Abstract
Purpose :
To develop and validate a risk score for age-related macular degeneration (AMD), based on a simple self-administered questionnaire.
Methods :
Risk factors having shown the most consistent associations with AMD in the literature were included in the STARS (Simplified Théa AMD Risk-assessment Scale) questionnaire. Two studies were conducted using similar methodology, in Italy (127 participating ophthalmologists) and in France (80 participating ophthalmologists). Briefly, during one week, ophthalmologists invited all their patients aged 55 years or more to fill in the STARS questionnaire. Based on fundus examination, early AMD was defined by the presence of soft drusen and/or pigmentary abnormalities, and late AMD by the presence of geographic atrophy and/or neovascular AMD. A multivariate logistic regression for late AMD was first conducted on the Italian sample, including all risk factors. The risk scores were estimated from rounded values of the parameters of the logistic regression. The discrimination of AMD by the risk score was determined using ROC curves in the Italian (derivation) and French (validation) samples. Patients were classified as low risk for a score of 0 to 9, moderate risk for scores of 10 to 19 and high risk for scores of 20 or more.
Results :
The Italian sample consisted of 12,639 patients, of whom 6061 with early and 1483 with late AMD. The French sample consisted of 6837 patients, of whom 1250 with early and 687 with late AMD. All 13 risk factors included in the STARS questionnaire (age, gender, ethnicity, family history of AMD, smoking, BMI, history of myocardial infarction, history of hypertension, history of atherosclerosis, history of hypercholesterolemia, iris colour, myopia, hyperopia) showed significant associations with AMD in the Italian sample. The area under the ROC curve for the STARS score derived from the multivariate logistic regression in the Italian sample was 0.78 in the Italian sample and 0.72 in the French sample. In both samples, less than 10 % of patients without AMD were classified at high risk, and less than 13 % of late AMD cases were classified as low risk, with a more intermediate situation in early AMD cases (Figure).
Conclusions :
This 13-item simple, self-assessed score shows good discrimination, and high feasibility in two large European samples and may allow in the future a simple and fast identification of patients at risk of AMD.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.