Screening for diabetic retinopathy in the UK (all studies included people with type 1 and type 2 diabetes). Studies in alphabetical order by name of first author
| First author | Screening method | Screener | Number screened | Severity of retinopathy | Sensitivity % (95% CI) | Specificity % (95% CI if reported) | Comparison (”gold standard”) | Comments |
|---|---|---|---|---|---|---|---|---|
| Burnett15 | Ophthalmoscopy: no details given | Optometrists | 536 | Referable | 100 | 94 (90 to 98) | Ophthalmoscopy by ophthalmologist | Screeners (community optometrists) trained and accredited, paid £20 for each examination |
| Buxton8 | Direct ophthalmoscopy | GP | 2350 | Sight-threatening | 53 (44 to 62) | 91 (90 to 92) | Ophthalmoscopy by trained clinical assistant | Cost effectiveness studies based on same data |
| Optician | 307 | 48 (26 to 70) | 94 (92 to 97) | |||||
| Hospital doctor | 416 | 67 (50 to 84) | 96 (94 to 98) | |||||
| Polaroid camera, no mydriasis | Ophthalmologist in GP practice or hospital clinic, photos read by ophthalmologist | 2799 | Sight-threatening | 56 (49 to 72) | 97 (96 to 98) | 5% of photos unusable, 90% “assessable” | ||
| Forrest16 | Ophthalmoscopy | Diabetologist | 282 | Any | 51 (35 to 68) | 99 (97 to 100) | Five field stereoscopic fundus photography | Confidence intervals reported only for diabetologist, any retinopathy |
| Nurse | 50 | 99 | ||||||
| Diabetologist | Sight-threatening | 27 | 99 | |||||
| Nurse | 55 | 92 | ||||||
| Gibbins17 | 35mm camera, mydriasis | GP. | 143 | Any | 87 (66 to 97) | 77 (70 to 85) | Same photos assessed by ophthalmologist | Sensitivity based on “good quality” photos — 78% of total |
| Proliferative | 100 | 96 (92 to 99) | ||||||
| Gibbins18 | Direct ophthalmoscopy | GP | 613 in first phase of study, | Any | 63 (56 to 69) | 75 (70 to 80) | Photos assessed by trained graders | |
| Optician | 74 (67 to 81) | 80 (75 to 85) | ||||||
| GP | Sight-threatening | 66 (54 to 77) | 94 (91 to 96) | |||||
| Optician | 82 (68 to 92) | 90 (87 to 93) | ||||||
| 35mm camera, mydriasis | GP | 644 in second phase. | Any | 79 (74 to 85) | 73 (68 to 79) | Same photos assessed by trained graders | ||
| Optometrist | 88 (83 to 93) | 68 (62 to 74) | ||||||
| Diabetologist | 73 (66 to 79) | 93 (89 to 96) | ||||||
| GP | Sight-threatening | 87 (77 to 94) | 85 (81 to 88) | |||||
| Optometrist | 91 (79 to 87) | 83 (79 to 87) | ||||||
| Diabetologist | 89 (79 to 95) | 91 (88 to 94) | ||||||
| Harding19 | 35mm camera, mydriasis | Ophthalmological clinical assistant | 358 | Sight-threatening | 89 (80 to 98) | 86 (82 to 90) | Slit lamp biomicroscopy by retinal specialist | 3.75% of photos “unobtainable” |
| Direct ophthalmoscopy | Ophthalmologist | 65 (51 to 79) | 97 (95 to 99) | |||||
| O'Hare7 | Direct ophthalmoscopy | Optician | Referable | 73 | 93 | Ophthalmoscopy by ophthalmologist | Only opticians using both methods achieve BDA criteria | |
| Direct ophthalmoscopy plus photo with mydriasis. | GP | 517 | 60 | 98 | ||||
| Optician | 493 | 88 | 99 | |||||
| Taylor20 | Polaroid camera | District retinal screener | 197 | Any | 72 (66 to 78) | 88 (85 to 91) | Seven field stereo photography (118 patients, randomly selected) | Results for referable retinopathy consistently meet BDA criteria. |
| Digital camera | 534 | Referable | 90 (86 to 94) | 97 (95 to 99) | ||||
| Polaroid plus ophthalmoscopy | Unclear | Any | 74 (68 to 80) | 96 (94 to 98) | Patients preferred digital; 2.6% discomfort versus 17% with polaroid | |||
| Referable | 85 (80 to 90) | 98 (96 to 100) | ||||||
| Any | 92 (86 to 98) | 92 (86 to 98) | ||||||
| Referable | 95 (91 to 99) | 97 (95 to 99) | ||||||
| Williams21 | 35mm or polaroid camera, no mydriasis | Ophthalmological clinical assistant | 62 | Any | 96 (88 to 99) | 98 (87 to 100) | Ophthalmoscopy by ophthalmologist | Unusually high levels of accuracy—but a small study |