Sensitivity and specificity of the red reflex in Costa Rican newborns

To compare the results of indirect ophthalmoscopy versus determination of the red reflex in newborns admitted to the intermediate neonatal intensive care unit.

All the patients (71) admitted to the intermediate neonatal intensive care unit from July to September of 2017 were screened with the red reflex. Immediately afterwards, they were pharmacologically dilated and indirect ophthalmoscopy was performed. All of the red-reflex screening and the funduscopic examination were performed by the same evaluator.

A total of 14.1% (20/142) of the eyes had an altered red reflex. Indirect ophthalmoscopy revealed that 21.8% (31/142) of the eyes had intraocular abnormalities. Two of these eyes had optic nerve colobomas, two eyes had cataracts, and the rest had intra-retinal haemorrhages. Indirect ophthalmoscopy revealed that only 35% (7/20) of the eyes with an altered red reflex had intraocular disease. Indirect ophthalmoscopy also revealed that intraocular disease was present in 77.4% (24/31) of the eyes despite a normal red reflex. The false positive rate of the red reflex was calculated to be 9.2%, and the false negative rate was 16.9%. The sensitivity of the red reflex was only 56.4%, and the specificity was 89.5%. The positive predictive value was 70.4%, and the negative predictive value was 82.2%.

Neonatal screening with the red reflex is not sufficient to detect intraocular disease. A normal neonatal red reflex does no exclude intraocular disease. It has a specificity of nearly 90%, but its sensitivity is only 56%. Most retinal haemorrhages are undiagnosed as they cannot be detected with a red reflex.