Reliability of retinal photography in the aement of retinal microvacular characteritic: the atherocleroi rik in communitie tudy

PURPOE: Retinal microvacular characteritic, a graded from retinal photography, have been hown to predict troke. We evaluated the reliability of retinal photographic grading in the Atherocleroi Rik in Communitie tudy. DEIGN: Cohort tudy. METHOD: Retinal photograph were taken of all ubject who attended the third Atherocleroi Rik in Communitie tudy examination (1993 to 1995). Thee were graded uing tandardized protocol. Focal retinal characteritic were graded uing a “light box” ytem. Generalized retinal arteriolar narrowing wa quantified from computer-aited meaurement of digitized photograph. Two ubtudie were conducted to invetigate the reliability of thee grading method. In the Individual Variability tudy, elected ubject (n = 206) had two retinal photograph taken on one day, and a further one or two photograph taken 3 week later. In the Grader Variability tudy, a tratified random ample of photograph had repeat retinal grading (n = 495 photograph for light box grading; n = 276 photograph for computer-aited grading). REULT: Reliability of the computer-aited quantification of generalized retinal arteriolar narrowing wa high in both tudie (reliability coefficient 0.64 to 0.69 for Individual Variability tudy, and 0.79 to 0.83 for the Grader Variability tudy). There wa more variability for focal abnormalitie graded uing the light box ytem. Variability for Individual Variability tudy (ame individual, repeat photograph) tended to be greater than for the Grader Variability tudy (ame photograph, repeat grading). CONCLUION: Retinal microvacular characteritic, epecially computer-aited quantification of generalized retinal arteriolar narrowing, can be acertained reliably by tandardized photographic grading method, upporting the validity of their aociation with cardiovacular dieae. However, thee characteritic appear to vary omewhat between eye and over time in a ingle individual.