To the Editor:

Bartlett et al. argue agreement between optometrists and ophthalmologists for sight impairment (SI) registration [1]. Raters were compared to a consensus opinion of four consultant ophthalmologists and three optometrists for each case rather than to each other. Agreement with the panel was 52% and 72% for SI registration; 76% and 62% for SSI registration for ophthalmologists and optometrists, respectively. These appear to be low agreement levels for a quantitative assessment of vision loss. We feel that when comparing to a standard, (the consensus group), sensitivity and specificity would be more informative. We calculate that the sensitivity and specificity of an ophthalmologist rating of any SI versus ineligible are 75.1% and 94.2%, respectively. For an optometrist rating these are 88.0% and 87.8%, respectively. Again, we consider this low for a quantitative assessment of visual acuity (VA). Thus, in this cohort optometrists over-diagnose and ophthalmologists under-diagnose SI relative to each other. Both groups were equally poor.

We wonder how the background information graders were provided with about social and living status, age, and occupation influenced graders [2]? It is possible that this information could introduce subjective individual rater bias into the process of eligibility accounting for some of the disagreement and making total agreement almost impossible. It does, however, reflect the current registration process [2].

Interpretation of cut-offs for VA eligibility should be done with relative certainty due to its quantitative nature leading to a greater likelihood of those with VA loss being registered [3]; unlike visual field (VF) loss where there is significant delay and disagreement regarding eligibility [4, 5]. We wonder whether any differences in interpretation of VF eligibility also led to disagreement in this study?

Terminology regarding VF loss eligibility within the explanatory document for registration is ambiguous. A more pragmatic approach may be to adopt a quantitative approach. The United States Social Security Administration has included the following VF criteria to identify those eligible for blindness registration: a mean deviation of ≥−22 decibels in the better eye, determined by automated static threshold perimetry measuring the central 30° of VF [6]. This may improve levels of agreement.