Exploring Emergency Physician–Hospitalist Handoff Interactions: Development of the Handoff Communication Assessment

Introduction 

Materials and Methods 

Results 

Participant volunteers consisted of 20 physicians (12 emergency physicians, 8 hospitalists). Emergency physicians had a median of 16 years' postgraduate experience (range 1 to 27 years), and their median age was 48 years (range 34 to 58 years). Hospitalists had a median of 6 years' postgraduate experience (range 1 to 18), and their median age was 37.5 years (range 32 to 48 years).

We obtained 24 recordings. Of these, we coded 15 conversation transcripts because technical difficulties in audio-recording made 9 transcripts unusable (background noise made conversations unintelligible, voices too faint, recording malfunctions). Emergency physicians initiated all handoffs and all handoff calls studied were specifically for the purpose of admission. Interactions ranged from 1 to 3 minutes and contained a median of 53 utterances, with a range of 24 to 105 utterances. A total of 742 utterances were coded. Eighty-five utterances were uncodable because of background noise and interruptions.

The Handoff Communication Assessment instrument is presented in Table 1. The instrument consists of 2 typologies. One typology is composed of 11 content categories representing topics of patient presentation (a description of the patient, consisting of identifiers, history, symptoms, and past procedures), assessment (statements about future treatment, clinical impression, prognosis, outcome, admission status, and transfer of responsibility), and professional environment (descriptive or evaluative talk about the clinical environment, such as logistics and bed availability, as well as courtesy comments). The other typology is made up of 11 language form categories representing information-seeking (eg, closed, open, and clarifying questions/requests), information-giving (eg, description, explanation, social amenities) and information-verifying (such as read-back) communication behaviors. Each utterance has a content and language form category. Interrater reliability using Cohen's κ was sufficient for the content (0.71) and language form (0.84) categories.25, 26, 27 Table 2 presents an example demonstrating how we coded handoff discourse to test the Handoff Communication Assessment categories.

Table 1. Utterance coding structure and frequency of occurrence.

	Coding Category	Operational Definition	Example Utterances	No. (N=742)	%
	Content
	Patient presentation
	Patient identifiers	Statements that convey patient's specific room location or name.	“I'm calling about patient [name] in bed 8.” “What room is she [patient] in?”	59	8.0
	History	The patient's past medical problems/conditions that are pertinent to the current diagnosis or clinical impression. Includes events that occurred before patient presented to ED. Patient comments about history. Physician comments about patient personality, lifestyle, demeanor.	“Patient is a 51-year-old male with a history of high cholesterol, high blood pressure, and some diabetes.” “Patient said he had trouble walking up stairs.”	113	15.2
	Symptom	Descriptions, explanations that address current clinical scenario in regard to providing information about symptoms of concern (medical or psychosocial).	“[Patient] presents here with some upper gastric and right flank pain.”	76	10.2
	Procedure	Statements about pertinent laboratory data, pertinent test results, medications, and evaluation that have already been performed to address the patient's current condition.	“Ultrasound showed no blockage or dilation of the biliary system. CT does show acute pancreatitis with possible necrosis in the tail.”	76	10.2
	Assessment
	Treatment	Statements about future medical procedures to be taken, if such steps are deemed necessary, to address the patient's current problem.	“Obviously, [the patient] needs to be monitored very closely tonight.”	59	8.0
	Clinical impression	Identification of the current clinical impression, naming the problem or reasons for the problem.	“I'm a little suspicious that he may have a stuttering gall bladder the way he's [taking] Vicodin.”	34	4.6
	Prognosis	Probabilistic statement about patient's future condition, based on completed or proposed treatments.	“I think it's best if you take him. I don't think he'll be able to take care of himself.”	4	0.5
	Outcome	Definitive utterances about the result of the handoff: accept, not accept (eg, check with another service), wait and see.	“Okay. So [patient] will be the only one then. Okay?” “Okay. I'll get him.”	24	3.2
	Transfer of responsibility	Statements about what was being asked of the hospitalist (eg, patient admission, clinical consult, other reason).	“Sounds to me you can't get anybody else to [admit] him [patient].” “Do you know if they [surgery] are going to take [patient] or should I keep him active on my list?”	30	4.0
	Professional environment
	Logistic processes	Descriptive or evaluative talk about logistic or procedural issues in ED, hospital, or greater health care system.	“I'm [behind] right now. I walked in at 4:15 PM and [received] 7 patients and then the ER called again with another admission.”	98	13.2
	Courtesies	Statements such as thank you, affirming/supporting remarks (okays/yeses), greeting and closing remarks that provide a context of professional courtesy.	“Hi. How are you?” “Thanks a lot. Bye.”	169	22.8
	Total			742	100
	Language form
	Information seeking
	Closed question	An utterance that is a question designed to solicit specific information.	“What did the CAT scan show?” “Did you talk to radiology?”	56	7.5
	Open question	An utterance that is a question designed to solicit information in a manner that affords the respondent any opportunity to elaborate.	“Why did they send him over? Is he worse symptomatically?”	3	0.4
	Clarifying question/request	An utterance that is a question designed to seek clarification of another's immediately preceding utterance. May take the form of a request.	“She [patient] came in with abdominal pain?”	7	0.9
	Information giving
	Description	The primary function of these utterances is pure description about the patient. Description of observations about the patient's past or present condition/state of affairs.	“91-year-old female came in complaining of low back pain and abdominal pain coming around her abdomen. Has not had a fever.”	229	30.9
	Explanation	Utterances that state the facts and make an inference about the patient. A physician explains the patient's condition and draws a conclusion or medical judgment based on data.	“Did x-rays and didn't see anything obvious and urine doesn't show significant urinary tract infection.”	94	12.7
	Rationale	A justification is offered to account for any medical procedures, tests, medications, or recommendations concerning the patient. The intent is to justify why an action has been taken or will be taken in the future.	“My big concern is that you [remove] 5 to 6 liters off [patient]. I'm a little worried about electrolyte shift.”	28	3.8
	Directive	Advisements, orders, or recommendations that inform/guide patient evaluation, treatment, and disposition.	“I [would be] comfortable if he was observed at least overnight.”	27	3.6
	Context talk	Talk about contextual issues in clinical environment such as logistics and procedures.	“It's just that I'm not going to get to him for a while. I do have two PAs so it's probably going to take us 2 to 3 hours to get caught up.”	107	14.4
	Social amenities	Utterances in which physicians exchange courtesies and talk that tells the sender that the receiver is paying attention.	“Thanks,” “Okay,” “Uh-huh”	158	21.3
	Decision	Utterances in which the physician accepts or does not accept the patient. May be directly stated or implied.	“Okay, all right, we'll get him.”	30	4.0
	Information verifying
	Read-back	Statements that paraphrase or restate what the other has said. A physician summarizes the gist of what has been said.	“What room is he in?” “Seven.” “Seven.”⁎	3	0.4
	Total			742	100

Handoff Communication Assessment categories derived from discourse coding established in physician-patient communication literature.17, 26, 28, 29, 30, 31, 32, 33, 34 We expanded the coding categories to physician-physician handoff communication by using our previous work17 and handoff research in medicine, health communication, and health systems engineering.1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15

⁎ Example of read-back.

Table 2. Example call coded by utterance.

	Utterance	Physician	Language Form	Content
	H3: okay hi	H	Social amenities	Courtesies
	Emergency physician 1: hi how are you?	E	Closed question	Courtesies
	H3: good	H	Social amenities	Courtesies
	Emergency physician 1: I have got a gentleman here	E	Description	Logistic processes
	[background noises and voices]
	named P14	E	Description	Patient identifier
	a patient of Dr.[outside physician]	E	Description	Logistical processes
	H3: um-hum	H	Social amenities	Courtesies
	Emergency physician 1: he has a history of COPD	E	Description	History
	H3: [um-hum]	H	Social amenities	Courtesies
	Emergency physician 1: he also has a cardiac history	E	Description	History
	he came in for chest pain	E	Description	Symptom
	and shortness of breath	E	Description	Symptom
	H3: [um-hum]	H	Social amenities	Courtesies
	Emergency physician 1: and to make a long story short he's still wheezing quite considerably	E	Description	Procedure
	H3: [um-hum]	H	Social amenities	Courtesies
	Emergency physician 1: his chest pain workup has turned out to be negative	E	Description	Procedure
	H3: [um-hum]	H	Social amenities	Courtesies
	Emergency physician 1: in fact he's been seen by cardiology	E	Description	Procedure
	but he still continues to wheeze	E	Description	Symptom
	he looks like a heavy smoker	E	Description	History
	Emergency physician 1: I mean he has nicotine staining on his fingers and so on and so forth	E	Description	History
	he is on home oxygen	E	Description	History
	Emergency physician 1: his BNP is 165	E	Description	Symptom
	it was 380 last admission	E	Description	History
	so he's not in failure	E	Description	Symptom
	Emergency physician 1: his pulse ox is reasonable	E	Description	Symptom
	but he still is wheezing	E	Description	Symptom
	and you know he has come back fairly frequently	E	Description	History
	he did have a stress test 8 days ago that was negative	E	Description	History
	so I think it's basically an exacerbation of COPD	E	Explanation	Symptom
	H3: [ah-huh]	H	Social amenities	Courtesies
	Emergency physician 1: I think that's what's causing his chest pain too	E	Explanation	Symptom
	H3: what room?	H	Closed question	Patient identifier
	Emergency physician 1: he's in room 6	E	Description	Patient identifier
	H3: room 6	H	Description	Patient identifier
	thank you [background noise makes it hard to hear this response]	H	Social amenities	Courtesies
	Emergency physician 1: thank you	E	Social amenities	Courtesies

H, Hospitalist; E, emergency physician; COPD, chronic obstructive pulmonary disease.

Table 1 provides frequencies and percentages for all coding categories. Regarding content, patient presentation issues constitute the majority of handoff talk (43.6%), followed by physician discussion of the professional environment (36%) and assessment issues (20.3%). Regarding language form, information-giving utterances (90.7%) far surpass information-seeking utterances (8.8%), with very few information-verifying utterances (0.4%). Because information giving is the dominant language form, we took a closer look at these utterances, finding that physicians tended to mostly use description (30.9%) (discussion of the patient's past and present condition, followed by social amenities [21.3%]), where physicians show courtesies (eg, “hellos,” “thank yous”), as well as demonstrate they are paying attention (eg, “uh-huhs,” “okays”) and context talk (14.4%) (discussion of clinical environment issues such as logistics and procedures). Closed, open, and clarifying questions/requests accounted for less than 10% of all utterances, and physicians did not specifically prompt each other for questions.

Read-back, the repeating back of information offered in an attempt to ensure precise communication, was observed in only 3 utterances (all 3 belonged to hospitalists). Regarding admission decisions, in 6 of the 15 study calls, hospitalists used indirect statements that appeared to tacitly accept patients (eg, “Thanks a lot,” or “Okay”), most frequently by asking for the patient's location or room number. One handoff had no clear resolution.

Examination of conversational roles (Table 3) revealed that emergency physicians talk more in handoffs (502 utterances, 67.7% of all utterances) compared with hospitalists (240 utterances, 32.3% of all utterances). In regard to content categories, emergency physicians showed talk dominance—discursive control over the conversation—in patient presentation (87.6%), assessment (61%), and logistic processes (60%). Hospitalists more actively use courtesies (62%). Sixty-six percent of outcome utterances and 56% of responsibility transfer utterances were offered by hospitalists. Hospitalists communicated a total of 20 decision utterances compared with 10 by emergency physicians. Regarding language form categories, hospitalists asked 71.2% of all questions. Hospitalist questions typically concerned patient identifiers (name or location) and information about patient symptoms, procedures, and treatments.

Table 3. Frequencies and proportions of content and language form utterances by physician role.

	Categories	Emergency Physicians		Hospitalists		Totals
		f	Proportion	f	Proportion
	Content
	Patient presentation
	Patient identifiers	37	0.07	22	0.09	59
	History	107	0.21	6	0.03	113
	Symptoms	70	0.14	6	0.03	76
	Procedure	70	0.14	6	0.03	76
	Assessment
	Treatment	42	0.08	17	0.07	59
	Clinical impression	28	0.06	6	0.03	34
	Prognosis	4	0.01	0	0.00	4
	Outcome	8	0.02	16	0.07	24
	Transfer of responsibility	13	0.03	17	0.07	30
	Professional environment
	Logistical processes	59	0.12	39	0.16	98
	Courtesies	64	0.13	105	0.44	169
	Total utterances	502		240		742
	Language form
	Information seeking
	Closed question	13	0.03	43	0.18	56
	Open question	1	0.002	2	0.01	3
	Clarifying question/request	5	0.01	2	0.01	7
	Information giving
	Description	219	0.44	8	0.03	229
	Explanation	89	0.18	5	0.02	94
	Rationale	22	0.04	6	0.03	28
	Directive	17	0.03	10	0.04	27
	Context talk	68	0.14	39	0.16	107
	Social amenities	56	0.11	102	0.43	158
	Decision	10	0.002	20	0.08	30
	Information-verifying
	Read-back	0	0.00	3	0.01	3
	Total utterances	502		240		742

Limitations 

Although this study provides a positive first step in assessing actual emergency physician–hospitalist handoff communication, we acknowledge several limitations. The Handoff Communication Assessment was developed and tested by the same investigators, using only emergency physician–hospitalist handoff interactions collected at the same, single center. We recognize that other handoffs may have a very different structure and content, even within the same institutions. As such, there may be assumptions built into the Handoff Communication Assessment that are unrecognized or open to disagreement. We used the Handoff Communication Assessment to analyze 15 handoffs; we recognize our results are not representative of all handoffs at the study institution, nor are they generalizable to other contexts. The handoff was evaluated as a series of utterances, but ultimately the handoff is a conversational unit, and the unit as a whole needs to be explored in future research. Finally, our study was limited to telephone conversations, some difficult to hear because of recording quality. We did not collect data on additional handoff conversations that may have occurred in other channels (eg, face to face, computer-mediated communication technology).

Discussion 

Handoff communication is a central activity in emergency medical care, yet there is a lack of sufficient research that studies actual ED physician handoff discourse. Thus, we developed the Handoff Communication Assessment to provide a way to analyze the content and language of handoffs. This study was grounded in the principles of discourse analysis: to explore uses of language by studying the central themes and functions of verbal communication in action.31 Discourse analysis techniques enabled us to take the necessary first step of describing what is said in handoffs and how information is communicated. We used several well-established discourse coding tools in patient-physician communication to derive the Handoff Communication Assessment.21, 22, 23, 24, 25, 26, 27, 28, 29, 30 These existing tools provided an excellent foundation for Handoff Communication Assessment coding categories, yet none specifically addressed the content and linguistic forms found in the unique context of physician-physician handoffs. The Handoff Communication Assessment was subsequently created to fill an identified research gap. This tool had good interrater reliability for both content and language form.

In this preliminary study of a single hospital, our analysis of handoffs using the Handoff Communication Assessment suggests that handoffs leave little space for collaboration or questioning. Emergency physicians talk the most in a handoff communication; hospitalists functioned more in a listening mode. Emergency physicians convey information and sometimes participate in decisionmaking; hospitalists communicate their decisions and occasionally ask questions. Study handoffs lacked regular read-back or information verifying utterances, as well as explicit statements accepting patient transfer. These findings draw attention to the complexity of interpersonal communicative activity that constitutes medical environments.22, 23, 24, 25, 26, 27, 31 Handoffs are not simply modes of information transfer; rather, they are active discursive exchanges that require involvement of multiple individuals using a wide range of communicative behaviors.34

In previous research, we found that emergency physicians expect handoffs to emphasize immediate care decisions, whereas hospitalists expect handoffs to convey information necessary for long-term inpatient care.28 In the current study, approximately two thirds of all utterances focused on patients' immediate conditions, likely to justify admission. Postpresentation utterances consisted mainly of courtesies and professional environment issues rather than postadmission care. In our previous study, emergency physicians reported dissatisfaction with the lack of definitive admission decisions from hospitalists. Emergency physicians believed it contributed to handoff ambiguity.28 In the present study, we found a preponderance of indirect acceptance utterances, and this type of communication could exacerbate physician uncertainty about future patient responsibility and care. Such ambiguity may also affect more than patient care, extending to the professional peer relationships. Communication scholarship shows that conveying ambiguous information affects the quality and quantity of interpersonal communication and can potentially undermine organizational relationships.35, 36

We also found that 90% of utterances were information giving and came from the emergency physician. Multiple factors may contribute to this finding: (1) the hospitalist initially knows nothing about the patient or the ED course; (2) the 1-sided structure promotes conversational efficiency; and (3) it may reduce ambiguity introduced by give and take. Although this “1-way street” has significant practical value, it may weaken the potential for back-and-forth questioning that could clarify areas of uncertainty,16 as well as limit participation by the hospitalist, who may offer a fresh set of interpretations about the patient.2, 12 Insights from communication theory and organizational studies also suggest that a “2-way street” approach may benefit handoffs. Scholars argue that dialogue—interactions that foster empathy, equity, and common ground—promotes collaboration that leads to enhanced decision, greater innovation, and heightened flexibility.35, 36, 37, 38 Such organizational characteristics are particularly desirable in high-reliability organizations such as EDs.14, 34

Although emergency physicians talk the most, it appears that hospitalists may have more control of the conversation. With a few “okays” and “uh-huhs,” the hospitalist puts pressure on the emergency physician to keep explaining and “selling” why the hospitalist should either treat or admit the patient. Emergency physicians may talk the most to “sell” the patient, but it is the silence of the hospitalist, who has ultimate decisionmaking authority, that controls the conversation. The hospitalist typically grants the admission request by asking for the patient room number and name (when it was given in the start of the conversation) as though the hospitalist has to be persuaded about the need for the admission before paying attention to the patient room and name. We argue that these conversational roles may reinforce hospitalists' legitimate power—the power associated with the importance of a particular position39—a particularly intriguing finding, given that we studied only attending-attending handoffs and that emergency physicians have more expert power because of their familiarity with the patient.39, 40 The power differential evident in the handoff may decrease openness between emergency physicians and hospitalists. Drawing on the literature in organizational communication, we know that lack of open communication can negatively affect organizational relationships in areas such as trust, feedback, and collaboration,35, 36 outcomes that have far-reaching outcomes for medicine. For example, reduced openness reduces admission of lack of experience, knowledge, or mistakes.34

Whether these findings will hold true in other settings remains to be explored. Nevertheless, the Handoff Communication Assessment tool has a number of uses for improving handoffs and, potentially, patient care. For example, Handoff Communication Assessment data may be useful in quality assurance reviews by providing insights into how specific communication behaviors play a role in medical errors. Joint Commission 2009 National Patient Safety Goals mandate that handoffs provide the opportunity to ask and answer questions.7 The Handoff Communication Assessment could be used to analyze handoff conversations for the frequency and type of information-seeking categories. Such data could demonstrate the need for improvement or provide evidence of achievement of the National Patient Safety Goals requirement. In addition, individual institutions could use Handoff Communication Assessment–generated information about discursive displays of power to better understand possible connections between decisionmaking authority and quality of care.

The Handoff Communication Assessment tool can also be used to teach and assess communication for medical students and physicians in training. For example, the Handoff Communication Assessment could function as a pretest/posttest assessment tool in conjunction with specific training interventions (eg, role-playing handoffs with attending physicians). Students and residents could also listen to audio-recordings of their actual handoff conversations and use the Handoff Communication Assessment in self-critique. Such education/training efforts provide students and residents with opportunities to master key communication skills so they are better prepared to communicate handoffs effectively.

The present study represents our initial effort to better understand the communication profile of handoffs. As such, we envision future refinements of the Handoff Communication Assessment. For example, the tool may benefit from addition, deletion, or modification of categories according to use at other centers. Other changes could include adding evaluative components, such as a 1 to 5 scale of communication indicators, to facilitate assessment of the quality of handoffs. We also see a need in subsequent research to study the unit of analysis beyond utterances to individual handoffs to capture the clinical “big picture” of handoff interactions. By looking at utterances clustered in a handoff event, researchers may be able to note some interesting differences or similarities across events, as well as gather data at multiple points within the same patient transfer.

Further tests of the tool's validity and reliability are also necessary. To this end, the Handoff Communication Assessment should be utilized in other settings, with both novice and experienced physicians, and using larger data sets consisting of different types of handoffs/cases. Models for emergency physician–hospitalist interactions vary by setting, so the inclusion of these settings would allow for broader generalizability of the Handoff Communication Assessment and its findings. Future research needs to also validate the Handoff Communication Assessment in other handoff channels beyond the telephone exchanges used in the current study. For example, studies could compare handoff communication found in face-to-face interactions and computer-mediated communication technology. Data sets generated across a larger sample of physicians, settings, and communication channels should allow for greater reliability and validity of the Handoff Communication Assessment and its findings.

In summary, we found that the Handoff Communication Assessment is an evaluation tool that can be used to more fully understand actual handoff communication. The Handoff Communication Assessment instrument enables analysis about what issues are discussed in handoffs and how physicians talk about those issues, yielding data that may be applied to explore the authentic discourse composing emergency physician–hospitalist handoff practice and revealing patterns of communication that might be subject to improvement. Future research in other hospital settings is needed to further assess and confirm the Handoff Communication Assessment.