Introduction
Materials and methods
Literature search
Study selection
Data abstraction and validity assessment
Data analysis
Results
Study flow
Description of included studies
Source | ICUs/hospitals, n | Patientsa(total/pre/post), n | Age (years) | Sex (% male) | Illness severity | |||
---|---|---|---|---|---|---|---|---|
Control | Telemedicine | Control | Telemedicine | Control | Telemedicine | |||
Rosenfeld et al. 2000 [10] | 1/1 | 628/227/201 | 61 | 61 | 56 | 57 | APACHE III 37 | APACHE III 38 |
Breslow et al. 2004 [8] | 2/1 | 2,140/1,396/744 | 61 | 60 | 56 | 50 | APACHE III APS 39 | APACHE III APS 38 |
Marcin et al. 2004 [27] | 1/1 | 296/249/47 | 5.5 | 5.3 | NR | NR | PRISM III 7.5 | PRISM III 9.6 |
Kohl et al. 2007 [30] | 1/1 | 2,811/189/2,622 | NR | NR | NR | NR | NR | NR |
Vespa et al. 2007 [28] | 1/1 | 1,218/578/640 | NR | NR | NR | NR | NR | NR |
Norman et al. 2009 [31] | 1/1 | 1,275/356/919 | NR | NR | NR | NR | APACHE IV 57 | APACHE IV 53 |
Thomas et al. 2009 [9] | 6/5 | 4,142/2,034/2,108 | 60 | 59 | 51 | 53 | SAPS II 35 | SAPS II 34 |
McCambridge et al. 2010 [11] | 3/1 | 1,913/954/959 | 65 | 64 | 50 | 50 | APACHE IV APS 57 | APACHE IV APS 58 |
Morrison et al. 2010 [12] | 4/2 | 4,088/1,371/2,717 | 64 | 65 | 56 | 52 | APACHE III 49 | APACHE III 48 |
Lilly et al. 2011 [29] | 7/1 | 6,290/1,529/4,761 | 62 | 64 | 57 | 57 | APACHE III 45 | APACHE III 58 |
Willmitch et al. 2012 [32] | 10/5 | 24,656/6,504/18,152 | NR | NR | NR | NR | CMI 2.68 | CMI 2.77 |
Details of the interventions
Source Study periods | Type of hospital/ICU ICU staffing model Type of interventiona | Intervention details | Intervention dose | Equipment cointerventions |
---|---|---|---|---|
Rosenfeld et al. 2000 [10] Pre 1: 1 Sept-18 Dec 1996 Pre 2: 1 Feb-18 May 1997b Post: 1 Sept-18 Dec 1997 | Academic-affiliated community hospital; surgical ICU Open modelc Low-intensity passive | Tele-intensivist interacted with patients and healthcare personnel via dedicated video conferencing and data transmission equipment 24 hours/day Clinical and stored physiologic data reviewed q2hours | Formal video conferencing rounds occurred on 50% of days; otherwise, intensivist discussed each case with senior housestaff or attending physician Tele-intensivists spent 4 to 5 hours/day on clinical care | Spacelabs Medical, Seattle WA None |
Breslow et al. 2004 [8] Pre: 1 July 1999-20 June 2001 Post: 1 Jan-30 June 2001 | Tertiary care, teaching; medical and surgical ICUs Closed unit for teaching team of medical ICU patients (40%); open model for remaining medical ICU patients and surgical ICUc High-intensity passive or active (alerts not clearly described) | Tele-ICU staff (board certified intensivist, nurse) monitored all patients 19 hours/day (1200-0700) Admitting physician determined tele-ICU decision-making authority (all versus some versus off-hours) Tele-ICU reviewed patient data q4hours | Not described | VISICU Inc. (eICU CARE), Baltimore MD None |
Marcin et al. 2004 [27] Pre: Oct 1997-Sept 1998 Post: Apr 2000-Apr 2002d | Tertiary referral; adult ICU (with some pediatric patients) Pediatric intensivist during baseline period only Low-intensity passive | Consultation (at discretion of admitting physician) with tele-pediatric intensivist using portable telemedicine unit in pediatric ICU and five consultants' homes available 24 hours/day within 15 minutes | Number of consultations, one to seven per patient (median, 1; mean, 1.5) | Tandberg 800 video conference units None |
Kohl et al. 2007 [30] Dates not reported | Academic; surgical ICU Staffing model not described High-intensity passive or active (based on vendor) | Tele-ICU staffed by board certified intensivists; no further details provided | Not described | VISICU Inc. (eICU CARE), Baltimore MD None |
Vespa et al. 2007 [28] Pre: 2003-2004 fiscal year Post: June 2005-June 2006 | Academic; neurologic ICU Staffing model not described; tele-intensivist same as on-site intensivist Low-intensity passive | Robotic telepresence program for live interactive consultation and review of physiologic trends with intensivist [2000-0000 (weekdays); 1800 (weekends)] Each patient reviewed for ≥ 5 minutes | Mean, two sessions/day Mean night-time rounding session, 52 minutes | Robot: InTouch Health, Santa Barbara CA Informatics system: Global Care Quest, Aliso Viejo CA Integrated clinical information system Paging protocol with goal of attending physician response within 15 minutes |
Norman et al. 2009 [31] Pre: Jan-Mar 2008 Post: Jan-Mar and Apr-June 2009e | Hospital not described; medical-surgical ICU Staffing model not described High-intensity passive or active (alerts not clearly described) | Tele-ICU staff ("team" included nurse; intensivist presence not specifically stated) reviewed patients; no further details provided | Not described | VISICU Inc. (eICU CARE), Baltimore MD Electronic discharge management tool |
Thomas et al. 2009 [9] Pre: Jan 2003-Aug 2005 Post (staggered roll-out): July 2004-July 2006 | Closedf medical and trauma/surgical ICU in tertiary care teaching hospital; two open medical-surgical ICUs in two small community hospitals; two open medical-surgical ICUs in two large urban hospitals Active | Tele-ICU staffed by two physicians (noon -7 am Monday-Friday, 24 hours/day weekends), four registered nurses, and two administrative technicians Rounds frequency: severely ill q1 hour, moderately ill q2 hours, relatively stable q4 hours Local physicians delegated to tele-ICU authority for full treatment (31% of patients) or for intervention only for life-threatening events (66%) | Tele-ICU physicians gave 1,446 orders in 60 days (four ICUs) Two closed ICUs, 5.3 orders/day (7% high-level interventions, (for example, code supervision, ventilator management) Two open ICUs, 18.5 orders/day (26% high-level) | VISICU Inc. (eICU CARE), Baltimore MD None |
McCambridge et al. 2010 [11] Pre: Sept 2002-Dec 2003 Post: Oct 2004-July 2005 | Academic community hospital; three ICUs Closed modelf Active | Tele-ICU team (intensivist and critical care nurse) (1900-0700) admitted new patients and responded to phone calls from ICU nurses, computer-generated alerts, and radiographic abnormalities Rounds for all monitored patients q2 hours | Not described | Vistacom Inc, Allentown PA Health information technology bundle: EMR with automatic alerts (iMDsoft, Needham MA); CPOE, electronic MAR and bar-coded medication administration, PACS (GE Healthcare, Fairfield CT) |
Morrison et al. 2010 [12] Pre: Dec 2002-Mar 2003 Post 1: Dec 2004-Mar 2004 Post 2: July-Oct 2004g | One community teaching hospital (medical ICU, surgical ICU, cardiac ICU) and one community nonteaching hospital (medical-surgical ICU) Open modelc Active | Admitting physician responsible for care plan and determined involvement of tele-ICU (four categories from emergency care only to no restrictions) Tele-intensivist reviewed all patient data at least q4 hours (q1 hour for sickest patients) At teaching hospital, tele-intensivist supervised and taught housestaff "real-time" | Physician adoption of high-level (unrestricted) tele-ICU care differed (teaching hospital, 25% of physicians [post one], 57% [post two]; nonteaching hospital, 9% [post one], 27% [post two]) | VISICU Inc. (eICU CARE), Baltimore MD, including "Sentry Alerts" software |
Lilly et al. 2011 [29] Pre: April 2005-Feb 2007 Post: (staggered roll-out) Aug 2006-Sept 2007 | Academic medical center; seven ICUs: three medical, three surgical, and one mixed cardiovascular Closed modelf Active | Tele-ICU (hospital staff intensivist, affiliate practitioner, systems analyst, ≥ one data clerks), 24 hours/day Tele-ICU monitored 5-minute timed median vital sign values on electronic flow sheet; reviewed care; audited best-practice adherence real-time; reviewed night-time admissions; monitored electronic alerts, intervened when responses of bedside clinicians to in-room alarms delayed | Tele-ICU reviewed care plan for 48% of after-hours admissions (46% reviewed by other methods in pre period) 23 943 tele-ICU initiated interventions for physiologic instability that affected care plan (76% "major") | VISICU Inc. (eICU CARE), Baltimore MD; APACHE (Cerner Healthcare Solutions, Kansas City MO) Criticalware (UMass) software package to audit best practices (glycemic control; prevention of DVT, CRBSI, VAP) None |
Willmitch et al. 2012 [32] Staggered roll-out: Dec 2005-July 2007 Pre: 1 year before roll-out Post 1: year 1 after roll-out Post 2: year 2 after roll-out Post 3: year 3 after roll-outh | Five community hospitals with 10 ICUs Closed modelf in largest hospital (28% of ICU beds in the study); otherwise open modelc Active | Tele-ICU, staffed by one intensivist, three critical care nurses, and one secretary, 24 hours/day | All admitting and consulting physicians (n = 2,607) indicated level of tele-ICU intervention for their patients: 1% selected level I (emergency care only), 97% level II (best-practices adjustments), 2% level III (no restrictions) | Philips VISICU eCare Manager (Admission, discharge and transfer interfaces), Philips Smart Alerts, Philips VISICU camera system (Philips, Amsterdam, Netherlands) None |
Study quality
Primary and secondary outcomes
Subgroup analyses
Discussion
Conclusions
Key messages
-
We found 11 before/after observational studies including 49,457 patients that examined the effect of telemedicine on clinically important outcomes.
-
Pooled unadjusted data from nine studies showed that telemedicine reduced ICU mortality (RR, 0.79; 95% CI, 0.65 to 0.96; P = 0.02) and hospital mortality (RR, 0.83; 95% CI, 0.73 to 0.94; P = 0.004); reductions in ICU and hospital lengths of stay were also statistically significant.
-
The effect of telemedicine on ICU mortality was similar in active or high-intensity passive systems (continuous patient-data monitoring with or without electronic alerts) compared with low-intensity passive systems (remote intensivist consultation only), but this subgroup analysis was underpowered.
-
Future research should establish the optimal telemedicine technology configuration and dose tailored to ICU organization and case mix.