Referral Communication for Pediatric Intensive Care Unit Admission and the Diagnosis of Critically Ill Children: A Pilot Ethnography

Abstract

Purpose:

The effect of communication between referring and accepting clinicians during patient transitions to the pediatric intensive care unit (PICU) on diagnostic quality is largely unknown. This pilot study aims to determine the feasibility of using focused ethnography to understand the relationship between referral communication and the diagnostic process for critically ill children.

Materials and Methods:

We conducted focused ethnography in an academic tertiary referral PICU by directly observing the referral and admission of 3 non-electively admitted children 0–17 years old. We also conducted 21 semi-structured interviews of their parents and admitting PICU staff (intensivists, fellows/residents, medical students, nurses, and respiratory therapists) and reviewed their medical records post-discharge.

Results:

Performing focused ethnography in a busy PICU is feasible. We identified three areas for additional exploration: (1) how information transfer affects the PICU diagnostic process; (2) how uncertainty in patient assessment affects the decision to transfer to the PICU; and (3) how the PICU team’s expectations are influenced by referral communication.

Conclusions:

Focused ethnography in the PICU is feasible to investigate relationships between clinician referral communication and the diagnostic process for critically ill children.

INTRODUCTION

Diagnostic errors can harm critically ill children [1]; the first 12 hours after admission to the pediatric intensive care unit (PICU) is particularly high-risk for misdiagnosis [2]. Most children are transferred to the PICU from another healthcare setting and these transitions are fraught with opportunities for diagnostic error [3]. We know that communication between clinicians during care transitions affects diagnostic quality [3,4]; however, we do not know how referral communication affects the PICU diagnostic process.

Qualitative methods have been used to study many aspects of critical care including the lived experience of patients and staff, complex phenomena such as end-of-life decision-making, and ICU team cognition [5,6]. Focused ethnography combines observation, interviews, and document review limited to specific participants, events, and context to study a distinct concept [7]. Given ethnography’s inductive, immersive, and multimodal approach to studying highly specialized phenomena, it is well-suited to study referral communication and the diagnostic process [7]. Information about the effect of referral communication between frontline settings and the PICU on the diagnostic process will provide a foundation for future study of targeted interventions to improve communication and prevent diagnostic errors in critically ill children. However, the PICU’s dynamic environment may pose challenges to ethnography.

The objective of this pilot study was to determine the feasibility of conducting focused ethnography in the PICU and to identify areas that should be studied so that we can fully describe the relationship between referral communication and the diagnostic process for critically ill children.

MATERIALS AND METHODS

We conducted a pilot study using focused ethnography concentrated on clinician referral communication for PICU admission and its relationship with the diagnostic process upon admission of critically ill children. We reported this study in accordance with the Consolidated Criteria for Reporting Qualitative Studies (COREQ) guidelines [8]. This work was approved by the local Institutional Review Board.

Study Setting and Subjects

We conducted the study at an academic tertiary referral PICU. In this institution, referring clinicians – emergency department (ED) physicians or hospitalists from the same or other institutions – contact the PICU fellow to request patient transfer. The PICU fellow (supervised by the PICU attending) decides whether the patient requires PICU care (no set criteria) and accepts the patient for admission.

We prospectively recruited a convenience sample of 3 patients at referral during weekday day shifts of 3 pre-determined study days. We included only patients with acute problems and excluded planned admissions (e.g., post-operative patients). Each patient’s parent/guardian and all of the admitting PICU team (at minimum, an attending pediatric intensivist, fellow, resident, bedside nurse, and charge nurse) were included in the study as well.

Data Collection

Two medical anthropologists (KD, HR) guided data collection and analysis. We developed and iteratively revised data collection guides for observations and interviews based on the National Academy of Medicine’s (NAM) conceptual model of the diagnostic process [9] and preliminary results of chart review to determine diagnostic errors in the same PICU [10]. Two data collectors (pre-medical and nursing students) were trained on direct observation and interviewing techniques and underwent a one-day orientation in the PICU to learn unit workflow and meet staff.

The data collectors observed diagnosis-related events (6–8 hours per patient) during communication between the referring clinician and accepting PICU fellow and the initial patient-PICU team bedside encounter upon patient arrival. Data collectors conducted semi-structured interviews (15–30 minutes each, audio-recorded and transcribed) with the observed PICU staff. The audio-recordings of referral calls were not available when we sought to analyze them and we did not interview referring clinicians because their responses likely would have been affected by recall bias. The principal investigator (PI) interviewed parents/guardians. Interviews focused on subjects’ perceptions of the referral, transfer, and diagnostic process. To avoid biasing staff behavior/responses, the PI, an attending pediatric intensivist at the same institution, did not observe or interview PICU staff. We completed all interviews within 12 hours of patient admission to preserve subjects’ recall. We reviewed each patient’s medical record to determine the final discharge diagnosis (Figure). Research team members debriefed after each study day to clarify and integrate data.

Figure. Timeline of Data Collection Activities per Patient.

Direct observation of diagnosis-related events started at referral and ended approximately 6–8 hours after pediatric intensive care unit (PICU) admission. Interviews were then conducted with the patient’s parent/guardian and members of the PICU team and completed within 24 hours of admission. Medical records were reviewed after the patient was discharged from the hospital.

To minimize PICU workflow disruptions, we interviewed only two PICU staff at a given time and we interviewed non-physicians during scheduled breaks. We ensured that other staff members could fill in for staff during interviews. Direct observers did not interrupt PICU staff at the bedside.

Qualitative Data Analysis

We uploaded field notes and transcribed interviews into MAXQDA©, a qualitative data analysis software program. The PI (CC), an anthropologist (KD), and a data coder (BA) developed preliminary codes based on the NAM conceptual model of diagnosis [9]. Two coders (CC, BA) assigned these codes to text segments (deductive coding) and developed new codes (inductive coding), iteratively revising the codebook. Since this was a pilot study, we did not collect data until theoretical saturation. We explored the coded data to identify themes regarding the relationship between referral communication and the diagnostic process on PICU admission.

RESULTS

We observed diagnosis-related events for 3 patients admitted to the PICU between April 2017 and February 2018 (23 observation hours). We conducted 21 semi-structured interviews (8 hours total; average duration 23 minutes) with 2 patients’ parents and 19 PICU staff (3 attending pediatric intensivists, 3 fellows, 3 residents, 1 medical student, 6 nurses, and 3 respiratory therapists). We identified three areas for additional exploration (Table):

Table.

Preliminary Areas for Exploration of the Relationship Between Referral Communication and the Diagnostic Process on Admission to the Pediatric Intensive Care Unit

Preliminary Areas for Additional Exploration	Illustrative Quotes
Characteristics of information transfer during referral communication affected the PICU diagnostic process.
Inaccurate or incomplete information was sometimes relayed, which required more effort to make an accurate diagnosis and can cause delays.	“…you can get the world’s worst sign-out and you can still probably come up with the diagnosis because the physical exam and labs… are important. But history is a huge portion of that so if you don’t get a good one, it can be really difficult so it might take you a little longer to get to your conclusion…” (Resident)
Information was typically passed down among members of the PICU team resulting in some degree of data loss. The perspectives of clinicians “upstream” also influenced the conclusions made by clinicians “downstream.”	“…I had heard bits and pieces but things always get filtered when it gets told to me in terms of like, the extraneous stuff gets filtered out and only the pertinent stuff gets told.” (Attending Physician)
	“Everybody was aware of the patient and what the history was… So everybody knew… it was maybe a potential diagnosis of sepsis, so you kind of know what to look for as far as clinical signs, vital signs…” (Nurse)
Uncertainty in patient assessment affected the decision to transfer patients to the PICU.
PICU staff perceived that referring clinicians’ decisions to transfer patients to the PICU were informed both by objective data and subjective feelings of comfort about caring for a potentially sick patient. Referring clinicians’ discomfort stemmed from uncertainty in assessments of patients’ physiologic stability.	(Asked if the referring clinician provided adequate information for the patient to be transferred to the PICU) “Well, …there are two parts to that. One is, do they (the patient) have a certain combination of signs and symptoms that automatically makes them come (to the PICU)? Then there is the level of comfort or the level of concern that the (referring) unit has with regards to (the patient). And that is irrelevant of what I think.” (Attending Physician)
	“…if our perception is that they (referring clinician) are not comfortable with the patient, even if we think that there is nothing wrong with them, then we will just accept (the patient).” (Attending Physician)
Uncertainty also affected accepting PICU physicians’ decision-making. Because accepting physicians could not perform a thorough assessment remotely, they relied on the referring clinician’s judgment tending to err on the side of caution, i.e., accepting the patient for PICU admission.	“I can go either way (in terms of admitting to the PICU vs. the floors), but I just don’t want to send this to (the) floor and then they say, ‘This is a disaster’.” (Fellow)
Expectations of the PICU team were influenced by referral communication before patients were admitted.
The PICU team’s mental image of the patient (physiologic stability and anticipated medical issues requiring intensive care) were influenced by the referring clinician’s working diagnosis. The PICU team’s failure to revisit expectations to check diagnostic accuracy resulted in anchoring bias (tendency to rely heavily on initial information obtained) and premature closure (tendency to accept a diagnosis before it has been fully verified) [11].	“So I basically took the call from the outside physician. And he gave me the data. And based on the data, we were able to make the diagnosis. I guess he already had that diagnosis, so we just confirmed it.” (Fellow)
	Asked if the patient/family contributed adequate information to help the PICU team come to the diagnosis) “Uh, yes. Although I don’t think we needed any information from them to make the diagnosis.” (Attending Physician)

PICU - pediatric intensive care unit

Characteristics of information transfer during referral communication affected the PICU diagnostic process.

PICU staff universally viewed information from referring clinicians as very important in guiding diagnosis and treatment upon patient arrival. Accuracy and completeness of information relayed affected the effort required to make a correct and timely diagnosis. Information was typically passed down among PICU team members resulting in some erosion of accuracy; in addition, perspectives of “upstream” clinicians influenced “downstream” clinicians’ conclusions.

Uncertainty in patient assessment affected the decision to transfer patients to the PICU.

PICU staff perceived that referring clinicians’ decisions to transfer patients to the PICU were influenced by uncertainty in their assessments of physiologic stability, with most erring on transferring patients to the PICU.

Expectations of the PICU team were influenced by referral communication before patients were admitted.

The PICU team’s mental image of the patient before admission (e.g., physiologic stability) was influenced by the referring clinician’s working diagnosis. The PICU team’s failure to assess their initial expectations regarding patients’ diagnoses resulted in anchoring bias (tendency to rely heavily on initial information obtained) and premature closure (tendency to accept a diagnosis before it has been fully verified), affecting their decisions [11].

DISCUSSION

Qualitative research benefits patients by elucidating complex processes that affect outcomes and by providing insight into potential causative mechanisms, thereby helping researchers understand complex phenomena and develop sound quantitative metrics [12,13]. Only a few qualitative studies have focused on the diagnostic process in inpatient settings [14,15]. To our knowledge, this pilot study is the first diagnosis-focused qualitative exploration in the ICU.

Our pilot study demonstrated that real-time observations and interviews can be conducted in a busy PICU during the brief period when a critically ill patient is admitted. To do so, we obtained informed consent from PICU staff in advance and the IRB waived initial consent for patients so we could commence data collection before patients arrived. We found that 3 data collectors per patient was optimal, allowing us to collect a large amount of data within 12 hours from admission. Informal queries of PICU staff found that study procedures did not disrupt workflow.

Our results are hypothesis-generating and serve as a roadmap for further studies. Given the preliminary areas of focus we identified, future work should: 1) evaluate how the accuracy and completeness of information transfer affects the PICU diagnosis, 2) describe how PICU teams critically re-evaluate the referring clinician’s diagnosis, and 3) examine how uncertainty affects the diagnostic process during PICU admission.

Our study has limitations. Observations may have introduced a Hawthorne effect, which we mitigated by triangulating data from different sources to confirm our findings. We could mitigate it further by extending the orientation period to increase PICU staff’s familiarity with the data collectors. In addition, we may have missed findings unique to weekends/night shifts. As noted, we did not obtain data directly from referring clinicians and we did not systematically collect data confirming that study procedures did not disrupt PICU work. Thus, future studies should include analysis of both recorded referral phone calls and interviews with referring clinicians plus data on PICU workflow. We did not collect data to theoretical saturation, thus our findings are incomplete. The PI is a pediatric intensivist in the study PICU, which may have biased data analysis; however, we mitigated this bias by having two non-PICU-affiliated team members analyze the data and by creating an audit trail to enhance confirmability [5].

CONCLUSIONS

Focused ethnography is feasible in the PICU. We will use this methodology to further investigate relationships between clinician referral communication and the diagnostic process for critically ill children.

ABBREVIATIONS

PICU

pediatric intensive care unit

ED

emergency department

COREQ

Consolidated Criteria for Reporting Qualitative Studies

NAM

National Academy of Medicine