A Health System-Based Critical Care Programwith a Novel Tele-ICU: Implementation, Cost,

and Structure Details

Spyridon Fortis, MD, Craig Weinert, MD, MPH, Robyn Bushinski, MA, NE, PHN, RN,Alison Greiner Koehler, MHA, Greg Beilman, MD, FACS

BACKGROUND: Improving the efficiency of critical care service is needed as the shortfall of intensivists isincreasing. Standardizing clinical practice, telemedicine, and organizing critical care serviceat a health system level improves outcomes. We developed a health system Critical Care Pro-gram based at an academic medical center. The main feature of our program is an intensivistwho shares on-site and telemedicine clinical responsibilities. Tele-ICU facilitates the stan-dardization of high-quality critical care across the system. A common electronic medicalrecord made the communications among the ICUs feasible. Combining faculty from medicaland surgical critical care divisions increased the productivity of intensivists.

STUDY DESIGN: We retrospectively reviewed the administrative database data from 2011 and 2012, includingmean census, number of transfers, age, sex, case mix index, mortality, readmissions, andfinancial data.

RESULTS: The Critical Care program has 106 adult ICU beds; 54 of those beds can be managedremotely using tele-ICU based at the main University hospital. The mean midnight censusof the system for 2012 was 69.44 and total patient-days were 34,406. The capital cost of thetele-ICU was $1,186,220. The annual operational cost is $1,250,112 or $23,150 permonitored ICU-bed. Unadjusted mortality was 6.5% before and 4.9% after implementation(p < 0.0002).

CONCLUSIONS: We describe a novel health system level ICU program built using “off the shelf” technologybased on a large University medical center and a tele-ICU with a full degree of treatmentauthority across the system. (J Am Coll Surg 2014;219:676e683. � 2014 by the AmericanCollege of Surgeons)

Management of critically ill patients by intensivists im-proves outcomes1 including mortality2,3 and ICU lengthof stay.4,5 The Society of Critical Care Medicine recom-mends that an intensivist, usually unit-based, have theauthority to intervene and directly care for critically illpatients in urgent and emergent situations 24 hours perday, 7 days per week.6

The demand for critical care services in the UnitedStates is anticipated to increase due to aging of the

Disclosure Information: Nothing to disclose.

Received February 11, 2014; Revised April 15, 2014; Accepted April 29,2014.From the University of Minnesota Medical Center-Fairview (Fortis,Weinert, Bushinski, Koehler, Beilman), Critical Care and Acute CareSurgery Division (Fortis, Beilman), and Pulmonary and Critical CareDivision (Fortis, Weinert), University of Minnesota, Minneapolis, MN.Correspondence address: Spyridon Fortis, MD, University of MinnesotaMedical Center-Fairview, 420 Delaware St SE, MMC 276, Minneapolis,MN 55455. email: [email protected]

676ª 2014 by the American College of SurgeonsPublished by Elsevier Inc.

population and the 35% shortfall of intensivists that isanticipated by 2030.7 New methods are needed toimprove the efficiency of critical care service deliveryand increase the number of critically ill patients thatcan be managed around-the-clock per each intensivist.Recent innovations have the potential to improve the

efficiency of care for patients in the ICU. These includethe electronic medical record (EMR), computerizedphysician order entry, and use of standardized proto-cols.8,9 Another technological development with potentialfor improving care is the use of telemedicine that canremotely provide care to critically ill patients located incommunity hospitals.10

Military traumas systems have improved their clinicalperformance mainly by using evidence-based practiceguidelines and telemedicine under the umbrella of a largetrauma system.11 Evidence-based guidelines have reducedthe variation in clinical practice in military trauma, and

http://dx.doi.org/10.1016/j.jamcollsurg.2014.04.015

ISSN 1072-7515/14

Abbreviations and Acronyms

CMI ¼ case mix indexEMR ¼ electronic medical recordFHS ¼ Fairview Health SystemMICU ¼ medical ICUSICU ¼ surgical ICUUMMC ¼ University of Minnesota Medical Center

Vol. 219, No. 4, October 2014 Fortis et al Health System-Based Critical Care Program 677

telemedicine allows the providers to discuss their patientsacross the continuum of care.11 Trauma patients are trans-ferred within the trauma system from the battlefield tothe tertiary hospital when necessary to escalate theircare. Civilian hospitals are increasingly organized in verti-cally integrated systems to improve care but also to alignwith economic and regulatory incentives to reduce healthcare costs.12

The Fairview Health System (FHS), like many others,has increasingly moved toward a stronger interaction be-tween hospitals of the system to improve economies ofscale and to share lessons learned about best practices inpatient care and other areas. As the ICUs of the FHSbecame increasingly busy, and as more patients werecared for by one group of intensivists, both administratorsand intensivists believed that there were considerable ben-efits to be gained by efforts to more effectively use theICU resources across the system. Using the example ofthe US Military Joint Theater Trauma System,13 in whichpatients are cared for within one system across 3 conti-nents, we sought to set up, although on a much smallerscale, an ICU program to provide needed patient careat the appropriate level across the FHS. This programincluded components that allowed appropriate optimiza-tion and standardization of patient care through multidis-ciplinary team efforts. We will describe its components.

METHODS

Critical Care Program hospitals

The Fairview Critical Care Program includes 5 hospitalsand 7 ICUs in FHS. Fairview operates 7 hospitals in

Table 1. Characteristics of Fairview System Intensive Care Un

Hospital ICU beds, n Da

UMMC-E 52 Daytime acad

UMMC-W 6 Nurse practit

Lakes 10 Hospitalist þNorthland 4 Hospitalist þRidges 12 Hospitalist þSouthdale 22 Intensivist þUMMC-E, East bank campus of University of Minnesota Medical Center (mMedical Center.

Minnesota with 22,000 employees, 74,649 acute careadmissions in 2012, and annual revenue of $3.22 billion.The hospitals of the program are Lakes, Northland,Ridges, Southdale, and the East and West bank campusesof University of Minnesota Medical Center (UMMC-Eand UMMC-W). The UMMC-E is the primary teachinghospital of the University of Minnesota Medical Schooland is operated by Fairview and administratively linkedwith the UMMC-W. The hospital characteristics are inTable 1 and the components of the Fairview CriticalCare Program are summarized in Table 2.

History of Critical Care Program development tocurrent state

For many years, the ICUs of FHS, apart from those inUMMC-E, functioned without fellowship-trained inten-sivists. Fairview-employed hospitalists and private prac-tice internal medicine, general surgery, or subspecialtyphysicians managed ICU patients. At UMMC-E, medicalICU (MICU) and surgical ICU (SICU) were staffed bytraditional academic teams including medical students,residents, and fellows and were led by a faculty attendingduring the day. At night, residents managed ICU patientswith fellows or attendings taking call from home andcoming to the hospital if needed. In 2008, SouthdaleHospital contracted with University of Minnesota Physi-cians (ie, a faculty practice plan of the Medical School) toprovide dedicated intensivist services from 7 AM to 11 PMwith intensivist call from home after 11 PM with rapidability to return to the ICU to deliver face-to-face careif needed. Hospitalists are also available to provide carefor emergencies. In 2009, Ridges Hospital contractedwith University of Minnesota Physicians to provide on-site intensivist services for 4 hours in the morning and24-hour availability by phone and return when needed.All intensivists were full-time faculty from the Depart-ments of Medicine and Surgery at University of Minne-sota with fellowship training in medical or surgicalcritical care. In 2011, all FHS hospitals started using acommon EMR (Epic). In October of 2011, an in-house

its

ytime care model Nighttime care model

emic intensivist þ trainees On-site intensivistioner þ intensivist Tele-ICUopen ICU Tele-ICU

open ICU Tele-ICU

Intensivist Tele-ICU

open ICU Tele-ICU

ain hospital); UMMC-W, West bank campus of University of Minnesota

Table 2. Components of the Fairview Critical Care Program

Component Roles/representation

Critical Care Program leadership Medical director of the Fairview ICU system, medical and nursing directors of tele-ICU,nursing directors of medical ICU and surgical ICU at UMMC-E and project manager

Physician leadership group Chiefs of Divisions of pulmonary and critical care and surgical critical care at UMMC,directors of ICUs from all hospitals

Steering committee Critical Care Program leadership, CNOs from all hospitals, presidents of Southdale andUMMC-E, VP nursing of UMMC-E, medical directors of medical ICU and surgicalICU at UMMC-E

ICU physician group Peer review, practice issues, guidelines, break down specialty barriers

Tele-ICU Active patient monitoring at all sites, data gathering for PI, nurse, and physician education,“eyes on ground”

Medical informatics Quality measures, technology management (eg, Epic, cameras, computer issues)

Critical Care Program working group Multidisciplinary, multisite working group, data sharing, PI, protocol, and order setdevelopment and implementation

Research and education ICU for hospitalist course, anesthesia rotations, pulmonary and surgical critical carefellowship programs, funded ICU research

CNO, chief nursing officer; PI, performance improvement; UMMC, University of Minnesota Medical Center; UMMC-E, East bank campus of University ofMinnesota Medical Center (main hospital); VP, Vice President.

678 Fortis et al Health System-Based Critical Care Program J Am Coll Surg

intensivist service, staffed by faculty alternating everyweek between the 2 departments, began serving as theadmitting attending for both MICU and SICU atUMMC-E from 6 PM to 7 AM.The need for intensivist direction in the other ICUs

within Fairview, especially at night, became apparent.At that time, the two options were to contract for tele-ICU coverage from an out-of-state vendor or build ourown tele-ICU program. To reduce the per-bed tele-ICU cost to the remote hospitals, at the end of 2011,the nighttime intensivist at UMMC-E started serving asthe tele-intensivist using a newly built tele-ICU hub.Costs to the remote ICUs were lowered because 66%of the nighttime intensivist compensation is paid byUMMC-E and the program did not have to use a com-mercial tele-ICU care management program in additionto the existing EMR, thereby avoiding licensing fees.All critical care providers have educational responsibil-

ities at UMMC through residency or fellowship programsand many of them work at the bedside in 2 or 3 of theFairview ICUs. They also share nighttime ICU dutiesat UMMC-E and tele-ICU responsibilities. Combiningthe staff of the departments reduces the nighttime callcommitments to 1 to 3 in-house nights per month and2 nights per month taking call from home with availabil-ity to come to the bedside at Southdale or Ridges for newadmissions or for issues requiring direct intensivistpresence.

Critical Care Program leadership

The medical director of the Fairview ICU system, themedical and nursing directors of the tele-ICU, the

nursing directors of MICU and SICU at UMMC-E,and the project manager of the Fairview Critical Careprogram meet twice a month. The ICU leadership com-municates with the individual ICUs and reports to thehospital and system leadership group.

Physician leadership group

The ICU leadership meets every 3 months with the physi-cian leadership from all the hospitals (Division Chiefs ofpulmonary and critical care and acute care surgery/criticalcare at UMMC and ICU directors from all hospitals) todiscuss the practice of critical care medicine, staff, andcontract issues.

Steering committee

The ICU leadership brings issues for discussion to thesteering committee, which meets every 3 months. Hospi-tal and system executives and ICU leadership constitutethe steering committee (Table 2), which is responsiblefor the management of financial aspects and approvesthe decisions of ICU system leadership.

Intensive care unit physician group

The ICU physician group is composed of academic phy-sicians from the Departments of Anesthesiology (n ¼ 1),Medicine (n ¼ 21), and Surgery (n ¼ 10) in the MedicalSchool and convenes every other month. The goal of thisgroup is to minimize differences in clinical practice byadopting common diagnostic and treatment protocols.This is especially important as the training history andclinical practice of medical and surgical intensivists canresult in different approaches to clinical problems that

Vol. 219, No. 4, October 2014 Fortis et al Health System-Based Critical Care Program 679

could confuse the users of the tele-ICU program. Theyalso share their expertise to optimize protocols.

Critical Care Program working group

The meeting site for this group rotates among all the tele-ICU hospitals to facilitate face-to-face communicationand to gain familiarity with the physical layout of themonitored ICUs. To improve and standardize criticalcare practice throughout the system, ICU leadership as-signs tasks to multidisciplinary working groups composedof physicians, nurses, pharmacists, nutritionists, andadministrative employees from the hospitals. Topicshave included standardizing vasoactive medicationdosing, glucose control, mechanical ventilation bundle,sedation protocols, pain assessment, and ICU admissionorder sets. The group regularly monitors “customer satis-faction” by surveying the medical and nursing staff thatinteracts with the tele-ICU program.

Tele-intensive care unit program

The cornerstone of our Critical Care Program is the tele-ICU, which provides support to the ICUs of 5 Fairviewhospitals, but not for UMMC-E ICUs (Table 1). Thetele-ICU command center (hub) is a 444-square footroom located adjacent to the SICU at UMMC-E. Atele-ICU registered nurse (tele-RN) is in the hub 24hours per day and does not have bedside patient care re-sponsibilities when in the hub. All tele-RNs maintainactive bedside ICU practice at other times. The mainduties of tele-RNs are patient monitoring every 2 or 4hours, depending on severity of illness, updating patientlists, and phone call assistance for nursing issues or triag-ing to tele-intensivists for medical issues. Because of lowerpatient care needs during the daytime, the tele-RN alsohas quality-improvement duties, such as abstractingcompliance data from the EMR as requested by the work-ing group.The tele-intensivists are full-time faculty in the Medical

School and have ongoing bedside critical care practices atthe ICUs within FHS. The daytime tele-intensivist iseither the MICU or SICU attending, who is also leadingthe traditional fellow-resident-student team. Becausethere is an on-site intensivist for all the ICUs excepttwo (Lakes and Northland), the tele-intensivist workload is low during the day, allowing the MICU orSICU attending to have this dual role. At night, theintensivist also has two roles: admitting and attendingphysician for both MICU and SICU (assisted by over-night medical or surgical residents with fellows takingcall from home) and tele-intensivist for the rest of the sys-tem’s critical care beds. Based on a work-frequency anal-ysis we performed for 2 weeks in 2013, the UMMC-E

and tele-ICU time commitment is split about equallythroughout the night. The tele-intensivist has permissionfor full “treatment authority” for all patients in the moni-tored ICUs and can call the at-home intensivist to cometo the bedside for issues that cannot be managedremotely. Mechanisms for provision of needed services(eg, intubation, central-line placement) vary betweeneach site. We have identified mechanisms to provide theseservices 24 hours daily. The back-up intensivist can bephysically present for unstable patients and new ICUadmissions.

Medical informatics

An important element of the Critical Care Program is thecommon EMR (Epic), which allows immediate access tovital signs, flow sheets, medication lists, laboratory results,and radiology images in the same EMR that clinicians useduring bedside care and ambulatory care. We created atele-ICU module within Epic that aggregates patientsfrom all the Fairview ICUs (except for UMMC-EICUs) into a single patient list with flow sheets thathave simple “one-click” buttons indicating when a tele-ICU patient was reviewed and which interventions wereperformed. There is also a free-text progress note sectionthat is prominently titled as “Tele-ICU” so the daytimeclinicians can read about patient care issues overnightand the interventions made by the tele-ICU.Fifty-four ICU rooms have moveable and zoom-

capable cameras (Video Guidance) controlled from thehub. When the video system is on, the patient’s televisionmonitor displays a video image of the tele-ICU clinicians,allowing two-way video and audio conversation betweenthe tele-ICU and patients, family, or bedside nurses.Real-time ECG telemetry, blood pressure, respiratoryrate, and oxygen saturation are also displayed in thetele-ICU hub (SpaceLabs monitors) with capability to re-view ECG recordings for up to 24 hours in the past. Tele-ICU staff use 3 Epic workstations, 2 monitors for viewingpatient rooms remotely, and 1 high-resolution radiologymonitor.

Education and research

Apart from sharing clinical responsibilities, the Depart-ments of Medicine and Surgery organize education activ-ities, such as critical care fellowship rotations and thecombined monthly medicine-surgery critical care confer-ence for staff, fellows, residents, and students. Our Crit-ical Care Program also provides education for cliniciansoutside of the departments. In the last 2 years, >50 prac-titioners have participated in a 4- or 5-day “Critical Carefor the Hospitalist” course. In addition, for the last 12years, there has been a joint, 1- or 2-day, annual

Table 3. Mean Daily Census and Total Patient Days perYear Monitored by the Tele-Intensive Care Unit Physician in2012

HospitalsMean ICUcensus, n

Total patient-daysper year

Tele-ICU (off-site) 32.64 16,973

Northland 1.46 865

Lakes 4.32 2,413

Ridges 8.48 4,654

Southdale 15.97 7,705

UMMC-W 2.4 1,336

UMMC (on-site) 36.81 17,433

Total 69.44 34,406

UMMC, University of Minnesota Medical Center (main hospital);UMMC-W, west part of University of Minnesota Medical Center.

Table 4. Capital and Annual Operational Cost of Tele-Intensive Care Unit

Cost, $

Capital cost

Spacelab monitors (vitals, ECG, etc) 323,000

Facilities (build out of the hub) 290,000

Cisco systems for the hub (cameras in the hub) 4,000

Computer monitors (hub) 8,000

Video guidance charges (set up, maintenance) 1,277

Video guidance 54 cameras in ICU rooms 414,456

Network costs 144,000

Imaging computers 1,487

Total 1,186,220

Total cost per ICU bed 21,967

Operational annual cost

Tele-ICU nurse 550,200

Intensivist* 547,675

Administrative cost 152,237

Total 1,250,112

Total cost per ICU bed 23,150

*The tele-intensivist cost was subsidized by 66% from medical ICU andsurgical ICU for on-site service at the East bank campus of University ofMinnesota Medical Center (main hospital). Tele-ICU nurse is in the tele-ICU hub 24 hours per day.

Table 5. Patient Outcomes Across Fairview Intensive CareUnits

Outcomes 2011 2012 p Value

ICU admissions, n 6,063 6,097

Age, y, mean � SD 58.1 � 19.2 58.2 � 19.7 0.9774Female, n (%) 2,876 (47.4) 2,894 (47.5) 0.6258

CMI � SD 5.69 � 6.1 5.89 � 6.01 0.0559Death, n (%) 394 (6.5) 299 (4.9) 0.0002

ICU readmissions, n (%) 54 (0.89) 29 (0.49) 0.0064

CMI, case mix index.

680 Fortis et al Health System-Based Critical Care Program J Am Coll Surg

continuing medical education course that includes pre-sentations of relevance to a multidisciplinary group.The goal of these educational activities is to improvethe care of critically ill patients within Fairview or otherhospital systems that have hospitalists who care for ICUpatients. In addition, the Critical Care Program funds 4research projects per year ($15,000 to $20,000 each) sub-mitted by the intensivist faculty via a competitive grantprocess.Although, our objective is mainly to provide informa-

tion about how we built our health systemewide CriticalCare Program, we reviewed administrative data of FHS toinvestigated whether our intervention improves outcomesacross the health system.

Data analysis

We retrospectively reviewed the hospitals’ administrativedatabase from January 2011 to December of 2012,including mean census, number of transfers, age, sex,case mix index (CMI), mortality and readmissions.Case mix index is a measure of patient acuity based ondiagnosis-related group and hospital characteristics (eg,diversity of populations, hospital facilities, etc).14

Diagnosis-related group is determined by patient charac-teristics, comorbidities, and diagnoses.15 We also reviewedfinancial data of our tele-ICU. Continuous or scale vari-ables were compared using Student’s t-test. Categorical ornominal variables were compared using chi-square. A pvalue <0.05 was considered statistically significant.

RESULTSIn 2012, the Fairview Critical Care program had 52 ICUbeds at UMMC-E and 54 tele-ICU beds remotely(Table 1). Mean midnight census for the entire programwas 69.44 and total patient-days were 34,406 (Table 3).Mean daily census of the tele-ICU program only was

32.64 patients and the total number of patient-days was16,973.The capital cost of the tele-ICU was $1,186,220

(Table 4). The annual operational cost is $1,250,112 or$23,150 per monitored ICU-bed. The annual cost ofthe tele-ICU nurses is $550,200. Sixty-six percent ofthe cost of the tele-intensivist is paid by UMMC-E forproviding in-house service, which reduces the cost forthe remote hospitals for the tele-ICU physicians to$547,675 (Table 4).The ICU mortality and 48-hour readmission rates were

slightly lower in 2013 compared with 2012, although theacuity of the patients was similar, as reflected by CMI(Table 5). Two hundred and thirty-three critically ill pa-tients from Fairview hospitals were transferred in 2012.

Vol. 219, No. 4, October 2014 Fortis et al Health System-Based Critical Care Program 681

One hundred and ninety-five (84%) were transferred toanother ICU within our system and 38 (16%) were trans-ferred for medical care in facilities outside of Fairview.

DISCUSSIONThis study describes the development and operational de-tails of a health care system level ICU program patternedoff the much larger Military Joint Theater Trauma Sys-tem.13 To our knowledge, it is one of the few studiesthat provide details about the structure and the operationof ICUs at a health system level.16

Our system-wide ICU program has led to standardiza-tion of some aspects of critical care in the Fairview sys-tem. Traditionally, the decisions at a system level affectfinancial and organization issues, and critical care serviceare organized at an ICU level by the ICU directors. Here,ICU leadership participates in strategic decisions aboutthe entire critical care service of the system. This kindof multidisciplinary approach to critical care is knownto be beneficial at a hospital level.16 In addition, becauseall of the intensivists are in the same physician practicegroup and medical school, there is greater ability to adoptcommon critical care protocols or order sets,16 which de-creases unnecessary variation. This is not a one-way pro-cess originating only from the University hospital;UMMC-E has adopted protocols and practice standardsthat started in the community hospitals.Our novel Critical Care program is based on the pres-

ence of an intensivist at UMMC-E around-the-clock whoalso serves as the tele-ICU physician. This dual-rolemodel would not be feasible if there were so much tele-ICU work that direct patient care was neglected or viceversa. Covering approximately 36 ICU patients atUMMC-E (with house-staff assistance) and 32 ICU pa-tients remotely makes for a balanced and busy 13-hournighttime shift.Although the benefit of a nocturnal intensivist is

controversial,1,4,17 at UMMC-E, the severity of illness ishigh and there are no medicine or surgery critical care fel-lows in house overnight. Around-the-clock intensivistpresence is the care model in many ICUs in the UnitedStates18 and Europe,19 and is recommended by the Societyof Critical Care Medicine.6 To avoid intensivist burn out,ICU physicians from Medicine, Surgery, and Anesthesi-ology joined together to share nighttime work. Other au-thorities have reported the same approach to increase thestaff of intensivists.16

Telemedicine can provide care to remote critically illpatients around the clock, thereby fulfilling the intentof the Society of Critical Care Medicine’s vision tohave intensivist input into the care of critically ill patients

anywhere. Many ICUs in the United States do not haveon-site intensivist assistance currently and many neverwill for various reasons. Telemedicine allows an intensiv-ist to review information for remote patients in the sameformat (ie, vitals, laboratory tests, and imaging throughEMR) as if she/he were at the bedside and provide criticalcare through on-site provider such as staff nurses, hospi-talists, or other advanced practice providers.Some reports show that tele-ICU programs decrease

mortality and improve outcomes,10,20-24 and others havenot shown a benefit.25 It is possible that the beneficial ef-fects of tele-ICU are related to the extent that the tele-ICU clinician is permitted to actually make changes incare for a patient in a monitored ICU.23,26 In the Fairviewsystem, the tele-ICU staff is given full authority for pa-tient care by the remote sites, including medication orderentry or having on-site providers perform emergency pro-cedures, such as endotracheal intubation. This authoritywas granted in part because the majority of tele-intensivists also maintain a bedside practice in severalICUs of the system9 and therefore have a face-to-face pro-fessional relationship with the clinicians that use the tele-ICU program. Additionally, the ICU leadership visits allhospitals served by the ICU program on a routine basis toensure that issues are addressed in a timely fashion.A financially sustainable stand-alone tele-ICU program

is only possible when the tele-ICU monitors a large num-ber of tele-ICU beds, which might explain why manytele-ICU programs include >100 beds.27 Our programdemonstrates that a smaller tele-ICU program can besustained with low “per-monitored bed” charges to theremote facilities, if all the ICUs use the same existingEMR technology and if the cost of the tele-ICU physicianis defrayed by professional fee revenue doing face-to-faceICU care at the hub hospital and also doing tele-ICU.Costs are reduced even more by avoiding licensing feesrequired to install a separate tele-ICU care managementsoftware program.Payment for professional work by the intensivist staff is

currently reimbursed only when provided directly at thebedside, therefore, the tele-ICU work is not reimbursed.However, indirect financial benefits, such as increasedrevenue from a higher CMI score; improved efficiencyoutcomes from 24-hour intensivist availability, and abil-ity to keep ICU patients in their local hospital (wherecosts are lower than in the teaching hospital to wherethe patient would be transferred) helps compensate forlack of reimbursement.The capital cost of a tele-ICU command center ranges

between $2,000,000 and $6,000,000.28 Fairview’s costswere $1,186,220. A recent meta-analysis reported thatthe estimated combined implementation and first-year

682 Fortis et al Health System-Based Critical Care Program J Am Coll Surg

operation cost per tele-ICU bed ranges between $50,000and $100,000.29 The authors of this meta-analysis suggestthat these data are of ambiguous validity due to the lowquality of the accounting methodology. The investigatorsalso reviewed data from Veterans Affairs hospitals andestimated a combined capital and first-year operationalcost from $70,000 to $87,000 per tele-ICU bed. TheFairview combined implementation and first-year opera-tional cost was $45,117 and the annual operational costfor one tele-ICU bed was only $23,150.Although this study objective was not to demonstrate

that the Fairview critical care and tele-ICU program im-proves the care of critically ill patients but rather to provideinformation about implementation of a system-wide pro-gram, the outcomes data support its functionality. Despitethe acuity of the patients admitted in 2012 being similaror higher than those admitted in 2011 based on CMI,ICU mortality was significantly lower in 2012 aftertele-ICU implementation. We observed similar findingsin ICU readmission rates. Eight-three percent of ICUtransfers in 2012 were able to be managed within the Fair-view ICU system. However, we cannot conclude with cer-tainty that our intervention leads to reduction of mortalityor other outcomes for several reasons. First, the interven-tion is a continuous process; it is still in progress and iscomposed of many smaller interventions.30 Second, theimplementations of our system-based ICU program canaffect the triage and selection of patients. For example,tele-ICU availability might have led to more judicious de-cisions about which patient can benefit from ICU. Thismight have been the reason that previous studies haveshown a reduction in ICU but not hospital mortality.21

In addition, the improved outcomes might reflect a Haw-thorne effect.31 The ICU staff might have improved theirproductivity because of the implementation of the CriticalCare Program, and the intervention itself has no effect onoutcomes. Although, this is less likely because our inter-vention took place gradually, such a possibility cannotbe excluded entirely. This does not undermine the impor-tance of our report, which describes how we built a healthsystemelevel Critical Care Program.Our Critical Care Program has also created a framework

that allows academic physician faculty to work closely withcommunity ICUs and understand their patient problems.This is important because most patient-level ICU researchtake place in large academic centers only.32

Our study has several limitations. Although the pa-tients admitted in 2011 had CMIs similar to thoseadmitted in 2012, mortality was not adjusted for initialseverity of illness. We also did not provide outcomesdata, such as ICU length of stay, need for mechanicalventilation, or transfers, for 2011, as they are not available

in the administrative database. Another limitation is thatthe complexity of the ICU system intervention makes itdifficult to causally link one intervention to one discreteresultda problem that has been described previously.23,30

We also did not systematically collect data on clinicalconflicts between the tele-ICU clinicians and the on-siteclinicians. We did not examine whether our interventionhas affected the clinical practice of different subspecialtiesor whether outcomes are different among the differentspecialists (ie, internists, surgeons, and anesthesiologist).These will be the subjects of future studies. In addition,we did not report the direct and indirect ICU patientcare costs before and after the tele-ICU intervention, asthose costs are not reliably assigned within the currentFairview cost accounting system.

CONCLUSIONSWe describe building a health system level ICU programusing “off the shelf” technology based at a Universitymedical center and implementing a tele-ICU programwith full treatment authority in all monitored ICUs.The tele-ICU is the “eyes on the ground” for the systemand facilitates the Critical Care Program’s efforts to stan-dardize critical care practices across 7 ICUs. The tele-ICUbusiness model is based on an intensivist who shareson-site and telemedicine clinical responsibilities, therebysubsidizing substantially the operational cost to theremote ICUs. Combining the faculty clinical effort ofacademically distinct departments into one Critical CareProgram reduces nighttime work and creates new educa-tion and research opportunities.

Author Contributions

Study conception and design: Weinert, BeilmanAcquisition of data: Fortis, Weinert, Bushinski, Koehler,Beilman

Analysis and interpretation of data: Fortis, Weinert,Beilman

Drafting of manuscript: Fortis, Weinert, BeilmanCritical revision: Fortis, Weinert, Beilman

Acknowledgment: The authors are indebted to Jeff Woodfor data acquisition support.

REFERENCES

1. Gajic O, Afessa B, Hanson AC, et al. Effect of 24-hourmandatory versus on-demand critical care specialist presenceon quality of care and family and provider satisfaction in theintensive care unit of a teaching hospital. Crit Care Med2008;36:36e44.

2. Gajic O, Afessa B. Physician staffing models and patient safetyin the ICU. Chest 2009;135:1038e1044.

Vol. 219, No. 4, October 2014 Fortis et al Health System-Based Critical Care Program 683

3. Garland A, Roberts D, Graff L. Twenty-four-hour intensivistpresence: a pilot study of effects on intensive care unit patients,families, doctors, and nurses. Am J Respiratory Crit Care Med2012;185:738e743.

4. Blunt MC, Burchett KR. Out-of-hours consultant cover andcase-mix-adjusted mortality in intensive care. Lancet 2000;356:735e736.

5. Pronovost PJ, Angus DC, Dorman T, et al. Physician staffingpatterns and clinical outcomes in critically ill patients: a sys-tematic review. JAMA 2002;288:2151e2162.

6. Haupt MT, Bekes CE, Brilli RJ, et al. Guidelines on criticalcare services and personnel: Recommendations based on a sys-tem of categorization of three levels of care. Crit Care Med2003;31:2677e2683.

7. Angus DC, Kelley MA, Schmitz RJ, et al. Caring for the crit-ically ill patient. Current and projected workforce require-ments for care of the critically ill and patients withpulmonary disease: can we meet the requirements of an agingpopulation? JAMA 2000;284:2762e2770.

8. Cummings J, Krsek C, Vermoch K, et al. Intensive care unittelemedicine: review and consensus recommendations. Am JMed Qual 2007;22:239e250.

9. Nguyen YL, Kahn JM, Angus DC. Reorganizing adult criticalcare delivery: the role of regionalization, telemedicine, andcommunity outreach. Am J Respiratory Crit Care Med2010;181:1164e1169.

10. Lilly CM, Cody S, Zhao H, et al. Hospital mortality, length ofstay, and preventable complications among critically ill pa-tients before and after tele-ICU reengineering of critical careprocesses. JAMA 2011;305:2175e2183.

11. Blackbourne LH, Baer DG, Eastridge BJ, et al. Military med-ical revolution: military trauma system. J Trauma Acute CareSurg 2012;73[Suppl]:S388eS394.

12. Hill R. Understanding the M and A trends in Hospital mergersand acquisitions. Opportunities and challenges. Online report.Dixon Hughes Goodman; 2010.

13. Eastridge BJ, Jenkins D, Flaherty S, et al. Trauma systemdevelopment in a theater of war: experiences from OperationIraqi Freedom and Operation Enduring Freedom. J Trauma2006;61:1366e1372; discussion 1372e1373.

14. Centers for Medicare and Medicaid Services. Fiscal Year 2011:Final Rule Data Files. Baltimore, MD: Centers for Medicareand Medicaid Services; 2011.

15. Fetter RB, Shin Y, Freeman JL, et al. Case mix definition bydiagnosis-related groups. Med Care 1980;18:iii, 1e53.

16. Irwin RS, Flaherty HM, French CT, et al. Interdisciplinary collab-oration: the slogan that must be achieved for models of deliveringcritical care to be successful. Chest 2012;142:1611e1619.

17. Kerlin MP, Small DS, Cooney E, et al. A randomized trial ofnighttime physician staffing in an intensive care unit. N Engl JMed 2013;368:2201e2209.

18. Diaz-Guzman E, Colbert CY, Mannino DM, et al. 24/7in-house intensivist coverage and fellowship education: across-sectional survey of academic medical centers in theUnited States. Chest 2012;141:959e966.

19. Vincent JL, Suter P, Bihari D, et al. Organization of intensivecare units in Europe: lessons from the EPIC study. IntensiveCare Med 1997;23:1181e1184.

20. Breslow MJ, Rosenfeld BA, Doerfler M, et al. Effect of amultiple-site intensive care unit telemedicine program on clin-ical and economic outcomes: an alternative paradigm forintensivist staffing. Crit Care Med 2004;32:31e38.

21. Young LB, Chan PS, Lu X, et al. Impact of telemedicine inten-sive care unit coverage on patient outcomes: a systematicreview and meta-analysis. Arch Int Med 2011;171:498e506.

22. Rosenfeld BA, Dorman T, Breslow MJ, et al. Intensive careunit telemedicine: alternate paradigm for providing contin-uous intensivist care. Crit Care Med 2000;28:3925e3931.

23. Willmitch B, Golembeski S, Kim SS, et al. Clinical outcomesafter telemedicine intensive care unit implementation. CritCare Med 2012;40:450e454.

24. Lilly CM, McLaughlin JM, et al. A multi-center study of ICUtelemedicine reengineering of adult critical care. Chest 2014;145:500e507.

25. Morrison JL, Cai Q, Davis N, et al. Clinical and economicoutcomes of the electronic intensive care unit: results fromtwo community hospitals. Crit Care Med 2010;38:2e8.

26. Thomas EJ, Lucke JF, Wueste L, et al. Association of telemed-icine for remote monitoring of intensive care patients withmortality, complications, and length of stay. JAMA 2009;302:2671e2678.

27. The New England Healthcare InstituteMassachusetts Technol-ogy Collaborative. Critical Care, Critical Choices: The Case forTele-ICUs in Intensive Care. 2010. Available at: http://www.nehi.net/writable/publication_files/file/teleicu_critical_care_critical_choices.pdf. Accessed June 9, 2014.

28. The New England Healthcare InstituteMassachusetts Technol-ogy Collaborative, Health Technology Center. Tele-ICUs:Remote Management in Intensive Care Units. March 2007.Available at: http://www.nehi.net/writable/publication_files/file/tele_icu_final.pdf. Accessed June 9, 2014.

29. Kumar G, Falk DM, Bonello RS, et al. The costs of criticalcare telemedicine programs: a systematic review and analysis.Chest 2013;143:19e29.

30. Yoo EJ, Dudley RA. Evaluating telemedicine in the ICU.JAMA 2009;302:2705e2706.

31. McCarney R, Warner J, Iliffe S, et al. The Hawthorne Effect: arandomised, controlled trial. BMC Med Res Methodol 2007;7:30.

32. Iyegha UP, Asghar JI, Habermann EB, et al. Intensivistsimprove outcomes and compliance with process measures incritically ill patients. J Am Coll Surg 2013;216:363e372.