impact of a computerized physician order-entry system

8
Impact of a Computerized Physician Order-Entry System William M Stone, MD, FACS, Benn E Smith, MD, Judd D Shaft, BS, Richard D Nelson, BS, Samuel R Money, MD, MBA, FACS BACKGROUND: The Institute of Medicine has urged the adoption of electronic prescribing systems in all health-care organizations by 2010. Accordingly, computerized physician order entry (CPOE) warrants detailed evaluation. Mixed results have been reported about the benefit of this system. No review of its application in surgical patients has been reported to date. We present the implementation of CPOE in the management of surgical patients within an academic multi- specialty practice. STUDY DESIGN: Retrospective and prospective analyses of patient-safety measures were done pre- and post-CPOE institution, respectively. Other metrics evaluated included medication errors, order-implementation times, efficiencies, personnel requirements, and physician time. Sampling of time span for the order placement process was assessed with direct hidden observation of the provider. RESULTS: A total of 15 (0.22%) medication errors were discovered in 6,815 surgical procedures performed during the 6 months before CPOE use. After implementation, 10 medication errors were found (5,963 surgical procedures [0.16%]) in the initial 6 months and 13 (0.21%) in the second 6 months (6,106 surgical procedures) (p NS). Mean total time from placement of order to nurse receipt before implementation was 41.2 minutes per order (2.05 minutes finding chart, 0.72 minutes writing order, 38.4 minutes for unit secretary transcription) compared with 27 seconds per order using CPOE (p 0.01). Four additional informational technology specialists were temporarily required for assistance in implementing CPOE. After CPOE adoption, 11 of 56 (19.6%) ancillary personnel positions were eliminated related to order-entry efficiencies. CONCLUSIONS: Present CPOE technology can allow major efficiency gains, but refinements will be required for improvements in patient safety. (J Am Coll Surg 2009;208:960–969. © 2009 by the American College of Surgeons) The quality of health care provided in the US has come under increased scrutiny during the last several years. Spe- cifically, patient safety has become a national initiative driven not only by patients but government agencies as well. It has been well documented that medication errors in particular are potentially avoidable and harmful. 1 Most health-care organizational efforts have involved some sort of information technology solution for elimination of such errors. 2 In 2004, President Bush established, by Executive Order, the Office of National Coordinator for Health In- formationTechnology.This office was created to ensure the establishment of electronic forms of medical records throughout the US by 2014. In addition, the Institute of Medicine has called for use of electronic prescribing sys- tems in all health-care organizations by 2010. 3 This com- bination is believed to provide the appropriate platform for reduction or potential elimination of medication errors. Recommendations from the industry-based Leapfrog Group have additionally enhanced interest in technological solutions for prescribing errors. 4 Computerized physician order-entry (CPOE) systems have proved to be successful in reducing medication errors in hospital settings. 5 Most published reviews of the impact of CPOE implementation to date describe the processes for establishment of CPOE systems and reduction of medica- tion errors, physician behavioral change, and costs. These reviews involve inpatient medical services, including ICU settings, psychiatric wards, pediatric services, and various Disclosure Information: Nothing to disclose. Presented at the Southern Surgical Association 120th Annual Meeting, West Palm Beach, FL, December 2008. Received November 26, 2008; Accepted January 14, 2009. From the Division of Vascular Surgery, Departments of Neurology, Information Technology, and Systems and Procedures, Mayo Clinic, Phoenix, AZ. Correspondence address: William M Stone, MD, FACS, Division of Vascular Surgery, Mayo Clinic, 5777 E Mayo Blvd, Phoenix, AZ 85525. email: [email protected] 960 © 2009 by the American College of Surgeons ISSN 1072-7515/09/$36.00 Published by Elsevier Inc. doi:10.1016/j.jamcollsurg.2009.01.042

Upload: william-m-stone

Post on 05-Sep-2016

218 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Impact of a Computerized Physician Order-Entry System

IPWS

Tucdwphoe

DPP

RFIPCVe

©P

mpact of a Computerizedhysician Order-Entry Systemilliam M Stone, MD, FACS, Benn E Smith, MD, Judd D Shaft, BS, Richard D Nelson, BS,

amuel R Money, MD, MBA, FACS

BACKGROUND: The Institute of Medicine has urged the adoption of electronic prescribing systems in allhealth-care organizations by 2010. Accordingly, computerized physician order entry (CPOE)warrants detailed evaluation. Mixed results have been reported about the benefit of this system.No review of its application in surgical patients has been reported to date. We present theimplementation of CPOE in the management of surgical patients within an academic multi-specialty practice.

STUDY DESIGN: Retrospective and prospective analyses of patient-safety measures were done pre- andpost-CPOE institution, respectively. Other metrics evaluated included medication errors,order-implementation times, efficiencies, personnel requirements, and physician time.Sampling of time span for the order placement process was assessed with direct hiddenobservation of the provider.

RESULTS: A total of 15 (0.22%) medication errors were discovered in 6,815 surgical procedures performedduring the 6 months before CPOE use. After implementation, 10 medication errors were found(5,963 surgical procedures [0.16%]) in the initial 6 months and 13 (0.21%) in the second 6months (6,106 surgical procedures) (p � NS). Mean total time from placement of order tonurse receipt before implementation was 41.2 minutes per order (2.05 minutes finding chart,0.72 minutes writing order, 38.4 minutes for unit secretary transcription) compared with 27seconds per order using CPOE (p � 0.01). Four additional informational technology specialistswere temporarily required for assistance in implementing CPOE. After CPOE adoption, 11 of56 (19.6%) ancillary personnel positions were eliminated related to order-entry efficiencies.

CONCLUSIONS: Present CPOE technology can allow major efficiency gains, but refinements will be required forimprovements in patient safety. ( J Am Coll Surg 2009;208:960–969. © 2009 by the American

College of Surgeons)

OfetMtbrRGs

hioetr

he quality of health care provided in the US has comender increased scrutiny during the last several years. Spe-ifically, patient safety has become a national initiativeriven not only by patients but government agencies asell. It has been well documented that medication errors inarticular are potentially avoidable and harmful.1 Mostealth-care organizational efforts have involved some sortf information technology solution for elimination of suchrrors.2 In 2004, President Bush established, by Executive

isclosure Information: Nothing to disclose.resented at the Southern Surgical Association 120th Annual Meeting, Westalm Beach, FL, December 2008.

eceived November 26, 2008; Accepted January 14, 2009.rom the Division of Vascular Surgery, Departments of Neurology,nformation Technology, and Systems and Procedures, Mayo Clinic,hoenix, AZ.orrespondence address: William M Stone, MD, FACS, Division ofascular Surgery, Mayo Clinic, 5777 E Mayo Blvd, Phoenix, AZ 85525.

smail: [email protected]

9602009 by the American College of Surgeons

ublished by Elsevier Inc.

rder, the Office of National Coordinator for Health In-ormation Technology. This office was created to ensure thestablishment of electronic forms of medical recordshroughout the US by 2014. In addition, the Institute of

edicine has called for use of electronic prescribing sys-ems in all health-care organizations by 2010.3 This com-ination is believed to provide the appropriate platform foreduction or potential elimination of medication errors.ecommendations from the industry-based Leapfrogroup have additionally enhanced interest in technological

olutions for prescribing errors.4

Computerized physician order-entry (CPOE) systemsave proved to be successful in reducing medication errors

n hospital settings.5 Most published reviews of the impactf CPOE implementation to date describe the processes forstablishment of CPOE systems and reduction of medica-ion errors, physician behavioral change, and costs. Theseeviews involve inpatient medical services, including ICU

ettings, psychiatric wards, pediatric services, and various

ISSN 1072-7515/09/$36.00doi:10.1016/j.jamcollsurg.2009.01.042

Page 2: Impact of a Computerized Physician Order-Entry System

mtm

MMCftwtrnwbtmowpldncetbpotottaloed“s(wvcs

RIgrc

STps(d6cinowwawowtltC

OFFaourflfdmetaso

tasmteowsm

961Vol. 208, No. 5, May 2009 Stone et al Impact of a Computerized Physician Order-Entry System

edical specialties. Accordingly, we conducted a review ofhe impact of implementation of a CPOE system within aultispecialty academic surgical practice.

ETHODSayo Clinic Hospital in Phoenix, AZ, implemented aPOE system for its inpatient practice in a staged manner

rom May 8, 2007, through April 30, 2008. A retrospec-ive and prospective analysis of patient-safety measuresas done pre- and post-CPOE implementation, respec-

ively. Other metrics evaluated included medication er-ors, order-implementation times, efficiencies, person-el requirements, and physician time. Medication errorsere recorded using a self-reporting technique that hadeen implemented on inception of Mayo Clinic Hospi-al in 1998. For purposes of our review, we tabulatededication errors as “any medication error that may cause

r lead to inappropriate medication use or patient harm,hile the medication is in the control of the health-carerofessional, patient, or consumer. Such event may be re-

ated to professional practice, health-care products, proce-ures, and systems, including prescribing, order commu-ication, product labeling, packaging, and nomenclature,ompounding, dispensing, distribution, administration,ducation, monitoring and use.” Order-implementationimes (defined as time from initiation of the order to ordereing available to nursing, radiology, or laboratory) androvider time were recorded for both written and electronicrdering. Using a personal digital assistant, order-entryime (time for provider to find patient record and write therder) was measured for written orders by hidden observa-ion of providers in the ordering process. Data for the elec-ronic order-entry time (time for provider to access recordnd enter the order) was obtained electronically. Deviceogon time was obtained through electronic measurementf time to access the patient’s medical record on variouslectronic devices used by the providers. Four electronicevices were used, including standard desktop computers,modified” desktop computers with streamlined accessoftware, standard laptop computers, and “thin” laptopspersonal computers [PCs]). Order-implementation timesere obtained electronically and with hidden direct obser-ation. Human resources information was reviewed, in-luding work assignments, to obtain information on per-onnel requirements.

ESULTSmplementation of a CPOE system in a multispecialty sur-ical practice was evaluated using both prospective andetrospective evaluation. Impact outcomes measured in-

luded safety, efficiency, and financial data. 1

afetyotal order numbers during the CPOE implementationrocess were 1,836,239. This consisted of 1,202,330 phy-ician orders (66%), 589,196 mid-level practitioner orders32%), 22,969 phone orders (1%), and 21,774 verbal or-ers (1%). Six months before CPOE implementation,,815 surgical procedures were performed in our multispe-ialty surgical practice. There were 15 medication errorsdentified (15 of 6,815 [0.22%]). No specific trends wereoted in the cause of these errors. After implementationf CPOE, 10 medication errors (10 of 5,963 [0.16%])ere identified, with 5,963 surgical procedures performedithin the first 6 months. During the second 6 months

fter CPOE implementation, 6,106 surgical proceduresere performed, with 13 medication errors identified (13f 6,106 [0.21%]). When compared, medication errorsere not statistically significantly different during any of

hese time periods. The patient-safety score metric estab-ished by Mayo Clinic Hospital in 2002 revealed no statis-ically significant change when comparing pre- and post-POE scores (Fig. 1).

rder-implementation timesor written orders, the systematic process is outlined inigure 2. Blinded, direct observation with personal digitalssistant recording of data entry was performed on 53rder-entry sessions. This included samples from 6:45 AM

ntil 7:30 PM. Staff surgeons performed 21 (39.6%) of theecorded order-entry sessions. Physician assistants per-ormed 22 (41.5%) sessions and surgical residents and fel-ows performed 10 (18.9%). Mean time for a provider toind a chart was 2 minutes and 3 seconds, with a standardeviation of 4 minutes and 1 second (range, 1 second to 30inutes and 1 second) (Table 1). Mean number of orders

ntered per written order session was 4.3, with a range of 1o 34 orders. Mean time for writing all of the orders duringn order session was 2 minutes and 39 seconds (range, 3econds to 16 minutes and 2 seconds), with mean time perrder of 37 seconds.

The next step in the written ordering process involveshe unit secretary, who transcribes the order and makes itvailable to nursing, radiology, and laboratory. The unitecretary’s mean time for routine written orders was 51inutes and 24 seconds (medications, 45 minutes; labora-

ory orders, 41 minutes; radiology orders, 56 minutes; di-tary orders, 55 minutes; nursing orders, 47 minutes; andther orders, 62 minutes). If orders were given verbally (notritten), mean unit secretary time was 10 minutes and 30

econds (medications, 12 minutes; laboratory orders, 10inutes; radiology orders, 3 minutes; and nursing orders,

6 minutes). For all written orders (eg, routine, admission,

Page 3: Impact of a Computerized Physician Order-Entry System

do3

svodDflst(treo

to2otap

PBtvmriI(wctbnpCtt

DChncC

ospit

962 Stone et al Impact of a Computerized Physician Order-Entry System J Am Coll Surg

ischarge, verbal, emergency), mean time from completionf writing the orders until available for nursing services was8 minutes and 24 seconds (Table 2).Electronic orders were tabulated for 100 sample order

essions on 4 random days during a 4-month period. Pro-iders had access to four electronic devices to enter ordersn patients. These included a standard laptop, standardesktop, thin PC, and a modified desktop for easier access.evice logon and patient activation times were recorded

or each electronic device. Logon and activation times wereongest using standard desktop devices (1 minute and 1econd) followed by thin PC (57 seconds), modified desk-op (yellow screen) (30 seconds), and laptop (28 seconds)p � NS) (Fig. 3). Mean number of electronic orders en-ered per session was 8.12, with a range of 2 to 60. Timeequired to enter an individual order per order with thelectronic ordering system had a mean of 27 seconds perrder (range, 1 second to 19 minutes).

When evaluating time in aggregate, before implementa-ion of CPOE, time from physician needing to place anrder to receipt by nurse was 41.2 minutes. This included.05 minutes to find the chart, 0.72 minutes to write therder, and 38.4 minutes for unit secretary clarification andranscription. After CPOE, time from physician enteringn order to having it accessible to the nurse was 27 seconds

CL0.00131

0.00080 0

0.00000

0.00050

0.00100

0.00150

0.00200

0.00250

0.00300

0.00350

Jan-02

Apr-02

Jul-02

Oct-02

Mar-03

Jun-03

Sep-03

Dec-03

Feb-04

May-04

Aug-0

Mont

Clin

ical

ly S

igni

fican

t Eve

nt R

ate

Figure 1. Mayo Clinic H

er order (p � 0.01) (Fig. 4). o

ersonnel requirementsefore implementation of CPOE, four additional informa-

ion technology personnel were temporarily added to pro-ide support for the implementation process. After imple-entation of CPOE, the unit secretary was no longer

equired to clarify orders and transcribe the written ordersnto a format for nursing, radiology, or laboratory to access.n lieu of decreased unit secretary work load, 11 of 5619.6%) ancillary personnel positions (unit secretaries)ere eliminated after CPOE implementation. The mean

ompensation package for a unit secretary at our institu-ion is $40,500. This would translate into a yearly financialenefit of $445,500. The four additional information tech-ology positions added before CPOE have been rede-loyed within other areas of the practice as familiarity withPOE processes has been established. Total capital cost for

he implementation project was $2.9 million, with addi-ional operating cost of $2.3 million.

ISCUSSIONPOE systems, as with any other disruptive technology,ave specific, intended consequences. The addition of anyew technology also has unexpected consequences, whichan play a more pivotal role. The intended consequence ofPOE systems is to provide a safer environment for the

0.00026

3

0.00033 0.00041

an-05

Apr-05

Jul-05

Oct-05

Mar-06

Jun-06

Sep-06

Dec-06

Feb-07

May-07

Aug-07

Nov-07

ar

U

UCL

+2 Sigma

+1 Sigma

Average

-1 Sigma

-2 Sigma

LCL

al patient safety score.

.0008

4

Nov-04

J

h/Ye

rdering process, thereby decreasing medication errors.

Page 4: Impact of a Computerized Physician Order-Entry System

MCrticocs

c

BtrdatgctfrtrodmJSt

senta

TV

MMRMMn

*

963Vol. 208, No. 5, May 2009 Stone et al Impact of a Computerized Physician Order-Entry System

ost published reviews that have evaluated the impact ofPOE implementation have revealed beneficial effects on

eduction of medication errors. In our review, we foundhat there was no substantial impact on the rate of med-cation errors, but we found considerable gains in effi-iency for the ordering process. Although not easily provedbjectively, these time improvements can lead to patientare improvements and potentially a decrease in length oftay.

Since its introduction in 1969, only 5% of medical fa-ilities to date have instituted some form of CPOE system.6

Figure 2. Diagrammatic repre

able 1. Physician Written Order Processariable Time* to find a chart

ean 0:02:03edian 0:00:43

ange 0:30:01inimum 0:00:01aximum 0:30:02

90

mh:min:s.

ased on the most optimistic of predictions, it appears ashough diffusion across the US will take at least 20 years toeach as many as 50% of the medical facilities. Rogers’iffusion of innovations model suggests five patterns ofdoption for any new technology.7 He identified innova-ors, early adopters, early majority, late majority, and lag-ards as the five patterns of adoption. In light of the highosts, complexity, and physician skepticism, it is doubtfulhat the CPOE systems will gain much traction in the nextew years. But if both direct and indirect cost savings areealized, more institutions will likely show increased accep-ance of implementation. This increased acceptance willesult from the perceived cost savings, but it might also beffset by the recent economic downturn and substantialecreases in available capital funds to purchase and imple-ent the system. In addition, if regulatory agencies, such as

oint Commission or Centers for Medicare & Medicaidervices, mandate electronic ordering systems, all institu-ions will most likely migrate quickly to implementation.

Most of the outcomes measures evaluated from imple-

tion of written order process.

entation of CPOE systems have revolved around patient

Page 5: Impact of a Computerized Physician Order-Entry System

sgsblTuhnlCms

tfimteedrrqe

mrcvgtAncwegdoinetF

d

T

V

RRASSDTP

PG

*I gency

964 Stone et al Impact of a Computerized Physician Order-Entry System J Am Coll Surg

afety. The available reviews on CPOE impact have sug-ested that it has a positive effect on patient safety. Patientafety has been investigated primarily by measuring num-ers of medication errors. Each study identifies and tabu-

ates medication errors using its own distinct methodology.here is no consistent standard for identification and tab-lation of these errors. Some medication errors result inarm to patients (adverse drug events), and some result ino actual or potential harm. In a systematic review by Es-

ami and colleagues,8 21 studies evaluating the impact ofPOE on medication errors, adverse drug events, andortality were analyzed. In all studies, there was either a

tatistically significant benefit on medication errors or a

Standard Workstation0:00:00

0:00:43

0:01:26

0:02:10

0:02:53

0:03:36

0:04:19

0:05:02

ThinPC

Yellow Screen Laptop

igure 3. Device logon and patient record activation times by

able 2. Time Measurement Results for Written Orders

ariable nNo. oforders

Mean writetime*

Mean timMedicatio

orders

outine written orders 86 319 0:00:43 0:45:14outine verbal orders 7 11 0:02:10 0:12:30dmission orders 4 110 0:00:21 0:43:48TAT written orders 7 17 0:01:32 0:13:58TAT verbal orders 3 6 0:00:53 0:06:23ischarge orders 1 4 0:01:15 0:26:41otal/mean 108 467 0:01:09 0:24:45ostop ordersPostop to floor 5 132 0:00:16 7:02:05Postop to PACU 1 2 0:00:16Subtotal 113 601 0:00:16

reop orders 1 12 0:00:37rand total 114 613

h:min:s.DX, software utilized by IBM; PACU, postanesthetic care unit; STAT, emer

revice (h:mm:s).

rend toward benefit. All trials reviewed were not per-ormed in a prospective randomized fashion, but they didnvolve comparisons between pre- and post-CPOE imple-

entation. Although evidence is present to demonstratehe beneficial effect of CPOE on medication errors, littlevidence is present to demonstrate its effect on adverse drugvents. Bates and colleagues9 performed a prospective ran-omized comparative study that was the first review toeveal a substantial reduction in medication errors (55%eduction of serious medication errors), yet was not ade-uately powered to reveal any statistically significant differ-nce in adverse drug events.

Although most reviews have found beneficial effects onedication errors, recent evidence suggests that drug-event

ates and mortality can be adversely affected. Koppel andolleagues10 performed a qualitative and quantitative re-iew of CPOE implementation using five different focusroups. They identified 22 types of medication errorshat were believed to be facilitated by the CPOE system.lthough there were medication errors that were elimi-ated, these other additional errors were unintendedonsequences of the system. The potential errors thatere unintended included orders on the wrong patient,

rrors of omission, nurses not knowing an order had beenenerated, desensitization to alerts, loss of informationuring care transitions, wrong medication dosing, andverlapping medication orders. In our review of CPOEmpact on medication errors in the surgical practice, weoted no statistically significant difference in medicationrrors when comparing pre- and post-CPOE implementa-ion. The capture and reporting of medication errors in our

rom start of writing order to time entered in IDX for:

Meanaboratoryorders

Radiologyorders

Nursingorders

Dietaryorders

Otherorders

0:41:33 0:56:56 0:47:35 0:55:04 1:01:59 0:51:240:10:06 0:03:00 0:16:25 0:10:301:06:06 1:28:40 0:57:03 1:12:29 1:03:31 1:05:161:13:14 0:37:39 0:03:12 0:19:00 0:29:250:00:25 0:06:21 0:04:23

0:26:410:38:17 0:46:34 0:37:00 0:43:35 0:40:14 0:38:24

7:56:58 7:56:18 7:49:58 8:26:08 8:02:206:43:28

.

e* fn L

eview were performed using a self-reporting process prac-

Page 6: Impact of a Computerized Physician Order-Entry System

tMnwdir0drdsnm

mifppftiacctrofa

isS

pcapgt

aotocaopTmltdmspsasticdbqcor

gdwEotgsmdtta

965Vol. 208, No. 5, May 2009 Stone et al Impact of a Computerized Physician Order-Entry System

iced in our inpatient hospital since its inception in 1998.ost reviews have used differing methodologies and defi-

itions for medication errors. These definitions includerong dosing, route and interval errors, drug interactions,rug allergies, wrong medication, formulary errors, illegibil-

ty, transcription errors, administration errors, dispensing er-ors, and contraindications. Medication errors were found in.22% of patients in 6,815 surgical procedures in the imme-iate 6 months before CPOE implementation. After CPOEollout, there were 0.16% of patients in 5,963 surgical proce-ures in the first 6 months and 0.21% of patients in 6,106urgical procedures in the second 6 months. As stated here, theumber of errors was not substantially impacted by imple-entation of CPOE in our practice.Several factors can contribute to the lack of improve-ent. Our baseline rate of medication errors before CPOE

mplementation was quite low. Sharek and colleagues11

ound the incidence of medication-related harm in the in-atient pediatric population to be as high as 11.1 per 100atient admissions. In addition, Dean and colleagues12

ound prescribing error rates of 0.3% to 39% of medica-ion orders written in the adult inpatient setting. Unantic-pated errors with CPOE implementation could have offsetny gains made in the intentional aversion of errors. Be-ause our system of medication error is self-reported, thereould have been increased vigilance in reporting and cap-uring of these errors after implementation of CPOE. Theesults could have been influenced by behavioral changes inur organization, but we are possibly not substantially dif-erent from other institutions that do not find a consider-ble decrease in medication errors.

Patient safety is not simply reducing the number of med-cation errors. Obtaining treatment more efficiently alsohould improve patient outcomes and possibly reduce costs.

Figure 4. Times for ordering process in aggregate.

everal other measures impacted by CPOE have been re- b

orted, including adherence to guidelines, time, costs, effi-iency, and satisfaction. All of the reviews to date have evalu-ted the impact on either a pediatric or medical specialtyractice. Our review involved a multispecialty academic sur-ical practice. Although our practice includes medical special-ies, our review focused on the surgical cohort.

The ordering process can be broken into small identifi-ble steps that can be individually analyzed. For the writtenrdering process (Fig. 5), the most time-consuming por-ion involves the unit secretary and transcription of therder into a form so that nursing, radiology, and laboratoryan be notified. The process is initiated when the providerttempts to find the patient chart and then transcribes therder for the unit secretary to process. Mean time for therovider to find the patient chart was slightly � 2 minutes.he longest time finding a chart was documented at 30inutes. In the electronic environment, this process corre-

ates to finding an electronic device to log on and activatinghe patient’s medical record. This time was electronic-evice dependent, but even using the slowest device, theean time to log on and patient record activation was

lightly � 1 minute. Comparing the electronic and writtenrocesses was comparable before the time required for unitecretary processing. The electronic order process had thedvantage of not requiring the unit secretary input andaved approximately 38 minutes (Fig. 2). This unit secre-ary processing time included not only time required tonput the particular orders but, more importantly, it in-ludes time for clarification of the written order. Not onlyoes poor handwriting necessitate frequent clarification,ut medication dosing, allergies, and other factors fre-uently result in the need to contact the ordering physi-ian. In addition, by removing the unit secretary from therdering chain of events, human error could potentially beeduced. But our data did not support that hypothesis.

The impact on patient care of this 38-minute efficiencyain with CPOE systems in the overall ordering process isifficult to quantitatively measure. In three reviews, CPOEas found to increase time in the ordering process.13-15

ach of these three reviews evaluated physician time as onef the primary outcomes measured, not the total time forhe ordering process. These reviews did show considerableains in efficiency (time) in the pharmacy and in nursingervices. Drug “turnaround” time, defined as time fromedication order entry to the time the medication was

elivered to the ward, was measured and evaluated. Thisime, which is more consistent with our methodology forime and efficiency outcomes, was found to be consider-bly improved with implementation of a CPOE system.

Overall financial impact of CPOE implementation has

een reported to have a positive effect on the net operating
Page 7: Impact of a Computerized Physician Order-Entry System

isIliwttseftpoiqq[ccbcbtmel

myafmte

pmnaiiwfCmfb

A

SA

rderi

966 Stone et al Impact of a Computerized Physician Order-Entry System J Am Coll Surg

ncome of an institution.5 Costs of information technologyystems can be divided into capital and operating expenses.n our institution, both of these costs were substantial. Theeadup phase before implementation of CPOE was approx-mately 28 months. This includes assuring that all supportas available for the providers (additional laptops, addi-

ional thin PCs, educational modules, and so forth) andhat adequate technical support and buildup of the CPOEystem was in place. Our inpatient hospital has used anlectronic medical record since its inception in 1998. Be-ore implementation of the inpatient hospital CPOE sys-em, an outpatient CPOE system was introduced and ex-erience had already been gained in this process. Althoughverall costs appear easier to obtain than overall financialmpact, the efficiencies gained are even more difficult touantify. Personnel changes have occurred as a conse-uence of work-load redistribution (eliminated 11 of 5619.6%] unit secretarial positions). This resulted in a finan-ial gain of $445,500 for our institution. The overall finan-ial impact of this redistribution and efficiency gain has noteen established. Several reviews have focused on the finan-ial impact of CPOE. Most have established a financialenefit to CPOE in at least one parameter studied; buthese were primarily focused reviews without establish-ent of overall impact. It appears CPOE has a beneficial

ffect on hospital and pharmacy costs. Kaushal and col-

Figure 5. Diagrammatic representation of electronic o

eagues16 found the capital and operational costs of imple- A

enting a CPOE system were $11.8 million during a 10-ear period. When factoring the overall gains in efficiencynd personnel, it was believed to have an overall positiveinancial impact. Our costs were likewise substantial ($2.9illion capital, $2.3 million operating during implemen-

ation phase), but the full financial effect has not beenvaluated and might not truly be objectively quantifiable.

Implementation of a CPOE system in our inpatient hos-ital practice resulted in no substantial change in the rate ofedication errors. Considerable gains in efficiency were

oted, which included the time necessary to have ordersccessible to nursing, radiology, and laboratory. This gainn efficiency will likely result in longterm cost savings andncreased quality of care. Additionally, personnel needsere reduced, which subsequently resulted in additional

inancial benefit for our institution. Although the currentPOE systems have shown benefit, additional enhance-ents of the software to provide a more user-friendly plat-

orm will result in wider adoption of CPOE and greaterenefit to our patients.

uthor Contributions

tudy conception and design: Stone, Smith, Moneycquisition of data: Smith, Shaft, Nelson

ng process (postcomputerized physician order entry).

nalysis and interpretation of data: Stone, Smith, Nelson

Page 8: Impact of a Computerized Physician Order-Entry System

DC

R

1

1

1

1

1

1

1

D

Di

sobeocra

aoiwiUpwvtifpCa

cerrfsCiae

rCDhwoamtwo

DtaaAi

i

967Vol. 208, No. 5, May 2009 Stone et al Impact of a Computerized Physician Order-Entry System

rafting of manuscript: Stone, Moneyritical revision: Stone, Smith, Shaft, Money

EFERENCES

1. Institute of Medicine. Preventing medication errors: qualitychasm series. Washington, DC: National Academy Press; 2007.

2. Kohn LT, Corrigan JM, Donaldson MS. To err is human: build-ing a safer health system. Washington, DC: National AcademyPress; 1999.

3. Institute of Medicine. Crossing the quality chasm: a new healthsystem for the 21st century. Washington, DC: National Acad-emy Press; 2001.

4. Birkmeyer JD, Birkmeyer CM, Wennberg MP, et al. Leapfrogsafety standards: potential benefits of universal adoption. Wash-ington, DC: The Leapfrog Group; 2000.

5. Eslami S, de Keizer NF, Abu-Hanna A. The impact of computer-ized physician medication order entry in hospitalized patients—asystematic review. Int J Med Inform 2007;77:365–376.

6. Ford EW, McAlearney AS, Phillips MT, et al. Predicting comput-erized physician order entry system adoption in US hospitals: canthe federal mandate be met? Int J Med Inform 2008;77:539–545.

7. Rogers EM. Diffusion of innovations. Glencoe, NY: Free Pressof Glencoe; 1962.

8. Eslami S, Abu-Hanna A, de Keizer NF. Evaluation of outpatientcomputerized physician medication order entry systems; a sys-temic review. J Am Med Inform Assoc 2007;14:400–406.

9. Bates DW, Leape LL, Cullen DJ, et al. Effect of computerizedphysician order entry and a team intervention on prevention ofserious medication errors. JAMA 1998;280:1311–1316.

0. Koppel R, Metlay JP, Cohen A, et al. Role of computerizedphysician order entry systems in facilitating medication errors.JAMA 2005;293:1197–1203.

1. Sharek PJ, Classen D. The incidence of adverse events and med-ical error in pediatrics. Pediatr Clin North Am 2006;53:1067–1077.

2. Dean FB, Vincent C, Schachter M, et al. The incidence ofprescribing errors in hospital inpatients: an overview of the re-search methods. Drug Saf 2005;28:891–900.

3. Bates DW, Boyle DL, Teich JM. Impact of computerized phy-sician order entry on physician time. Proc Annu Symp ComputAppl Med Care 1994:996.

4. Shu K, Boyle D, Spurr C, et al. Comparison of time spentwriting orders on paper with computerized physician order en-try. Proc Med Inform 2001:1207–1211.

5. Banet GA, Jeffe DB, Williams JA, et al. Effects of implementingcomputerized practitioner order entry and nursing documentationon nursing workflow in an emergency department. J Healthc In-form Manage 2006;20:45–54.

6. Kaushal R, Jha AK, Franz C, et al. Return on investment for acomputerized order entry system. J Am Med Inform2006;13:261–266.

iscussion

R SUSAN GALANDIUK (Louisville, KY): Dr Stone presents an

ntriguing study of the effect of computerized order entry on patient e

afety measures, specifically in the surgical patient. The focus here isn the reduction of medication errors in surgical patients in the timeefore and after the introduction of CPOE. Because medicationrrors at his institution are self-reported and occurred in less thanne-third of 1% in the baseline period, it not surprising there was nohange in the incidence after the introduction of CPOE. The onlyeal medication errors that CPOE reduces are those OF transcriptionnd illegibility.

In the article the authors focus on reporting the time from writingn order to when it is “available” for the nurse, laboratory, or radiol-gy, and how this differed before and after CPOE. I argue that themportant case in point is not when the order is available, but ratherhen it is “carried out.” Is the patient getting more prompt service, or

s this really only “an electronic mirage?” One case in point is our ownniversity Hospital, where we actually had to install a special com-uter screen so the nurses knew when an electronic order was beingritten for their patients. We also noted that several consulting ser-ices neglected to check electronic orders and therefore delayed pa-ient care. Dr Stone, do you have any data on this aspect in your ownnstitution? CPOE also can have some interesting “controlling” af-ects on resident behavior. For example, in our own University Hos-ital Pharmacy, it includes only a very limited formulary on itsPOE system, so it makes it very difficult to order medications that

re not listed in this framework.CPOE does not prevent medication or transfusion errors. Re-

ently, one of my patients with rectal cancer who was to undergo anlective loop ileostomy closure after a colo-anal anastomosis nearlyeceived a preoperative blood transfusion when a family practiceesident who wasn’t familiar with the CPOE system, and even lessamiliar with surgery, was thinking to order blood to be available forurgery when instead he ordered a preoperative blood transfusion.POE processed things so quickly that the blood was already hang-

ng ready to be transfused when the patient questioned the nursebout this and mentioned all that I had told him about the adverseffects of blood transfusion in patients with rectal cancer.

One problem I have with CPOE is time. In view of the 80-houresident work week that we have been discussing this morning,POE takes longer than handwriting of orders despite the data thatr Stone has shown. Dr Stone in the article gives information about

ow many staff surgeons actually wrote orders in the pre-CPOE era,hat is that now? There’s information about a log-on time of 27 sec-nds and how it takes about 27 seconds to write each order. With anverage of eight orders per session, that comes out to about fiveinutes per patient. At our institution, it takes residents much longer

o write orders electronically, and this is almost like electronic editinghere the work of editorial assistant has actually been put in the handf a journal reviewer.

R EUGENE M LANGAN III (Greenville, SC): I would like tohank the Association for the pleasure of the floor to review Dr Stonend colleagues’ work on a computer order entry system installationnd the impact on the Department of Surgery at the Mayo Clinic inrizona. I would like to thank the authors for sending me a copy well

n advance.This article describes a single institution’s implementation of an

npatient CPOE on a department of surgery and the outcomes gen-

rated by a change from handwritten orders. Unfortunately, I have