Você está na página 1de 13

Journal of Biomedical Informatics 59 (2016) 76–88

Contents lists available at ScienceDirect

Journal of Biomedical Informatics

journal homepage: www.elsevier.com/locate/yjbin

Characterizing the structure and content of nurse handoffs: A Sequential

Conversational Analysis approach
Joanna Abraham a,⇑, Thomas Kannampallil b, Corinne Brenner c, Karen D. Lopez d, Khalid F. Almoosa e,
Bela Patel e, Vimla L. Patel c
Department of Biomedical and Health Information Sciences, College of Applied Health Sciences, University of Illinois at Chicago, Chicago, IL, United States
Department of Family Medicine, College of Medicine, University of Illinois at Chicago, Chicago, IL, United States
Center for Cognitive Studies in Medicine and Public Health, The New York Academy of Medicine, 1216 5th Avenue, New York, NY, United States
Department of Health Systems Science, College of Nursing, University of Illinois at Chicago, 845 S. Damen Avenue, Chicago, IL 60612, United States
Department of Internal Medicine, Division of Critical Care Medicine, University of Texas Health Science Center, Houston, TX, United States

a r t i c l e i n f o a b s t r a c t

Article history: Effective communication during nurse handoffs is instrumental in ensuring safe and quality patient care.
Received 20 July 2015 Much of the prior research on nurse handoffs has utilized retrospective methods such as interviews, sur-
Revised 19 November 2015 veys and questionnaires. While extremely useful, an in-depth understanding of the structure and content
Accepted 20 November 2015
of conversations, and the inherent relationships within the content is paramount to designing effective
Available online 26 November 2015
nurse handoff interventions. In this paper, we present a methodological framework—Sequential
Conversational Analysis (SCA)—a mixed-method approach that integrates qualitative conversational anal-
ysis with quantitative sequential pattern analysis. We describe the SCA approach and provide a detailed
Conversational analysis
Sequential analysis
example as a proof of concept of its use for the analysis of nurse handoff communication in a medical
Care transitions intensive care unit. This novel approach allows us to characterize the conversational structure, clinical
Communication content, disruptions in the conversation, and the inherently phasic nature of nurse handoff communica-
Communication errors tion. The characterization of communication patterns highlights the relationships underlying the verbal
Nurse handoffs content of nurse handoffs with specific emphasis on: the interactive nature of conversation, relevance of
Intensive care role-based (incoming, outgoing) communication requirements, clinical content focus on critical patient-
Shift report related events, and discussion of pending patient management tasks. We also discuss the applicability of
the SCA approach as a method for providing in-depth understanding of the dynamics of communication
in other settings and domains.
Ó 2015 Elsevier Inc. All rights reserved.

1. Introduction [8], time constraints [9,10], interruptions and distractions

[11,12], lack of training [13], and communication bottlenecks
Patient handoffs involve the transfer of information, responsi- [14–18]. These failures have been associated with incorrect or
bility, and control between care providers and are viewed as an delayed diagnosis and treatment, prolonged morbidity, increased
ubiquitous, clinical and organizational activity [1]: ubiquitous as they patient length of stay, clinician and patient dissatisfaction, and
occur across various groups of clinicians; clinical as they serve as a increased costs [19–21].
forum for sharing patient-related information; and organizational, To promote safer handoffs, The Joint Commission (TJC) spear-
as they occur at all levels of a hospital. Nurse handoffs (also headed efforts [22–24] to standardize communication activity
referred to as ‘‘shift reports”) are the most frequent [2], averaging [15,25], leading to the development and implementation of new
approximately 2 million per year in a mid-size hospital [3,4]. Fail- or re-designed handoff strategies and processes [26,27]. These
ures during handoffs contribute to nearly 35% of sentinel events efforts have also led to standardization initiatives and the develop-
and medical errors [5,6]. These failures arise from limited structure ment of handoff tools based on templates, spreadsheets, checklists
during communication [7], multiple communication approaches and mnemonics [28,29]. While these solutions have been widely
implemented, there is still a lack of consistency in their adoption
and use [30]. It has generally been acknowledged that the stan-
⇑ Corresponding author at: 1919 West Taylor Street, Chicago, IL 60612, United dardization efforts should rely on understanding the content and
States. Tel.: +1 312 413 4623. structure of handoff communication [31]. Although TJC has
E-mail address: abrahamj@uic.edu (J. Abraham).

1532-0464/Ó 2015 Elsevier Inc. All rights reserved.
J. Abraham et al. / Journal of Biomedical Informatics 59 (2016) 76–88 77

launched such efforts to standardize the content of communication, workflow efficiency. Baldwin and McGinnis [22], using a
there is little agreement, as to ‘‘what” needs to be standardized questionnaire-based study, reported that 86% of their participants
and ‘‘how to” standardize—both in terms of clinical content, and preferred the problem-based sign-out tool as it reduced chances of
structure [21]. overtime, increased direct patient care time, and improved com-
Developing an informed understanding of the content and munication of patient data. Interviews have also been used for eval-
structure of handoff communication can have implications not uating the nurse perceptions of the quality of handoff
only for minimizing communication errors in nursing practice, communication. For example, Roberts et al. [45] used interviews
but also for developing evidence-based guidelines for handoff to illustrate the effectiveness of an SBAR (Situation, Background,
training, and the design of nurse handoff interventions. However, Assessment and Recommendation) tool for handoffs. Interviews
much of the prior research on handoffs, with some notable excep- with nurses highlighted the importance of structured communica-
tions (e.g., [32]), are ‘‘descriptive” relying on retrospective evalua- tion tools in facilitating collection and transfer of critical safety
tions, rather than developing a ‘‘deeper understanding of what data that otherwise may have been lost.
they [handoffs] are, what needs they serve, and what actually hap- As highlighted above, much of the prior research on nurse hand-
pens in them” [33]. Part of the reason for such evaluations can be off communication, both with and without tools, has relied primar-
attributed to the lack of methodological approaches for: (a) tracing ily on obtaining perspectives of the participant nurses using user-
the evolution of handoff conversations, (b) investigating the inher- based surveys, questionnaires, and interviews [26,33]. While these
ent relationships within these conversations, and (c) identifying studies are useful in characterizing the nature and perception of
causes of disruptions during conversations. In this paper, we intro- handoff quality, and care transition activities of nurses (e.g.,
duce a methodological framework, Sequential Conversational Analy- [46]), they provide limited understanding of the dynamics of hand-
sis (SCA) that relies on a mixed-method approach that integrates a off conversations, especially with regards to the structure and con-
qualitative conversational analysis with a quantitative sequential tent of communication.
pattern analysis for evaluating interactions during handoffs. Exceptions are studies by McCloughen et al. [2] and Carroll et al.
[32] that have used complementary methods to understand the
actual content of nurse communication. McCloughen et al. [2] used
2. Methodological approaches for studying handoff
content analysis of nurse communication to explore the purpose,
intent, practice, processes, and quality of handoffs in an inpatient
mental health setting. They found that handoffs were retrospec-
In this section, we first describe the prior research on nurse
tive, problem-focused, and inconsistent as they lacked structure
handoff communication, focusing on the evaluation techniques
and context. Similarly, Carroll et al. [32] investigated the nature
that have been previously used. Next, we describe the SCA
of communication exchanges between nurses at shift change.
approach and its applicability for analyzing communication.
Based on a multi-method study using audio recording of handoffs,
interviews, surveys and review of patient records, they identified
2.1. Background: communication during nurse handoffs significant variations in the nature of communication based on role
(incoming vs. outgoing nurses), and years of clinical experience.
Nurse shift report serves as an interactive forum for transfer of They also found that incoming nurses wanted an interactive con-
information regarding the patient, family and other contextual versation about the patient, while outgoing nurses wanted to pre-
issues [34,35]. With the sheer volume and frequency of shift sent information without being interrupted. Similarly, experienced
changes, ensuring safe, effective and efficient handoffs is challeng- nurses provided shorter reports than less experienced ones, which
ing, but critical for maintaining care continuity across shifts [36]. led to significant number of follow-up questions, especially, by less
Besides supporting patient care, researchers have emphasized that experienced incoming nurses. While these studies highlight the
nurse handoffs provides social and psychological support for importance of evaluating the content of clinical conversations, an
achieving nursing staff cohesiveness, professionalism, socialization in-depth understanding of the communication process and conver-
and learning. Given the primary function of information transfer, sational strategies that are instrumental in designing appropriate
nurse handoffs have been characterized as a ‘‘ritual” [37,38]. handoff interventions is limited [2,22,47,48].
Studies of nurse handoffs have been conducted using several In the following section, we present a new methodological
methods including interviews, case narrative reports, question- approach, Sequential Conversational Analysis, to evaluate the con-
naires and surveys (e.g., [39,40]). Using these methods, most tent, structure and nature of roles and relationships within handoff
research studies have highlighted perceived barriers for nurse conversations. Such an approach also helps in examining the
handoffs including general communication problems, human fac- understated functions of handoffs such as social interactions,
tors issues, social and hierarchical problems, time and environ- shared understanding, and distributed cognition [26,31,33,49,50].
mental constraints [22,41]. Evaluation studies of handoff tools
using these methods have shown improvements in nurses’ percep-
tions of increased usefulness and efficiency, improved standardiza- 2.2. Sequential Conversational Analysis
tion of communication, improved information sharing skills, and
most importantly, nurses’ confidence and satisfaction with com- SCA is an exploratory mixed-method approach that combines
munication [22,42]. For example, Nelson and Massey [42] con- in-depth qualitative analysis (using conversational analysis) with
ducted a survey-based study to investigate nurse satisfaction on statistical temporal pattern analysis (using sequential analysis).
the use of a handoff tool. They found that the handoff tool SCA is based on the exploratory data analysis paradigm that relies
improved perceived usefulness and efficiency of the overall nurse on summarizing data through an open-ended approach, and does
workflow by reducing time spent on preparatory handoff commu- not rely on a priori hypotheses or statistical models [51–53]. Three
nication activities. Similarly, Jukkala et al. [43] used surveys to inherent attributes of communication make SCA a viable and effec-
report on nurse perceptions regarding a handoff tool based on tive approach to study handoffs: (a) structural and semantic prop-
the body-systems format. They found that standardization of con- erties of communication; (b) inherent temporal properties of
tent using the tool enhanced perceptions of communication effec- interactive communication; and finally, (c) breakdowns during
tiveness. Other survey-based studies (e.g., see [44]) showed the conversations, and its relationships with the structural and
perceived improvements in standardized communication and temporal characteristics.
78 J. Abraham et al. / Journal of Biomedical Informatics 59 (2016) 76–88

SCA uses a combination of conversational and sequential analy- 3. Evaluation of nurse handoff communication: a case study
sis techniques. Conversational analysis (CA) is a methodological
approach for studying verbal interactions, and relies on procedural In the rest of this paper, we describe a case study on the use of
analysis of audio or video-recorded communication [54,55]. It the SCA approach on evaluating and analyzing nurse handoff com-
involves analysis of how participants organize their conversations munication. Our focus with the case study is twofold: first, to
including the content of their conversations, turn-taking behaviors, describe how the SCA approach can be used to analyze interactive
collaborative development of actions and plans, and problem solv- communication data; and second, to highlight the insights that can
ing. CA is a form of discourse analysis, but utilizes a more holistic be drawn regarding the nuances of handoff communication, specif-
perspective, relying on both the verbal and contextual aspects dur- ically related to the structure and content of interactive nurse
ing interactions [56]. CA has been used in a variety of contexts to handoff communication using the SCA approach. This study was
evaluate physician–patient communication [57], the effect of part of a larger, longitudinal effort on investigating the nature of
decision-making technology on primary care consultation [58], care transitions in critical care settings.
team situation awareness, and communication in operating rooms In the rest of this section, we describe the participants, setting,
[59]. Sequential analysis approaches are used to identify temporal data collection, and the use of the SCA approach for analyzing
patterns of human interactions [60–62]. Given the temporal and nurse handoff communication. The Institutional Review Board
interactive style of handoff conversations (i.e., that involves pre- approved the study and written consents were obtained from all
senting and receiving of information), sequential analysis study participants.
approaches provide an effective basis for identifying repetitive pat-
terns within temporal data. For example, Kannampallil et al. [51]
used this approach to compare the strategies based on identifica- 3.1. Study setting and participants
tion of information seeking patterns among nurse practitioners
and residents. The study was conducted at a 16-bed medical intensive care
Applying the SCA approach involves three steps: (a) data collec- unit (MICU) at an academic medical center in Texas. Nurses
tion and processing of audio or video recorded verbal communica- worked 12-h shifts and were responsible for two patients during
tion data, (b) conversational coding and analysis, and (c) their shift. Handoffs occurred twice a day (at 6 AM and 6 PM) at
communication event transformations and sequential analysis. shift changes. There were several factors that influenced nurse
Data collection and processing involves the collection and tran- assignments: patient acuity (e.g., new nurses were not assigned
scription of verbal data and its segmentation into functional units very sick patients), prior knowledge of a patient, (e.g., if a nurse
of conversation. Functional units are the psychological analogs of a had provided care for a patient previously, then s/he would likely
single unit of experience and are syntactically coherent units that be assigned the same patient); and spatial considerations (e.g., a
correspond to speech acts such as statements or commands nurse was likely to have two co-located patients). Other consider-
[63,64]. A number of coding schemes can then be applied to cate- ations for nurse assignment included patient requests and cultural
gorize the functional units. For example, conversations can be factors (e.g., language). The average patient length of stay in the
coded based on their clinical content, or conversational strategies, MICU was 4 days.
or disruptions (breakdowns). The coding of the functional units Sixteen (N = 16) MICU nurses participated in the study over a 2-
provides a mechanism to categorize the conversational content month period on 15 handoffs (see Table 1 for the list of nurses,
at the level of the smallest coherent conversational level (i.e., func- coded as N1–N16, the 15 handoff sessions, and the 7 patients,
tional units). P1–P7). All participants were critical care nurses; some of the par-
Sequential analysis can then be used to characterize the under- ticipants had CCRN certification (Certification for Adult, Pediatric
lying patterns of the conversations within each of the considered and Neonatal Critical Care Nurses).
dimensions (e.g., structure or clinical content). For sequential anal-
ysis, coded categories are converted into event streams (i.e., a
sequences of coded categories). For example, consider the coding 3.2. Nurse handoff tool
of a handoff conversation on the clinical content dimension was
along 5 variables: cc1 through cc5. Each functional unit will be All participants used a paper-based body systems-based hand-
assigned a category resulting in the entire conversation having a off tool to document patient care information for handoffs. Clini-
clinical content categorization. For example, an event stream for cal content included patient administrative data, code status,
this conversation could be cc1, cc3, cc2, cc5, cc1 . . . cc3. For statisti- clinician orders, problem list, assessment and plan, medications
cal analysis purposes, event streams need to be converted into and treatments, lines and invasive devices, and lab results. The
antecedent–consequent event transition matrices. These transition content was organized into a head-to-toe body system-based
matrices are the basis for conducting categorical analysis to iden- (or organ) information format that included neurological, pul-
tify temporal relationships. Several techniques can be applied to monary, cardiovascular, gastrointestinal, and genitourinary
the antecedent–consequent matrix to characterize sequential rela- systems.
tionships: sequential pattern analysis [51,62], lag sequential anal- The clinical information within each body system was orga-
ysis [65], log-linear modeling [66], and information theoretic nized in a narrative format. For example, within the neurology sys-
approaches [67]. The right half of Fig. 1 shows the sequential anal- tem, nurses documented the problems, assessments, results, and
ysis component—from creating events to developing transition plans related to the nervous system (and patient’s psychological
matrices and various sequential pattern analyses. The left half of behaviors). In addition, the tool included a section for incorporat-
Fig. 1 illustrates the organization and processing of data followed ing a brief history of the patient, primary diagnosis, allergies, code
by conversational analysis. In the next section, we present a case status, and other care related information such as labs, radiology
study on the use of the SCA approach to identify the key character- reports and medication orders and treatments (each of which
istics of nurse communication: to identify transitions in communi- was organized in a temporal fashion). Prior evidence suggests that
cation (i.e., turn taking behavior between incoming and outgoing nurses are trained using the body systems format (as part of their
nurses), structure of communication (i.e., conversational strate- nursing curriculum), and hence such a format is an effective mech-
gies), the clinical content exchanged, and disruptions during anism for presenting clinical information [68]. The handoff tool
handoffs. used by nurses is shown in Fig. 2.
J. Abraham et al. / Journal of Biomedical Informatics 59 (2016) 76–88 79

Record verbal conversations

Data Collection &


Sequential Analysis

Transcribe and segment

conversations into functional
Conversational Analysis

Code functional units

(e.g., communication events,
disruptions, clinical content)

Converting coded functional units Statistical or descriptive

(communication events, clinical analysis of event streams
Qualitative summary of
content, disruptions) into temporal (Chi-square and Lag
conversational patterns
event streams Sequential Analysis)

Fig. 1. Sequential Conversational Analysis framework highlighting the three stages – data collection, coding using the conversational analysis and the temporal sequential

3.3. Nurse handoff communication non-verbal behaviors and actions (e.g., accessing the patient chart
or monitor, level of eye-contact) during handoffs. Given that the
During nurse handoff sessions, the outgoing nurse verbalized field notes were hand-written, the granularity of these notes was
patient information to the incoming nurse in a narrative style, while not detailed enough to systematically evaluate specific non-
referring to their personalized hand-written notes. Incoming nurses verbal interactions. However, these notes provided socio-
noted down relevant details on their own body-system based tool. contextual details around the observed handoffs. The average
The incoming nurse’s notes were updated during their shift with length of a handoff was 516s [SD = 161s].
new and relevant patient information, and were used for their
own ensuing handoff session at the end of their shift. The majority 3.4.2. Observations and interviews
of information exchanged during handoff occurred outside the In order to contextualize handoffs within nurse activities and
patient’s room at the nurse’s workstation. An example snippet of workflow, we (first author and a graduate student in biomedical
the transcribed handoff communication is shown in Fig. 3. informatics) conducted over 20 h of ethnographic observations.
These observations were on days when shift reports were
recorded. The purpose was to understand nurses’ tasks and activi-
3.4. Data collection
ties, both before and after handoffs. Three formal (i.e., scheduled,
semi-structured) and four informal (i.e., during observations after
Audio recording of communication during shift reports, general
handoffs) interviews were conducted with participating nurses.
observations of nurse workflows, and semi-structured nurse inter-
The purpose of these semi-structured, formal interviews was to
views were the primary sources of data for this study.
obtain an understanding of the nurses’ perceptions on the handoff
process including pre-handoff preparatory activities, post-handoff
3.4.1. Audio recording of shift reports
management activities, information needs during handoffs, com-
Fifteen nurse handoffs were audio-recorded (Nmorning = 8;
monly encountered challenges, and workaround strategies that
Nevening = 7). We also took field notes regarding the nurses’
were used to alleviate bottlenecks in workflow. Nurses were asked
to describe these within the context of their incoming and outgo-
Table 1
Distribution of participants across the 15 data collection sessions; Nurses are given ing handoff roles. The interview guide that was used can be found
unique identifiers from N1 through N16 (16 participants). Table shows the outgoing in the supplementary material (see Appendix A). These interviews
nurse and incoming nurses for each session, their shift, repetition of participants and were scheduled and conducted in the nurse’s lounges or break
the patients (P1 through P7) that were discussed.
rooms and lasted an average of approximately 15 min.
Handoff Outgoing Incoming Time of handoff Patient Informal interviews were conducted prior to, or after handoffs
no. nurse nurse [Handoff] no. to gather insights regarding work activities around handoffs or
1. N1 N2 6:00 PM [Evening] P1 clarify specific observed instances. These instances often involved
2. N2 N3 6:00 AM [Morning] P1 clarifications regarding why certain tasks were performed in a par-
3. N4 N2 6:00 PM [Evening] P1 ticular way (e.g., use of electronic tools during handoffs) or ques-
4. N5 N6 6:00 PM [Evening] P1
5. N6 N5 6:00 AM [Morning] P1
tions regarding unique emergencies (e.g., a code event), and its
6. N7 N8 6:00 AM [Morning] P2 impact on care continuity.
7. N9 N10 6:00 PM [Evening] P3
8. N11 N9 6:00 AM [Morning] P3 3.5. Data analysis
9. N9 N11 6:00 AM [Morning] P3
10. N12 N13 6:00 PM [Evening] P4
11. N13 N10 6:00 AM [Morning] P5 The coding and analysis process involved bi-weekly meetings
12. N13 N14 6:00 AM [Morning] P4 for peer de-briefing. We used the grounded theory approach for
13. N2 N15 6:00 AM [Morning] P6 analyzing open-ended field notes and interviews [69]. Audio-
14. N15 N2 6:00 PM [Evening] P6
recorded handoff communication data was analyzed using the
15. N5 N16 6:00 PM [Evening] P7
SCA approach. Each of these is described in further detail.
80 J. Abraham et al. / Journal of Biomedical Informatics 59 (2016) 76–88

Patient Info (Admission)

Diagnosis & History

Medications & Organ/Body Systems:

Treatments Neurology, Respiratory, Cardio-
Vascular, GI/GU, Skin


Lab Results

Fig. 2. Structural organization of the body-system based handoff tool (with the different areas highlighted) used in the MICU.

Outgoing Nurse: He is a 62 year-old,

....deleted [age]... obese, CHF probably; hyperthyroid.
He presented with pneumonia, they diagnosed it as COPD, I mean CHF exacerbation.
So they gave him lasix, no fluid, dropped his BP and then under main propofol and
sent him to us...

Incoming Nurse: Is this the ER? or...

....deleted [location]
Outgoing Nurse: No, this is Livingston

....deleted [location]
Incoming Nurse: Oh, Livingston. Livingston, OK.

Outgoing Nurse: Neuro-wise, he is very pleasant. He is awake and is thankful for

everything you do.

Incoming Nurse: Cool

Outgoing Nurse: Respiratory-wise, he is on a couple of liters of Oxygen on 99-100%.

However, tonight if he has episodes of de-sating, you probably need a bi-pap for sleep
apnea. He has not been diagnosed with it. But he is like 400+. He is clear for me. CV-
wise he has regular heart rate at 80s, 90s. He is on amiodarone at home. So I guess
he is probably at it for hypertension. His blood pressure is like 90 over 60 levo. I have
give 3 liters of fluid and i am only giving him 1 every hour.

Incoming Nurse: you said 3?

Outgoing Nurse: I said 3, but I have given him 1 per hour and then kind of re-
evaluating and let him do at 9,9,9 rather than like pressure bagging him in fast.

Incoming Nurse: Right, right, right

Fig. 3. An example snippet from our data showing the handoff communication.

3.5.1. Analysis of observations and interviews workflow, and the use of the system-based tool to support infor-
We used a grounded theory based open coding approach to ana- mation exchange during handoffs.
lyze various aspects of nurse handoff workflow activities. Examples
of open codes included nurses’ information and communication
needs, interruptions, tasks (before and after handoffs), chart use, 3.5.2. Analysis of audio recorded handoffs
and interactions with members of the care team and the patient’s Verbal communication from each nurse handoff session was
family. Next, we performed axial coding that involved the identifi- transcribed verbatim for analysis. Data analysis involved two
cation of relationships between previously identified open codes to phases: a qualitative conversational analysis phase, where tran-
generate a core set of categories. A prototypical example of an axial scripts were segmented, and coded along multiple dimensions of
code was nurse’s information documentation tasks on patient’s communication; and a quantitative sequential analysis phase,
EHR (Electronic Health Record) before completing the shift. Once where temporal sequential pattern analysis was performed to
the axial coding was completed, we reviewed our field notes to identify potential patterns in communication across nurse hand-
selectively code any data that related to the core categories to gen- offs. The various steps in each of these phases are described in
erate a theory or a collection of explanations that described nurse detail below.
J. Abraham et al. / Journal of Biomedical Informatics 59 (2016) 76–88 81

Table 2
Segmentation of functional units in nursing handoff communication.

Speaker Description Example (from data)

A (Outgoing Nurse) A meaningful piece of information from the outgoing nurse A: ‘‘I was drawing labs sometime, I drew labs and sent them – they wanted
to do magnesium so the bag is hanging down.”
B (Incoming Nurse) A meaningful piece of information from the incoming nurse B: ‘‘So did we ever get the sputum?” Conversational analysis. Qualitative analysis followed a con- provides insights into the underlying structure of communication.
versational analysis approach [70,71]. This approach has been pre- We used a modified version of the structured communication cod-
viously used to explore the purpose, intent, practice, processes, and ing framework by Clark and Schaefer [77] to categorize the types of
quality of nurse handoff [2], and the content analysis demon- conversational moves during handoffs (see Table 3).
strated that these handoffs were retrospective, problem-focused The first author (JA) coded all transcripts for conversational
and inconsistent as they lacked structure and context. Other stud- strategies; the third author (CJB) coded one representative tran-
ies have utilized similar discourse and content analysis approaches script (11.32% of total functional units) with high inter-rater agree-
within the context of handoffs [72,73]. ment (Cohen’s Kappa = 0.89, 96.18% agreement).
Segmentation of Content: Verbal content was segmented into Clinical content of communication: By analyzing each functional
functional units of conversation, which are defined as the psycho- unit, we categorized the clinical information exchanged during
logical analogs of a single unit of experience [64,72]. While these handoffs. For this purpose, we adapted the coding scheme devel-
were usually syntactically coherent units corresponding to speech oped by Berkenstadt et al. [78] for analyzing the clinical content
acts such as statements or commands, single words could also be of handoff communication. The framework included a comprehen-
units if they functioned as separate speech acts (e.g., ‘‘okay”). For sive list of specific clinical elements relevant to critical care
each functional unit, we recorded the speaker (i.e., A: outgoing; patients (see Table 4 for all the categories and their description).
and B: incoming), and the position of the spoken content in the The first author (JA) coded the transcripts for content of clinical
transcript (i.e., line number). Table 2 provides an example of the communication (using medical reference books with joint consul-
segmentation of the content. To evaluate the reliability of the seg- tation with KFA, an ICU physician); given the clinical nature of the
mentation process, two coders (JA, CJB) separately coded func- data, a nurse collaborator (KDL, an RN, PhD) coded two transcripts
tional units in two transcripts. The two coders then compared (23.2% of total functional units). Comparisons of the coding showed
their coding of the functional units. Disagreements regarding the high inter-rater reliability (Cohen’s Kappa = 0.97, 94.9% agreement).
classification of functional units were resolved through discussion After discussion and further clarification of the coding scheme,
and a mutually agreed segmentation was derived. The ratio of inter-rater agreement reached 100%.
changes that were made to the segmentation was used as a mea- Disruptions in communication: Communication failures have
sure of reliability. The two coders had 94% agreement. been cited as a primary contributor to sentinel events and medical
Segmented transcripts were further coded along three dimen- errors [1]. We identified the communication disruptions during
sions: structure, clinical content and disruptions in communica- nurse shift reports to evaluate the quality of communication.
tion. The structure of communication dimension was used to A disruption was characterized as a breakdown (or gap) during
identify the conversational strategies that were used during collab- information exchange. For example, failure to share admission
orative interactions for transferring patient related information, information of a newly admitted patient by the outgoing nurse
and establishing common ground. In other words, this dimension can result in such a disruption. We adapted the information break-
can be used to characterize the nature of collaboration during com- down classification framework developed by Abraham et al. [1,27]
municative interactions. The content of communication dimension to analyze the handoff communication disruptions. Each functional
was for identifying the nature of clinical content during the con- unit was classified as either ‘‘no disruption” or as one of five types:
versations. The disruptions in communication were used as a surro- doubtful information, missing information, incorrect/conflicting
gate measure for representing the communication gaps during information, repetitive information or misinterpreted information
handoffs. (see Table 5).
Structure of communication: During collaborative interactions, The first author (JA) coded all transcripts for disruptions identi-
conversations coordinate both content and processes—whereby fied in the conversation; the second author (TGK) coded three tran-
speakers develop a shared understanding of the exchanged infor- scripts (29.42% of functional units) with a relatively high inter-
mation. The process of creating such a shared understanding is rater agreement (Cohen’s Kappa = 0.76, 97.7% agreement). Differ-
the basis for effective collaborative interaction [74–76]. Capturing ences were resolved by discussion, with final inter-rater agreement
the strategies used to maintain the continuity of conversations reaching 100%.

Table 3
Structure of communication during collaborative interaction in nurse handoffs (modified from [77]).

Conversational Definition Example (from data)

moves during
collaborative interaction
Present Presenting patient information ‘‘Neuro-wise, she’s been following commands for me”
Seek Requesting additional information ‘‘They didn’t perform the drip?”
Propose Provide recommendation for care delivery (e.g., plan) ‘‘and ...after that you can turn it off, okay?”
Critique Rejecting a some part presented information A: ‘‘and she’s on Levo at [PAUSE].”
B: ‘‘I think it’s off.”
Revise Revise the information based on the critique ‘‘Oh, okay. I had it at 5.”
Accept Accept presented information ‘‘Yeah, I’ll keep it on.”
Meta-Cognition Information about regarding reasoning or thoughts ‘‘If his cardiac output starts to drop, then you will say his heart is acting up.”
82 J. Abraham et al. / Journal of Biomedical Informatics 59 (2016) 76–88

Table 4
Clinical content of communication exchanged during nurse handoffs (from [78]).

Content categories Description Examples of content elements

Administrative data Patient demographics, and patient admission information to the unit Patient name, age and sex
Family background
Date of admission
Length of stay
Problem List Medical observations of patient conditions and patient history Medical and surgical history
Reason for admission to unit
Current medical problems
Current assessment and Patient status and plan related to body systems including neurology, Neurology: consciousness and sedation score,
plan (including cardiovascular, respiratory/pulmonary, infection disease, GI/GU, skin extremities, psychological status
diagnosis) Cardiovascular: Blood pressure; mean arterial pressure;
heart rate; Central venous pressure, rhythm
Respiratory/Pulmonary: respiratory rate, oxygen
saturation, mechanical ventilation
Infectious disease: Temperature max, Lactic acid, ScVO2
(venous oximetry catheter), Cultures
GI/GU: Diet, Abdominal status, Input/Output fluid Skin:
Lesions, redness
Medications and treatment Medications and treatments for each body system problem Regular medications
Medications in continuous feeding
Lines and invasive devices Invasive lines and devices IV lines and Fluid input
Nasogastric tube and input/output
Urinary catheter and urine output
Endotracheal tube and secretions
Labs Lab work status Lab work including blood, sputum cultures, etc.
Results Results from lab work and imaging studies Laboratory results
Radiology results
Events during the last shift Patient unexpected response to intervention Hemodynamic, respiratory, other
Code Status Status of patient Full code, DNR (Do-not-resuscitate), DNI (Do-not-
Order Review Checking orders together on patient chart (toward the end of handoff) Pending and new medication and procedure orders
Tasks expected to be done Pending tasks Laboratory test, imaging studies
Family concerns Concerns about patient status/care from family Family issues/questions about care process
Special Order Orders for special conditions Isolation, fall precautions, restraints
Allergies Patient allergies Allergies to medicines, latex, food
Other Any other situational awareness information that do not fall in the above Psycho-social and contextual dynamics of
categories (and unrelated to patient’s diagnostic condition) patient/family

Inflection point in conversations: During data analysis, we The quantitative analysis involved the computation of mean
observed changes in the structure and content of communication length of turns of the speakers, and the number of turns to analyze
during the latter stages of each shift report. To further examine the transitions in communication. Turn length represents the num-
this, we developed a post hoc hypothesis, where we identified an ber of consecutive functional units attributed to a speaker before
inflection point in each transcript that signified a conversational the second participant speaks, while number of turns reflects the
transition. For each handoff transcript, the inflection in the conver- switches between speakers during a conversation.
sation was marked as a point in the conversation where there was
a clear distinction in the conversational structure and content—the Sequential analysis. Verbal data that were coded along the
outgoing nurse paused after their information presentation giving three dimensions, structure, clinical content and disruptions were
the incoming nurse an opportunity to ask questions regarding the used for sequential analysis. We used Lag Sequential Analysis (LSA)
presented information. We provide an example how the inflection to analyze the evolution of repetitive patterns within the structure
point was coded in the transcripts. and content of handoff communication. LSA has been widely used
Table 6 provides a segment of the conversation where an inflec- in general behavioral research [66,79], and more specifically in
tion in conversation was identified. After the ‘‘that’s all” dialogue studying physician–patient interactions [65,80,81].1 We performed
by the outgoing nurse (see line 4 in Table 6), there was a long pause LSA only on the structure of communication categories (see Table 3).
in the conversation indicating the end of information transfer. This In order to perform LSA, the structure of communication data was
pause was often followed by the incoming nurse’s questions transformed into an antecedent–consequent transition matrix. The
regarding the presented information. In the example in Table 6, raw sequences of structure of communication data included 7 items
the questions were related to the administered medications. of outgoing nurse (A) conversational structure, and 7 items of incom-
Two authors coded all transcripts (JA, CJB) indicating the inflec- ing nurse (B) conversational structure (see Table 3). For example, the
tion point in the conversation, with a high degree of inter-rater
agreement (Cohen’s Kappa = 0.93, 97.56% agreement). Differences
were resolved by discussion such that final agreement reached 1
Performing LSA requires individual cells to not be sparsely populated. Given that
100%. there were 15 clinical content categories, the cells were sparsely populated.
J. Abraham et al. / Journal of Biomedical Informatics 59 (2016) 76–88 83

Table 5
Disruptions in communication during nurse handoffs (modified from [27]).

Communication disruption type Definition Example (from data)

Doubtful information Communication characterized by uncertainty in patient A: ‘‘The way, um, I was, uh, helping, uh, I think, um, K said it’s a
information and is subject to question multiple resistance to the ___ and some other stuff that she could
not pronounce.”
Missing information Communication characterized by incomplete or loss of A: ‘‘An NG tube? I don’t know. I don’t know.”
Incorrect/Conflicting information Communication characterized by inaccurate information A: ‘‘Going into this one.”
and erroneous data B: ‘‘No, they want the, that’s what I told you, they don’t want to do
a, they want a new line.”
Repetitive information Communication characterized by a complete match B: ‘‘You said 3?”
(duplicate) or similarity in information exchanged earlier
in the conversation
Misinterpreted Information Communication characterized by inaccurate explanation B: ‘‘I thought you are suggesting me to tell him not to?”
and understanding of the information A: ‘‘No. That’s what he said. Let me wrap it so that you don’t get
bed bugs.”

Table 6 our qualitative evaluation that illustrates the use of the body-
Example of how the inflection in the conversation was coded. systems based tool by the nurses before and after handoffs.
Speaker Content Pre or
4.1. Transitions during communication
Outgoing Ativan 1 milligram titrate to myoclonic seizures or Pre
Mean length of turns for the outgoing nurse (M = 3.50,
Outgoing Um, here. Continued, No myoclonic jerks. Continuous Pre
one and one
SD = 1.63) was greater than the incoming nurse for the overall
Outgoing do not interrupt patient’s sedation Pre (M = 1.09, SD = 0.08) [t(14) = 5.74, p < 0.001], pre-[t(14) = 4.89,
Outgoing And that’s all Pre p < 0.01)], and post-inflection phases [t(14) = 3.78, p < 0.001]. The
Pause in conversation [Inflection Point] mean number of turns across all the transcripts was 59.20
Incoming So she’s no longer on . . ..? Post (SD = 33.20). The mean number of turns during the pre-inflection
Outgoing She’s not on it anymore Post phase (M = 42, SD = 24) was significantly greater than that during
Incoming Or the propofol? Post
the post-inflection phase (M = 17.87, SD = 17.44) [t(14) = 3.63,
Outgoing Or the propofol. [UNCLEAR] DC’d. Post
Outgoing Anything else I need to tell you about her? Post p = 0.002].
Incoming [UNCLEAR]-no Post

4.2. Structure, content and disruptions in communication

outgoing nurse would have the following categories: A-Present, Structure of communication: Conversational strategies during
A-Seek, A-Propose, A-Critique, A-Accept, A-Revise, and A- collaborative interaction included: presentation of information
Metacognition. In this transition matrix, initial communication (69.7% of functional units), accepting information (16.3%), and
structure categories were listed by row (antecedent), and subse- seeking of additional information (8.1%). There were limited
quent communication structure categories were listed by column meta-cognitive (2.5%) or critiquing (0.2%) aspects during these
(consequent). Each cell in the antecedent–consequent matrix repre- interactions (see Fig. 4).
sents the frequency of transitions between two considered cate- There was a significant association between the speakers and
gories (e.g., A-Present and B-Accept with cell count of 5 indicates their structure of communication (v2(6) = 1267.03, p < 0.001), with
five instances of A-Present followed by B-Accept conversations). the outgoing nurse focusing on information presentation (89% of
In order to perform Chi-square analysis on the tables, the spar- functional units), and the incoming nurse on accepting (59%) or
sely populated rows and columns were eliminated (i.e., those with seeking additional information (26%). The structure of communica-
marginal frequencies less than 35). After the removal of the sparse tion was significantly different between pre- and post-inflection
cells, the following categories remained: ‘‘A-Present”, ‘‘A-Propose”, phases (v2(6) = 34.27, p < 0.0001) with the post-inflection phase
‘‘A-Metacognition”, ‘‘B-Accept”, and ‘‘B-Seek” (i.e., a 5  5 matrix). having 20% less information presentation and 50% more informa-
We performed a likelihood Chi-square test on the antecedent–con- tion seeking than the pre-inflection phase.
sequent transition matrix to ascertain whether the cell frequencies Clinical content of communication: The content of communica-
were distributed across the contingency table by chance or not. If tion was distributed across information related to current assess-
the rows and columns were independent, expected cell counts ments and plan (total = 35.1%) related to the cardio-vascular
and adjusted residuals were calculated for each cell in the table. (7.4%), neurology (7.3%), pulmonary (6.9%) and GI (6.7%) systems;
Each cell was compared to a criterion value of 2.58 (i.e., a = 0.01). lines (16.0%), medications and treatment (9.6%), order reviews
Positive values of adjusted standardized residuals indicated that (8.6%), and problems (8.1%). There was limited discussion of labs
observed values were larger than expected values for each cell, (1.7%), special orders (0.7%), allergies (0.5%) or code status (0.4%).
and were evidence that the transition between conversational cat- A summary of the distribution across all the considered clinical
egories represented by that cell occurred more frequently than categories can be found in Table 7.
would be expected by chance. There was a significant association between speakers and their
content of communication (v2(19) = 43.91, p < 0.001) with both
4. Results speakers focusing on lines (Outgoing: 16%, Incoming: 15%) and
medications (Outgoing: 16%, Incoming: 15%). Additionally, there
In this section, we describe the analysis of transitions, interac- was a significant focus by the incoming nurse on order review
tions and sequential structure that were observed during the (13%). The content of communication was significantly different
verbal handoff communication. Next, we report on the results from in the pre- and post-inflection phases (v2(19) = 578.68,
84 J. Abraham et al. / Journal of Biomedical Informatics 59 (2016) 76–88

Percentage of Communication Category


70.0 0.35

Proportion of Disruptions
60.0 0.3

50.0 0.25

40.0 0.2

30.0 0.15

20.0 0.1

10.0 0.05

0.0 0
Doubtful Duplicate Incorrect/Conflicting Misinterpreted Missing
Present Seek Propose Critique Revise Accept Meta-
Fig. 5. Distribution of the various types of disruptions (see the list of disruption
Fig. 4. Distribution of the various categories related to the structure of categories in Table 5).

words, longer conversation length did not lead to increased

Table 7 communication disruptions.
Distribution of the various clinical content categories discussed during nurse handoff
4.3. Patterns within the structure of communication: sequential
Clinical category % of clinical content analysis
Administrative data 2.2
Problem list 8.1 Chi-square likelihood ratio tests were significant for the transi-
Current assessment and plan tion matrix representing the structure of communication—indicat
CV 7.4 ing possible association between the structure of communication
GI/GU 6.7 categories (v2(25) = 265.93, p < 0.001). There was a positive rela-
ID 3.4
Neurological 7.3
tionship between six pairs of categories (see Table 7, cells shaded
Pulmonary 6.9 in gray). These were (‘‘?” indicates followed by): (1) A-
Skin 3.4 Present ? B-Accept; (2) A-Present ? B-Seek; (3) B-Accept ? A-
Medications and treatment 9.6 Present; (4) B-Accept ? A-Meta-Cognition; (5) B-Accept ? A-
Lines and invasive devices 16.0
Propose and (6) B-Seek ? A-Present.
Labs 1.7
Results 1.6 These sequences on structure of communication can add
Events during the last shift 1.6 nuance to our understanding of interaction between the nurses.
Code status 0.4 For example, the outgoing nurse’s role is most often to present
Order review 8.6 information to the incoming nurse, who accepts or seeks additional
Tasks expected to be completed 3.8
Family concerns 4.2
information (sequences: A-Present ? B-Accept; A-Present ? B-
Special order 0.7 Seek). The incoming nurse’s conversations are dominated by ‘‘ac-
Allergies 0.5 cept” conversations (i.e., acknowledgments), which are followed
Other 5.7 by the outgoing nurse presenting new information (B-Accept ? A-
Present), offering insightful responses (B-Accept ? A-Meta-
Cognition) or proposing new information (B-Accept ? A-Propose)
p < 0.0001). While the content of pre-inflection conversations (see Table 8).
focused on lines (19%), medications and treatment (10%), post- Two of these sequences were symmetric (and bidirectional), i.e.,
inflection conversations focused on order review (31%), and transitions occurred in both directions. These were: (a) A-Present,
patient/family’s psycho-social and contextual dynamics (17%; i.e., B-Accept and (b) A-Present, B-Seek. These symmetric, bidirectional
‘‘Other” in Table 4).2 sequences provide evidence for the presence of longer sequences
A preliminary analysis of the comparison between the morning that contain Present-Accept and Present-Seek transitions, showing
and evening handoffs can be found in Appendix A. consistent interactivity during handoff communication. For exam-
Disruptions in communication: 58 disruptions in communication ple, there are likely to be longer sequences of [A-Present, B-Accept,
were identified (3.1% of functional units). Most of these disruptions A-Present, B-Accept. . .] and [B-Seek, A-Present, B-Seek, A-Present. . .].
were caused by either doubtful or missing information (37.9% Further evidence for interactivity also exists in the fact that consec-
each), followed by repetitive information (15.5%), incorrect/con- utive conversational categories were not spoken by the same
flicting information (6.9%), and misinterpreted information (1.7%) speaker, showing the absence of long monologues. In one of our
(see Fig. 5). There was no statistically significant association interviews, a nurse remarked that ‘‘whenever, say the nurse is giving
between the speakers and disruptions (v2(1) = 0.67, p = 0.41), or report, she may say something that triggers something that may lead
between the phases of the conversation (v2(1) = 1.89, p = 0.17). to additional questions. Let’s say she says his CVP (Central Venous
Based on a linear regression analysis, we found that disruptions Pressure) has been 2 or 3 and I then ask – is his heart rate elevated?
were not associated with the mean length of turns for either the Is his blood pressure lower? I automatically think of volume status.”
outgoing or the incoming nurse (p > 0.05). Additionally, over 80%
of the disruptions occurred when the length of the previous 4.4. Qualitative evaluation of nurse shift reports
conversation was less than or equal to 5 functional units. In other
In this section, we report on our results from the observations
and interviews that add insights on the nature of handoff commu-
Comparisons based on the derived groups (i.e., speakers, pre- and post-inflection) nication using the system-based tool. Primarily, we focus on the
were based on normalized percentages of the categories within each group. nurse handoff workflow and tool use.
J. Abraham et al. / Journal of Biomedical Informatics 59 (2016) 76–88 85

Table 8
Frequencies, conditional probabilities, and adjusted standardized residuals for the overall structure of conversation variables (the significant pairs of cells are shaded).

Nurse handoff communication was situated within the clinical the tool helped them in having a consistent and systematic conver-
workflow of the nurses. As has been previously reported regarding sation, in spite of the variability within patient, clinician or other
resident handoffs [1,82], nurse handoffs in the MICU consisted of contextual aspects. Several nurses also commented on the role of
three phases: a pre-handoff phase, a handoff phase and a post- handoff tool as a cognitive aid, especially to support information
handoff phase. During the pre-handoff phase, nurses focused on seeking and documentation in preparation for handoffs.
completing the patient care tasks and activities, updating the
EHR, and other documentation efforts. As the body systems-
4.5. Summary of results: nurse handoffs
based structure of handoff tool supported the constant update dur-
ing the course of their shift, it was a ‘‘live” document and required
We used a new methodological framework, Sequential Conver-
minimal updates (and additional effort and time) prior to the shift
sational Analysis, to evaluate nurse handoff communication. The
report (i.e., supporting information updates). The handoff phase
SCA framework based analysis showed differences between the
was primarily focused on the exchange of patient related informa-
outgoing and incoming nurses’ conversational patterns, conversa-
tion. In addition to the structured presentation, there were also
tional strategies and content of information exchanged during
informal information exchanges that provided contextual informa-
the handoffs. Analysis of conversational strategies showed signifi-
tion for smooth care transitions. For example, an outgoing nurse
cant interactivity during conversations with limited monologues.
asked the incoming nurse to be on alert for an ‘‘unstamped” physi-
There were two distinct phases in the conversation based on a shift
cian order that was sent to the pharmacy. The outgoing nurse also
in the conversational pattern and strategy. While the pre-
advised the incoming nurse to re-send the order such that patient
inflection phase involved information presentation/acknowledgment,
medications can be administered in a timely manner. Such
the post-inflection phase consisted of interactive review and revi-
exchanges highlight the role of contextual and situated informa-
sion. Predominant patterns of information exchange involved infor-
tion for the incoming nurse to more effectively perform their
mation presentation by the outgoing nurse followed by
post-handoff tasks.
acknowledgments (and acceptance of presented information) by
The post-handoff phase was primarily focused on the incoming
the incoming nurse with limited focus on evaluation or reflection
nurses’ tasks such as patient chart review, nurse–physician and
regarding the exchanged information. While outgoing and incoming
nurse–respiratory therapist communication (in instances where
nurses used different conversational strategies during handoffs,
the patient was on mechanical ventilation) regarding the patient’s
these differences had no significant effect on causing disruptions
care plan.
in conversations. Correspondingly, longer monologues by a
One of the important takeaways from our interviews and obser-
speaker did not lead to more disruptions. Clinical content during
vation sessions was how the nurses used the body-system based
pre-inflection conversations was related to information regarding
handoff tool. Most nurses described the body-system based tool
lines and patient medications, while post-inflection conversations
as an effective mechanism to develop a holistic overview of the
were related to order reviews and discussion of the patient/family’s
patient—both in their roles as an incoming and outgoing nurse.
psycho-social and contextual aspects, unrelated to patient clinical con-
For example, one of the nurses remarked that the patient informa-
dition. Additionally, the use of the body-systems based tool was
tion on the EHR system is typically fragmented across multiple
perceived to enhance a nurse’s ability to quickly develop a holistic
sources, ‘‘forms on the EHR are not consistent, it doesn’t show up
perspective of the patient contributing to thorough handoff
and it [information] is in multiple places,. . . hard to find.” As a result,
the handoff tool provided a standardized template for gathering
and organizing the information for presentation during handoffs.
Additionally, as previously described, the handoff tool was con- 5. Discussion
stantly updated during the shift and integrated into the nurse
workflow. One of the nurses stated that the body-systems based Care transitions have been identified as a vulnerable period dur-
tool helped to ‘‘review the systems from head to toe, highlight any ing care delivery with risks for communication breakdowns [83].
changes during my shift, and order review.” However, there is a lack of comprehensive understanding of the
In addition to being used for information updates during the shift, nature of communication during handoffs [33]. The current focus
the handoff tool helped them in structuring the discussion during on handoffs has centered on ‘‘information transfer,” with limited
handoffs. Most nurses described that the standardized structure of focus on the nature and patterns of the dynamics of such an infor-
86 J. Abraham et al. / Journal of Biomedical Informatics 59 (2016) 76–88

mation transfer [31,32,49]. Although one of the goals of handoffs ance the collaborative effort. In nurse handoff contexts, it is poten-
emphasizes the importance of ensuring ‘‘common ground” tially possible to break the information presentation into shorter
[63,84] or a shared understanding of the patient case between clin- installments; for example, the outgoing nurse can present informa-
icians [85], there has been limited investigation on the collaborative tion on a single body system, and then allow the incoming nurse to
aspects of handoff interactions [32,86–88]. Using the SCA approach, interject (or ask clarification questions) before presenting the next
we investigated the process and structure of information sharing body system. These shorter communication spans can increase the
between the incoming and the outgoing nurses. number of turns, but provide more opportunities for clarifications,
Based on our results, we describe two aspects of conversational and balance the overall collaborative effort.
patterns—the nature of interactivity and collaborative effort during
nurse handoffs, and we highlight the potential of the SCA frame- 5.2. Significance of using the SCA approach for studying verbal
work for investigating temporal verbal interaction in pairs and communication
small groups.
In a recent commentary, Enrico Coiera argued for a paradigm
5.1. Interactivity and collaborative effort shift in the design thinking related to informatics—arguing for
the importance of understanding the dynamics of communication
Nurse handoffs have primarily been characterized as an unidi- for designing information systems in healthcare [95]. His argu-
rectional transfer of information (from the outgoing to the incom- ments were predicated on how errors in communication lead to
ing nurse) [15]. Our analyses showed that nurse handoffs are clinical morbidity and mortality. He presented a ‘‘continuum view”
relatively interactive. We use the term, relatively interactive to that relied on understanding conversational structure and model-
highlight that the incoming nurses’ interactions during the conver- ing the flow of conversations. The SCA approach described in this
sation are in the form of acknowledgments of receiving informa- paper mirrors this perspective combining seemingly orthogonal
tion (e.g., ‘‘OK”), rather than any confirmatory processes (e.g., methodological approaches to characterize conversational
‘‘read back” of presented information, as is often the case in Air patterns.
Traffic Control communication). Though TJC suggests the use of While our focus in this paper has primarily been on analyzing
read-backs as a mechanism for effective grounding and interactiv- dyadic handoff communication, the general SCA approach can be
ity during handoffs, we found no evidence for the use of this strat- utilized for investigating any type of clinical communication—
egy in our data [63]. This is likely because the narrative e.g., during workflow, decision-making or reasoning tasks. There
presentation style used by the outgoing nurses may not be amen- are two important considerations regarding the SCA approach:
able to a stringent read-back strategy [20]. first, as with any in-depth conversational analysis approach, there
Prior research has provided evidence on the importance of two- is significant effort involved in transcribing, segmenting, organiz-
way conversation and interaction between nurses to promote team ing and coding the communication content. Researchers consider-
cohesion, teaching, and collaborative reflection of patient care tasks ing the use of this approach should account for the time and effort
of the previous shift [36]. Our results point to a two-way interaction investment. Second, the approach requires an interdisciplinary
between nurses with limited reflective or critical review of pre- team that includes experts in the clinical subject matter, commu-
sented information (see Section 4.2). Researchers have shown that nication, and statistical analysis. Nevertheless, the SCA approach
explicit metacognitive activities can improve the reasoning process provides an opportunity to conduct in-depth analysis on interac-
leading to effective joint actions and improved collaborative activ- tive conversations, helping in the identification of temporally situ-
ities [89–91]. Although limited, the transfer of meta-cognitive effort ated, and repetitive patterns in language.
observed in our data can potentially minimize opportunities for
bringing in a ‘‘fresh eye” perspective—often considered to be a pos- 5.3. Conclusion and limitations
itive aspect of handoffs enabling the detection and recovery from
errors [92]. It is also relevant to mention that during handoffs The role of effective handoffs in ensuring the quality and safety
(and similar care transition events), clinicians perform a balancing of the patient care is well recognized [96]. Using a novel SCA
act between information comprehensiveness and limited available framework, we identified several nuances of the nurse handoff
resources (i.e., time and effort). However, the transfer of outgoing communication that can have a significant impact on patient care
nurse’s metacognitive activities and critiques regarding patient across the care continuum. Insights on the nature and characteris-
assessment and plan can improve shared decision making, poten- tics of handoff communication such as its inherent interactivity,
tially resulting in better continuity of care [93]. two-phased communication processes supporting different func-
Closely related to interactivity is the concept of collaborative tions can inform the development of training and design interven-
effort for achieving grounding (or shared understanding) during tions for improved handoff communication. For example, training
communication [84]. Based on our analysis, we found that the interventions can focus on educating nurses on the importance of
mean length of turns for the outgoing nurse was greater than that transfer of self-reflective and metacognitive aspects. Learners
of the incoming nurse, highlighting the incongruent distribution of who engage in self-reflective and critical thinking (i.e., metacogni-
effort during information sharing. In other words, it seems that the tion) have been found to be better at processing and storing new
burden of collaborative effort is skewed toward the incoming nurse information. Several tools have been developed for nurses that
where the incoming nurse has the task of receiving, filtering, and help in critiquing diagnosis and care plans [97], and for improved
processing the presented information. This highlights the predom- judgment and reasoning regarding patient conditions [98].
inant role of the outgoing nurse in these conversations, and the rel- In terms of design interventions, information on handoff tools
atively lower overall collaborative effort between the nurses. should highlight the critical and noteworthy clinical events related
Morrow et al. [64,94] reported similar results regarding differ- to patient treatments, systems-based assessment and plan, and
ential collaborative effort between speakers during air traffic con- orders within the context of the overall patient condition and sta-
troller communication. They found that reduced turn length (and tus. This work extends previous work that argues for the need for
therefore increase in the content presented in a single message) considering the social and contextual aspects of handoffs while
lead to reduced common understandings between speakers. Using designing tools and strategies for handoffs [32]. The presentation
a collaboration scheme that supports natural transitions in of information using the body-system or medical format allows
communication has been suggested as a potential strategy to bal- for a holistic overview of the patient, and also provides a frame-
J. Abraham et al. / Journal of Biomedical Informatics 59 (2016) 76–88 87

work for the receiver to detect potential discrepancies within the Appendix A. Supplementary material
presented data. For example, the handoff tool should allow for easy
flagging of key pieces of information that may require a ‘‘fresh Supplementary data associated with this article can be found, in
eye”. Additionally, tools must provide support for various phases the online version, at http://dx.doi.org/10.1016/j.jbi.2015.11.009.
in the conversation—for example, the discussion of the tasks (i.e.,
post-inflection phase in our study) needs to be interactive and References
streamlined to aid nurses to provide a more-comprehensive and
collaborative care. The SCA approach, though exploratory, provides [1] J. Abraham, T.G. Kannampallil, V.L. Patel, Bridging gaps in handoffs: a
a viable mechanism for characterizing the evolution of such con- continuity of care approach, J. Biomed. Inform. 45 (2) (2012) 240–254.
[2] A. McCloughen et al., Nursing handover within mental health rehabilitation:
versations in pairs and small groups.
an exploratory study of practice and perception, Int. J. Mental Health Nurs. 17
We acknowledge the following limitations of our exploratory (2008) 287–295.
study. First, we did not evaluate the impact of unresolved disrup- [3] E.T. Eggland, D.S. Heinemann, The reporting process, in: D.L. Hilton, B. Ryalls
(Eds.), Nursing Documentation, J.B. Lippincot, Philadelphia, PA, 1994, pp. 209–
tions, and the unintended effects of the use of the specific nurse
handoff tool on clinical outcomes. Given the focus of this paper [4] I. Ekman, K. Segesten, Disputed power of medical control: the hidden message
was on the development and use of the SCA approach to first iden- in the ritual of oral shift reports, J. Adv. Nurs. 22 (1995) 1006–1011.
tify the nature and quality of nurse communication during shift [5] The Joint Commission on Accreditation of Healthcare Organizations (JCAHO),
National Patient Safety Goals, in: Critical Access Hospital and Hospital National
changes, we hope to assess the impact of the disruptions and unin- Patient Safety Goals, 2006.
tended consequences related to the tool in our future work. Second, [6] V.L. Patel, T.G. Kannampallil, E.H. Shortliffe, Role of cognition in generating and
we did not evaluate the non-verbal cues (such as gestures, or direct mitigating clinical errors, BMJ Qual. Safety 24 (7) (2015) 468–474.
[7] D.T. Bomba, R. Prakash, A description of handover processes in an Australian
eye contact) that are likely to have had an impact on the quality and public hospital, Aust. Health Rev. 29 (2005) 68–79.
efficacy of grounding during communication. While we did make [8] M.M. Benham-Hutchins, J.A. Effken, Multi-professional patterns and methods
field notes regarding some non-verbal cues, they were not detailed of communication during patient handoffs, Int. J. Med. Inform. 79 (4) (2010)
enough as those captured using video recordings or through simu- [9] B. O’Connell, K. Macdonald, C. Kelly, Nursing handover: it’s time for a change,
lated scenarios in experimental settings (see e.g., [99]). While we Contemp. Nurse: J. Aust. Nurs. Prof. 30 (1) (2008) 2–11.
acknowledge that non-verbal cues have a significant role in interac- [10] A. Laxmisan et al., The multitasking clinician: decision-making and cognitive
demand during and after team handoffs in emergency care, Int. J. Med. Inform.
tive communication, the constraints of the clinical environment 76 (11) (2007) 801–811.
(i.e., the intensive care unit) and the institutional restrictions pre- [11] E.W. Coiera et al., Communication loads on clinical staff in the emergency
vented the video recording of our sessions. Additionally, we also department, Med. J. Aust. 176 (2002) 415–418.
[12] B. Hedberg, U.S. Larsson, Environmental elements affecting the decision-
did not perform any secondary analysis based on the expertise of
making process in nursing practice, J. Clin. Nurs. 13 (2004) 316–324.
the nurses, or weekday/weekend handoffs as we used a relatively [13] Patterson et al., Handoff strategies in settings with high consequences for
small sample of handoffs for our analysis. Third, the nurses used failure: lessons for health care operations, Int. J. Qual. Health Care 16 (2)
the body-systems based tool for their handoffs in the MICU. The (2004) 125–132.
[14] P. Calleja, L.M. Aitken, M.L. Cooke, Information transfer for multi-trauma
results presented in this paper are likely applicable only to critical patients on discharge from the emergency department: mixed-method
care settings (such as surgical intensive care unit, and transplant narrative review, J. Adv. Nurs. 67 (1) (2011) 4–18.
care unit) that follow a system-based information format. Fourth, [15] L.A. Riesenberg, J. Leisch, J.M. Cunningham, Nursing handoffs: a systematic
review of the literature, Am. J. Nurs. 110 (4) (2010) 24–34.
we only evaluated the metacognitive processes during handoff [16] C.A. Welsh, M.E. Flanagan, P. Ebright, Barriers and facilitators to nursing
communication. It is possible (and likely) that the nurses engaged handoffs: recommendations for redesign, Nurs. Outlook 58 (3) (2010) 148–
in metacognitive activities on their patient care activities prior to 154.
[17] J. Abraham et al., Comparative evaluation of the content and structure of
handoff communication. More research is needed to establish the communication using two handoff tools: implications for patient safety, J. Crit.
role of metacognition during a care transition discussion, and its Care 29 (2) (2014). 311e1–e7.
potential effects on the quality and safety of ensuing patient care [18] T.G. Kannampallil et al., Considering complexity in healthcare systems, J.
Biomed. Inform. 44 (6) (2011) 943–947.
activities. Finally, we would also like to highlight the fact that the [19] M. Freitag, V.S. Carroll, Handoff communication: using failure modes and
focus of this paper was on highlighting the applicability of the effects analysis to improve transition in care process, Qual. Manage. Health
SCA approach in characterizing the nuances of interactive nurse Care 20 (2) (2011) 103–109.
[20] E.S. Patterson, E.M. Roth, M.L. Render, Handoffs during nursing shift changes in
handoff communication. The case study was a single site study,
acute care, in: Proceedings of the Human Factors and Ergonomics Society, vol.
and further research maybe required in drawing generalizable con- 49, 2005, pp. 1057–1061.
clusions regarding the nature and patterns of nurse communica- [21] N. Staggers, J.W. Blaz, Research on nursing handoffs for medical and surgical
tion. However, we have identified several nuances of settings: an integrative review, J. Adv. Nurs. 69 (2) (2013) 247–262.
[22] L. Baldwin, C. McGinnis, A computer generated shift report, Nurs. Manage. 25
communication that could result in concerted research efforts. (1994) 61–64.
[23] J. Cahill, Patient’s perceptions of bedside handovers, J. Clin. Nurs. 7 (1998)
Funding [24] B. Payne, H. Hardey, P. Coleman, Interactions between nurses during
handovers in elderly care, J. Adv. Nurs. 32 (2000) 277–285.
[25] S.M. Hohenhaus, S. Powell, J.T. Hohenhaus, Enhancing patient safety during
This research was supported in part by a grant from the James S. hand-offs: standardized communication and teamwork using the ‘‘SBAR”
McDonnell Foundation (Grant No. 220020152), and by a training method, Am. J. Nurs. 106 (8) (2006) 72A.
fellowship from the Keck Center AHRQ Training Program in Patient [26] J. Abraham, T.G. Kannampallil, V.L. Patel, A systematic review of the literature
on the evaluation of handoff tools: implications for research and practice, J.
Safety and Quality of the Gulf Coast Consortia (AHRQ Grant No. 1 Am. Med. Inform. Assoc. 21 (1) (2014) 154–162.
T32 HS017586-02). [27] J. Abraham et al., Comparative evaluation of the content and structure of
communication using two handoff tools implications for patient safety, J. Crit.
Care 29 (2) (2014) 311.e1–311.e7.
Conflict of interest [28] J. Blaz, N. Staggers, The format of standard tools for nursing handoff: an
integrative review, in: Proceedings of the Annual Conference of American
Medical Informatics Association (AMIA), Chicago, IL, 2012.
All the authors declare that they have no conflicts of interest [29] S.A. Collins et al., Content overlap in nurse and physician handoff artifacts and
with this work: none of the authors have any financial the potential role of electronic health records: a systematic review, J. Biomed.
relationships with any organizations that might have an interest Inform. 44 (4) (2011) 704–712.
[30] J. Abraham, T.G. Kannampallil, V.L. Patel, Towards an ontology for
in the submitted work in the previous three years; no other rela- interdisciplinary handoff communication in intensive care: implications for
tionships or activities have influenced the submitted work. tool resiliency and patient safety, in: International Symposium on Human
88 J. Abraham et al. / Journal of Biomedical Informatics 59 (2016) 76–88

Factors and Ergonomics in Healthcare, Human Factors and Ergonomics Society [66] G. Olson, J. Herbsleb, H. Reuter, Characterizing the sequential structure of
(HFES), Chicago, IL. interactive behaviors through statistical and grammatical techniques, Hum.–
[31] E.S. Patterson, R.L. Wears, Patient handoffs: standardized and reliable tools Comput. Interact. 9 (3 and 4) (1994) 427–472.
remain elusive, Joint Comm. J. Qual. Patient Safety 36 (2) (2010) 52–61. [67] T.G. Kannampallil et al., Understanding the nature of information seeking
[32] J.S. Carroll, M. Williams, T.M. Gallivan, The ins and outs of change of shift behavior in critical care: implications for the design of health information
handoffs between nurses: a communication challenge, BMJ Qual. Safety 21 (7) technology, Artif. Intell. Med. 57 (1) (2013) 21–29.
(2012) 586–593. [68] American Association of Critical Care Nurses, AACN: ESSENTIALS OF CRITICAL
[33] R. Wears, Poverty amid plenty, BMJ Qual. Safety 21 (7) (2012) 533–534. CARE ORIENTATION (ECCO), 2015, AACN.
[34] M.P. Kerr, A qualitative study of shift handover practice and function from a [69] A. Strauss, J. Corbin, Basics of Qualitative Research: Techniques and Procedures
socio-technical perspective, J. Adv. Nurs. 37 (2) (2002) 125–134. for Developing Grounded Theory, SAGE Publications, Thousand Oaks, CA, 1998.
[35] S. Lally, An investigation into the functions of nurses’ communication at the [70] O.R. Holsti, Content Analysis for the Social Sciences and Humanities, Addison-
inter-shift handover, J. Nurs. Manage. 7 (1) (1999) 29–36. Wesley, Reading, MA, 1969.
[36] R. Randell, S. Wilson, P. Woodward, The importance of the verbal shift [71] K. Krippendorff, Content Analysis: An Introduction to Its Methodology, Sage,
handover report: a multi-site case study, Int. J. Med. Inform. 80 (11) (2011) Newbury Park, CA, 1980.
803–812. [72] J. Apker et al., Exploring emergency physician–hospitalist handoff
[37] C.K. Holland, An ethnographic study of nursing culture as an exploration for interactions: development of the handoff communication assessment, Ann.
determining the existence of a system of ritual, J. Adv. Nurs. 18 (9) (1993) Emerg. Med. 55 (2) (2010) 161–170.
1461–1470. [73] R. Wears, Discourse and process analysis of shift change handoffs in
[38] F. Strange, Handover: an ethnographic study of ritual in nursing practice, emergency departments, in: Proceedings of the Human Factors and
Intens. Crit. Care Nurs. 12 (2) (1996) 106–112. Ergonomics Society, vol. 54(12), 2010, pp. 953–956.
[39] M.K. Anthony, G. Preuss, Models of care: the influence of nurse communication [74] S.R. Fussell, R.E. Kraut, J. Siegel, Effects of shared visual context on
on patient safety, Nurs. Econ. 20 (5) (2002) 209–215. collaborative work, in: Proceedings of the Conference on Computer
[40] P.K. Patterson et al., Nurse information needs for efficient care continuity Supported Cooperative Work (CSCW), ACM Press, 2000.
across patient units, J. Nurs. Adm. 25 (10) (1995) 28–36. [75] D. Gergle, R.E. Kraut, S.R. Fussell, Action as language in a shared visual space,
[41] M.M. Hays, The phenomenal shift report: a paradox, J. Nurs. Staff Dev. 19 in: Proceedings of the Conference on Computer Supported Cooperative Work
(2003) 25–33. (CSCW), 2004.
[42] B.A. Nelson, R. Massey, Implementing an electronic change-of-shift report [76] R.E. Kraut, M.D. Miller, J. Siegel, Collaboration in performance of physical tasks:
using transforming care at the bedside processes and methods, J. Nurs. Adm. effects on outcomes, in: Proceedings of Conference on Computer Supported
40 (4) (2010) 162–168. Cooperative Work (CSCW), ACM Press, 1996, pp. 57–66.
[43] A.M. Jukkala et al., Developing a standardized tool to improve nurse [77] H.H. Clark, E.F. Schaefer, Collaborating on contributions to conversations, Lang.
communication during shift report, J. Nurs. Care Qual. 27 (3) (2012) 240–246. Cogn. Process. 2 (1) (1987) 19–41.
[44] L. Wentworth et al., SBAR: electronic handoff tool for noncomplicated [78] H. Berkenstadt et al., Improving handoff communications in critical care:
procedural patients, J. Nurs. Care Qual. 27 (2) (2012) 125–131. utilizing simulation-based training toward process improvement in managing
[45] M. Roberts, J. Putnam, G.H. Raup, The interdepartmental ticket (IT) factor: patient risk, Chest 134 (1) (2008) 158–162.
enhancing communication to improve quality, J. Nurs. Care Qual. 27 (3) (2012) [79] C.A. Bowers et al., Analyzing communication sequences for team training
247–252. needs assessment, Hum. Factors 40 (4) (1998) 672–680.
[46] A.J. Spooner et al., Understanding current intensive care unit nursing handover [80] H. Eide et al., Physician–patient dialogue surrounding patients’ expression of
practices, Int. J. Nurs. Pract. 19 (2) (2013) 214–220. concern: applying sequence analysis to RIAS, Soc. Sci. Med. 59 (1) (2004) 145–
[47] C.S. Priest, S.K. Holmberg, A new model for the mental health nursing change 155.
of shift report, J. Psychosoc. Nurs. Ment. Health Serv. 38 (8) (2000) 36–43. [81] C. Zimmermann, L. Del Piccolo, M. Mazzi, Patient cues and medical
[48] A. Sexton et al., Nursing handovers: do we really need them?, J Nurs. Manage. interviewing in general practice. examples of the application of sequential
11 (1) (2004) 37–42. analysis, Epidemiologia e Psichiatria Sociale 12 (2003) 115–123.
[49] B. Hilligoss, Selling patients and other metaphors: a discourse analysis of the [82] A.C.R. Streeter, What Nurses Say Communication Behaviors Associated with
interpretive frames that shape emergency department admission handoffs, the Competent Nurse Handoff, University of Kentucky, 2010.
Soc. Sci. Med. 102 (2014) 119–128. [83] D.J. Solet et al., Lost in translation: challenges and opportunities in physician-
[50] Y. Xiao, Artifacts and collaborative work in healthcare: methodological, to-physician communication during patient handoffs, Acad. Med. 80 (12)
theoretical, and technological implications of the tangible, J. Biomed. Inform. (2005) 1094–1099.
38 (1) (2005) 26–33. [84] H.H. Clark, S.A. Brennan, Grounding in communication, in: L.B. Resnick, J.M.
[51] T.G. Kannampallil et al., Comparing the information seeking strategies of Levine, S.D. Teasley (Eds.), Perspectives on Socially Shared Cognition, APA
residents, nurse practitioners, and physician assistants in critical care settings, Books, Washington DC, 1991.
J. Am. Med. Inform. Assoc. 21 (2014) e249–e256. [85] L. Groah, Handoffs – a link to improving patient safety, Assoc. Perioperative
[52] R. Bakeman, J.M. Gottman, Observing Interaction: An Introduction to Nurs. J. 83 (2006) 227–230.
Sequential Analysis, first ed., Cambridge University Press, New York, NY, 1986. [86] M.D. Cohen, B. Hilligoss, A.C. Kajdacsy-Balla Amaral, A handoff is not a
[53] J.W. Tukey, Exploratory Data Analysis, Pearson, 1977. p. 688. telegram: an understanding of the patient is co-constructed, Crit. Care 16 (1)
[54] H. Sacks, Lectures on Conversation, in: G. Jefferson (Ed.), University Press, New (2012) 303.
York, 1992. [87] E.A. Greenstein et al., Characterising physician listening behaviour during
[55] H. Sacks, S. Emmanuel, G. Jefferson, The simplest systematics for turntaking in hospitalist handoffs using the HEAR checklist, Qual. Safety Health Care 22
conversation, Language 50 (1977) 696–735. (2013) 203–209.
[56] A. Pomerantz, B.J. Fehr, Conversation analysis an approach to the study of [88] D. Flemming, U. Hübner, How to improve change of shift handovers and
social action as sense making practices, Discourse Soc. Interact. 2 (64–91) collaborative grounding and what role does the electronic patient record
(1997). system play? Results of a systematic literature review, Int. J. Med. Inform. 82
[57] R. Cameron, J. Williams, Sentence to ten cents: a case study of relevance and (7) (2013) 580–592.
communicative success in nonnative-native speaker interactions in a medical [89] N. Sebanz, H. Bekkering, G. Knoblich, Joint action bodies and minds moving
setting, Appl. Linguist. 18 (4) (1997) 415–445. together, Trends Cogn. Sci. 10 (70–76) (2006).
[58] E. Kaner et al., Medical communication and technology: a video-based process [90] B. Bahrami et al., Optimally interacting minds, Science 329 (2010) 1081–1085.
study of the use of decision aids in primary care consultations, BMC Med. [91] B. Bahrami et al., Together, slowly but surely: the role of social interaction and
Inform. Decis. Mak. 7 (2) (2007). feedback on the build-up of benefit in collective decision-making, J. Exp.
[59] A. Parush et al., Communication and team situation awareness in the OR: Psychol. Hum. Percept. Perform. 38 (2011) 3–8.
implications for augmentative information display, J. Biomed. Inform. 44 (3) [92] M.D. Cohen, P.B. Hilligoss, The published literature on handoffs in hospitals:
(2011) 477–485. deficiencies identified in an extensive review, Qual. Safety Health Care 19 (6)
[60] T. Kannampallil et al., Making sense: sensor-based investigation of clinician (2010) 493–497.
activities in complex critical care environments, J. Biomed. Inform. 44 (3) [93] C.D. Frith, The role of metacognition in human social interactions, Philos.
(2011) 441–454. Trans. Roy. Soc. B: Biol. Sci. 367 (1599) (2012) 2213–2223.
[61] P.M. Sanderson, C. Fisher, Exploratory sequential data analysis: foundations, [94] D. Morrow, M. Rodvold, A. Lee, Nonroutine transactions in controller–pilot
Hum.–Comput. Interact. 9 (3 and 4) (1994) 251–317. communication, Discourse Process. 17 (2) (1994) 235–258.
[62] K. Zheng, R. Padman, M.P. Johnson, An interface-driven analysis of user [95] E. Coiera, When conversation is better than computation, J. Am. Med. Inform.
interactions with an electronic health records system, J. Am. Med. Inform. Assoc. 7 (3) (2000) 277–286.
Assoc. 16 (2) (2009) 228–237. [96] J. Abraham, T.G. Kannampallil, K.F. Almoosa, Metrics for evaluating the quality
[63] H.H. Clark, Using Language, Cambridge University Press, Cambridge, MA, 1996. of handovers, JAMA Int. Med. 175 (4) (2015) 654–655.
[64] D. Morrow, A. Lee, M. Rodvold, Analyzing problems in routine controller–pilot [97] J. Hernan, D.J. Pesut, L.P. Conard, Using metacognitive skills: the quality audit
communication, Int. J. Aviat. Psychol. 3 (1993) 285–302. tool, Nurs. Diag. 5 (2) (1994) 56–64.
[65] E. Montague et al., Modeling eye gaze patterns in clinician–patient interaction [98] L. Chartier, Use of metacognition in developing diagnostic reasoning skills of
with lag sequential analysis, Hum. Factors: J. Hum. Factors Ergonom. Soc. 53 novice nurses, Nurs. Diag. 12 (2) (2001) 55–60.
(5) (2011) 502–516. [99] R.M. Frankel et al., Context, culture and (non-verbal) communication affect
handover quality, BMJ Qual. Safety 21 (Suppl. 1) (2012) i121–i128.