“To Call or Not to Call” Communication Preferences Among Pediatric Hospital Medicine Team Leaders

We conducted a cross-sectional survey study at 6 institutions nationwide. The Institutional Review Boards at all sites deemed this study exempt. This work is part of a larger study investigating supervision preferences among SRs, fellows, and hospitalists and defining the ideal role of a PHM fellow on clinical service.²³  The findings presented here are distinct as they focus on communication preferences during specific clinical scenarios.

Similar to our prior work,²³  we applied self-determination theory as a conceptual framework since it involves the 3 needs of competency, relatedness, and autonomy, and when these needs are fulfilled, one is motivated to perform.^24,25  We queried how the competency of a learner (resident or fellow), relationship with the team (relatedness between SR, fellow, and/or hospitalist), and autonomy provided to these learners will impact one’s motivation and performance in the clinical setting. We considered these needs when designing our surveys as we wanted to learn more about how best to provide supervision and what motivates one to communicate or want to be contacted. We developed our surveys to explore the balance of supervision and autonomy, learner competency, and the relationships among attending hospitalists, fellows, and SRs, as all may impact communication preferences.^24,25 

We created 3 complementary surveys, 1 for each subject population: hospitalists, PHM fellows, pediatric SRs (Appendices 3–5). Also, similar to our prior methodology, we used Messick’s content and response process criteria to collect validity evidence.²⁶  Our surveys were modeled after prior research and surveys that were previously distributed among internal medicine residents and attending physicians.⁷  Our clinical scenarios were based off this previous survey work, using similar categories (transfer of care, diagnostics, therapeutics, adverse events), but we added a “consult” category and replaced some scenarios with pediatric situations.⁷  Our surveys and clinical scenarios were revised after review by the 6 site leads and an expert survey methodologist. Lastly, we conducted cognitive interviews with each survey population for a total of 25 interviews, which led to minor survey revisions, to ensure response process validity.^26,27 

The survey instruments included demographic questions including but not limited to: years of clinical experience, academic rank, number of weeks per year supervising residents and fellows on service, and completion of a PHM fellowship. This paper examines a series of questions in which: (1) hospitalists and fellows were asked about their level of agreement regarding the PHM fellow contacting the hospitalist for several clinical scenarios; (2) hospitalists, fellows, and SRs were asked to indicate their level of agreement regarding the SR contacting the PHM fellow for the same scenarios. These scenarios focused on communication regarding transfers of care during the day and overnight (such as transfer to or from the ICU, admissions, and discharges), diagnostic decisions, therapeutic decisions, consults, and adverse events. Although agreement was measured on a 4-point unipolar ordinal response scale in the survey, we dichotomized this scale for final statistical analyses, with the top-2 categories indicating agreement. We performed all statistical analyses with the ordinal response scale and the dichotomized version, confirming that dichotomization did not change study results.

Six institutions were chosen as a convenience sample and as study sites based on their differing fellowship program characteristics, geographic locations (ranging from the southeast, Midwest, and west coast), and hospital types (3 free-standing children’s hospitals and 3 children’s hospitals within larger adult systems). Fellowship programs had 1 to 3 fellows per year, and level of fellow supervision when completing hospital medicine rotations varied from direct to indirect to no supervision at 3 institutions.³  For the latter institutions, the fellow would be paired with an attending hospitalist during the beginning of their fellowship but then function independently for the remainder of fellowship as an instructor with billing privileges. At time of survey distribution, 2 programs had been in existence for 9 years, whereas the remaining 4 had existed between 2 and 7 years. None had been accredited by ACGME, as accreditation was not yet possible during this study. A summary of relevant institutional characteristics is provided in Appendix 1.

We included SRs, defined as second- through fifth-year postgraduate level residents in general or combined pediatric residency programs. We excluded chief residents because of variability among levels of responsibility and supervision at different institutions. We sent surveys to all attending physicians; however, following receipt of the surveys, we excluded 6 hospitalists who neither attend on resident teams nor supervise fellows. We included hospitalists who only attend on resident teams, because those hospitalists and the learners whom they supervise may interact with fellows in certain clinical situations, such as for admissions, consults, or overnight.

All individuals in each role category at the 6 study sites were invited to participate. We invited a total of 200 hospitalists (range between institutions: 6–54), 20 fellows (range between institutions: 1–9 fellows), and 380 SRs (range between institutions: 23–142) by e-mail to complete the REDCap websurvey between September and October 2019.²⁸  Two reminder emails were sent to nonrespondents. Participation was voluntary and anonymous; no incentives for participation were offered.

We calculated univariate descriptive statistics for hospitalist, fellow, and SR data to evaluate levels of agreement. We then assessed the bivariate relationships between 2 types of team members at a time (eg, SRs and fellows) and their agreement with a variety of clinical scenarios using χ² tests with Rao-Scott correction to account for the clustering of the data by institution. Such analyses revealed differences in agreement between respondents. Statistical significance was achieved at P < .05. All analyses were conducted using Stata 17.0 (StataCorp, 2021; Stata Statistical Software: Release 17).

Survey respondents included 106 of 200 hospitalists (53% American Association for Public Opinion Research [AAPOR] response rate 2), 20 of 20 fellows (100% AAPOR response rate 2), and 149 of 380 SRs (39% AAPOR response rate 2).²⁹  Demographic characteristics of survey respondents are presented in Table 1.

Percent agreement (strongly agree and agree on the original response scale), or the level of agreement, of hospitalists, PHM fellows, and SRs for different clinical scenarios are shown in Appendix 2. Notable results for each grouping of clinical scenarios (transfers and escalations of care, diagnostics and therapeutics, consults, navigating conflict and adverse events) are provided below.

Regarding survey results from PHM fellows and hospitalists, all fellows and most hospitalists (82% to 97%) agreed that fellows should make their attending hospitalist aware when transferring an existing patient to another service or to the ICU during the daytime (Appendix 2). This level of agreement did not hold true for these scenarios overnight. Fewer fellows (35%) and hospitalists (44%) felt that the PHM fellow should contact their attending hospitalist when a Rapid Response Team (RRT), or ICU consult equivalent, is called.

Fewer fellows and hospitalists also reported that it was necessary for the PHM fellow to make their attending hospitalist aware of initiation of heated high-flow nasal cannula (HHFNC) (40% fellows, 42% hospitalists) and of anticoagulation (20% fellows, 28% hospitalists).

The great majority of both fellows (90%) and hospitalists (88%) agreed that the fellow should communicate with the hospitalist before obtaining consultation from another subspecialty.

All fellows and hospitalists agreed the fellow should call their attending hospitalist if a patient or family is upset, regardless of cause, but there was less agreement for other adverse events. Only 45% of fellows but 77% of hospitalists preferred the fellow to call the hospitalist for any adverse event.

Figure 1 displays differences in agreement between PHM fellows and hospitalists regarding when the fellow should contact the hospitalist. Hospitalists generally preferred less communication during the day but more communication overnight than fellows for a variety of clinical situations (Fig 1). Our results also revealed that hospitalists do prefer to be notified less frequently when transferring a patient to a new service than expressed by fellows during the day (P < .05). They want to be notified more than desired by fellows if discharging a patient who is unexpectedly ready for discharge overnight (P < .05) and if a patient or family is upset, regardless of cause (P < .01) (Fig 1).

When considering survey results from SRs, fellows, and hospitalists, almost all respondents (97% to 100% of SRs, fellows, and hospitalists) agreed that the SR should make the fellow aware when transferring a PHM patient to a new service or to the ICU and when discharging a patient who unexpectedly met discharge criteria during the daytime (Appendix 2). Fewer survey respondents agreed that such communication was necessary for these scenarios overnight, with the exception being transferring a patient to the ICU, then 89% to 97% of respondents still agreed that the SR should contact the fellow. Nearly all SRs, fellows, and hospitalists (95% to 100%) agreed that SRs should make the fellow aware if an RRT is called (Appendix 2).

Almost all agreed that the SR should contact the fellow if a resident will be performing an invasive procedure on a stable patient (99% to 100%). Fewer respondents agreed communication should occur from the resident to the fellow when the SR is ordering a noninvasive diagnostic procedure (48% hospitalists, 40% fellows, 38% residents) or when discontinuing HHFNC (39% to 40% for all respondents) (Appendix 2).

Ninety-five percent of fellows and 86% of hospitalists but only 64% of SRs desired the SR to contact the fellow before obtaining consultation from another subspeciality (Appendix 2).

There was near universal agreement (99% to 100% of respondents) that if a family wants to leave against medical advice or experiences an adverse event, the SR should escalate to the fellow. Although fewer residents (74%), fellows (85%), and hospitalists (92%) agreed with the SR contacting the fellow when a patient is upset regardless of cause, the majority still preferred such communication to occur (Appendix 2).

Figure 2 displays differences in agreement between SRs and hospitalists regarding when the SR should contact the PHM fellow for specific scenarios; Fig 3 illustrates the differences in agreement between fellows and hospitalists for the same chain of communication. Hospitalists preferred more communication between the SR and fellow than expressed by residents for multiple scenarios, including transferring a patient to the ICU overnight (P < .05), placing a consult, initiating HHFNC, ordering behavioral restraints, and addressing an upset patient or family (all P < .01) (Fig 2). For most overnight scenarios, hospitalists also preferred more communication between the SR and fellow than expressed by fellows (Fig 3). They preferred more communication when admitting patients from the emergency department at the same institution (P < .05) but less communication when transferring a patient to a new service overnight (P < .05) (Fig 3). Fellows, however, wanted more communication between the SR and fellow for the latter scenario and before placing a consult compared with SRs (P < .05) (Appendix 6).

SRs, PHM fellows, and hospitalists on an inpatient medical team may have differing preferences regarding communication during certain clinical scenarios that can impact supervision, provision of autonomy, and escalation of care. Hospitalists generally prefer more communication between the SR and fellow and between the fellow and attending overnight than expressed by fellows and residents; thus, it is critical for supervisors to communicate their preferences to team members. Fellows may also consider calling their attending sooner overnight and for other specific situations, such as before obtaining consultation from another service. For the adverse event vignettes provided, communication from resident to fellow and fellow to attending was desired; however, hospitalists preferred even more communication than fellows and residents when a patient or family is upset, regardless of cause. It is also important to note that despite these differences in preferences, we did find agreement, such as when transferring a patient to a new service or to the ICU or when calling an RRT. Leaders of fellowship and residency programs should consider such perspectives when developing expectations for faculty, fellow and resident workflow, and when creating communication guidelines as mandated by ACGME.

Our prior work revealed that the majority of hospitalists, fellows, and SRs agreed the SR should always call the fellow first.²³  We also found that both hospitalists and fellows expressed a desire for more frequent communication when the attending did not round with the medical team. Although the majority of hospitalists and fellows perceived communication can be “as needed” overnight,²³  the current study added to these results by inquiring about specific scenarios and revealed agreement across the 3 surveyed populations varied depending on the situation. When asked about specific scenarios, the majority of hospitalists actually desired more communication overnight, especially for transfers to the ICU or before discharging a patient unexpectedly. Although our study did not explore why this is the case, contributing factors may include availability of 24/7 in-house nocturnist coverage, differences in expectations,³⁰  supervisors’ assessments of learner trustworthiness,³¹  institutional policies or guidelines around care escalation, medico-legal concerns, or personal demographics, such as years of experience or completion of a PHM fellowship; exploration of these factors are next steps worthy of investigation. In addition, during the daytime, hospitalists may be more frequently checking the electronic medical record and hearing updates during rounds, so unlike overnight, they are more aware of changes in patient status or plans during the day. Farnan et al. also found that attending physicians reported that they changed care plans more at night than appreciated by residents⁷ ; such perceptions may explain why hospitalists want to be involved more at night as they believe that they are impacting plans. Our prior findings coupled with this study’s results suggest residents and fellows should clarify with their attending when to call—whereas the attending should also proactively share when they want to be notified—as many hospitalists may want more frequent communication if they do not round with the team, for certain situations overnight, if a patient or family is upset, and before consulting another subspecialty.

Effective communication with consultants, such as articulation of a question for a consulting service or sharing patient information that is required for consultation, is emphasized as an ACGME milestone for both pediatric residents and PHM fellows.³²  It is interesting to note differences in respondent preferences, as markedly fewer residents advocated for communication between the SR and fellow before calling a consult compared with hospitalists and fellows, with SRs preferring more autonomy in this situation. Supervising physicians may be less comfortable with offering more autonomy in this situation because of perception of decreased learner competency or prior experiences with or concern for miscommunication with consultants.¹⁷  The literature has shown that communication during consultations may be prone to errors and can be quite variable depending on the subspecialties involved.^17,33,34  It is also possible that the attending physician wants to be notified first, as they may not agree that calling the consult is necessary. Additional work is warranted to explore how residents and fellows progress on communicating with consultants and why some PHM supervisors want to be more closely involved.

Our findings highlight the importance of not only expectation-setting but also the desire for clear communication and escalation to help ensure positive learning environments as well as patient safety. Although our results describe preferences for communication, poor communication and adverse events are often interrelated,⁵  and structured communication tools and a standardized escalation process have been shown to improve patient safety.^4,35  Most fellows and hospitalists desire escalation of communication when a patient or family is upset, regardless of cause, but hospitalists prefer more communication between team members than expressed by SRs and fellows in this scenario. On day 1 of service, hospitalists should clarify their desire for such communication as well as their preference to be notified when behavioral restraints are ordered, for example. Similarly, they should stress that they want to be called when a patient experiences an adverse event, since only 45% of fellows agreed with calling their attending hospitalist in this situation. These findings emphasize the need for supervisors to outline specific communication guidelines, particularly for upset patients or families and adverse events. One may consider creating a checklist outlining when one should call their supervisor; this would be a useful tool for the residents, fellows, and hospitalist to review at the beginning of a service week. This checklist, perhaps using the scenarios from Appendix 2 may serve as a guide for communication among team members. This need is further supported by ACGME as it requires PHM fellowship programs to set guidelines, including circumstances or events, for when fellows should communicate with their supervising faculty.³ 

Limitations of this research include this was a survey study investigating perspectives, not true behaviors. Although we had a 100% response rate for the fellows, only 20 fellows were surveyed so it is possible we did not uncover some statistical significance because of lack of power. Results pertain to the institutions surveyed, all of which were associated with universities, and some were free-standing children’s hospitals, so findings may not be generalizable to other sites. None of the programs were accredited by ACGME at time of study so we noted differences in supervision in Appendix 1, given the lack of any standardized requirements; however, we asked respondents to consider a medical team with a SR, fellow, and attending on service together, as fellows must be paired with an attending for accreditation. It would be interesting to determine if participant preferences have changed once most fellowship programs are accredited and adhering to ACGME PHM requirements. It is worth noting that the response rates varied between institutions, and that we did not have data for all invited participants so we could not assess potential nonresponse bias or apply poststratification weights. Nevertheless, the occurrence of nonresponse bias in statistical estimates may occur with both low and high response rates.^36,37  Since the majority (44%) of hospitalist respondents were practicing PHM 5 or less years, and another 29% practiced for 10 years or less, results may not be generalizable to more experienced faculty. Additionally, 36% of hospitalists did not supervise fellows and another 39% supervise fellows 1 to 2 weeks per year, likely because of the overall low number of fellows (1–3 per year per program) relative to the number of possible service weeks; it may be worth resurveying once more fellowship programs are established and more faculty have worked with fellows. Lastly, we did not define “invasive” versus “noninvasive” diagnostic procedures and “adverse events” in our surveys. It is possible respondents envisioned different procedures or events that may have skewed their preferences; however, our literature review and cognitive interviews did not support the need to further define or provide examples.

Given that inpatient teams consist of more than a SR, fellow, and attending, an important next step would be to investigate communication preferences among other members of the multidisciplinary medical team, such as interns, students, nurses, and other providers. Such findings would help clarify communication expectations and create shared mental models^38,39  around care escalation in the inpatient setting. Qualitative interviews or focus groups would also be worthwhile; for example, to explore why hospitalists want to be contacted in certain situations and to inquire about the proposed factors for why hospitalists want to be called overnight. Additionally, determining what other factors, such as nocturnist availability, impact communication preferences would be valuable.

Our results emphasize that SRs, PHM fellows, and hospitalists may have differing preferences regarding communication, impacting the balance of supervision and autonomy for the learner as well as patient safety. Level of agreement and differences in agreement across the 3 surveyed populations varied depending on the clinical scenario. Although more work is needed to gain a greater understanding of these preferences, it is insightful and critical to know learners’ and hospitalists’ desires for when to call or not to call a supervisor, particularly for a growing field such as Pediatric Hospital Medicine.