October 29, 2026

ACCESS2: High screening completion, with diagnostic trade-offs

Conceptual illustration separating screening completion from diagnostic performance. — Original editorial illustration, created for this article. Not patient imagery or a study result.

Risa M. Wolf and colleagues’ ACCESS2 study appeared in Nature Communications on 23 October 2026 as a peer-reviewed article in press.

This prospective, nonrandomized study enrolled 397 young people at two Baltimore sites. It assessed diabetic eye screening, not prevention of vision loss.

ACCESS2 findings. Source: Results and Tables 1–3.
Measure	Result
Screening completed	392/397 (98.7%)
White non-Hispanic / non-White or Hispanic groups	99.5% / 98.0%; prespecified equivalence criterion met
Diagnostic analysis	390 participants; eight reference-positive cases
Sensitivity / specificity for more-than-mild retinopathy	100% (95% CI 67.6–100%) / 76% (95% CI 71.4–79.9%)
False positives against reading-center grading	92 among 390 participants

What this does not establish

Few positive cases make sensitivity uncertain. Screening was free, and participants already attended care. Completion equivalence does not establish equal diagnostic performance or improved vision.

The authors report that the device was not labeled for people under 22; a retina specialist overread every image during enrollment. Commercial interests are disclosed in the paper.