Managers on the business side of a hospital or health service keep running into cyber words such as containment, forensics, dwell time and tabletop. Few arrive with a definition, and the meetings that use them rarely pause. Here we slow down. The subject is ResponseOps, the Ops Atlas shorthand for treating a security incident as a capability the organisation always keeps ready, and twenty key terms are explained below. The article concerns non-clinical operations only; it does not discuss patient care, and it is background reading, not professional advice.
First, the big idea in three sentences
Security incidents are inevitable at some point for most organisations. What separates a rough week from a damaging one is whether people know in advance who decides, what to protect first and what must be reported. ResponseOps is the Atlas’s name for preparing exactly that, and it is worth noting that the entry itself says the word is its own shorthand and not an official standard.
Where the term comes from
According to the Atlas, none of the incident-response standards adopt the word. The international text that covers the practice, ISO/IEC 27035, instead speaks of information and communications technology incident response operations. U.S. guidance on the topic, NIST SP 800-61, was reissued in April 2025 as a third revision that links response to a wider cybersecurity framework; its predecessor from 2012 was withdrawn on 3 April 2025. Emergency-management agencies use the phrase “response operations” for disasters, so context matters when you search. For contracts, use incident response.
One point of orientation before the lists. The glossary follows the order in which a real event unfolds, which mirrors the eight practices the Atlas lists, starting with declaring and ending with learning. You can therefore read it as a timeline, and a team can print each group as a one-page card to pin beside a phone.
Mini-glossary, part one: the first hour
- Declaration. Someone with the authority says aloud that an incident is open, which begins the log and the timers.
- Incident lead. The coordinator of the effort, who need not be the most technical person present.
- Out-of-band channel. A route for talking that avoids the systems being doubted, for instance a dial-in call.
- Break-glass account. A sealed emergency login, kept for the moment ordinary accounts cannot be trusted.
- Chain of custody. The paper trail showing every person who held a given item of evidence, at what time and for which purpose.
The Atlas’s suggested first month for any organisation centres on these items: a lead and executive sponsor, a short written plan, an independent communication line and break-glass accounts. None needs a large budget.
Mini-glossary, part two: evidence and scope
- Order of volatility. A rule from a 2002 internet RFC: capture the data that disappears fastest first. Memory comes before disk.
- Hash. A digital fingerprint taken at the moment of collection, to show later that a copy has not changed.
- Write-once storage. Media that does not allow edits after saving.
- Scoping. Working out how many systems, accounts and records were touched. Early estimates tend to be too small; the Atlas’s sample case grew from six systems to fifteen.
- Dwell time. How long an intruder operated before being spotted. The Atlas’s worked example lands slightly above eleven hours, whereas a 2025 industry report it cites puts the global median at a fortnight.
Why do administrators need to understand these? Since the decisions about whether to shut a system down, call an outside adviser or tell a regulator often land on non-technical desks. Knowing that a reboot erases memory, for example, helps a manager ask the right question before agreeing to one.
Mini-glossary, part three: removing and restoring
- Containment. Limiting the damage without destroying evidence. The 2012 U.S. guidance suggested judging each step on possible harm, evidence, service uptime, effort, success rate and staying power.
- Eviction. Removing the intruder completely, after every access route has been found, and doing it in one coordinated move.
- Persistence. Any hidden way back in: spare accounts, scheduled tasks, forwarding rules.
- Double extortion. Stealing data before locking it, so that restoring from backup does not undo the theft.
- Dead-man switch. A booby trap left by an attacker, which wipes data when a machine is cut off or switched off.
The Atlas lists several classic errors in this territory: evicting before scoping is complete, restoring from tainted backups and resetting a key once where twice was needed. Each can be prevented by a simple pre-written checklist.
Mini-glossary, part four: duties and practice
- Notification clock. The time limit for informing a regulator, partner or the public. Depending on the regime, it starts at awareness, classification or a materiality decision.
- Awareness. The point at which an organisation is treated as knowing. Regulators interpret it with some nuance, and the Atlas says deadlines are often missed because nobody owns that judgement.
- Playbook. A short, scenario-specific procedure, for example for ransomware, account takeover or a leaked cloud key.
- Tabletop exercise. A talk-through of a made-up incident with leadership present, which ends with a list of fixes.
- Maturity ladder. The Atlas’s proposed scale: ad hoc, defined, practised, measured, adaptive.
Three misunderstandings worth correcting
Terms are easier to remember when attached to a mistake. First, many people assume reinstalling a computer settles an incident. The Atlas says it erases the evidence and removes one entry point, but not the person behind it. Second, people assume that a restore from backup returns everything to normal; with double extortion, stolen data stays stolen. Third, some managers think routing a report through lawyers makes it untouchable. A 2020 U.S. court decision described in the Atlas ordered production of a consultant’s report even though counsel was involved, so the entry treats privilege labels with caution. As ever, this is general information rather than legal advice.
Why the deadlines deserve special attention
The Atlas’s snapshot for October 2026 shows how much the timing differs. One European privacy rule allows three days from awareness. A European directive on network security asks for an early heads-up inside a day. India’s national agency wants a report within six hours. A health-privacy rule in the United States gives up to sixty days. That spread is the reason to build a deadline map with legal counsel and to name the person who judges awareness. The entry’s own caution applies: rules change, so check current text each year.
A final practical suggestion for the administrative team: pick five of the twenty terms and test whether everyone in the next planning meeting can explain them in a sentence. Start with declaration, chain of custody, containment, awareness and tabletop exercise. Where the answers differ, you have found where a short training session would help. Vocabulary alignment is a modest goal, yet it prevents the confusion that turns a routine call into an argument at the worst possible time.
How the glossary becomes a routine
Terms are only useful when they appear in a plan. A reasonable path for a small administrative team is to work through the Atlas’s staged build. In the first thirty days, appoint the lead and write the plan. By sixty, draft the three playbooks and a custody template, and agree the deadline map with counsel. By ninety, run an executive tabletop, correct the main gaps and begin recording simple measures such as time to declare and time to contain. Readers who like to see how a personal routine is built from scratch can find a parallel in our guide to setting up a first monthly money plan, and anyone interested in technology work might browse the practice schedule for getting through technical interview rounds.
For the complete discussion, the source is the Atlas’s long entry on ResponseOps, which sets out scope boundaries, standards and tools in detail. A quick last reminder: count incidents if you like, but remember that, as the Atlas says, a raw count measures workload, not quality. Time to declare and time to contain are the figures that show whether a team is getting better.