Objective: The aim of this study was to compare entoptic perimetry, using conventional television, to Amsler grid and patient-reported visual loss for the detection of functional diabetic maculopathy and macular edema.
Design: Observational case series.
Participants: A single eye from each of 104 consecutive patients with diabetes in an academic retina clinic.
Intervention: Each eye was screened by Amsler grid, entoptic perimetry, and Humphrey 10-2 threshold visual field testing (HVF 10-2; Humphrey Instruments Inc., San Leandro, CA) in random order. Eyes were then examined clinically.
Main outcome measures: The presence or absence of new visual decline since the patient's last clinical examination, the presence or absence of central visual field abnormalities using an Amsler grid, entoptic perimetry, HVF 10-2, and the presence or absence of clinically significant macular edema (CSME).
Results: The sensitivities and specificities for the detection of central diabetic scotomas as evidenced by HVF 10-2 abnormalities were: subjective impression, 31 of 90 eyes (34.4%) and 11 of 14 eyes (78.6%); Amsler grid, 29 of 90 eyes (32.2%) and 13 of 14 eyes (92.9%); and entoptic perimetry, 58 of 90 eyes (64.4%) and 11 of 14 eyes (78.6%). Entoptic perimetry was statistically more sensitive than both subjective impression (P < 0.001) and Amsler grid (P < 0.001), but the specificities were statistically indistinguishable. The sensitivities and specificities for the detection of CSME were: subjective impression, 6 of 24 eyes (25.0%) and 52 of 80 eyes (65.0%); Amsler grid, 9 of 24 eyes (37.5%) and 59 of 80 eyes (73.8%); and entoptic perimetry, 17 of 24 eyes (70.8%) and 44 of 80 (55.0%) eyes. These results are also statistically significant, with entoptic perimetry being more sensitive and less specific than both subjective impression (P = 0.007 and P = 0.011, respectively) and Amsler grid (P = 0.008 and P < 0.001, respectively) in this subset of patients.
Conclusions: Entoptic perimetry is 87% more sensitive than the subjective impression of visual decline (P < 0.001) and 100% more sensitive than Amsler grid (P < 0.001) for the detection of central scotomas in diabetic patients. For the detection of CSME, entoptic perimetry is 183% more sensitive than subjective impression (P = 0.007) and 89% more sensitive than Amsler grid (P = 0.008). Hence, entoptic perimetry, performed using conventional television, has the potential to be an effective, inexpensive, and widespread adjunct to surveillance examinations for the early detection of diabetic maculopathy.