What a Near Miss Proves, and What It Does Not

There is enormous commercial temptation in this category to convert near misses into saved lives. It is an easy sentence to write and it is not supportable. Nobody can establish what would have happened, and a claim that cannot be examined is a claim that will eventually be examined by someone hostile.

It also creates a specific risk for the facility, not just the vendor. A jail that has publicly asserted that its system prevented deaths has set a standard it will be measured against the first time something goes wrong. That is a bad trade for a county.

The honest framing is stronger anyway, because it is defensible. The chain worked. Here is the timestamp. Here is who responded. Here is how long it took. That holds up under examination in a way that a prevention claim does not.

That the event was detectable. Something observable occurred and it was observed, which is not a given.

That awareness reached a person, at a recorded time.

That a human made a decision and acted, and how long that took.

That the facility's chain from observation to response functioned in real conditions, rather than in a drill.

Over time and in aggregate, that the chain functions consistently rather than occasionally, which is the more meaningful claim and the one that requires a series rather than an anecdote.

Record near misses as their own event class, distinct from incidents. Facilities that file them as incidents lose them in the noise; facilities that do not record them at all lose them entirely.

Capture the same fields as any event: time of onset if known, time raised, time acknowledged, time on scene, what was found, what was done, and by whom.

Write what was observed factually and avoid characterizing the outcome that did not occur. The record should say a person was found unresponsive and medical was summoned, not that a death was averted.

Review them in the same forum as incidents. A monthly review that looks only at failures teaches the organization that only failures are informative.

Track the pattern. Near misses cluster by area and by hour in the same way incidents do, and they arrive in greater numbers, which makes them better data for identifying where attention should go.

Virtual Patrol produces the timestamped chain that makes a near miss documentable: what was raised, when, who acknowledged it, and what they did. Without that, near misses exist only as staff recollection and mostly evaporate.

The company does not claim prevented deaths and does not publish detection or accuracy figures. What is claimed is earlier awareness and a structured record, and both of those are things a facility can verify for itself.

A person reviews every raised event and makes every decision. No facial recognition is used.

Why not say a life was saved?

Because it cannot be established. The counterfactual is unknowable, and a claim that cannot be examined becomes a liability for the facility that repeats it.

Is a near miss still valuable evidence?

Yes, and more so than most facilities realize. It demonstrates a working detection and response chain under real conditions, with times attached.

Should near misses be public?

That is a facility and county decision involving privacy and legal considerations. The internal record should exist regardless of whether anything is published.

How should a near miss be written up?

Factually, describing what was observed and what was done, without characterizing the outcome that did not happen.

Do near misses help identify risk areas?

Considerably. They occur more frequently than serious incidents and cluster in the same places, which makes them better data for deciding where to focus.

Does VPT publish prevention claims?

No. Earlier awareness and a structured record are the claims, and both are verifiable by the facility itself.