Utilization of Robotic “Remote Presence” Technology Within North American Intensive Care Units

Abstract

Objective: To describe remote presence robotic utilization and examine perceived physician impact upon care in the intensive care unit (ICU). Study Design: Data were obtained from academic, university, community, and rural medical facilities in North America with remote presence robots used in ICUs. Objective utilization data were extracted from a continuous monitoring system. Physician data were obtained via an Internet-based survey. Results: As of 2010, 56 remote presence robots were deployed in 25 North American ICUs. Of 10,872 robot activations recorded, 10,065 were evaluated. Three distinct utilization patterns were discovered. Combining all programs revealed a pattern that closely reflects diurnal ICU activity. The physician survey revealed staff are senior (75% >40 years old, 60% with >16 years of clinical practice), trained in and dedicated to critical care. Programs are mature (70% >3 years old) and operate in a decentralized system, originating from cities with >50,000 population and provided to cities >50,000 (80%). Of the robots, 46.6% are in academic facilities. Most physicians (80%) provide on-site and remote ICU care, with 60% and 73% providing routine or scheduled rounds, respectively. All respondents (100%) believed patient care and patient/family satisfaction were improved. Sixty-six percent perceived the technology was a “blessing,” while 100% intend to continue using the technology. Conclusions: Remote presence robotic technology is deployed in ICUs with various patterns of utilization that, in toto, simulate normal ICU work flow. There is a high rate of deployment in academic ICUs, suggesting the intensivists shortage also affects large facilities. Physicians using the technology are generally senior, experienced, and dedicated to critical care and highly support the technology.

Introduction

Telemedicine in the intensive care unit (ICU) (tele-ICU) dates back 2–3 decades,^1,2 with more rapid growth in the last decade following commercial development.³ Tele-ICU includes complex systems like Philips/VISICU (Andover, MA),^3
–5 Cerner (Kansas City, MO),⁶ and IMDsoft (Needham, MA)⁷ and systems from Polycom (Pleasanton, CA), Tandberg (Oslo, Norway), and GlobalMedia (Scottsdale, AZ). Alternatively, “remote presence,” using medical robotics (InTouch Health, Santa Barbara, CA),^8,9 provides unique capabilities that add value in the chaotic environment of the ICU. From 2000 to 2010, a single design (Philips/VISICU) largely drove growth in the tele-ICU arena. As a consequence, the majority of the tele-ICU literature has evolved from the Philips/VISICU technology. However, recently, there has been significant growth with other vendors and technologies. As of 2011, robotic technology known as “remote presence” has been installed in more than 450 hospitals (information obtained from InTouch Health). The purpose of this article is to provide the first large-scale review of robotic “remote presence” technologies in the ICU.

The site from which ICU telemedicine services are delivered has been designated the tele-ICU.^10

–13 Growth of the tele-ICU has been driven by manpower maldistribution,^14

–17 shortage of intensivists,^18

–22 quality-of-care issues,^23

–26 success of the tele-ICU–structured care model,^27

–32 and measurable positive outcomes.^3,4,33
–35 More recently, a large study from the University of Massachusetts³⁶ indicated substantial improvement in mortality of patients managed with tele-ICU technology, largely the result of adherence to established standards of care. Following national political events in 2009 regarding healthcare reform, telemedicine has taken greater prominence as part of this evolution.³⁷ Furthermore, with a growing shortage of physicians expected to top 125,000 by 2020–2025,³⁸ maximizing efficiency of physicians is mandatory to include greater use of telemedicine technologies. Finally, with recommendations for further study of the tele-ICU concept,³⁹ we believe it is imperative to examine other modalities for the delivery of remote care in the ICU.

Process

Technology

The technology used is a semi-autonomous, Internet-enabled, real-time, two-way audiovisual telecommunications platform that moves about in a wireless environment. The devices are casually referred to as “robots” providing “remote presence.” The robots are under the control of a remote clinician who can drive the device to the bedside, maneuver around the bed, perform visual physical examination, coach and observe a “surrogate”⁴⁰ examiner, and review graphical information on monitors, ventilators, balloon pumps, etc. (Fig. 1).

Fig. 1.

(A) The remote clinician is controlling the robot at the bedside of a critically ill patient. (B) The monitor on the robot shows the face of the remote clinician performing real-time activities. Staff at the bedside can function as the “surrogate” examiner.

The “robots” are enhanced with electronic stethoscopes (Fig. 2 shows a telephonic stethoscope from TeleHealth Technologies [Viera, FL]) and may be further enhanced with lights to illuminate the retina, pharynx, ear canal, or tympanic membrane. In some cases, the devices have been enabled to transmit bedside ultrasonography in real time.

Fig. 2.

“Robot” with electronic stethoscope mounted on back. Use requires an on-site assistant such as a nurse to place the stethoscope over the heart, lungs, airways, or vascular structures.

The robots are controlled real-time by a remote physician at a “control station.” The control stations may be either portable or stationary. Portable control stations are laptop computers enhanced with necessary software, cameras, and headsets. Additionally, the laptops have “joy-sticks” for real-time control of the robot body and head and its direction of movement. All laptops are further equipped with WiFi cards, and some are further enhanced with cellular broadband capability.

Nonportable systems are traditional desktop computers with the above capabilities and generally with dual-view or triple-screen systems. Desktop systems generally have simultaneous access to hospital clinical information system and digitized radiography.

To provide care, networks are developed to accommodate local manpower supply and staffing patterns. Generally, one of two network patterns may be deployed: “centralized” or “decentralized” systems.^40,41 In general, the centralized programs supported multiple facilities, whereas the decentralized programs may have had the physicians rendering care to a single hospital.

Connectivity between the robot and hospital network is via an IEEE 802.11a, b, or g WiFi. Connectivity between physician control stations and hospital networks uses the bidirectional User Datagram Protocol (UDP). When institutional firewalls block UDP connections, communications servers broker the connection using Transmission Control Protocols. Data are encrypted with AES 256-bit symmetric encryption protocols. Video is captured at 30 frames/s and optimally requires 600 kilobits per second (Kbps) both up and down but will provide satisfactory performance at 400 Kbps and unsatisfactory performance at less than 250 Kbps. Optimization of available bandwidth and compression/decompression (CoDec) is provided by corporate proprietary methodology. Video CoDec is H.264 compliant and standards based. All control stations have echo-cancellation technology.

The remote presence robot is compliant with the Food and Drug Administration (FDA): Medical Device Data System Final Rule. Specifically, the device complies with FDA Class II requirements for devices that transmit, store, or display medical device information intended for immediate clinical action by healthcare providers.⁴²

Data Collection

The Fleet Monitoring System (FMS) is a proprietary (InTouch Health) corporate tool for tracking utilization and functional characteristics of remote presence robots in real-time, 24 h/day for every deployed device. The FMS provides status reports for robots and control stations by monitoring characteristics like battery strength, signal strength, software version, most recent use, most recent failure, and last report to the FMS. The FMS also monitors utilization and supports a database search to tabulate frequency of use by care provider, date and time, duration, etc. Finally, the FMS alerts the remote technical support team to take corrective action when a device experiences a failure such as a loss of connectivity or has reached a low battery level.

Remote presence robotic utilization and deployment were reviewed for a 12-month interval, from June 1, 2009 to May 31, 2010, via the FMS. Only robots known to be exclusively deployed in ICUs were studied. Users designating utilization primarily for stroke, emergency medicine, acute long-term care, translation services, or undefined services were excluded from review. Data were analyzed by facility, robot, month, and time of day. The study group was restricted to users accounting for 90% of the activations (the bottom 10% were not studied based on the assumption that they were in the training phase of technology acquisition). Data are presented by time of utilization. All data were de-identified as to facility or site.

Physician Survey

A Web-based survey was performed using “Survey Monkey” (Survey Monkey, Palo Alto, CA). The survey had 30 listed questions, with some of the questions having as many as seven embedded questions. The survey was validated internally. The questions were divided into six general categories: (1) Demographics of Tele-ICU Remote Presence Users; (2) General Utilization of Remote Presence; (3) Description of ICUs with Deployed Robots; (4) Application and Impact of Remote Presence in the ICU; (5) Physician Perspective and Views; and (6) an Optional section for identification.

Users and directors of the designated ICUs with remote presence robots deployed were identified. The identified individuals were e-mailed three times at weekly intervals and finally 4 weeks later. When possible, phone calls were made directly to nonresponders. Data were analyzed with Survey Monkey proprietary software.

Results

FMS Data

Review of the FMS data (June 1, 2009–May 31, 2010) revealed that there were approximately 350 remote presence robots deployed in North America; of these, 56 robots were dedicated to ICUs in 25 North American facilities. Because of the rapid growth of the market, some robotic devices have been included in the survey but only recently installed and therefore have minimal or no patient-related utilization. When there was less than 12 months of utilization, despite relatively high utilization for a short period (training and mock activations), the programs were excluded from analysis (Program 10). Programs outside North America were eliminated. Therefore, only the facilities with higher-level utilization were analyzed, which limits the analysis to 13 programs.

One program represents six facilities (Program 1), and another represents five facilities (Program 3). Otherwise, programs have one or two robots deployed and represent one or two facilities. Data are presented per robot. For example, Program 1 had 2,612 activations over six robots. Therefore, the average from Program 1 calculates to 47 sessions/month/robot. In summary, the programs included are Programs 1–9, 11, 13, 15, and 16.

During this 12-month period, there were 10,872 robotic sessions. Once the bottom 10% plus new/developing programs were excluded, 13 programs performed 10,065 activations (92.5%). The general overview of robotic utilization is presented in Figure 3.

Fig. 3.

Intensive care unit–designated use of remote presence robotic technology presented as average number of sessions per month per facility. All specific facilities were de-identified.

Upon review of each facility, three patterns of use emerged. The first pattern is usage primarily during the early nighttime hours as seen in Figure 4. Two different facilities with almost identical patterns of usage yet substantially different programmatic structure were discovered: Program 2 is a large urban, tertiary academic medical center, whereas Facility 6 is a small urban, private, nonacademic hospital with intensivists in-house only during daytime hours only.

Fig. 4.

Total number of sessions, expressed as utilization by time of day, for (A) Facility 2, a large, West Coast, tertiary, academic teaching hospital, and (B) Facility 6, a small, East Coast, urban primary care hospital. Peaks are in the early evening hours, suggesting remote review of patient activity in the absence of on-site staffing, scheduled evening rounding, or academic teaching/supervisorial rounds.

The second utilization pattern is high level during the daytime hours with minimal to no utilization during the evening and nighttime hours. Of the 13 institutions reviewed, 6 institutions fit the daytime-only utilization. Five of these six programs are large academic tertiary medical facilities. Figure 5 demonstrates one of the six programs with a typical daytime usage pattern.

Fig. 5.

Total number of sessions, expressed as utilization by time of day, for Facility 4, a large, Northeast, tertiary academic hospital. The daytime-only utilization pattern of remote presence technology, peaking during typical maximal intensive care unit activity, suggests that the remote presence is used to support on-site staff during times of maximal load.

The third pattern of utilization is 24-h coverage. The 24-h pattern was discovered in five programs. Two of the programs have 24/7 contractual relationships with multiple facilities to cover multiple ICUs. These two contractual programs provided the two highest utilization rates of 2,612 and 2,199 tele-ICU sessions per year. Another program (Program 9) is the center of a statewide program providing acute services, and Program 15 is a multi-institutional program. Figure 6 demonstrates the near-continuous nature of remote presence utilization. Of note, even in the “continuous coverage” pattern, there is a very substantial slowdown during the hours from midnight to 5:00 a.m.

Fig. 6.

Total number of sessions, expressed as utilization by time of day, for (A) Facility 1, a multi-institutional, nonacademic, North American program, and (B) Facility 3, a private, commercial, West Coast, nonacademic, intensivist coverage program. Both exhibit Pattern 3, with broad usage both midday and into evening and night hours. Note the different y-axis scale for Facility 3.

When all utilization data from all programs are combined, a pattern emerges of diurnal variation with maximum utilization midday (from approximately 9:00 to 16:00 h) and a second rise in the evening (approximately 20:00 through 23:00 h). In all patterns, the utilization is remarkably low during the hours from 00:00 to 05:00 h (Fig. 7).

Fig. 7.

Overall remote presence robotic utilization in the intensive care unit by time of day combining data from “high users” during a 12-month period. Note the diurnal pattern of variation and minimal utilization during the hours from midnight to 5:00 a.m.

Closer analysis of the utilization patterns segregated by “day-shift,” “evening-shift,” and “night-shift” has been performed in Table 1.

Table 1.

Breakdown of Robotic Activations by Shift

	ACTIVATIONS
PROGRAM	6:00 A.M.–5:59 P.M. (“DAY-SHIFT”)	6:00 P.M.–11:59 P.M. (“EVENING-SHIFT”)	12:00 A.M.–5:59 A.M. (“NIGHT-SHIFT”)	OVERALL ORGANIZATIONAL UTILIZATION PATTERN
1	1,872	603	137	24-h
2	179	312	20	Evening-shift
3	1,596	468	135	24-h
4	294	33	0	Day-shift
5	1,776	85	48	Day-shift
6	186	430	5	Evening-shift
7	127	32	1	Day-shift
8	195	0	0	Day-shift
9	193	39	9	24-h
11	274	26	5	Day-shift
13	164	38	0	Day-shift
15	403	105	7	24-h
16	144	89	4	24-h
Totals	7,403 (74%)	2,260 (22%)	371 (4%)

Each program is defined based upon pattern as “evening-shift,” “day-shift,” or “24-h” coverage.

Overall, the vast majority of the utilization of remote presence occurs during the “day-shift.” On the other hand, the utilization during the “night-shift” is extremely small, at 4% of the total utilizations. Even within the two commercial programs (Programs 1 and 3), with contractual agreements for 24/7 coverage, the “night-shift” utilization is 5% and 6%, respectively.

Physician Survey

The physician survey yielded the following results. For the 25 institutions, 56 individuals were polled with 22 respondents.

Demographics

Of the survey respondents, approximately two-thirds of the care providers were trained in internal medicine, and one-fourth had surgical training. Over 80% had additional training in straight critical care medicine, and about one-sixth were pulmonary trained. Over 75% of the physicians were over 41 years of age. Approximately 60% of the care providers had more than 16 years in the general practice of medicine, while nearly 50% had been focused strictly in the field of critical care for 16 years or more. Regarding focus in critical care medicine, 59% dedicated 75–100% of their clinical practice to critical care medicine. In summary, the clinical staff is more likely internal medicine based, critical care trained, senior, and highly dedicated to critical care medicine.

Utilization

Over 70% of the respondents have been using remote presence robotic devices for more than 3 years, with half of those more than 5 years. Over 70% use only one device. Only one respondent indicated utilizing both “Robots and Carts.” Eighty percent of the physicians initiated remote presence robotic care from home, and additionally 17 of 22 respondents also initiated remote presence robotic care from a medical facility, suggesting that many have multiple control stations located at multiple sites.

In general, 87% of remote presence physicians are located in cities with a population >50,000, but surprisingly 80% of remote presence patient encounters are also in cities >50,000. Two-thirds of the clinicians provide service to only one ICU, but another combined 26% provide coverage to two or three ICUs. In summary, the remote presence programs are relatively matured and operate in a decentralized architecture from multiples sites. For the most part, remote presence physicians are located in larger cities, and remote care is provided in cities that are at least midsized.

ICU Descriptors

Of the primary sites where remote critical care services are delivered, 46.6% are either academic centers or university affiliates. Most of the remaining 53% are described as “community” facilities. As programs reach further out, there is a progressive shift to fewer or no academic facilities and predominately community facilities. Only three facilities were identified as critical access hospitals. Functionally, the vast majority of the ICUs were described as “Mixed Medical Surgical,” whereas only two sites were listed as neurosurgical and two as trauma. Of the 15 respondents, only 1 program was covering five ICUs simultaneously, with 33% covering two or more, 25% covering three or more, and only 20% covering four or more. Table 2 shows the spread of covered ICU beds and average number of beds.

Table 2.

Beds Covered by Robotic Remote Presence Among the 15 Respondents

COVERED ICU	CUMULATIVE NUMBER OF BEDS	RANGE OF ICU BEDS	AVERAGE NUMBER OF ICU BEDS COVERED/ROBOT	% OF RESPONDENTS
Primary	167	1–28	11.3	100% (15/15)
Second	54	2–19	10.8	33% (5/15)
Third	58	6–36	14.5	25% (4/15)
Fourth	17	1–12	5.6	20% (3/15)
Fifth	4	4	4.0	7% (1/15)

ICU, intensive care unit.

The survey further revealed that within these ICUs, at the primary (first listed) ICU, 80% of the respondents provide both on-site and remote care. In summary, a significant amount of remote care is provided to academic facilities and/or university affiliates. Otherwise, essentially all care is rendered at community facilities and generally not in specialty care units. A majority of physicians provide remote and on-site care at the primary covered ICU, but the further removed the ICU is, the more likely that care will only be delivered remotely.

Application and Impact of Remote Presence in the ICU

Sixty percent of the respondents indicated that they use the robotic remote presence technology for routine “rounds” either very often or regularly. When rounds are performed, they are “scheduled” (73%), and over one-half of these scheduled rounds are in the evening. Otherwise, there was an even spread of rounding for “work,” “collaboration,” “teaching,” or “safety” rounds. Urgent response was performed either very often or regularly in 53%. Using the robotic technology for education was evenly spread across the spectrum, from 21% “very often” to 21% “never.” Family meeting were judged as “sometimes” among 53%. Two of the respondents indicated using the robotic technology for ICU “Triage” and “Consultation.”

Regarding perceptions of impact, 100% of respondents believe that remote patient care is either “very valuable” or “valuable” in improving patient care and patient/family satisfaction. Seventy-eight percent believe there was an associated reduction in ICU length of stay. Ninety-three percent (93%) believe remote presence improves compliance with best practices, and 87% feel that it improves nursing education. Impact upon physician quality of life was less certain, with 43% believing they had experienced an improvement, whereas 50% recognized no benefit. The data have been included in Table 3.

Table 3.

Respondents' Perception of Impact of Remote Presence Robots in the Intensive Care Unit

OUTCOMES AND FUNCTIONAL AND SOCIAL VARIABLES	VERY VALUABLE (%)	VALUABLE (%)	NO IMPACT (%)	DETRIMENTAL (%)
Improving patient care	64.3	35.7	0.0	0.0
Reducing patient LOS	28.6	50	21.4	0.0
Improving adherence to best practice	50	42.9	7.1	0.0
Improving on-site staff education/skills	46.7	40	13.3	0.0
Improving patient satisfaction	23.1	76.9	0.0	0.0
Improving family satisfaction	42.9	57.1	0.0	0.0
Improving physician quality of life	28.6	14.3	50	7.1

LOS, length of stay.

Physician Perspective and Views

The vast majority of physicians (87.5%) perform robotic remote presence without additional compensation. Sixty-six percent (66%) perceived the robotic technology as a “blessing,” 33% see the technology as a mixed “blessing and burden,” while none perceived the robotic technology as only a “burden.” Despite these perceptions, 100% of the respondents intend to continue to deliver services by robotic remote presence.

Discussion

To date, this represents the largest study of the applications of “robotics” in the ICU. This study of robotic utilization is designed to develop a descriptive picture of how robots are currently being deployed in ICUs. The FMS-derived utilization represents relatively hard data derived from an electronic database. However, the data only report the activation of the robotic device and do not offer insight into what is being done once it is activated. Therefore, it was imperative to add the user's description of the critical care environments and how the devices were being used. In addition, we have added the subjective element of how the user views the impact upon patient care, patient/family interactions, education, and impact upon the physician.

This is not an efficacy study. There are no objective data reflecting changes in mortality, morbidity, or length of stay. Despite the lack of objective data, there is a stunningly high and consistent impression from the care providers of improved patient care, shortened length of stay, greater adherence to best practices, and impact upon social issues such as patient and family satisfaction. This study did not query whether there are internal hospital data to support such impressions. Knowing that facilities in the United States are mandated by regulatory agencies^43,44 to collect, analyze, and retain quality data, it is presumed that internal data do exist that could support clinicians' impression of positive impact upon care.

Regarding the FMS utilization data, we have recognized three different utilization patterns based upon visual recognition. The pattern described as the “evening-shift” is most recognizable and objective, with very little daytime activity and a significant peak in the early evening hours. A “day-shift” pattern is recognized with little evening or nighttime activity and a dominant peak in utilization during daytime hours. The third pattern is characterized by 24-h activity. There may be some overlap between the “day-shift” pattern and the “24-h” coverage pattern. In certain cases, the pattern blurred between these two groups, and a judgment call was made by at least two observers/co-authors. The clinical significance of the various patterns is unclear except that the pattern paints a picture of how the robots are being used. In fact, it is unknown if the patterns reflect clinical needs or reflects programmatic design to satisfy some study or “pilot” evaluation.

What is remarkable is the high utilization of robotics by “day-shift” programs at academic and university affiliate centers, environments that would be assumed to be more than adequately staffed. In this review, approximately 50% of the covered facilities were academic centers. Perhaps this speaks to the ongoing shortage of specialists in critical care despite the otherwise well-staffed environments. In support of this concept, Lafit et al.⁴⁵ reported, in abstract form, an evaluation of robotics in an environment known to be heavily staffed and yet demonstrated a positive impact when remote clinicians provided an additional remote structured review of “best practices.”

When all the utilization data are combined, a distinct diurnal pattern evolves with two peaks of utilization: morning and evening. The ICU typically functions around an established diurnal flux of activity. Morning rounds with the intensivists, consultants, etc., lead to substantial activity as the days' plans are activated. However, the evening peak of activity may be driven more by nursing than physicians. ICUs are typically staffed with 12-h nursing shifts. With the transition to the evening, nursing staff typically review patients again and develop questions, leading to a secondary flux of activity.

The robotic utilization pattern imitates this waxing and waning of ICU activity, suggesting that the remote presence clinicians are closely mimicking on-site reality or needs. Furthermore, while nursing activity remains high during the late night,⁴⁶ physician activity seems to be minimal during the late night hours from midnight to 5:00 a.m. (Table 1). This may be pertinent from the perspective of administrators staffing a remote tele-ICU. Specifically, staffing could be heaviest during the day shift, intermediate during the evening shift, and “skeleton” for the night shift. Despite multiple articles in the literature describing work flow,^47
–49 to our knowledge, this is the most descriptive pattern of physician work flow or demands in the ICU.

In review of the physician survey data, staff working in a robotic remote presence–based “tele-ICU” are more likely internal medicine based, more senior by years of practice, and highly focused on critical care medicine. The greater adoption of the tele-ICU technology by senior staff has been recognized before⁵⁰ and may represent different clinical commitments in different generations, different views of the systems by senior staff, or lack of financial incentive for the younger staff. Most of the programs are also relatively mature, existing generally more than 3–5 years, and generally are single-ICU programs operating a single robot. All but one “tele-ICU” programs are decentralized, in that “rounds” or consultations may be provided from a multitude of different sites. This contrasts with the Philips/VISICU, Cerner, or iMDSoft programs with large, centralized tele-ICUs. The Web or Internet-enabled technology of the robotic remote presence devices supports an “open architecture” network⁴¹ versus the more commonly used “closed architecture” of the Philips/Cerner/iMDSoft programs.

Regarding where services are delivered, remote care is more often delivered to moderately large cities (population >50,000) and generally to community facilities. Critical access hospitals, generally thought to be the most likely to deploy such remote technology, actually represented a minority of sites. We have previously observed and reported⁵¹ a similar phenomenon in that a federally designated “Medically Underserved” facility may exist within a large urban environment, literally blocks from major medical centers. In fact, surprising numbers of facilities were “academic” or “university affiliated,” suggesting the locations would be urban. Finally, most were described as “Mixed Medical-Surgical” ICUs, while very few were specialty ICUs.

Physician work patterns indicated that most provide remote presence to a single ICU and that over 80% provide both on-site and remote care to those ICUs. Most of the rounds are scheduled and in the evening. This may suggest that the same physician who knows the patient by day could be performing some sort of routine or scheduled follow-up on patients in the evening. If this is indeed the case, the physician carries the patient “database” mentally with him or her. Therefore, large databases such as those found with the VISICU/Cerner/iMDSoft systems, which are necessary for staff covering unknown patients, would be less necessary for this cadre of physicians.

Physician perceptions of impact are highly positive despite lack of objective data. It is unknown whether there are internal hospital data supporting these perceptions, and the survey failed to interrogate for such supporting data. However, all these programs require financial investment and are unlikely to continue to exist without some sort of “value added.” Most facilities will search for a return on investment and may demand hard outcomes such as reduced length of stay but may also accept soft data such as patient/family/staff satisfaction. Therefore, it is assumed that value to the facility exists, but we failed to establish that value in an objective fashion. In final summation of physician perspectives, 33% reported the technology as a mixed “blessing and burden,” none reported getting extra compensation for robotic remote presence rounds, yet 100% intend to continue to deliver medical care via this technology.

A primary intent of this study was to determine the number of remote patient encounters or beds covered. We felt it important to generate an estimate to provide the reader with a general sense of the actual magnitude of remote patient encounters. To do so, the following assumptions have been made to create lower and upper estimates:

1. During the day shift, all patients are seen in each ICU.

2. During evening rounds, only one-half of the patients are seen.

3. During the night shift, only one patient is seen per activation.

4. The ICU has 80% occupancy.

The upper estimate is then generated using an upper average number of beds covered as 10 (rounded upper averages in Table 2). The lower average is taken as 5 (rounded lower average from Table 2) and is shown in Table 4.

Table 4.

Estimated Patient Encounters During 1 Year of Robotic Remote Presence Utilization

ACTIVATIONS	ASSUMPTION	PATIENT ENCOUNTERS USING UPPER AVERAGE NUMBER OF BEDS (N=10)	LOWER AVERAGE NUMBER OF BEDS (N=5)
7,403 daytime activations	100% ICU patients seen, 80% occupancy	59,224	29,612
2,260 evening activations	50% ICU patients seen, 80% occupancy	9,040	4,520
371 nighttime activations	1 ICU patient seen	371	185
Estimated total patient encounters		68,560/year	34,317/year

ICU, intensive care unit.

Of note, critical care services may be delivered in a variety of sites within the hospital to include the ICU, Emergency Department, and postanesthesia care unit or on the general medical or surgical wards, particularly during a “Rapid Response.” This review of utilization of the robotic technology for critical care services was designed only to assess critical care services performed within the ICU and therefore will underestimate the total volume of critical care services.

Conclusions

The robotic remote presence has substantial utilization in the ICU environment. The utilization includes a wide spectrum of types of ICUs to include large academic and referral centers and community centers. It is surprising that very few “critical access” hospitals are covered. The time-utilization pattern seems to mimic the natural ebb and flow of routine on-site activities of an ICU, suggesting that the robotic remote presence technology is supplementing typical day-to-day ICU activities.

The physicians involved in robotic remote presence tend to be more senior and have a high dedication to the field of critical care medicine. The physicians tend to be internal medicine based with additional critical care medicine training. There is a strong belief in impact on both clinical and social issues. Despite receiving no extra salary for services rendered by the robotic remote presence, most intend to continue using the robots.

The networking structure is greatly different from the larger centralized “hub and spoke,” closed architecture, networks. Clinicians using robotic remote presence technology tend to use a reverse hub-and-spoke configuration with a decentralized, open architecture network and may work from multiple different sites. Despite the absence of hard data, the number of actual patient encounters seems quite substantial.

Footnotes

Acknowledgments

The authors would like to thank Intouch Health for access to the data from the Fleet Monitoring System.

Disclosure Statement

E.M.R. had a 3-month employment as a summer intern with InTouch Health and has had no subsequent financial or working relationship. A.G., T.W., and M.F. are full-time administrative employees of InTouch Health. H.N.R. is a non-paid member of the Board of Clinical Advisors of InTouch Health and is a full-time staff member of the R. Adams Cowley Shock Trauma Center.

References

Grundy

, Jones

, Lovitt

. Telemedicine in critical care: Problems in design, implementation, and assessment. Crit Care Med, 1982; 10:471–475.

Grundy

, Crawford

, Jones

et al. Telemedicine in critical care: An experiment in health care delivery. JACEP, 1977; 6:439–444.

Breslow

, Rosenfeld

, Doerfler

et al. Effect of a multiple-site intensive care unit telemedicine program on clinical and economic outcomes: An alternative paradigm for intensivist staffing. Crit Care Med, 2004; 32:31–38.

Rosenfeld

, Dorman

, Breslow

et al. Intensive care unit telemedicine: Alternate paradigm for providing continuous intensivist care. Crit Care Med, 2000; 28:3925–3931.

Breslow

. Remote ICU care programs: Current status. J Crit Care, 2007; 22:66–76.

Cerner. Critical care. www.cerner.com/solutions/Hospitals_and_Health_Systems/Critical_Care/. 2011 July 26.

McCambridge

, Jones

, Paxton

et al. Association of health information technology and teleintensivist coverage with decreased mortality and ventilator use in critically ill patients. Arch Intern Med, 2010; 170:648–653.

New England Healthcare Institute. Tele-ICUs: Remote management in intensive care units. Cambridge, MA: Massachusetts Technology Collaborative and Health Technology Center, 2007.

Garingo

, Friedlich

, Tesoriero

et al. The use of mobile robotic telemedicine in the neonatal intensive care unit. J Perinatol, 2012; 32:55–63.

10.

SCCM. MyICUCare. www.myicucare.org/Critical_Care_Questions/Pages/default.aspx. 2011 July 27.

11.

Chu-Weininger

, Wueste

, Lucke

et al. The impact of tele-ICU on provider attitudes about team work and safety climate. Qual Saf Health Care, 2010; 19:1–5.

12.

Young

, Chan

, Cram

. Staff acceptance of tele-ICU coverage. A systemic review. Chest, 2011; 139:279–288.

13.

Forni

, Skeham

, Hartman

et al. Evaluation of the impact of a tele-ICU pharmacist on the management of sedation in critically ill mechanically ventilated patients. Ann Pharmacother, 2010; 44:432–438.

14.

Health care workforce distribution and shortage issues in rural America. Kansas City, MO: National Rural Health Association, 2003.

15.

Zawada

, Kapasla

, Herr

et al. Prognostic outcomes after the initiation of an electronic telemedicine intensive care unit in a rural health system. South Dakota Med, 2006; 59:391–393.

16.

Zawada

, Aaronson

, Herr

, Erickson

. Relationship between levels of consultative management and outcomes in a telemedicine intensivist staffing program (TISP) in a rural health system [abstract] Chest, 2006; 130:226S.

17.

Zawada

, Herr

, Erickson

, Hitt

. Financial benefit of a teleintensivist program to a rural health system. Chest, 2007; 132:444.

18.

Angus

, Kelley

, Schmitz

, Whiet

, Popovich

. Current and projected workforce requirements for care of the critical ill and patients with pulmonary disease: Can we meet the requirements of an aging population? JAMA, 2000; 284:2762–2770.

19.

Health Resources and Services Administration report to Congress: The critical care workforce. A study of the supply, demand for critical care physicians. http://bhpr.hrsa.gov/healthworkforce/reports/criticalcare/default.htm. 2012 June 17.

20.

Kelley

, Angus

, Chalfin

et al. The critical care crisis in the United States: A report from the profession. Crit Care Med, 2004; 32:1219–1222.

21.

Grover

. Critical care workforce: A policy perspective. Crit Care Med, 2006; 34,3 Suppl:S7–S11.

22.

Krell

. Critical care workforce. Crit Care Med, 2008; 36:1350–1353.

23.

Reynolds

, Haupt

, Carlson

. Impact of critical care physician staffing on patients with septic shock in a university medical care unit. JAMA, 1988; 260:3446–3450.

24.

, Phillips

, Shaw

et al. On site physician staffing in a community hospital intensive care unit: Impact on test and procedure use and on patient outcome. JAMA, 1984; 252:2023–2027.

25.

Young

, Birkmeyer

. Potential reduction in mortality rates using an intensivist model to manage intensive care units. Eff Clin Pract, 2000; 3:284–289.

26.

Engoren

. The effect of prompt physician visits on intensive care unit mortality and cost. Crit Care Med, 2005; 33:727–732.

27.

Patel

, Kao

, Thomas

et al.

Improving compliance with surviving sepsis campaign guidelines via remote electronic ICU monitoring [abstract]

Crit Care Med, 2007; 35:A275.

28.

Giessel

, Leedom

. Centralized, remote ICU intervention improves best practices compliance [abstract] Chest, 2007; 132:444a.

29.

Youn

. ICU process improvement: Using telemedicine to enhance compliance and documentation for the ventilator bundle [abstract] Chest, 2006; 130:226S.

30.

Badawi

, Shemmeri

. Greater collaboration between remote intensivists and on-site clinicians improves best practice compliance [abstract] Crit Care Med, 2006; 34:A20.

31.

Rincon

, Bourke

, Ikeda

. Centralized, remote care improves sepsis identification, bundle compliance and outcomes [abstract] Chest, 2007; 132:557S–558S.

32.

Aaronson

, Zawada

, Herr

. Role of a telemedicine intensive care unit program (TISP) on glycemic control (GC) in seriously ill patients in a rural health system [abstract] Chest, 2006; 130,Suppl:226S.

33.

Vespa

, Miller

, Hu

, Nenov

, Buxey

, Martin

. Intensive care unit robotic telepresence facilitates rapid physician response to unstable patients and decreased costs in neuroIntensive care. Surg Neurol, 2007; 67:331–337.

34.

Meyer

, Raman

, Hemmen

et al. Efficacy of site-independent telemedicine in the STRroke Doc Trial. A randomized, blinded, prospective study. Lancet Neurol, 2008; 7:787–795.

35.

Youn

. Utilizing robots and an ICU telemedicine program to provide intensivist support for rapid response teams [abstract A8] Chest, 2006; 130:102s.

36.

Lilly

, Cody

, Zhao

, Landry

, Baker

, McIIwaine

, Chandler

, Irwin

. Hospital mortality, length of stay, and preventable complications among critically ill patients before and after tele-ICU reengineering of critical care processes. JAMA, 2011; 301:2175–2178.

37.

Bashshur

, Shannon

et al.

National telemedicine initiatives: Essential to healthcare reform (Policy White Paper for the American Telemedicine Association)

Telemed J E Health, 2009; 15:1–11.

38.

Dill

, Salsberg

. The complexities of physician supply and demand: Projections through 2025. AAMC Center for Workforce Studies, November 2008. www.aamc/initiatives/workforce/reports/. 2011 October 13.

39.

Kahn

, Hill

, Lilly

et al. The research agenda in ICU telemedicine. A statement from the Critical Care Collaborative Societies. Chest, 2011; 140:230–238.

40.

Reynolds

. The tele-ICU: Why, how and what. What the future may look like. Latifi

, Poropatich

, Hadeed

. Telemedicine for trauma, emergencies, and disaster management. Norwood, MA: Artech House Publishers, 2011; 303–330.

41.

Reynolds

, Rogove

, Bander

, McCambridge

, Cowboy

, Niemeier

. A working lexicon for the tele-ICU: We need to define tele-ICU to grow and understand it. Telemed J E Health, 2011; 17:773–783.

42.

Food and Drug Administration, Department of Health and Human Services21 CFR Part 880 [Docket No. FDA-2008-N-0106] (formerly Docket No. 2007N–0484)Fed Register, 2011; 76,31:8637–8649.

43.

The Joint Commission. Specifications manual for national hospital inpatient quality measures. www.jointcommission.org/specifications_manual_for_national_hospital_inpatient_quality_measures/. 2011 July 24.

44.

Centers for Medicare & Medicaid Services. Quality initiatives—General information. www.cms.gov/QualityInitiativesGenInfo/. 2011 July 24.

45.

Latif

, Romig

, Barasch

, Pronovost

, Sapirstein

. A consultative tele-ICU service improves compliance with best practice guidelines in a highly staffed ICU. Presented at the 40th Society of Critical Care Medicine Symposium: San Diego, CA, 2011.

46.

Tamburri

, DiBrienza

, Zozula

, Redeker

. Nocturnal care interactions with patients in critical care units. Am J Crit Care, 2004; 13:102–112quiz 114–115.

47.

Ogle

, Copley

, Berthune

, Parkin

. Intensive care workforce and workflow. Literature review. July 2004. www.pdfio.com/k-19898.html. 2011 July 27.

48.

Hashimoto

, Bell

, Marshment

. A computer simulation program to facilitate budgeting and staffing decisions in an intensive care unit. Crit Care Med, 1987; 15:256–259.

49.

Doyle

. Changing patterns of work. BMJ, 1991; 303:982–984.

50.

Reynolds

, Reynolds

. The non-adopter phenomenon. A case where being “senior” may enhance technological adoption [abstract] Telemed J E Health, 2010; 16,Suppl 1:S-81.

51.

Reynolds

. Underserved does not necessarily mean rural: Remote telepresence in downtown Baltimore [abstract] Telemed J E Health, 2008; 14,Suppl 1:103.