Outcome measures in amblyopia: what to track, and when

Amblyopia follow-up tends in practice to collapse into a single number. Visual acuity earns its place as the primary measure, but if the goal of treatment is not acuity alone then follow-up cannot consist of acuity alone either.

The frame below is intended to separate which measure answers which question. Clinical judgement always belongs to the whole patient.

Choosing an age-appropriate optotype is as decisive as the result itself. Preferential looking at pre-verbal ages, symbol-based tests in the pre-school years, letter-based tests once reading has begun.

The mode of presentation is a critical variable in amblyopia. Acuity measured with a single optotype can come out better than acuity on linear presentation because of crowding. Linear or surrounded presentation should therefore be preferred at every age that allows it.

In follow-up, the interocular difference is watched more closely than the absolute value. Where both eyes climb together but the gap does not close, the gain specific to amblyopia treatment is limited.

Stereoacuity is a direct measure of the two eyes working together and can move independently of acuity. In approaches with a binocular target, dichoptic programs in particular, recording it at baseline and at follow-up is meaningful.

Depth of suppression and binocular balance form the axis dichoptic treatment works on directly. The contrast balance a program requires shifting towards equality over time is not in itself a clinical outcome, but it is a practical indicator of response.

Angle of deviation and fusional ranges are followed separately where strabismus accompanies the picture.

Contrast sensitivity is informative where acuity has plateaued while a functional complaint persists. It need not be routine; it is used with a question in mind.

Fixation stability can be assessed where eye-tracking technology is available. It is increasingly used in research and is not standard in routine clinic.

Reading speed adds information where the aim is to objectify a functional complaint reported by the family.

Changing the test type between visits is the commonest and most misleading error. The difference between two optotype systems can be mistaken for real change.

Memorisation of the same chart inflates the result, particularly in frequently reviewed children. Tests that allow the line order to be varied should be preferred.

A change of examiner, of distance or of illumination are also variables worth noting. Recording the testing conditions beside the result spares the next assessor the guesswork.

The state of correction should be stated at every measurement: with or without glasses, and with which prescription.

Where a refractive error is found, allowing time for refractive adaptation prevents starting additional treatment unnecessarily early. Measuring at intervals during that stretch shows where the gains level off.

Once further treatment has started, the review interval follows the method and the child's age. Where penalisation is used, measuring the sound eye at every visit is the only way to catch reverse amblyopia early.

The definition of plateau is worth setting in advance. Where two consecutive measurements show no meaningful gain, the question is whether to change method or to raise the dose, and that distinction cannot be made without data on what was actually done.

Adherence is one of the variables that determines the result, and the gap between reported and actual treatment is a known problem. An objective record of the dose delivered is therefore not a detail but a precondition for interpretation.

The size of the gap has been measured: in a randomised trial using dose monitors, the group prescribed 6 hours of patching a day received a mean of 4.2 hours and the group prescribed 12 hours received 6.2 (Stewart et al., BMJ, 2007). Follow-up that relies on self-report cannot see that gap.

This is the practical advantage of digital programs: session count, duration and difficulty settings can be logged. At a plateau it then becomes possible to separate "no response" from "not enough treatment delivered".

Access to that data must rest on the patient's explicit and revocable consent. Who can see it and how long it is kept belong in the conversation as part of the treatment plan.