Effects of interruptions on prospective memory performance in ...

PROCEEDINGS of the HUMAN FACTORS AND ERGONOMICS SOCIETY 52nd ANNUAL MEETING—2008

808

Effects of interruptions on prospective memory performance in anesthesiology T. Grundgeiger1, D. Liu1, P.M. Sanderson1, S. Jenkins2, & T. Leane2 1

The University of Queensland, Brisbane, Australia; 2 Royal Adelaide Hospital, Adelaide, Australia

Copyright 2008 by Human Factors and Ergonomics Society, Inc. All rights reserved. 10.1518/107118108X353147

Interruptions have been associated with adverse events in healthcare. However, supporting studies are descriptive and atheoretical rather than explanatory, and they seldom show that interruptions compromise patient safety. Prospective memory may provide useful theoretical background. We analyzed video from a full-scale patient simulator for factors enhancing or inhibiting anesthesiologists’ prospective memory performance. The critical task was to remember to cross check a unit of blood against the patient before administering the blood. All 12 participants were interrupted by the surgeon when the blood arrived. Only participants who self-initiated the retrieval (n = 3), or returned their full attention to the transfusion task and saw the blood bag label (n = 7), remembered the check. The result can be explained with findings from prospective memory literature. INTRODUCTION

Prospective memory (PM)

Interruptions can be described as “an external intrusion of a secondary, unplanned, and unexpected task, which leads to a discontinuity in task performance” (Brixey et al., 2007). Research on interruptions and distractions in laboratory settings (Monsell, 2003), driving related research (Monk, Boehm-Davis, & Trafton, 2004) and aviation (Dismukes, Young, & Sumwalt, 1998) shows that interruptions negatively affect performance. Motivated by these results and the fact that the medical domain is a highly interrupted workplace (e.g., Drews, 2006), researchers have started to investigate interruptions and distractions in healthcare. However, these studies are not guided by a theory of how interruptions lead to negative effects and seldom report evidence for a relation between interruptions and negative outcomes. A common and critical task in healthcare is blood transfusion. Transfusing the wrong blood to a patient is the leading cause of transfusion related serious incidents and accounts for 37 % of all transfusion-related errors (Linden, Wagner, Voytovich, & Sheehan, 2000). The bedside check verifying that the patient is the correct recipient for the blood is considered to be “the most critical task to prevent mistransfusion events” (Dzik, 2007, p. 186). In this paper, we use prospective memory theory to analyze video data from a full-scale patient simulator study with anesthesiologists and show how interruptions can lead to forgetting of the bedside check.

PM is the ability to recall a previously formed intention at a specific time or in response to a specific cue in the future, without being encouraged to recall the intention (McDaniel & Einstein, 2000). Examples include remembering to call a colleague at 4 pm (time) or to give the colleague a paper the next time you see them (cue). Research on PM is receiving growing interest and researchers have shown that PM performance is influenced by factors such as importance of the intention, availability and properties of cues, and attentional resources (for an introduction and review see McDaniel & Einstein, 2007). Only Dieckmann, Reddersen, Wehner, and Rall (2006) have studied PM in healthcare. They investigated the effects of intention importance and type with medical students training in simulators, and reported a trend for important intentions to be executed more often. Despite some methodological limitations, they showed that patient simulators can be used to run PM experiments. We briefly review four influencing factors of PM performance that are relevant to our study: habitual tasks, task context, divided attention and cue-task association. First, habitual tasks are performed often and usually do not need attentional resources. It can be argued that habitual tasks do not involve PM since the intention is not explicitly formed. Dismukes (2008) disagrees, however, and considers them PM tasks because participants report that they had intended to do the task if asked afterwards. The execution of habitual tasks depends upon cuing from events in the environment or preceding task steps (Meacham & Leiman, 1982, cited in Dismukes, 2008). Second, the probability of remembering to do a PM task increases if the context of encoding matches the context of retrieval (e.g., Nowinski & Dismukes, 2005). In standard laboratory studies involving event-based PM tasks, participants work on an ongoing task, such as rating word pleasantness. The PM task is to press a designated key when a certain cue is encountered (e.g., the word “milk”). Nowinski and Dismukes (2005) used two different ongoing tasks (Task A and B) in a task context experiment. Introducing the PM task with an example of Task A led to better PM performance in Task A blocks vs. Task B blocks because the task context matched.

Interruptions in healthcare Research on interruptions in healthcare is typically descriptive and reports the number of interruptions that occur, who interrupts whom, for how long, and so forth (e.g., Chisholm, Collison, Nelson, & Cordell, 2000; Drews, 2006; Spencer, Coiera, & Logan, 2004). Because the research is descriptive, it is not possible to draw conclusions about whether and how interruptions and distractions can jeopardize patient safety. Parker & Coiera (2000) are the only authors to propose a theoretical explanation of how interruptions may lead to forgetting of intended actions. However, Parker and Coiera’s (2000) model is based on basic psychology and findings on prospective memory are not considered (Grundgeiger & Sanderson, 2008).

809

PROCEEDINGS of the HUMAN FACTORS AND ERGONOMICS SOCIETY 52nd ANNUAL MEETING—2008

Detected

Engages P sees AN do first immediately steps of in HDU task transfusion task

Time P looks towards label (seconds)

P looks at label when remembering check

Time between interruption & detection (seconds)

#4

No

Yes

No, talking to S

2 (+5 where label at screen border)

# 10

No

Yes

No, talking to S

2

/

/

#3

Yes

No

No, talking to S

7

Yes

73

#6

Yes

No

No, talking to S

3

Yes

10

#7

Yes

No

No, talking to S

3

Yes

22

# 12

Yes

No

No, talking to S

5

Yes

7

#1

Yes

No

Yes

7

Yes

no audio

#2

Yes

No

Yes, but talking to S

5

Yes

7

#8

Yes

No

Yes, but talking to S

4

Yes

4

#5

Yes

No

Yes, but talking to S

0

No

6

#9

Yes

No

Yes

0

No

2

# 11

Yes

No

Yes, but talking to S

0

No

1

/

/

Step at which P remembers check

Description of case

Analysis of case

Agrees to HDU transfer; waits for AN to finish transfusion task to immediately delegate HDU task to AN; stands in 90 degree angle to blood label; misses check

Misses first steps of transfusion task (no cuing, no task context); not engaged in transfusion task indicated by immediate task delegation (divided or / no attention); stands in 90 degree angle to blood label (no specific cue) Talking to S; agrees to HDU transfer and Misses first steps of transfusion task (no cuing, no immediately engages in HDU task; P ensures task context); engages immediately in HDU task that AN is watching monitor before P is indicated by behavior and utterances (divided or / calling HDU; misses check no attention); label out of view while on phone (no specific cue) Talking to S; label is facing away from P or is Misses first steps of transfusion task (no cuing, no covered by AN's arm; denies request and task context); returns late to transfusion task (full Transfusing returns to transfusion task; asks for pressure attention); view on label obscured (no specific cue for 74 seconds bag on blood bag; after appling pressure bag, available); looks at label (specific cue) label is facing P; looks at label; asks if checked Talking to S; agrees to HDU transfer; returns Misses first steps of transfusion task (no cuing, no AN taking to transfusion task; looks at label; asks if task context); returns attention on transfusion task down IV bag checked (full attention); looks at label (specific cue) Talking to S; turns head to AN from time to Misses first steps of transfusion task (no cuing, no Transfusing time; denies request and finishes discussion; task context); returns attention on transfusion task for 9 seconds looks at label after transfusion started; asks if (full attention); looks at label (specific cue) checked Talking to S; agrees to transfer; watches AN Misses first steps of transfusion task (no cuing, no AN taking set up blood; looks at label; asks if checked task context); returns attention on transfusion task down IV bag (full attention); looks at label (specific cue) Brief distraction of S; agrees and delays HDU task; returns to blood task early; takes other blood bag out of esky; looks at label; asks if checked Talking to S while checking IV lines; denies AN taking request and returns to transfusion task; looks down IV bag at label; asks if checked Talking to S; turns head to AN from time to Transfusion time; denies request and finishes discussion; about to start looks at label just before transfusion starts; asks if checked Talking to S; rejects transfer; returns to AN looking for transfusion task; looks at esky; gives order to clamp check blood Brief distraction of S; agrees and delays HDU AN taking bag task; immediately engages in transfusion task out of and delegates blood check to AN container AN taking down IV bag

AN taking down IV bag

Talking to S; turns head to AN from time to time; denies request and finishes discussion; immediately orders check

Table 1: Measures, verbal descriptions and analyses of the single cases (AN = anesthetic nurse, P = participant, S = surgeon).

Aware of transfusion task (preceding steps cue, task context); returns attention on transfusion task (full attention); looks at label (specific cue) Engages in transfusion task early (no cuing, but task context); full attention on transfusion task at detection Aware of transfusion task (preceding steps cue, task context); returns attention on transfusion task (full attention); looks late at label (specific cue) Aware of transfusion task (preceding steps cue, task context); returns attention on transfusion task (full attention); self initiated intention Aware of transfusion task (preceding steps cue, task context); returns to transfusion task (full attention); immediately delegates blood check to AN; self initiated intention Aware of transfusion task (preceding steps cue, task context); returns attention on transfusion task (full attention); self initiated intention

PROCEEDINGS of the HUMAN FACTORS AND ERGONOMICS SOCIETY 52nd ANNUAL MEETING—2008

Third, dividing attention worsens PM performance (e.g., Logie, Maylor, Della Sala, & Smith, 2004; Marsh & Hicks, 1998). Consider an experiment involving an event-based PM task. If an additional task (e.g., judging whether the answer to an arithmetic problem is correct) is added to the standard procedure, the participant must then divide their attention between the ongoing and the additional task. Dividing attention lowers the probability of remembering the PM task. Fourth, stronger associations between the PM cue and the PM task leads to better PM performance (e.g., Cherry et al., 2001; Nowinski & Dismukes, 2005). For example, the instruction may be to press a certain key whenever a word from a certain category is encountered. More typical examples from the category (e.g., animal - dog) lead to better PM performance than less typical members (e.g., animal wombat) because of stronger prior associations between the category and the words. In the following section, we give a brief description of the scenarios, the nature of the event with the PM task, and how we analyzed the video data. METHOD Scenarios Three highly detailed 35-40 min scenarios were designed for a study on head mounted displays in anesthesia. The scenarios were run in a full-scale patient simulator and each scenario contained eight unique events including the “Failure to check blood” event described below. Video recordings were captured using two cameras in the simulated operating room and a head-mounted camera worn by participants. Failure to check blood event The event was preceded by major hemorrhage (see Figure 1). All participants decided to order blood from the blood bank, began treating the hypovolemia with IV fluids, and waited for the blood to arrive. Finally, the blood arrives and is accepted by the anesthetic nurse who was waiting near the door. Concurrently, the surgeon distracts the participants by asking them to organize a bed in the High Dependency Unit (HDU). After accepting a cooler containing the blood, the nurse walks straight to the drug trolley on the opposite end of the room, takes the blood out of the cooler and hangs it on the infusion stand. The nurse then clamps and disconnects the IV fluid, spikes the blood bag and starts the transfusion. Finally,

15 mins before the event: 12 mins before the event: Participant notes evidence Participant arranges for blood from blood bank of major hemorrhage

810

the nurse stands back and waits. The nurse deliberately omits validating the blood product against the patient in a clear breach of standard operating procedure. The event was classified as missed if the participant did not detect the omitted check within three minutes after the arrival of the blood. Analysis of event videos Two researchers analyzed the video in three stages and resolved discrepancies via discussion. No formal reliability tests were conducted. First, the 12 event instances were watched to see which PM influencing factors reported in the literature (McDaniel & Einstein, 2007) were relevant and applicable to the event. Second, the descriptive measures and information represented in Table 1 were extracted (excluding the far right column). In addition, we checked whether the initial distraction by the surgeon led to an interruption of the participant. We used Brixey et al.’s (2007) definition of interruptions: “an external intrusion of a secondary, unplanned, and unexpected task, which leads to a discontinuity in task performance”. Third, the information from the preceding steps was used to write an analysis of each event instance (far right column of Table 1). We explained whether the PM performance influencing factors introduced earlier have either an enhancing or an inhibiting effect in each instance. The essence of these analyses is represented as a flow diagram in Figure 2. RESULTS AND DISCUSSION The analysis of the videos showed that every participant engaged in a conversation with the surgeon. The conversation lasted between 6 and 48 seconds. All participants were therefore interrupted in accordance with our definition. Most striking is the fact that only the two participants who immediately engaged in the HDU task failed to check the blood (left columns of Table 1). We make two assumptions in our analysis. First, we assume that all participants intended to check the blood, as they are responsible for the patient. One may question whether participants see the blood check as part of the scenario. We do not believe that this is the case because of the highly detailed scenarios and the fact that most participants checked the blood (10 of 12). Furthermore, anesthesiologists in patient simulators expect events to occur and are therefore often more vigilant.

10 mins before the event: IV fluids administered to treat hypovolemia

Event: Blood arrives in OR while surgeon requests transfer to HDU (concurrent)

Figure 1: Screens shots of participant’s head mounted camera as event unfolds.

811

PROCEEDINGS of the HUMAN FACTORS AND ERGONOMICS SOCIETY 52nd ANNUAL MEETING—2008

Event increasing probability of remembering

1,2,3,6, 7,8,12

1,2,5, 8,9,11 1,2,5, 8,9,11

Event decreasing probability of remembering

REMEMBER N = 10 5,9,11

P sees first steps of transfusion task

Blood arrives Distraction starts

3,4,6, 7,10,12

P disengages from HDU task (attention on transfusion task)

P looks at blood bag label

3,6,7,12 4

4 P does not see first steps of transfusion task

4,10

P engages in HDU task (attention on HDU task)

10

P does not look at bag label

FORGET

Factors influencing PM

N=2 Cuing of preceding task steps

Divided attention & Task context

Cue-task association

(A)

(B)

(C)

Figure 2: Flow diagram of how different events increase or decrease the probability of remembering to check the blood and the PM influencing factors. Upper shaded areas describe the observable event, while lower shaded areas indicate theoretical mechanisms. The numbers on the arrows indicate participant numbers from Table 1. Second, we consider blood transfusions to be a routine task for anesthesiologists, and the blood check to be a subtask of the transfusion task. Local subject matter experts confirm that this is the case in the hospital where the study was performed. Our analysis of the event starts with the arrival of the blood and how it subsequently unfolded. We discuss how the four PM influencing factors (habitual tasks, divided attention, task context, cue-task association) either increases or decreases the probability of participants remembering to do the blood check. First, the nurse accepted the blood while the participant was concurrently distracted. These events led to a disadvantageous situation in which the first steps of the transfusion task happened behind the participant’s back. Hence, the blood check subtask was not cued from the previous steps of the task because the participant was not doing or even seeing the steps (Dismukes, 2008) (see Figure 2A). Second, the two participants who immediately engaged in the HDU task were either dividing or turning their full attention to the HDU task. In general, divided attention leads to decreased PM performance in event-based PM tasks (Marsh & Hicks, 1998). The likelihood that these two participants would be reminded/cued by the nurse setting up the transfusion or by the label on the blood bag decreased because of their immediate engagement in the HDU transfer task (see Figure 2B). Third, the same two participants who engaged in the HDU task were not in the context of the blood transfusion task, which further decreased the likelihood of them remembering to do the blood check (Nowinski & Dismukes, 2005) (see Figure 2B). This interpretation is supported by the cases in which participants later remembered to perform the blood

check. In these cases, it is only after the end of the interruption, when they returned to the transfusion task, that the participants remembered the check. Fourth, seven participants asked the nurse whether the blood had been checked immediately after looking at the label on the blood bag. In these cases, the retrieval of the task was initiated by a retrieval cue. Furthermore, the sight of the nurse setting up the blood transfusion may not have been specific enough to cue the participants to do the check. They remember to perform the check only when they see the label, which is a more specific cue than the nurse. This explanation is supported by the results on cue-task association reported earlier (Nowinski & Dismukes, 2005) (see Figure 2C). However, participant 4 seems to be an exception. This participant was dividing attention and was in a different task context. Under these conditions, even a specific cue might have been not strong enough to cue the check. Moreover, the label was only half visible on the head-mounted camera footage and so the participant might have not seen it at all. With regard to PM, it is worth noting that four participants decided to defer the HDU task. In studies where task execution was deferred by as little as 10 seconds, participants were less likely to remember to complete the task (e.g., Einstein, McDaniel, Williford, Pagan, & Dismukes, 2003). CONCLUSION Research on interruptions in healthcare typically does not refer to theory to explain how interruptions compromise patient safety. We used PM theory to explain why participants either remember or forget to check the patient data before a blood

PROCEEDINGS of the HUMAN FACTORS AND ERGONOMICS SOCIETY 52nd ANNUAL MEETING—2008

transfusion. A clear pattern was found in the data. Remembering the check depended upon self-initiation or returning full attention to the transfusion task and being cued by a specific cue (blood bag label). Forgetting occurred with immediate engagement in the interrupting HDU task. Overall, applying basic findings of PM turned out to be a successful way of demonstrating and explaining how interruptions can lead to safety-compromising events. Our conclusions are limited by the small sample size, a missing baseline of uninterrupted trials, and the general limitations of patient simulators as reported earlier. Although one could argue that these results are specific to the anesthesiology context, our explanations are based on theory and are applicable to other contexts. Finally, these results are based on a post-hoc analysis and not a prospective experiment. Nonetheless, if compared to other retrospective analyses (e.g., incident reporting systems), the data we present offers the unique features of a tightly controlled scenario with “identical” cases and detailed video data. In future studies, an eye tracker would allow an even more detailed analysis. Since interruptions are a common form of communication and can have positive effects on patient safety, they cannot, and should not, be prevented (Grundgeiger & Sanderson, 2008). Thus, the question is how we can help healthcare staff overcome the negative effects of interruptions. One approach is to train healthcare staff to avoid fixating on the interruption and to encourage double checks. A second approach is to design equipment and information technology in a way that ensures general execution of tasks and resumption of interrupted tasks. Effective techniques for improving task resumption include placing reminders (Dodhia & Dismukes, 2005; McDaniel, Einstein, Graham, & Rall, 2004) or adding a break after an interruption (Dodhia & Dismukes, 2005). A third approach is to design information technology to change the cognitive demands of the task. Turner, Casbard, and Murphy (2003) reported that using a barcode patient identification system “appeared to dissuade individuals from becoming distracted and interrupted” (Turner et al., 2003, p. 1206). A reduction in cognitive demands associated with the new system probably helped to preserve individuals’ attention to the primary task. Overall, we should seek innovative ways to help healthcare workers experience the benefits of interruptions while minimizing the dangers. ACKNOWLEDGEMENTS This research is supported by Australian Research Council Discovery Project grant ARC DP0559504 to Sanderson, Watson, and Russell, an APA scholarship to Liu and an Endeavour IPRS to Grundgeiger. Grundgeiger’s project is supported by the National Health and Medical Research Council (NHMRC) Center of Research Excellence in Patient Safety. The Center is funded by the Australian Council for Safety and Quality in Health Care (the Safety and Quality Council) and is designated as a NHMRC Center of Research Excellence. The Safety and Quality Council is a joint initiative of the Australian, State and Territory Governments.

812

REFERENCES Brixey, J. J., Robinson, D. J., Johnson, C. W., Johnson, T. R., Turley, J. P., & Zhang, J. (2007). A concept analysis of the phenomenon interruption. Advances in Nursing Science, 30(1), E26-42. Cherry, K. E., Martin, R. C., Simmons-D'Gerolamo, S. S., Pinkston, J. B., Griffing, A., & Gouvier, D. (2001). Prospective remembering in younger and older adults: Role of the prospective cue. Memory, 9(3), 177-193. Chisholm, C. D., Collison, E. K., Nelson, D. R., & Cordell, W. H. (2000). Emergency Department Workplace Interruptions: Are Emergency Physicians "Interrupt-driven" and "Multitasking"? Academic Emergency Medicine, 7(11), 1239-1243. Dieckmann, P., Reddersen, S., Wehner, T., & Rall, M. (2006). Prospective memory failures as an unexplored threat to patient safety: results from a pilot study using patient simulators to investigate the missed execution of intentions. Ergonomics, 49(5-6), 526-543. Dismukes, R. K. (2008). Prospective Memory in Aviation and Everyday Settings. In M. Kliegel, M. A. McDaniel & G. O. Einstein (Eds.), Prospective memory: cognitive, neuroscience, developmental, and applied perspectives (pp. 411-431). New York: Lawrence Erlbaum Associates. Dismukes, R. K., Young, G. E., & Sumwalt, R. L. (1998). Cockpit interruptions and distractions: Effective management requires a careful balancing act. ASRS Directline, 10, 4-9. Dodhia, R. M., & Dismukes, R. K. (2005). A task interrupted is a prospective memory task. Paper presented at the Meeting of the Society for Applied Research in Memory and Cognition. Drews, F. A. (2006). The frequency and impact of task interruptions on patient safety in the ICU. Paper presented at the 16th World Congress on Ergonomics. Dzik, W. H. (2007). New technology for transfusion safety. British Journal of Haematology, 136(2), 181-190. Einstein, G. O., McDaniel, M. A., Williford, C. L., Pagan, J. L., & Dismukes, R. K. (2003). Forgetting of intentions in demanding situations is rapid. Journal of Experimental Psychology Applied, 9(3), 147-162. Grundgeiger, T., & Sanderson, P. (2008). Interruptions in healthcare: Theoretical views. Manuscript submitted for publication. Linden, J. V., Wagner, K., Voytovich, A. E., & Sheehan, J. (2000). Transfusion errors in New York State: an analysis of 10 years' experience. Transfusion, 40(10), 1207-1213. Logie, R., Maylor, E., Della Sala, S., & Smith, G. (2004). Working memory in event- and time-based prospective memory tasks: Effects of secondary demand and age. European Journal of Cognitive Psychology, 16(3), 441456. Marsh, R. L., & Hicks, J. L. (1998). Event-based prospective memory and executive control of working memory. Journal of Experimental Psychology-Learning Memory and Cognition, 24(2), 336-349. McDaniel, M. A., & Einstein, G. O. (2000). Strategic and automatic processes in prospective memory retrieval: A multiprocess framework. Applied Cognitive Psychology, 14, 127-144. McDaniel, M. A., & Einstein, G. O. (2007). Prospective Memory: An overview and synthesis of an emerging field. Thousand Oaks, Calif.: Sage Publications. McDaniel, M. A., Einstein, G. O., Graham, T., & Rall, E. (2004). Delaying execution of intentions: Overcoming the costs of interruptions. Applied Cognitive Psychology, 18(5), 533-547. Monk, C. A., Boehm-Davis, D. A., & Trafton, J. G. (2004). Recovering from interruptions: Implications for driver distraction research. Human Factors, 46(4), 650-663. Monsell, S. (2003). Task switching. Trends in Cognitive Sciences, 7(3), 134140. Nowinski, J., & Dismukes, R. K. (2005). Effects of ongoing task context and target typicality on prospective memory performance: The importance of associative cueing. Memory, 13(6), 649-657. Parker, J., & Coiera, E. (2000). Improving clinical communication: A view from psychology. Journal of the American Medical Informatics Association, 7(5), 453-461. Spencer, R., Coiera, E., & Logan, P. (2004). Variation in communication loads on clinical staff in the emergency department. Annals of Emergency Medicine, 44(3), 268-273. Turner, C. L., Casbard, A. C., & Murphy, M. F. (2003). Barcode technology: its role in increasing the safety of blood transfusion. Transfusion, 43(9), 1200-1209.