PHOENIX — The use of tele-ICU in a system of urban safety net hospitals resulted in shorter hospital stays, stable ventilator use, and stable median ICU stays 18 months after the program was launched, according to a retrospective cohort study presented on October 20 at the American College of Chest Physicians (CHEST) 2026 Annual Meeting in Phoenix, Arizona.

“Mortality models showed no credible increase in mortality,” Chidinma Chima-Melton, MD, MBA, a pulmonologist and medical director of telehealth at the University of California Los Angeles, told attendees. “Tele-ICUs can be implemented in resource-constrained systems without increasing invasive ventilatory support or inducing system-level deterioration.”

While tele-ICU programs have expanded to address shortages of intensivists, the evidence on these programs is sparse and heterogenous, often with only short-term studies that do not account for preexisting mortality trends or consider safety net systems that serve high-risk populations, Chima-Melton said.

She had previously reported findings that a tele-ICU program reduced ICU length of stay at 5 months after its September 2024 launch in an urban safety net system with four hospitals in Los Angeles. The new data she was presenting was an evaluation at 18 months after launch.

The four hospitals had no in-house intensivists at launch and 62 total ICU beds, with 93.2% of admissions publicly insured. They ran a closed ICU model where remote intensivists directed care and wrote orders with 24/7 coverage, including structured night rounds and contiguous intensivist scheduling. Additional quality improvement included standardized workflow, protocoled order sets, holding family meetings within 72 hours, and remote nursing education. 

The analysis included all 6330 ICU admissions from November 2022 to February 2026. The pre-implementation period included the 22 months from November 2022 to August 2024. The prespecified ramp period was the 6 months from September 2024 to February 2025, and post-implementation maturation period included the 12 months from March 2025 to February 2026.

Baseline characteristics of the patients were relatively similar across the full time period in terms of age, sex, facility, and mortality risk. Medicare representation in payers slightly increased from 22% pre-intervention to 27% post (P = .001), and mortality index score slightly decreased from 19% to 18% (P = .015).

At 18 months after launch, geometric mean hospital lenth of stay was 7.1 days, an approximately 19% decrease from 8.8 days pre-implementation (P = .032). ICU length of stay remained steady at 5 days pre- and post-implementation, and mechanical ventilation exposure was similar before (44.1%) and after (44.2%) the program’s launch. 

Crude rates of mortality rose nonsignificantly during the ramp-up and showed no significant increase through maturation.

Mortality trends did differ, however, by hospital. There was no significant difference in mortality at Huntington Park, Coast Plaza, or Gardena, but odds of mortality significantly dropped by approximately 6% at East LA Doctors Hospital (odds ratio, 0.936, P = .006).

Cost of implementation for this program was low, Chima-Melton said, because they built the system on existing technology at the hospitals. “They were already using some telephone formulary consultations to ER [emergency room] programs, so the actual technology investment was going to be very minimal,” she said.

Rachael W. Starcher, MD, MHI, a pulmonary, critical care, and clinical informatics physician with Respiratory Consultants in Robbinsdale, Minnesota, viewed the presentation and found the study “compelling.”

“Particularly with regards to safety net hospitals, you have to consider the costs of implementing technology and weigh that with the possible risks and benefits that patients will potentially face,” Starcher said. “With tele-ICU implementation, upfront costs can be huge. However, the way they utilized easily available technologies and leveraged staff training with standardized protocols allowed them to improve access to high-quality ICU care without breaking the bank.”

Despite that, Starcher said the modest outcome results made her question whether the safety net hospitals in this system were already operating at a high level before the intervention began, potentially explaining why they did not see more robust improvements in ICU metrics.

“With any tele-ICU implementation, outcomes are going to be highly dependent upon the hospitals at the spokes and hub, staff training and protocol interventions, and the gap you are trying to fill with telecommunications,” Starcher said. “For them, my biggest concern is that perhaps there wasn’t a gap that needed filling in the first place. Only their full and final analysis will help answer that question.

Chima-Melton acknowledged that the study’s observational design had no concurrent control group, and residual confounding and uncertainty limit any claims regarding causation or safety.

“These findings address key gaps in prior literature by incorporating extended follow-up, accounting for preexisting mortality trends, and modeling facility-level heterogeneity within a high-risk safety net population,” Chima-Melton and her colleagues wrote. She advised evaluation frameworks for other tele-ICU programs to pre-specify ramp periods and rely on longitudinal modeling instead of early comparisons of pre- and post-implementation.

No external funding for the research was noted. Chima-Melton and Starcher had no disclosures. 

Tara Haelle has been an independent science/health journalist for over 15 years, with work appearing in over two dozen trade and consumer publications. She is author of Vaccination Investigation: The History and Science of Vaccines and The Informed Parent: A Science-Based Guide to the First Four Years. Tara is the Public Health Beat Leader for the Association of Health Care Journalists, where she develops resources to help other journalists report on medical studies and public health. She is based in Dallas.