The use of statistical methodology to determine the accuracy of grading within a diabetic retinopathy screening programme.

AIMS: We aimed to use longitudinal data from an established screening programme with good quality assurance and quality control procedures and a stable well-trained workforce to determine the accuracy of grading in diabetic retinopathy screening. METHODS: We used a continuous time-hidden Markov model with five states to estimate the probability of true progression or regression of retinopathy and the conditional probability of an observed grade given the true grade (misclassification). The true stage of retinopathy was modelled as a function of the duration of diabetes and HbA1c . RESULTS: The modelling dataset consisted of 65 839 grades from 14 187 people. The median number [interquartile range (IQR)] of examinations was 5 (3, 6) and the median (IQR) interval between examinations was 1.04 (0.99, 1.17) years. In total, 14 227 grades (21.6%) were estimated as being misclassified, 10 592 (16.1%) represented over-grading and 3635 (5.5%) represented under-grading. There were 1935 (2.9%) misclassified referrals, 1305 were false-positive results (2.2%) and 630 were false-negative results (1.0%). Misclassification of background diabetic retinopathy as no detectable retinopathy was common (3.4% of all grades) but rarely preceded referable maculopathy or retinopathy. CONCLUSION: Misclassification between lower grades of retinopathy is not uncommon but is unlikely to lead to significant delays in referring people for sight-threatening retinopathy.

Delay in diabetic retinopathy screening increases the rate of detection of referable diabetic retinopathy.

AIMS: To assess whether there is a relationship between delay in retinopathy screening after diagnosis of type 2 diabetes and level of retinopathy detected. METHODS: Patients were referred from 88 primary care practices to an English National Health Service diabetic eye screening programme. Data for screened patients were extracted from the primary care databases using semi-automated data collection algorithms supplemented by validation processes. The programme uses two-field mydriatic digital photographs graded by a quality assured team. RESULTS: Data were available for 8183 screened patients with diabetes newly diagnosed in 2005, 2006 or 2007. Only 163 with type 1 diabetes were identified and were insufficient for analysis. Data were available for 8020 with newly diagnosed type 2 diabetes. Of these, 3569 were screened within 6 months, 2361 between 6 and 11 months, 1058 between 12 and 17 months, 366 between 18 and 23 months, 428 between 24 and 35 months, and 238 at 3 years or more after diagnosis. There were 5416 (67.5%) graded with no retinopathy, 1629 (20.3%) with background retinopathy in one eye, 753 (9.4%) with background retinopathy in both eyes and 222 (2.8%) had referable diabetic retinopathy. There was a significant trend (P = 0.0004) relating time from diagnosis to screening detecting worsening retinopathy. Of those screened within 6 months of diagnosis, 2.3% had referable retinopathy and, 3 years or more after diagnosis, 4.2% had referable retinopathy. CONCLUSIONS: The rate of detection of referable diabetic retinopathy is elevated in those who were not screened promptly after diagnosis of type 2 diabetes.

Updating diabetic retinopathy screening lists using automatic extraction from GP patient records.

OBJECTIVES: Diabetic Retinopathy screening services aim to reduce the risk of sight loss amongst patients with diabetes. The rising incidence of diabetes in England and the operational need to ensure the accuracy and timeliness of screening lists led to a pilot study of electronic extraction of data from primary care. This study aimed to evaluate the effectiveness of updating the single collated list of patients eligible for diabetic eye screening using extracts from electronic patient records in primary care. SETTING AND METHODS: The Gloucestershire Diabetic Eye Screening Programme (GDESP) provides screening for 85 General Practices in the county. Of these, 54 using Egton Medical Information Systems (EMIS) practice management system software agreed to participate in this study. The screening list held in 2009 by the Gloucestershire DESP of 14,209 patients known to have diabetes was audited against a list created with automatic extraction from General Practice records of patients marked with the diabetes Read Code C10. Those subsequently screened and referred to the Hospital Eye service were followed up. RESULTS: The Gloucestershire DESP manual list covering the 54 EMIS practices comprised 14,771 people with diabetes. The audit process identified an additional 709 (4.8%) patients coded C10, including 23 diagnosed more than 5 years ago, and 20 patients under the age of 20 who were diagnosed more than a year ago. CONCLUSION: Automatic extraction of data from General Practice identified 709 patients coded as having diabetes not previously known to the Gloucestershire DESP.

Quality assurance for diabetic retinopathy telescreening.

AIMS: TOSCA was an EU-Commission supported international research project designed to develop telescreening services in diabetic retinopathy and glaucoma. This paper describes the quality assurance methods developed for the diabetic retinopathy telescreening service within the TOSCA project. SETTING: The study was performed in 1895 patients with diabetes between 2000 and 2002 at diabetic retinopathy screening sites in five European countries. Data were analysed centrally. METHODS: Patients attending each clinic's diabetic retinopathy screening service received standardized retinal photography. The images and associated data were transferred electronically to a remote location for grading. Each photographer uploading images and each grader downloading images for assessment was controlled by a systematic quality management approach. The quality assurance measures defined were image quality, intragrader reliability. A cockpit chart was developed for the management and presentation of relevant results and quality measures. For the intragrader reliability tests, 10% of the images were processed for a second grading. An algorithm for calculating differences between repeated gradings was developed. RESULTS: The assessment of image quality for the different sites showed that only 0-0.7% were unassessable. One hundred per cent agreement for both gradings was achieved in 50-85% of graded cases, depending on site and grader, and an agreement better than 95% in 71-100% of cases. CONCLUSIONS: A telemedicine-supported quality assurance process is practical and advantageous. The cockpit charts have proven to be useful tools when monitoring the performance of a telescreening service. Grader feedback showed high satisfaction with the quality assurance process.

Comparison of two reference standards in validating two field mydriatic digital photography as a method of screening for diabetic retinopathy.

AIM: To compare two reference standards when evaluating a method of screening for referable diabetic retinopathy. METHOD: Clinics at Oxford and Norwich Hospitals were used in a two centre prospective study of 239 people with diabetes receiving an ophthalmologist's examination using slit lamp biomicroscopy, seven field 35 mm stereophotography and two field mydriatic digital photography. Patients were selected from those attending clinics when the ophthalmologist and ophthalmic photographer were able to attend. The main outcome measures were the detection of referable diabetic retinopathy as defined by the Gloucestershire adaptation of the European Working Party guidelines. RESULTS: In comparison with seven field stereophotography, the ophthalmologist's examination gave a sensitivity of 87.4% (confidence interval 83.5 to 91.5), a specificity of 94.9% (91.5 to 98.3), and a kappa statistic of 0.80. Two field mydriatic digital photography gave a sensitivity of 80.2% (75.2 to 85.2), specificity of 96.2% (93.2 to 99.2), and a kappa statistic of 0.73. In comparison with the ophthalmologist's examination, two field mydriatic digital photography gave a sensitivity of 82.8% (78.0 to 87.6), specificity of 92.9% (89.6 to 96.2), and a kappa statistic of 0.76. Seven field stereo gave a sensitivity of 96.4% (94.0 to 98.8), a specificity of 82.9% (77.4 to 88.4), and a kappa statistic of 0.80. 15.3% of seven field sets, 1.5% of the two field digital photographs, and none of the ophthalmologist's examinations were ungradeable. CONCLUSION: An ophthalmologist's examination compares favourably with seven field stereophotography, and two field digital photography performs well against both reference standards.

The effectiveness of screening for diabetic retinopathy by digital imaging photography and technician ophthalmoscopy.

AIMS: To evaluate the introduction of a community-based non-mydriatic and mydriatic digital photographic screening programme by measuring the sensitivity and specificity compared with a reference standard and assessing the added value of technician direct ophthalmoscopy. METHODS: Study patients had one-field, non-mydriatic, 45 degrees digital imaging photography prior to mydriatic two-field digital imaging photography followed by technician ophthalmoscopy. Of these patients, 1549 were then examined by an experienced ophthalmologist using slit lamp biomicroscopy as a reference standard. The setting was general practices in Gloucestershire. Patients were selected by randomizing groups of patients (from within individual general practices) and 3611 patients were included in the study. Patients for reference standard examination were recruited from groups of patients on days when the ophthalmologist was able to attend. The main outcome measure was detection of referable diabetic retinopathy (DR) as defined by the Gloucestershire adaptation of the European Working Party guidelines. RESULTS: For mydriatic digital photography, the sensitivity was 87.8%, specificity was 86.1% and technical failure rate was 3.7%. Technician ophthalmoscopy did not alter these figures. For non-mydriatic photography, the sensitivity was 86.0%, specificity was 76.7% and technical failure rate was 19.7%. CONCLUSIONS: Two-field mydriatic digital photography is an effective method of screening for referable diabetic retinopathy. Non-mydriatic digital photography has an unacceptable technical failure rate and low specificity.

Relationship between the severity of retinopathy and progression to photocoagulation in patients with Type 2 diabetes mellitus in the UKPDS (UKPDS 52).

AIM: to establish the degree to which the severity of retinopathy determines the risk for the need for subsequent photocoagulation in those with newly diagnosed Type 2 diabetes mellitus. METHODS: Of 5102 patients entered into the UK Prospective Diabetes Study (UKPDS), 3709 had good quality retinal photographs that could be graded at entry. They were followed until the end of the study or until lost to follow-up, or until they received photocoagulation. Retinopathy severity was categorized as no retinopathy, microaneurysms (MA) only in one eye, MA in both eyes or more severe retinopathy features. The risk of photocoagulation was assessed in relation to severity of retinopathy at baseline, 3 and 6 years. RESULTS: Of the 3709 patients assessed at entry to the UKPDS, 2316 had no retinopathy. Of these 0.2% needed photocoagulation at 3 years, 1.1% at 6 years and 2.6% at 9 years. Those with MA in one eye only (n = 708) were similar, with 0%, 1.9% and 4.7% needing photocoagulation by 3, 6 and 9 years, respectively. Amongst those who had more retinopathy features at entry (n = 509), 15.3% required photocoagulation by 3 years, and 31.9% by 9 years. When those without retinopathy at 6 years (n = 1579) were examined 3 and 6 years later (9 and 12 years after diagnosis), 0.1% and 1.8% required photocoagulation. Those with more severe retinopathy (n = 775) needed earlier treatment, 6.6% after 3 years and 13.3% after 9 years. The commonest indication for laser therapy was maculopathy, but those with more severe retinopathy were more likely to be treated for proliferative retinopathy and to need both eyes treated. CONCLUSION: Few type 2 diabetic patients without retinopathy progress to photocoagulation in the following 3-6 years, while patients with more severe retinopathy lesions need to be monitored closely.

UKPDS 50: risk factors for incidence and progression of retinopathy in Type II diabetes over 6 years from diagnosis.

AIMS/HYPOTHESIS: To determine risk factors related to the incidence and progression of diabetic retinopathy over 6 years from diagnosis of Type II (non-insulin-dependent) diabetes mellitus. METHODS: This report describes 1919 patients from within the United Kingdom Prospective Diabetes Study (UKPDS), with retinal photographs taken at diagnosis and 6 years later and with complete data available. Photographs were centrally graded for lesions of diabetic retinopathy using the modified Early Treatment of Diabetic Retinopathy Study Final scale. Risk factors were assessed after 3 months diet from the time of diagnosis of diabetes. Patients were seen every 3 months in a hospital setting. Biochemical measurements were done by a central laboratory. End points of vitreous haemorrhage and photocagulation were confirmed by independent adjudication of systematically collected clinical data. The main outcome measures were incidence and progression of retinopathy defined as a two-step Early Treatment of Diabetic Retinopathy Study (ETDRS) final scale change. RESULTS: Of the 1919 patients, 1216 (63 %) had no retinopathy at diagnosis. By 6 years, 22 % of these had developed retinopathy, that is microaneurysms in both eyes or worse. In the 703 (37 %) patients with retinopathy at diagnosis, 29 % progressed by two scale steps or more. Development of retinopathy (incidence) was strongly associated with baseline glycaemia, glycaemic exposure over 6 years, higher blood pressure and with not smoking. In those who already had retinopathy, progression was associated with older age, male sex, hyperglycaemia (as evidenced by a higher HbA1c) and with not smoking. CONCLUSION/INTERPRETATION: The findings re-emphasise the need for good glycaemic control and assiduous treatment of hypertension if diabetic retinopathy is to be minimised.

Markers of insulin resistance are strong risk factors for retinopathy incidence in type 1 diabetes.

OBJECTIVE: To determine the incidence of retinopathy and the relative importance of its risk factors in type 1 diabetes. RESEARCH DESIGN AND METHODS: This is a 7.3-year follow-up of 764 of 1,215 (63%) people with type 1 diabetes across Europe, aged 15-60 years at baseline with no retinopathy (the EURODIAB Prospective Complications Study). Retinal photographs were taken at baseline and follow-up and risk factors were assessed to a standard protocol. RESULTS: Retinopathy incidence was 56% (429/764, 95% CI 52-59%). Key risk factors included diabetes duration and glycemic control. We found no evidence of a threshold effect for HbA1c on retinopathy incidence. Univariate associations were observed between incidence and albumin excretion rate, cholesterol, triglyceride, fibrinogen, von Willebrand factor, gamma-glutamyltransferase, waist-to-hip ratio, and insulin dose. No associations were observed for blood pressure, cardiovascular disease, or smoking. Independent risk factors, as assessed by standardized regression effects, were HbA1C (1.93, P = 0.0001), duration (1.32, P = 0.008), waist-to-hip ratio (1.32, P = 0.01), and fasting triglyceride (1.24, P = 0.04). CONCLUSIONS: Retinopathy incidence in type 1 diabetes remains high. Key risk factors include diabetes duration and glycemic control, with no evidence of a threshold for the latter. Other independent risk factors, such as waist-to-hip ratio and triglyceride levels, both markers of insulin resistance, were strongly related to incidence.

Microaneurysms in the development of diabetic retinopathy (UKPDS 42). UK Prospective Diabetes Study Group.

AIMS/HYPOTHESIS: To determine whether microaneurysms, in the absence of other lesions, have a predictive role in the progression of diabetic retinopathy in Type II (non-insulin-dependent) diabetes mellitus. METHODS: Retinal photographs taken at diagnosis in patients participating in the United Kingdom Prospective Diabetes Study, and thereafter at 3 yearly intervals, were assessed using a modified Early Treatment of Diabetic Retinopathy grading system for lesions of diabetic retinopathy and end points of vitreous haemorrhage and photocoagulation. The number of microaneurysms in each eye was recorded. RESULTS: The changes between diagnosis and later photographs were analysed in 2424 patients at 6 years, 1236 at 9 years and 414 at 12 years. Of the 2424 patients studied in the 6 year cohort 1809 had either no retinopathy or microaneurysms only at entry. In these patients the presence of microaneurysms alone and also the number of microaneurysms had a high predictive value for worsening retinopathy at 3, 6, 9, and 12 years after entry into the study (e. g. at 6 years chi(2) for trend = 75 on 1 df, p < 0.001). The predictive value of the presence or absence of microaneurysms and their number at 3 years from diagnosis and subsequent worsening retinopathy was similar to that at entry. CONCLUSION/INTERPRETATION: Microaneurysms are important lesions of diabetic retinopathy and even one or two microaneurysms in an eye should not be regarded as unimportant.

A comparison of digital retinal images and 35 mm colour transparencies in detecting and grading diabetic retinopathy.

We compared digital retinal images and 35 mm colour transparencies taken with the Canon CR5 retinal camera for the detection and grading of diabetic retinopathy in a clinical setting, in a randomized, blinded study of diabetic patients with a spectrum of severity of diabetic retinopathy. Forty patients were photographed, giving a total of 75 eyes including non-diabetic eyes as controls. Images were graded according to the validated European guidelines. There was exact agreement between grades obtained from both the 2 field 45 degrees 35 mm colour transparencies and digital images in 93.3% (70/75) of eyes, with Cohen's Kappa statistic for the comparison being 0.92. Overall, when grading from the digital images 5.3% (4/75) eyes were undergraded with three cases of sight threatening diabetic retinopathy (STDR) graded as non-sight threatening (NSTDR) (3/48, 6.3%). One eye was overgraded (1/75, 1.3%). Two of the three cases of STDR undergraded as NSTDR had small numbers of intra-retinal microvascular abnormalities (IRMA) discernible on the colour transparencies but which were not visible from the digital image. The third had multiple small cotton wool spots graded as laser photocoagulation scars from the digital images. In conclusion there is good to excellent agreement between retinopathy grades using the Canon CR5 digital retinal imaging system compared to 35 mm colour transparencies.

United Kingdom Prospective Diabetes Study, 30: diabetic retinopathy at diagnosis of non-insulin-dependent diabetes mellitus and associated risk factors.

OBJECTIVES: To report on the prevalence of retinopathy in patients with newly diagnosed non-insulin-dependent diabetes mellitus (NIDDM) and to evaluate the relationship of retinopathy to clinical and biochemical variables. DESIGN: A multicenter, randomized, controlled clinical study of therapy in patients with NIDDM. SETTING AND PATIENTS: Patients were part of the United Kingdom Prospective Diabetes Study, a 23-center study of 2964 white patients who had both eyes photographed and assessed. OUTCOME MEASURES: The presence and severity of diabetic retinopathy were evaluated by sex, and the relationship of retinopathy to medical and biochemical parameters was assessed. RESULTS: Retinopathy, defined as microaneurysms or worse lesions in at least 1 eye, was present in 39% of men and 35% of women. Marked retinopathy with cotton wool spots or intraretinal microvascular abnormalities was present in 8% of men and 4% of women. The severity of retinopathy was related in both sexes to higher fasting plasma glucose levels, higher systolic and diastolic blood pressure, lower serum insulin levels, and reduced beta-cell function. In addition, in men, increased alcohol consumption was related to increased severity of retinopathy, while leaner women had more severe eye lesions. Visual acuity was normal in most patients, but in men there was a trend for those with more severe retinal lesions to have worse visual acuity. CONCLUSIONS: Diabetic retinopathy is common in patients with newly diagnosed NIDDM. Careful ophthalmic assessment at diagnosis is important.

The accurate assessment of changes in retinal vessel diameter using multiple frame electrocardiograph synchronised fundus photography.

PURPOSE: Detection and precise quantification of changes in retinal vessel diameter by image analysis techniques is important in a number of fields of research. The retinal vessels have been shown to exhibit pulse related changes in diameter. These may need to be taken into account when studying diameter changes due to other causes. This study examined the effect of using multiple fundus photographs with and without electrocardiographic synchronisation on the size and statistical significance of changes in mean retinal vessel diameter. METHODS: Twelve fundus photographs spaced throughout the cardiac cycle by electrocardiographic synchronisation were taken in 10 normal volunteers: (a) at rest, (b) during isometric exercise, and (c) during oxygen inhalation. Vessel diameters were measured using a computer assisted image analysis system. Subsequently smaller sample sizes, with and without electrocardiograph synchronisation were modeled from the available data. RESULTS: With a group of ten subjects six or more electrocardiograph synchronised photographs enable reliable detection of small diameter changes (1.4%) induced by isometric exercises while other methods either failed to detect change or were unreliable at doing so. With six subjects twelve synchronised photographs were required to reliably detect a change of the same magnitude. Larger diameter changes (5.4%) were detected by any method including a single unsynchronised photograph. CONCLUSIONS: Multiple frame electrocardiograph synchronized fundus photography permits more accurate detection of small changes in retinal vessel diameter.

Methodology for retinal photography and assessment of diabetic retinopathy: the EURODIAB IDDM complications study.

We present the methodology for 45 degrees retinal photography and detail the development, application and validation of a new system of 45 degrees field grading standards for the assessment of diabetic retinopathy. The systems were developed for the EURODIAB IDDM Complications Study, part of a European Community funded Concerted Action Programme into the epidemiology and prevention of diabetes (EURODIAB). Assessment of diabetic retinopathy was carried out centrally by a trained reader of colour retinal photographs using the newly-developed system. The system proved to be acceptably accurate, repeatable repeatable and relatively simple to apply. It compared well with the recognised 'gold standard' 7-field 30 degrees stereo photography (assessed using a modified Airlie House classification scheme), against which the new system was validated in a series of 48 eyes. Selection was as a stratified random sample based on clinical retinopathy status: 5, no retinopathy; 25, non-proliferative retinopathy; 16, proliferative or photocoagulated; plus 2, eyes with potentially confounding lesions (vein occlusion). Simple presence of retinal lesions was correctly detected by both systems in 43 of the 48 eyes, giving 100% agreement on detection. Both systems correctly identified the two known cases of confounding vein occlusion. In eyes with diabetic retinopathy (n = 41), when severity was expressed in three groups: mild background, moderate/severe background and proliferative/photocoagulated, at least one grader (out of five) using the new system matched the verified results in 38 out of 31 (93%) eyes and three or more graders matched in 31 (76%) eyes.(ABSTRACT TRUNCATED AT 250 WORDS)

Evaluation of a non-mydriatic camera in Indian and European diabetic patients.

The value of the Canon CR3-45NM non-mydriatic fundus camera in screening for diabetic retinopathy has been assessed in two ethnic groups, namely, 45 Indian and 40 European diabetic patients. There was 72% agreement between ophthalmoscopy by an ophthalmologist and the photographic assessment, and 100% agreement on clinically important lesions requiring treatment. However, peripheral retinal lesions, seen on ophthalmoscopy but lying outside the photographic field, occurred in 4% of eyes. 6% of photographs were totally unassessable and an additional 12% were only partially assessable. The prevalence and severity of retinopathy was similar in Indian and European patients. The advantages and disadvantages of the non-mydriatic camera in screening for diabetic retinopathy are discussed.

Diabetic retinopathy: II. Assessment of severity and progression from fluorescein angiograms.

A quantitative method for assessing the severity of retinopathy from fluorescein angiograms was applied to a group of patients starting continuous subcutaneous insulin infusion. There was a highly significant correlation between the area of nonperfusion as measured in the 20 degrees field centered on the right macula, and the mean retinopathy score derived by a standard method from a color retinal photographic survey. In those subjects who deteriorated more, there was an increase in the nonperfused areas, whereas the foveal avascular zone remained unchanged.

Computer-aided digitisation of fundus photographs.

A method is described for the analysis of standard retinal colour photographs with the facility to compare accurately photographs taken serially, thus yielding information on the change in the retinal disease over a period of several years. The reproducibility of placement of a microaneurysm on the same photograph by the same observer is equivalent to +/- 6.6 microns at the retina and inter-observer variation is equivalent to +/- 7.2 micron at the retina. Overall accuracy of placement of a point is 10 micron at the retina.