STUDY OF CLINICAL WORKFLOW AND INFORMATION

FLOW OF A BREAST CARE UNIT

Vítor Silva

, Maria João Cardoso

, João Fonseca

and Ricardo Cruz-Correia

Magnetic Resonance Department, Hospital São João, Oporto, Portugal

Center for Research in Health Technologies and Information Systems (CINTESIS), Portugal

Breast Center, Hospital São João, Oporto, Portugal

INESC, Breast Research Group, Portugal

Faculdade de Medicina da Universidade do Porto (FMUP), Oporto, Portugal

Center for Research in Health Technologies and Information Systems (CINTESIS), Portugal

Faculdade de Medicina da Universidade do Porto (FMUP), Oporto, Portugal

Keywords: Clinical protocols, Information flows, Processes, Breast diseases, Hospital communication systems.

Abstract: Introduction. The study of the clinical workflows and information flows in healthcare institutions is of vital

importance to improve its effectiveness and efficiency. At Hospital São João, Oporto – Portugal, a

Diagnostic Breast Unit (DBU) was recently created. The implementation of a new Electronic Patient Record

(EPR) called Breast.Care triggered the need to better understand the DBU processes and suggest

improvements. Aim. To describe clinical workflows and information flows in DBU, detect problems and

propose solutions for better communication among different actors. Methods. The study started with a direct

observation period with a total of 24 hours. The observed processes and flows were transcribed into free text

and then into structured text, tables and Unified Modeling Language (UML) diagrams (activity and

partition-activity). The structured text and diagrams were analysed to find possible improvements. Results.

Seven main processes were identified representing how different actors (humans and computers) work

together. Three communication process improvements between humans were detected (e.g. changing timing

of patient data insertion to facilitate reading access to others), three human–machine improvements (e.g.

changing computer medical forms) and one between machines (creating specific links between information

systems). Discussion. Analysing workflow and information flow in DBU allowed the detection of

communication problems and the improvement of those through changes in EPR and in DBU current

processes.

1 INTRODUCTION

The Diagnostic Breast Unit (DBU), a part of the

Hospital Breast Center was created to improve

health care delivery to patients with breast problems

at Hospital São João (HSJ). A team of experts on

breast diseases work together in the Unit optimizing

the diagnosis of breast problems. The Unit provides

an integrated monitoring of patients, creates greater

comfort and prevents increase of hospital patient’s

visits, greatly reducing the waiting time to diagnosis.

This unit aims at being credited by the European

Society of Mastology (EUSOMA) and therefore

meets all the requirements needed. As a new unit

within the community hospital and having their own

team of experts, problems and difficulties on

communication and information flow patterns with

the “outside world”, including other hospital

departments, are to be expected.

Managing the diversity of medical data and its

communication in DBU, is not an easy task, due to

the existence of many heterogeneous processes.

Understanding those processes, workflows and its

information flows is of vital importance to improve

its effectiveness and efficiency. Currently, most of

clinical data are handled by software applications,

which are known to become the engine of data

management (Keizer 2005) (Oroviogoicoechea

2007).

Often the terms process and workflow are used

as synonyms or side by side without any distinctive

differences (Knolmayer 2000) (Reijers 2003). In

294

Silva V., João Cardoso M., Fonseca J. and Cruz-Correia R. (2010).

STUDY OF CLINICAL WORKFLOW AND INFORMATION FLOW OF A BREAST CARE UNIT.

In Proceedings of the Third International Conference on Health Informatics, pages 294-301

DOI: 10.5220/0002744502940301

 SciTePress

this paper, the following definitions will be used: a)

Process - a way for an enterprise to organize work

and resources (people, equipment, information,

computer applications, etc.) to accomplish its aims.

Nowadays, organizations try to arrange work and

resources to achieve a specific output – a result – for

a specific costumer. It can be simply defined as a

collection of interrelated work tasks, initiated in

response to an event, that achieve a specific result

for the process’ costumer (Sharp 2001); b)

Workflow

- simply refers to tasks, resources and

triggers associated with a specific process. It is

dependent on the process and contains the sequence

of tasks and information about data and facilities the

execution of functions (Becker 1999); c)

Information flow

- can be defined as a serie of

successive events, between the generation of

information from a source of information and

acceptance by a receiver. It is designed to promote

some kind of change (Silva 2008). With the advance

of information technologies, information flows tend

to be a multifocal and multioriented. All

organizations need systematic and consistent efforts

to influence and monitor all their information flow

by promoting strategies for that purpose (Krovi

2003). The retrieve of information flow is essential

to recognize its efficiency (Jacoski 2005).

Numerous authors have discussed the theoretical

benefits of a well established clinical workflow

(Duitshof 1995; Holland 2006; Becker 2007;

Malhotra 2007). It leads to detailed, organized,

formalized, communicated, managed and

streamlined processes. The workflow becomes clear

and potentiates process efficiency, better

organization, improved quality and better

identification and characterization of possible

medical errors (Becker 2007; Malhotra 2007; Peleg

2007).

In healthcare today we are faced with the

problem of departmental applications that support

specific functions and limited workflow. Most of the

data captured by these systems reside within the

application and are not easily retrieved and made

available to other processes that may require them

(Berg 1999; Nielsen 2000; Lorenzi 2003).

Delivering healthcare today is a complex task

often undertaken by multidisciplinary teams, whit

the obvious output of a great amount of information.

Additionally organizations have complex

infrastructures with poor communication between

different departments and staff. To optimize

operational efficiencies they require means to

automate and manage these processes making sure

that the right information is made available to the

right individual at the right time. An established

workflow can support such care processes by

executing agreed models of care (Berg 1999;

Nielsen 2000; Lorenzi 2003).

For information management and

communication with other healthcare institutions

(e.g. hospitals or primary care) and other hospital

departments (e.g. radiology or pathology), DBU

uses the following computer applications: a)

SONHO – the ADT (Admission-Transfer-

Discharge) system of the hospital, which

manipulates administrative data (e.g. demographic

patients’ data); b) SAM (Portuguese initials for

Medical Support System) - the outpatient medical

scheduling system and multipurpose hospital

patient; c) Siima (Portuguese initials for Clinical

Imaging Management System) - the imaging system

responsible for the record and visualizing of

mammography and ultrasound exams at DBU; d)

Breast.Care - a specific departmental electronic

patient record built for the DBU, to record and

analyse clinical data from patients with breast

problems; and e) Alert P1 – a system that receives

requests for consultation from primary care

physicians (this system is not available on DBU).

All described applications help in the integration

of outside and inside information contributing to

improve its efficiency and effectiveness

(Ammenwerth 2004; Keizer 2005).

2 AIM

To describe and analyse clinical workflows and

information flows in DBU, detect problems in the

system and propose solutions for better

communication among different actors, thereby

trying to improve performance.

3 METHODS

Study Design. This study was undertaken in

different days totalising a 24 hours period of

participant, direct and semi-structured observation

during January 2009.

Study Participants. Participants were all potencial

users of computer applications existing in the DBU.

Data Method Collection. The main data collection

method used was direct observation. This method

allowed a thorough analysis of clinical workflows

and information flows between different actors.

STUDY OF CLINICAL WORKFLOW AND INFORMATION FLOW OF A BREAST CARE UNIT

295

During a 24 hours period, one researcher (the

first author of this paper) observed interactions,

information flow, clinical workflow and processes

between different actors, in different offices

(consultation rooms, waiting rooms, main reception

areas, and technical work areas). Information was

gathered while shadowing study participants. During

this process, open-ended questions were asked to

develop insights about how events happened,

experienced and reported. In this phase extensive

field notes were collected.

Variables Description. Free-text was used to

extract the following variables: i) local of

observation; ii) actors observed; iii) stage of clinical

workflow; iv) software applications used.

Data Analysis. Data were collected through

observation method with field notes. The gathered

notes were then passed into structured text to allow

better analysis and comprehension. That structured

text described and analysed the different clinical

workflows and information flows in DBU. UML

diagrams, like activity and partition activity

diagrams, and tables were used and constructed.

4 RESULTS

We identified various types of clinical processes

(n=7) with different actors and corresponding

workflow (Table 1). Data about, where, how and

when they occurred was registered. Multiple actors

were present in all processes.

UML diagrams were used to analyse and

illustrate all processes. As an example, Figure 1

represents an activity diagram of Process I described

on Table 1. Table 2 shows the different types of

information flows that exist in the DBU. To clarify

some existing information flow described above, we

used UML partition activity diagrams. The first

partition diagram (see Figure 2) describes when the

surgeon goes directly to the radiologist to know his

opinion about a patient’s case (communication

between humans). The second diagram (see Figure

3) is related to radiologist performing examination

reports in Breast.Care application and copying and

pasting them to SIIMA application (communication

between humans and computers). After analysing

the processes, we have detected several possible

improvements to information flows. Improvements

were grouped together based on the type of actors

involved (Table 3).

Figure 1: UML activity diagram of Process I (Table 1) -

Patients (first consultation), do not perform invasive

exams and are discharged from the Unit in the same day.

Figure 1 is an activity UML diagram that is

recognized as the most suitable for modeling the

functional vision of a system, because describes the

logic of procedures or duties(Silva 2005).

These two tables show the most important clinical

workflows and information flows that occur in

DBU. Table 1 has seven processes with all

description, the different actors involved and how,

where and when they interact. Table 2 shows some

information flows associated to clinical workflows

and processes described on table 1, describing all

information flow and the sequence of actors.

Figures 2 and 3 show two UML partition activity

diagrams, which include division lines (called swim

lanes), that describe in greater detail and specificity

certain classes or subsystems, in this case actors

during information flows.

HEALTHINF 2010 - International Conference on Health Informatics

296

Table 1: Description of the different processes in DBU.

N Description Actors How Where When

I Patients

(first

consultation)

do not

perform

invasive

exams and

are

discharged

from DBU

same visit.

Patient,

Administra

tive staff,

Surgeon,

Radiologist

Radiograph

Patient arrives at DBU, gives all documents to administrative

staff, who confirms the patient’s arrival on SONHO application.

Reception Patient’s

arrival

Surgeon, through SAM application, confirms arrival and calls

patient to consultation room. Next he starts Breast.Care

application to introduce demographic and clinical data. Surgeon

performs clinical breast exam on patient.

Consultation

Room

Surgeon’s

interview

and exam

Surgeon observation of patients previous imaging exams.

Discussion and observation of exams with Unit Radiologist.

Lecture

Room

After

consultation

Radiologist could have two opinions: i) the patient doesn’t need to

perform any exam and is discharged from consultation; ii) patient

needs to perform some extra non-invasive diagnostic exams,

which will be reported through Breast.Care application.

Imaging

Rooms

At imaging

Patient is informed by surgeon of normal or benign results.

Discharge and reference to attending physician.

Consultation

Room

End of

consultation

Patients

perform

invasive

studies

(breast

biopsy) and

have a

subsequent

visit for

results.

Patient,

Administra

tive staff,

Surgeon,

Radiologist

Radiograph

er,

Pathologist

Similar to first 3 steps of process I. Reception,

Consultation

Room

Patient’s

arrival, at

consultation

A guided breast biopsy (imaging rooms) and a histological exam

will be undertaken (pathology department).

Imaging

Rooms

Pathology

Department

At imaging

At pathology

after

specimen

prepared

Patient informed of another visit date for results by the

administrative staff.

Consultation

Room

consultation

Patients

come to be

informed of

biopsy

result.

Surgeon,

Patient

The biopsy’s result can be benign (with or without surgery) or

malignant (surgery or primary non surgical treatments). If results

are benign, patients may be discharged from consultation and be

followed by attending physician. In some cases, surgery is advised

or desired.

Consultation

Room

consultation

Patients

scheduled

for surgical

intervention.

Surgeon,

Patient

All these patients must have a surgery scheduled in SAM

application (national waiting list). In some cases where primary

treatment is decided before surgery patient must be referred to

oncologist. Surgery is schedulled subsequentially (7 months later).

Consultation

Room

consultation

V Patients who

are referred

for a

Genetics

consultation.

Surgeon,

Patient,

Nurse in

charge of

genetics

consultatio

Patients, who have high risk for breast cancer (e.g. family

history), are referred to Genetics consultation in DBU, which

currently only occurs on Fridays. Surgeon refers those patients

directly to nurse in charge of genetics appointments.

Consultation

Room

DBU

consultation

Patients who

will be

followed in

DBU

(patients

with breast

cancer that

have already

been treated)

Surgeon,

Patient,

Administra

tive staff

Patient arrives to DBU, gives and the administrative confirms the

patient’s arrival on SONHO application.

Reception Patient’s

arrival

Surgeon, through the SAM application, sees that the patient has

arrived and calls him into consultation room. A clinical

questionnaire and physical examination completes patient’s

clinical data into the Breast.Care application. Patients are referred

for an outpatient consultation in HSJ. The surgeon usually uses

SAM application to record and print drugs prescription.

Consultation

Room

consultation

Diagnosis

Multidiscipli

nary Team

Meeting

Surgeon,

Radiologist

Pathologist

All cases of patients who were submitted to biopsy are discussed

in a diagnosis multidisciplinary team meeting.

Lecture

Room

Periodically,

proportional

to number of

biopsies

STUDY OF CLINICAL WORKFLOW AND INFORMATION FLOW OF A BREAST CARE UNIT

297

Table 2: Different types of information flow and actors that exist in DBU.

Information Flow Actors

Surgeon searchs and requests, directly and personally, radiologist

opinion about patient case. Radiologist reads patient’s exams and

informs surgeon. Surgeon informs patient (to perform or not to

further exams).

Surgeon → Radiologist → Surgeon → Patient

Patient is referred to genetics consultation. Surgeon refers patient

directly to nurse in charge of those consultations, who makes an

appointment. Patient will receive notification at home.

Surgeon → Patient → Nurse → Patient

Patients need to perform a breast MRI (Magnetic Ressonance

Imaging) to better characterize or stage the lesion. Surgeon informs

patient of the decision and requests exam at SAM application. MRI

unit receives that request, appoints it and notifies patient.

Surgeon → Patient → MRI Unit → Patient

Patient has benign breast disease and no need for further

intervention she will be referred to her attending physician.

Surgeon informs patient of decision and prints a report that will be

delivered by hand to attending physician.

Surgeon → Patient → Attending Physician

Majority of requests for first appointment in DBU are requested by

Alert P1 (system is not available directly in DBU). Surgeon in

charge prints breast referrals from general surgery department and

brings them to DBU. A classification of urgent, medium and non-

urgent priority is done based on attending physician information.

Surgeon → Surgery Unit → Surgeon

(Requests)

Radiographer knows of patient arrival, when personally warned by

administrative staff.

Patient → Administrative → Radiographer

During consultation, surgeon fills data relative to that episode in

Breast.Care application. It will create a PDF document with this

data in SAM application.

Surgeon → Breast.Care Application → SAM

Application

Radiologist has to copy and paste imaging reports in Breast.Care

application to SIIMA application.

Radiologist → Breast.Care Application → SIIMA

Application

Administrative staff takes patient information on printed sheets to

surgeon.

Administrative (information in paper) → Surgeon

Figure 2: Partition Activity Diagram of an information flow in which the surgeon goes personally to know the radiologist’s

opinion about a patient.

HEALTHINF 2010 - International Conference on Health Informatics

298



Figure 3: Partition Activity Diagram of an information flow in which the radiologist fills the report in Breast.Care

application and has to copy and paste that report to another computer application (SIIMA).

Table 3: Suggested Improvements on communication processes.

Current Suggested Improvement

Communication

between humans

Normally information related to patient is

printed to attending physician.

Deliver all patient information in

electronic format through one of the

currently available systems.

Radiographer only knows of patient

arrival to DBU when administrative staff

warns him personally.

Administrative staff could insert a

patient’s arrival hour at DBU, in SIIMA

application.

Surgeon has to search and ask, personally

radiologist’s opinion about a patient’s

case.

Surgeon’s clinical evaluation is reported

in Breast.Care, Radiologist could access

all abstract and patient’s exam over the

same application. Radiologist would write

his report back in Breast.Care.

Communication

between human-

machine

Alert P1 system is not available in DBU. Information on Alert P1 system should be

available to DBU users. This would avoid

going to another department and also

avoid printing requests.

Surgeon has to print reports to patient.

He has to go to reception room to get

those reports.

Some printers could be available in some

key-places, like in consultation rooms.

Printed reports could be mailed in

envelopes by administrative staff to

handle to patients on their way out.

Surgeons filling Breast.Care forms using

free text components.

Implement sctrutured forms that force a

logical sequential order (according to

consultation progress).

This modification in the software was

suggested to Breast.Care development

team and it is already implemented in

DBU.

Communication

between machines

(Integration of IS)

Radiologist has to copy and paste all

examination reports from Breast.Care to

SIIMA application.

A link should be available in Breast.Care

application connecting it to SIIMA

application.

STUDY OF CLINICAL WORKFLOW AND INFORMATION FLOW OF A BREAST CARE UNIT

299

5 DISCUSSION

This method allowed detailed description of

workflows and information flows, and also the

detection of communication problems, suggesting

some improvements. We feel that this methods

when properly applyed provide useful results for

improvement of information flow. With good

observational methods, we can also identify

appropriate and inappropriate patterns of

communication (Caris-Verhallen 2004; Unertl

K.M. 2006; Ash 2007). Through this method, we

can analyse various information flows in a health

unit, in order to develop general models of

workflow for the improvement of process

automation (Unertl K.M. 2006).

This study identified many processes involved

in the unit that showed the high diversity and

quantity of information that staff and computer

applications deal with. This fact reinforces the

need for well established clinical workflows and

information flows, in order to provide better

healthcare.

With suggested improvements, costs could be

reduced (e.g. reducing the need to print in paper)

and also additional time spared (e.g. the time

needed for the radiologist to fill the examination

reports in SIIMA and Breast.care software

applications). Simple information flows would

allow also avoid unnecessary participant travels

between rooms.

With observation method, authors were able to

see how doctors filled patients’ data into

Breast.Care application and thereby suggest some

changes. Based on those suggestions, Breast.Care

development team created a sequential method to

enter patients’ data in Breast.Care. This change

affected positively DBU doctors’ team, by

fastening data entering into the application.

Future research will use other methods of data

collection, like interviews and focus groups and

computer systems log analysis, aiming at enriching

the quality and quantity of data, and enable a

process of triangulation when analysis the

processes of the DBU (Bottorf 1993; Hewinson

1995; Ash 2007; Lessard-Hébert 2008).

ACKNOWLEDGEMENTS

We are grateful to all DBU staff for all the help

and availability provided.

REFERENCES

Ammenwerth, E., Brender J., Nykanen P., Prokosch HU,

Rigby M., Talmon J. (2004). "Visions and strategies to

improve evaluation of health information systems.

Reflections and lessons based on the HIS-EVAL

workshop in Innsbruck." International Journal of

Medical Informatics 73: 479-491.0

Ash, J., Guappone, K. (2007). "Qualitative evaluation of

health information exchange efforts." Journal of

Biomedical Informatics 40: S33-S39.

Becker, J., Jamiesch, C. (2007). Resctrictions in Process

Design: A Case Study on Workflows in Healthcare. 5th

International Conference on BPM, Australia.

Becker, J., zur Mulhen, M. (1999). Towards a Classification

Framework for Application Granularity in Workflow

Management Systems. Proc. 11th International

Conference on Advanced Information Systems

Engineering (CAiSE).

Berg, M. (1999). "Patient care systems and healthcare work:

a socialtechnical approach." Int J.Med.Inf. 55: 87-101.

Bottorf, J. L. (1993). "The use and meaning of touch in

caring for patients with cancer." Oncol Nurs Fórum 20:

1531-8.

Caris-Verhallen, W., Timmermans, L., van Dulmen, S.

(2004). "Observation of nurse-patient interaction in

oncoloy: review of assessment instruments." Patient

Education and Counseling 54: 307-320.

Duitshof, M. (1995). Workflow Automationin Three

Administrative Organizations. Department of Computer

Science – Section Information Systems. Twente,

University of Twente - The Netherlands. Masther's

Thesis.

Hewinson, A. (1995). "Nurses’ power in interactions with

patients." J Adv Nurs 1995 21: 75-82.

Holland, M., Young, J. (2006). "Improving clinical

workflow through effective context and identity

management." Health Industry Insights.

Jacoski, C. (2005). Peculiaridades do fluxo de informações

em pequenos escritórios de projeto de edificações. V

Workshop Brasileiro da Gestão do Processo na

Construção de Edifícios, Brazil.

Keizer, N., Ammenwerth, E. (2005). "The effects and

quality of IT evaluation studies: Trends in 1982-2002."

AMIA Annu Symp Proc.: 186-90.

Knolmayer, G., Endl, R., Pfahrer, M. (2000). Guidelines for

Business Process Modeling. Business Process

References 315 Management. Berlin, Springer-Verlag.

1806: 16-29.

Krovi, R., Chandra, A., Rajagolapan, B. (2003).

"Information Flow Parameters for Managing

Organizational Processes." Communications of the

ACM 18(3): 286-292.

Lessard-Hébert, M., Goyette, G., Boutin, G. (2008).

Investigação Qualitativa Fundamentos e Práticas,

Instituto Piaget.

Lorenzi, N., Riley, R. (2003). "Organizational issues =

changes." International Journal of Medical Informatics

69: 197-203.

HEALTHINF 2010 - International Conference on Health Informatics

300

Malhotra, S., Jordan, D., Shortliffe, E., Patel, V. (2007).

"Workflow modelling in critical care: Piecing

together your own puzzle." Journal of Biomedical

Informatics 40(2): 81-92.

Nielsen, P., Thomson, B., Jackson, R., Kosman, K.,

Kiley, K. (2000). "Standard Obstetric Record

Charting System: Evaluation of a New Electronic

Medical Record." Obstet & Gynecol 96: 1003-8.

Oroviogoicoechea, C., Elliott, B., Watson, R. (2007).

"Review: evaluating information systems in

nursing." J Clin Nurs. 17: 567-575.

Peleg, M. (2007). Guidelines and Workflow Models.

Clinical Decision Support, The Road Ahead: 281-

306.

Reijers, H. (2003). Design and Control of Workflow

Processes. Germany, Springer.

Sharp, A., McDermott, P. (2001). Workflow Modelling

– Tools for Process Improvement and Application

Development. London, Artech House.

Silva, A., Videira, C. (2005). UML Metodologias e

Ferramentas CASE. Portugal, CentroAtlantico.PT.

Silva, J., Oliveira, A., Costa, J.A., Lima, J.E., Penha, R.

(2008). Fluxo da Informação, a partir dos

prontuários, para a gestão das organizações em

saúde: um estudo de Caso na Liga Norte

Riograndense Contra o Câncer. XI SEMEAD –

Seminários em Administração FEA-USP, Brazil.

Unertl K.M., W. M. B., Johnson K.B. (2006). Applying

Direct Observation to Model Workflow and Assess

Adoption. AMIA 2006 Symposium Proceedings.



STUDY OF CLINICAL WORKFLOW AND INFORMATION FLOW OF A BREAST CARE UNIT

301