Arfken 1998.

Study characteristics
Methods	Study design: retrospective cohort longitudinal Location: USA Time period: unclear Outcome: PDR Outcome measurement: incidence and OR (95% CI)
Participants	Number of participants: 312 Diabetes type: T1D Inclusion criteria: T1D (defined as age of onset ≤ 40 years and continuous insulin usage); at least two visits with gradable fundus photographs; African‐American or white; duration of DM ≤ 16 years (relaxed for African‐American group) Exclusion criteria: haemoglobinopathies; PDR; evidence of treatment for PDR at baseline
Notes	Link to data extraction table, risk of bias assessment, and results: osf.io/krc53/?view_only=bf305ee63ef54bdca777f003087b567c
Item	Authors' judgement	Support for judgement
Study participation	Unclear	See ‘Notes’ above for link to risk of bias assessment
Study attrition	No	See ‘Notes’ above for link to risk of bias assessment
Prognostic factor measurement	Unclear	See ‘Notes’ above for link to risk of bias assessment
Outcome measurement	Yes	See ‘Notes’ above for link to risk of bias assessment
Adjustment for other prognostic factors	Unclear	See ‘Notes’ above for link to risk of bias assessment
Statistical analysis and reporting	Unclear	See ‘Notes’ above for link to risk of bias assessment