The Pediatric Heath Information System (PHIS) developed by the Child Health Corporation is a robust database with detailed and clinically relevant patient-level data culled from 36 freestanding children’s hospitals. Researchers from the University of Washington and Children’s Hospital and Regional Medical Center in Seattle utilized this database to perform a retrospective, descriptive study of the variation in the diagnostic approach, treatment, length of stay (LOS), and readmission rate for children less than 1 year of age hospitalized for bronchiolitis. Data for 17,397 patients admitted between October 2001 and September 2003 with a primary discharge diagnosis of bronchiolitis were analyzed. Diagnostic measures included chest radiograph, rapid viral testing, and various bacterial cultures. Treatment options measured included antibiotics, racemic epinephrine, and inhaled and/or systemic steroids.
In the aggregate, the mean patient age was 3.96 months (standard deviation [SD] 2.92), and these patients had a mean LOS of 2.97 days (SD 2.52). Mean LOS varied from 2.4 to 3.9 days across hospitals. Logistic regression analysis showed that increased LOS was associated with chest radiograph, blood or urine culture, and treatment with antibiotic or systemic steroids. Forty-five percent of patients received antibiotics (range 28% to 62%). Antibiotic treatment was associated with having a chest radiograph and bacterial cultures of blood, urine, or spinal fluid. Although not statistically significant, use of viral testing was associated with decreased use of antibiotics (OR=0.91; 95% CI, 0.81–1.01). Chest radiograph ordering ranged from 38% to 89% (mean=72%). After controlling for age, gender, Medicaid status, illness severity, and month of admission, the hospital remained a statistically significant contributor to variation in LOS, diagnostic testing, pharmacologic treatment, and readmission rate (P<.001).
Dr. Pate has disclosed no financial relationships relevant to this commentary.
The inherent limitations in this type of study design are justified by the unique opportunity to analyze such a large sample size. Using the PHIS database, these authors present compelling evidence of unnecessary variation across hospitals in the approach to one of the most common inpatient pediatric problems.
These results highlight several important issues related to the management of inpatient bronchiolitis. First, decision-making concerning the diagnosis and management of bronchiolitis should be guided by data, including recent systematic reviews1,2 (see