A quantitative system to evaluate diabetic retinopathy from fundus photographs.

PURPOSE: To evaluate a quantitative system to measure the early lesions of diabetic retinopathy seen in stereoscopic fundus photographs. METHODS: Using a quantitative classification system, photographs of 4657 eyes (7 stereo pairs of 35-mm slides per eye) were scored for 16 diabetic lesions. A single severity level (identical to the ETDRS Interim Scale) was calculated for each eye. The reliability of this technique, and its reproducibility by independent examiners, was evaluated for individual lesions and severity levels using percent agreement, kappa, and weighted kappa statistics. RESULTS: This quantitative technique demonstrated an "almost perfect" agreement (weighted kappa > or = 0.810) on all but one lesion by independent observers. For the severity levels, there was a 95.7% perfect agreement (kappa = 0.9428). The reproducibility of agreement over time was "almost perfect" on all but four lesions; with 88% perfect agreement (kappa = 0.8394) for severity levels. CONCLUSIONS: When used to evaluate the early lesions of diabetic retinopathy, the Vanderbilt Classification System is highly reliable between graders and over time. This system can gather quantitative data and evaluate incremental changes in an accurate, reproducible manner.

The topographic distribution of the first sites of diabetic retinal neovascularization.

PURPOSE: To examine the topographic distribution of the origin of diabetic retinal neovascularization. METHODS: The eyes of 3,121 patients with background diabetic retinopathy were investigated. These patients were volunteers in systemic medical therapy experiments. Color stereo photographs were obtained annually. The first retinal neovascularization sites were identified and the distances from the optic nerve measured. RESULTS: In 1 year, neovascularization originated in 282 eyes. The superotemporal quadrant, at 6 mm from the optic disk, was the most frequent initial site. CONCLUSIONS: The first retinal neovascularization sites cluster around specific anatomic foci. This information should influence retinopathy monitoring protocols.

The Vanderbilt Classification System in the evaluation of diabetic retinopathy patients treated with Alredase.

BACKGROUND: The Vanderbilt Classification System is a quantitative method of measuring features detected in diabetic retinopathy photographs. It does not require comparisons to preexisting standard photographs. This is the first report of the application of this system to a large-scale, multiple-medical-center drug trial. METHODS: This was a prospective, randomized, double-masked, placebo-controlled study that involved 74 medical centers. There were 3,679 out-patients followed for more that 4 years, with some observed for over 9 years. The Vanderbilt Classification System generated patient data for the Early Treatment Diabetic Retinopathy Study (ETDRS) and the Diabetes Control and Complication Trial (DCCT) scales. The deterioration rate was one variable used to assess drug effect. A comprehensive Quality Assurance Program evaluated intergrader and intragrader reliability. RESULTS: Target values for reliability and reproducibility were met or exceeded on all measures of agreement between photo readers and over time. Kappa statistics were 0.610 or greater, with most weighted kappa values greater than 0.810. This represents "almost perfect agreement" and compares favorably with previous reports from the ETDRS and DCCT. CONCLUSION: Diabetic retinopathy can be evaluated in a reliable and reproducible manner with the VCS. The VCS is unique in that it produces a quantitative analysis of retinal lesions. Subtle variations that might be influenced by systemic medications can be measured accurately with this technique.

Screening for amblyogenic factors using a volunteer lay network and the MTI photoscreener. Initial results from 15,000 preschool children in a statewide effort.

PURPOSE: To describe the results from a statewide preschool vision screening program using the MTI PhotoScreener (Medical Technology and Innovations, Inc., Cedar Falls, IA). DESIGN: Cross-sectional study. PARTICIPANTS: A total of 15,059 children aged 6 to 47 months enrolled in childcare and preschool settings throughout the state of Tennessee. METHODS: Volunteers from local Lions Clubs took photoscreening photographs of children in a statewide effort. Photographs were interpreted at the Vanderbilt Ophthalmic Photography Reading Center using predetermined criteria. Children who failed the screening were referred to community ophthalmologists or optometrists who performed a comprehensive evaluation and forwarded the results to the authors. MAIN OUTCOME MEASURES: Referral rate, unreadable rate, and predictive value positive (PVP). RESULTS: During the 2 years of the screening program, 15,059 children were screened in 850 screenings. The screening referred 1013 children (6.7%), and 704 photographs (4.7%) were unreadable. Children who failed the screening had a significant abnormality (strabismus, anisometropia, high hypermetropia, high astigmatism, or high myopia) in 320 of the 531 cases where adequate follow-up results were reported. The PVP ranged from 84% when a diagnosis of strabismus was suggested by the photoscreen reading to 41% for astigmatism. Despite intense attention to follow-up, many children who failed the screening never received a formal eye examination. CONCLUSIONS: The MTI PhotoScreener can be used by volunteers to screen preschool children and can have a high PVP in organized settings, provided that meticulous attention is paid to photograph interpretation and quality control. The PVP of the MTI PhotoScreener depends on the diagnosis suggested when the photograph is read. Significant obstacles exist in obtaining care for those who fail screening.