Using cycloplegia, the change in ametropia of 113 infants was followed at 3 month intervals over the first year of life. Scatterplots of the spherical equivalent power show that the dioptric differences exhibit a significant myopic shift of -0.38 ds between 26 and 36 weeks and -0.38 ds between 36 and 52 weeks. The spread of the dioptric differences (95% CI) does not appear to be related to the magnitude of the ametropia present and decreases with time. By 12 months of age the frequency distribution of the spherical equivalent appears to become leptokurtic as it is in the adult. On average the astigmatism was of low degree (less than 1 dioptre cylinder) and with the rule. Anisometropia was rarely seen. The results of this longitudinal study point to an optimal time for screening and perhaps prescribing for 'abnormal' refractive error between 9 and 12 months of age.

Photorefraction has been suggested as a suitable method of screening for refractive error in infants. The relative performance of cycloplegic and non-cycloplegic videorefraction and cycloplegic retinoscopy was investigated on 150 infants. Under cycloplegic conditions the correlation between findings for spherical error (Rxy = 0.70) was compatible with a previous study. However, where cycloplegia was not used for videorefraction, there was poor agreement between the two techniques in the case of astigmatic error, and all types of ametropia. Interobserver repeatability was very high both for cycloplegic retinoscopy (Rxy = 0.96 spherical error, and Rxy = 0.75 astigmatic error) and for videorefraction measurements (Rxy = 0.95 horizontal meridian of photograph and Rxy = 0.85 vertical meridian). Intraobserver repeatability was also good, both for cycloplegic retinoscopy (Rxy = 0.91 spherical error and 0.82 astigmatic error) and for videorefraction with regard to spherical errors (Rxy = 0.84). Throughout the experiments videorefraction measurements of astigmatic errors proved less consistent when compared with cycloplegic retinoscopy, and to its internal reliability.

Isotropic photorefraction has been suggested as a suitable method for screening infants for refractive error. Recently published data suggested that reasonable consistency with retinoscopy results might be achieved using cycloplegic videophotorefraction (VPR) for spherical refractive error but that results might be unreliable for astigmatic errors. Non-cycloplegic VPR did not appear to produce results consistent with retinoscopy. A practical idea of how many children might be identified using this technique and how many missed was needed by personnel designing screening projects. Hence the VPR was tested by screening a population of 247 infants for significant refractive error, and comparing the results with cycloplegic retinoscopy. Sensitivity and specificity scores were calculated for a range of test levels of ametropia. Without cycloplegia, sensitivity of VPR was poor. With cycloplegia the situation was much improved, with sensitivity for hyperopia +4.00 D or over of 83.3% and specificity of 90.6%. Sensitivity for astigmatism of 1 D or greater (84.6%) was high but specificity was poor (45.6%). Acceptable sensitivity was achieved for identifying children in this age group at risk of developing squint and amblyopia due to refractive error, providing cycloplegia was used.

Agreed clinical guidelines would make the prioritisation of most referral letters straightforward by a trained ophthalmic practitioner. However, traditional methods are inherently variable. To assist the untrained prioritiser there may need to be a priority scoring system, or availability of a triage assessment clinic.

No abstract