General operational and educational information for corrections professionals. Not legal, medical, or compliance advice, and not a certification of compliance with any law or standard. Policies and standards vary by agency and jurisdiction; follow your facility's policy and your own legal, medical, and professional advisors.
After a self-harm incident the facility has a narrow window in which the review is worth something. Done within days it produces a record and a change. Done months later under legal pressure it produces testimony about memories. The difference is almost entirely about when it starts, not how thorough it eventually becomes.
Every self-harm event, including attempts and interruptions, contains information the facility will not get any other way. The conditions, the timing, the position, what was visible, what was known beforehand, and who knew it.
Most of that information degrades quickly. Staff memory hardens into a version within weeks, and the version people sincerely believe at month nine is not the one they would have given at day three. Video expires. Housing changes. The cell gets reassigned.
The facilities that improve are the ones that treat every event, including the ones with no injury, as a review trigger. The facilities that do not are examining only their catastrophes, which is both too few data points and the worst possible sample.
The preceding period, not the event. Start the window hours before, at minimum. The behaviour that mattered usually began well before anything looked like an emergency.
What was known beforehand: screening, prior history, watch level, recent case or family events, and critically whether that information reached the officer working the post.
The observation record: what interval was ordered, what evidence exists that it occurred, and what the last confirmed contact was as distinct from the last round.
The physical environment: what the observation position could actually see, and whether the person was in a part of the cell nobody could view.
Timing: onset to awareness, awareness to response, response to medical intervention. Those three intervals do more work in any subsequent review than the narrative around them.
The handoffs: what was passed at shift change and between custody and clinical, and whether anything was requested and not answered.
Separate fact from analysis. Build a factual timeline first with a time, a source, and a plain statement per entry. Put conclusions in a separate section. A timeline containing argument stops being useful as a record.
Mark the gaps explicitly with their length and cause. No coverage in that area. Camera out of service. Retention expired. No contemporaneous record created. Each is a different problem with a different remedy, and an acknowledged gap is far better than one discovered later by someone else.
Write a finding and an owner, with a date. A review with no action item is an account, not a review.
Involve your county attorney early on privilege and process, particularly where litigation is reasonably anticipated. How and when statements are taken is a legal judgment, not an operational one.
Feed the result into the standing safety review so that the pattern across events becomes visible. Single reviews improve one thing. A series improves the facility.
Most of the difficulty in this work comes from assembling records that were never designed to be assembled. Virtual Patrol produces a timestamped, structured stream of what was raised, who acknowledged it, and what they did, which supplies a large part of the timeline as a byproduct of ordinary operation rather than as a reconstruction project.
It also reduces the number of unknown spans, because coverage in designated areas is observed continuously rather than only when someone happens to be watching.
A person reviews every event before action and no facial recognition is used. Nothing here makes any judgment about a person in crisis; clinical assessment belongs to clinicians.
Yes, and they are the more useful data. They are more numerous, they carry the same information about conditions and timing, and they cost nothing to learn from.
Within days. Preservation of video should happen immediately and wide, camera by camera, confirmed in writing.
Elapsed time from the first visible sign to the first staff response. It is objective and directly relevant, and most facilities cannot state it.
No. Keep entries factual and put analysis separately. Reviewers discount a timeline that argues.
Mark it with its length and cause. An acknowledged gap with a stated remediation is defensible; one found by opposing counsel is not.
Someone not directly involved, working with counsel where litigation is reasonably anticipated.
