Alternative venues handled a quarter of acute ED care without routine hospital admission
A large retrospective cohort from Kaiser Permanente Northern California shows that noninpatient treatment venues are already carrying a meaningful share of acute care once considered hospital-bound. Among 26,618 emergency department encounters for heart failure, pneumonia, cellulitis, urinary tract infection, COVID-19, or anemia, one-quarter were transitioned to an ambulatory treatment center, an advanced care at home program, or an emergency department observation unit rather than a standard inpatient bed.
The cohort was older, medically complex, and heavily concentrated in a population that often strains emergency and inpatient capacity. Median age was 77 years, 52.6% were female, and most patients had only one encounter during the study period, which ran from January 1, 2024, through June 30, 2025. Those details matter because they place the findings in a real-world setting where clinicians are deciding, often quickly, which patients need admission and which can safely complete treatment elsewhere.
Escalation was uncommon in the most ambulatory settings
The clearest signal came from the need to move patients back into the hospital. Inpatient admission rates were 4.3% for ambulatory treatment center encounters and 5.3% for advanced care at home encounters, compared with 28.0% for emergency department observation unit encounters. The contrast is stark and clinically intuitive: patients placed into more deliberately outpatient-oriented pathways were far less likely to deteriorate enough to require inpatient escalation than those first routed through observation status.
A single-sentence reading of the data gets to the practical bedside point. Patients selected for ambulatory treatment center or home-based acute care usually stayed there.
That pattern does not imply the alternative venues were treating trivial illness. Heart failure, pneumonia, cellulitis, urinary tract infection, COVID-19, and anemia all can destabilize older adults quickly, but they also include a subset of patients whose oxygenation, hemodynamics, response to antibiotics or diuresis, and follow-up needs can be managed without a conventional ward bed when systems are built for rapid reassessment.
Short-term revisits were not markedly higher outside the hospital
Rates of 7-day ED return visits were similar across care settings. They were 10.4% after ambulatory treatment center encounters, 5.3% after advanced care at home encounters, 9.2% after emergency department observation unit encounters, and 8.5% after hospital encounters. The spread does not suggest a large penalty for moving care out of the hospital, at least over the short horizon measured here.
The biology behind this model is straightforward even if the logistics are complex. For selected patients, the initial problem is not an uncontrolled disease state requiring continuous inpatient monitoring; it is a condition that benefits from prompt treatment, close surveillance, and a low threshold to adjust therapy. Delivering intravenous medications, monitoring symptoms, and arranging near-term follow-up in a less resource-intensive setting can be sufficient when the illness trajectory is predictable and the patient can be safely observed outside a bedded unit.
Mortality and 30-day hospitalization remained lower than in hospital-treated encounters
The longer-horizon outcomes were also favorable in the alternative venues. Thirty-day mortality was 2.1% for ambulatory treatment center encounters, 1.7% for advanced care at home encounters, and 2.8% for emergency department observation unit encounters, compared with 5.3% for hospital encounters. Thirty-day hospitalization followed the same direction: 11.4%, 8.3%, and 12.7% across the three noninpatient settings, versus 15.3% after hospital care.
These are unadjusted rates, so they should not be mistaken for causal proof that the venue itself lowered death or readmission. Selection almost certainly played a role. Patients chosen for home-based or ambulatory pathways were likely those judged stable enough to avoid the ward, while hospitalized patients probably carried greater severity, frailty, or physiologic instability at baseline. Even so, the absolute event rates are important because they show that these pathways were not associated with obvious excess short-term harm in this integrated system.
A shift in resource allocation rather than a universal replacement for admission
The study’s most consequential message may be structural rather than procedural. Among 22,755 unique patients, 87.4% had one encounter, 9.8% had two, and 2.8% had more than two, suggesting that the system was reaching a broad outpatient population rather than a narrow repeat-user subgroup. Within that population, 728 encounters occurred in ambulatory treatment centers, 880 in advanced care at home, 5,133 in emergency department observation units, and 19,877 in hospital settings.
Those numbers suggest a tiered care architecture already functioning at scale. Emergency departments and inpatient teams were not simply diverting low-risk patients away from admission; they were actively sorting acute illness into the venue best matched to monitoring intensity, treatment needs, and anticipated recovery time. For congestive heart failure, infection, and anemia in particular, the line between inpatient and noninpatient management often depends less on diagnosis alone than on symptom burden, lab trends, social support, transport reliability, and the ability to reassess quickly if the patient worsens.
What this means for systems trying to preserve hospital beds for the sickest patients
In a health system with integrated membership, shared records, and organized follow-up, alternative venues can absorb a substantial fraction of acute care demand. The potential operational value is obvious: hospital beds can be preserved for patients with refractory hypoxemia, hemodynamic instability, complicated infection, severe decompensated heart failure, or other problems that truly require inpatient infrastructure.
The study also points to a narrower clinical question that every health system adopting these pathways must answer locally: which patients can be shifted safely, and how should escalation thresholds be enforced when symptoms evolve after the initial ED decision? The current data do not resolve that question, because the cohort design reports unadjusted outcomes rather than a randomized comparison. Differences in baseline severity, social risk, and clinician selection remain unavoidable sources of confounding.
Still, the signal is strong enough to matter for practice planning. In this cohort, novel treatment spaces were not fringe options; they accounted for a meaningful share of ED-to-completion care, and short-term outcomes were not worse than hospital care on the measures reported here. The remaining uncertainty is not whether these venues can function, but how to define the safest patient selection rules as adoption expands beyond a single integrated delivery system.