Evaluation of a Staff-Only Hospitalist System in a Tertiary Care, Academic Children’s Hospital
Dwight P, MacArthur C, Friedman JN, et al. Evaluation of a staff-only hospitalist system in a tertiary care, academic children’s hospital. Pediatrics. 2004;114:1545-9.
The division of pediatrics at The Hospital for Sick Children in Toronto, a tertiary care, academic center, operates a pediatric inpatient unit with approximately 4000 admissions each year. In 1995, limited resident duty hours led to a reorganization of inpatient pediatric teams to include two distinct hospitalist models: A hospitalist/housestaff model (CTU) and hospitalist staff -only model (CPU). The authors review recent research that has demonstrated the efficiency of hospitalist/housestaff systems in both adult and pediatric medicine and accurately point out that published data is lacking assessing the staff -only pediatric hospitalist model. Therefore, the authors designed a cohort study of 3807 admissions to the general inpatient pediatric unit between July 1, 1996 and June 30, 1997.
The primary outcome measure was length of stay, and secondary outcome measures included frequency of subspecialty consultation, readmission to the hospital, and death. Consultations were measured as none or >1 and readmissions were defined as admission within 7 days of discharge for the same or a related diagnosis. Clinically relevant information collected for each patient included age, gender, referral source, stay in a special care unit, most responsible diagnosis, and comorbidity. Comorbidity was defined as either an uncomplicated stay or a stay complicated by a chronic illness, series or important conditions, and/or a potentially life-threatening condition. The CTU team had a daily census of 24 to 30 patients and consisted of 1 attending pediatrician, 3 or 4 pediatric residents, and 2 medical students. CTU pediatricians attended this service 4–8 weeks each year. The CPU was staffed with 3 pediatricians who were responsible for all aspects of care Monday through Friday from 0800 to 1700 and on weekends. Medical students were included on this team. Overnight and weekends clinical fellows not part of the CPU team provided coverage. Each CPU physicians maintained a daily census of 8–10 patients. These physicians spent approximately 11 months of the year providing inpatient care.
During the study there were 3807 admissions, of which 33% were to the CPU and 67% were to the CTU. Patients admitted to the CPU were older (median age: 95 weeks vs. 69 weeks, p < .01) and less likely to have comorbidity (24% vs. 30%; p < .01). The diagnoses admitted to the two teams were not significantly different. The median length of hospital stay for the CPU team was 2.5 days (interquartile range [IQR]: 1.6–4.4 days) versus 2.9 days (IQR: 1.8–4.9) for the CTU team (p < .01). Multivariate linear regression showed a significant difference in length of stay after adjustment for age, gender and comorbidity (p < .04). The authors performed a stratified analysis of the 10 most frequent diagnoses admitted during the study period, and the median length of stay for these groups combined was shorter on the CPU team compared with the CTU team (2.1 days vs. 2.6 days, p < .01). There was no significant difference between the two teams with readmissions, frequency of consultation, or death.
The authors discuss some important limitations to the study. First, the unique characteristics of the individual unit studied inhibit the ability to generalize the results. Second, there were some differences in the baseline characteristics between the two groups, although multivariate analysis of theses differences did not change the statistical significance of the results. Finally, satisfaction of patients, families, and care providers was not measured.
These researchers conclude that within this system the hospitalist staff -only team reduced the length of stay by 14% compared with an attending staff/housestaff team. Although statistically significant, the clinical significance of this reduction in length of stay is unclear and the authors did not include financial data in the study design. Despite these facts, the authors make an important assertion that the difference in stay of 8 hours may be enough to promote throughput by decreasing wait times for admissions from the emergency departments and/or special care units. Additional studies are required to test this assertion.