Skip to main content

Table 3 Risk of cardiovascular events by the deep-learning-based retinal CVD biomarker (Reti-CVD) in participants who had 7.5–10% QRISK3 score

From: Validation of a deep-learning-based retinal biomarker (Reti-CVD) in the prediction of cardiovascular disease: data from UK Biobank

QRISK3

Reti-CVDa

N

Cases

Person-years

Incidence rate at year 10 (95% CI)

Non-statin cohort (n = 45,473)

 ≥ 5 to < 7.5

 

7842

478

83,295

5.6 (5.1–6.1)

 ≥ 7.5 to < 10

Low

652

47

6868

7.2 (5.4–9.6)

 ≥ 7.5 to < 10

Moderate

4526

401

47,286

7.9 (7.1–8.8)

 ≥ 7.5 to < 10

High

518

62

5318

11.5 (8.9–15.0)

 ≥ 10 to < 12.5

 

3353

394

34,314

11.0 (9.9–12.2)

Total

 

16,891

1382

177,081

7.8 (7.4–8.2)

Stage 1 Hypertension cohort (n = 11,966)

 ≥ 5 to < 7.5

 

2622

154

27,891

5.5 (4.6–6.5)

 ≥ 7.5 to < 10

Low

230

16

2427

6.8 (4.1–11.3)

 ≥ 7.5 to < 10

Moderate

1525

143

15,821

8.3 (6.9–9.9)

 ≥ 7.5 to < 10

High

186

22

1903

11.5 (7.4–17.8)

 ≥ 10 to <1 2.5

 

974

115

9990

10.9 (9.0–13.3)

Total

 

5537

450

58,032

7.8 (7.1–8.5)

Middle–aged cohort (40 to 64 years, n = 38,941)

 ≥ 5 to < 7.5

 

7097

442

75,354

5.7 (5.2–6.3)

 ≥ 7.5 to < 10

Low

497

30

5291

5.8 (4.0–8.4)

 ≥ 7.5 to < 10

Moderate

2856

272

29,813

8.5 (7.5–9.7)

 ≥ 7.5 to < 10

High

287

43

2914

14.2 (10.3–19.5)

 ≥ 10 to < 12.5

 

1327

153

13,658

11.2 (9.5–13.2)

Total

 

12,064

940

127,030

7.4 (6.9–7.9)

  1. aReti-CVD only applied on borderline-QRISK3 groups. Incidence per 1000 person-years at year 10. CI confidence interval. N number at risk. Reti-CVD deep-learning-based retinal CVD biomarker