"How long should it last?" is among the first questions any team asks, and the honest answer starts with an admission: there is no established dose of virtual reality, in the sense that there is a dose of a drug.
It is worth understanding why, because it changes how you decide.
What the literature uses, and what that is
In published work on virtual reality in health contexts, sessions usually run between ten and twenty minutes, with some shorter ones during procedures and some longer ones in rehabilitation. That is the range you will see quoted everywhere.
What is almost never said is where that range comes from. It does not come from trials that compared durations to find the best one. It comes from logistical decisions: how long the procedure took, how much time the research team had, the battery life of the equipment, what someone judged a participant would tolerate. It is what fitted, not what was optimised.
So when a supplier tells you "twenty minutes, twice a week", ask where that came from. It is an extrapolation from ranges in studies with different designs, different populations and different goals. We gathered the work in the study library precisely so that it can be checked rather than believed.
Three situations, three different logics
Grouping everything under "session duration" is the mistake. There are at least three cases and they are decided differently.
1. During a procedure. Here duration is not a choice: it is the length of the procedure. It starts before (putting the headset on is already part of it) and ends when the procedure ends. Asking "how many minutes?" makes no sense — the answer is "as long as the cannulation, the dressing or the dialysis session takes".
2. As a moment of calm or stimulation. Here there is a choice, and the practical rule teams converge on is short: end before the person wants to end. A session that finishes with someone asking for one more is the session that makes sure there is a next one. One that drags into tiredness is the one that gets refused next time.
3. In rehabilitation. Here duration is subordinate to the exercise and to whoever is guiding it. The system does not decide, the professional's plan does — and fatigue is the real limit, not the clock.
Frequency is a scheduling question, not a pharmacological one
On "how many times a week", the honest answer is shorter still: there is no evidence establishing an optimal frequency. What exists is what the institution can sustain.
And here is the most useful observation we have: the frequency you keep beats the ideal frequency. A care home doing two sessions a week for a year does more than one that plans five and gives up in the second month. The practical question is not "what is the best frequency?", it is "what is the highest frequency this team can keep up in February, with someone off sick?".
Signals that end a session early
Whatever was planned, the session ends when:
- the person asks to stop — no negotiation, and this comes before all others;
- signs of discomfort or motion sickness appear: pallor, sweating, complaints of dizziness, reaching for the chair;
- the person stops responding to what is happening, or seems lost rather than engaged;
- agitation is increasing rather than decreasing;
- whoever is running it has a doubt. The doubt is enough.
None of these is a failure. They are information, and they are worth more logged than remembered.
What the log solves, and what it does not
Since there is no reference dose, the only route is to look at what happens in your institution. Every session is logged — what ran, how long, when it was stopped. After a few months that is a series you can argue about: are this person's sessions systematically stopped at five minutes? Does this scenario hold attention longer than that one?
What this is not: proof of effect. A log says what was done, not that it worked. A supplier presenting session counts as a clinical result is changing the subject — we have written about what cannot be proven.
In short
- There is no established dose. The quoted 10–20 minutes is study logistics, not an optimum.
- During a procedure, the duration is the procedure's.
- As a moment of calm, end before the person wants to end.
- Sustainable frequency beats ideal frequency.
- Stop signals always beat the plan — and the doubt of whoever is running it is one of them.