Safer Surgery 344 it to be labelled a PM situation, there would not be an additional reminder about the intention. The execution action, putting the intention into practice, is considered a separate phase in the model. It can contain specic errors, e.g., related to the ability to perform the intended task or meta-cognitive abilities to monitor the success of the intention implementation. Finally, the acting person must note whether or not the intention was fullled and, if yes, also to what extent. On occasion, intentions are only partially fullled but seem to the acting person as though the task has already been completed. Depending on the phase of the cycle in which this disturbance occurs, different error types would be responsible for the failed execution of the intention as error form. Future research should aim to better understand the error types. We will use the framework introduced here to more closely analyse the examples presented at the start of the chapter. This will necessarily involve some assumptions and even speculations – as PM failures are related to intentions, which can not be observed directly. Analysis of the Clinical Examples within the PM Framework Example 1 – Treatment in Time The situation where the anaesthesia team must remember the antibiotic can be classied as an event-based PM task. This case is particularly difcult because the intention should be fullled before another event (skin cut) takes place. Therefore, the anaesthesia crew would need to create some kind of cue that could help them remember to provide the antibiotic before the skin cut. The skin cut is not an appropriate cue to trigger the action, as it occurs too late. The newly created cue could be ‘internal’: the crew could try to anticipate in detail the moment in which they want to remember to administer the drug. They could thus make their internal intention stronger, so that they would be more alert about the beginning of the window of opportunity (Gollwitzer and Schaal 1998, Schweiger Gallo and Gollwitzer 2007). Another strategy which is more ‘external’ and is already partially implemented according to the task description. The team can create some kind of external reminder (here the bottle that is taped to the chart). They could also use a combination of internal and external strategies when trying to remember to execute the intention: they could anticipate the moment in which they take the bottle off the chart and try to use this act as a cue to perform the intention of administering the drug as well. Potentially, it might be easy to forget to actually administer the drug if the bottle is taken away from the chart but not used immediately. Just handling the bottle carries the risk of psychologically ‘nishing’ the task (pars-pro-toto error type). 4. 5. 6. Remembering To Do Things Later and Resuming Interrupted Tasks 345 Example 2 – Recurrent Measurements As opposed to the rst example, the intention in this example (anaesthetist checking blood glucose) needs to be remembered by a single person, not a team. This has advantages and disadvantages. Research shows that asking other people to help remember might, at times, not be effective. One underlying mechanism might be the shared responsibility for forming and retaining the intention leads to lowered cognitive energy for all involved (Shaefer and Laing 2000). This example further involves a recurring time-based intention: the blood glucose needs to be checked at regular intervals. The anaesthetist relies on the alarm clock which acts as an external reminder. However, the effectiveness of this reminder might be compromised by the concurrent tasks during the operation – which might be even more obviously demanding (the negative effects of low blood glucose are not immediately obvious in anaesthetized patients). So, when the timer clock goes off and the anaesthetist is just performing another task, a new PM challenge is posed – the intended action is again interrupted or cannot even be started. The reminder does not have a memory for the execution of the task in the sense of the PM Phase 5, described above. Using such external reminders might have the drawback that their potential to help might not be usable in a specic moment and, in addition, using them requires resources in itself, thus adding to the complexity of the situation. Example 3 – Dynamic Change of Plans Unlike the rst two examples, Example 3 (remembering to perform an appendectomy) involves changes of plans during the actions – a common situation in medicine as a dynamic eld. During the demanding operation, the decision is made dynamically to extend the treatment by including an additional treatment after the originally planned steps are fullled. This situation requires either modication of the original intention or the formation of a new intention – a differentiation that might be difcult to make in practice but could be relevant for the strengths of the intention and thus the likelihood of the execution of this activity-based intention. In either case, due to the demanding operation, the forming of the intention might be weak and with little energy, increasing the likelihood of ‘forgetting’ it. In addition, this situation is error prone because of its ‘pars-pro-toto’ character (cf. (Dieckmann et al. 2006). When the originally planned part of the operation is nished, the fth phase of the PM cycle becomes very relevant: the evaluation of the results. As the original intention is fullled completely, it might be easy to mistakenly see the whole, now enlarged intention as fullled, thus missing the newly formed part. Fullling a part of the intentions might feel as if they were being fullled completely. Safer Surgery 346 Results from a Pilot Study After the theoretical considerations during the analyses of the examples, we report empirical data from a pilot study about PM in acute care settings; it was meant as a rst approximation to PM in acute medical care – a so far neglected connection (Dieckmann et al. 2006). Simulator Study In one study we used a patient simulator to investigate situations prone to PM failures and the effects of the missed execution of intentions (Dieckmann et al. 2006). In summary, we created ten different simulation scenarios for acute care settings. Each scenario contained between one and three PM situations, which differed in how closely they were related to the context of the scenario. The match to the context was the rst independent variable. Participants were medical students who were close to or within their nal year, and worked through the scenarios in pairs. The dependent variable was whether the team (i.e., one or both members) executed the intentions during the scenario. After each scenario, each student rated the importance of the intention and several other elements of the scenario. We used this rating as a second independent variable. We analysed the number and percentage of executed tasks in dependency of type and importance. No signicant differences were found with regard to either factor. Subjectively, important tasks tended to be executed more often than unimportant tasks and the closer the task matched the overall goal of the situation (learning), the more often it tended to be executed. Despite the lack of signicant results, since it was possible to trigger PM errors in these scenarios, we believe that the simulator is an appropriate research setting for studies on prospective memory. Questionnaire Study In another part of the study, we assessed the attitudes and experiences of elements of prospective memory in anaesthetists and intensive care physicians from diverse healthcare institutions in Germany. Based on the results of a workshop with eight anaesthetists of different experience levels, we developed a questionnaire, which described 24 PM-prone situations. The questionnaire briey explained the concept of PM. Respondents rated the relevance of PM for their practice and estimated how often they had encountered each situation during the preceding month (see Table 20.1, mean count) and how often they failed to execute the specied intention. We computed percentages from these counts for every participant and averaged these percentages across situations (Table 20.1, mean percentage). Questionnaires were sent to 680 anaesthesiologists and intensive care physicians in German healthcare institutions of different sizes and types (university hospitals, community hospitals, private practice) from all 16 federal Remembering To Do Things Later and Resuming Interrupted Tasks 347 states in Germany. We sent every institution a letter that described the study, accompanied by the appropriate number of blank questionnaires and stamped return envelopes for the relevant staff. There were 112 questionnaires returned (16.4 percent response rate). The mean duration of job experience of the respondents was 10 ± 8 years. A total of 69 percent of respondents rated the inuence of PM failures on the development of critical incidents as ‘very big’, ‘big’ or ‘somewhat big’ (mean and standard deviation: 4.03 ± 1.01; n=107), while 59 percent estimated the threat of such failures to patient safety as ‘somewhat big’ or greater (3.78 ± .98; n=107). ‘No impact at all’ was not checked by any of the respondents. Table 20.1 shows the experiences of situations prone to failure of PM and the respondents’ estimate of how often they are connected with missed executions of the intention. Situation Mean count ± SD experienced Mean % ± SD forgotten Number of valid estimations (n) A medical intervention is interrupted by a more important task and you want to correctly resume the interrupted treatment 36.3 ± 41.7 11.3 ± 9.0 98 You need to check whether another person executed the intention 28.9 ± 43.8 16.0 ± 17.0 61 You are disturbed by another person while you are about to execute an intention 25.7 ± 58.8 18.8 ± 19.5 83 You are disturbed by an alarm while you are about to execute an intention 24.7 ± 66.9 11.6 ± 14.6 78 You are not certain whether you already executed the intention or not 15.1 ± 36.6 15.5 ± 20.4 100 The intention concerns an action that is unimportant for you 10.5 ± 15.4 40.9 ± 29.7 82 You need to postpone a routine action and perform it later 10.2 ± 14.5 16.4 ± 19.4 81 You executed a very similar but not the intended action 9.3 ± 18.2 15.4 ± 20.0 86 Table 20.1 Selected estimations of frequency of prospective memory based situations in medicine (mean count), error proneness of situations (mean %), and valid number of estimations for each situation (n) Safer Surgery 348 Because of the low response rate and an unrepresentative sampling, generalizations need to be drawn carefully from this questionnaire. Nevertheless, we think the results can generate hypotheses and emphasize the need for more research in this regard. One of the formed hypotheses is that interruptions appear to impact PM in different ways. Interruptions by persons are estimated to be more disruptive than those by alarms. This difference might be due to the fact that interruptions by other people tend to be stronger than those by alarms. Where it is possible to take a short time to nish a thought, task or part of a task when the alarm goes off, this is often more difcult with interruptions by other persons. Observations in the intensive care setting support this assumption: interruptions by other persons were followed by immediate reactions by those who were interrupted. These results t also with the tendency for face-to-face communication in medical settings, described above. It is a question of further research to discover the underlying error types for this phenomenon. Discussion of Research Related to Prospective Memory and its Failures in Acute Care Settings There is a large body of basic psychological research on prospective memory, relying to a large extent on highly controlled studies in laboratories. While this type of research allows for looking more closely at the different error types underlying the missed execution of intentions, it seems difcult at times to use the results from those studies to improve patient care because it is difcult to directly apply the results from articial research situations. The discussions showed that the theoretical PM framework is helpful to have a closer look at PM and how it relates to patient safety. However, there are challenges when trying to strictly apply the denitions within the PM framework to the dynamic settings of eld research in Situation Mean count ± SD experienced Mean % ± SD forgotten Number of valid estimations (n) Y ou know that you wanted to do something but cannot remember what it was 8.8 ± 11.4 36.2 ± 31.7 88 The intention concerns an action that is unpleasant for you 7.7 ± 9.0 8.1 ± 16.7 82 Problems using a device cause delays in executing an intention 6.9 ± 12.1 18.4 ± 20.8 91 You need to postpone an exceptional task and do it later 7.2 ± 11.6 16.0 ± 19.9 82 Table 20.1 Concluded Remembering To Do Things Later and Resuming Interrupted Tasks 349 the acute care settings. Intentions, as inner processes, are not directly observable but need to be inferred. In addition, they are not always conscious in all parts. It seems necessary to also conduct more applied studies under more ecologically valid conditions and to progress the knowledge by looking at prospective memory from both the applied and the basic science perspective. As with all studies of errors, it is also a challenge in PM research, to dene the ‘unit of analysis’, i.e., what is to be considered an error and what is correct performance (Wehner et al. forthcoming). In most PM studies, ours included, the investigation relies on the assumption that participants form an intention from instructions. If the instruction is never transformed into an intention, it would be wrong to assume an error, if no related task is executed. For this reason, we think it is benecial to have a closer look at interruptions. It seems reasonable that tasks which are not yet completed are related to the intention of resuming and nishing it. In some cases there might also be the conscious decision not to pursue the intention any more, so that supplementary interviews and other data would still be helpful. In this regard, we acknowledge the smart methodological aspects of the rst group studying PM-related issues in the 1920s around Kurt Lewin (Birenbaum 1930, Ovsiankina 1928, Lewin 1926, Zeigarnik 1927), who investigated, using interviews, how easily participants could remember intentions that were fullled as opposed to those that were not fullled. Another challenge for research in PM and its failures stem from the fact that, in many cases, the window of opportunity is articially strictly dened under study conditions, possibly decreasing the ecological validity of the results. In many clinical situations, the borders of the window of opportunity are not dened sharply, but more in terms of transitions (see Example 2 above). In many (certainly not all) cases it is not of signicance whether a certain task is fullled in a specic time frame. The sequence of events might be more important in many cases (e.g., Example 1) than the absolute time and in some cases there might be discussions about the existence of a window of opportunity in the rst place (i.e., in those cases where it is difcult to dene the gold standard of medical practice). There is much debate about the memory component of PM. As discussed above, the phenomenon of executing or not executing a certain intention involves many different processes during the phases of a PM cycle. In order to understand the underlying error mechanisms (error types), we would need a very ne-grained analysis – which is often not possible in applied studies. In this regard it might be better to (a) talk about intentions and their execution instead of PM in applied settings and (b) to integrate basic, laboratory-based research with more applied studies, with the aim of validating the ndings. Implications for Clinical Practice For dealing with each threat to patient safety, it is important to work through at least three steps. The rst step involves the removal of the threat from the Safer Surgery 350 system, reducing the number of PM-prone situations. One could analyse and optimize communication and interaction patterns, potentially decreasing the number of interruptions and the amount of time spent waiting for team members. This would be a question of work system design. In a second step, which should only be taken if the rst step is not possible, one could try to build in safety measurements, to prevent errors in PM-prone situations. This can be external reminder systems, checklists and other (technical) devices that might help in optimizing PM performance. Thirdly, and this step should only be taken if the rst two steps have failed, one could train people to handle the remaining PM situations better. The priority should be to systematically design weak points out of the system instead of training people to handle them. Such training could involve strategies that can help remembering and also showing the sometimes counter-intuitive effects of some of those strategies (e.g., asking someone to help remember, which might decrease the likelihood of executing the intention as discussed above). Here we do need deeper analyses, trying to understand how people redene tasks into primary, secondary, etc. and decide what is the disturbance and what is the disturbed element. Work system design can tailor the interplay of humans, technology and organization in a PM friendly and PM error friendly way (Wehner 1992, von Weizsäcker and von Weizsäcker 1984). Depending on the perspective of this interplay and the underlying error types, we need different ways for improvement. Those ways would necessarily involve cognitive, but also social and organizational, elements. References Beyea, S.C. (2007) Distractions, interruptions, and patient safety. Aorn J 86, 109– 12. Birenbaum, G. (1930) Das Vergessen einer Vornahme [The forgetting of an intention]. Psychologische Forschung 13, 218–84. Brandimonte, M., Einstein, G.O. and Mcdaniel, M. (eds) (1996) Prospective Memory: Theory and Applications. Mahwah, NJ: Erlbaum. Chisholm, C.D., Collison, E.K., Nelson, D.R. and Cordell, W.H. (2000) Emergency department workplace interruptions: Are emergency physicians ‘interrupt- driven’ and ‘multitasking’? Academic Emergency Medicine 7, 1239–43. Dieckmann, P., Reddersen, S., Wehner, T. and 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, 526–43. Dismukes, K. and Nowinski, J. (2006) Prospective memory, concurrent task management, and pilot error. In A. Kramer and D. Wiegmann (eds) Attention: From Theory to Practice. New York: Oxford. Remembering To Do Things Later and Resuming Interrupted Tasks 351 Dyrløv Madsen, M. and Schou, B. (2008) Forstyrrelser, afbrud og travlhed: Resultater af en pilotundersøgelse i et onkologisk ambulatorium [Interruptions, disruptions, and workload: Results of a pilot project in an oncological ambulatory department]. Unpublished internal work report, Danish Institute for Medical Simulation, Herlev Hospital, Herlev Denmark. Dyrløv Madsen, M., Boje Andersen, H. and Itoh, K. (2007) Assessing safety culture and climate in health care. In P. Carayon (ed.) Handbook of Human Factors and Ergonomics in Health Care and Patient Safety. Mahwah, NJ: Erlbaum. Ellis, J. (1996) Prospective memory or the realization of delayed intentions: A conceptual framework for research. In M. Brandimonte, G.O. Einstein and M. Mcdaniel (eds) Prospective Memory: Theory and Applications. Mahwah, NJ: Erlbaum. Gaba, D.M., Howard, S.K., Fish, K.J., Smith, B.E. and Sowb, Y.A. (2001) Simulation based training in anesthesia crisis resource management (ACRM): A decade of experience. Simulation and Gaming 32, 175–93. Gollwitzer, P.M. and Schaal, B. (1998) Metacognition in action: The importance of implementation intentions. Pers Soc Psychol Rev 2, 124–36. Goschke, T. and Kuhl, J. (1993) Representation of intentions: Persisting activation in memory. Journal of Experimental Psychology: Learning, Memory and Cognition 19, 1221–6. Groopman, J. (2007) How Doctors Think. Boston: Houghton Mifin. Harris, J.E. and Wilkins, A.J. (1982) Remembering to do things: A theoretical framework and an illustrative experiment. Human Learning 1, 123–6. Healey, A.N., Sevdalis, N. and Vincent, C.A. (2006) Measuring intra-operative interference from distraction and interruption observed in the operating theatre. Ergonomics 49, 589–604. Itoh, K., Boje Andersen, H. and Dyrløv Madsen, M. (2007) Safety culture in health care. In P. Carayon (ed.) Handbook of Human Factors and Ergonomics in Health Care and Patient Safety. Mahwah, NJ: Erlbaum. Jauhar, S. (2007) Intern. A Doctor’s Initiation. New York: Farrar, Straus and Giroux. Kvavilashvili, L. and Ellis, J. (1996) Varieties of intention: Some distinctions and classiactions. In M. Brandimonte, G.O. Einstein and M. Mcdaniel (eds) Prospective Memory: Theory and Applications. Mahwah, NJ: Erlbaum. Lewin, K. (1926) Vorsatz, Wille und Bedürfnis [Intention, will, and need]. Psychologische Forschung 7, 330–85. Manser, T., Thiele, K. and Wehner, T. (2003) Soziotechnische Systemanalyse im Krankenhaus – Eine Arbeispsychologische Fallstudie in der Anästhesiologie [Sociotechnical system analysis in the hospital – a work psychological case study in anaesthesiology]. In E. Ulich (ed.) Arbeitspsychologie in Krankenhaus und Arztpraxis. Arbeitsbedingungen, Belastungen, Ressourcen. Bern: Huber. Ovsiankina, M. (1928) Die Wiederaufnahme unterbrochener Handlungen [The resumption of interrupted tasks]. Psychologische Forschung 11, 302–79. Safer Surgery 352 Parker, J. and Coiera, E. (2000) Improving clinical communication: A view from psychology. J Am Med Inform Assoc 7, 453–61. Schaefer, E.G. and Laing, M.L. (2000) ‘Please Remind Me …’: The role of others in prospective remembering. Applied Cognitive Psychology 14, S99–S114. Schweiger Gallo, I. and Gollwitzer, P.M. (2007) Implementation intentions: A look back at fteen years of progress. Psicothema 19, 37–42. US Pharmacopeia (2000) Summary of the 1999 Information Submitted to MedMARx - A National Database for Hospital Medication Error Reporting. Available at: <http://www.usp.org/reporting/medmarx/1999/nal.pdf>. Rockville: US Pharmacopeia. Von Weizsäcker, C. and Von Weizsäcker, U. (1984) Fehlerfreundlichkeit [error friendliness]. In K. Kornwachs (ed.) Offenheit – Zeitlichkeit – Komplexität: Zur Theorie der offenen Systeme [Openness – Timeliness – Complexity: On the Theory of Open Systems]. Frankfurt: Campus. Wehner, T. (1992) Sicherheit als Fehlerfreundlichkeit. Arbeits- und Sozialpsychologische Befunde für eine kritische Technikbewertung [Safety as Error Friendliness. Work- and Social Psychological Findings for a Critical Assessment of Technology]. Opladen: Westdeutscher Verlag. Wehner, T., Mehl, K. and Dieckmann, P. (forthcoming) Fehlhandlungen und Prävention [errors and prevention]. In U. Kleinbeck (ed.) Enzyklopädie der Psychologie. Band Ingenieurpsychologie. Zeigarnik, B. (1927) Über das Behalten von erledigten und unerledigten Handlungen [on the retention of executed and not executed tasks]. Psychologische Forschung, 9, 1–85. Weimer, H. (1925) Psychologie der Fehler [Psychology of the errors]. Leipzig: Klinkhardt. Chapter 21 Surgical Decision-Making: A Multimodal Approach Nick Sevdalis, Rosamond Jacklin and Charles Vincent Introduction Surgeons, in conjunction with their patients and colleagues, have to make risky and irreversible choices. The ability to make clinical decisions, both within and outside the operating theatre, is a key feature of surgical expertise, which has been highlighted in attempts to dene the essential competencies of a surgeon. For example, the CanMEDS model of clinical competency (Frank and Langer 2003), asserts that ‘the role of medical expert/clinical decision-maker is central’ (Frank et al. 1996, p. 4), and includes elements such as diagnostic reasoning, clinical judgement and clinical decision-making (Frank 2005). The CanMEDS model was inuential in the development of the Intercollegiate Surgical Curriculum Project, which forms the framework for the postgraduate training of surgeons in the UK (Canter and Kelly 2007). This chapter discusses surgical decision-making in both a clinical and research context and introduces a multimodal approach to measuring and assessing surgical decision-making. In what follows, we begin by outlining the role of decision- making within the ‘systems approach’ to surgical safety. Next, we describe some important features of surgical decision-making in its clinical context. Thirdly, we provide a brief overview of some of the methods available for studying surgical decision-making. Fourthly, we introduce the multimodal approach that we have taken to investigate, measure and train surgical decision-making in our research group and the results we have achieved so far. We conclude with a discussion of different research approaches, including limitations and implications for research in and assessment of surgical decision-making. (i) The ‘Systems Approach’ to Surgical Safety In recent years, there have been shifts in the understanding of surgical performance and surgical outcomes. Traditionally, surgical performance and outcomes were understood as a function of the patient’s risk factors (i.e., the severity of the disease and existing co-morbidities). In the past ten years or so, with the surge in the use of surgical simulators and the associated development of the evidence . as fullled, thus missing the newly formed part. Fullling a part of the intentions might feel as if they were being fullled completely. Safer Surgery 346 Results from a Pilot Study After. Safer Surgery 344 it to be labelled a PM situation, there would not be an additional reminder about. error proneness of situations (mean %), and valid number of estimations for each situation (n) Safer Surgery 348 Because of the low response rate and an unrepresentative sampling, generalizations