Inpatient Cross-Cover Consensus Recommendations for Medical and Surgical Residents: A Delphi Analysis

Medical and surgical residents regularly provide care to hospitalized patients for whom they are not the primary provider. This practice, known as cross-cover,¹  is widespread among residency programs in the United States, in part due to program compliance with the Accreditation Council for Graduate Medical Education (ACGME) work hour restrictions.^2–4  In a shift-based day team and night team model, residents who are assigned to the night team provide cross-cover care for patients admitted by the primary day team.⁵  Cross-cover care also occurs routinely during intrateam handoffs between residents during the day.⁶  A multisite study demonstrated that the primary intern assigned to a patient is available in the hospital for only 39% of a patient's hospital stay.⁷  Hence, the majority of a patient's time in the hospital is spent being cared for by a cross-covering physician. Despite this, there appear to be no recommendations regarding best practices or standards of care for resident cross-coverage.

Cross-coverage creates unique challenges; the cross-covering resident often has not met the patient or participated in generating the care plan during daily rounds. When fielding pages about these patients, cross-covering residents must make decisions regarding when a patient should be seen at the bedside, when an issue can be addressed by placing an order in the electronic health record, and when a task can be deferred to the primary team. The cross-covering resident must also have a sense of when to notify the supervising faculty member, and what additional documentation may be needed. Many graduating medical students have not received formal training in cross-cover care,⁸  yet this is a skill they are expected to perform proficiently on the first day of residency.

The purpose of this study was to define key components of safe and efficient cross-cover care based on input from faculty and chief residents in medical and surgical fields.

We invited 43 medicine physicians and 17 surgeons representing 11 academic institutions in the Midwest region of the United States to participate in September 2017. This included practicing academic hospitalists (n = 24), medical and surgical subspecialists (n = 21), chief internal medicine residents (n = 9), and chief surgery residents (n = 6). All physicians invited were contacts of the authors and were selected because of their leadership roles in undergraduate or graduate medical education or because of their direct cross-cover experience. Direct cross-cover experience indicates that the physicians were responsible for performing cross-coverage themselves (ie, holding a cross-cover pager or telephone). We intentionally sought to include chief residents and hospitalists because these roles involve routine direct cross-cover experience. We invited 60 participants in order to obtain a goal of at least 20 responses, because a panel of 15 to 30 participants is recommended to effectively perform the Delphi technique.⁹ 

We used a modified Delphi technique⁹  consisting of 2 anonymous electronic surveys created in Qualtrics (Qualtrics LLC, Provo, UT). The process is provided as online supplemental material.

In the first survey, respondents were asked the following free-text questions:

1. When should a resident evaluate a patient at bedside?

2. When is it appropriate to defer a question or concern to the primary team?

3. When should a note about cross-coverage be documented in the medical record?

4. When should an attending be contacted overnight about a cross-cover patient?

5. What aspects of communication with nurses are important for safe and efficient cross-cover care?

The authors developed these questions based on their clinical experience, and they categorized the responses based on themes. In response to the question on physician-nurse communication, 3 subsets were created: (1) physician communication with nurses, (2) nurse communication with physicians, and (3) shared features of both physician and nurse communication.

The respondents were also asked if they had direct cross-cover experience over the past year.

In the second survey, respondents were asked to review each categorized response and provide the frequency using a 5-point Likert scale (from 1, never, to 5, every time) for the following: when the patient should be evaluated at the bedside, the issue should be deferred to the primary service, additional documentation should be completed, or the supervising faculty member should be contacted.

Respondents were also asked to evaluate communication elements for their importance in cross-coverage using a 5-point Likert scale (from 1, not at all important, to 5, very important). The survey was developed by the authors without further testing.

Agreement was defined as positive or negative. Positive agreement meant that the majority of respondents chose either a 4 or 5 on the Likert scale; negative agreement meant that the majority of experts chose either a 1 or 2 on the Likert scale.¹⁰  High consensus was defined as ≥ 80% agreement among the respondents, approaching consensus was defined as 51% to 78% agreement, and nonconsensus was defined as ≤ 50% agreement.¹⁰ 

The study was determined to be exempt by the University of Michigan Institutional Review Board.

Agreement in each domain was analyzed by specialty (medicine versus surgery) and by clinical experience with direct cross-cover (none versus any in the past year). To determine differences, a Pearson χ² test or a Fisher exact test was completed depending on the number of response categories. A Bonferroni adjustment for multiple comparisons was completed for each domain. JMP Pro 13.1.0 (SAS Institute Inc, Cary, NC) was used for analysis.

Of the 60 invited physicians, 40 (67%) agreed to participate, including 24 internal medicine physicians (60%) and 16 surgeons (40%) representing 8 academic institutions across the Midwest. This group included 10 chief residents, 14 hospitalists, 8 current or former residency program or assistant program directors, 6 current or former clerkship directors, and 10 physicians who hold inpatient leadership positions or medical school faculty roles (n = 7). A total of 22 respondents (55%) reported no direct cross-cover experience in the past year, whereas 18 (45%) had provided cross-coverage. A total of 30 invited physicians (50%) completed the second part of the Delphi survey; this was composed of 18 internal medicine physicians (60%) and 12 surgeons (40%). Of the 30 respondents, 12 (40%) had no cross-cover experience within the past year, whereas 18 (60%) had provided direct cross-cover care within the past year.

Categorization of free text responses identified 47 total categories (provided as online supplemental material). One category—medical error—was inadvertently not included in Delphi survey round 2.

Twenty-eight items achieved high consensus (tables 1 and 2). Several items achieved 100% consensus: residents should evaluate a patient at bedside whenever asked by the nurse; documentation should occur for change in level of care (ie, transfer to ICU), death, a code, or when the rapid response team is activated; and physician-nurse communication should be respectful and closed loop (acknowledging information and repeating back to confirm accuracy).

Eight approached consensus (tables 2 and 3). There were 10 items that did not achieve consensus (table 4).

There was no difference in consensus based on medical versus surgical specialty or those faculty with versus without direct cross-cover experience in the past year (provided as online supplemental material).

This modified Delphi survey study using Midwestern medical and surgical physician faculty and chief residents demonstrated high agreement in several areas for cross-cover care in the hospital setting. There was 100% agreement seen in several categories: residents should evaluate a patient at bedside when asked by the nurse; documentation should occur for change in level of care, death, a code, or when the rapid response team is activated; and physician-nurse communication should be respectful and closed loop. High consensus was also seen in other areas: for example, attendings should be contacted for change in level of care, death, code activation, or unanticipated discharges. Of the areas that achieved consensus, there was no statistically significant difference in item ranking based on medical or surgical specialties or based on prior cross-cover experience, suggesting that these cross-cover care features are common across diverse inpatient settings.

Our results highlight clinical scenarios in which supervising faculty would like to be contacted by a resident; these are consistent with previous findings.¹¹  Prior work also suggests that a preexisting relationship between a supervisor and trainee is necessary for these communications to occur.¹²  However, in a shift-based model of care it is possible that an attending has not met or has not had significant contact with the cross-covering resident. Hence, the use of explicit and specific recommendations may improve patient safety.¹¹  Trainees also must be able to identify key communication principles when working with nursing colleagues. Past work has demonstrated that communication with nurses in the hospital setting is inadequate,^13,14  but evaluation of communication specifically during times of cross-cover is a relatively unexplored area. One study evaluated resident, attending, and nurse perceptions of night float rotations and found that nurses perceived suboptimal physician-nurse communication and continuity of care in these cross-cover situations.¹⁵  This is not surprising given that the cross-cover mindset can lead to a lack of ownership of a patient.¹  As front-line care providers, nurses may be the first to detect this difference in the clinical approach. Our findings reinforce the value of communication, particularly situational awareness and closed-loop communication.¹⁶ 

Our study has limitations. Training programs that participated in our study were all located within the Midwestern United States, and it is possible that the physician consensus represents a regional perspective. We sought only the viewpoints of physicians, not other members of the interprofessional care team. A significant proportion of respondents did not have cross-cover experience within the past year, and may not be fully informed about cross-cover issues. Finally, the survey was developed without testing, hence respondents may have interpreted questions differently than the authors intended.

Future steps involve seeking input from other health care professionals, specifically the nursing perspective. Currently, we are working to integrate the consensus statements into our medical school and residency curricula and to evaluate the impact. To facilitate implementation, we have created a final list of the high consensus cross-cover recommendations (provided as online supplemental material). We believe that institutional adaptation of cross-cover recommendations will help improve the transparency of expectations regarding attending supervision of significant cross-cover events and help to align the desire of faculty to be involved in cross-cover decisions with the notifications they receive from trainees in real time.

We have provided a set of general recommendations regarding cross-cover care that had high consensus among a panel of physicians from medical and surgical specialties. Implementation of these recommendations in the educational curricula for undergraduate and graduate medical education can offer a shared foundation for providing care in this role.