We conducted interviews of a cross-sectional sample of patients’ nurses and physicians in June 2007 and June 2008, before and after an intervention designed to localize physicians to specific patient care units. Each day during the study months, we interviewed patients admitted to general medical services, their nurses and physicians in the afternoon of the patient’s 2nd hospital day. Patients were randomly selected from daily hospital census reports for inclusion in the study using a random number generator.
Setting and Participants
The study was conducted at Northwestern Memorial Hospital (NMH), an 897-bed tertiary care teaching hospital in Chicago, Illinois, and was approved by the Institutional Review Board of Northwestern University. General medical patients were admitted to one of two physician services: a housestaff-covered teaching service or a non-housestaff-covered hospitalist service. Teaching service teams consisted of one attending, one resident, one or two interns, and one or two third-year medical students. Hospitalists cared for patients independently without the assistance of housestaff physicians or mid-level providers (i.e., physician assistants or nurse practitioners).
Prior to January 2008, medical patients were first assigned to a physician service, and subsequently admitted to one of six units with no attempt to localize physician teams to specific units. Therefore, physicians may have cared for patients on as many as six units on four different floors. In January 2008, an additional medical unit was made available, and physicians were designated to specific units. Three units were designated for the teaching service, and four were designated for hospitalists. Patients requiring admission from the emergency department or a physician’s office were first assigned a hospital bed and subsequently a physician service (teaching service or hospitalist) based on their assigned unit. Each teaching service unit was staffed by three teaching service physician teams who cared for all patients admitted to their unit. Each hospitalist service unit was covered by two or three hospitalists. In an effort to optimize localization, physicians did not care for patients off of their assigned unit except to allow readmissions from other services or the outpatient setting to be cared for by the same physician team. No other interventions to improve familiarity, communication, or collaboration among team members were initiated between June 2007 and June 2008.
Interviews of Patients, Nurses, and Physicians
We used a structured interview instrument previously designed by our research team to characterize nurse-physician communication and assess understanding of patients’ plan of care (See Appendix available online).8
Standardized interviews of patients’ nurses and physicians were conducted by a research assistant in the afternoon of the patients’ 2nd hospital day. Patients who were unable to speak and/or understand English, or had cognitive impairment were excluded. Nurses and physicians of excluded patients remained eligible for participation in the study.
Patients’ nurses were interviewed and asked whether they knew the name of the physician primarily responsible for the care of the patient and whether they had discussed the patient’s plan of care with the physician on that day. If discussion had taken place, nurses were asked whether communication had taken place face-to-face, via telephone, or via textpage. Nurses were then asked specific questions about six aspects of the plan of care for the patient that day, including: the main diagnosis, planned tests, procedures, medication changes, which physician consulting services were expected to see the patient, and the expected length of stay. Nurses’ responses to questions about the plan of care were recorded verbatim by the research assistants.
For patients admitted to the teaching service, we defined the inpatient physician as the intern. For patients admitted to the hospitalist service, we defined the inpatient physician as the hospitalist. The rationale for this designation was that these were the physicians primarily responsible for nurse-physician communication. Patients’ physicians were interviewed and asked whether they knew the name of the nurse taking care of the patient and whether they had discussed the patient’s plan of care with the nurse on that day. The form of communication and specific questions regarding the plan of care for the patient that day were assessed in the same fashion as for the nurses.
In order to minimize confounding due to evolving plans of care and various levels of physician training, all interviews were conducted during the afternoon of patients’ 2nd hospital day. This ensured that housestaff physicians were interviewed after their supervising attending had rounded with the team and discussed the plan of care for patients admitted the day prior.
Physician Review for Agreement on the Plan of Care
Two board certified internists (M.P.L. and N.K.) reviewed nurse and physician responses and rated nurse-physician agreement on each aspect of the plan of care as no agreement, partial agreement, or complete agreement. For example, if the nurse responded that the patient’s main diagnosis was fainting and the physician responded that it was syncope, complete agreement was given for that aspect of care (primary diagnosis). If the nurse responded that the patient’s planned testing for the day included only a lower extremity venous duplex study and the physician responded that the patient was planned for a lower extremity venous duplex study and a CT scan of the chest, partial agreement was given. For anticipated length of stay, we defined complete agreement as an exact match between the nurse and physician, partial agreement as a difference between the nurse and physician of 1 day, and no agreement as a difference between the nurse and physician of more than 1 day. We calculated a Nurse-Physician Summary Agreement Score by assigning 0, 1, or 2 points for no agreement, partial agreement, and complete agreement, respectively, to each of the six aspects of the plan of care. To assess inter-rater reliability, approximately 30% of participant responses underwent duplicate review.
Demographic data on patients, nurses, and physicians was obtained from administrative databases and interviews, and compared using chi-square and t-tests. We used chi square to evaluate the number and percent of nurses and physicians reporting that communication had occurred, the form of communication, and whether health-care professionals knew each others’ names before and after localization of physicians to specific units. We assessed inter-rater reliability on the plan of care using the weighted kappa statistic. We compared nurse-physician agreement on aspects of the plan of care using chi square and compared the Nurse-Physician Summary Agreement Score using the Mann-Whitney U test. All analyses were conducted using Stata version 10.0 (College Station, Texas).