Background Population-based diabetes eye screening aims to reduce the occurrence of visual loss from diabetic eye disease. This study evaluated whether repeated non-attendance for diabetes eye screening is associated with risk of sight-threatening diabetic retinopathy (STDR).

Methods A population-based cohort study was conducted using diabetes eye screening appointments and results bare presented for N = 35,253 patients resident in three South London boroughs invited for screening between 2008 and 2011. Analysis examined associations between the number of years in which screening was missed and risk of sight-threatening diabetic retinopathy. We also describe uptake and socio-demographic and clinical associations with non-attendance.

Results There were 16,985 participants who were invited for screening in all four years, from 2008 to 2011, including 10,208 (60%) who were screened in every year, 6,564 (39%) who did not attend screening for up to three years, and 213 (1%) who did not attend screening in any year. Compared with participants who were screened annually, the adjusted relative odds of STDR were 2.28 ([95% CI 1.86, 2.78], p < 0.001) for participants who did not attend one year’s screening and 5.51 ([3.89, 7.80], p < 0.001) for participants who did not attend two year’s screening. In participants with mild non-proliferative retinopathy at their first screen, the adjusted relative odds of moderate or severe non-proliferative retinopathy or proliferative retinopathy at the next screen were 8.35 ([4.56, 15.26], p < 0.001) when one year of screening was missed and 24.08 ([11.52, 50.31], p < 0.001) when two years of screening was missed.

Conclusion Patients who do not attend diabetes eye screening are at increased risk of developing sight threatening diabetic retinopathy. This risk is greater after two years of non-attendance or when background retinopathy has already been diagnosed. Quality assurance in the screening programme should focus on repeat non-attenders who already have some evidence of retinopathy.