Normal Findings

1. 0:00Before you can recognise an abnormal DVA, you have to know what normal looks like across age and across paradigms. The largest reference dataset comes from Li, Beaumont, Rine, Slotkin, and Schubert in 2014 — the NIH Toolbox normative study of nearly four thousand individuals aged three to eighty-five.
2. 0:25Two findings dominate. First, between ages three and forty-nine, there is no significant change in DVA performance. A nine-year-old and a thirty-five-year-old score similarly. Second, from age fifty onwards, performance declines progressively. By the seventies and eighties, the upper limit of normal has widened well beyond the 0.2 logMAR threshold used at the bedside.
3. 0:55This matters because applying a single bedside threshold to a seventy-five-year-old can falsely label them abnormal. The age-matched mean plus two standard deviations is the right reference for the computerised test, and most clinical labs maintain their own age-banded norms.
4. 1:25Sex matters slightly. Males score worse than females in the NIH Toolbox dataset, a small but statistically significant difference. Ethnicity and education show no meaningful effects after adjusting for the visual acuity floor.
5. 1:50Test-retest reliability of computerised DVA is high. Intraclass correlations cluster around 0.83 for the active yaw protocol. The bedside test is less reliable — examiner-dependent variation in head motion frequency and amplitude introduces noise. Reproducibility over weeks is good in healthy subjects and poor in active disease, which is itself diagnostically useful.
6. 2:25Three practical reminders. Always pair a static visual acuity measurement with the dynamic one — DVA loss is the difference, not the absolute. Always test both directions of head motion. And always interpret the result against the right paradigm-specific and age-banded cut-off.

What normal looks like

In a healthy adult under fifty, DVA loss is small — typically under 0.1 logMAR (one line on a Snellen chart) at the standard bedside oscillation frequency of two hertz. The NIH Toolbox normative study, the largest available dataset, reported a population mean DVA loss across all ages of 0.116 ± 0.184 logMAR (n = 3,992).²¹

That mean conceals an important age effect. Between ages three and forty-nine, performance is essentially flat; a child and a middle-aged adult score similarly. From age fifty onwards, DVA loss climbs progressively, and the upper limit of the normal range widens accordingly.^21,23

Static visual acuity comes first

DVA loss is a difference, not an absolute. Static visual acuity must be measured first; the dynamic measurement is interpreted relative to it. A patient with uncorrected refractive error or cataract may have a low static acuity that produces a floor effect — the smallest optotype they can read with the head still is already large enough that head motion makes little additional difference, and the DVA loss looks falsely normal.

Correct vision first. Reading glasses worn for distance during the test. Then measure static, then dynamic. The Herdman 1998 protocol, the NIH Toolbox protocol, and the Schubert head-thrust paradigms all enforce this order for the same reason.^1,7,18

Age effects in detail

The Li 2014 dataset combined the paediatric population (ages 3–17) into a single band and analysed adults in decade-wide bands thereafter. Three findings deserve emphasis.²¹

1. No paediatric handicap. Three-year-olds with normal vestibular function score similarly to teenagers and young adults. This makes DVA usable as a screening test in children — a rare attribute among vestibular tests, most of which require adult cooperation. The Rine paediatric protocol carries 100% sensitivity and high specificity in children with bilateral vestibulopathy.⁹
2. A plateau through middle age. Between 18 and 49, the mean and SD are essentially stable. A 35-year-old with a DVA loss of 0.2 logMAR is borderline regardless of which decade of adulthood they sit in.
3. Decline from fifty onwards. By the 60–69 band, the mean has roughly doubled; by 70–85 it is approaching the bedside threshold even in healthy individuals. The SD widens in parallel, so the +2 SD upper limit of normal moves up faster than the mean itself.²¹

Sex and other factors

Males scored worse than females in the NIH Toolbox dataset (p < 0.001). The magnitude is small but statistically significant; the effect was present at every age band.²¹ Ethnicity, dominant language, and education showed no meaningful effects after adjustment.

Refractive error and visual function should be corrected before testing. Cervical spine instability, severely limited neck range of motion, and oculomotor impairment that prevents stable fixation are exclusion criteria — the test cannot validly interrogate the VOR if the head or eyes cannot do what the protocol requires.²¹

Reliability and reproducibility

Computerised DVA is reproducible to a degree that makes it suitable for monitoring response to vestibular rehabilitation. The Herdman 1998 protocol reported test-retest ICCs around 0.83 for the active yaw paradigm.^1,19 Mohammad et al. (2011) confirmed that both cDVA and GST are stable across visits in patients with peripheral vestibular disorders.¹⁹

Bedside DVA is less reproducible. Examiner-dependent variation in head oscillation frequency and amplitude introduces noise that the computerised paradigm eliminates. For monitoring change over time — rehabilitation, post-surgical recovery, ototoxic risk surveillance — stick with the computerised test and the same operator if possible.⁶

Choosing the right cut-off

Three reasonable strategies exist for declaring a cDVA score abnormal, and a clinician should know which their lab uses.^1,21

• Fixed cut-off — 0.2 logMAR per direction. The Herdman 1998 paradigm cut-off, equivalent to two chart lines. Simple, widely-used, but over-calls older adults.¹
• Age-matched mean + 2 SD. Preferred when an age-banded normative dataset is available. The Li 2014 dataset is the largest published reference for the NIH Toolbox protocol; lab-specific norms should otherwise be derived from at least 30 controls per age band.²¹
• Asymmetry of ≥ 0.1 logMAR between directions. Reads the test against itself, so it is robust to age effects and to floor effects in the static measurement. The most useful criterion for screening unilateral vestibular loss; insensitive to symmetric bilateral loss.⁷

Diagnostic accuracy at the published thresholds

For computerised DVA against a reference battery of caloric testing and rotational chair, the Herdman 1998 paradigm reported sensitivity 94.5% and specificity 95.2% for identifying any vestibular hypofunction in a mixed cohort.¹ The Vital 2010 protocol using 100°/s and 150°/s active yaw impulses reported 100% sensitivity against scleral search-coil VOR gain measurement.²²

In the paediatric population, Rine and Braswell 2003 reported 100% sensitivity and specificity for identifying bilateral vestibulopathy in children with sensorineural hearing impairment.⁹These published values come from research-grade protocols and well- characterised cohorts; real-world clinic performance is typically lower, particularly for unilateral subtle loss.

DVA in the context of the full vestibular battery

A normal DVA does not rule out vestibular pathology. The test probes the high-frequency VOR; isolated otolith dysfunction, low-frequency caloric weakness, and well-compensated chronic unilateral loss can all present with normal DVA scores.¹³ The right interpretive frame is:

• DVA abnormal, vHIT abnormal → high-frequency VOR loss with functional consequence; classic peripheral vestibular hypofunction pattern.
• DVA abnormal, vHIT normal → consider central origin, oculomotor disorder, or visual processing problem. Recheck refractive correction.
• DVA normal, vHIT abnormal → well-compensated peripheral loss, often with effective covert catch-up saccades. Patient may still be symptomatic at extremes of head velocity (GST helps here).^5,17
• DVA normal, vHIT normal, caloric abnormal → low-frequency canal loss only. Hydrops should be considered.

DVA during walking and passive translation

Routine clinical DVA is performed seated, with rotational head motion. Walking-DVA — performed on a treadmill while the patient reads a chart — interrogates the translational VOR in addition to the rotational, and is more sensitive to chronic bilateral loss than seated DVA. It is rarely available outside research settings.¹⁹

A version of DVA using passive vertical motion of the patient on an oscillating chair has been described for the bilaterally hyporeflexic population. It picks up patients in whom seated rotational DVA appears falsely normal because of robust efference-copy preprogramming. Specialised laboratories only.^19,22