Performing the test

1. 0:00DVA testing comes in four closely related forms. All four share the same logic — read an optotype during head motion — but they differ in how the head is moved, how the optotype is shown, and what claim the test can support.
2. 0:20The bedside test is the simplest. The patient sits in front of a Snellen chart at two to three metres. Static visual acuity is established first — the smallest line the patient can read with the head still. Then the examiner stands behind the patient, grips the head firmly at the temples, and oscillates it horizontally at about two cycles per second. The patient reads down the chart. A drop of more than two lines from static is abnormal.
3. 0:55Two practical points for the bedside test. First, the oscillation must be passive — the examiner moves the head, not the patient. Active motion lets the brain preprogramme compensatory eye movements and inflates the score. Second, the frequency must be at least two hertz. Below that, smooth pursuit can keep up with the chart and the VOR is not specifically tested.
4. 1:25Computerised DVA replaces the human examiner's eye with a rate sensor and an optotype-flashing computer. The patient wears a head-mounted gyroscope. An optotype — typically a randomly rotated letter E — appears on the screen only when head velocity exceeds a threshold. The Herdman protocol uses 120 degrees per second; the NIH Toolbox protocol uses 180. Optotype dwell is around 80 milliseconds — short enough to exclude pursuit.
5. 2:00The gaze stabilisation test inverts the question. Instead of asking how small an optotype the patient can read at a fixed velocity, it asks how fast the head can move while the patient still reads a fixed optotype. The outcome measure is peak head velocity, not visual acuity. Reduced gaze stabilisation velocity on one side compared to the other is the signature of unilateral vestibular loss.
6. 2:35Head-thrust DVA is the most demanding paradigm. A single passive, rapid, unpredictable head impulse replaces the sinusoidal oscillation. The impulse is delivered in the plane of a single semicircular canal — horizontal for the lateral canals, RALP for the right anterior and left posterior, LARP for the left anterior and right posterior. Because the canals are stimulated one at a time, the test localises loss to individual canals.
7. 3:10Use the interactive simulator on this page to feel how gain, frequency, and amplitude interact. The simulator is a teaching caricature, but the qualitative relationship — slip rises as gain falls, and acuity drops with slip — is the substrate of every clinical DVA test you will perform.

The DVA simulator

Three sliders — VOR gain, head frequency, head amplitude — drive a live optotype and a head silhouette. Watch what happens to retinal slip as gain drops, and how the optotype blurs as slip rises. The numeric read-out estimates the steady-state DVA loss in logMAR.

1. Bedside DVA

The bedside test requires only a wall-mounted Snellen chart. Static visual acuity is established first — the smallest line the patient can read with the head still. The examiner then stands behind the seated patient, grips the head firmly at the temples, and oscillates it horizontally at approximately two cycles per second. The patient reads down the chart. A loss of more than two lines from the static baseline is taken as abnormal.^6,11

Step-by-step

1. Seat the patient at two to three metres from a wall-mounted Snellen chart, with corrective lenses worn for distance if normally used.
2. Record static visual acuity — the smallest line the patient can read with the head still. This is the reference.
3. Stand behind the patient, grip the head firmly at the temples or occiput, and oscillate horizontally at approximately 2 Hz with peak displacement of ±20–30 degrees. Avoid pausing at the turn-arounds.
4. Ask the patient to read down the chart. Record the smallest line they can identify during oscillation.
5. Calculate the lines lost. ≥3 lines is abnormal; 2 lines is borderline and warrants further testing.⁶

2. Computerised DVA

The computerised test replaces the human examiner with a head-mounted rate sensor and a screen that flashes optotypes only when head velocity exceeds a pre-set threshold. The Herdman protocol uses ≥120°/s; the NIH Toolbox uses ≥180°/s with an 83 ms optotype dwell — short enough to exclude smooth pursuit and pre-programmed saccades.^1,18

A randomly oriented letter E (or Landolt C) appears briefly when the velocity gate triggers. The patient identifies its orientation; an adaptive algorithm converges on the smallest reliably identified size. Static visual acuity is measured first, dynamic acuity is measured per direction (rightward and leftward head motion separately), and the per-direction loss in logMAR is reported.⁷

Normative values

The NIH Toolbox normative study (n = 3,992, ages 3–85) is the largest available reference dataset. cDVA was worse in males than females, and began to decline from age 50 onwards; younger age bands (3–49) showed no significant difference.¹⁸ Most clinical labs use a per-direction loss of ≥0.2 logMAR as the cut-off for abnormality, with asymmetry of ≥0.1 logMAR between sides considered significant.

3. Gaze stabilisation test (GST)

The GST inverts the DVA question. Instead of asking how small an optotype the patient can read at a fixed velocity, it asks how fast the head can move while the patient still reads a fixed optotype. The outcome is peak head velocity at which the patient reliably identifies the optotype, reported per direction.¹⁷

A fixed optotype size — typically 0.2 logMAR above the patient's static acuity — is used.²⁰ The patient performs head motion at progressively increasing velocity bands (60–99, 100–139, 140–179, 180–219, ≥220°/s in the original Goebel protocol). The test stops when the patient can no longer identify the optotype at three out of five presentations in the current band; the maximum velocity for the preceding successful band is recorded.¹⁷

GST values vary substantially across labs because protocol details differ — testing distance, optotype size, the increment between velocity bands. Reduced contralesional GST velocity is a sensitive marker of unilateral vestibular dysfunction; an asymmetry index of ≥25% between sides is commonly considered abnormal.^17,20

4. Head-thrust DVA (htDVA)

The most demanding paradigm. A single passive, rapid, unpredictable head impulse — in the plane of a specific semicircular canal — replaces sinusoidal oscillation. Acuity is measured during the impulse window. Because the canals can be stimulated one at a time, the test localises loss to individual canal pairs.⁴

Schubert and colleagues proposed a head-thrust DVA cut-off of 0.158 logMAR (mean + 2 SD in their healthy control cohort) for abnormality.⁴ The impulse must be brisk: an acceleration of 3,000–6,000°/s² is typical, with peak velocity ≥200°/s.

Which test, when?

The four paradigms are not interchangeable. They probe the VOR at different operating points and report different things, so the right test depends on the clinical question.^7,19

• Bedside DVA — for the initial screen in any patient with dizziness, oscillopsia, or imbalance. Fast, free, and 80% sensitive for bilateral vestibular loss in experienced hands. Not sensitive enough for unilateral subtle loss.⁶
• Computerised DVA — for quantitative documentation (research, medico-legal, rehabilitation monitoring). Reproducible (ICC ~0.8) and well-normed across ages.^18,19
• GST — for patients whose static DVA is normal but who complain of oscillopsia at high head velocities. The GST extends the test into a higher-velocity regime than standard cDVA.¹⁷
• htDVA — for topographic questions. Selective inferior-division neuritis, post-SCD plugging follow-up, partial canal palsy. Pair with video head-impulse testing.⁴

Technical considerations

Three engineering details that disproportionately affect test performance, in order of impact:

1. Velocity gating. If the optotype is displayed at head velocities below ~120°/s, smooth pursuit and pre-programmed saccades contribute and the test loses VOR specificity. Set the threshold high.^2,18
2. Optotype dwell. Display times under 100 ms exclude pre-programmed compensatory saccades, which have a minimum latency of about 100 ms. The NIH Toolbox standard of 83 ms is a sound default.¹⁸
3. Optotype family. Letter E and Landolt C tests require the same orientation judgement and are functionally equivalent. Mixed letter sets (ETDRS) increase difficulty but allow normative comparison with ophthalmic acuity testing.¹⁸

Using DVA to monitor vestibular rehabilitation

DVA improves with vestibular rehabilitation in both unilateral and bilateral loss.^3,5 Schubert and colleagues (2008) attributed the improvement to two mechanisms: a small rise in active VOR gain and an increase in the number of well-timed compensatory saccades during the head motion. The latter often dominates in chronic peripheral loss.⁵

When monitoring a rehabilitation programme, the same paradigm should be used at each visit — bedside is not interchangeable with computerised. A loss reduction of 0.1 logMAR (one chart line) over four to six weeks of structured gaze-stabilisation exercise is a typical responder pattern.⁵