colour vision’ (any score above 0).

The study by Matsuo et al. 51 did not present diagnostic accuracy estimates but reported that total error scores on the NCT were significantly greater among participants with DR than in those without retinopathy (p < 0.01), suggesting a possible correlation between colour vision deficits and the presence of retinopathy (Table 6).

Mecca et al. 49 reported an overall sensitivity of 79% (95% CI 69% to 87%) and specificity of 60% (95% CI 48% to 72%) for the NCT. The authors reported that combining the NCT results with findings of the D-15 test (how these were combined was not clear) increased overall sensitivity (86%) but decreased specificity (36%). Both positive and negative LRs were better for NCT alone: LR+ 1.97 (95% CI 1.45 to 2.68), LR– 0.35 (95% CI 0.22 to 0.55). Although these values are slightly more promising than for some other arrangement tests, they still indicate a poor ability to rule in or rule out disease, and it should be noted that they are derived from the results of a single, small, poor-quality study.

Mollon–Reffin Minimalist Test

One study32 evaluated the Mollon–Reffin Minimalist Test (Table 7).

A combination of slit-lamp biomicroscopy, retinal photography and fluorescein angiography was used to identify CSMO. An error score greater than 1 was considered a ‘fail’ on the Mollon-Reffin test.

Although exact numbers of patients required to calculate diagnostic accuracy measures were not reported, the authors stated that no errors were made on the protan or deutan axes by any patient. The overall sensitivity and specificity of the Mollon-Reffin test were reported to be 89% and 93% respectively. These values would suggest a LR+ of 12.7 and a LR– of 0.12. Despite the positive conclusions of this study, there is no further evidence to corroborate these findings on the diagnostic accuracy of the Mollon-Reffin test for the detection of CSMO.

Search strategy

MEDLINE, EMBASE, CINAHL, NHS Economic Evaluation Database (NHS EED) and the Health Economic Evaluation Database (HEED) were searched for economic evaluations of CVT for the diagnosis of DR. Searches were performed on 15 November 2007. The full search strategy is described in Appendix 1. In brief, the search strategy for the systematic review of CVT diagnostic accuracy was adapted by including economics search filter terms and excluding diagnostic accuracy terms. In total, after excluding duplicate articles, we identified 356 potentially relevant publications.

Model structure

We developed a decision tree and Markov model to estimate the costs and effects of adding CVT to the current NSPDR using digital photography of the retina. An NHS perspective was taken for the cost-effectiveness analysis. The model was developed in Microsoft excel 2000 and run with 1-year cycles for a time horizon of 50 years. The time horizon of 50 years was felt adequate to represent a lifetime horizon of the vast majority of the cohort. Half-cycle corrections have been used to improve the precision of cost-effectiveness estimates. To represent the uncertainty surrounding our base-case estimates of cost-effectiveness we used probabilistic sensitivity analyses (PSA), based on 1000 Monte Carlo second-order simulations, to estimate the possible distribution of cost-effectiveness. PSA results are presented graphically on the cost-effectiveness plane and cost-effectiveness acceptability curve (CEAC). 64

The hypothetical cohort consisted of patients over the age of 12 years with a recorded diagnosis of diabetes, on a centralised register and invited for screening by the English NSPDR. Two models were created, one each for type 1 diabetes and type 2 diabetes.

Current screening

The current English national screening programme invites eligible individuals with diabetes to referral units typically based at hospitals although they can also be based in mobile units, GP surgeries or optometrists offices. 65 Patients are invited for screening at annual intervals. Screening is performed by a trained and accredited technician68 using a digital non-mydriatic fundus 45-degree field camera with ophthalmological grading. If screening indicates STDR then patients are referred to an ophthalmologist for further assessment and treatment planning. If screening indicates no retinopathy or background retinopathy the individual is invited back for screening at the next annual appointment (Figure 7). In practice, policy varies by region: in some areas individuals graded with background retinopathy may be referred to an ophthalmologist.

FIGURE 7.

Screening pathway. STDR, sight-threatening diabetic retinopathy.

Disease progression

The possible health state (Markov) transitions are depicted in Figure 8. We assumed that the eyesight of patients with diabetes progresses through the various stages of retinopathy, in some cases leading to blindness. Treatment of STDR with laser photocoagulation slows down the progression from STDR to blindness, but regression from more severe to less severe retinopathy or visual impairment is assumed not to occur.

FIGURE 8.

Markov transitions. STDR, sight-threatening diabetic retinopathy.

Source data

Initial prevalence

The initial prevalence of diabetic eye disease in patients with type 1 and type 2 diabetes (Table 11) was taken from a large study of individuals attending a primary care-based DR screening programme. 67 The study was based in Liverpool and included patients with diabetes who were not under the continuing care of an ophthalmologist and who, at baseline, had not been in the systematic screening program before. Patients with type 1 (n = 831) and type 2 (n = 7231) diabetes were recruited between 1991 and 1999. Retinopathy and maculopathy were graded from dilated three-field non-stereoscopic photography by trained graders. Younis et al. 67 reported diabetes-related eye disease using an adaptation of the ETDRS gradings,66 which we mapped to the English NSC Retinopathy Grading Standard65 so that all grading definitions are consistent across the model (Table 10). For example, 442 out of 822 individuals (i.e. 53.8%) with type 1 diabetes and gradable biomicroscopy had no retinopathy (level 10).

Images from photographic screening may, because of technical failures, be ungradable. In these circumstances individuals are automatically referred to the hospital eye service for ophthalmologist grading. Data on the rate of technical failure have been taken from the Younis et al. study,67 in which 36 out of 822 (4.4%) images were ungradable for type 1 patients and 796 out of 7231 (11%) for type 2. The higher rate of ungradable images in the type 2 population reflects the older age of patients. It was assumed that ungradable images are equally likely to occur across all DR grades. The impact of uncertainty about the prevalence of retinopathy was assessed in our PSA; prevalence probabilities were assumed to follow a Dirichlet distribution based on the source data in the Younis et al. 67 study.

Disease transition

Annual transition rates for the progression through the various stages of DR, within a screened population, have been calculated separately for type 1 (Table 12) and type 2 diabetes (Table 13). 72,73 Type 1 (n = 501) and type 2 (n = 4770) patients from the Liverpool Diabetic Eye Study were screened at annual intervals for up to 6 years after first entering the screening programme. We converted the 6-year cumulative incidences of DR reported in the Liverpool study into annual transition probabilities using a standard conversion formula. 83 For example, the 6-year cumulative incidence of STDR in patients with type 1 diabetes and initially no retinopathy was 5.4%. This is equivalent to an annual transition probability of 0.92%. 83 These transition probabilities have been deflated to take account of transitions to mortality and blindness detailed below.

A proportion of patients with preproliferative STDR are not initially offered laser photocoagulation80 but remain under ophthalmologist monitoring until they develop proliferative retinopathy. We used data from the Liverpool study to estimate the annual transition from preproliferative to proliferative STDR72,73 (Table 11).

Data for the annual probabilities of transition to blindness from either no retinopathy or background retinopathy were taken from the economic evaluation of DR screening of Dasbach et al. 84 and used data from the Wisconsin Epidemiologic Study of Diabetic Retinopathy (WESDR). Dasbach et al differentiated the severity of DR into high risk and low risk of progression to blindness. For our model, no retinopathy and background retinopathy have been considered equivalent to low risk in the Dasbach model.

Effectiveness of laser photocoagulation

The transition probabilities from STDR to blindness in treated and untreated patients have been taken from Odland et al. 71 This study was a randomised controlled trial involving 107 patients with a similar degree of STDR in both eyes, followed for 5–7 years after treatment, testing the efficacy of photocoagulation. One eye chosen at random was treated whilst the other was used as a control. In total, 77 individuals completed the 5-year follow-up. At follow-up, 44% (34/77) of eyes in the control group were legally blind compared with 13% (10/77) in the treatment group, suggesting annual transition probabilities of 10.9% and 2.7% respectively (see Tables 12 and 13). Odland et al. did not differentiate between individuals with type 1 and type 2 diabetes and so, in our model, both have an identical transition probability before deflating for mortality.

Health-related quality of life

Utility values have been taken from Lloyd et al. 76 (Table 11). Utilities in this study were obtained from a group of patients with type 1 or type 2 diabetes and retinopathy in the UK (n = 122). Patients rated their current health using the EuroQol 5 dimensions (EQ-5D) questionnaire, which provides a single utility score [anchored at 1 (best health) and 0 (health state as bad as death)] for current health. Lloyd et al. found a mean utility score of 0.83 in patients with no DR and a utility score of 0.34 for patients with severe vision loss (counting fingers – hand motion). We assumed that most patients attending screening would be asymptomatic (0.83 utility score) and that the benefit of screening and treatment is through reducing the probability that a patient with STDR becomes legally blind (0.34 utility score).

Although it would have been more realistic to model differing severities of vision loss, we were unable to do so. The randomised controlled trial of photocoagulation71 did not report results in sufficient detail to be combined with the utility scores for imperfect vision estimated by Lloyd et al. 76

Mortality

Age-specific mortality rates are based on general population UK life tables (ONS, 2008, www.statistics.gov.uk/) and have been inflated to reflect the higher mortality rates of individuals with diabetes. Mortality rates gradually increase throughout the lifetime of the modelled cohort as it ages. We adjusted general population mortality rates using hazard ratios taken from Soedamah-Muthu et al. 77 for type 1 diabetes and Mulnier et al. 78 for type 2 diabetes (Table 11). Both studies were prospective cohort studies following patients from 1992 to 1999 matching a cohort with diabetes (n = 7713,77 n = 44,23078) to an age and sex-matched cohort with no history of diabetes (n = 38,518,77 n = 219,79778). Mortality hazard ratios between the two groups were calculated using Cox proportional hazard models. The mortality hazard ratio in type 1 diabetes was 3.7 times higher than that in the matched general population and for type 2 diabetes the hazard ratio was 1.93 times higher.

These mortality rates were then adjusted further to reflect positive correlation between the severity of diabetes-related eye disease and mortality based on Klein et al. 79 (Table 11). This was a population-based cohort study following 996 younger-onset and 1370 older-onset patients. Klein et al. reported the mortality hazard ratios for patients with mild retinopathy and proliferative retinopathy compared with a reference group with diabetes but no retinopathy. In our model we assumed that the relative mortality of patients with background retinopathy was equal to that of patients with ‘mild retinopathy’ and the mortality of patients with STDR was equal to that of patients with proliferative retinopathy. As there is no evidence to suggest that photocoagulation affects mortality, we assumed that mortality did not increase in patients with STDR who subsequently became blind.

Attendance at hospital eye clinics and compliance with therapy

Previous decision-analytical models have often assumed that 100% of patients who have positive screening results will attend the ophthalmology clinic for further diagnostic workup and, if indicated, undergo therapy for retinopathy. 62,85,86 We have also assumed in our base case that there is 100% attendance for both, but have used a uniform distribution between 95% and 100% to assess the impact of imperfect compliance on our conclusions in the sensitivity analysis.

Costs

All current screening costs have been taken from the 2004 costings of diabetic retinal photography screening in England of Garvican. 68 That paper estimated a cost of £23 per patient on a diabetes register and £29 per patient screened if we assume an attendance rate of 80% for registered patients at screening appointments (similar to the attendance levels used in the base-case estimate). This is made with the assumption of a geographic region containing 15,000 individuals with diabetes eligible for screening and included all costs associated with equipment (e.g. cameras, eye drops), staffing, screening invitation, administration and quality assurance. A full breakdown of the costs is shown in Table 14.

We are not aware of any studies on the costing or resource use of any CVT in a clinical setting. To minimise administrative and patient burden, we have assumed that the CVT is conducted during the same screening appointment as the retinal photography. Therefore, administration and management costs would not increase greatly. Many of the automated CVTs, including the TCT used by Ong et al. ,31,60 are not commercially available. Therefore, as an example of the likely cost of CVT, we have estimated the cost of including the FM-100 test alongside retinal photography screening. We vary the cost of CVT widely in our sensitivity analysis to establish a threshold cost at which CVT might be cost-effective.

The capital cost of a FM-100 test kit is £350 (Richmond Products), and a lamp to provide the appropriate lighting costs £140 (Macbeth Lighting). It is assumed that, at most, four CVT sets and lamps would be purchased each year to be used alongside the four retinal cameras. Over time, and through use, the testing kit hues degrade; therefore, we have assumed that the kits need replacing each year. The introduction of the test would lead to an increase in staff time because of extra time spent administering the CVTs and automated recording of the results. We have made the assumption that additional staff time would be equivalent to 3.5 full-time technicians. Effectively, the introduction of CVT would double the amount of technician time compared with a screening programme based solely on retinal photography.

The extra costs associated with using the FM-100 test alongside photographic screening are presented in Table 15. For a geographical area of 15,000 people eligible for screening this will lead to an increased cost of £7.80 per patient screened. The FM-100 probably represents a lower bound for the cost of CVT. For automated CVTs, the equipment and maintenance costs are likely to be much higher, although technician costs could be lower.

Based on previous work,68 the cost for a referral to an ophthalmologist is estimated to be £65 and bilateral photocoagulation costs are £815, which is made up of 1.5 treatments (average number of treatments needed per patient) and 8 outpatient follow-ups. Those who are screened as having preproliferative retinopathy are assumed to be recalled to an ophthalmologist annually, incurring an ophthalmologist cost of £65. This is incurred until the patient progresses to proliferative retinopathy or until they go blind or die. All patients who are blind also incur an annual cost of £872. 81 An annual ophthalmologist cost of £65 has been added to post laser photocoagulation patients to represent the cost of continued monitoring. All costs have been varied in the PSA, assuming uniform distributions from 50% to 200% of the primary estimate. This is to reflect the large uncertainties regarding the costs.

All costs have been inflated to 2007 prices using the hospital and community health services price indices. All costs and quality-adjusted life-years (QALYs) after the first year of the model have been discounted at 3.5% per annum.

Analysis

We calculated the incremental cost-effectiveness ratio (ICER) of screening using CVT as an adjunct to retinal photography versus retinal photography alone. Although there is no uniformly accepted ICER threshold defining cost-effective care, NICE87 has cited a £20,000–30,000 per QALY threshold below which an intervention is generally accepted as cost-effective.

Threshold/scenario analysis

Figures 9 and 10 depict the impact of different levels of sensitivity of CVT on the ICER, keeping all other base-case estimates constant. The ICER rises sharply as the sensitivity of combined CVT and retinal photography falls towards 0.88 (the sensitivity of retinal photography alone). However, because CVT screening is relatively cheap and photocoagulation is effective at reducing progression to blindness, the sensitivity of CVT and retinal photography screening has to fall to 0.896 (type 1 diabetes) or 0.916 (type 2 diabetes) before the ICER for type 1 diabetes exceeds the £20,000 threshold.

FIGURE 9.

Two-way threshold analysis of the sensitivity and cost of colour vision testing for sight-threatening diabetic retinopathy (STDR)versus the incremental cost-effectiveness ratio (ICER) (type 1 diabetes).

FIGURE 10.

Two-way threshold analysis of the sensitivity and cost of colour vision testing for sight-threatening diabetic retinopathy (STDR) versus the incremental cost-effectiveness ratio (ICER) (type 2 diabetes).

At an additional cost of CVT testing of £29, doubling the overall cost of the screening workup, the ICER still remains below the £20,000 threshold for our base case of 94% sensitivity for STDR in type 1 diabetes (Figure 9). However at this cost a small reduction in sensitivity below the base case 94% (0.924 for type 1 diabetes) would render the use of CVT not cost-effective. If CVT costs £29, the ICER for type 2 diabetes is not below the £20,000 threshold unless the sensitivity is increased to 0.984 (Figure 10).

Probabilistic sensitivity analysis

The cost-effectiveness planes in Figures 11 and 12 show that, under our structural assumptions, the inclusion of CVT in screening always results in increased costs and is as effective or more effective than retinal photography screening alone. This is because the results of CVT only affect the management of patients with negative retinal photography screening results, in effect increasing the sensitivity and reducing the specificity of screening. The increased sensitivity leads to increased effectiveness, but the cost of the CVT itself and the additional ophthalmologist visits because of reduced specificity lead to increased costs overall. The increased number of false-positive cases referred to the ophthalmologist may also increase unnecessary anxiety, although this is difficult to quantify and is not included in our QALY calculations. The uncertainty about whether CVT is more accurate than current retinal photography screening, because of the paucity of diagnostic accuracy studies, is depicted by the number of points that lie on or very close to the vertical axis (where the CVT strategy is not effective).

FIGURE 11.

Cost-effectiveness plane – type 1 diabetes.

FIGURE 12.

Cost-effectiveness plane – type 2 diabetes.

The CEACs (Figures 13 and 14) derived from the PSA indicate a 52.9% probability (type 1) or 39% probability (type 2) that the addition of CVT to retinal photography will be cost-effective at a cost per QALY threshold of £20,000. The fact that the probability of cost-effectiveness depicted in the CEACs never increases beyond 70%, no matter how much society is willing to pay for a QALY, reflects the considerable uncertainty that CVT combined with retinal photography is more sensitive than retinal photography alone. If CVT combined with retinal photography is not more sensitive, it is unlikely to be effective or cost-effective in improving patient outcomes.

FIGURE 13.

Cost-effectiveness acceptability curve – type 1 diabetes.

FIGURE 14.

Cost-effectiveness acceptability curve – type 2 diabetes.

Quantity and quality of evidence

The limited available evidence on the diagnostic accuracy of CVT in DR is generally of poor methodological quality. With a few exceptions, the potential for multiple biases could not be excluded. For example, spectrum of disease is an important consideration in the evaluation of diagnostic accuracy, yet on the QUADAS evaluation only one included study met the ‘appropriate spectrum composition’ criterion. In addition, the majority of studies excluded groups of patients (i.e. those with poorer visual acuity) that form part of the typical diabetes screening population, thereby limiting the generalisability of these findings to a screening context.

Pseudoisochromatic plates

A very small amount of poorly reported evidence is available on the diagnostic accuracy of pseudoisochromatic plates. Ishihara plates are not designed to detect tritan defects, and the available evidence suggests that they are of little value in screening for DR.

Arrangement tests

A small number of studies show a correlation between FM-100 score and degree or stage of retinopathy; however, diagnostic accuracy estimates were highly variable and no consistent cut-off value for FM-100 score was identified among the included studies. Even for the detection of more advanced retinopathy, accuracy estimates were poor, which suggests that FM-100 alone would not be useful as a screening tool for the detection of DR.

Although diagnostic accuracy estimates in adults with diabetes were more consistent across studies of the D-15 than across those of the FM-100, specificities were generally low and, on the basis of the available evidence, the D-15 alone does not appear to be a promising screening tool for the detection of DR.

The NCT showed little promise as a screening test and, although a single study reported promising results for the Mollon–Reffin Minimalist Test in detecting CSMO, no data were available on its ability to detect less severe forms of retinopathy. In addition, this test is not currently commercially available.

Computerised/automated tests

Of the six studies evaluating computerised or automated CVTs, four focused on tritan testing using the ChromaTest or variants of the SGM. Specificities were generally good for detecting more advanced retinopathy and CSMO, although sensitivities were more variable. As with the surface arrangement tests, these methods appear less sensitive and specific in detecting milder forms of retinopathy. In one study, the combination of positive results on both TCT and retinal photography produced an extremely high positive likelihood ratio for the detection of sight-threatening retinopathy, but did not appear to improve the ability of retinal photography alone to rule out disease. The optimal performance thresholds reported for these tests were derived from the data collected in the evaluation, thereby potentially biasing the diagnostic accuracy results in favour of the test. It should also be noted that these studies were all conducted in collaboration with the developers of the ChromaTest/SGM tests; there is no independent evaluative evidence available.

Anomaloscopes

Although generally considered the gold standard for detecting (typically red–green) colour deficiencies, the evidence on anomaloscopes was limited in both quality and quantity. The two available studies gave conflicting findings, one reporting that yellow–blue discrimination was the best single predictive factor of retinopathy and the other reporting no differences in anomaloscope performance between diabetes patients with and without retinopathy. Therefore these methods cannot currently be recommended as a tool for screening for DR.

MEDLINE and MEDLINE In-Process & Other Non-Indexed Citations (OVID gateway) 1950 to September Week 3 2007

Searched 4 October 2007.

643 records were retrieved in MEDLINE and 6 in MEDLINE In-Process & Other Non-Indexed Citations.

1. Color Vision Defects/

2. Color Perception/

3. Color Perception Tests/

4. Vision Screening/

5. Vision Tests/

6. Contrast Sensitivity/

7. ((color$or colour$or vision or visual) adj3 (test$or screen$)).ti,ab.

8. ((color$or colour$) adj3 (blind$or deficien$or defect$or loss or impair$or perception)).ti,ab.

9. (monochromatopsia or achromatopsia or deutan or protan).ti,ab.

10. anomaloscop$.ti,ab.

11. pseudoisochromatic$.ti,ab.

12. hue$discrimination$.ti,ab.

13. lantern$.ti,ab.

14. (tritan or TCT).ti,ab.

15. Ishihara$.ti,ab.

16. Kojima$.ti,ab.

17. Ohkuma$.ti,ab.

18. Matsubara$.ti,ab.

19. Dvorine$.ti,ab.

20. cvtmet.ti,ab.

21. (Hardy Rand Rittler$or AOHRR or AO H-R-R).ti,ab.

22. Lanthony$.ti,ab.

23. (panel d 15 or panel d15 or d 15 panel or d15 panel).ti,ab.

24. Farnsworth$.ti,ab.

25. ((color$or colour$) adj plate).ti,ab.

26. Mollon Reffin$.ti,ab.

27. Pease Allen$.ti,ab.

28. Giles Archer$.ti,ab.

29. Adams Desaturated.ti,ab.

30. (City University or CUCVT).ti,ab.

31. Velhagen Pflugertrident$.ti,ab.

32. (contrast adj3 sensitivit$).ti,ab.

33. or/1–32

34. exp Diabetes Mellitus/

35. diabet$.ti,ab.

36. (IDDM or NIDDM or T2DM).ti,ab.

37. or/34–36

38. 33 and 37

39. exp Diabetic Retinopathy/

40. retinopath$.ti,ab.

41. exp Macular Degeneration/

42. Retinal Hemorrhage/

43. Retinal Neovascularization/

44. ((optic$or ocular$or intraocular$or macula$or retina$) adj3 (edema$or oedema$)).ti,ab.

45. maculopath$.ti,ab.

46. (microaneurism$or micro aneurism$or microaneurysm$or micro aneurysm$).ti,ab.

47. ((ocular$or intraocular$or optic$or retina$or eye$) adj3 (hemorrhag$or haemorrhag$or neovascular$or leak$or perme$or bleed$or neo vascular$)).ti,ab.

48. or/39–47

49. 38 and 48

50. Animals/

51. Humans/

52. 50 not (50 and 51)

53. 49 not 52

EMBASE (OVID gateway) 1980 to 2007 Week 39

Searched 4 October 2007.

419 records were retrieved.

1. Color Vision Defect/

2. Color Vision/

3. Color Vision Test/

4. Vision Test/

5. Color Blindness/

6. Contrast Sensitivity/

7. ((color$or colour$or vision or visual) adj3 (test$or screen$)).ti,ab.

8. ((color$or colour$) adj3 (blind$or deficien$or defect$or loss or impair$or perception)).ti,ab.

9. (monochromatopsia or achromatopsia or deutan or protan).ti,ab.

10. anomaloscop$.ti,ab.

11. pseudoisochromatic$.ti,ab.

12. hue$discrimination$.ti,ab.

13. lantern$.ti,ab.

14. (tritan or TCT).ti,ab.

15. Ishihara$.ti,ab.

16. Kojima$.ti,ab.

17. Ohkuma$.ti,ab.

18. Matsubara$.ti,ab.

19. Dvorine$.ti,ab.

20. cvtmet.ti,ab.

21. (Hardy Rand Rittler$or AOHRR or AO H-R-R).ti,ab.

22. Lanthony$.ti,ab.

23. (panel d 15 or panel d15 or d 15 panel or d15 panel).ti,ab.

24. Farnsworth$.ti,ab.

25. ((color$or colour$) adj plate).ti,ab.

26. Mollon Reffin$.ti,ab.

27. Pease Allen$.ti,ab.

28. Giles Archer$.ti,ab.

29. Adams Desaturated.ti,ab.

30. (City University or CUCVT).ti,ab.

31. Velhagen Pflugertrident$.ti,ab.

32. (contrast adj3 sensitivit$).ti,ab.

33. or/1–32

34. exp Diabetes Mellitus/

35. diabet$.ti,ab.

36. (IDDM or NIDDM or T2DM).ti,ab.).ti,ab.

37. or/34–36

38. 33 and 37

39. exp Retinopathy/

40. retinopath$.ti,ab.

41. Retina Maculopathy/

42. Retina Edema/

43. Retina Hemorrhage/

44. Retina Neovascularization/

45. Retina Macula Degeneration/

46. Microaneurysm/

47. ((optic$or ocular$or intraocular$or macula$or retina$) adj3 (edema$or oedema$)).ti,ab.

48. maculopath$.ti,ab.

49. (microaneurism$or micro aneurism$or microaneurysm$or micro aneurysm$).ti,ab.

50. ((ocular$or intraocular$or optic$or retina$or eye$) adj3 (hemorrhag$or haemorrhag$or neovascular$or leak$or perme$or bleed$or neo vascular$)).ti,ab.

51. or/39–50

52. 38 and 51

53. Animal/or Animal Experiment/or Nonhuman/

54. (rat or rats or mouse or mice or murine or rodent or rodents or hamster or hamsters or pig or pigs or porcine or rabbit or rabbits or animal or animals or dogs or dog or cats or cow or bovine or sheep or ovine or monkey or monkeys).ti,ab,sh.

55. or/53–54

56. exp Human/or Human Experiment/

57. 55 not (55 and 56)

58. 52 not 57

CINAHL (OVID gateway) 1982 to September Week 4 2007

Searched 4 October 2007.

115 records were retrieved.

1. Color Perception/

2. Color Perception Tests/

3. Color Vision Defects/

4. Vision Screening/

5. Vision Tests/

6. ((color$or colour$or vision or visual) adj3 (test$or screen$)).ti,ab.

7. ((color$or colour$) adj3 (blind$or deficien$or defect$or loss or impair$or perception)).ti,ab.

8. (monochromatopsia or achromatopsia or deutan or protan).ti,ab.

9. anomaloscop$.ti,ab.

10. pseudoisochromatic$.ti,ab.

11. hue$discrimination$.ti,ab.

12. lantern$.ti,ab.

13. (tritan or TCT).ti,ab.

14. Ishihara$.ti,ab.

15. Kojima$.ti,ab.

16. Ohkuma$.ti,ab.

17. Matsubara$.ti,ab.

18. Dvorine$.ti,ab.

19. cvtmet.ti,ab.

20. (Hardy Rand Rittler$or AOHRR or AO H-R-R).ti,ab.

21. Lanthony$.ti,ab.

22. (panel d 15 or panel d15 or d 15 panel or d15 panel).ti,ab.

23. Farnsworth$.ti,ab.

24. ((color$or colour$) adj plate).ti,ab.

25. Mollon Reffin$.ti,ab.

26. Pease Allen$.ti,ab.

27. Giles Archer$.ti,ab.

28. Adams Desaturated.ti,ab.

29. (City University or CUCVT).ti,ab.

30. Velhagen Pflugertrident$.ti,ab.

31. (contrast adj3 sensitivit$).ti,ab.

32. or/1–31

33. exp Diabetes Mellitus/

34. diabet$.ti,ab.

35. (IDDM or NIDDM or T2DM).ti,ab.

36. or/33–35

37. 32 and 36

38. Diabetic Retinopathy/

39. retinopath$.ti,ab.

40. Macular Degeneration/

41. Eye Hemorrhage/

42. ((optic$or ocular$or intraocular$or macula$or retina$) adj3 (edema$or oedema$)).ti,ab.

43. maculopath$.ti,ab.

44. (microaneurism$or micro aneurism$or microaneurysm$or micro aneurysm$).ti,ab.

45. ((ocular$or intraocular$or optic$or retina$or eye$) adj3 (hemorrhag$or haemorrhag$or neovascular$or leak$or perme$or bleed$or neo vascular$)).ti,ab.

46. or/38–45

47. 37 and 46

Science Citation Index (Web of Science) 1900 to 30 September 2007

Searched 4 October 2007.

463 records were retrieved.

TS=(color* SAME test*) or TS=(color* SAME screen*) or TS=(colour* SAME test*) or TS=(colour* SAME screen*) or TS=(vision SAME test*) or TS=(vision SAME screen*) or TS=(visual SAME test*) or TS=(visual SAME screen*)

TS=(color* SAME blind*) or TS=(color* SAME deficien*) or TS=(color* SAME defect*) or TS=(color* SAME loss*) or TS=(color* SAME impair*) or TS-(color* SAME perception) or TS=(colour* SAME blind*) or TS=(colour* SAME deficien*) or TS=(colour* SAME defect*) or TS=(colour* SAME loss*) or TS=(colour* SAME impair*) or TS=(colour* SAME perception)

TS=(monochromatopsia or achromatopsia or deutan or protan)

TS=(anomaloscop* or pseudoisochromatic*) or TS=(hue* discrimination*) or TS=(lantern* or tritan or TCT or Ishihara* or Kojima* or Ohkuma* or Matsubara* or Dvorine* or cvtmet)

TS=(Hardy Rand Rittler*) or TS=(AOHRR or AO H-R-R or Lanthony*) or TS=(panel d 15) or TS=(panel d15) or TS=(d 15 panel) or TS=(d15 panel) or TS=(Farnsworth*)

TS=(color* plate) or TS=(colour* plate) or TS=(Mollon Reffin*) or TS=(Pease Allen*) or TS=(Giles Archer*) or TS=(Adams Desaturated) or TS=(City University) or TS=(CUCVT) or TS=(Velhagen Pflugertrident*) or TS=(contrast SAME sensitivit*)

#1 or #2 or #3 or #4 or #5 or #6

TS=(diabet* or IDDM or NIDDM or T2DM)

#7 and #8

TS=(retinopath*) or TS=(macula* SAME edema*) or TS=(macula* SAME oedema*) or TS=(retina* SAME edema*) or TS=(retina* SAME oedema*) or TS=(optic* SAME edema*) or TS=(ocular* SAME edema*) or TS=(optic* SAME oedema*) or TS=(ocular* SAME oedema*)or TS=(intraocular* SAME edema*) or TS=(intraocular* SAME oedema*) or TS=(maculopath*) or TS=(microaneurysm*) or TS=(micro aneurysm*) or TS=(microaneurism*) or TS=(micro aneurism*) or TS= (retina* SAME hemorrhag*) or TS=(retina* SAME haemorrhage*) or TS=(retina* SAME neovascular*) or TS=(retina* SAME leak*) or TS=(retina* SAME perme*) or TS=(retina* SAME bleed*) or TS= (eye* SAME hemorrhag*) or TS=(eye* SAME haemorrhage*) or TS=(eye* SAME neovascular*) or TS=(eye* SAME leak*) or TS=(eye* SAME perme*) or TS=(eye* SAME bleed*)

TS= (ocular* SAME hemorrhag*) or TS=(ocular* SAME haemorrhage*) or TS=(ocular* SAME neovascular*) or TS=(ocular* SAME leak*) or TS=(ocular* SAME perme*) or TS=(ocular SAME bleed*) or TS= (intraocular* SAME hemorrhag*) or TS=(intraocular* SAME haemorrhage*) or TS=(intraocular* SAME neovascular*) or TS=(intraocular* SAME leak*) or TS=(intraocular* SAME perme*) or TS=(intraocular* SAME bleed*) or TS= (optic* SAME hemorrhag*) or TS=(optic* SAME haemorrhage*) or TS=(optic* SAME neovascular*) or TS=(intraocular* SAME leak*) or TS=(optic* SAME perme*) or TS=(optic* SAME bleed*)

#10 or #11

#9 and #12

TS=(rat or rats or mouse or mice or murine or rodent or rodents or hamster or hamsters or pig or pigs or porcine or rabbit or rabbits or animal or animals or dogs or dog or cats or cow or bovine or sheep or ovine or monkey or monkeys)

#13 NOT #14

CDSR and CENTRAL (Cochrane Library) 2007 Issue 3

Searched 4 October 2007.

38 reviews were retrieved in CDSR and 84 records in CENTRAL.

1. #1 MeSH descriptor Color Vision Defects explode all trees

2. #2 MeSH descriptor Color Perception explode all trees

3. #3 MeSH descriptor Color Perception Tests explode all trees

4. #4 MeSH descriptor Vision Screening explode all trees

5. #5 MeSH descriptor Vision Tests explode all trees

6. #6 MeSH descriptor Contrast Sensitivity explode all trees

7. #7 (color* NEAR/3 test*) or (colour* NEAR/3 test*) or (color* NEAR/3 screen*) or (colour* NEAR/3 screen*) or (vision NEAR/3 test*) or (vision NEAR/3 screen*) or (visual NEAR/3 test*) or (visual NEAR/3 screen*)

8. #8 (color* NEAR/3 blind*) or (color* NEAR/3 deficien*) or (color* NEAR/3 defect*) or (color* NEAR/3 loss) or (color* NEAR/3 impair*) or (colour* NEAR/3 blind*) or (colour* NEAR/3 deficien*) or (colour* NEAR/3 defect*) or (colour* NEAR/3 loss) or (colour* NEAR/3 impair*)

9. #9 monochromatopsia or achromatopsia or deutan or protan or anomaloscop* or pseudoisochromatic* or “hue* discrimination*” or lantern* or tritan or TCT

10. #10 ishihara* or kojima* or ohkuma* or matsubara* or dvorine* or cvtmet or “hardy rand rittler*” or AOHRR or AO H-R-R or lanthony* or “panel d 15” or “panel d15” or “d 15 panel” or “d15 panel” or farnsworth* or “color* plate” or “colour* plate” or “mollon reffin*” or “pease allen*” or “giles archer*” or “adams desaturated” or “City University” or CUCVT or “velhagen pflugertrident*” or (contrast NEAR/3 sensitivit*)

11. #11 (#1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9 OR #10)

12. #12 MeSH descriptor Diabetes Mellitus explode all trees

13. #13 (diabet* or IDDM or NIDDM or T2DM)

14. #14 (#12 OR #13)

15. #15 (#11 AND #14)

16. #16 MeSH descriptor Diabetic Retinopathy explode all trees

17. #17 (retinopath*)

18. #18 MeSH descriptor Macular Degeneration explode all trees

19. #19 MeSH descriptor Retinal Hemorrhage explode all trees

20. #20 MeSH descriptor Retinal Neovascularization explode all trees

21. #21 (macula* NEAR/3 edema*) or (macula* NEAR/3 oedema*) or (retina* NEAR/3 edema*) or (retina* NEAR/3 oedema*) or (optic* NEAR/3 edema*) or (optic NEAR/3 oedema*) or (ocular* NEAR/3 edema*) or (ocular* NEAR/3 oedema*) or (intraocular* NEAR/3 edema*)

22. #22 (maculopath* or microaneurysm* or “micro aneurysm*” or microaneurism* or “micro aneurism*” or (retina* NEAR/3 hemorrhag*) or (retina* NEAR/3 haemorrhag*) or (retina* NEAR/3 neovascular*) or (retina* NEAR/3 leak) or (retina* NEAR/3 perme*) or (eye* NEAR/3 hemorrhag*) or (eye* NEAR/3 haemorrhag*) or (eye* NEAR/3 neovascular*) or (eye* NEAR/3 leak) or (eye* NEAR/3 perme*) or (ocular* NEAR/3hemorrhag*) or (ocular* NEAR/3 haemorrhag*) or (ocular* NEAR/3 neovascular*) or (ocular* NEAR/3 leak*) or (ocular* NEAR/3 perme*) or (ocular* NEAR/3 bleed*) or (intraocular* NEAR/3hemorrhag*) or (intraocular* NEAR/3 haemorrhag*) or (intraocular* NEAR/3 neovascular*) or (intraocular* NEAR/3 leak*) or (intraocular* NEAR/3 perme*) or (intraocular* NEAR/3 bleed*)

23. #23 (#16 OR #17 OR #18 OR #19 OR #20 OR #21 OR #22)

24. #24 (#15 AND #23)

DARE and HTA (CRD internal databases) 1994 to September 2007

Searched 4 October 2007.

0 records were retrieved in DARE and 1 record was retrieved in HTA.

s (color$or colour$or vision or visual)(w3)(test$or screen$)

s (color$or colour$)(w3)(blind$or deficien$or defect$or loss or impair$or perception)

s monochromatopsia or achromatopsia or deutan or protan or anomaloscop$or pseudoisochromatic$or hue$(w)discrimination$or lantern$or tritan or TCT

s ishihara$or kojima$or ohkuma$or matsubara$or dvorine$or cvtmet or hardy(w)rand(w)rittler$or AOHRR or AO(w)H(w)R(w)R or lanthony$or panel(w)d(w)15 or panel(w)d15 or d(w)15(w)panel or d15(w)panel or farnsworth$or (color$or colour$)(w)plate or mollon(w)reffin$or pease(w)allen$or giles(w)archer$or adams(w)desaturated or City(w)University or CUCVT or velhagen(w)pflugertrident$or contrast(w3)sensitivit$

s s1 OR s2 OR s3 OR s4

s diabet$or IDDM or NIDDM or T2DM

s s5 and s6

s retinopath$

s (optic$or ocular$or intraocular$or macula$or retina$)(w3)(edema$or oedema$)

s maculopath$or microaneurysm$or micro(w)aneurysm$or microaneurism$or micro(w)aneurism$or (ocular$or intraocular$or optic$or retina$or eye$)(w3)(hemorrhag$or haemorrhag$or neovascular$or leak or perme$or bleed$or neo(w)vascular$)

s s8 OR s9 OR s10

s s7 AND s11

LILACS (Birme Virtual Health Library) 1982 to September 2007

Searched 4 October 2007.

1 record was retrieved.

(color test$) OR (color screen$) OR (colour test$) OR (colour screen$) OR (vision test$) OR (vision screen$) OR (visual test$) OR (visual screen$) OR (color blind$) OR (colour blind$) OR monochromatopsia OR achromatopsia OR deutan OR protan OR anomaloscop$OR pseudoisochromatic$OR (hue$discrimination$) OR lantern$or tritan OR TCT OR ishihara$OR kojima$OR ohkuma$OR matsubara$OR dvorine$OR cvtmet OR (hardy rand rittler$) OR AOHRR OR lanthony$OR (panel d 15) OR (panel d15) OR (d 15 panel) OR (d15 panel) OR farnsworth$OR (color plate) OR (colour plate) OR (mollon reffin$) OR (pease allen$) OR (giles archer$) OR (adams desaturated) OR (City University) OR CUCVT OR (velhagen pflugertrident$) OR (contrast sensitivit$)[words]

AND

diabet$OR IDDM OR NIDDM OR T2DM[words]

AND

retinopath$OR (macula$edema$) OR (macula$oedema$) OR (retina$edema$) OR (retina$oedema$) OR maculopath$OR microaneurysm$OR (micro aneurysm$) OR microaneurism$OR (micro aneurism$) OR (retina$hemorrag$) OR (retina$haemorrhage$) OR (retina$neovascular$) OR (retina$leak) OR (retina$perme$) OR (eye$hemorrhag$) OR (eye$haemorrhag$) OR (eye$neovascular$) OR (eye$leak) OR (eye$perme$)[words]

BIOSIS (Dialog) 1926 to September Week 5 2007

Searched 5 October 2007.

325 records were retrieved.

s (color? or colour? or vision or visual)(3n)(test or tests or testing or tested or screen or screens or screening or screened)

s (color? or colour?)(3n)(blind or blindness or deficient or deficiency or defect or defects or defection or loss or impair or impaired or impairment or perception)

s monochromatopsia or achromatopsia or deutan or protan

s anomaloscop? or pseudoisochromatic? or hue(w)discrimination? or lantern? or tritan or TCT or Ishihara? or Kojima? or ohkuma? or Matsubara? or Dvorine? or cvtmet or Hardy(w)Rand(w)Rittler? or AOHRR or Lanthony? or panel(w)d(w)15 or panel(w)d15 or d(w)15(w)panel or d15(w)panel or Farnsworth? or color?(w)plate or colour?(w)plate or Mollon(w)Reffin? or Pease(w)Allen? or Giles(w)Archer? or Adams(w)Desaturated or City(w)University or CUCVT or Velhagen(w)Pflugertrident? or contrast(3n)sensitivit?

s s1:s4

s diabet? or IDDM or NIDDM or T2DM

s retinopath? or macula?(3n)edema? or macula?(3n)oedema? or retina?(3n)edema? or retina?(3n)oedema? or optic?(3n)edema? or optic?(3n)oedema? or ocular?(3n)edema? or ocular?(3n)oedema? or intraocular?(3n)edema? or intraocular?(3n)oedema? or maculopath? or microaneurysm? or micro(w)aneurysm? or microaneurism? or micro(w)aneurism? or retina?(3n)hemorrhag? or retina?(3n)haemorrhage? or retina?(3n)neovascular? or retina?(3n)leak or retina?(3n)perme? or eye?(3n)hemorrhag? or eye?(3n)haemorrhag? or eye?(3n)neovascular? or eye?(3n)leak or eye?(3n)perme? or ocular?(3n)hemorrhag? or ocular?(3n)haemorrhag? or ocular?(3n)neovascular? or ocular?(3n)leak? or ocular?(3n)perme? or ocular?(3n)bleed? or intraocular?(3n)hemorrhag? or intraocular?(3n)haemorrhag? or intraocular?(3n)neovascular? or intraocular?(3n)leak? or intraocular?(3n)perme? or intraocular?(3n)bleed? or optic?(3n)hemorrhag? or optic?(3n)haemorrhag? or optic?(3n)neovascular? or optic?(3n)leak? or optic?(3n)perme? or optic?(3n)bleed?

s s5 and s6 and s7

Pascal (Dialog) 1973 to September Week 4 2007

Searched 5 October 2007.

83 records were retrieved.

s (color? or colour? or vision or visual)(3n)(test or tests or testing or tested or screen or screens or screening or screened)

s (color? or colour?)(3n)(blind or blindness or deficient or deficiency or defect or defects or defection or loss or impair or impaired or impairment or perception)

s monochromatopsia or achromatopsia or deutan or protan

s anomaloscop? or pseudoisochromatic? or hue(w)discrimination? or lantern? or tritan or TCT or Ishihara? or Kojima? or ohkuma? or Matsubara? or Dvorine? or cvtmet or Hardy(w)Rand(w)Rittler? or AOHRR or Lanthony? or panel(w)d(w)15 or panel(w)d15 or d(w)15(w)panel or d15(w)panel or Farnsworth? or color?(w)plate or colour?(w)plate or Mollon(w)Reffin? or Pease(w)Allen? or Giles(w)Archer? or Adams(w)Desaturated or City(w)University or CUCVT or Velhagen(w)Pflugertrident? or contrast(3n)sensitivit?

s s1:s4

s diabet? or IDDM or NIDDM or T2DM

s retinopath? or macula?(3n)edema? or macula?(3n)oedema? or retina?(3n)edema? or retina?(3n)oedema? or optic?(3n)edema? or optic?(3n)oedema? or ocular?(3n)edema? or ocular?(3n)oedema? or intraocular?(3n)edema? or intraocular?(3n)oedema? or maculopath? or microaneurysm? or micro(w)aneurysm? or microaneurism? or micro(w)aneurism? or retina?(3n)hemorrhag? or retina?(3n)haemorrhage? or retina?(3n)neovascular? or retina?(3n)leak or retina?(3n)perme? or eye?(3n)hemorrhag? or eye?(3n)haemorrhag? or eye?(3n)neovascular? or eye?(3n)leak or eye?(3n)perme? or ocular?(3n)hemorrhag? or ocular?(3n)haemorrhag? or ocular?(3n)neovascular? or ocular?(3n)leak? or ocular?(3n)perme? or ocular?(3n)bleed? or intraocular?(3n)hemorrhag? or intraocular?(3n)haemorrhag? or intraocular?(3n)neovascular? or intraocular?(3n)leak? or intraocular?(3n)perme? or intraocular?(3n)bleed? or optic?(3n)hemorrhag? or optic?(3n)haemorrhag? or optic?(3n)neovascular? or optic?(3n)leak? or optic?(3n)perme? or optic?(3n)bleed?

s s5 and s6 and s7

Dissertation Abstracts (Dialog) 1861 to July 2007

Searched 5 October 2007.

4 records were retrieved.

s (color? or colour? or vision or visual)(3n)(test or tests or testing or tested or screen or screens or screening or screened)

s (color? or colour?)(3n)(blind or blindness or deficient or deficiency or defect or defects or defection or loss or impair or impaired or impairment or perception)

s monochromatopsia or achromatopsia or deutan or protan

s anomaloscop? or pseudoisochromatic? or hue(w)discrimination? or lantern? or tritan or TCT or Ishihara? or Kojima? or ohkuma? or Matsubara? or Dvorine? or cvtmet or Hardy(w)Rand(w)Rittler? or AOHRR or Lanthony? or panel(w)d(w)15 or panel(w)d15 or d(w)15(w)panel or d15(w)panel or Farnsworth? or color?(w)plate or colour?(w)plate or Mollon(w)Reffin? or Pease(w)Allen? or Giles(w)Archer? or Adams(w)Desaturated or City(w)University or CUCVT or Velhagen(w)Pflugertrident? or contrast(3n)sensitivit?

s s1:s4

s diabet? or IDDM or NIDDM or T2DM

s retinopath? or macula?(3n)edema? or macula?(3n)oedema? or retina?(3n)edema? or retina?(3n)oedema? or optic?(3n)edema? or optic?(3n)oedema? or ocular?(3n)edema? or ocular?(3n)oedema? or intraocular?(3n)edema? or intraocular?(3n)oedema? or maculopath? or microaneurysm? or micro(w)aneurysm? or microaneurism? or micro(w)aneurism? or retina?(3n)hemorrhag? or retina?(3n)haemorrhage? or retina?(3n)neovascular? or retina?(3n)leak or retina?(3n)perme? or eye?(3n)hemorrhag? or eye?(3n)haemorrhag? or eye?(3n)neovascular? or eye?(3n)leak or eye?(3n)perme? or ocular?(3n)hemorrhag? or ocular?(3n)haemorrhag? or ocular?(3n)neovascular? or ocular?(3n)leak? or ocular?(3n)perme? or ocular?(3n)bleed? or intraocular?(3n)hemorrhag? or intraocular?(3n)haemorrhag? or intraocular?(3n)neovascular? or intraocular?(3n)leak? or intraocular?(3n)perme? or intraocular?(3n)bleed? or optic?(3n)hemorrhag? or optic?(3n)haemorrhag? or optic?(3n)neovascular? or optic?(3n)leak? or optic?(3n)perme? or optic?(3n)bleed?

s s5 and s6 and s7

Inside Conferences (Dialog) 1993 to 3 October 2007

Searched 5 October 2007.

0 records were retrieved.

s (color? or colour? or vision or visual)(3n)(test or tests or testing or tested or screen or screens or screening or screened)

s (color? or colour?)(3n)(blind or blindness or deficient or deficiency or defect or defects or defection or loss or impair or impaired or impairment or perception)

s monochromatopsia or achromatopsia or deutan or protan

s anomaloscop? or pseudoisochromatic? or hue(w)discrimination? or lantern? or tritan or TCT or Ishihara? or Kojima? or ohkuma? or Matsubara? or Dvorine? or cvtmet or Hardy(w)Rand(w)Rittler? or AOHRR or Lanthony? or panel(w)d(w)15 or panel(w)d15 or d(w)15(w)panel or d15(w)panel or Farnsworth? or color?(w)plate or colour?(w)plate or Mollon(w)Reffin? or Pease(w)Allen? or Giles(w)Archer? or Adams(w)Desaturated or City(w)University or CUCVT or Velhagen(w)Pflugertrident? or contrast(3n)sensitivit?

s s1:s4

s diabet? or IDDM or NIDDM or T2DM

s retinopath? or macula?(3n)edema? or macula?(3n)oedema? or retina?(3n)edema? or retina?(3n)oedema? or optic?(3n)edema? or optic?(3n)oedema? or ocular?(3n)edema? or ocular?(3n)oedema? or intraocular?(3n)edema? or intraocular?(3n)oedema? or maculopath? or microaneurysm? or micro(w)aneurysm? or microaneurism? or micro(w)aneurism? or retina?(3n)hemorrhag? or retina?(3n)haemorrhage? or retina?(3n)neovascular? or retina?(3n)leak or retina?(3n)perme? or eye?(3n)hemorrhag? or eye?(3n)haemorrhag? or eye?(3n)neovascular? or eye?(3n)leak or eye?(3n)perme? or ocular?(3n)hemorrhag? or ocular?(3n)haemorrhag? or ocular?(3n)neovascular? or ocular?(3n)leak? or ocular?(3n)perme? or ocular?(3n)bleed? or intraocular?(3n)hemorrhag? or intraocular?(3n)haemorrhag? or intraocular?(3n)neovascular? or intraocular?(3n)leak? or intraocular?(3n)perme? or intraocular?(3n)bleed? or optic?(3n)hemorrhag? or optic?(3n)haemorrhag? or optic?(3n)neovascular? or optic?(3n)leak? or optic?(3n)perme? or optic?(3n)bleed?

s s5 and s6 and s7

NTIS (National Technical Information Service) (US Department of Commerce) 1990 to September 2007

Searched 5 October 2007.

0 records were retrieved.

Each line searched separately:

color test diabetic [With all of the words]

color test diabetes [With all of the words]

color testing diabetic [With all of the words]

color testing diabetes [With all of the words]

color screen diabetic [With all of the words]

color screen diabetes [With all of the words]

color screening diabetic [With all of the words]

color screening diabetes [With all of the words]

colour test diabetic [With all of the words]

colour test diabetes [With all of the words]

colour testing diabetic [With all of the words]

colour testing diabetes [With all of the words]

colour screen diabetic [With all of the words]

colour screen diabetes [With all of the words]

colour screening diabetic [With all of the words]

colour screening diabetes [With all of the words]

color retinopathy diabetes [With all of the words]

color retinopathy diabetic [With all of the words]

colour retinopathy diabetes [With all of the words]

colour retinopathy diabetic [With all of the words]

ReFer (Research Findings electronic Register) (ReFer website) September 2007

Searched 5 October 2007.

0 records were retrieved.

Each line searched separately:

color AND diabetic AND retinopathy

color AND diabetes AND retinopathy

colour AND diabetic AND retinopathy

colour AND diabetes AND retinopathy

ClinicalTrials.gov (US National Library of Medicine) September 2007

Searched 5 October 2007.

0 records were retrieved.

Each line searched separately:

color, diabetic, retinopathy

color, diabetes, retinopathy

colour, diabetic, retinopathy

colour, diabetes, retinopathy

Current Controlled Trials (MetaRegister of Current Controlled Trials – mRCT) September 2007

Searched 5 October 2007.

0 records were retrieved.

Each line searched separately:

retinopathy AND colour

retinopathies AND colour

retinopathy AND color

retinopathies AND color

screening AND color

screen AND color

screening AND colour

screen AND colour

ClinicalStudyResults.org (ClinicalStudyResults website) September 2007

5 October 2007.

0 records were retrieved.

Each line searched separately:

color, diabetes, retinopathy

retinopathy

retinopathies

ClinicalTrialResults.org (Clinical Trial Results website) September 2007

5 October 2007.

0 records were retrieved.

Each line searched separately:

retinopathy

retinopathies

NHS EED (CRD internal databases) 1994 to September 2007

5 October 2007.

0 records were retrieved.

s (color$or colour$or vision or visual)(w3)(test$or screen$)

s (color$or colour$)(w3)(blind$or deficien$or defect$or loss or impair$)

s monochromatopsia or achromatopsia or deutan or protan or anomaloscop$or pseudoisochromatic$or hue$(w)discrimination$or lantern$or tritan or TCT

s ishihara$or kojima$or ohkuma$or matsubara$or dvorine$or cvtmet or hardy(w)rand(w)rittler$or AOHRR or AO(w)H(w)R(w)R or lanthony$or panel(w)d(w)15 or panel(w)d15 or d(w)15(w)panel or d15(w)panel or farnsworth$or (color$or colour$)(w)plate or mollon(w)reffin$or pease(w)allen$or giles(w)archer$or adams(w)desaturated or City(w)University or CUCVT or velhagen(w)pflugertrident$or contrast(w3)sensitivit$

s s1 OR s2 OR s3 OR s4

s diabet$or IDDM or NIDDM or T2DM

s s5 and s6

s retinopath$

s (optic$or ocular$or intraocular$or macula$or retina$)(w3)(edema$or oedema$)

s maculopath$or microaneurysm$or micro(w)aneurysm$or microaneurism$or micro(w)aneurism$or (ocular$or intraocular$or optic$or retina$or eye$)(w3)(hemorrhag$or haemorrhag$or neovascular$or leak or perme$or bleed$or neo(w)vascular$)

s s8 OR s9 OR s10

s s7 AND s11

HEED (Wiley online) September 2007

5 October 2007.

2 records were retrieved.

AX=‘color test’ within 3 OR ‘color testing’ within 3 OR ‘color tests’ within 3 OR ‘color screen’ within 3 OR ‘color screening’ within 3 OR ‘color screened’ within 3 OR ‘colour test’ within 3 OR ‘colour testing’ within 3 OR ‘colour tests’ within 3 OR ‘colour screen’ within 3 OR ‘colour screening’ within 3 OR ‘colour screened’ within 3

AX=‘vision test’ within 3 OR ‘vision testing’ within 3 OR ‘vision tests’ within 3 OR ‘vision screen’ within 3 OR ‘vision screening’ within 3 OR ‘vision screened’ within 3 OR ‘visual test’ within 3 OR ‘visual testing’ within 3 OR ‘visual tests’ within 3 OR ‘visual screen’ within 3 OR ‘visual screening’ within 3 OR ‘visual screened’ within 3

CS=1 or 2

AX=diabetes or diabetic or IDDM or NIDDM or T2DM

AX=retinopathy or retinopathies or (macular edema) or (macular oedema) or (retina edema) or (retina oedema) or (macular edemas) or (macular oedemas) or (retina edemas) or (retina oedemas) or maculopathy or microaneurysm or (micro aneurysm) or microaneurism or (micro aneurism) or (retina hemorrhage) or (retina haemorrhage) or (eye hemorrhage) or (eye haemorrhage)

CS=3 and 4 and 5

Internet sites searched

Websites were browsed (publications and research) and searched using a variety of combinations of the following terms: ‘colour vision screening’, ‘color vision screening’, ‘colour vision’, ‘color vision’, ‘colour screening’, ‘color screening’, ‘retinopathy’, ‘diabetes’, ‘diabetic’.

MedlinePlus: www.nlm.nih.gov/medlineplus/medlineplus.html

intute: www.intute.ac.uk/

National Library for Health (NLH) Diabetes Specialist Library: www.library.nhs.uk/diabetes/

National Library for Health (NLH) Screening Specialist Library: www.library.nhs.uk/screening/

National Screening Programme for Sight-threatening Diabetic Retinopathy: www.nscretinopathy.org.uk/

British Association for Retinal Screeners: www.eyescreening.org.uk/

Diabetes UK: www.diabetes.org.uk/

American Diabetic Association (US): www.diabetes.org/home.jsp

Royal College of Ophthalmologists: www.rcophth.ac.uk/

American Academy of Ophthalmology (US): www.aao.org/

College of Optometrists: www.college-optometrists.org/

Association of Optometrists: www.assoc-optometrists.org/

National Eye Institute (US): www.nei.nih.gov/

Conference proceedings

American Academy of Ophthalmology annual meeting (1999–2006) (www.aao.org/meetings/annual_meeting/) – Searchable archive available: posters 1999–2006; programs 2001–6. Searched using various combinations of the following terms: ‘color’, ‘colour’, ‘color vision’, ‘colour vision’ and ‘retinopathy’.

American Diabetic Association Annual Scientific Sessions (2003–7) (www.diabetes.org/for-health-professionals-and-scientists/profed.jsp) – Searchable archive available: 2003–7. Searched using various combinations of the following terms: ‘color vision’, ‘colour vision’, ‘retinopathy screening’ and ‘retinopathy testing’. Browsed results for potentially relevant abstracts.

European Association for the Study of Diabetes annual meeting (2001–7) (www.easd.org/) – Abstracts available in a variety of PDF, web page and searchable formats: 2001–7. Browsed/searched: ‘color’, ‘colour’, ‘retinopathy’.

Royal College of Ophthalmologists Annual Congress (2004–7) (www.rcophth.ac.uk/scientific/) – Abstracts available in PDF format: 2004–7. Browsed PDFs using search facility: ‘color’, ‘colour’, ‘retinopathy’.

Search alerts

Search alerts were set up in MEDLINE and EMBASE to run every time the databases were updated (usually weekly). The following strategy was used:

(colo?r vision and retinopathy).ti,ab.

Search alerts were also created in the following journals:

American Journal of Ophthalmology (http://archopht.ama-assn.org/) – Alert criteria: Title or Abstract: (color or colour) and vision and retinopathy.

British Journal of Ophthalmology (http://bjo.bmj.com/) – Alert criteria: Title or Abstract: (color or colour) and vision and retinopathy.

Clinical and Experimental Ophthalmology (www.blackwell-synergy.com/loi/ceo) – Alert criteria: Full text: (color or colour) and vision and retinopathy.

Diabetes (http://diabetes.diabetesjournals.org/) – Alert criteria: Anywhere: (color or colour) and vision and retinopathy.

Diabetes Care (http://care.diabetesjournals.org/) – Alert criteria: Anywhere: (color or colour) and vision and retinopathy

Diabetic Medicine (www.blackwell-synergy.com/toc/dme/0/0) – Alert criteria: Full text: (color or colour) and vision and retinopathy

Investigative Ophthalmology and Visual Science (www.iovs.org/) – Alert criteria: Title or Abstract: (color or colour) and vision and retinopathy.

Ophthalmology (www.sciencedirect.com/science/journal/01616420) – Alert criteria: Title-Abstr-Key: ((color or colour) and vision and retinopathy)

Diabetic retinopathy screening (economic/cost) search strategies