i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s x x x ( 2 0 0 6 ) xxx–xxx

journa l homepage: www. int l .e lsev ierhea l th .com/ journa ls / i jmi

When requests become orders—A formative investigationinto the impact of a computerized physician order entrysystem on a pathology laboratory service

Andrew Georgioua,!, Johanna Westbrooka, Jeffrey Braithwaiteb, Rick Iedemab,Sangeeta Raya, Rowena Forsythb, Anthony Dimosc, Tony Germanosc

a Centre for Health Informatics, University of New South Wales, NSW 2052, Australiab Centre for Clinical Governance Research in Health, University of New South Wales, NSW 2052, Australiac Central Sydney Laboratory Services, Camperdown, RPAH, NSW 2050, Australia

a r t i c l e i n f o

Article history:Received 13 May 2005Received in revised form 22December 2005Accepted 19 April 2006

Keywords:PathologyEvaluation studiesLaboratoriesHospital information systemsQualitative research

a b s t r a c t

Purpose: The purpose of this study was to identify the key implications of the implementationof a computerized physician order entry (CPOE) system on pathology laboratory services.Methods: An in-depth qualitative study using observation, focus groups and interviews withpathology staff, managers, clinicians and information systems staff during implementationof a CPOE system in 2004 at a major Australian teaching hospital.Results: Pathology laboratories experienced a shift in their work roles resulting in alteredwork practices, responsibilities and procedures. These changes were marked by terminolog-ical and procedural changes in the test order process from when clinicians issued a requestfor a test, to the new system that established clinical orders at the point of care. This changewas accompanied by some organizational dysfunctions including the emergence of a newcategory of “frustrated” orders without specimens; problems with the procedure of addingtests to previously existing specimens; the appearance of discrepancies in the recorded timeof specimen collection. In response to these changes, hospital and pathology staff adopteda variety of means to cope with their changed circumstances. These ranged from effortsto increase clinical awareness to compensatory laboratory workarounds and enforced rulechanges.Conclusions: CPOE systems can have a major impact on the nature of the work of pathologylaboratories. Understanding how and why these changes occur can be enhanced throughconsidering the organizational and social contexts involved. The effectiveness of CPOE sys-tems will rely on how administrators and staff approach and deal with these challenges.

© 2006 Elsevier Ireland Ltd. All rights reserved.

1. Introduction

Computerized physician order entry (CPOE) systems automatethe clinical ordering process [1–3]. They incorporate features,such as decision support mechanisms, built-in alerts and

! Corresponding author. Tel.: +61 2 9385 9040; fax: +61 2 9385 9006.E-mail address: a.georgiou@unsw.edu.au (A. Georgiou).

rule-based prompts, which provide the potential to improvethe quality and safety of patient care [4–7]. The introductionof CPOE systems is widely perceived to be an importantbuilding block for establishing electronic medical records[8,9].

1386-5056/$ – see front matter © 2006 Elsevier Ireland Ltd. All rights reserved.doi:10.1016/j.ijmedinf.2006.04.002

IJB-2230; No. of Pages 9

2 i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s x x x ( 2 0 0 6 ) xxx–xxx

The introduction of CPOE systems is still in its infancy [10].In the USA less than 10% of inpatient settings have adoptedCPOE [11] and a growing number of studies of CPOE systemshave highlighted the significant costs involved and the majorchanges required in work processes and hospital departmen-tal interaction [10,12,13]. So far there has been relatively littleattention on the effect of CPOE on ancillary hospital depart-ments like pathology laboratories, which constitute a largecomponent of a hospital’s information repository [14]. Some ofthe early enthusiasm for CPOE has been tempered by reports ofimplementation problems (such as limited success in gettingdoctors to use the system), unexpected outcomes and fail-ures [8]. In many cases the issues involved are both complexand emotional [15], with some institutions reported to be sur-prised and unprepared for the major unforeseen challengesconfronted [16].

Studies employing qualitative methods incorporating awide range of theoretical approaches and methods havebecome more numerous in the recent past. In large part, thisdevelopment has been spurred by increased interest in thewhy and how questions relevant to understanding the unan-ticipated consequences of implementation. The purpose ofthis study was to identify the key implications and impact ofa CPOE system on pathology laboratory services. Qualitativemethods including observation, focus groups and interviewswith clinicians, pathology laboratory personnel and hospi-tal managers were undertaken to compare the situation inpathology laboratories during the implementation of the CPOEsystem at a large teaching hospital in Sydney.

1.1. Theoretical considerations

Clinical laboratories can be said to represent a scaled downversion of the automation requirements across the whole hos-pital. This is because they confront, on a smaller scale, many ofthe same data-handling questions faced in the hospital-widesetting [17,18]. Some early work in the area of Informationand Communications Technologies (ICT) evaluation focusedon ancillary hospital departments like pathology laboratoriesor radiology to explore how computer systems can be cata-lysts for new interactions between departments and health-care organizations [17,19–21]. These approaches were stronglymotivated by a concern about the lack of research into howcomputer systems affect the nature and definition of jobs andthe work process [17,22,23].

Aydin’s [22] work on organization adaptation to comput-erized medical information systems highlights the ability ofnew information systems to, on the one hand, threaten anoccupational community’s sovereignty in its work domain,and on the other hand improve its ability to conduct itswork effectively. In some cases this has brought issues likethe control of work processes to the forefront [20]. Barleydescribes a dyadic approach to the question looking both athow new technologies initially modify tasks, skills and othernon-relational aspects of roles, while also investigating the rela-tional components, such as who one interacts with, what theinteractions involve, and their affect on organizational andoccupational structure [23–25]. Changes in dyadic relationscan lead to structural changes in work roles, creating newdependencies or even requiring interactions among people

who did not previously relate. They can also alter levels ofsupervision, collaboration and other social and work relations.Within this framework, work and organization are bound indynamic tension and the introduction of new technologiesforces organizational structures to either adapt or risk becom-ing misaligned with the activities they organize [23]. Aydinand Barley provide valuable complementary frameworks toexplore the organizational implications of the introductionof CPOE systems for pathology laboratory services. This isparticularly relevant given the large investments into pathol-ogy CPOE systems planned in Australia, Britain and the USA[4,26,27].

2. Methods

This is an in-depth qualitative study using observation(including video), focus groups and interviews to provide richcontextual data. Three field researchers collected data. Theywere part of a larger team of researchers who contributed tothe discussion and analysis of the data. During each stageof the study process we incorporated feedback and regularinput from hospital pathology laboratory managers andscientists.

2.1. Setting

Our study site was the pathology laboratory at a major650-bed Sydney teaching hospital housing one of the state’slargest pathology test processing sites. The laboratoriesare divided into a number of sections dealing with speci-fied pathological processes. The hub of the department islocated within the Central Specimen Reception that handlesspecimens for general laboratory tests like biochemistry,haematology, immunology and endocrinology. It deals withsome 3700 test orders per day containing an average of 4.5discreet test assays per order. Other sections include theBlood Bank for pre-transfusion testing; Microbiology fortesting of infectious agents, such as bacteria, viruses, fungiand parasites; Anatomical Pathology for tissue diagnosis ofdisease.

In November 2003, the hospital began to implement theCerner Millennium PowerChart (Version 7.8) system thatallows clinicians to order pathology laboratory tests electron-ically. The initial implementation began in a small numberof wards and was progressively rolled out across all hospitalwards during 2004. The new CPOE system was linked to theexisting laboratory-based system—Cerner PathNet (Version7.8). Within 2 months the new system had led to a significantreduction in the average laboratory turnaround time from thetime of receipt of a specimen in the laboratory to availabilityof a test result [28]. Our study was carried out over the periodMarch to December 2004 during the peak implementationperiod of the new CPOE system. This time frame providedus with the opportunity to witness and document issues,challenges and changes that occurred in the course of theestablishment of the system. In this way, it gave us the scopeto undertake a formative evaluation [29–31] investigatingissues and themes as they arose. The research was approvedby the University of New South Wales Human Research

i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s x x x ( 2 0 0 6 ) xxx–xxx 3

Ethics Committee and the relevant Area Health Service EthicsReview Committee.

2.2. Data collection

Data collection for the study was based on an initial phaseof observations and interviews with laboratory-based person-nel, followed by a set of focus groups and semi-structuredinterviews with clinicians, laboratory staff and managers car-ried out between August and September 2004. The first phaseinvolved researchers interviewing laboratory personnel andobserving laboratory work processes using formal and infor-mal interview techniques. This approach provided our teamwith the iterative scope and latitude to pursue ideas, explorepotential issues and dimensions [32]. Recordings of observa-tions and interviews were collected in researcher logs withmemos that also noted reflections and documented changesin the investigation process [33]. This log formed an audittrail of the study while simultaneously helping to shape theresearch process. We chose informants using theoretical sam-pling techniques using observations and emerging themes toguide follow up choices and to help develop and challengeemerging hypotheses [34].

Our research log recorded over 40 observation and inter-view sessions within the laboratories. These included reportsof meetings, informal interviews and extensive observationsessions and videoing of laboratory work during and outside ofbusiness hours. They involved a range of personnel includingsenior laboratory managers, laboratory scientists, technicians,laboratory IT staff and receptionists. Most of our observationsrelated to the Central Specimen Reception area of the labora-tory but also included visits to other sections of the depart-ment including the Blood Bank, Anatomical Pathology andMicrobiology departments. Our observation work was impor-tant in helping us obtain contextually sensitive data on workpractices and relationships, beyond just identifying what peo-ple say they do [23,35,36].

The second phase of the study was based on taped focusgroups and interviews with doctors, nurses, pathology man-agement staff and IT personnel, carried out after an intensiveperiod of interaction and observation within the laboratoryhad reached saturation. In this way, the interviews and focusgroups were used to provide the study with a means of check-ing emerging themes and helping to test and improve the cred-ibility of findings [37]. The focus groups and interviews werebased on open-ended questions, which aimed to explore andclarify participants’ attitudes and experiences [38]. This phaseof the research included interviews with 11 physicians drawnfrom intensive care, emergency and orthopaedics wards. Onefocus group (four participants) and two further interviewsinvolved nurse unit managers from gastro-intestinal, trans-plant, plastics and intensive care wards. Other focus groupsincluded one centred on pathology management (six par-ticipants) and one with senior hospital information man-agement personnel (four participants). A further set of twointerviews was held with pathology staff and one with ahealth information manager. This phase of the work accumu-lated data from 19 different sessions that resulted in 9 h oftaped recordings transcribed by a person experienced in thetask.

2.3. Data analysis

Qualitative data analysis software NVivo 2.0 [39] was usedto assist our analysis. A grounded theory approach identi-fied emergent themes, using the informants’ words to guidethe assignment of codes. Themes were then reported anddiscussed in the course of four formal sessions with othermembers of the research team. Our analyses were also dis-cussed with, and checked by, a selection of key informants.This course of action was enhanced by our relationship withsenior pathology personnel who in the course of the studydeveloped into active collaborators and participants in the dis-covery and assessment of emerging analyses, while providinga valuable feedback mechanism to enhance the validity of ourfindings [40].

3. Results

3.1. Before and after CPOE implementation

Prior to CPOE the pathology test process began with a handwritten request (HWR) prepared by a clinician, which accom-panied the specimen delivered to pathology laboratories. Therequests and samples were delivered either by dedicated hos-pital staff, or through an air tube system if the tests were fromcritical care wards. Once a request was received within the lab-oratory, laboratory technicians checked that the request formcontained all relevant information including patient identi-fiers and the requesting doctor’s signature. They also con-firmed that the specimen was appropriate for the type andnumber of tests requested. Laboratory staff then entered theinformation into the pathology information system (CernerPathNet Version 7.8), which assigned a laboratory accessionnumber (sample identification) to the sample. From here thesample could be analysed and the results reported back to theclinician.

A laboratory manager explained the pre-implementationprocess in the following way:

“It is like when you look at a warehouse; you have a requestfrom a client and it comes on a purchase docket or whatevertheir piece of paper is, and the order is only placed when itis in [the] computer . . . it’s been ordered.”

With the introduction of CPOE the above entry process wasreplaced by a new set of procedures. Clinicians now place anelectronic test order within the CPOE system, which automat-ically generates a laboratory accession number. Cliniciansthen print a paper copy of the order, and it is forwarded alongwith the sample to the pathology laboratory. The laboratoryreceives this paper copy of the electronic test order togetherwith the sample and proceeds to match the order in thepathology laboratory system before printing a label andperforming the test. A laboratory scientist described it thus:

“Now it is a different process, because before you had arequest and you created the order in the lab and the orderwas created when you had all the samples. Now the orderis created and then the order is received in the lab, ready,pending for testing . . . only when we’ve got the samples.”

4 i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s x x x ( 2 0 0 6 ) xxx–xxx

Fig. 1 – Comparison of work flows before and after CPOE.

Fig. 1 provides a comparison of the before and after pro-cesses. It shows how the previous paper ordering process hasbeen replaced by a simpler, streamlined and more straight-forward approach that eliminates the need to transfer infor-mation from the handwritten form onto the laboratory-basedcomputer.

This change in the ordering process represented a sub-tle shift in responsibility and control away from the labora-tory to clinicians on the ward. Laboratory staff reported thatthe corrective role they once played, ensuring that all sam-ples were properly requested, signed and accounted for, hadbeen reduced. They nevertheless appreciated that the types ofrequest errors previously encountered (e.g. illegible handwrit-ing, missing signatures and incorrectly requested tests) hadnow been virtually eliminated.

When we asked clinicians about the new system they werepleased with the streamlining of the process and welcomedthe ability to check and verify the status of tests. As one doctornoted:

“Previously . . . if a form had been handwritten and it wentto the laboratory, you know, once it leaves you, you don’thave any record of where it is, who has got it or that it hasbeen done. So it’s good from that point of view.”

3.2. Frustrated orders

During the early months of the new system, the pathologylaboratories encountered an accumulation of unfulfilled testorders within the database. One senior laboratory managerdescribed these as “frustrated orders”. Those orders had beenentered into the computer system by clinicians, but were notassociated with a specimen taken from the patient. This canhappen when the request is forgotten or misplaced, or if thepatient had changed ward or been discharged prior to speci-men collection. Alternatively, the doctor may have decided towithdraw the order after realising that it was a duplicate.

In the previous HWR system, an order (accession num-ber) was only generated when a request and specimen werepresent in the laboratory. If a HWR arrived without a sample itwould be followed up, generally through a telephone call to therequesting clinician or ward to ensure, either that the requestwas no longer necessary, or that the sample was on its way.

For laboratory managers this procedure maintained databaseintegrity, which meant that they could account for all orderswithin the system.

The new CPOE system altered the above procedure by gen-erating an accession number as soon as a clinician enteredan order into the system. This meant that the new systemimmediately assumed that a sample was either present orsoon to be present. However, if the clinician decided not togo ahead with the order, for whatever reason, the order wouldremain on both the Pathnet and PowerChart databases unlesscancelled by the issuing clinician. The build up of these frus-trated or un-cancelled tests was a cause of growing confusionwithin pathology laboratories. Clinicians and laboratory staffsoon began to notice multiple instances of frustrated ordersappearing on their computer screens. A doctor described theproblems thus:

“In the old days, throwing out one of the [request] formsinto the bin was a pretty efficient solution, but now youhave this issue where the labs . . . have got these initiatedorders that haven’t been acted on and they’re not quite surewhether they’re real or they’re duplicates.”

One intensive care unit (ICU) doctor described the problemthis way:

“Doctors are notoriously lazy, will do things in the quickesttime possible and if something takes a long time to do thatis not necessarily beneficial or necessary for patient carethen we probably wouldn’t do it. Computer systems are thelast things on our mind when we’ve got sick people to lookafter.”

This issue had a particular resonance with managers of theBlood Bank, where any error matching a test order with a spec-imen would have severe consequences for patient care. BloodBank managers described their concern about the growinguntidiness in the database caused by the build up of frustratedorders. It meant that the system could show a patient with alist of unexplained and unfulfilled tests, some of them quitedated. Blood Bank managers were worried that this situationincreased the potential for error, i.e. the possibility of BloodBank staff matching a test specimen with an incorrect order.Initially the Blood Bank sent lists of frustrated orders to clini-

i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s x x x ( 2 0 0 6 ) xxx–xxx 5

cians to try and communicate the problem to them, but whenthings did not dramatically improve, pathology laboratoriesinstalled a procedure to check, and if necessary, cancel frus-trated orders after 3 days.

3.3. Add-on tests

An unforeseen issue that emerged from the CPOE systeminvolved add-on tests. An add-on test occurs when a clini-cian requires an additional test assay to be carried out on aspecimen that has already been delivered to the laboratory.Before CPOE implementation an add-on test was requested bya clinician by phone and a new handwritten request form wassigned and faxed by the clinician to the laboratories. Since anyone patient may have multiple specimens taken, this commu-nication with laboratories was important as it helped staff toidentify the correct specimen for an add-on and assign it thesame laboratory accession number as the initial test order.

The introduction of CPOE initially allowed clinicians toorder electronically an add-on test to an initial test order with-out consulting the laboratories. But in doing so the computersystem automatically assigned a new laboratory accessionnumber, which now appeared as a new test order. Pathologystaff explained that they were unable to distinguish betweenan add-on from a new test order. As described below thismeant the laboratories were forced to spend more time chas-ing up, or waiting for new specimen collections. Clinicians, onthe other hand, were making more calls to the laboratoriesinquiring about the delay in the order results.

Researcher: “What were the implications of this probleminitially?”

Laboratory scientist: “It caused big problems because wewere looking for specimens we didn’t have.”

Laboratory manager: “We didn’t know because the doctorwould send down the form with no blood, and we’d say, ‘OK,there’s no blood, let’s call the doctor. Where’s the blood?”

In the absence of an immediate technical solution to this prob-lem, pathology managers enforced a return to the before CPOEsystem of clinicians phoning and faxing signed requests foradd-on tests. This procedure was formally reinstated throughan official hospital-wide memo sent by the Pathology ClinicalDirector to all medical staff.

The problem was also one described by ordering clinicians,one of whom explained:

“There were some other problems with ordering extra testsafter the sample has gone down. We resorted to a backuppaper system order . . .. You think of an extra test and thesample has already gone and you have to write it on a bitof paper and then fax it down to them, so that’s a bit of aproblem. It would be easier if you could just do it on thecomputer like all the other ordering.”

3.4. Accurate collection times

Another consequence of the implementation of the newCPOE system was in situations when a discrepancy occurredbetween the time the system recorded that a specimen was

drawn from the patient, and the time it arrived within the lab-oratory. Laboratories pay a lot of attention to turnaround timeof tests, aiming to ensure that test results are returned withinsafe and efficient time frames. Turnaround times are thereforeimportant measures of quality performance because of theirpotential consequences for patient care [41]. Laboratory man-agers were concerned to discover that one of the unexpectedresults of the new system was the occurrence of significanttime discrepancies between when a specimen was drawn andwhen it arrived in the laboratory.

A doctor described the problem thus:

“So everyday in ICU people have to get a full range of bloodsand an X-ray done and if you’ve got 20 patients in the Unitthat whole process on the electronic ordering, can end uptaking you half an hour, just to click the buttons. And thetime that something is ordered, versus the time it is actu-ally taken, is often not cross-referenced, so if I order some-thing at 1.00 a.m. but actually [it is] taken at 3.00 a.m., itmight show up on my screen as being collected at 1.00 a.m.,which might not make a big difference, 2 h, but with sometests it is very relevant at actually what time the blood wastaken, as opposed to when it was ordered.”

The CPOE system asks clinicians to specify whether they:(1) intend to collect the specimen and (2) whether they intendto “Collect Now”. If clinicians answer “Yes” to “Collect Now”,the system automatically assigns the date and time at whichthe test order was generated as the collection date and time.But if the clinician was unable, or forgot to organize animmediate collection, the test specimens would be collectedat a later time, which sometimes resulted in a discrepancybetween the actual collection time and the time indicated onthe CPOE form. This meant that pathology laboratories wereforced to check regularly, and if need be, modify the actualcollection time and date in cases (e.g. over 30 min) of widevariation. Clinicians were alerted to the problem and askedto be more vigilant in order to minimize the number of testorders requiring modification.

4. Discussion

We considered the four major themes that emerged from ourstudy from three interrelated vantage points as depicted inTable 1: (1) the impact of CPOE implementation on lines ofresponsibility; (2) any unexpected changes or dysfunctionaloccurrences; (3) consequences and related changes.

4.1. Responsibility shifts

The experience of our study strongly reinforces Barley andAydin’s proposition that new IT systems have the potentialto shift patterns of responsibility, dependencies and in somecases, control, sometimes in unanticipated ways [22,23]. Wecan make an additional and complementary observation: thatinformation systems developed in one country are also likelyto incorporate assumptions about the nature of work whichcan cause problems when transferred [42]. Massaro [43], forinstance, notes that much of what professionals do is com-monly understood and enforced, but is not always the prod-

6 i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s x x x ( 2 0 0 6 ) xxx–xxx

Table 1 – Implications of major themes

Responsibility shifts Organizational dysfunctions Consequences

Request vs. order: Unclear lines of responsi-bility

Decreased laboratory scrutiny oftest orders

Laboratory regains some levels ofmonitoring and control

Frustrated orders: Clinicians responsibilityfor monitoring orders

Build up of unfulfilled ordersIncreased potential for error

Laboratory establishes protocol to cancelfrustrated orders after three days

Add-on tests: No laboratory scrutiny andcontrol

Unclarity within the laboratoryabout what is a new or add-onorder

Enforced return to pre-CPOE status quo

Accurate collection times: Clinicians need tomonitor distinction between order andcollection

Time discrepancies Laboratories monitor excessive times andchange them when needed

uct of a rigid guideline or protocol. Computer systems haveno such flexibility and do not make mutually understoodassumptions. In this case the request versus order issue tapsinto a much broader discussion about the role of laboratorymedicine itself, and whether its role is solely technical – pro-viding accurate and speedy analytical results, or medical –issuing the correct answer to a medical question. Plebani notesthat this conflict is reflected in the fact that British labo-ratories receive “requested” tests, while US clinicians preferto “order” tests [44]. This suggests that the new CPOE sys-tem may have inadvertently imposed the US style definitionof test ordering from its North American creators to Aus-tralian settings, more accustomed to British-style patterns.Such nuances in language are an interesting area for furtherstudy. They may reflect broader organisational and contextualdifferences, such as the inter-relationship between cliniciansand laboratory personnel [45]. As we saw in this study, termi-nology can also serve to reinforce or shift the balance of powerand control between departments.

4.2. Organizational dysfunctions

An important point emphasized by Barley and Kunda is thatthere is a dynamic tension between work and organisation[23] and new technologies can perturb the ‘negotiated order’of the hospital [46]. Shifts in work patterns and responsibil-ities can lead to dysfunctional work patterns accompaniedby task uncertainty and unplanned changes [19]. There werethree instances where the computerized ordering process dis-rupted previous operating norms—the occurrence of frus-trated orders; add-on tests; discrepancies between recordedand actual specimen collection times.

The appearance of these unanticipated problems indicatesthat the new work processes planned for the electronic sys-tem failed to take into account important aspects of previouswork practices, or to predict or compensate for the changes.This finding parallels results from a qualitative study usingfocus groups, participant observation and oral history tech-niques undertaken by Dykstra [16], which concluded thatCPOE systems can replace previous communication channelswith new and not clearly understood channels, no matterhow much planning is undertaken. As our examples show,the ordering process is not always linear and straightforwardas is often assumed by CPOE systems [47]. For instance, incomplex patient care situations it is not always correct toassume that a doctor’s decision to tick a box that says “Col-

lect Now” means that the specimen will necessarily occurimmediately.

4.3. Consequences

A clear advantage of undertaking a formative study of CPOEimplementation is the ability to examine how the pathologydepartment and hospital responded over time to the issuesthey confront. In the case of frustrated orders, we noticed thatthe Blood Bank initially sent lists of frustrated orders to clin-icians to try and increase their awareness of the problem. Inthe case of add-on tests, the Pathology Department formallyissued a memorandum restricting the ability of clinicians toorder add-on tests electronically. For the problem of accu-rate times, the laboratories established a procedure to changetimes when the discrepancy was in excess of 30 min. Thiswas important in maintaining the integrity of their monitoringprocesses. The hospital also undertook a training and aware-ness program among clinicians to alert them to the problemof incorrect specimen collection times.

One of the central points underscored by Aydin is theimportance of understanding how different groups negotiatetheir responses to the demands of computerization [22]. Bar-ley and Kunda’s contribution to this discussion emphasizesthe dynamic and behavioural components of work roles, nego-tiated and renegotiated in the flow of activity over time [23].Change in how departments deal with issues is to be expected.Previous studies have noted that problems confronted by CPOEcan be transitory and may decrease with greater experienceand knowledge [20–22]. Time is also a crucial element in useracceptability. It takes time to implement CPOE and it takestime for organizations to feel comfortable with it [48].

At our study site, it was possible to identify differentresponses to the way the laboratory managers and the hospitalresponded to the issues they faced. These are depicted in Fig. 2.They ranged from “soft” responses, such as bringing prob-lems to the attention of clinicians, or additional training andreminders to clinicians about collection times. Workaroundscan be described as ways of achieving things that the sys-tem does not let you readily achieve [49]. These occurred withthe decision to cancel frustrated orders after three days, orthe decision to modify times when there was a major dis-crepancy. Workarounds were a consequence of inadequatemechanisms for negotiation between departments. The thirdapproach can be termed a “hard” response, as in the laboratorydepartment’s formal refusal to accept add-on tests ordered via

i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s x x x ( 2 0 0 6 ) xxx–xxx 7

Fig. 2 – Categories of responses to areas of dysfunction and change.

CPOE. Each example represents a response to a shift in respon-sibility involving some level of negotiation among a range ofindividuals, groups and departments.

4.4. Limitations of the study

It is important to note the limitations of this study. It wascarried out at one Australian site during a formative periodof implementation. This had the advantage of providing richdata about how departments and individuals responded to thechanges and challenges of the new CPOE system. Its disad-vantage is it lacks the benefit of hindsight and overview oftenidentified with summative evaluation studies. Our study siteis located in a city location, containing a large and busy pathol-ogy department. The experiences described here may helpto illustrate and amplify issues that will be confronted else-where, but may be less relevant when applied to smaller sites.The study centred on observing and interviewing pathologylaboratory staff, but also included doctors, nurses and hospitalIT planners. We do not claim to have a broad cross section of allhospital perspectives but we have, as far as possible, sought toenhance the validity of our findings by comparing them acrossdifferent departments and professions. Despite limitations ofgeneralisability we believe that the findings outlined in thispaper are indicative of the type of issues confronted by hospi-tals undergoing CPOE implementation.

The purpose of the study was to identify the key implica-tions of the implementation of CPOE. While this enabled usto identify a number of shifts and changes in work respon-sibilities associated with some organizational dysfunctions,it limited our ability to comment comprehensively on whythings may have happened the way they did.

5. Conclusion

This study shows that large and complex computer systemslike CPOE have a major impact on the nature of work andhow it is defined. They can also affect levels of supervision,departmental and social relationships. In this case the seem-ingly subtle distinction between a request and an order withinthe laboratory test process had important repercussions onhow aspects of work were carried out. It also threatened theboundaries between different occupational communities andcontributed to dysfunctions and altered work practices andrelationships. Several researchers have emphasized the ability

of CPOE systems to replace previous communication channelswith new, often poorly understood channels, in some casescreating ambiguity and uncertainty about order information[21,43,50]. Recent trends expressed within international infor-matics literature point to the potential value of numerous elec-tronic tools including mobile computing devices, such as Per-sonal Digital Assistants (PDAs) [51], and silicon chip technol-ogy [52] to better manage labelling and time stamping proce-dures within hospitals, possibly overcoming the dysfunctionsreported in this study. However, the effectiveness of CPOE willcontinue to rely heavily on how planners approach and han-dle changes in work practices and departmental relationships.This points strongly to the need for clear processes throughwhich different groups can re-negotiate work practices inorder to avoid unsatisfactory solutions, such as workarounds.

Understanding how and why this occurs requires goingbeyond simple measures of performance and acceptability.How a new CPOE system functions within its specified settingdoes not just depend on its design and how it is implemented,it needs to be understood within an organizational and socialcontext [53]. Studying this complexity requires a plurality ofmethods including interviews and focus groups that exam-ine how people understand the issues, alongside observationtechniques investigating what people actually do and howthey do it [23].

Lastly, we note that studies of CPOE systems can beenhanced by the adoption of a formative approach to researchand evaluation. This gave us the means to track the unfold-ing processes at the centre of this study and provided uswith insights into the issues and problems as they were beingencountered. It also laid the basis for interactions with pathol-ogy managers and staff, providing us with valuable feedbackabout emerging themes and ideas.

Acknowledgements

This study is part of an Australian Research Council Link-age Grant funded project to evaluate the impact of informa-tion and communication technologies on organizational pro-cesses and outcomes using a multidisciplinary, multi-methodapproach. The partners in the study are the Centre for HealthInformatics and Centre for Clinical Governance Research inHealth from the University of New South Wales and the NewSouth Wales Health Department.

8 i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s x x x ( 2 0 0 6 ) xxx–xxx

What was known before this study?• Computerized physician order entry (CPOE) system

implementation involves a myriad of complex decision-making processes.

• Qualitative studies of CPOE have helped to highlightsome of the previously unforeseen problems involved inimplementation.

• Pathology laboratories have to deal with many of themajor data-handling issues confronted by hospital-wideinformation systems, albeit on a smaller but more inten-sive scale.

What this study has added to the body of knowledge?• CPOE system implementation has important terminolog-

ical and procedural implications for the process of testordering within the hospital.

• CPOE systems have the potential to create levels of organi-zational dysfunction leading to problems not previouslyencountered in the test ordering process.

• Pathology laboratory and hospital management oftenhave to embrace a variety of means to cope with thechanged circumstances brought about by CPOE.

• Clear re-negotiation processes between groups need to bein place to deal with problems and issues encountered inCPOE implementation.

r e f e r e n c e s

[1] G.J. Kuperman, R.F. Gibson, Computer physician orderentry: benefits, costs, and issues, Ann. Intern. Med. 139(2003) 31–39.

[2] National Electronic Decision Support Taskforce, ElectronicDecision Support for Australia’s Health Sector,Commonwealth of Australia, Canberra, 2003.

[3] D.F. Doolan, D.W. Bates, Computerized physician orderentry systems in hospitals: mandates and incentives,Health Aff. (Millwood) 21 (2002) 180–188.

[4] First Consulting Group, Computerized Physician OrderEntry: Costs, Benefits and Challenges, Available at:www.fcg.com/healthcare/ps-cpoe-research, AccessedMarch 2004.

[5] The Leapfrog Group for Patient Safety, Fact Sheet:Computerized Physician Order Entry (CPOE), Available at:www.leapfroggroup.org/Factsheet.pdf, Accessed May 2004.

[6] W. Tierney, J. Overhage, B. Takesue, L. Harris, M. Murray, D.Vargo, C. McDonald, Computerizing guidelines to improvecare and patient outcomes: the example of heart failure, J.Am. Med. Inform. Assoc. 2 (1995) 316–322.

[7] D.L. Hunt, R.B. Haynes, S.E. Hanna, K. Smith, Effects ofcomputer-based clinical decision support systems onphysician performance and patient outcomes: a systematicreview, J. Am. Med. Inform. Assoc. 280 (1998) 1339–1346.

[8] J.S. Ash, D.W. Bates, Factors and forces affecting EHRsystem adoption: report of a 2004 ACMI discussion, J. Am.Med. Inform. Assoc. 12 (2005) 8–12.

[9] J.I. Hwang, H.A. Park, S. Bakken, Impact of a physician’sorder entry (POE) system on physicians’ ordering patternsand patient length of stay, Int. J. Med. Inf. 65 (2002)213–223.

[10] D. Stablein, E. Welebob, E. Johnson, J. Metzger, R. Burgess,D.C. Classen, Understanding hospital readiness for

computerized physician order entry, Jt. Comm. J. Qual. Saf.29 (2003) 336–344.

[11] J.S. Ash, P.N. Gorman, V. Seshaddri, W.R. Hersh,Computerized physician order entry in U.S. hospitals:results of a 2002 Survey, J. Am. Med. Inform. Assoc. 11(2004) 95–99.

[12] Institute for Clinical Systems Improvement, TechnologyAssessment—Computerized Physician Order Entry,Available at: www.icsi.org, Accessed May 2004.

[13] C.M. Birkmeyer, J. Lee, D.W. Bates, J.D. Birkmeyer, Willelectronic order entry reduce health care costs? Eff. Clin.Pract. 5 (2002) 67–74.

[14] M.J. Becich, Information management: moving from testresults to clinical information, Clin. Leadersh. Manage.Rev. 14 (2000) 296–300.

[15] J.S. Ash, P.N. Gorman, M. Lavelle, J. Lyman, Multipleperspectives on physician order entry, AMIA Annu. Symp.Proc. (2000) 27–31.

[16] R. Dykstra, Computerized physician order entry andcommunication: reciprocal impacts, AMIA Annu. Symp.Proc. (2002) 230–234.

[17] B. Kaplan, Initial impact of a clinical laboratory computersystem. Themes common to expectations and actualities,J. Med. Syst. 11 (1987) 137–147.

[18] T.L. Lincoln, R.A. Korpman, Computers, health care, andmedical information science, Science 210 (1980) 257–263.

[19] C.E. Aydin, R.E. Rice, Bringing social worlds together:computers as catalysts for new interactions in health careorganizations, J. Health Soc. Behav. 33 (1992) 168–185.

[20] C.E. Aydin, Computerized order entry in a large medicalcenter, in: J.G. Anderson, C.E. Aydin, S.J. Jay (Eds.),Evaluating Health Care Information Systems, Sage,Thousand Oaks, 1994, pp. 260–275.

[21] E.J. Davidson, W. Chismar, Examining the organizationalimplications of IT use in hospital-based health care: acase study of computerized order entry, in: Proceedings ofthe 32nd Hawaii International Conference on SystemSciences, Hawaii, 1999.

[22] C.E. Aydin, Occupational adaptation to computerizedmedical information systems, J. Health Soc. Behav. 30(1989) 163–179.

[23] S.R. Barley, G. Kunda, Bringing work back in, Org. Sci. 12(2001) 76–95.

[24] S.R. Barley, The alignment of technology and structurethrough roles and networks, Adm. Sci. Q. 35 (1990) 61–103.

[25] S.R. Barley, Technology as an occasion for structuring:evidence from observations of CT scanners and the socialorder of radiology departments, Adm. Sci. Q. 31 (1986)78–108.

[26] M. Humber, National programme for informationtechnology, Br. Med. J. 328 (2004) 1145–1146.

[27] Victorian Government Department of Human Resources,Health Smart—Strategy for the modernisation andreplacement of information technology, Available at:www.health.vic.gov.au/healthsmart/documents/healthsmart-strategy.pdf, Accessed May2004.

[28] J. Westbrook, A. Georgiou, A. Dimos, T. Germanos,Computerised pathology test order-entry reduceslaboratory turnaround times and influences test orderedby hospital clinicians: a controlled before and after study,J. Clin. Pathol. (published online 3 Feb 2006),doi:10.1136/jcp.2005.029983.

[29] J. Westbrook, A. Gosling, The Impact of Point of CareClinical Systems on Health Care: A Review of the Evidenceand a Framework for Evaluation, Centre for HealthInformatics, University of New South Wales, Sydney,2002.

i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s x x x ( 2 0 0 6 ) xxx–xxx 9

[30] A.P. Stoop, M. Berg, Integrating quantitative and qualitativemethods in patient care information system evaluation:guidance for the organizational decision maker, MethodsInf. Med. 42 (2003) 458–462.

[31] C.P. Friedman, J.C. Wyatt, Evaluation Methods in MedicalInformatics, Springer-Verlag New York Inc., New York,1997.

[32] E. Murphy, R. Dingwall, D. Greatbatch, S. Parker, P. Watson,Qualitative research methods in health technologyassessment: a review of the literature, Health Technol.Assess. 2 (1998) 1–274.

[33] S. Gifford, Analysis of non-numerical research, in: C. Kerr,R. Taylor, G. Heard (Eds.), Handbook of Public HealthMethods, McGraw Hill, Australia, Sydney, 1998, pp.543–554.

[34] B. Glaser, A. Strauss, The Discovery of Grounded Theory:Strategies for Qualitative Research, Aldine PublishingCompany, New York, 1967.

[35] J. Braithwaite, M. Westbrook, R. Iedema, N. Mallock, R.Forsyth, K. Zhang, A tale of two hospitals: assessingcultural landscapes and complexities, Soc. Sci. Med. 60(2005) 1149–1162.

[36] R. Iedema, P. Degeling, J. Braithwaite, L. White, It’s aninteresting conversation I’m hearing: the doctor asmanager, Org. Stud. 25 (2004) 15–34.

[37] R. Dingwall, Accounts, interviews and observations, in: G.Miller, R. Dingwall (Eds.), Context and Method inQualitative Research, Sage Publications, London, 1997, pp.51–65.

[38] J. Kitzinger, Focus groups with users and providers ofhealth care, in: C. Pope, N. Mays (Eds.), QualitativeResearch in Health Care, BMJ Books, London, 2000, pp.20–29.

[39] P. Bazeley, L. Richards, The NVivo Qualitative Project Book,Sage, Oxford, 2000.

[40] E. Guba, Y. Lincoln, Fourth Generation Evaluation, Sage,Newbury Park, CA, 1989.

[41] P. Manor, Turnaround times in the laboratory: a review ofthe literature, Clin. Lab. Sci. 12 (1999) 85–89.

[42] R. Heeks, S. Bhatnagar, Understanding success and failurein information age reform, in: R. Heeks (Ed.), ReinventingGovernment in the Information Age—InternationalPractice in IT-enabled Public Sector Reform, Routledge,London, 1999, pp. 49–74.

[43] T.A. Massaro, Introducing physician order entry at a majoracademic medical center: I. Impact on organizationalculture and behavior, Acad. Med. 68 (1993) 20–25.

[44] M. Plebani, The clinical importance of laboratoryreasoning, Clin. Chim. Acta 280 (1999) 35–45.

[45] P.N. Gorman, M.B. Lavelle, J.S. Ash, Order creation andcommunication in healthcare, Methods Inf. Med. 42 (2003)376–384.

[46] A. Strauss, L. Schatzman, D. Ehrlich, R. Bucher, M.Sabshin, The hospital and its negotiated order, in: E.Friedson (Ed.), The Hospital in Modern Society, Free Press,New York, 1961, pp. 147–169.

[47] J.S. Ash, P.Z. Stavri, R. Dykstra, L. Fournier, Implementingcomputerized physician order entry: the importance ofspecial people, Int. J. Med. Inf. 69 (2003) 235–250.

[48] J.S. Ash, J. Lyman, J. Carpenter, L. Fournier, A diffusion ofinnovations model of physician order entry, AMIA Annu.Symp. Proc. (2001) 22–26.

[49] J.S. Ash, P.N. Gorman, M. Lavelle, T.H. Payne, T.A. Massaro,G.L. Frantz, J.A. Lyman, A cross-site qualitative study ofphysician order entry, J. Am. Med. Inform. Assoc. 10 (2003)188–200.

[50] B. Cheng, M. Golstein, E. Geller, R. Levitt, The effects ofCPOE on ICU workflow: an observational study, AMIAAnnu. Symp. Proc. (2003) 150–154.

[51] Y.-C. Lu, Y. Xiao, A. Sears, J.A. Jacko, A review and aframework of handheld computer adoption in healthcare,Int. J. Med. Inf. 74 (2005) 409–422.

[52] M. Collen, The origins of informatics, J. Am. Med. Inform.Assoc. 1 (1994) 91–107.

[53] B. Kaplan, Organizational evaluation of medicalinformation resources, in: C.P. friedman, J.C. Wyatt (Eds.),Evaluation Methods in Medical Informatics,Springer-Verlag, New York, 1997, pp. 255–280.