You are on page 1of 7

Impact of the Electronic Medical Record on

Mortality, Length of Stay, and Cost in the Hospital


and ICU: A Systematic Review and Metaanalysis*
Gwen Thompson, MD, MPH1; John C. O’Horo, MD, MPH2; Brian W. Pickering, MBBCh, MSc3;
Vitaly Herasevich, MD, PhD, MSc3

Objective: To evaluate effects of health information technology information technology interventions effect on mortality and length
in the inpatient and ICU on mortality, length of stay, and cost. of stay were not statistically significant. Cost was unable to be
Methodical evaluation of the impact of health information technol- quantitatively evaluated. Qualitative synthesis of studies of each
ogy on outcomes is essential for institutions to make informed outcome demonstrated significant study heterogeneity and small
decisions regarding implementation. clinical effects.
Data Sources: EMBASE, Scopus, Medline, the Cochrane Review Conclusions: Electronic interventions were not shown to have a
database, and Web of Science were searched from database substantial effect on mortality, length of stay, or cost. This may be
inception through July 2013. Manual review of references of iden- due to the small number of studies that were able to be aggre-
tified articles was also completed. gately analyzed due to the heterogeneity of study populations,
Study Selection: Selection criteria included a health information interventions, and endpoints. Better evidence is needed to iden-
technology intervention such as computerized physician order tify the most meaningful ways to implement and use health infor-
entry, clinical decision support systems, and surveillance systems, mation technology and before a statement of the effect of these
an inpatient setting, and endpoints of mortality, length of stay, or systems on patient outcomes can be made. (Crit Care Med 2015;
cost. Studies were screened by three reviewers. Of the 2,803 43:1276–1282)
studies screened, 45 met selection criteria (1.6%). Key Words: costs and cost analysis; electronic health records;
Data Extraction: Data were abstracted on the year, design, inter- length of stay; medical informatics; mortality
vention type, system used, comparator, sample sizes, and effect
on outcomes. Studies were abstracted independently by three
reviewers.

I
Data Synthesis: There was a significant effect of surveillance n recent years, the U.S. government has invested billions of dol-
systems on in-hospital mortality (odds ratio, 0.85; 95% CI, lars on the advancing of health information technology (HIT)
0.76–0.94; I2  =  59%). All other quantitative analyses of health through the Health Information Technology for Economic and
Clinical Health (HITECH) Act of 2009 (1, 2). This has been done
with the hope that HIT will improve the health of Americans by
*See also p. 1342.
providing better care while simultaneously lowering costs. Propo-
1
Division of General Internal Medicine, Mayo Clinic, Rochester, MN.
nents of the electronic medical record (EMR) such as politicians,
2
Division of Infectious Diseases, Mayo Clinic, Rochester, MN.
journalists, and the EMR industry claim that EMRs are able to ful-
3
Multidisciplinary Epidemiology and Translational Research in Intensive
Care and Department of Anesthesiology, Division of Critical Care Medi- fill this expectation. Statements by these groups are often made that
cine, Mayo Clinic, Rochester, MN. EMRs “save lives” (3). However, there has never been a systematic
Supplemental digital content is available for this article. Direct URL cita- review in inpatient settings supporting this claim.
tions appear in the printed text and are provided in the HTML and PDF Other systematic reviews of HIT have been conducted.
versions of this article on the journal’s website (http://journals.lww.com/
ccmjournal). However, they have focused on a particular intervention such
Drs. Pickering and Herasevich and their institutions licensed technology. as computerized physician order entry (CPOE) (4, 5), different
Drs. Pickering and Herasevich receive royalties and have stock with settings such as ambulatory care (6, 7), a particular popula-
Ambient Clinical Analytics Inc. Dr Pickering additionally is member on the tion (7), different endpoints such as medication prescription
Board of Directors of Ambient Clinical Analytics Inc. The remaining authors
have disclosed that they do not have any potential conflicts of interest. errors, medication safety (4, 5), or efficiency (6). No review has
For information regarding this article, E-mail: herasevich.vitaly@mayo.edu been conducted that evaluates all HIT interventions across all
Copyright © 2015 by the Society of Critical Care Medicine and Wolters inpatient settings. The effect of various HIT interventions such
Kluwer Health, Inc. All Rights Reserved. as CPOE, clinical decision support (CDS) systems, and surveil-
DOI: 10.1097/CCM.0000000000000948 lance systems or “sniffers” are heterogeneous, each affecting a

1276 www.ccmjournal.org June 2015 • Volume 43 • Number 6


Copyright © 2015 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Review Articles

different aspect of healthcare delivery, making it important to RESULTS


individually evaluate the impact of each system on patient out-
Literature Review
comes. There is also a lack of evidence whether EMRs affect an
The initial search yielded 2,736 articles. Sixty-seven additional
area where mortality is often an unfortunate reality, ICUs. The
studies were identified through manual review of references of
aim of this systematic review and metaanalysis is to evaluate
included studies. Of all studies identified, 45 studies met eligibil-
the effects of HIT in the inpatient and ICU setting on mortal-
ity criteria and were included in this review. Twenty-six contained
ity, length of stay (LOS), and cost.
quantitative data to be included in the metaanalysis (Fig. 1).
Of the studies, five described electronic health record ­systems
METHODS (11–15), 11 CPOE (16–26), 17 CDS systems (27–43), six sur-
veillance systems or “sniffers” (44–49), and six electronic medi-
Search Strategy
cation reconciliation tools not fitting into the prior categories
The search strategy and literature search was performed by a
(50–55). Almost all of the studies were peri-­implementation
medical reference librarian with input from the investigators.
designs comparing preintervention with postintervention.
A search of EMBASE, Scopus, Medline, the Cochrane Review
database, and Web of Science from database inception through
Mortality
July 2013 was conducted (8).
Of the included studies, 33 had mortality as an endpoint (11,
Eligibility criteria included studies that used a form of HIT
13–15, 17–20, 23, 24, 26, 27, 29–31, 33–37, 39, 40, 42–49, 51–53).
as the intervention, was conducted in an inpatient setting, and
evaluated mortality, LOS, or cost as an endpoint. Exclusion cri-
teria were studies where there was no control group, that were
conducted in an acute care or outpatient setting, and that did
not evaluate mortality, LOS, or cost as an outcome. Full search
strategy is available as supplemental data (Supplemental
Digital Content 1, http://links.lww.com/CCM/B218).
Abstracts were independently screened, and potentially rel-
evant articles were identified for full-text review. References
from included studies were manually inspected to identify all
potentially relevant studies for screening and review.

Data Abstraction
From each study, data were abstracted on the study year, loca-
tion, design, setting, inclusion criteria, type of intervention
(CPOE, EMR, or CDS), specific system used, implementa-
tion timeframe, comparator, population, total study size, size
of control group and comparator groups, and effect found on
outcomes. Outcomes were categorized as mortality, LOS, and
cost. The abstraction template can be found as Supplemental
Table 1 (Supplemental Digital Content 2, http://links.lww.com/
CCM/B219). Studies were abstracted and screened by three
reviewers, and disagreements were resolved by discussion.

Statistical Analyses
Dersimonian and Laird random effects models were used for
quantitative synthesis of the data where there were at least
three articles meeting inclusion criteria for that outcome. This
modeling uses standard inverse variance weighting followed by
an “unweighting” based on the extent of variability observed,
making for larger CIs and less weight to individual studies to
make more conservative estimates of pooled effects.
Heterogeneity was assessed using an I2 statistic, where 0–30%
indicates low heterogeneity, 31–60% moderate heterogeneity,
and more than 60% high statistical heterogeneity (9). These anal-
yses were performed with Cochrane Review Manager Software
(RevMan, version 5.3; Nordic Cochrane Center, Copenhagen,
Denmark) and the Stata Metan package (Stata Statistical Figure 1. Preferred items for reporting in systematic reviews and
Software, Release 13; StataCorp LP, College Station, TX). (10). metaanalyses flow diagram.

Critical Care Medicine www.ccmjournal.org 1277


Copyright © 2015 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Thompson et al

Of these 33, 24 (11, 17–20, 23, 24, 27, 29–31, 33–36, 39, 40, 44, Nine studies (13–15, 37, 42, 43, 51–53) provided informa-
46–49) included data that were able to be quantitatively evalu- tion on mortality that could not be quantitatively synthesized
ated. Studies looked at in-hospital mortality, in-ICU mortality, via metaanalysis. This included one cross-sectional study (51),
or 90-day mortality. one nationwide registry review (52), one clustered random-
Of the studies that could be quantitatively analyzed and ized trial (37), five pre-post studies (13, 15, 42, 43, 53), and
examined in-hospital mortality, five evaluated CPOE systems one interrupted time series study (14). Of the nine studies,
(17, 18, 24, 26, 45), one EHR (11), seven CDS systems (27, four demonstrated a decrease in mortality (13, 14, 51, 52),
30, 31, 33, 35, 39, 40), and three sniffers (46, 48, 49). Overall, four showed no difference (15, 42, 43, 53), and one (37) had
CPOE systems did not show a significant effect on this out- no mortalities in their study population making a difference in
come (odds ratio [OR], 0.91; 95% CI, 0.75–1.10; I2 = 66%), nor mortality difficult to assess.
did EHRs alone (OR, 0.96; 95% CI, 0.77–1.19). CDS systems The methods and quality of these studies varied significantly
had slightly better performance overall (OR, 0.83; 95% CI, making a conglomerate statement difficult. In addition, the
0.69–1.00; I2 = 66%). The best overall performance was seen electronic intervention assessed and study population varied
in “sniffer” systems, which had a pooled OR for in-hospital between and within studies. For example, one study (51) stated
mortality of 0.85 (95% CI, 0.76–0.94) with moderate hetero- that there was a decrease in mortality; however, this varied sig-
geneity (I2 = 59%). This is ­summarized in Figure 2. nificantly by the electronic intervention assessed and the disease
Of the studies addressing ICU mortality as an endpoint, process used for the inclusion criteria for the study population
there were four using CPOE systems (17, 19, 20, 23), three CDS even within that individual study. Study interventions ranged
(30, 34, 40), and two sniffers (46, 47). Heterogeneity was high from CDS related to antibiotic guidance (37, 42) to full EMRs
in all subgroups and none trended toward a significant effect. (14). Another study (52) demonstrated a statistically significant
This is summarized in Figure 3. decrease in mortality; however, the clinical applicability of the
Three studies reported 90-day mortality, two CDS (29, 36), small decrease in mortality rate needs to be considered as the
and one sniffer (44). There was no evidence of effect of any of largest decrease in mortality was from 16.4% to 16.0%. There
the interventions on this outcome. were also differences in the definition of mortality. For instance,
one study only commented
on 30-day mortality and not
overall mortality (43). Another
study (13) demonstrated a
decrease in 30-day mortality
but no statistically significant
difference in inpatient mor-
tality. Hence, which form of
mortality is being evaluated
between studies could affect
the results.
Sensitivity analyses excluding
older studies failed to decrease
heterogeneity or change the
overall impact observed for
each system.

LOS
Of the included studies, 35
(11–15, 17–19, 21, 22, 24,
25, 27, 29, 30, 32–35, 37–42,
45–54) evaluated LOS as an
outcome. Of these 35, nine (17,
19, 21, 22, 24, 27, 33, 34, 46)
included data that were able
to be quantitatively evaluated.
Studies looked at either hospi-
tal LOS or ICU LOS.
Hospital LOS was evaluated
by eight studies (17, 21, 22, 24,
Figure 2. In-hospital mortality. Overall, there was a trend towards better mortality with sniffer systems and clinical
decision support (CDS); computerized physician order entry (CPOE) did not achieve significance. EHR = electronic 27, 33, 34, 46), including four
health record. CPOE studies (17, 21, 22, 24),

1278 www.ccmjournal.org June 2015 • Volume 43 • Number 6

Copyright © 2015 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Review Articles

cross-sectional studies (51, 52,


54), one cluster (12), one cluster
randomized trial (37), two ran-
domized control trials (35, 50),
and two interrupted time series
(14, 38). Sixteen studies dem-
onstrated no difference in LOS
after an electronic interven-
tion (15, 18, 29, 30, 32, 35, 38,
40, 41, 45, 47, 48, 51–54). One
study showed a decrease LOS in
one of the institutions studied
but not in the other (25). Eight
studies concluded that there
was a significant decreased LOS
after intervention (11, 13, 14,
37, 39, 42, 49, 50). One study
showed an increase in LOS with
EMR (12).
Qualitatively, this seems
Figure 3. ICU mortality. Neither sniffers, computerized physician order entry (CPOE), or clinical decision support to suggest a no-impact to
(CDS) outperformed controls significantly. EHR = electronic medical record. potential benefit to electronic
systems, but the clinical sig-
one sniffer (46), and three CDS systems (27, 33, 34). CPOE nificance and validity of results of the studies that showed a
trended toward a reduction in LOS (mean decrease, 0.67 d; decreased LOS should be considered. One study stated that
95% CI, –2.07 to 0.73), though with significant heterogeneity there was a decreased LOS; however, this was not statistically
(I2 = 82%). Neither CDS nor sniffers trended toward changes significant within hospitals that implemented the EMR with a
in hospital LOS, and the net pooled effect was not significant pre-post study design but was significant with a decreased LOS
(Fig. 4). Three studies (17, 19, 46) addressed ICU LOS, none of by 0.11 days when comparing hospitals with EMRs against
which found any significant impact. those that did not adopt EMRs (13). Hence, there may be an
Twenty-six studies (11–15, 18, 25, 29, 30, 32, 35, 37–42, 45, inherent difference between hospitals that adopt EMRs and
47–54) that evaluated LOS as an outcome were unable to be quan- those that do not that account for this difference. Another of
titatively synthesized. Of these studies, there were 17 pre-post the statistically significant studies showed a decreased LOS
studies (11, 13, 15, 18, 25, 29, 30, 32, 39–42, 45, 47–49, 53), three with CPOE from 3.91 to 3.71 days in one of the two institu-
tions evaluated (25). One
study showed a decreased LOS
after CDS implementation for
potential adverse drug events
that was statistically significant
after severity adjustment with
a ratio of expected to actual
LOS decreasing from 1.02 to
0.996 days (39). The study by
Sintchenko et al (42) showed
a decrease in bed days in an
ICU from 7.15 to 6.22 after
implementation of a handheld
CDS to assist in antimicrobial
choice. One study determined
that LOS decreased at a rate
of 0.043 bed-days per month
after EMR implementation
(14). The clinical significance
of these statistically significant
Figure 4. Hospital length of stay. Neither sniffers, computerized physician order entry (CPOE), or clinical deci- but small decreases in LOS
sion support (CDS) outperformed controls significantly. needs to be considered.

Critical Care Medicine www.ccmjournal.org 1279


Copyright © 2015 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Thompson et al

One study showed an increased LOS with EMR. This study In regard to mortality, there was no net effect of CPOE- or
compared a Veteran Affairs (VA) hospital with an EMR to a EHR-based interventions on ICU mortality or 90-day mortal-
non-VA hospital without an EMR (12). Therefore, confound- ity. Sniffers were the only intervention that had a statistically
ers may be present, and the conclusion that the increased LOS significant effect on hospital mortality. Qualitative synthesis
is attributable to EMR cannot be made. supported these conclusions. Although the majority of stud-
ies included in the qualitative synthesis stated that there was a
Cost decrease in mortality, this often varied by the electronic inter-
There were 14 studies that addressed cost as an outcome. These vention studied and had small statistically significant differ-
were unable to be quantitatively synthesized due to heteroge- ences that may not be clinically relevant.
neity of study design and endpoints. Of the 14 studies, there Similar results were obtained when looking at LOS. No
were nine pre-post studies (15, 25–28, 30, 39, 53, 55), one impact of electronic interventions on LOS, either hospital
cross-sectional study (51), one randomized control trial (35), or ICU, was found across all interventions. CPOE did trend
one cross-sectional study (54), one cluster randomized trial toward a decrease in hospital LOS, but this was not statistically
(37), and one interrupted time series (14). Eight of the studies significant and clinically may not be applicable as it was a little
concluded that there was a decrease in cost after an electronic over half of a day decrease in LOS in our metaanalysis, and
intervention (15, 27, 28, 30, 35, 37, 39, 51). Four studies showed overall, most studies reported no effect on LOS.
no significant difference (25, 26, 53, 54) and two stated that The endpoint of cost was unable to be evaluated quantita-
there was an increase in costs after intervention (14, 55). tively. Although most studies stated that there was a decrease
Study populations and endpoints varied considerably in cost, the implementation process was not included in the
between studies. The study populations in the studies ranged analysis of any study, and therefore, overall cost effectiveness
from adult patients with specific disease processes such as was not evaluated. Also, the interventions evaluated were often
heart failure and pneumonia (51) to infants (53). The end- very narrow in focus, and secondary endpoints were often used
points were also heterogeneous. One study addressed the cost making a final conclusion of effect difficult.
per admission (51). Other studies examined a specific CDS and The major limitation of this study is the heterogeneity of
therefore only commented on cost related to that one aspect interventions and outcome assessments of the available lit-
of care such as blood transfusions, laboratory testing, antibi- erature. This made overall conclusions difficult. Attempts
otic use, or pharmacy costs (27, 28, 30, 35, 37, 39). One study were made to control for this limitation by assessing different
examined decreased costs in specific areas such as radiology interventions separately. However, even within categories not
utilization or paper use (15) and another only evaluated the all interventions are equal. For instance, a CDS to assist with
number of pathology tests used (26). One study quantified antibiotic use is very different than one to assist in blood trans-
cost in terms of resource intensity weights and not dollars (53). fusion practices. In addition, EMR, CPOE, CDS, and sniffer
The results also varied within studies. For instance, one study implementation is not an isolated event. All electronic inter-
showed a decrease in pharmacy costs after CDS implementa- ventions inevitably lead to other practice changes. Endpoints
tion but an increase of total hospitalization costs (39). The two such as mortality, LOS, and cost are affected by a large number
studies that stated that there was an increase in costs associated of factors, many of which were likely not assessed by the stud-
with EMRs (14, 55) had many confounding variables making ies. In addition, the majority of the studies were pre-post stud-
this information difficult to interpret. One even demonstrated ies, and due to the nature of the intervention, they could not
that after controlling for specific factors that could influence be blinded. This may have created a Hawthorne effect thereby
cost, the difference was no longer significant (55). confounding the effect on the outcomes attributed to the elec-
tronic interventions. Publication bias may occur as negative
DISCUSSION findings are not often published or sought.
Overall, electronic interventions were not shown to have a sub-
stantial effect on mortality, LOS, or cost. This may in part be due CONCLUSIONS
to the small number of studies that were able to be aggregately The HITECH of 2009 made billions of Medicare and Medicaid
analyzed due to the heterogeneity of study populations, interven- dollars available to hospitals who engage in meaningful use of
tions, and endpoints. Also, most studies were pre-post studies. EHRs. In order to fall under meaningful use, five goals have
Many changes occur at the time of implementation and attrib- to be met: improving the quality, safety, and efficiency of care
uting all changes in outcomes to a single intervention may not while reducing disparities; engaging patients and families in
be appropriate. In addition, failure to observe patient-level out- their care; promoting public and population health; improv-
comes does not necessarily indicate a failure of EMRs. The lack ing care coordination; and promoting the privacy and security
of impact may be attributable to how the EMR is being imple- of health records (1). There is not enough evidence to con-
mented, the interpretation of correct use of EMRs, or the analysis fidently state that electronic interventions have the ability to
of the results in individual studies. This review highlights the het- achieve the goal of improving quality and safety. There may be
erogeneity of the current evidence to support the implementation other benefits of HIT, such as efficiency, patient engagement,
of EMR to improve patient outcomes. Further research is needed and healthcare information protection that were not evalu-
to determine the true impact of HIT on healthcare quality. ated by this review. In addition, failure to observe changes in

1280 www.ccmjournal.org June 2015 • Volume 43 • Number 6

Copyright © 2015 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Review Articles

patient-level outcomes does not necessarily indicate a failure of 19. Del Beccaro MA, Jeffries HE, Eisenberg MA, et al: Computerized pro-
vider order entry implementation: No association with increased mor-
EHRs in and of themselves, but rather may reflect an upstream tality rates in an intensive care unit. Pediatrics 2006; 118:290–295
problem with how EHRs are being used in specific processes. 20. Han YY, Carcillo JA, Venkataraman ST, et al: Unexpected increased
The quantity and quality of studies related to HIT’s impact mortality after implementation of a commercially sold computerized
on mortality, LOS, and cost are lacking, and there is significant physician order entry system. Pediatrics 2005; 116:1506–1512
interstudy and intrastudy variability. This study demonstrates 21. Holdsworth MT, Fichtl RE, Raisch DW, et al: Impact of computerized
prescriber order entry on the incidence of adverse drug events in
that future research is needed with standardized interventions pediatric inpatients. Pediatrics 2007; 120:1058–1066
and endpoints that also evaluate processes such as EMR use 22. Hwang JI, Park HA, Bakken S: Impact of a physician’s order entry
and implementation before a final statement on the effect of (POE) system on physicians’ ordering patterns and patient length of
stay. Int J Med Inform 2002; 65:213–223
HIT on patient outcomes such as mortality, LOS, and cost can
23. Keene A, Ashton L, Shure D, et al: Mortality before and after initiation
be made. of a computerized physician order entry system in a critically ill pedi-
atric population. Pediatr Crit Care Med 2007; 8:268–271
24. Longhurst CA, Parast L, Sandborg CI, et al: Decrease in hospital-
REFERENCES wide mortality rate after implementation of a commercially sold com-
1. Blumenthal D: Launching HITECH. N Engl J Med 2010; 362:382–385 puterized physician order entry system. Pediatrics 2010; 126:14–21
2. Lenert L, Sundwall DN: Public health surveillance and meaningful 25. Mekhjian HS, Kumar RR, Kuehn L, et al: Immediate benefits realized
use regulations: A crisis of opportunity. Am J Public Health 2012; following implementation of physician order entry at an academic
102:e1–e7 medical center. J Am Med Inform Assoc 2002; 9:529–539
3. Still T: Electronic health records can save lives and improve medical 26. Westbrook JI, Georgiou A, Rob MI: Test turnaround times and mortal-
care. WTN News, February 8, 2005 ity rates 12 and 24 months after the introduction of a computerised
4. Kaushal R, Shojania KG, Bates DW: Effects of computerized provider order entry system. Methods Inf Med 2009; 48:211–215
physician order entry and clinical decision support systems on 27. Evans RS, Pestotnik SL, Classen DC, et al: A computer-assisted
medication safety: A systematic review. Arch Intern Med 2003; management program for antibiotics and other antiinfective agents.
163:1409–1416 N Engl J Med 1998; 338:232–238
5. van Rosse F, Maat B, Rademaker CM, et al: The effect of comput- 28. Levick DL, Stern G, Meyerhoefer CD, et al: Reducing unnecessary
erized physician order entry on medication prescription errors and testing in a CPOE system through implementation of a targeted CDS
clinical outcome in pediatric and intensive care: A systematic review. intervention. BMC Med Inform Decis Mak 2013; 13:43
Pediatrics 2009; 123:1184–1190
29. Austrian JS, Adelman JS, Reissman SH, et al: The impact of the
6. Chaudhry B, Wang J, Wu S, et al: Systematic review: Impact of health heparin-induced thrombocytopenia (HIT) computerized alert on pro-
information technology on quality, efficiency, and costs of medical vider behaviors and patient outcomes. J Am Med Inform Assoc 2011;
care. Ann Intern Med 2006; 144:742–752 18:783–788
7. Shekelle PG, Morton SC, Keeler EB: Costs and benefits of health 30. Fernández Pérez ER, Winters JL, Gajic O: The addition of decision
information technology. Evid Rep Technol Assess (Full Rep) 2006; support into computerized physician order entry reduces red blood
132:1–71 cell transfusion resource utilization in the intensive care unit. Am J
8. Liberati A, Altman DG, Tetzlaff J, et al: The PRISMA statement for Hematol 2007; 82:631–633
reporting systematic reviews and metaanalyses of studies that evalu- 31. Filice GA, Drekonja DM, Thurn JR, et al: Use of a computer deci-
ate healthcare interventions: Explanation and elaboration. BMJ 2009; sion support system and antimicrobial therapy appropriateness. Infect
339:b2700 Control Hosp Epidemiol 2013; 34:558–565
9. Higgins JP, Thompson SG, Deeks JJ, et al: Measuring inconsistency 32. Holroyd-Leduc JM, Abelseth GA, Khandwala F, et al: A pragmatic
in metaanalyses. BMJ 2003; 327:557–560 study exploring the prevention of delirium among hospitalized older
10. Centre TNC: Review Manager (RevMan) Version 5.0. Copenhagen: hip fracture patients: Applying evidence to routine clinical practice
The Nordic Cochrane Centre, The Cochrane Collaboration, 2008 using clinical decision support. Implement Sci 2010; 5:81
11. Schenarts PJ, Goettler CE, White MA, et al: An objective study of 33. Lakshminarayan K, Rostambeigi N, Fuller CC, et al: Impact of an elec-
the impact of the electronic medical record on outcomes in trauma tronic medical record-based clinical decision support tool for dyspha-
patients. Am Surg 2012; 78:1249–1254 gia screening on care quality. Stroke 2012; 43:3399–3401
12. Boockvar KS, Livote EE, Goldstein N, et al: Electronic health records 34. Mazars N, Milési C, Carbajal R, et al: Implementation of a neonatal
and adverse drug events after patient transfer. Qual Saf Health Care pain management module in the computerized physician order entry
2010; 19:e16 system. Ann Intensive Care 2012; 2:38
13. Lee J, Kuo YF, Goodwin JS: The effect of electronic medical record 35. McGregor JC, Weekes E, Forrest GN, et al: Impact of a computerized
adoption on outcomes in US hospitals. BMC Health Serv Res 2013; clinical decision support system on reducing inappropriate antimicro-
13:39 bial use: A randomized controlled trial. J Am Med Inform Assoc 2006;
14. Xue Y, Liang H, Wu X, et al: Effects of electronic medical record 13:378–384
in a Chinese hospital: A time series study. Int J Med Inform 2012; 36. Mosen D, Elliott CG, Egger MJ, et al: The effect of a computerized
81:683–689 reminder system on the prevention of postoperative venous thrombo-
15. Zlabek JA, Wickus JW, Mathiason MA: Early cost and safety ben- embolism. Chest 2004; 125:1635–1641
efits of an inpatient electronic health record. J Am Med Inform Assoc 37. Paul M, Andreassen S, Tacconelli E, et al; TREAT Study Group:
2011; 18:169–172 Improving empirical antibiotic treatment using TREAT, a computer-
16. Collins S, Currie L, Bakken S, et al: Interruptions during the use of a ized decision support system: Cluster randomized trial. J Antimicrob
CPOE system for MICU rounds. AMIA Annu Symp Proc 2006;895 Chemother 2006; 58:1238–1245
17. Al-Dorzi HM, Tamim HM, Cherfan A, et al: Impact of computerized 38. Peterson JF, Kuperman GJ, Shek C, et al: Guided prescription of psy-
physician order entry (CPOE) system on the outcome of critically ill chotropic medications for geriatric inpatients. Arch Intern Med 2005;
adult patients: A before-after study. BMC Med Inform Decis Mak 165:802–807
2011; 11:71 39. Piontek F, Kohli R, Conlon P, et al: Effects of an adverse-drug-event
18. Collin S, Reeves BC, Hendy J, et al: Implementation of computerised alert system on cost and quality outcomes in community hospitals.
physician order entry (CPOE) and picture archiving and communica- Am J Health Syst Pharm 2010; 67:613–620
tion systems (PACS) in the NHS: Quantitative before and after study. 40. Rana R, Afessa B, Keegan MT, et al; Transfusion in the ICU Interest
BMJ 2008; 337:a939 Group: Evidence-based red cell transfusion in the critically ill: Quality

Critical Care Medicine www.ccmjournal.org 1281


Copyright © 2015 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Thompson et al

improvement using computerized physician order entry. Crit Care 48. Sawyer AM, Deal EN, Labelle AJ, et al: Implementation of a real-time
Med 2006; 34:1892–1897 computerized sepsis alert in nonintensive care unit patients. Crit Care
41. Roberts GW, Farmer CJ, Cheney PC, et al: Clinical decision support Med 2011; 39:469–473
implemented with academic detailing improves prescribing of key 49. Jones S, Mullally M, Ingleby S, et al: Bedside electronic capture of
renally cleared drugs in the hospital setting. J Am Med Inform Assoc clinical observations and automated clinical alerts to improve compli-
2010; 17:308–312 ance with an Early Warning Score protocol. Crit Care Resusc 2011;
42. Sintchenko V, Iredell JR, Gilbert GL, et al: Handheld computer-based 13:83–88
decision support reduces patient length of stay and antibiotic pre- 50. Schnipper JL, Hamann C, Ndumele CD, et al: Effect of an electronic
scribing in critical care. J Am Med Inform Assoc 2005; 12:398–402 medication reconciliation application and process redesign on poten-
43. Thursky KA, Buising KL, Bak N, et al: Reduction of broad-spectrum tial adverse drug events: A cluster-randomized trial. Arch Intern Med
antibiotic use with computerized decision support in an intensive care 2009; 169:771–780
unit. Int J Qual Health Care 2006; 18:224–231 51. Amarasingham R, Plantinga L, Diener-West M, et al: Clinical informa-
44. Paterno MD, Cina JL, Goldhaber SZ, et al. Preventing DVT and PE in tion technologies and inpatient outcomes: A multiple hospital study.
hospitalized patients: Improving a successful electronic alert. AMIA Arch Intern Med 2009; 169:108–114
Annu Symp Proc 2006:1058 52. Jones SS, Heaton P, Friedberg MW, et al: Today’s ‘meaningful use’
45. Thompson W, Dodek PM, Norena M, et al: Computerized physician standard for medication orders by hospitals may save few lives; later
order entry of diagnostic tests in an intensive care unit is associ- stages may do more. Health Aff (Millwood) 2011; 30:2005–2012
ated with improved timeliness of service. Crit Care Med 2004; 53. King WJ, Le Saux N, Sampson M, et al: Effect of point of care informa-
32:1306–1309 tion on inpatient management of bronchiolitis. BMC Pediatr 2007; 7:4
46. Herasevich V, Tsapenko M, Kojicic M, et al: Limiting ventilator-induced 54. Teufel RJ, Kazley AS, Basco WT Jr: Is computerized physician order
lung injury through individual electronic medical record surveillance. entry use associated with a decrease in hospital resource utilization
Crit Care Med 2011; 39:34–39 in hospitals that care for children? J Med Syst 2012; 36:2411–2420
47. Meyfroidt G, Wouters P, De Becker W, et al: Impact of a computer- 55. Teufel RJ 2nd, Kazley AS, Ebeling MD, et al: Hospital electronic
generated alert system on the quality of tight glycemic control. medical record use and cost of inpatient pediatric care. Acad Pediatr
Intensive Care Med 2011; 37:1151–1157 2012; 12:429–435

1282 www.ccmjournal.org June 2015 • Volume 43 • Number 6

Copyright © 2015 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.

You might also like