CIN: Computers, Informatics, & Vol. 29, No. 12, 706–713 & Copyright B 2011 Wolters Kluwer Health | Lippincott Williams & Wilkins

FEATURE

ARTICLE

Informing the Design In the ICU, an extensive array of variables from the of Hemodynamic hemodynamic monitoring display is routinely ana- lyzed. However, the development of new display Monitoring Displays technologies is proceeding without adequate study of the monitoring tasks and behaviors of a primary user group—critical-care nurses. Semi- ALEXA K. DOIG, PhD, RN structured interviews focusing on the cognitive FRANK A. DREWS, PhD aspects of the hemodynamic monitoring task were MAUREEN R. KEEFE, PhD, RN, FAAN conducted with 14 critical-care nurses. A system- atic content analysis of qualitative data identified cognitive tasks that had applicability to the design The ICU monitoring display provides a visual repre- of monitoring displays. The cognitive tasks of hemodynamic monitoring were (1) selective data sentation of an array of physiological variables and acquisition, (2) applying meaning to the vari- thus is the interface between the patient and the critical- ables and understanding relationships between care clinician. The format of the display affects the parameters, (3) controlling hemodynamics by clinician’s ability to detect changes in physiological var- titrating medications and intravenous fluids, and iables, interpret data, and identify interventions. A (4) monitoring complex trends of multiple inter- number of investigators have designed and evaluated acting variables and patient response to inter- monitoring displays for clinical data presentation (see ventions. Recommendations include designing Gorges and Staggers1 for a systematic review). How- the monitoring display to match the mental con- ever, the development of new display technologies is structs and cognitive tasks of the user by applying proceeding without adequate study of the monitoring conceptual meaning to the variables, highlighting tasks and behaviors of the primary users—critical-care relationships between variables, and presenting a ‘‘big picture’’ view of the patient’s condition. nurses. This article addresses how nurses use hemody- Monitoring displays must also present integrated namic monitoring displays and the cognitive tasks in- trends that illustrate the dynamic relationship be- volved in their use. tween interventions and patient response.

KEY WORDS

BACKGROUND Critical care & Data display & Nursing & Task performance and analysis Although nurses assess the physical, functional, and cognitive state of the patient using hands-on data acqui- sition such as health assessment and auscultation, critical-care monitoring practice includes hemodynamic monitoring, which consists of acquiring and interpreting Author Affiliations: College of Nursing (Drs Doig and Keefe) and data from single-sensor single-indicator devices. While Department of Psychology (Dr Drews), University of Utah, Salt Lake many professional activities in critical-care nursing are City. procedural, hemodynamic monitoring appears to be This research was funded by a Ruth L. Kirschstein Predoctoral more complex. Aitken2 demonstrated that nurses inte- National Research Service Award (F31 NR008832-01) from the National Institute of at the National Institutes of grate a wide array of physiological data, information Health that was awarded to the first author. from the patient’s medical history, and current treat- The authors have disclosed that they have no significant relation- ments regimes, describing hemodynamic monitoring as ship with, or financial interest in, any commercial companies pertain- ‘‘a balancing act between a variety of different attri- ing to this article. 2(p216) Corresponding author: Alexa K. Doig, PhD, RN, University of butes and concepts.’’ Another body of research Utah, 10 South 2000 East, Salt Lake City, UT 84112 (alexa.doig@ has described expert nursing practice as being based on nurs.utah.edu). contextual intuition, where the nurse makes a decision DOI: 10.1097/NCN.0b013e3182148eba

706 CIN: Computers, Informatics, Nursing & December 2011

Copyright @ 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. because they instinctively ‘‘know’’ what to do.3–5 blood pressure, the nurse is expected to respond. How- Others attempted to dissect this phenomenon and ever, what that response will be and what cognitive identified data-driven patterns that the expert recognizes processes lead to that choice is difficult to predict. The in the situation, but cannot necessarily explain.6 How- nurse may also anticipate a certain course of events and ever, in Aitken’s study, all but one of the nurses were able will intervene before the patient’s condition demonstrates to consistently describe the rationale basis for their as- substantial changes. Goal parameters also change as a sessment. Furthermore, the nurse who described having a patient’s state worsens, and in some cases, the original ‘‘gut feeling’’ was the only one to perform erratically. Thus, goal is unattainable when a adverse effect of the chosen it appears that even when an expert nurse describes an treatment causes injury to other organ systems. intuitive feeling about the patient’s state, his/her assess- To describe context-dependent human behavior, ment still depends on how hemodynamic data are perceived Hollnagel10 focuses on three components associated and his/her ability to interpret patterns in those data. with human performance—competence, control, and When gathering information for the purpose of de- constructs. In hemodynamic monitoring, competence is signing a user interface, the focus can be on modeling gained though education, training, and experience and the system or modeling the user’s behavior. System can be evaluated though testing or observation. Practice modeling succeeded in the domain of engineering with standards are an attempt to raise competence by de- the application of abstraction hierarchy methods and scribing context-dependent procedures and defining the development of monitoring displays using principles minimum standards for monitoring and intervention. of ecological interface design.7,8 In the ICU, the moni- Part of the hemodynamic monitoring task also involves toring task is particularly complex since clinicians control, for example, responding to changes in patient monitor a dynamic physiological system that does not state and predicting declines in patient status or adverse necessarily respond within the normal constraints. Un- physiological events. A construct describes ‘‘the mental like engineered systems, the human body has interindi- representations of what a person may know or assume vidual variability that cannot always be predicted (for a about the situation in which the action takes place.’’10(p81) general perspective on this issue, see Durso and Drews9). Nurses may use constructs to assess and understand the Consequently, two patients with the same disease will patient’s hemodynamic state and to make intervention not necessarily respond in the same manner. The physi- decisions. They may also use constructs to predict the ological systems in the body also have the ability to patienttrajectoryorresponsetotreatmentandtoan- compensate, although with severe disease or physio- ticipate adverse events. One opportunity to affect nurse logical exhaustion those compensatory mechanisms performance by supporting task-oriented constructs is to often fail. Furthermore, when critically ill or injured, design display interfaces that promote data perception or under the influence of medications that influence the and support integration functions. However, it is unclear cardiovascular system, physiological response may vary which constructs are most useful in the task of hemo- from day to day or minute to minute in the same pa- dynamic monitoring. tient. Hence, the constraints between variables in en- gineered systems that provide the basis for ecological interface design may not be as predictable in unstable RESEARCH OBJECTIVES physiological systems. Clearly, with our ability to con- sistently and predictably model the system being limited, The goal of the present study was to acquire informa- to drive interface design a focus on the user’s behavior tion about critical-care nurses’ monitoring practices by must be considered. exploring, from a functional perspective, how nurses use Hollnagel10 describes a functional approach to study- monitoring displays and the cognitive tasks involved in ing context-dependent human behavior that is more their use. The findings were used to identify areas of flexible than traditional information processing models. improvement of hemodynamic monitoring displays to In this model of human action, the traditional end-point support the efficient monitoring behaviors. Finally, a goal such as ‘‘achieving stability in the system’’ is re- number of recommendations for display design strat- placed by a process-oriented goal such as ‘‘achieving egies were formulated. stability as efficiently as possible.’’ Additional aspects of performance-related behavior include anticipation and planning. Using nursing practice as an example, it is METHODS recognized that there are many ways for a nurse to accomplish the control task of maintaining a patient’s Study Design and Procedures blood pressure and cardiac output (CO) within a defined range through the use of medications and intravenous An exploratory, descriptive design was used to capture fluid administration. If there is a drop in the patient’s the thoughts, perceptions, and cognitive processes of

CIN: Computers, Informatics, Nursing & December 2011 707

Copyright @ 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. each nurse participant. Individual, semistructured inter- of technology. These themes were either merged with or views lasting 30 to 45 minutes were conducted with 14 added to the list that had been previously developed experienced critical-care nurses. Interviews were audio during the global reading. recorded and transcribed. The focus of the interview With this preliminary list of themes, the interview was on knowledge regarding current hemodynamic data were reanalyzed, and individual statements were monitoring practices, use of technologies, and sugges- coded. Many themes derived from the tions for designing new displays for patient monitoring. would have required an observation, and others such as The structure of the questions was open ended to allow the use of intuition did not emerge as a consistent ap- for follow-up probes to elicit further information. This proach to patient monitoring. As new themes emerged, study was approved by the University of Utah’s institu- previously analyzed data were reevaluated. After this tional review board, and written informed consent for process was completed, the coded data were examined conducting and recording the interviews was obtained to determine the structure and nature of the themes. At from the participants. this point, the idea of these themes describing primarily cognitive tasks emerged, resulting in the final organiz- ing framework. Sample Within each cognitive task, the data were examined again to identify statements made by nurse participants Fourteen participants who met the following inclusion that were congruent or contradictory. Challenges and criteria were recruited: an active RN license, at least 3 barriers were also identified and described. Lastly, the years’ experience in critical care, currently working in data coded under each category of cognitive task were an ICU, and familiarity with hemodynamic monitoring. then examined to determine the frequency of each re- Two of the nurse participants worked in a burn-trauma sponse or comment. Since the nature of this study was ICU, one was from a medical ICU, and the remaining purely exploratory, all comments and ideas presented 11 worked in a surgical ICU setting. All had current or by the nurses were treated with equal weight. recent clinical experience. Average experience in critical care was 11.6 (SD, 9.3) years (range, 3–30 years), and six of the 14 nurses were male. RESULTS AND DISCUSSION Qualitative Data Analysis An overview of the four categories of cognitive tasks This study uses a realist approach to analyzing the identified during the analysis of the interview tran- interview data wherein the descriptions are interpreted scripts is presented in Table 1. as the reality of the nurses’ monitoring practice and could be verified by observational study.11 Interview data were analyzed according to the guidelines for content analysis Task Category 1: Selective and theme development of Ryan and Bernard.12 In the Data Acquisition first step of the data analysis, the transcripts were read several times to obtain a global overview of the interview Nearly all interviewed nurses had a set of parameters in data. General categories and themes associated with mind when they approached their hemodynamic mon- each interview questions were recorded. Emerging itoring task. Five of the surgical ICU nurses reported themes were use of global concepts in hemodynamic routine use of all the basic parameters obtained from monitoring, applying meaning to the numerical data, a pulmonary artery (PA) . As one nurse stated, how trend data are acquired and used, and how previous ‘‘With the sickest patients, I’m pretty much using every- knowledge (and bias) affects variable selection. thing.’’ The other seven surgical ICU nurses appeared to Next, general concepts about nurses’ use of hemody- focus on subsets of data, especially blood pressure, namic monitoring data were derived from the nursing heart rate, and CO. These three hemodynamic param- literature on hemodynamic and other types of ICU pa- eters were routinely used to assess heart function and to tient monitoring.2,13–15 Themes that emerged from this monitor the patient’s postoperative recovery. In addi- review of the literature included use of limited data sets tion to variables obtained from a PA catheter, many by expert nurses, structure of knowledge required for nurses mentioned other sources of data central to he- hemodynamic monitoring, use of intuition, holistic ap- modynamic monitoring: physical and cognitive assess- proaches to monitoring a complex system, nonlinear ment findings, urine output, and laboratory data. thinking, the relative value of clinical attributes, dif- Lack of perceived value and the fact that some pa- ficult concepts for novice nurses, complexity of the rameters are not displayed on the main screen were two monitoring and decision-making task, and current use reasons given for the limited use of indexed variables

708 CIN: Computers, Informatics, Nursing & December 2011

Copyright @ 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. Table 1 any additional information to the hemodynamic picture; however, there may be key variables that novice nurses Categories of Cognitive Tasks in or nurses who do not routinely perform hemody- Hemodynamic Monitoring namic monitoring do not understand and therefore do Cognitive Task Summary of Issues not use. Selective data & Selecting the relevant data for acquisition each patient & Limited by ability to understand Task Category 2: Data Interpretation and/or conceptualize variables Data interpretation & Applying meaning to variables The interviews provided strong evidence that data inter- & Visualizing the big picture pretation is a central task during hemodynamic mon- & Understanding interrelationships itoring. Nurses used phrases such as ‘‘trying to put it all between variables together,’’ ‘‘understanding or interpreting the numbers,’’ & Diagnostic responsibilities and ‘‘tying the numbers to the picture’’ to describe their & Time constraints cognitive activities. One nurse acknowledged that Controlling & Titrating medications and ‘‘critical-care nurses are in diagnosis mode.’’ hemodynamics intravenous fluids to achieve a In response to being asked to identify difficulties that physiological goal novices have interpreting hemodynamic data, one nurse & Dynamic nature of treatment goals stated, ‘‘They can look at a number, some can tell me the & Predicting the effects of normal range, but then how it relates to a patient’s normal treatments on other systems Monitoring complex & Available computerized trend physiology is a difficult concept.’’ Rather than identifying trends functions not routinely used one particular skill or variable, many nurses remarked & Need to visualize and chart that novices had difficulty integrating all of the numbers interrelationships between into a ‘‘big picture’’ representation of patient state. Upon interventions and physiological further exploration, several key factors emerged as un- trends derlying barriers to constructing an overall appreciation & Defaulting to memory for trend of patient state. Four participants felt that new nurses assessment miss potential problems by getting caught up in all of the numbers coming from the monitoring display. These such as the various ventricular work and vascular resis- nurses and almost all of the others reported that new tance indices. As one surgical ICU nurse explained, ‘‘The nurses often struggle to apply meaning to the numbers. ones I focus on are the ones I see. I’m sure there’s some One stated, ‘‘The hardest part of this job is learning what kind of function to [the indices] as far as the patient’s the numbers mean and how you can project to avert a condition goes, but as far as monitoring our drugs, they crisis.’’ It was generally felt that if the nurse could con- are not that important.’’ ceptualize what the variable represented and how it related to other variables, then they could understand how to use it. Three participants stated that new nurses have difficulty understanding the interdependence (or in ISSUES IN SELECTIVE DATA ACQUISITION some cases uncoupling) of variables. It was also remarked When nurses use all of the data acquired from a PA that new nurses have trouble linking changes as a series of catheter, they are integrating more than of eight hemo- cues to a developing problem. A new nurse may detect dynamic variables with additional data obtained from changes in a monitored variable, but may be unable to put the electrocardiogram, pulse oximeter, and dis- the pieces of the puzzle together. Other nurses suggested play. Hemodynamic monitoring can justifiably be classified that novices get caught up in the numbers, but do not as a complex cognitive task. Complexity is heightened by know what to do with them. Another emerging theme interindividual variability related to prior health status, was the negative impact of time pressure on the quality of severity of disease or injury, and age. patient monitoring. There is technology to acquire data, Several nurses reported using a narrow range of but in the end, as one nurse commented, ‘‘We have to be variables. Since this study relied on subjective descriptions able to put it all together, and sometimes this doesn’t of the monitoring task, it is not clear whether these nurses happen soon enough for the patient.’’ routinely ignore other acquired variables or these are considered only when abnormal. The hemodynamic ISSUES IN DATA INTERPRETATION monitors also provide several indices located on second- ary menus or screens that nurses do not routinely use. In The monitoring task in the ICU goes beyond observa- some instances, the indices mentioned may not contribute tion and documentation, although one nurse cautioned

CIN: Computers, Informatics, Nursing & December 2011 709

Copyright @ 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. that physicians do not want nurses to be in ‘‘diagnosis cation infusion titration or fluid administration, while mode.’’ Nevertheless, it was apparent from the interviews they assess parameters on monitoring display. Clearly, that critical-care nurses are frequently interpreting data nurses develop a conceptualization of the patient’s status for the purpose of identifying and understanding their and condition to adjust treatments with the goal of patient’s pathophysiological condition. achieving or maintaining homeostasis in the patient. The Regardless of which term is used to describe the cog- physiological goals in the ICU are multifocal, and many nitive process, when utilizing data from the PA catheter, of the treatments induce secondary effects. Critical-care nurses are required to apply meaning to the variables to nurses are expected to anticipate the effect of treatment construct a picture of the patient’s hemodynamic con- and may have to monitor for changes in other organs or dition. Understanding the interrelationships between the systems. If adverse effects are detected, they are required different variables is also an essential part of developing to intervene again, perhaps with a new goal in mind. the mental model. The observations of experienced nurses were that novices have difficulty abstracting the big pic- ture of the patient’s state or condition because they strug- Task Category 4: Monitoring Complex Trends gle to apply meaning to the numbers and have trouble recognizing the interdependence of variables. The novice All nurses emphasized the importance of monitoring may have a conceptual understanding of the acquired trends in critically ill, hemodynamically unstable pa- variables, but they often cannot conceptualize the under- tients. Despite this assertion, only one nurse used the lying mechanisms pertaining to the patient’s hemo- monitoring display’s trending functions routinely. Other dynamic state. Nurses who do not perform frequent nurses reported knowing of only a few staff members hemodynamic monitoring may have similar problems who regularly examine computerized trends. with data interpretation. Several nurses stated that they primarily kept track of general trends in their head. As one nurse put it, ‘‘When I’m taking care of patients, I’m in the room all the time. I’m aware of what’s going on, so I don’t find [the trend- Task Category 3: Controlling Hemodynamics ing functions] useful.’’ However, the handwritten flow sheet, it was explained, was the tool they consistently The most common uses of hemodynamic monitoring used for recording and monitoring trends. Even the sole were vasoactive or inotropic infusion titration and intra- nurse who used the monitor’s trending functions stated venous fluid administration. All nurses reported that they were expected to titrate medications and fluids according that, to fully evaluate trends, she deferred back to the flow sheet. Nurses verbalized that they were very re- to parameter-based medical orders or unit protocols. In luctant to give up the handwritten record of trends, one ICU, the nurses were also expected to initiate intra- despite having computerized charting. The flow sheet venous infusions based on unit protocols. As one nurse was especially important as a tool for describing trends described the control task: ‘‘You’ve got to keep the wedge to the oncoming nurse during shift report. (PAWP) above 16, or you need to wean dobutamine to Recording trends on a flow sheet involves writing down maintain a particular CO.’’ Another nurse remarked that numerical values for each monitored parameter at speci- novice nurses may understand the hemodynamics, but have trouble understanding how the infusions relate to fied intervals. This manner of monitoring trends in patient status has obvious drawbacks including the time required what they are seeing on the monitor. for data transfer, a lack of information between the times Several nurses described the task of understanding the data happened to be recorded, and the potential for how the different medications affect the different errors in transcription. Given these problems, why do hemodynamic variables. As one nurse stated, ‘‘We play these nurses persist with the task of manually writing the numbers game. If we have a high SVR [systemic down values rather than use the computerized trending vascular resistance] and low CI [cardiac index], then we functions on the monitoring display? Obtaining answers give them something to reduce the afterload, which should raise the CI.’’ If the nurse attempted to treat the to this question required an exploration of how these nurses use trending information and why current display low CI independent of the elevated SVR, then they technologies fail to meet their needs. might induce worsening heart failure. Whereas several nurses expressed interest in how a single hemodynamic variable such as CO changed over a 12- to 24-hour period, nearly all suggested that trends ISSUES IN CONTROLLING HEMODYNAMICS were most valuable in conjunction with a treatment Critical-care nurses use clinical data obtained from the goal. Two specific uses of goal-oriented trend monitor- monitors to drive treatment decisions. Most nurses ex- ing emerged from the interview data. First, short-term pressed having an intervention in mind, such as medi- trends were used to make decisions about future

710 CIN: Computers, Informatics, Nursing & December 2011

Copyright @ 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. treatment and interventions. Second, and most fre- performance lies in the development of constructs by de- quently, trends were used to monitor a patient’s ongoing signing display interfaces that promote data perception hemodynamic response to intravenous fluid and medi- and data interpretation. One of these constructs in- cation administration, often in the form of constant volves trying to apply a conceptual meaning to the var- infusions that the nurse titrates. In addition, the nurse iables. Another is the need to understand what some must communicate the patient’s response to other mem- nurses described as the big picture. This global perspec- bers of the healthcare team. As one nurse stated, ‘‘Our tive involved the incorporation of the relevant parame- monitors store all of the information, but not in a mean- ters, combined with information concerning the patient’s ingful way. It’s really important for an ICU nurse to be history and current treatment regimens. It appears that if able to see everything at once.’’ a nurse cannot conceptualize the meaning of a parameter A nurse who worked as an educator in a burn-trauma or see the big picture, then the data will go unused. The ICU hypothesized about another reason for the limited hemodynamic monitoring display should also attempt to use of available trending functions. ‘‘Most ICU nurses highlight relationships between associated variables. know how to evaluate their patient and know what’s Under conditions of normal homeostasis, relationships going on with them at that moment. Whatever happened between physiological variables can be accurately mod- in the past is not at the forefront of their mind.’’ Despite eled. However, with severe disease or injury, it is difficult rating trend monitoring as a highly useful tool, nurses to predict the physiological systems’ ability to respond tend to focus on the patient’s present state. since in many cases compensatory mechanisms are over- whelmed or nonfunctional. In addition, the body’s physiological mechanisms are influenced by medications and other treatments, such as the mechanical ventilator ISSUES IN MONITORING COMPLEX TRENDS and aortic balloon pump. Finally, since time pressures The majority of participants were not using avail- are common in the ICU, clinical data must also be pres- able computerized trending functions. One of the pri- ented in a manner that promotes rapid perception and, mary limitations of computerized trending functions is more importantly, rapid data interpretation. that they do not show the interrelationships between These findings support the development of integrated monitored variables and interventions. The ICU nurse graphical displays by depicting variables as shapes that wants to be able to easily observe the patient’s hemody- visually resemble and behave in a similar manner to the namic response to the titration of a particular drug or an physiological system.16 These graphical displays may adjustment of the ventilator setting. Charting and moni- help nurses see the big-picture view of a patient status, toring trends on a flow sheet are time consuming and thus reducing the amount of data processing they are incomplete. Furthermore, errors may occur during the required to perform during emergent events when time manual transfer of data. In addition, it is important for pressures and other tasks limit the resources available the entire healthcare team to have access to trending for step-by-step numerical data interpretation. Graph- information. ical displays may also help novice nurses learn about Another concern was the primary focus of some nurses relationships between constrained hemodynamic varia- on the present situation, with limited evaluation of trends. bles and develop an understanding of the links between Although an understanding of a patient’s current status is treatments and the range of possible hemodynamic crucial, many conditions evolve over time (eg, and responses. However, the method of graphically repre- heart failure). A seemingly stable patient, when only cur- senting variables and relationships between variables rent data are evaluated, may show signs of a developing should be thoroughly tested among diverse groups of condition if trends are closely evaluated. Because trending critical-care nurses to ensure that the data can be cor- data are not easily accessible or relevant to their task in rectly interpreted. its current form, critical-care nurses may be missing an One challenge of developing integrated graphical dis- important element of ICU monitoring. plays for monitoring physiological systems is that constraints are often difficult to accurately model. There- fore, the goal of these displays should be to provide cues to the relationships between associated variables, rather RECOMMENDATIONS FOR NEW than attempting to portray constraints as is done in the DISPLAY DESIGN STRATEGIES ecological interface design of engineered systems theor- ized by Vicente and Rasmussen.7 Reasons for the lim- It appears that critical-care nurses use mental models to itation of this approach are provided by Durso and Drews,9 understand the state of the patient and to predict the with one of the main problems being the differences in patient’s trajectory or response to treatment and to an- principle between monitoring technical and natural sys- ticipate adverse events. One opportunity to affect nurse tems that lead to different constraints and requirements.

CIN: Computers, Informatics, Nursing & December 2011 711

Copyright @ 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. Display interfaces for critical-care nurses must be de- monitoring display design such as the abstraction hier- signed to support active decision making rather than archy methods outlined by Vicente and Rasmussen.7 simply providing an indication of the patient’s status. A Using an information-processing approach to behavior monitoring display should present the data in the modeling is also limited since nurses are highly proactive context of the control tasks of the critical-care nurse. In and adaptive in their approach to patient monitoring. critical-care nursing practice, hemodynamic monitoring This study used a functional approach to behavioral is primarily used to titrate medications that affect the modeling as described by Hollnagel10 to describe the cardiovascular system or to administer intravenous cognitive tasks that critical-care nurses undertake dur- fluids. Monitoring displays therefore should provide a ing hemodynamic monitoring of severely ill and injured way for nurses to rapidly perceive the patient’s volume patients. This approach allows investigators to gather status and provide support for the control task of data regarding global behaviors, rather than focusing titrating vasoactive medications. on individual incidents, the variety of which is infinite. The present study demonstrated that critical thinking is The primary goal of the tasks described in this study was central to all aspects of monitoring task. The goal of the to achieve physiological stability in the patient as effi- monitoring display is not necessarily to provide the nurse ciently as possible. Knowledge of these context-dependent with a solution to the problem. Rather, the display should tasks can then guide the designers of patient monitoring highlight changes in patient status that may be missed displays to best support nurses’ need to be proactive and because of a lack of vigilance or inexperience and provide adaptive in the most efficient manner with minimal error. contextual support for data interpretation, decision mak- New display technologies must be developed with ing, and monitoring the treatments that are implemented. attention to the specific cognitive tasks of the intended Finally, it is important to emphasize that any automation user group. Nurses have a distinct focus in their ap- may create a situation that leads to deskilling of the proach to patient monitoring that may differ from that involved personnel. of physicians, technicians, and other critical-care clini- When monitoring critically ill, hemodynamically cians. Without rigorous evaluation of these underlying unstable patients, nurses need a data presentation that differences, features designed specifically for one do- explicitly shows the relationship between treatments main may not support the tasks of other users. Since and physiological changes over time. However, the is- nurses perform the vast majority of patient monitoring sue of monitoring complex trends in a critically ill in the ICU,17 display technologies must support the patient’s hemodynamic state has not been a major focus cognitive tasks of this particular user group to ensure of patient monitoring research and technology develop- that technology does not become a source of error and ment. Computerized trending functions on monitoring a threat to patient safety. displays appear not to be routinely used because they do not provide the integrated, response-to-treatment trends that critical care nurses require. New trend displays must show the relationships between key hemodynamic varia- REFERENCES bles and treatments such as medication and fluid admin- istration, intravenous infusion titration, or ventilator 1. Go¨rges M, Staggers N. Evaluations of physiological monitoring adjustments. Automation of trending information could displays: a systematic review. J Cin Monit Comput. 2008;22:45–66. 2. Aitken LM. Expert critical care nurses’ use of pulmonary artery reduce the risk of human error in the transfer of data, pressure monitoring. Intensive Crit Care Nurs. 2000;16:209–220. significantly decrease the time a nurse spends charting, 3. Agan RD. Intuitive knowing as a dimension of nursing. ANS Adv and would provide a continuous display of trended var- Nurs Sci. 1987;10:63–70. 4. Benner P. From Novice to Expert: Excellence and Power in Clini- iables. Nurses should also be able to use a summary of cal Nursing Practice. Menlo Park, CA: Addison-Wesley Publish- trending information to enhance communication with ing Co; 1984. other healthcare team members and during handoffs such 5. Rew L, Barrow EM. Intuition: a neglected hallmark of nursing knowledge. ANS Adv Nurs Sci. 1987;10:49–62. as the change of shift report. 6. Klein G. Developing expertise in decision making. Think Reason- ing. 1997;3:337–352. 7. Vicente KJ, Rasmussen J. Ecological interface design: theoretical foundations. IEEE Trans Syst Man Cybern. 1992;22:589–606. CONCLUSION 8. Effken JA, Kim NG, Shaw RE. Making the constraint visible: test- ing the ecological approach to interface design. Ergonomics.1997; To date, myriad monitoring displays and other display 40:1–27. 9. Durso FT, Drews FA. , aviation, and ecosystems—a interfaces have been developed without a thorough ex- socio-natural systems perspective. Curr Dir Psychol Sci. 2010;19: amination of nurses’ mental model of the task or behavior 71–75. that the interface is designed to support. 10. Hollnagel E. Modelling the orderliness of human action. In: Sarter NB, Amalberti R, eds. Cognitive Engineering in the Aviation The dynamics of monitoring a physiological system Domain. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.; 2000: make it impractical to use traditional approaches to 65–98.

712 CIN: Computers, Informatics, Nursing & December 2011

Copyright @ 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 11. Silverman D. Analyzing talk and text. In: Denzin NK, Lincoln Y, 14. Effken JA. Informational basis for expert intuition. J Adv Nurs. eds. Collecting and Interpreting Qualitative Materials. 2nd ed. 2001;34:246–255. Thousand Oaks, CA: Sage; 2003:340–362. 15. Redden M, Wotton K. Clinical decision making by nurses when 12. Ryan GW, Bernard HR. Data management and data analysis faced with third-space fluid shift. How well do they fare? Gastro- methods. In: Denzin NK, Lincoln Y, eds. Handbook of enterol Nurs. 2001;24:182–191. Qualitative Research. 2nd ed. Thousand Oaks, CA: Sage; 2000: 16. Drews FA, Westenskow, DR. The right picture is worth a thou- 769–802. sand numbers: data displays in . Hum Factors. 2006;48: 13. Currey J, Botti M. Naturalistic decision making: a model to 59–71. overcome methodological challenges in the study of critical care 17. Donchin Y, Gopher D, Olin M, et al. A look into the nature and nurses’ decision making about patients’ hemodynamic status. causes of human errors in the . Crit Care Med. Am J Crit Care. 2003;12:206–211. 1995;23:294–300.

CIN: Computers, Informatics, Nursing & December 2011 713

Copyright @ 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.