Background: Dedicated hospitalists are associated with improvements in healthcare quality and patient outcomes, but are not feasible for small-volume critical access hospitals (CAHs). A telemedicine “virtual hospitalist” model may expand the capabilities of a CAH at a fraction of the cost of an on-site provider.
Purpose: Our purpose was to provide hospitalist coverage to remote CAHs using a remote, “virtual hospitalist” teleprovider. Virtual hospitalists were equipped with full access to the local electronic medical record and were supported by local nursing staff and advance practice providers.
The primary outcomes were the percentage of ED visits resulting in acute inpatient admission and the percentage of ED visits resulting in transfer to an outside hospital. Secondary outcomes included mean daily census, mean length of stay, and virtual hospitalist time commitment.
Description: Virtual hospitalists were initially deployed at a single CAH. After 6 months of operation, the percentage of ED visits resulting in local admission increased from 4.1 to 6.7% (P = 0.014). Outside transfers from the ED decreased from 16.6 to 10.5% of all ED visits (P < 0.001). Despite this, mean daily census did not significantly increase (1.98 vs. 2.13 inpatients per day, P = 0.513), likely due to a lower rate of admissions from the ED. There were no significant changes in length of stay, case mix index, or inpatient mortality. Virtual hospitalists reported an average time commitment of 65 minutes per day, suggesting the potential to scale the intervention across multiple sites simultaneously. After six months, we expanded coverage to two additional CAHs. At one we served as an additional general medicine service and at the other site we provided general medicine co-management of surgical patients. We also expanded to provide skilled nursing facility coverage at all three facilities. This increased our mean daily census from 2.13 to 5.61 and the reported daily time commitment up to 192 minutes. Although the mean daily time commitment was low, outliers in excess of 600 minutes per day were reported. The ability to spread a single provider further is also limited by the desire for timely rounding and responsiveness to acute patient events spread across multiple institutions. Additional analysis is required to see if we achieved similar increases in the percentage of patients treated locally at our expansion sites. Keys to success in our model included a shared electronic medical record across all institutions and partnerships with local advanced practice and nursing staff to supplement telehealth with high quality local exams. Potential barriers to implementation include familiarizing providers with the varying capabilities of each specific institution and communicating effectively when patients have hearing or visual impairment. The impact on local institutions is limited by the specific roles asked of the virtual hospitalist providers, and individualized metrics for success may be necessary as the service expands to take on additional roles at additional institutions.
Conclusions: Our virtual hospitalist model was successful in increasing the percentage of ED patients who could safely receive inpatient care locally. Virtual hospitalists were able to cover multiple CAHs simultaneously, but logistic barriers may prevent matching the patient ratios seen in traditional inpatient models.