Sepsis recognition on the ward lessons from Hospital Medicine 2017

Sepsis rarely announces itself. By the time a general ward patient looks obviously unwell, the window for simple interventions has often closed, and the next step is an unplanned transfer to ICU. Hospital Medicine 2017 in Las Vegas dedicated substantial floor space to this problem, and Australian hospitalists who attended came home with a clearer sense of what works on Ward 6B of a suburban hospital rather than in a US tertiary referral centre.

The Australian context shapes how sepsis presents and how we catch it. Medical Emergency Team calls activate on objective criteria aligned with international definitions but flavoured locally. Standards set by the Australian Commission on Safety and Quality in Health Care push wards towards consistent vital sign documentation and explicit escalation pathways. Even with these scaffolds, recognising a patient who "just doesn't look right" remains the hardest variable to standardise.

Las Vegas sessions focused on pragmatic tools that survive contact with a busy ward. This article walks through those tools: screening triggers, early warning scores, bundle timing, and the cultural changes that make it stick. The aim is to translate conference discussion into something that suits a Tuesday morning huddle in Brisbane, Perth or Hobart.

The recognition gap on general wards

Sepsis on the ward rarely presents as textbook fever and tachycardia. The elderly patient recovering from a fractured neck of femur, the post-chemotherapy admission with subtle hypotension, or the person from a remote community can each deteriorate in ways that slip past routine observations. Conference speakers pointed to atypical presentations in immunocompromised, post-operative and older populations, where blunted febrile responses mask the clinical picture until compensation fails.

Australian data mirrors this. ANZICS registries show ward-originating ICU admissions for sepsis carry higher mortality than ED admissions, often because antibiotics and lactate arrive hours later. The NSQHS Standard on Recognising and Responding to Acute Deterioration addresses this directly, yet operationalising it on a 32-bed general medical ward at hospitals like Royal Adelaide or Westmead remains a daily challenge. The gap is rarely about knowledge; it is about cognitive load.

Speakers emphasised three recurring blind spots: relying solely on SIRS criteria, missing the trajectory of deterioration across a shift, and under-valuing nurse concern. Each maps directly to a tool discussed at the conference.

What the conference floor highlighted

The most visible theme in the exhibit hall was electronic health record-driven screening. Vendors showed automated algorithms that pull hourly observations, score them against an early warning system, and flag patients meeting locally configured sepsis criteria. The appeal for Australian IT directors is obvious: an algorithm running on existing infrastructure in hospitals like the Princess Alexandra in Brisbane or the Austin in Melbourne, without forcing nurses to learn a new paper chart.

Sessions revisited the qSOFA versus SIRS debate. Presenters argued qSOFA, intended as a prognostic tool for suspected infection, is poorly sensitive as a ward screening test. They recommended pairing it with a higher-sensitivity trigger such as the National Early Warning Score, then escalating to qSOFA to assess severity. Detailed bundle implementation coverage is on the main conference site, and practical takeaways around lactate timing and antibiotic choice drew the largest crowds.

The third strand was human factors. Workshops used simulation to show how ward teams miss sepsis when vital signs are taken less frequently than policy dictates, when handover omits a recent trend, or when the team adopts a "watch and wait" mindset. Recognition is as much a workflow problem as a clinical one.

Building a ward-based screening program

A working screening program has three layers. First, a trigger: defined vital sign thresholds, nurse concern, or a laboratory value prompting bedside assessment. Third, escalation: clear criteria for calling the MET, contacting the on-call registrar, or initiating the sepsis pathway. Middle, the bedside assessment itself, including focused examination, lactate measurement and blood cultures where appropriate.

Australian wards often default to the between-the-flags chart, which incorporates a colour-coded adult early warning system. Faculty endorsed this as a robust foundation but argued that adding an explicit sepsis trigger — two or more modified SIRS criteria plus nurse concern, or NEWS of 5 or more — improves sensitivity without overloading the system. The key was avoiding duplication: the sepsis trigger should sit on top of the existing chart, not beside it.

Trajectory matters too. A single borderline observation is easy to dismiss, but a patient whose respiratory rate climbs from 16 to 22 across four hours, whose systolic pressure drifts down by 20 mmHg, and whose urine output falls is a different story. Several electronic systems track this automatically, and the case for them is strongest when they integrate with the existing chart rather than creating a parallel workflow.

A practical look at common screening tools

Choosing a screening approach means balancing sensitivity, specificity and burden on frontline staff. The table summarises how the most commonly discussed tools performed in conference-presented data and published validation studies.

The table reinforces the conference's central argument: no score replaces clinical judgement. Strongest programs use NEWS or NEWS2 as the substrate, layer a sepsis-specific trigger on top, and reserve qSOFA for severity stratification once sepsis is suspected. Programs that failed tended to over-rely on one score, or to deploy electronic alerts without nursing engagement.

Connecting recognition to bundles and the first hour

Once sepsis is suspected, speed matters. Sessions on the Surviving Sepsis Campaign 1-hour bundle emphasised lactate measurement, blood cultures before antibiotics, broad-spectrum empiric antibiotics, fluid resuscitation for hypotension or elevated lactate, and vasopressor support if fluids fail to restore perfusion. The bundle is well known, yet Australian audit data shows median time-to-antibiotics on general wards consistently exceeds the one-hour target.

The barrier is rarely knowledge and almost always process. Workshops returned repeatedly to pre-prescribed antibiotic kits, rapid bedside lactate testing, and empowering nurses to draw blood cultures before registrar review. Practical guidance on implementing these bundles in a community hospital setting is discussed in bundle implementation insights, with case examples drawn from hospitals of varying size and resourcing.

A second recurring theme was "sepsis mimic" conditions. Pulmonary embolism, acute pancreatitis, adrenal insufficiency and severe diuretic-induced hypovolaemia can all trigger sepsis alerts. Faculty advised that bundles should be initiated while diagnostics proceed, then reviewed at four hours to confirm or redirect therapy. This dual approach respects both urgency and diagnostic uncertainty.

Tying sepsis detection to rapid response systems

Australia's Medical Emergency Team framework is one of the few areas where the country genuinely leads internationally. The challenge on general wards is that MET calls often activate late, after hours of gradual deterioration. Speakers presented root-cause analyses from their own institutions, summarised on the rapid response pattern review page, demonstrating that most calls for septic deterioration had at least one antecedent trigger documented in the chart but not acted upon.

Embedding sepsis triggers directly into the MET criteria creates a safety net. If the ward screen flags suspected infection with NEWS of 7 or more, the MET call should be automatic. Several presenters shared data showing automatic-trigger criteria reduced time-to-ICU transfer for ward-onset sepsis by between 30 and 90 minutes. That interval is clinically meaningful.

Equally important is post-event learning. Every MET call for sepsis should generate structured review: was the trigger recognised, was the bundle initiated, was antibiotics delivered in time, and what system factors contributed to delay? Hospitals that closed this loop, such as a regional Victorian service presenting at the conference, showed measurable improvement in subsequent quarters without additional staffing.

Training, simulation and cultural change

Tools fail when culture does not support them. The conference's most consistently repeated message was that sepsis recognition improves most when nursing staff feel psychologically safe to escalate, when junior medical officers are coached rather than reprimanded for calling early, and when senior clinicians model the behaviour they want.

Simulation-based education was a strong theme. Scenarios focused on the patient who looks "off" but does not meet strict criteria, the post-operative patient whose trajectory is worsening, and the family member who says "they are not themselves". These soft triggers, combined with objective data, formed the basis of education programs described at the conference. A tertiary hospital network in Western Australia shared their experience of in-situ ward simulation, which produced sustained improvement in time-to-recognition and time-to-antibiotics at twelve months.

Cultural change requires leadership. Clinical directors who review sepsis metrics monthly, ward pharmacists who flag delayed antibiotics, and educators who debrief after every MET call all contribute. The conference was clear that the return on these investments is measurable in lives, bed-days and dollars.

The practical takeaway is straightforward: pick one screening tool that integrates with your existing observation chart, configure it to trigger both bedside assessment and automatic MET escalation for high-risk patients, audit time-to-antibiotics monthly, and invest in simulation that builds confidence rather than blame. None of this requires new technology or a substantial budget — only the willingness to treat sepsis recognition as a system property rather than an individual achievement.