Article Text

Download PDFPDF

Differentiating between detrimental and beneficial interruptions: a mixed-methods study
  1. Robert A Myers1,
  2. Mary C McCarthy2,3,
  3. Amelia Whitlatch3,
  4. Pratik J Parikh1,2
  1. 1Department of Biomedical, Industrial and Human Factors Engineering, Wright State University, Dayton, Ohio, USA
  2. 2Department of Surgery, Wright State University, Dayton, Ohio, USA
  3. 3Miami Valley Hospital, Dayton, Ohio, USA
  1. Correspondence to Dr Pratik J Parikh, Department of Biomedical, Industrial and Human Factors Engineering, Wright State University, 207 Russ Engineering Center, 3640 Colonel Glenn Hwy, Dayton, OH 45435, USA; pratik.parikh{at}wright.edu

Abstract

Introduction Efforts to understand interruptions now span much of the last decade and a half. Often thought to negatively impact patient safety, some now acknowledge that interruptions may be beneficial and actually necessary for safety and high quality care. This study seeks a framework for differentiating between interruptions that are detrimental and those that are beneficial.

Methods A mixed-methods approach at a US Level 1 trauma centre included direct observation of 13 registered nurses (RNs), survey of 47 RNs, retrospective observation of hands-free communication devices, and modelling of observed interruptions to key performance measures.

Results On average, RNs were interrupted every 11 min, with 20.3% of their workload triggered by interruptions. While 85% of RNs agreed that interruptions place their patients at risk, only 21% agreed that all should be eliminated. During one 90-min period, 18 original events spawned 68 interruptions, 50 of these repeat messages. A statistical model, with patient measures of time and comfort, revealed that alarms and call lights returning RN's attention to the patient outside the patient room are beneficial, while interruptions in the patient room are generally detrimental. Triangulating the results, we present an emerging framework for differentiating between beneficial and detrimental interruptions based on the impact of interruptions on the RN's steady treatment and attention to the patient.

Conclusions A mixed-methods approach can help distinguish between detrimental and beneficial interruptions. While interruptions breaking the delivery of steady treatment and attention to the patient are detrimental, those returning the RN's focus to the patient, as well as those supporting patient-clinician and clinician-clinician communications are beneficial. This insight may be helpful to healthcare delivery teams tasked with improving interruption-laden processes.

  • Interruptions
  • Patient-centred care
  • Nurses
  • Quality measurement
  • Lean management

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Efforts to understand interruptions and their influence on patient safety and clinician workflow now span much of the decade and a half since the Institute of Medicine's landmark To Err Is Human and join a growing body of research addressing patient safety and medical errors.1–3 Experts suggest that pursuing systemic factors, such as interruptions, will lead to the substitution of new reliable healthcare delivery systems for old unreliable ones, a much more valid plan for reducing errors than just blaming clinicians and urging them to try harder.4 ,5

Often thought to negatively impact patient safety by disrupting clinicians’ memory, the phenomenon is the subject of scores of articles6–8 and labelled as: interruptions,4 distractions,9 workflow interruptions,10 intrusions,11 glitches12 and flow disruptions.13 While most focus on negative aspects, others present a broader view acknowledging that some interruptions may be beneficial and actually necessary for safety and high-quality care.8 ,14 ,15 While some suggest a rather nuanced stance when discussing interruptions based on their content, timing and perception by clinicians, others link interruptions’ value to their ability to change clinicians’ behaviour to meet emerging patient needs.10 ,16–19

In spite of much research, clear evidence linking interruptions with negative medical outcomes is still lacking, maybe due to the complex nature of interruptions and their almost always having positive and negative effects.20 ,21 Based on the perception that interruptions are generally detrimental, some have carried out improvement projects to reduce interruptions, but with the overall benefit of reducing interruptions still unclear and raising the question of possible unintended consequences.8

Efforts to categorise interruptions and develop taxonomies have led to a call for additional research to comprehend the extent to which interruptions contribute to medical errors, and for rigorous methodologies to differentiate between negative and positive interruptions.19 ,22 Accordingly, this work addresses the research question: “what is an effective framework for differentiating between interruptions that are detrimental and those that are beneficial?”

Value in healthcare

A natural question when differentiating between detrimental and beneficial interruptions is ‘of value to whom’.23 When judging interruptions, should value be defined from the clinician's perspective, the patient's, the payer's or some combination? Providing clarity, the Quality Chasm identifies the experience of patients as the fundamental source of the definition of quality, with many now recognising the need to define value around the customer (patient), not the supplier (clinician, payer).5 ,24–27

While value may be defined as ‘patient health outcomes achieved per dollar spent’, unfortunately, there is no single outcome that captures the results of care, and dollars spent are often unobserved by the patient.25 ,28 Patient safety has emerged as an important facet of these outcomes, but is often linked with adherence to evidence-based guidelines instead of actual patient outcomes.25 Until accepted patient outcomes emerge and the labyrinth of cost is unravelled, surrogates for value are needed to judge health processes including interruptions. A paradigm that appears helpful when confronting this enigma of healthcare value is its bifurcation into ‘content’ (evaluated primarily by physicians) and ‘delivery’ (evaluated primarily by patients).23 This view accommodates most patients who do not feel qualified to judge technical quality, but instead assess their healthcare by other dimensions that reflect what they personally value.27 ,29

Performance measures in healthcare

From a lean methodology perspective, value to the patient would then be considered as the objective of all activities surrounding patient care, and Womack and Jones’ proposal to put the patient in the foreground with time and comfort could then be considered as reasonable surrogates of value (p.289).30–33

Time

It can be envisioned that breaks in the steady treatment of the patient create delays, which degrade the patient service measure of time. Examples include delays from clinician or tool unavailability, the need to move the patient to tools that are too large to bring to the patient, and interruptions slowing workflow. These breaks in steady treatment may manifest as waiting and delays noticeable to the patient, and in aggregate as increased length of stay. As such, time may be recognised as a value important to patients and an important factor in judging the impact of interruptions on patient care delivery.

Comfort

The Swedish Health Care Act states that “healthcare shall fulfil the patient's needs of comfort and treatment.”31 Patients rely on caregivers to be attentive and present, recognising and alleviating their physical discomfort (p.130–131).34 Comfort has been called the most basic service that hospitals offer patients and the sick person's most fundamental right (p.147).34 This right is represented well in a patient quote: “What I wanted was someone with basic human kindness who would understand the fundamental factors of fatigue, need for sleep, personal privacy, and just being left alone from time to time (p.129)”.34 Eating, drinking, eliminating, sleeping, moving, bathing and grooming are key elements in providing for patients’ physical comfort (p.129). Interruptions, either supporting or breaking steady attention in support of any of these elements, may be understood as factors positively or negatively impacting the patient's experience and their comfort.34

Methods

In this study a mixed-methods approach, with time and comfort as key measures, used three modes of qualitative and quantitative data: direct observation of interruptions experienced by registered nurses (RNs), survey of RNs, and retrospective observation of hands-free communication device (HCD) data.

Participants

This study focused on RNs working at a Midwest US Level 1 trauma centre. The centre serves nearly 3000 trauma patients each year via an emergency trauma centre, surgical intensive care unit and trauma unit (for improving and less acute patients). Thirteen RNs in the trauma unit participated voluntarily in 48 h of direct observation and 47 responded voluntarily to an online survey. The RNs were observed from all hours of the day and all 7 days of the week in an attempt to capture the non-stationary nature of interruptions, which are linked to temporal tasks (such as medication and rounding) and also the cyclical workload typical of trauma services (increased admissions during evenings and on weekends).35 The RNs were enrolled upon obtaining informed consent via a printed copy of the observation protocol and a private opportunity to verbally opt in or out of the study. Nurses were also voluntarily enrolled in the online survey, with only the data of 47 RNs fully completing the survey included for analysis.

Data collection procedures

Direct observation

We constructed an observation data form from a priori categories reported by previous researchers, in particular J J Brixey, and with free-form text fields to capture details about unanticipated observations.22 ,36 Data recorded for each interruption included level, task interrupted, a description of the interrupting event, location, source, medium and time. Levels were recorded as emergent, urgent and routine. Direct care tasks, where RNs interact directly with the patient, were distinguished from indirect care tasks, where the RN is away from the patient to obtain supplies or to get more information needed to continue direct care. The event description included reason for the interruption (task request, receive info or provide info) and whether relocation or change of task was required of the RN. Free-form fields were used to record any observed impact of the interruption and interventions used by RNs to manage interruptions. To provide context for observed interruptions, we also noted the times and task categories the RN engaged in while being observed.

After enrolling the RN, the observer (first author) shadowed the RN, noting the time when the RN changed tasks and capturing data from observed interruptions. The observer followed without verbal interaction except when first entering each patient's room during the observation session, at which time the RN would ask the patient for permission to have the observer watch the RN during their care. No patients declined to allow the observer to enter their room and no patient information was collected. For rooms with isolation protocol, the RN was observed from the doorway without entering the room.

Survey of RNs

The purpose of the 55-question survey was to capture how interruptions are viewed by RNs in the trauma centre. Topics included how interruptions impact daily workload, patient safety, and care provider stress, as well as their perceived impact on patients and their families. Additionally, questions about how and where interruptions occur and the techniques used by RNs to manage interruptions were included. Participation was voluntary with each RN receiving a link via email from their nurse manager presenting the opportunity to anonymously complete the survey. In the survey instructions, interruptions were defined as “anything that takes your attention away from a task or communication activity that you were already engaged in as part of your job”.

Hands-free communications data

Nurses in the trauma centre wear HCDs to enhance communication and responsiveness to patients (eg, Vocera). These devices provide direct voice communication capability between staff, as well as real-time delivery of notifications and alarms from medical devices. Because it is difficult to identify and document HCD messages received by the RN during direct observations, we used retrospective HCD data for the RNs from the hospital's information technology department for the periods of direct observation.

Data analysis

Direct observation

Given the identification of comfort and time as measures of service important to the patient, we retrospectively mapped the observed interruptions to these two measures using the coding scheme presented below (−1, 0, 1).

For comfort:

  • (−1) Causes a break in steady attention, and/or negatively impacts control of pain, providing for patient bodily function, and/or results in a more stressful environment for patient,

  • (0) Neutral,

  • (+1) Supports steady attention and/or control of pain, patient bodily functions, and/or results in less stressful patient environment.

For time:

  • (−1) Causes a break in steady treatment or other delay noticeable to patient or extending their length of stay,

  • (0) Neutral,

  • (+1) Supports steady treatment.

Coded spreadsheet data were imported into a statistical analysis data table (SAS JMP V.11.0.0; Cary, North Carolina, USA) where relationships between observed factors and the outcomes of time and comfort were explored.

Statistical modelling

From the observed interruptions (n=259), 65 were excluded due to either incomplete records or only observed a few times (≤5) for a particular type of interruption, providing us 194 observations in the final data set.

A single response variable was derived from the sum of the coded values for the patient measures of time (−1, 0, 1) and comfort (−1, 0, 1) for each interruption. This sum was transformed into a binary variable, assigning a value of 1 to summed values >0 (beneficial to the patient, n=112) and a value of 0 to summed values ≤0 (not beneficial, n=82).

A nominal logistic regression model was used to identify statistically significant factors for location, task interrupted, source, medium, type and relocation. Included were interaction effects suggested by the study's direct observation, survey and HCD analysis; only significant effects (α=0.05) were retained in the model. An example of such an interaction was that of phone calls (medium) received by RNs while in the patient room (location).

Results

Direct observation

On average, RNs were interrupted every 11 min (5.4/h), with 20.3% of their workload triggered by these interruptions. Nearly half of these interruptions involved the RN providing information to others, 12% receiving information and 36% involved a task request. Over 35% of observed interruptions occurred during critical direct care and medication tasks in the patient room. Overall, 34% of interruptions caused the RN to relocate, while 85% of alarms and 80% of call lights triggered relocation. No negative clinical outcomes were noticed as direct result of observed interruptions. Table 1 shows a summary of observations by occurrence (a, b, c), along with column (d) depicting the portion of time RNs spent on each category of work during an average day as context.

Table 1

Observed occurrences of interruptions by (a) location, (b) medium, (c) task, depicted in (d) is an average RN day providing context

The trauma unit was staffed with patient care technicians (PCTs), RNs and advanced practice providers. Call light notifications arrived directly via visual and audible alarms in the hall above patient room doorways and from the unit clerk in the nurses’ station via face to face, phone or HCDs. Observed RNs carried cell phones (on the hospital network) and RNs and PCTs wore HCDs. No formal policies for protecting RNs from interruptions during medication or direct care tasks were observed.

RN survey

Several themes emerged from the 55-question survey. While 85% of RNs agreed that interruptions place their patients at risk, only 21% of RNs agreed that all interruptions should be eliminated. Nurses indicated that beneficial interruptions include requests for help (from patient or clinicians) and notification of changes in medical orders or patient status (eg, vital signs, bed alarms). They also indicated that detrimental interruptions include redundant communications and those occurring in the patient room, including those via HCDs while providing direct care to the patient. Nurses also identified several temporal conflicts that place their patients at risk, such as scheduled interruptions during medication hours (eg, rounding and audits) and those that wake their patients during sleeping hours.

Only 18% of RNs reported that phone calls put their patients at risk; however, in the survey's comment field related to interruptions that place their patients at risk, RNs stated that they are ‘interrupted while providing direct care (in the patient room) by phone calls that they must leave the room to answer’, suggesting that there may be some interaction between where and how interruptions arrive. Techniques to manage interruptions included telling other care providers they were busy, waiting to answer phone or HCD calls until the current task was completed, and writing notes to self.

Hands-free communication device

RNs receive, on average, one HCD message every 3 min. Of these, 23% are repeat messages linked to device alarms and automatically sent again by the system every 60 s. Nearly 21% of HCD messages arrive within 30 s of another message, creating periods when the RN is exposed to multiple, rapidly arriving, interruptions. Shown in figure 1 is one such 90-min period when 18 original events spawned 68 interruptions, 50 of these repeat messages generated automatically by the HCD system. Twenty-eight of these repeat messages arrived while the RN was in the patient room providing direct care or medication.

Figure 1

One registered nurse, 90 min—68 interruptions, 50 repeat messages from only 18 original events.

Mapping and modelling

Figure 2 shows mapping of observed interruptions onto the customer values of comfort (x axis) and time (y axis). Interruptions falling into the upper right quadrant (+, +) are beneficial to both patient measures (time and comfort), while those in the lower left (−, −) are detrimental to both. Interruptions falling in the upper left (−, +) and lower right (+, −) quadrants have offsetting qualities, having beneficial and detrimental effects for one or more patients’ measures of comfort and time.

Figure 2

Interruptions plotted against patient measures of comfort and time. (A) In/Out of patient room, (B) Out of patient room by medium, (C) In patient room by medium and (D) Source (patient or other).

From figure 2A, interruptions occurring outside the patient room fall in the (+, +) quadrant and are more beneficial (based on the combined measures of comfort and time) compared with those occurring inside the patient room (OR 3.5, 95% CI 1.6 to 7.4). Figure 2B shows a breakdown of interruptions outside the patient room by arrival medium. This expansion reveals that alarms, call lights and cell phones may be beneficial mediums for returning the RN’s attention to the patient while outside the patient room. Similarly, figure 2C shows a breakdown of interruptions inside the patient room, revealing that cellphone calls in the patient room contribute negatively to both measures of comfort and time. Figure 2D shows interruptions arriving from patients are on average more beneficial than those from other sources (OR 5.9, 95% CI 2.0 to 17.7).

Further exploring the relation between observed interruptions and the patient values of time and comfort, we built a nominal logistic regression model (p<0.0001) to the interruption effects, such as location, task, source, medium, type and relocation, for the response variable of the sum of comfort plus time.

This model showed that interruptions occurring outside the patient room are generally beneficial (p=0.0002), as are those arriving from the patient (p=0.0003). Though alarm and cellphone interruptions were not significant by themselves, their individual interactions with location (in or out of the patient room) were both significant (p=0.0337 and p=0.0004, respectively). Similarly, though not significant alone, the interaction of call lights with the nurses’ station location revealed that call lights were effective in returning RNs’ focus to the patient while in the nurses’ station (p=0.0111).

Triangulation of methods

Triangulating the results from mapping and modelling of direct observations to the patient measures of comfort and time, RN survey and retrospective HCD data, we are able to propose an emerging framework for differentiating between beneficial and detrimental interruptions as shown in figure 3.

Figure 3

An emerging framework from triangulation of methods. HCD, hands-free communication device; RN, registered nurse.

Discussion

This study proposes a patient-centred framework to distinguish between beneficial and detrimental interruptions. Viewing interruptions from such a systems perspective provides an important basis for healthcare delivery teams tasked with improving interruption-laden processes. As shown in figure 3, of greater importance than location alone (in or out of the patient room) is whether a particular interruption returns an RN's focus to the patient or causes a break in the steady attention or treatment of a patient. Interruptions providing value to the patient (beneficial and returning focus) should be supported through process improvements making them less disruptive and establishing them as standard components of the RN's workflow. Those detrimental to patient service (breaking the steady delivery of treatment and or attention) may be labelled as waste and should be targeted for elimination via continuous improvement efforts. Even so, one must be careful what gets labelled as waste. Important intangible values related to patient comfort survive in compartments sometimes labelled as inefficiency, (eg, listening, relationship building, learning, reflection and knowledge sharing.)26 While some interruptions may be easily classified as waste from a patient value perspective, a number of these may be driven by policies or organisational and clinical practices that may not be easily changed and may instead need to be managed until cultural changes allow for their reduction.

In practice, interruptions may arrive from a second or even third patient while serving the first, creating an ‘interruption conundrum’ for the care provider.11 If a care provider pre-empts their service to the first patient to refocus on another, they risk alienating the patient already being served. Likewise, continuing to serve the first patient while ignoring the requests of other patients may alienate the others. In the coding of observed interruptions, those identified as ‘offsetting’ included such interruptions. While there appears to be no win-win strategy once an interruption conundrum occurs, some consider them to be evidence of work system failures introducing unplanned work and suggest that there may be opportunity to pre-empt the occurrence of avoidable interruptions by modifying the clinician's workflow.37 Workflow improvements to prevent interruptions may be challenging, requiring new forms of asynchronous communications between team members or even smaller teams aligned around fewer patients to reduce the frequency of interruptions. An example involves pre-empting interruptions caused by family requests for patient information. By introducing periodic clinician initiated interactions with the family, these communication events may be incorporated as a component in the RN's regular workflow and serviced between other tasks, instead of arriving as interruptions.

Comfort, the second suggested measure, tends to fall outside of traditional flow models and measures of quality, but must not be ignored in understanding value as perceived by the patient.31 Opportunities for organisations to facilitate patient comfort include: (1) controlling acute pain, (2) providing basic nursing care to support and maintain normal body functions, and (3) minimising stress in the environment to promote healing and recovery (p.120).34 Clinicians of all types may be tempted to sideline patient experience, concluding that measures such as comfort are too subjective or mood oriented. Recent research shows that patient experience is positively associated with clinical effectiveness and patient safety, and that it should be included as one of the central pillars of healthcare.38 Additionally, focusing on patients’ experiences related to comfort and time has been shown to give clinicians needed impetus to improve their personal efficiency without sacrificing quality.39

This mixed-methods approach provided corroborating results and insights that may have been missed using a single method. As shown in figure 3, only the survey revealed the beneficial nature of interruptions involving patient-clinician and clinician-clinician requests for help and notifications. Informed by the RN survey, the benefit of these communication-driven interruptions are apparent in the team-based treatment and steady attention to the patient called for by lean methods. Likewise, the detrimental aspect of repeat/redundant communications found in the study of HCD messages is better understood as we consider the impact they have on timely delivery of care and the support of patient comfort.

There are several limitations of this study that must be noted. First, the data collection was limited to a trauma unit at a single Level 1 trauma centre in the USA, which may limit generalisability of results and conclusions. Second, the technique of shadowing RNs during direct observation may have altered their behaviour and, subsequently, the collected data. Additionally, this study's perspective is of the patient as customer, evaluating interruptions based on their impact on patient values. While providing value to the patient must be healthcare's primary focus, an important dimension we did not address is the impact of interruptions on the clinician whose typical day is often filled with stressors and interruptions. Some may be a brief hindrance, while others may cause significant delays leading to decreased patient satisfaction, opportunities for error and possible deterioration of patient condition. Frequent interruptions may contribute to the physical workload and psychological stress experienced by RNs, many now working 12–13 h shifts. Such stressors may have a cumulative effect on an RN's ability to manage tasks as frequent interruptions begin to overlap without sufficient recovery time. Like icebergs, the negative impact of interruptions on what patients value most and on the ability of clinicians and their organisations to provide access to quality affordable healthcare may not be immediately visible, but may manifest as lower patient experience survey scores, suboptimal clinical outcomes and higher cost of care.

Future studies should examine positive and negative effects of interruptions on accepted measures of patient outcomes as they emerge, but should also include the impact of interruptions on RN workload and psychological stress.34 Research is also needed to understand the effects of interruptions caused by on-the-job training of clinicians, and how to minimise any negative impact on patient care.

Conclusions

While some interruptions may lead to poor patient satisfaction and outcomes, waste valuable resources, and negatively impact clinicians’ workload and stress, others may be critical to providing timely, quality and affordable care.

Using a mixed-method approach based on lean principles, we proposed a framework that could distinguish between detrimental and beneficial interruptions. While interruptions breaking the delivery of steady treatment and attention to the patient are detrimental, those returning the RN's focus to the patient, as well as those supporting patient-clinician and clinician-clinician communications are beneficial.

This insight is expected to help healthcare delivery teams tasked with improving interruption-laden processes. Interruptions providing value to the patient should become a standard component of the RN's workflow, while minimising their disruptive nature. Those detrimental to patient service should be labelled as waste and targeted for elimination.

References

View Abstract

Footnotes

  • Twitter Follow Robert Myers at @fixineer

  • Contributors RAM: Study design, literature review, data collection, modelling and analysis, and manuscript preparation. MCM: Study design, provided clinical access, process insights, manuscript editing. AW: Provided insight from RNs’ perspective during interpretation of data and manuscript editing. PJP: Study design, interpretation of statistical analysis, and manuscript preparation and editing.

  • Competing interests None declared.

  • Ethics approval Wright State University IRB, in conjunction with Miami Valley Hospital's Human Investigation and Research Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.