TL;DR
This study evaluates a pilot program in New York City that diverted low-acuity 911 calls directly to hospital-based telemedicine during the COVID-19 pandemic. While the program successfully reduced ambulance dispatches for a subset of patients, it revealed that patient acceptance of "virtual-first" emergency care is highly dependent on the perceived risk of the surrounding environment.
What Problem Does This Paper Address?
During the height of the COVID-19 pandemic, the NYC Emergency Medical Services (EMS) system faced unprecedented strain, with call volumes hitting record highs and ambulance response times slowing by over 30 minutes. Traditionally, 911 systems are "high-touch": an ambulance is dispatched, a provider evaluates the patient in person, and the patient is usually transported to an Emergency Department (ED).
The authors sought to test a "direct-to-telemedicine" diversion model. The goal was to determine if low-acuity callers (those with minor symptoms or suspected mild COVID-19) could be safely and effectively managed via remote triage (RT) without ever sending an ambulance. This represents a significant shift from the standard "dispatch-first" mentality of emergency medicine.
Methods
The study tracked 454 calls diverted from the NYC 911 system to a hospital-based triage line at Weill Cornell Medicine between April and October 2020.
The workflow functioned as follows:
- Initial Screening: Emergency Medical Dispatchers (EMD) used a computerized algorithm to identify eligible patients (aged 5–59, in a safe location, with low-acuity complaints like "fever and cough" or minor medical issues).
- Remote Triage (RT): Calls were transferred to a hospital hotline staffed by Nurse Practitioners (NPs) or Physician Assistants (PAs).
- Disposition: The RT provider could either:
- Dispatch an ambulance (if the patient seemed too sick or insisted on one).
- Provide medical advice over the phone.
- Facilitate a formal video-based telemedicine visit with an emergency physician.
- Follow-up: A Registered Nurse (RN) attempted to contact all patients within 48 hours to assess outcomes and subsequent healthcare utilization.
Datasets
The researchers used a primary dataset of 454 diverted 911 calls captured via a secure REDCap database. They compared characteristics and outcomes across two distinct phases:
- Early Phase (April–June 2020): High COVID-19 prevalence and strict lockdowns.
- Late Phase (June–August 2020): Declining COVID-19 cases and easing restrictions.
Key Insights and Ideas
- The "Fear Factor" Drives Acceptance: One of the most striking findings was that patient acceptance of telemedicine was inversely proportional to the safety of the outside world. During the peak of the pandemic, patients were more willing to stay home and use telehealth to avoid the "risky" hospital environment. As the pandemic eased, the odds of a patient insisting on an ambulance (despite being eligible for telehealth) increased nearly seven-fold (OR 6.8).
- The Triage Gap: 44% of the calls diverted by 911 dispatchers were immediately sent back for ambulance dispatch by the hospital-based NP/PA. This suggests that current EMD algorithms are not yet sensitive enough to replace a clinical "eye" or ear.
- Telemedicine Friction: Despite being diverted for a "telemedicine" program, only 8.6% of patients actually completed a full video visit. Most were managed with simple phone advice. This highlights the technical and workflow barriers to implementing video-based care in an emergency context.
- Safety vs. Acuity: 26.1% of the patients reached during follow-up were eventually admitted to the hospital. While this indicates the program successfully identified some sick patients, it also suggests that "low-acuity" as defined by a 911 dispatcher often masks underlying complexity that requires in-person care.
Key Results
- Ambulance Reduction: Among the calls deemed "appropriate" for transfer, 31.7% were successfully managed without an ambulance dispatch.
- Total Dispatches: Overall, 60.4% of the diverted calls still resulted in an ambulance being sent, either due to medical necessity or patient preference.
- Clinical Outcomes: 37 patients were eventually admitted to the hospital, with one requiring ICU-level care.
- Follow-up Success: The team successfully contacted 45.1% of patients for the 48-hour follow-up, providing a rare window into what happens to 911 callers after the initial contact.
Limitations and Future Work
The study was limited by its reliance on self-reported data during the remote triage process, as providers lacked access to objective vital signs (heart rate, oxygen saturation, etc.). The low follow-up rate (under 50%) also means some adverse outcomes may have been missed.
The authors suggest that for these programs to succeed long-term, we need:
- More sophisticated screening: Moving beyond simple "yes/no" algorithms to better identify who truly needs a physical ambulance.
- Public Education: Changing the "911 = Ambulance" mindset to "911 = Access to the right care."
- Hybrid Models: Integrating EMS providers on the ground with remote physicians (similar to the ET3 model) to combine physical assessment with specialist oversight.
Summary updated