rying to improve quality by eliminatingprocess steps might appear to be a contra-

'diction. But this was not the case when a200-plus full service;hospital in Illinois targetedmedication error reduction as a 2005 goal.

In fact, the Joint Commission on Accreditation of

__ Inn 5OWordsOr Less

* An Illinois hospital reduced work while improving

process performance 90% during a medication error

reduction project.

e A Black Belt helped a multidisciplinary team use Six

Sigma's define, measure, analyze, improve and control

methodology.

* The team then used quality function deploymentto

design and develop the process functions.

Healthcare Organizations (GCAHO) accreditedAlton Memorial Hospital achieved a more than90% improvement in process performance as aresult of the project.

Background

Alton Memorial, part of the St. Louis based BJCHealthCare, naturally supports a culture of patientsafety as part of its mission. The Agency forHealthcare Research & Quality estimates the inci-dent rate for an adverse drug event (ADE) atbetween two and seven per 100 hospital admissionsnationwide, with a mean cost of $4,685 per event.'

At only 0.02%, Alton Memorial's medicationerror rate was low compared to those nationwidestatistics. But the potential cost savings to the insti-tution that would result from even a smallimprovement in ADEs could not be ignored.

Of those medication errors reported in the hospi-tal's risk management database, 43% were causedby transcription errors. Transcription involves thecopying of a physician order for the purpose of pro-cessing. The strict guidelines that define a transcrip-tion error are based on the National CoordinatingCouncil for Medication Error Reporting andPrevention (NCC MERP) error category indexý

38 J JANUARY 2007 J www.asq.org

Information copied incorrectly or omitted is con-sidered a transcription error. This includes missing,inaccurate or only partially provided special instruc-tions with a physician order. An example of a spe-cial instruction is: Give only if pain level greaterthan five one hour after Tylenol.

A multidisciplinary team was formed at the hos-pital to find ways to reduce the hospital's medica-tion errors. Team members included representativesfrom pharmacy, nursing, clinical informationsystems, nursing management, performanceimprovement and medication safety. Hospitalmanagement requested the help, of a Six SigmaBlack Belt (BB) to guide the team through theprocess. A BB was assigned to this team in a facilita-tor and trainer role.

Methodology

Initially, the team followed a define, measure,analyze, improve and control (DMAIC) methodol-ogy. In the define phase the team realized hospitalmanagement mandated two distinct goals:

"* Reduce the defect rate of the current processwith quick hit initiatives.

"* Develop a standardized procegs that works forall hospital units except the emergency unit

(EU), because of the immediate nature of itsneeds.

With these two goals, the team expected toimprove order entry accuracy related to transcrip-tion errors by 50%.

In the measure phase, the team flowcharted thecurrent medication order entry processes. Theteam quickly realized each-unit had developed itsown medication ordering process. Most used twodocuments to review each order: the medicationadministration record (MAR) and a chronologicalsheet, a tool devised and used only by nursingstaff.

The intent of the chronological sheet wastwofold:

* Provide a historical list of medicationsreceived throughout a patient stay.

SAct as a repeat check against the MAR eachshift.

The purpose of this double check system was tohelp ensure order accuracy. Although the MARinformation should have been verified against thephysician orders, the staff deemed this step toocumbersome and labor intensive and created thechronological sheet for nurses to rewrite everymedication order by hand.

QUALITY PROGESS I JANUARY 2007 1 39

Eliminating a DocumentIt became clear that data needed to be collected to

compare the accuracy of the MAR and the chrono-logical sheet. A small baseline study was performedin the third quarter of 2004. The phar,macist's orderentry accuracy in the MAR was 95% the first timethrough, yet once the order was verified by thepharmacist, the pharmacy software and the nursingstaff, the MAR accuracy increased to 99.98%.

On the other hand, the chronological sheet wasitself fairly inaccurate'(10% error rate). Unfortunately,nurses had developed a false sense of security with

These changes improved themedication order entryprocess, yet the issue of themultiple ways staff processedmedication errors remained.

the chronological sheets, assuming they were correct.Furthermore, this tool was used during the entirepatient stay, so mistakes could carry through multipledays.

Concurrently, interpretations of orders, abbrevia-tions and the definition of the policy guidelineshad differences between the two forms. These vari-ances were contributing to the frequency and typeof transcription errors.

The team concluded the chronological sheet hadto be retired because it added no value to medica-tion order entry accuracy. Eliminating it represent-

.ed an average seven minute workload reductionper patient per day.

Reducing Pharmacist Interruptions

By analyzing cases, the team identified disrup-tions to the pharmacist during the order entryprocess as a repeating cause of mistakes. Actionstaken to reduce disruptions included:

Establishing a process for all intravenous fluid

40' JANUARY 2007 wf,v.asq.org

orders to be sent to pharmacy before 6 a.m.daily so they could be prepared early.

e Establishing a missing medication sheet forthe clinical units to fax instead of phoningtheir questions and concerns to the pharmacy.In the pharmacy, a technician would reviewthe request and assess whether it could behandled by someone other than a pharmacist.After completing the order entry for a givenpatient, the pharmacist then would addressthe waiting requests prior to starting the orderentry for the next patient.

o Training pharmacists on a consistent medica-tion order entry format.

Another cause of transcription errors was theillegibility of the physician order. There was notenough space on the form for physicians to writemedication orders, forcing them to write outsidethe assigned box or in tiny script. This caused mul-tiple issues such as illegible orders, eliminatedwords and missing information.

Ten more lines were added to the form. Marginswere darkened to reduce the possibility of writingoutside the faxable area. The guidelines inside thewriting area were made to be invisible when faxedto pharmacy. Finally, fax machines were tested andadjusted to improve quality.

These changes improved the medication orderentry process, yet the issue of the multiple waysstaff processed medication errors remained.

Standardization

The team shifted its focus to the second goal:designing one standard medication order processfor all hospital units to use, except the EU. The newprocess had to function flawlessly in such diverseareas as critical care, where patient medicationschange constantly, and long-term care wherepatients tend to have a long list of medications.

To achieve this goal, the team decided to follow adesign for Six Sigma methodology, quality functiondeployment (QFD), to link the needs of the cus-tomers with the design and development of theprocess functions. QFD helps organizations identifyboth spoken and unspoken needs, translate theseinto actions and designs, and focus various busi-ness functions toward achieving this common goal.

The first question to answer was: Who is ourcustomer in the medication order entry process?

Even though the ultimate cus-tomer is the patient, the primaryand more immediate customerwas determined to be the nurse incharge of the patient.

The team then set about cap-turing the voice of the customer(VOC), getting insights into thenurses' needs through interviews.The top needs identified were:

" Quick access to medicationorder information. It shouldtake no more than oneminute and three steps toaccess a patienes medicationinformation.

" Quick pharmacy turn-around time. Orders shouldbe entered into the MAR sys-tem within 120 minutes. Areal-time copy of the order should be availablewhile the order is being entered into the MARsystem.

* The process must provide a history of pa-tient medications. Nurses should be able tosee all medication orders, including start andstop times for the entire patient stay.

* Information should be portable and mobile.Nurses should not need to return to the nurs-ing station to learn what medications are dueto be administered and when.

"* The new process should provide doublechecking capability for MAR information.

"* Order entry needs to be trustworthy. Itshould be at least 95% accurate the first timethey are entered in the MAR, Pharmacistsshould be consistent in the way they enterthe order.

A hierarchical value then was assigned to eachcustomer need, representing the relative impor-tance the need had with respect to other needs.For instance, a hierarchical value of 2% can beinterpreted as meaning only 2% of the customersconsidered it important. Table 1 summarizes theresults.

The team analyzed the current processes and theperceptions nurses from all units had of eachprocess's effectiveness. The medication order entryprocess flowchart for each unit also was revisited.

.ý t Hierarchy of Nurses' Needs

Nurses'"need Defined as: Hierarchy

Q s Available immediately for printing (less than one minute). 2%Less than three steps to access the document. 2%

Quick pharmacy Less than two hours. 38%turnaround time Fax of order available in real time. 7%

Provide history Able to see all medications forthe entire patient stay. 2%of meds Able to see medication start and stop dates. 2%

Portable/mobile Nurse should not need to go back to station for document. 2%

Serve to double check all orders in medication 12%Check on MAR administration record (MAR).

Medications sort in the same way as MAR. 7%

At least 95% accurate the first time. 19%

Consistency in order entry by all pharmacists. 7%

MAR= medicaton administration record.

Because each unit followed a different order entryprocess, its evaluation of the current process per-formance was relative to its own application of it.

The nurses' needs then were converted intoquantitative measures of success with establishedmetrics and targets. These would become thedesign requirements for the new process. Eachdesign requirement was rated against the customer(nurse) needs in terms of the strength of its correla-tion. The resulting QFD matrix is shown in Figure 1.

Analyzing Options

Now that the design requirements had beendefined and the nurses' needs measured, the nextstep was to analyze options. In this phase the teamgenerated alternative concepts of processes thatcould be designed. These concepts then were eval-uated against customer (nurse) needs from Table 1.

Some of the concepts considered were:1. Use the existing patient care activity record

(PCAR) system as the medication list, withsome minor enhancements. PCAR uses theMAR information to generate its medicationlist-one source of information. This processalready was scheduled to be replaced by anelectronic medication bar coding systemcalled the medication administration checkingsystem (MAK), slated for implementation m,the fall of 2005.

QUALITY PROGRESS I .JANUARY 2007 1 41

n Quality Function Deployment Matrix

+ +

-A- + I- �

+-

4 �-�-f

- L - - I

+

Correlation Matrix

+ I+

+ I Fax to MAR line

Desktop to document

MD order availability

Process to entermedication

Availability printedcopy of documentFirst pass percentageorders incompleteFirst pass numberof defects

-+-

Direction of improvement 4 .1, Q Q Q l. , Customer rating

requirements

Customerreouirements

a

a.32-

WE

a

ca

0..

C3 If

72

~ 0

a

0-

COc.c

0

aa -aaa aak..a. 0

aU, _

- a�IZ�

a

CAccCL U, -

U,a.0II

.- C�J C�2 �* 10

Available immediately 3 9 3 1 1 @ xfor printing

Quick access -

pf-ntingLess than three steps to 3 1 1 @ Xaccess the document

Quick pharmacy QuickTAT 50 9 1 3 9 3 3 3 @x I

turn around Facsimile of order available 10 3 9 3 1 3 3 @xin real time

Able to see all m'eds for 3 3 1 3 9 1 1 1 @ x#Provide history the entire patient stayof medications Able to see med start 3 3 1 3 9 1 1 1 x @1

and stop datesPortable/mobile Nurse should not3needtogo 1 1 9 1 1 x @

back to station for document

Servetodoublecheck 15 3 1 3 9 3 1 1 xf @all orders in MARCheck on MARMedications sort in the same 10 3 9 3 1 1 @x'

way as MAR

At least 95% accurate 25 3 1 1 3 9 9 @ # xTrustworthy the firsttime ,_____ I

Consistency in order entry 10 3 1 1 9 9 9 @ # x, by all pharmacists

-I Technical I Absoluteimportance I Relative (percentage)

714 , 14b 344 93U ZII 29 3292114110 271 8 115 115

MAR = medication administration record, @= in-patient unit, x = labor and delivery, I = long-term care.

42 1JANUARY 2007~ 1wwasq.or,g

2. Deploy the MAK systemearly. MAK would detectany discrepancy betweenthe medication and theorder at the bedside. It wasscheduled for implementa-tion in the fourth quarterof 2005. This proposalcalled for extra resourcesto expedite the system'slaunch by the spring of2005.

3. Use optical characterrecognition (OCR) tech-nology. OCR equipmentscans images of medica-tion orders as they arefaxed to the pharmacy. Itconverts handwriting intoelectronic lists of medica-tions stored in a folder thatis accessible with a mouseclick.

A Pugh selection matrix wasused to evaluate these conceptsagainst customer (nurse)needs? The Pugh matrix com-pares multiple concepts againsta baseline model in terms ofhow well they address VOC.The preferred concept can beeither the one with the highestnumber of plus signs minus thenumber of minus signs or anew concept that incorporates superitics of the proposed concept ideas.

Based on the Pugh matrix results irteam decided to implement the PCAIbecause it had the lowest number of rsame number of plus signs as MAK,easy to test and could be implemente(

Option two, MAT, would have reqcant hardware and structural changesmedication is ordered and delivered.time was required than was availabletechnology requires the purchase of hThere also were reliability and capabiregarding its ability to read physicianing.

ýý Pugh Selection Matrix

______I-

Available immediately for printing.

Less than three steps to access the document.

Quick turnaround time.

Facsimile of order available in real time.

Able to see all medications for the entire patient stay.

Able to see medication start and stop dates.

Nurse should not need to go back to station for document.

Serve to double check all orders in MAR.

Medications sort in the same way as in MAR.

At least 95% accurate the first time.

Consistency in order entry by all pharmacists.

=

E=

+ = Betterthan datum- = Worse than datumS = Same as datum

or characteris- Going Live

The team discussed verification plans, includingFigure 2, the a pilot program. A go live date of April 4 wasoption selected because of the usual low patient census on

ninus signs, Sunday nights. Mandatory training sessions wereret was very programmed throughout the month of March tod quickly. educate nursing staff on the new process.uired signifi- On April 3, the team performed a second hospi-to the way talwide baseline study to reassess the accuracy of

More lead the current process. For simplicity's sake, the met-.The OCR ric used was errors per bed instead of the moreardware. representative errors per order. The resulting base-lity concerns line value was 0.4 errors per bed, a value similar tos' handwrit- the results of the initial baseline study conducted at

the beginning of the project, which yielded 90%

QUAUTY PROGRESS J JANUARY 2007 1 43

Q[ Control Chart of Medical Transcription Errors per Bed

0.80-

0.70-

0.60 - -

0.50 . ........

SO 0.40-

"0.30-

Ln J I -n L I I n Ln Ln Ln I n L o . L n I Lo

CL 0 75 -: 7Ei p p p p< ?5L f- -1 <- ~ d c ~ ~~

<) < . -- ~

accuracy per order or' 0.5 errors per bed.At midnight, the chronological sheet was

removed from all patient charts and the PCARprocess was put into place. During the day theteam surveyed all units, answered questions, gath-ered feedback and supported the nurses through-out the transition.

The new process uses the physician orders asthe double check for the MAR accuracy. When anew medication order is written, the nurse keepsthe information in the PCAIZ Once pharmacyenters the order into the MAR, the nurse verifies itagainst what was in the PCAR and signs off theorders as entered in the MAR. In case of discrep-ancy, the nurse goes back to the physician for veri-fication. Any changes are recorded in the MARand corrected in the PCAR on the next printout.

Of course, the keystone of the PCAR initiative'ssuccess is for staff to actually verify new medica-tion orders in the computer before signing off onthem. This process was not happening with thechronological sheet, 'thus mistakes could remain

undetected for multiple days.The PCAR process is quite similar to the process

required by the new MAK. Starting it now wouldallow for a seamless transition to MAK later.

Results

A control plan was developed to collect data onthe current process and ensure the improvementsare maintained over the long term. The planaddresses issues regarding the type of randomaudits to perform, by whom, how often and thereaction plan if the process goes out of control.

These random audits are conducted weekly bythe medication safety officer. An average of 30chart audits are completed per week based on indi-vidual nursing unit census. The results of theaudits are shown in a control chart (see Figure 3).

The control chart allows the process owner tomonitor changes in medication errors by distin-guishing random day-to-day variation from vari-ation caused by a significant change in processperformance. In the control chart the average

44 1 JANUARY 2007 1 www.asq.org

DLea[

weekly errors per bed is plotted against a seriesof lines representing the overall mean and stan-dard deviations (+/- 1, 2 and 3 sigmas).

The control chart data show the error reductionteam's goal of 50% reduction was surpassed. Auditsrevealed the percenfage of order entry errors consis-tently improved by 90% to less than 0.04 errors perbed every month for four months after the processchanges.

Power of InformationNurses now ask the team, "Why did we do all

that extra work?" when referring to the formermedication order entry process. The new process issimpler, works on all units and is more accurate.But the power of information is what will keep thestaff from digressing back to old habits.

When audits showed an increase in transcriptionerrors during the third week after implementationof the initiative, the Alton Memorial knew exactlywhat to do. The reaction plan devised by the teamwas put into place, a root cause analysis was done,the source of error was removed and the processwent back in control in no time.

As an Alton Memorial administrator said:"Finding out we had a problem even before weknew it and knowing exactly what to do to fix itwas priceless."

REFERENCES

1. Medication Errors & Patient Safety, www.ahrq.gov/

qualVerrorsix.htm.2. National Coordinating Council for Medication Error

Reporting and Prevention, www.mccmerp.org.3. Pugh Matrix, www.isixsigma.com/dictionary/glossary.

asp.

BIBLIOGRAPHY

1. BMG University, "Transactional Design for Six Sigma

Course Manual," BMG, 2004, sections two and seven,

www.bmgu.com.

2. Crow, Kenneth, "Performing QFD Step by Step,"

www.npd-solutions.com/qfdcons.htbnl, 2005.

3. Evans, James, and William Lindsay, The Management &

Control of Quality, sixth edition, 2005, pp 568-578.

4. Mazur, Glenn, Jeff Gibson and Bruce Harries, "QFD.

Applications in Health Care and Quality of Worklife,"

International Symposium on Quality Function Deployment,

Union of Japanese Scientists and Engineers, Tokyo, March

1995.

5. Yang, Kai, and Basem EI-Haik, Design For Six Sigma: A

Roadmapfor Product Development, McGraw Hill, 2003, pp:49-68 and 173-196.

YANIRA BENITEZ is business process leader and Six Sigma

Black Belt for BJC HealthCare in St. Louis. She earned a

master's degree in industrial engineering Pennsylvania

State University and is pursuing a doctorate at the

University of Missouri. Benitez is a member ofASQ.

LESLIE FORRESTER is a registered nurse and clinical

information specialist/educator for Alton Memorial

Hospital, Alton, IL. She has a bachelor's degree in nurs-

ing from McKendree College, Lebanon, IL.

CAROLYN HURST is a registered nurse and medication safe-

ty officerfor Alton Memorial Hospital. She has an associate'sdegree in nursing from Lewis & Clark Community College,Godfrey, IL.

DEBRA TURPIN is a registered nurse and lead clinical

information specialist at Alton Memorial Hospital. Shehas a bachelor's degree from McKendree College.

QUALITY PROGRESS I JANUARY 2007 45

COPYRIGHT INFORMATION

TITLE: Hospital Reduces Medication Errors Using DMAIC andQFD

SOURCE: Qual Prog 40 no1 Ja 2007WN: 0700101027017

The magazine publisher is the copyright holder of this article and itis reproduced with permission. Further reproduction of this article inviolation of the copyright is prohibited.

Copyright 1982-2007 The H.W. Wilson Company. All rights reserved.