Four Letters From a Submarine: SBAR and the Discipline of Saying the Important Thing First

There is a particular sentence that every senior clinician has been woken by, and it goes something like this:

"Hi, sorry to bother you — it's about the gentleman in bed four, the one we admitted yesterday, I think it was yesterday, anyway he's the one with the chest thing, and the nurses are a bit worried about him, he's just not looking great, and I wondered whether you might want to come and have a look at some point?"

Everything wrong with that call is a communication-systems failure rather than a knowledge failure. The caller almost certainly knows what is happening. The information required to act is somewhere in that sentence, or was going to be by the end of the third one. What is missing is structure: the urgent thing is not first, the interpretation is never stated, and no request is ever actually made. The recipient is left to reconstruct a clinical picture from a narrative, in the dark, at 2 a.m., and then to guess what is being asked of them.

SBAR is the standard remedy. Four letters, four moves, and the sentence above becomes something a decision-maker can act on in fifteen seconds.

I want to do this one properly, because SBAR is taught almost everywhere and understood rather less often. Most teaching covers the mnemonic and stops. What it usually leaves out is where the framework actually came from (the popular story is close to right but overstated in a specific way worth knowing), what it is doing mechanically to make communication safer, and — the part I care most about — what the evidence does and does not support, which is a good deal less triumphant than the average induction slide implies.

1. The four letters

The R is worth pausing on. Some organisations expand it as Recommendation, others as Request, and the AHRQ TeamSTEPPS materials hedge by giving both — SBAR is defined there as "Situation, Background, Assessment, and Recommendation (or Request)," a structured framework for sharing information "about the condition of a patient or team member or about another issue your team needs to address."

That is not a trivial ambiguity, and it is not sloppiness. The two readings sit at different points on a hierarchy. A consultant calling a colleague makes a recommendation. A first-year nurse calling a consultant at night may not feel entitled to recommend anything, and telling them they must can turn a useful tool into an intimidating one — but they are always entitled to make a request. "I need you to see him within the next ten minutes" is a complete and legitimate R. In teaching I now give both words explicitly and let the caller take whichever one they can say out loud, because an R that goes unspoken because the caller felt presumptuous is the exact failure the framework exists to prevent.

2. The same call, structured

Here is the 2 a.m. call again, run through the framework.

Situation. "This is Dr Otieno in the emergency department. I'm calling about a patient with sudden chest pain and hypotension."

Background. "He is 68, has known coronary disease, and underwent PCI two years ago."

Assessment. "His blood pressure is 82/54, the ECG shows inferior ST elevation, and he is becoming confused."

Recommendation. "I recommend immediate cardiology review and activation of the STEMI pathway."

Fifteen seconds. Compare what the two versions demand of the listener.

The last row is the one that matters. A structured call converts the recipient's task from construction to adjudication. Construction at 2 a.m. is where errors live.

Note also what the format has forced out of the caller. The unstructured version contained no assessment at all — "not looking great" is an observation dressed as one. The structured version forces the speaker to commit: this is an inferior STEMI, and he is shocked. That commitment can be wrong. Being wrong out loud, early, and correctably is enormously safer than being vague and unfalsifiable, and this is the single most underrated property of the framework.

3. Where it actually came from

Ask around a hospital and you will be told that SBAR was invented by the US Navy for nuclear submarines. That is the right neighbourhood and the wrong level of precision, and since I have spent a fair amount of this blog complaining about frameworks with unexamined provenance, I should be careful here.

What is well documented: the modern healthcare SBAR framework was introduced at Kaiser Permanente in 2002. The people usually credited are Michael Leonard, Doug Bonacum and Suzanne Graham, and it appears in their widely-cited paper on human factors in patient safety, The human factor: the critical importance of effective teamwork and communication in providing safe care (Quality and Safety in Health Care, 2004;13(Suppl 1):i85–i90). Bonacum is a retired US Navy officer who worked in the nuclear submarine service before moving into healthcare patient safety, and a BMJ rapid response states the lineage directly: in 2002 SBAR was introduced by "retired United States Navy Captain Doug Bonacum, while working at Kaiser Permanente on patient safety," drawing on "his handoff experience from nuclear submarine crews needing to discuss strategies quickly during shifts changeover."

What is worth qualifying: the claim usually shortens to "the Navy invented SBAR," which implies a formal military checklist with those four words in it, from which healthcare copied. The stronger and more defensible statement is that the healthcare framework was adapted from communication practices associated with US Navy nuclear-submarine operations by someone who had lived inside them — not lifted from a published military document. The four-letter mnemonic as clinicians know it is a healthcare artefact with a naval intellectual debt. That distinction survives scrutiny; the folk version does not, and it is the sort of thing that gets picked apart in exactly the settings where you most want your framework to be credible.

The stated design problem at Kaiser was specific and is often forgotten. Leonard and colleagues framed SBAR as a way to bridge the gap between two professional communication styles: the narrative, descriptive mode that nursing training encourages, and the terse, problem-first, headline-led mode that medical training rewards. Two groups of competent professionals, trained to speak differently, failing to transfer information across the boundary. SBAR is a shared grammar for that boundary, not a nursing tool and not a medical one.

4. Why a submarine, specifically

The naval provenance is not decoration. It explains the design, because a submarine crew has a very particular combination of constraints:

• Small teams in confined spaces, with no option to escalate outside the boat.
• High-consequence failures, where the tail of the outcome distribution is the loss of the vessel.
• Complex technical systems whose state cannot be inferred by looking around.
• Steep and formal rank gradients, across which junior personnel nonetheless have to raise problems.
• Shift handovers that must transfer the operational picture completely, because the incoming watch cannot see what the outgoing watch saw.
• Ambiguity and delay as direct threats to survival.

Read that list again as a description of a night shift on a busy ward and the transfer becomes obvious. The reported naval usage — structuring reports on equipment abnormalities, reactor and propulsion concerns, navigation hazards, changes in operational status, watch turnover, and emergencies requiring a command decision — maps almost item for item onto clinical escalation, handover, referral and deterioration calls.

The underlying discipline is the same in both settings: state the problem first, supply only the context that bears on it, declare your interpretation, and name the decision you need.

5. Where else it went

Aeromedical evacuation

The US Air Force applied SBAR to one of the least forgiving handover environments there is. A 2012 Wright State University project by Karey M. Dufour, Implementation of the SBAR Checklist to Improve Patient Safety in the United States Air Force Aeromedical Evacuation, identified handoff communication deficiency as a leading patient-safety problem in Air Force patient transfers — 1,694 incident reports filed in 2010, and no standardised transfer protocol in existence — and developed an SBAR checklist adapted for aeromedical use, submitted to Air Mobility Command for implementation.

The setting is instructive because the handover chain is long and heterogeneous: ground medical team, flight nurse, aircrew, receiving facility, each with different training and different mental models. The information that has to survive that chain includes patient identity and current condition, mechanism of injury or diagnosis, treatments already given, airway/breathing/circulation and monitoring status, deterioration risks specific to flight, equipment and medication requirements, and what the receiving team must have ready on arrival. That last item is the R, and in a transfer it is the whole point.

Military health systems

The US Military Health System published a full SBAR Toolkit as part of its patient-safety resources — overview, user guide, facilitation guide, slides, handouts, quick-reference cards, action-planning guide and an evaluation form — on the stated basis that "use of a structured communication tool known as SBAR can improve information exchange among health care team members and reduce the rate of adverse events."

The past tense is deliberate: health.mil restructured its patient-safety section and the whole Toolkits branch, including this page, now returns a 404. The link above is the Internet Archive's capture of 2 June 2023, which is the version quoted here.

I flag the contents of that toolkit deliberately. Nine artefacts, of which exactly one is the mnemonic. The other eight are implementation. Hold that thought until section 7.

Civil aviation and other high-reliability settings

SBAR-like structure spread into civil aviation alongside crew resource management, and from there into fire and rescue, emergency medical services, industrial safety, utilities and electrical operations, maritime operations and control-room incident management. The common denominator is a team that must build a shared mental model quickly across a rank or specialty boundary, using speech, with a decision pending.

6. What the framework is actually doing

It is worth being precise about the mechanism, because "it improves communication" is not a mechanism.

It front-loads. Spoken language defaults to chronological narrative; SBAR overrides that with an inverted pyramid. The recipient can begin acting after the first sentence and refine afterwards.

It bounds the background. B is not the history — it is the history this decision needs. The instruction "only what bears on the situation you just stated" is what stops the call becoming a ward round.

It makes the sender's model explicit. This is the load-bearing letter. In an unstructured call the assessment is the thing most often withheld, either because the caller is unsure or because they feel it is not their place. A withheld assessment is information the system already possesses and cannot use.

It forces a request. Escalations fail silently when nobody states what they want. "I'm just letting you know" is not a handover; it is a transfer of anxiety.

It gives the receiver a predictable shape. Prediction is cheap attention. A listener who knows the next thing they will hear is the background does not have to hold open a slot for it.

It flattens the hierarchy — a little. This is the claim I would treat most carefully. A junior clinician who has been taught the format has a socially sanctioned script for saying something assertive to a senior one. The script does real work. It does not abolish a rank gradient, and organisations that believe a mnemonic has fixed their psychological safety problem have mistaken the symptom for the disease.

7. What the evidence actually shows

Here is where most SBAR teaching stops being honest, so let us do the opposite. The evidence base is real, mixed, and considerably more interesting than the induction slide.

The case for

The strongest single study is Randmaa et al. (2014) in BMJ Open — SBAR improves communication and safety climate and decreases incident reports due to communication errors in an anaesthetic clinic — a prospective intervention study with a comparison group across two Swedish anaesthetic clinics, covering theatres, ICU and post-anaesthesia care.

A drop from roughly a third of incidents to roughly a tenth, with a concurrent comparison group that did not move, is about as good as this literature gets.

The case for caution

The randomised trial is equivocal. Cunningham et al. (2012) in Postgraduate Medical Journal ran a randomised controlled trial of SBAR for telephone referrals by junior doctors in a simulated setting: 66 interns, two simulated referral scenarios, SBAR education delivered to the intervention arm between them, blinded raters. On the objective score — whether the critical data elements were actually conveyed, scored 1–12 — the SBAR group reached 8.5 against 8.0 for controls, p=0.051. Not significant. On the global rating of the call's impact, SBAR won clearly: 3.0 versus 2.0, p=0.003. Time to "first pitch" did not improve.

Read that carefully, because it is the most informative result in the whole literature. A single teaching intervention made the calls land better without reliably making them contain more. SBAR changed how the referral sounded more than what it carried. If your improvement plan is a one-hour lecture, that is your realistic ceiling.

The systematic reviews are hedged. Müller et al. (2018), Impact of the communication and patient hand-off tool SBAR on patient safety: a systematic review (BMJ Open 2018;8(8):e022202), found multiple studies supporting SBAR's effectiveness but concluded the evidence was moderate and that further research was needed on patient-safety outcomes specifically.

And the implementation finding is the one that should change practice. Lo, Rotteau and Shojania (2021), Can SBAR be implemented with high fidelity and does it improve communication between healthcare workers? A systematic review (BMJ Open 2021;11(12):e055247), reached a conclusion I would put on the wall of every quality-improvement office: fidelity with SBAR is highest in classroom settings, while studies in clinical contexts "either did not achieve sufficient improvements in fidelity or did not assess fidelity."

That is a two-part indictment. Part one: people use SBAR properly where they are being taught it and drift once they are not. Part two — and this is worse — most clinical implementations never measured whether anyone was using it. A great deal of what is published as "we implemented SBAR" is more accurately "we distributed SBAR."

What I take from this

The mnemonic is not the intervention. The mnemonic is cheap, memorable and, on its own, weakly effective. The intervention is the mnemonic plus deliberate practice, plus a locally adapted prompt at the point of use, plus measurement of whether calls actually contain the four elements, plus seniors who reinforce it by responding well when it is used. The Military Health System toolkit has nine components and one of them is the acronym. That ratio is roughly correct.

For context on why this is worth the trouble: CRICO Strategies' benchmarking analysis of 23,658 US malpractice cases filed between 2009 and 2013 found communication failure implicated in around 30%, associated with 1,744 deaths and $1.7 billion in costs — and provider-to-provider communication failures were more likely to result in payment than provider-to-patient ones. The Joint Commission devoted Sentinel Event Alert 58 (September 2017) entirely to inadequate handoff communication, recommending handoff training, leadership prioritisation and continuous measurement — and, notably, holding up I-PASS rather than SBAR as its worked example of a research-validated framework.

8. The variants, and which one to adopt

For clinical use I would default to ISBAR, and I would treat the I as non-negotiable. The four-letter version quietly assumes the recipient already knows who is speaking and about whom — an assumption that holds in a small unit where everyone recognises voices, and fails in exactly the large, cross-covering, multi-site settings where escalation errors actually happen. The first three seconds should establish who I am, where I am, and which patient this is, and everything downstream inherits that certainty.

Where the stakes are high enough to justify the extra seconds — drug doses, transfers, anything the receiver will act on without seeing the patient — add the read-back. I-SBAR-R is the maximal version and the closest to the closed-loop communication that aviation and the nuclear navy actually practise. A message is not delivered until the sender has heard it come back.

You will also encounter local variants, SABER among them, which add an element for expectations or the response required. I would treat these as organisational house style rather than established practice: they appear in institutional teaching materials but not in the peer-reviewed comparative literature, so adopt one because it fits your workflow, not because it is better evidenced. It is not.

One genuine alternative deserves naming. I-PASS (Illness severity, Patient summary, Action list, Situation awareness and contingency planning, Synthesis by receiver) was designed specifically for shift-to-shift handover of a patient list rather than for a single escalation call, and it has the better outcome evidence in that niche — which is why the Joint Commission cites it. The two are not competitors so much as tools for different jobs:

• ISBAR — one patient, one problem, one decision, usually urgent, often by telephone. Escalation, referral, deterioration.
• I-PASS — many patients, no single decision, transfer of an entire mental model at shift change. Note that its S is situation awareness and contingency planning — "if X happens overnight, do Y" — which SBAR has no slot for at all.

Using SBAR for a twenty-patient night handover is a common and quiet mistake. It has no mechanism for anticipatory guidance, and anticipatory guidance is most of what a night team needs.

9. Where SBAR stops

Four boundaries, in rough order of how often I see them crossed.

1. It is not a substitute for clinical reasoning. The A is a container; nothing about the format fills it correctly. A structured, confident, fluent transmission of a wrong assessment is more dangerous than a rambling transmission of the right one, because structure signals competence and buys the assessment unearned credibility. SBAR improves the fidelity of transmission and does nothing whatsoever for the quality of what is transmitted.

2. It is thin on contingency. Covered above: no slot for "what to do if." For anything forward-looking, extend it or use a framework built for the job.

3. It can become a barrier as easily as a bridge. Work in non-acute paediatric settings has asked directly whether SBAR functions as a bridge or a barrier to interprofessional collaboration, and the question is not rhetorical. A tool designed for urgent, decision-focused, hierarchical exchange can constrain the discursive, negotiated, multi-professional conversation that chronic and complex care actually depends on. Mandating a crisis-communication format for every interaction teaches people that all clinical conversation is a transaction with a request at the end. Much of it is not.

4. It does not survive contact with a culture that punishes escalation. If the predictable consequence of a well-structured 3 a.m. call is being told off, the framework will be abandoned within a fortnight and the incident report will say "communication failure." It will not have been.

10. If you are actually going to implement it

Compressed, from the evidence above rather than from enthusiasm:

• Teach ISBAR, not SBAR. The identification step is free and prevents the worst class of error.
• Give both words for R. Recommendation for those who can, Request for those who cannot yet.
• Practise in role-play against a resistant listener, not on a slide. The Cunningham trial suggests one exposure changes the delivery more than the content; content needs reps.
• Put the prompt where the call happens — the phone, the badge card, the ICU wall — not in the induction pack.
• Measure fidelity, or accept that you do not know whether you implemented anything. Audit twenty escalation calls against the five elements. This is the single step the Lo review found missing almost everywhere, and it is not difficult.
• Train the receivers too. SBAR is a two-person protocol taught almost exclusively to one of them. The senior who responds "thank you — say the assessment again?" is doing more for adherence than any poster.
• Use I-PASS for shift handover and stop asking SBAR to do a job it was not built for.

The framework is fifty years old in spirit and twenty-four in its clinical form. It survives because the underlying insight is durable and largely independent of medicine: when a decision is pending and attention is scarce, say the urgent thing first, say only what bears on it, say what you think it means, and say what you want.

The submarine service worked that out because the alternative was unacceptable. Medicine is still, in places, deciding whether it agrees.

Sources

• AHRQ TeamSTEPPS SBAR tool — the definition used above, including the Recommendation/Request ambiguity: ahrq.gov/teamstepps-program
• The founding healthcare paper — Leonard, M., Graham, S. and Bonacum, D. "The human factor: the critical importance of effective teamwork and communication in providing safe care," Quality and Safety in Health Care, 2004;13(Suppl 1):i85–i90. PMC1765783
• The naval attribution — BMJ rapid response to bmj.o2091, stating the Bonacum / nuclear-submarine lineage and the 2002 Kaiser Permanente date.
• Prospective intervention study — Randmaa, M., Mårtensson, G., Swenne, C.L. and Engström, M. "SBAR improves communication and safety climate and decreases incident reports due to communication errors in an anaesthetic clinic: a prospective intervention study," BMJ Open, 2014;4(1):e004268. doi:10.1136/bmjopen-2013-004268
• Randomised controlled trial — Cunningham, N.J. et al. "Telephone referrals by junior doctors: a randomised controlled trial assessing the impact of SBAR in a simulated setting," Postgraduate Medical Journal, 2012;88(1045):619–626. doi:10.1136/postgradmedj-2011-130719 — the source of the p=0.051 objective score and the p=0.003 global rating.
• Systematic review, patient safety — Müller, M., Jürgens, J., Redaèlli, M. et al. "Impact of the communication and patient hand-off tool SBAR on patient safety: a systematic review," BMJ Open, 2018;8(8):e022202. Summary via AHRQ PSNet
• Systematic review, fidelity — Lo, L., Rotteau, L. and Shojania, K.G. "Can SBAR be implemented with high fidelity and does it improve communication between healthcare workers? A systematic review," BMJ Open, 2021;11(12):e055247. Summary via AHRQ PSNet — the classroom-versus-clinic fidelity gap.
• Bridge or barrier — "The use of SBAR as a structured communication tool in the pediatric non-acute care setting: bridge or barrier for interprofessional collaboration?", Journal of Interprofessional Care, 2020. doi:10.1080/13561820.2020.1816936
• US Air Force aeromedical evacuation — Dufour, K.M. "Implementation of the SBAR Checklist to Improve Patient Safety in the United States Air Force Aeromedical Evacuation," Wright State University, 2012. corescholar.libraries.wright.edu
• Military Health System SBAR Toolkit — nine implementation artefacts, one of which is the mnemonic. The live page was withdrawn when health.mil restructured; cited here from the Internet Archive capture of 2 June 2023
• The IHI's SBAR tool — the Kaiser-derived guidelines and the one-page worksheet a clinician actually fills in before calling: ihi.org
• Handoff communication as a sentinel-event driver — Joint Commission Sentinel Event Alert 58, September 2017. Summary via AHRQ PSNet
• The malpractice figures — CRICO Strategies 2015 Annual Benchmarking Report, Malpractice Risks in Communication Failures: 23,658 cases filed 2009–2013, ~30% involving communication failure, 1,744 deaths, $1.7bn. Harvard RMF summary · AJMC breakdown by specialty
• Beyond healthcare — SBAR in utilities and electrical operations: incident-prevention.com

Drafted with AI assistance. Every statistic above is taken from the cited source rather than asserted, and the two systematic reviews are quoted for their hedges as well as their conclusions. The judgements — particularly the recommendation to default to ISBAR, the reading of the Cunningham trial, and the four boundaries in section 9 — are mine, as are any errors.