|Home | About | Journals | Submit | Contact Us | Français|
Healthcare settings can be hectic, demanding, time-constrained environments. Within these environments, health care professionals (HCP) are expected to perform tasks that often require their undivided attention. However, HCPs are frequently interrupted, which can distract their attention and add to the complexity of their work. That said, not all interruptions are bad; many interruptions are essential to the patient care process and provide HCPs with necessary information. This paper systematically reviews the peer-reviewed literature on interruptions in healthcare settings to determine the state of the science and to identify gaps. It then provides a complex sociotechnical systems approach to understanding interruptions in healthcare.
Healthcare settings can be hectic, demanding, time-constrained environments. Within these environments, health care professionals (HCPs) perform complex cognitive tasks[1, 2] that often require their undivided attention. Interruptions such as phone calls, pages, other HCPs’ requests, equipment failures, alarms, patients, and patients’ families disrupt HCPs throughout their day and potentially interfere with their already demanding workload.
The Institute of Medicine’s 2000 report, To Err Is Human, identified interruptions as a likely contributing factor to medical errors, and literature has reported that interruptions can be disruptive and can often hinder HCPs from successfully completing their tasks.[4–6] However, some interruptions are essential to the patient care process and provide HCPs with necessary information (e.g., a patient’s monitor alarming due to abnormal vital signs).
Interruptions have implications for safe and high-quality healthcare delivery, thus this paper systematically reviews the peer-reviewed literature on interruptions in healthcare settings to determine the state of the science and to identify gaps. It then discusses the implications of the reviewed literature and suggests directions for future research to develop a better understanding of interruptions in healthcare.
The inclusion criteria were 1) the article’s domain was healthcare; 2) one of the main focuses of the article was interruptions or the concept of shifting attention away from a primary task (related terms were disruptions, distractions, breaks-in-task, etc); 3) the article was published in a peer-reviewed journal; 4) the article presented empirical data; 5) the article was published prior to 1 August 2008; and 6) the article was available in the English language. Articles were excluded if they only contained conceptual or theoretical discussions of interruptions.
The online databases PubMed and Web of Knowledge–CrossSearch were searched (the latter simultaneously searched under Arts and Humanities Citation Index, Social Sciences Citation Index, Science Citation Index Expanded, Biological Abstracts, MEDLINE, and Zoological Record) using the following search phrases: 1) healthcare* AND interrupt*; 2) health care* AND interrupt* (which was subsequently disregarded because it provided too many irrelevant articles); 3) nurse* AND interrupt*; and 4) physician* AND interrupt*. These searches yielded a total of 2,387 articles. Colleagues were also requested to provide any relevant papers that might meet the inclusion criteria. Fourteen papers met the inclusion criteria. A search of their references yielded 19 additional articles meeting the inclusion criteria.
Next, a cited reference search in Web of Knowledge was performed on the 31 articles. This cited reference search produced two additional papers. Three papers focusing on conversational interruptions were eliminated because they focused on how individuals gained power over one another by studying the interruptive and overlapping speech patterns of physician-patient consultations. This focus was too narrow for the scope of this paper. A total of 32 papers were included in this review.
Table 1 summarizes the 32 articles and their main results. Table 2 provides methodological characteristics of the studies. These factors were chosen to highlight the similarities and differences among the studies. Table 3 presents sources of interruptions in the studies that provided those data. The source of an interruption is defined as the agent or event creating the interruption. The use of the phrase “cause of interruptions “was intentionally avoided because of ambiguity in the meaning of “cause” in interruption research; it is unclear that if a pager interrupts a nurse, whether the cause is the pager, the person who created the page or the event that led the person to create the page. Cleary these are all part of a causal chain. However, the page was the proximal source of the interruption.
Five studies reported the primary tasks their participants were performing when interrupted.[1, 7–10] It was reported that direct patient care tasks and/or patient interventions were interrupted 19%, 45%, 47% and 62% of the time.
Six of the 32 studies looked at their participants’ actions or responses to the interruptions.[8, 9, 11–14] Harvey et al. reported their participants’ responses to their pagers–51% of the time pager interruptions lead to new orders being written and 18% resulted in no action. Friedman et al. focused on travel distance and showed that 87.5% of interruptions required little or no movement while 9.75% of interruptions required three meters or more of travel. The other four studies[8, 12–14] reported whether or not their participants resumed the primary task after an interruption. For example, Westbrook et al. found that 74% of primary tasks that were interrupted were resumed within the observation period of one hour; and Brixey et al. found that after being interrupted, participants generally resumed the primary task, but only after completing one to eight other tasks.
Three studies implemented interventions to try to reduce interruptions[23, 24, 26] and two were successful. Pape found that providing nurses with a medication checklist, or a checklist and vest indicating not to interrupt them, significantly reduced interruption frequency over a control group. Another study implemented a process-improvement program and posted visible signage to reduce interruptions in areas where nurses handled medications. Survey results showed that the interventions reduced interruptions (p < 0.001). The third study focused on reducing the number of uninvited patients, incoming telephone calls, and urgent house calls for office-based physicians using a variety of redesign steps. No statistically significant improvement resulted.
One study calculated the cost of interruptions and estimated that each operational failure that resulted in an interruption cost the hospital a median of $117, or roughly $95 per hour per nurse.
Seven studies examined the impact of interruptions on safety or patient outcomes.[18, 19, 25, 27, 28, 32, 34] Flynn et al. found that interruptions during drug dispensing increased the error rate by 3.42%. Sevdalis et al. found that surgical team members perceived patient-related disruptions contributed most to errors (p < 0.01). Wiegmann et al. found a linear relationship between surgical flow disruptions and errors; as the number of disruptions increased, so did the number of errors (r = .47, p < 0.05). Tucker reported that interruptions caused short delays in patient care tasks, which caused minor inconvenience and discomfort to patients. That said, Paxton et al. and Dearden et al. reported from self-report surveys that only a few patients (4–18%, respectively) had negative feelings about interruptions and Rhoades et al. found that 59% of patients were generally satisfied with their visit despite interruptions during the physician-patient encounter.
This review identified several important findings. First, it provided evidence that interruptions occur frequently in healthcare regardless of the setting. Second, it highlighted an important gap that exists in research on interruptions in healthcare: only seven studies examined outcomes related to interruptions. Third, it emphasized that interruptions in healthcare have only been studied from the viewpoint of the person being interrupted, and not the perspective of the interrupter. Fourth, few studies explicitly or implicitly examined the cognitive implications of interruptions by measuring subsequent performance, such as errors or problem identification. These cognitive implications of interruptions are at the heart of why the study of interruptions is important.
When individuals are disrupted by an interruption (as opposed to when they completely ignore a potential interruption, and therefore are not disrupted, or when they take on the interrupting task in parallel with the primary task, which would then result in dual- or multitasking), their attention is shifted from the primary task (e.g., ordering a medication) to the interrupting task (e.g., responding to an alarm, or responding to a question from a colleague). Once this shift in attention occurs, memory of the primary task begins to decay in order to “make room” for the processes required to deal with the interrupting task.[38, 39] Thus, when the primary task is resumed, it is easy (and natural) for an individual not to remember which part of the primary task was last completed.[39, 40] The amount of memory loss of the primary task depends on the characteristics of the primary task itself and of the interrupting task. Although results have varied, in general, interruptions that occur in the middle of the primary tasks, that are more similar to the primary task (i.e., require the same cognitive processes), that are longer in duration, and that are more difficult for individuals to process, are the most disruptive.[40–44]
Interruptions have also been called distraction,[19–21] break-in-task,[8, 13] and disruption. However, no matter which term is used, the issue is that when an individual’s attention is shifted away from the primary task, the likelihood of an error occurring upon return to the primary task is increased. The same results can occur even when the shift in attention is volitional and initiated by the individuals themselves, such as when a driver chooses to look down at the radio or a cell phone.[45, 46]
From the discussion of interruptions thus far it might seem that interruptions are necessarily unsafe. Many of the reviewed studies took that point of view. However, interruptions may be beneficial to the interrupter and interruptee. After all, the interrupting agent may be interrupting to accomplish a particular goal, such as providing or gathering information. Interruption research might benefit from taking a more holistic view of interruptions, that is, one that takes a systems approach to understanding the multiple goals being pursued among the agents. Brixey et al. provide a good start by conceptualizing interruptions as a system of events and agents and providing a useful set of interruption attributes. However we feel a complex sociotechnical systems approach[47–50] to thinking about interruptions may provide researchers with new insights for studying interruptions.
The source of the interruption and the goal of the interruption provide insight into the emergence and implications of interruptions. At a basic level, the source can be internal or external to the interruptee.[4, 45] These have been referred to as breaks or intrusions, respectively. Internal interruptions can essentially occur in two ways: 1) an individual decides to take a break from what he or she is doing (e.g., a nurse stops charting for a bathroom break) or 2) an individual has a thought enter his or her working memory (e.g., “Uh oh, did I forget to log out of the computer?”). They can have positive outcomes such as remembering to do something nearly forgotten or negative outcomes such as forgetting the details of the primary task. External interruptions (or intrusions) occur when an agent external to the interruptee, such as another person, an alarm or a phone, disrupts the interruptee’s workflow. External interruptions can be initiated by an external agent or by the agent him or herself (e.g., doctor asks a lab technician to call him when the labs are back for his patient).
External interruptions can occur in order to achieve a goal or in the absence of a goal. An externally goal-driven interruption is one in which the initiator of the interruption creates an interruption to achieve a goal, such as when one person interrupts another to provide information, when one person asks another person to remind them later about something, or when an alarm sounds to provide information. On the other hand, some interruptions are devoid of goals, such as when the primary task is to scan a barcode and the interruption is that the barcode cannot be read by the scanner, or when the interrupting event is to stop the primary task to look for missing information. These external, non-purposeful interruptions should be designed out of the system to the greatest extent possible.
The focus of the remainder of the discussion is on external, goal-driven interruptions because current research has focused on these (with the exception of Tucker, and Tucker and Spear who included operational failures which were external interruptions without goals). While the extant research in healthcare has predominantly treated them as unsafe events, they are in fact much more complicated and nuanced. From a complex sociotechnical systems perspective, these external, goal-driven or purposeful interruptions may be necessary for the successful function of one or more parts of the healthcare system (e.g. interruptions for the purpose of preventing a medication overdose or interruptions for the purpose of obtaining time-critical, important information). Because of that, interruptions have emerged, been required, been designed into, and been encouraged throughout healthcare delivery because they can contribute to system safety and resilience. Vital monitor alerts are designed such that they have the capability to interrupt and refocus attention on patient conditions. The same is true for other technologies, such as pagers, which through vibration or sound are designed to have an interrupting capability. Healthcare professionals and healthcare staff are also encouraged to interrupt each other if the interrupter or interruptee requires time-critical information. That said, those same interruptions are potentially harmful. Using an interrupter-interruptee paradigm, external, goal-directed interruptions can result in many different outcomes. Table 4 shows several scenarios that can emerge from an interruption.
This table, although not all-inclusive, shows the complexity of interruptions and the simultaneous implications faced by interacting system elements during interruptions. The examples provided show how some interruptions can increase the risk of an error occurring, while others can be quite beneficial and in some cases can even prevent errors from occurring. This complex sociotechnical systems framing of interruptions also has implications for interpreting the results of the reviewed studies.
The results of this review indicate that interruptions are common occurrences in a variety of healthcare environments. However, the high frequency of interruptions is not unique to healthcare; the same is true in aviation and driving.[52–54] Also, the interruptions studied (see Table 2 and Table 3) were frequently information sharing events involving interruptions by other clinicians or patients, whether mediated by technology, such as pagers, or not. At least one study demonstrated that these interruptions could improve performance by correcting medication orders. Together, the high frequency of interruptions coupled with information content may simply be indicative of the high need for constant communication and coordination in healthcare. This should be expected; healthcare delivery, like all complex sociotechnical systems, relies on communication and coordination to maintain system performance. As such, the high frequency of interruptions need not necessarily be worrisome.
To that end, it is also not clear that interventions to eliminate interruptions are a good idea. Trying to eliminate all interruptions is unwise, because it may be either unfeasible or unsafe. On the other hand, there may be situations, such as during high risk procedures, when limiting interruptions may be warranted. This similarly calls into question what outcomes to measure with regard to interruptions. We agree with Tucker [32, 33] that non-purposeful interruptions, or operational failures that interrupt, are appropriate to measure as costs. We also agree that errors are appropriate outcomes. [19, 28, 34] However, goal-driven interruptions need to be studied as having potential performance benefits that may result in improved situation awareness,[55–58] appropriately refocused attention, problem identification, collaboration, communication, and forecasting / planning.
Admittedly, it can be difficult to study associations between interruptions and outcomes in healthcare field studies. Determining the effects of interruptions on the interrupter, interruptee, and the patient is especially difficult because some of what happens is not observable, but rather manifests as short-term cognitive effects (e.g., break in attention, increase in stress or cognitive workload, obtaining wrong information, etc.). However, some have used observations to examine interruptions in healthcare [8, 9, 11–14] and we believe this approach deserves further attention. Observations can be used to identify performance improvements or decrements, and can be complimented with other cognitive field research techniques[59, 60] to gain deeper insights into interruptions.
Future research must go into more depth to understand interruptions in light of the complexity of healthcare. Many interruptions may be necessary for safe, high quality care. However, there may be times, especially during tasks that require undivided attention, that interruptions should be proactively limited to only those that are clearly needed. Taking a complex sociotechnical systems approach will help researchers view interruptions more holistically and will result in more comprehensive studies that take into account the complexity of interruptions and the many variables in healthcare settings. This should lead to a deeper understanding of interruptions, and improved design of systems to support HCPs as they deal with interruptions in the course of their normal (that is to say, hectic) work.
The first author would like to thank Sam Alper, PhD, and Richard Holden, PhD, for their feedback and Colin Shapiro, BS, for his editorial assistance. Funding for this research was provided, in part, by research grants NLM 1R01LM008923, AHRQ 1R01HS013610, and Robert Woods Johnson Foundation Grant #61148.