Author: Claude (AI), from PubMed abstracts · Reviewed by: Dr Mina Abdelmseih · Reviewed: 2026-11-15 · Review due: 2027-11-15

Educational content only. Not personalised medical advice. Clinical decisions remain the responsibility of the treating clinician.

Retinopathy of prematurity

Retinopathy of prematurity (ROP) is a vision-threatening vasoproliferative disease of preterm infants [S3] [S2] and a leading cause of potentially avertable childhood blindness [S4]. Its rising prevalence in low- and middle-income countries, the "third epidemic", has increased the need to screen at-risk infants [S7]. Timely screening and treatment with laser or intravitreal anti-VEGF drugs are the core of care [S12].

Key points - Risk rises steeply with prematurity: each week less of gestational age carried a median 1.4 times the odds of ROP, and each 100 g less of birth weight 1.8 times [S9]. - Laser photocoagulation and intravitreal anti-VEGF injection are the standard treatments [S12]; anti-VEGF was not inferior to laser for retreatment in a network meta-analysis and gave a better refractive profile [S12]. - Anti-VEGF monotherapy does not reduce retinal detachment overall and may increase recurrence needing retreatment in zone II disease (Cochrane, low-quality evidence) [S2]. - Telemedicine, weight-gain algorithms and artificial intelligence can support screening in settings with few specialists, but AI is not yet sufficient as a sole diagnostic tool [S7] [S8] [S11].

Evidence last checked: 2026-10-17 (PubMed abstracts).

Red flags

• Treatment-requiring (type 1) ROP: zone I any stage with plus disease, zone I stage 3 with or without plus disease, or zone II stage 2 or 3 with plus disease [S2].
• Aggressive posterior ROP was included in treatment trials as requiring treatment [S10].
• Recurrence after anti-VEGF: recurrence risk was higher in eyes that received bevacizumab in eye-level analyses [S2], and lower doses were associated with shorter times to retreatment [S12]. Infants treated with anti-VEGF need continued follow-up.
• The most immature infants: at 23 weeks' gestation the risk of any, severe and treatment-requiring ROP was 66.5%, 40.3% and 39.4% [S9].

Recognise it: who is at risk

• Across 18 population-based studies (342,005 infants), the overall risk of ROP in preterm infants was 18.8% [S9].
• Gestational age and birth weight are the basis of national screening guidelines [S9]. Higher risk was also found with neonatal sepsis and bronchopulmonary dysplasia [S9].
• In regions with higher neonatal mortality, infants with ROP had the highest mean gestational age [S9]: more mature babies develop ROP in these settings, so screening criteria from high-income countries may miss them.
• Exact screening eligibility thresholds are not given in these abstracts: follow your national guideline.

Confirm it

• The recommended screening method is binocular indirect ophthalmoscopy, which needs frequent examinations and a heavy workload [S8].
• Weight-gain algorithms (WINROP, G-ROP, CHOP ROP, ROPScore, CO-ROP) had adequate sensitivity to rule out type 1 or severe ROP in a meta-analysis of 61 studies; only algorithms with sensitivity close to 100% should be used to reduce examinations safely [S8]. WINROP had a pooled sensitivity of 0.89 [S8], too low for that purpose.
• Telemedicine: low-income countries use available internet and communications technology for ROP screening; programmes need national support to maintain infrastructure [S7].
• Artificial intelligence on wide-field retinal images reported sensitivity of 71%-100% and specificity of 74-99%, comparable to ophthalmologists, but should be an adjunct rather than the sole diagnostic tool [S11].

Manage it

Before acting on this section, check the national and local guidelines, product information and regulatory approvals that apply where you practise. See the medical disclaimer.

Tier A (primary care / no specialist)

• Neonatal units: identify every infant meeting national screening criteria and make sure the examination happens [S9].
• Oxygen: in extremely preterm infants (under 28 weeks), restricted (SpO2 85%-89%) vs liberal (91%-95%) targets did not change ROP rates, but restricted targets increased mortality before discharge and necrotising enterocolitis (moderate to low quality) [S1].

Tier B (general eye clinic or screening service)

• Screen with indirect ophthalmoscopy or validated imaging and telemedicine where an ROP specialist is not on site [S7] [S8].
• After laser, arrange periodic cycloplegic refraction and early optical correction: laser was associated with myopia (OR 8.08) and astigmatism (OR 3.19) [S4].

Tier C (specialist centre)

1. Laser vs ranibizumab. In RAINBOW (225 infants under 1500 g, 26 countries), treatment success was 80% with ranibizumab 0.2 mg, 75% with 0.1 mg and 66% with laser (doses as reported in [S5]); superiority of 0.2 mg narrowly missed significance (p=0.051), with fewer unfavourable structural outcomes and similar adverse events at 24 weeks [S5].
2. Laser vs aflibercept. In FIREFLEYE (118 infants), aflibercept 0.4 mg (as reported in [S10]) gave 85.5% vs 82.1% success but did not meet noninferiority criteria; rescue treatment was needed in 4.8% vs 11.1% of eyes [S10].
3. Cochrane review. Bevacizumab or ranibizumab monotherapy did not reduce retinal detachment or overall recurrence, reduced recurrence in zone I but increased it in zone II, and lowered very high myopia at 30 months; evidence was very low to low quality and long-term systemic effects are unknown [S2].
4. Choice of agent and dose. A network meta-analysis of 68 studies (6445 infants) found no agent inferior to laser for retreatment; half doses retained efficacy but were associated with shorter times to retreatment [S12]. Exact bevacizumab doses are not stated in these abstracts: follow local protocols.
5. Neurodevelopment. A meta-analysis of 8 studies did not find increased severe neurodevelopmental impairment after bevacizumab (OR 1.39, 95% CI 0.98-1.97), with a minor difference in motor scores; evidence quality was low [S6].
6. Propranolol prophylaxis might reduce progression to stage 3 and the need for laser or anti-VEGF, but adverse events at 2 mg/kg/day (as reported in [S3]) raise concerns, and there is insufficient evidence for routine use [S3].

Refer when

Pitfalls

• Using high-income screening thresholds where more mature infants develop ROP [S9].
• Relying on an algorithm without near-100% sensitivity to skip examinations [S8].
• Using AI as the only diagnostic step [S11].
• Losing anti-VEGF-treated infants to follow-up despite recurrence risk [S2] [S12].
• Forgetting refraction after laser [S4].

Evidence notes

• Conflict: RAINBOW and FIREFLEYE suggest anti-VEGF performs at least as well as laser [S5] [S10] [S12], while the Cochrane review found no reduction in retinal detachment and warned of higher recurrence in zone II [S2]; the Cochrane review predates both trials.
• Funding: RAINBOW was funded by Novartis [S5].
• The network meta-analysis relied largely on non-randomised studies [S12].
• The UK 2022 screening guideline was in the search results but had no abstract, so it could not be cited.
• Not covered by these sources: national screening thresholds and examination intervals, ROP staging details (ICROP), vitrectomy for stage 4 and 5, and pain relief during screening.

Related articles

• The white pupil (leukocoria): urgent assessment
• Vision screening in children

References

See the numbered source list below. Each marker [S#] links to its PubMed and DOI record.