Objectives: Hospitalists are a new subgroup of pediatricians. However, data comparing quality of care between hospitalists and nonhospitalists are limited. Bronchiolitis, a common cause of pediatric hospital admission, is an appropriate condition for evaluating inpatient quality of care. The goal of this study was to identify quality indicators for the evaluation and treatment of children hospitalized with bronchiolitis based on the 2006 American Academy of Pediatrics guidelines and to assess differences in adherence rates to these indicators between hospitalist and nonhospitalist pediatricians.
Methods: This was a retrospective chart review of children admitted to 2 academic centers in 2007 and 2008. Reviewers selected charts with a discharge diagnosis of bronchiolitis and collected data on evaluation, treatment, length of stay, readmission, and adverse outcomes.
Results: Reviewers analyzed 713 charts. In children without objective response to bronchodilator therapy, hospitalists and nonhospitalists discontinued albuterol and racemic epinephrine similarly. Hospitalists discontinued unnecessary systemic corticosteroid therapy (75.0% vs 42.4%; P = .001) and antibiotic therapy (71.0% vs 48.6%; P = .007) more frequently than nonhospitalists.
Conclusions: These data suggest hospitalists better adhere to selected portions of the American Academy of Pediatrics bronchiolitis guidelines, thus providing higher quality of care. Quality indicators used in this study can distinguish physician performance in the inpatient management of bronchiolitis.
Pediatric hospitalists are a rapidly growing subset of providers. Previous studies reveal that hospitalists provide care associated with lower length of stay (LOS) and costs than traditional pediatric providers1–6 with few exceptions.7 In most cases, cost savings are due largely to decreased LOS. However, data comparing quality of care measures between hospitalists and nonhospitalists are lacking.8 Bronchiolitis, a reason for 16% of all pediatric hospital admissions,9 is an appropriate choice for evaluating quality of care.
Before 2006, physicians varied markedly in their management of children hospitalized for bronchiolitis.10 This variability led to the publication of national guidelines that clearly delineated evidence-based recommendations of care.11 A subset of these guidelines is especially pertinent to physician management of hospitalized patients (Table 1). However, results from previous studies suggest that development of bronchiolitis guidelines alone may not change physician behavior; an implementation strategy may also be necessary.12,13 Providing physician feedback on the quality of care being delivered could be part of such a strategy. The challenge in using national guidelines for the evaluation of quality of care is that they may not be appropriate for use as quality indicators. A quality indicator should be easily obtained and measured, complete, able to be measured repeatedly, and generalizable to multiple institutions.14,15 To date, no study has evaluated the American Academy of Pediatrics (AAP) guidelines for use as quality indicators in physician management.
Survey data suggest that hospitalists more often use evidence-based guidelines than nonhospitalists when treating children hospitalized with common conditions, including bronchiolitis.16 Although a previous retrospective study of children admitted with either bronchiolitis or asthma found no significant difference between hospitalist and community physician care in terms of cost of care, LOS, or clinical outcomes,17 quality of care data were limited to assessment of adherence to local care pathways at a single institution and not to national evidence-based recommendations.
The goal of our study was to develop a proposed set of quality of care indicators for the inpatient management of children hospitalized with bronchiolitis. By using these indicators, we aimed to compare the quality of physician care delivered at 2 geographically distinct institutions between hospitalist and nonhospitalist pediatricians.
Hasbro Children’s Hospital of Rhode Island Hospital is a tertiary care facility located in Providence, Rhode Island. This site admits ∼500 patients for bronchiolitis annually. It is the principal tertiary pediatric care center for the state of Rhode Island and bordering communities and serves a predominantly urban and suburban population. Nearly all hospital admissions are first triaged in the emergency department, including transfers from other facilities. University of Missouri Women and Children’s Hospital is a tertiary care facility located in Columbia, Missouri. This facility admits ∼130 patients for bronchiolitis annually and serves as the only pediatric hospital for central Missouri. It provides services for a geographically disparate suburban and rural population. A large proportion of admissions are directly admitted to the hospital from primary care clinics or as transfers from outlying hospitals. Both sites have full-time pediatric hospitalist services staffed by physicians with a minimum of 2 years’ inpatient attending experience as well as specialist and community pediatricians who admit children for general inpatient pediatric illnesses. All inpatient services are provided with the assistance of residents, medical students, and interns at both study sites. Admission to an inpatient service is determined by the patient’s primary care provider’s preference. During the study periods, neither site routinely used a standardized bronchiolitis care pathway. Bronchiolitis disease severity assessments were based primarily on clinical impression without universal adoption of a standardized scoring method at either institution.
We conducted a retrospective cohort study of children admitted from January 1, 2007, to December 31, 2008. All charts with a primary discharge diagnosis of International Classification of Diseases, Ninth Revision code 466.19 (non–respiratory syncytial virus bronchiolitis), 466.11 (respiratory syncytial virus bronchiolitis), 786.03 (apnea), 465.9 (acute upper respiratory infection), 493.9 (asthma, unspecified), or V73.99 (unspecified viral illness) were screened for inclusion by reviewers. All charts abstracted included a combination of paper-based and electronic medical record data. To prevent inclusion of patients with asthma or nonbronchiolitis acute respiratory tract infection, children aged >2 years were excluded. In addition, charts were excluded if patients were first admitted to the ICU. Finally, reviewers excluded patients without either a primary or secondary discharge diagnosis of bronchiolitis (codes 466.11 or 466.19) or documentation of symptoms and clinical findings consistent with bronchiolitis. All reviewers were unblinded to the nature of the study and were supervised directly by the primary researcher at each study site. The study was approved by the institutional review boards at both study sites.
Baseline characteristics assessed included age, gender, history and degree of prematurity, history of wheeze or reactive airway disease (RAD), tobacco exposure, day care exposure (defined as ≥1 day a week in a day care setting), and diagnosis of acute otitis media (AOM).
Quality Measures and Outcomes Assessed
We selected quality measures based on the AAP guidelines for bronchiolitis diagnosis and management.11,18 A subset of these guidelines was determined to be most applicable specifically to the management of hospitalized children, with minimal confounding from health care provider behavior before admission (Table 1). Furthermore, the selected guidelines were determined to have a goal adherence rate of 100%, allowing for their use as quality indicators.
Assessment of adherence to guideline 2b (minimizing bronchodilator use) was limited to those charts in which a negative or positive response to bronchodilator therapy was documented and a trial of bronchodilator therapy was attempted before, at, or during admission. Acceptable documentation of bronchodilator response included either use of a scoring system or recorded clinical impression of response from a care provider (inclusive of respiratory therapists, residents, or attending physicians). Bronchodilator therapy was considered continued if an order for its use or a recorded administration occurred after documentation of no effect. For assessment of guideline 3 (avoidance of corticosteroid use in patients without asthma or RAD), charts were included in the analysis if the patient had no documented history of RAD or asthma and received at least 1 dose of systemic corticosteroids at, before, or during inpatient admission. Corticosteroid therapy was considered stopped if the inpatient team discontinued corticosteroids during the admission and did not prescribe them at discharge. Charts were included for assessment of adherence to guideline 5 (antibiotic therapy avoidance in cases negative for bacterial infection) if patients received antibiotics before, at, or during admission without documented evidence of bacterial infection. Adequate documentation of infection included: (1) AOM; (2) a blood culture positive for a pathogen; (3) a positive urine culture; (4) a urinalysis positive for leukocyte esterase, nitrites, or white blood cell count >5 cells per high-powered field; (5) a clinical diagnosis of bacterial pneumonia, documented as a combination of positive chest radiography, sputum Gram stain and culture, and/or physician clinical impression; or (6) any other bacterial infection recorded in the chart for that admission. Antibiotics were considered appropriately discontinued if they were stopped within 24 hours of admission (or, if started in the hospital, within 24 hours of initial order being written) and none were prescribed at discharge.
Researchers also gathered data on evaluation, treatment, LOS, readmission within 4 weeks, and adverse outcomes (ICU admission or death during the hospitalization). LOS was calculated from recorded times of admission and discharge orders. Bronchodilator therapy included albuterol, levalbuterol, or racemic epinephrine given via nebulizer or meter-dose inhaler. Corticosteroid therapy included oral, intramuscular, or intravenous administration of any corticosteroid.
LOS, age, and continuous variables were analyzed by using the Wilcoxon rank-sum test; categorical variables were analyzed by using ϰ 2 analysis. Statistical analysis was performed by using Stata 11.1 (StataCorp, College Station, TX).
A total of 713 charts met inclusion criteria. Baseline characteristics of patients admitted to hospitalist and nonhospitalist pediatricians are listed in Table 2. Patients admitted to hospitalists were more likely to be exposed to tobacco (43.6% vs 34.2%; P = .01) and to undergo serum electrolyte testing (35.4% vs 27.8%; P = .03). Children admitted to nonhospitalist pediatricians more often underwent lumbar puncture (9.4% vs 3.4%; P = .001) and were more often diagnosed with AOM (12.4% vs 7.1%; P = .02). Although testing for urinary tract infection (UTI) was similar between physician groups, children more often had evidence of UTI on urinalysis or urine culture in the hospitalist group (6.1% vs 0.0%; P = .03). Of children receiving an initial dose of albuterol, corticosteroid, or antibiotic as part of their illness management, the proportion of times this occurred after hospital admission was 5.8%, 17.2%, and 24.9%, respectively. Baseline patient characteristics differing between study sites are listed in Table 3. All other measured baseline characteristics between study sites were similar.
Compliance With Quality Indicators
Adherence to quality indicators between hospitalists and nonhospitalists is listed in Table 4. Epinephrine efficacy results were recorded in all children who received epinephrine at both sites. However, albuterol efficacy was not recorded in 20.6% of children cared for by hospitalists and 26.3% of nonhospitalist admissions (P = .13). For children with no history of RAD or asthma and who had documented evidence of no response to bronchodilator therapy, physicians usually discontinued ineffective racemic epinephrine therapy (92.9% hospitalists, 91.4% non-hospitalists; P = .87) but less often stopped ineffective albuterol treatments (84.0% hospitalists, 80.5% nonhospitalists; P = .57). However, hospitalist and nonhospitalist pediatricians differed in discontinuing unnecessary corticosteroid therapy (75.0% hospitalists, 42.4% nonhospitalists; P = .001) and discontinuing antibiotic therapy in patients who were not diagnosed with a bacterial infection (71.0% hospitalists, 48.6% nonhospitalists; P = .007).
When assessing individual hospitals, there were no differences between stopping ineffective albuterol by hospitalists compared with nonhospitalists (site 1: 62.5% hospitalists and 56% nonhospitalists [P = .28]; site 2: 67.1% hospitalists and 48.4% nonhospitalists [P = .07]). In addition, at site 1, differences in discontinuing antibiotics when they were not indicated did not achieve statistical significance, although the trend remained in the direction seen in aggregate analysis of both sites (71.4% hospitalists, 54.9% nonhospitalists; P = .19). Likewise, at site 2, differences in discontinuing unnecessary systemic corticosteroids was not significantly different (79.0% hospitalists, 60.0% nonhospitalists; P = .38).
There was no difference in LOS between hospitalists and nonhospitalists (LOS 2.71 days hospitalists, 2.75 days nonhospitalists; P = .86). A greater proportion of patients cared for by nonhospitalist pediatricians were transferred to the ICU (16.3% nonhospitalists, 10.9% hospitalists; P = .046). There were no deaths in children diagnosed with bronchiolitis at either study site. Factors differing between study sites included LOS (2.48 days site 1 vs 3.83 days site 2; P = .0002) and rates of readmission within 4 weeks (7.3% site 1 vs 3.4% site 2; P = .031).
We used evidence-based quality indicators for inpatient management of bronchiolitis based on the 2006 AAP diagnosis and treatment guidelines.18 By using these quality indicators, we found that hospitalists more frequently adhered to selected therapeutic recommendations compared with nonhospitalist pediatricians. Specifically, hospitalists more often discontinued unnecessary systemic corticosteroid and antibiotic therapy.
A major strength of our study is its multisite nature. This detail allows us to demonstrate the generalizability of these quality indicators to other pediatric tertiary care centers. In addition, the wide variability of care suggests that utilization of these quality indicators may result in a general improvement of care once implemented. As with other previously identified quality indicators in infection control such as hand-washing, the mechanical ventilator bundle, and the central line insertion and maintenance bundles, our inpatient bronchiolitis quality indicators have a goal of 100% adherence. The fact that we identified measures that are clearly below this standard demonstrates their potential utility in evaluating physician practice and setting goals for practice improvement. It also stands to reason that when a quality indicator has already achieved near-universal compliance, it is less likely to highlight discrepancies in care. Finally, in cases in which there are baseline discrepancies between patient characteristics such as those found between hospitalist and nonhospitalist patients or those between study sites, the fact that adherence goals remain 100% for all providers and study sites helps to minimize bias incurred through retrospective assessment.
Our results revealed several differences in baseline characteristics and clinical outcomes between study sites. However, differences in demographic characteristics and referral source are likely due in part to regional differences and institutional context specific to each site. If implementation of a hospitalist service is viewed as a complex quality improvement initiative, then evaluating its effectiveness in a multisite study will naturally demonstrate differences in institutional context that cannot be controlled.19 In addition, with such interventions, focusing on a specific disease category, such as bronchiolitis, and examining process measures, rather than clinical outcomes, will more likely yield results that are both interpretable and which have the possibility of being individualized to other institutions.20 Therefore, focusing on differences between patients seen by hospitalist and nonhospitalist physicians provides more relevant and applicable data. Another possible limitation to these data would be that overall differences between hospitalist and nonhospitalist performance could have been driven by 1 study site. When assessing study sites independently, some differences were not found to be statistically significant in cases in which the overall number of patients assessed for that indicator was very small. For example, bronchodilator use was much less common in study site 2, limiting the ability to detect a difference between hospitalists and nonhospitalists in discontinuing unnecessary bronchodilator therapy at that site. However, the trends toward difference were similar to those detected at study site 1, supporting the overall difference detected in aggregate analysis. This finding demonstrates the differences in institutional context which should be expected between study sites that are geographically and culturally separate.
Of note, more children admitted to nonhospitalist pediatricians were transferred to the ICU than those admitted to hospitalists. Although this finding may indicate that children admitted to nonhospitalists were more severely ill, it is notable that there was no significant difference in age or history of wheeze or RAD between the 2 groups, all of which are known risk factors for severity of illness in bronchiolitis.11 In addition, patients admitted to hospitalists more often had a history of tobacco exposure, a known risk factor for developing bronchiolitis in this age group.21,22 The difference in ICU admission rates could be explained by increased accessibility of hospitalists to respond to acute changes in patient condition without competing outpatient responsibilities, a theory put forth for adult hospitalists,23 or it may represent a difference in “disease-specific physician experience”24 with inpatient bronchiolitis management. It would be interesting to perform a cost analysis as well; such an analysis was beyond the scope of this study.
Our study has several limitations. First, our data were restricted to retrospective observations. Next, although laboratory testing was initially identified as a process measure of interest and data on diagnostic testing were collected (guideline 1a), we could not compare rates of laboratory testing due to confounding by the high proportion of patients first seen in the emergency department. These patients likely had at least some laboratory and/or radiologic evaluation performed before hospital admission, and this testing could not be reliably separated from testing performed on or after admission. Differences in rates of lumbar puncture and electrolyte testing therefore could not be reliably attributed to admitting physician practice. In addition, in many cases accurate documentation of response to a trial of bronchodilator therapy was not available, limiting the number of cases in which physician practice could be evaluated on this process measure. Furthermore, continued administration of bronchodilator therapy was sometimes left to the discretion of the respiratory therapist, thus limiting the detection of practice patterns between physician groups. Prospectively gathered data from children randomized to admission to either hospitalist or nonhospitalist pediatrician care would help provide further insight but may be limited due to institutional policies and the widespread implementation of hospitalist services.
An important area of future research in hospital medicine involves studying indicators related to health care quality. These can be divided into outcomes measures (eg, ICU admission, death, rehospitalization, bloodstream infections) and process measures (eg, hand hygiene, vaccination).25 Little guidance exists for developing quality-of-care indicators in inpatient pediatric medicine. However, the Society for Healthcare Epidemiology of America provides several recommendations that are readily generalizable.14 First, the indicator must be clearly defined and measurable with numerators and denominators. Second, it should be easy to identify and collect data on the indicator. Third, data collection for the indicator must be sensitive enough to capture all data. Finally, the indicator and method of collection must be used across all institutions. By using the 2006 AAP guidelines on diagnosis18 and management of bronchiolitis, we were able to identify process measures that met these criteria and could be used as quality indicators.
Although process and outcome measures have been assessed to some extent for hospitalists caring for adults,26 evaluations of pediatric hospitalists have been limited primarily to outcome measures.1–7,27,28 Data from the few studies assessing differences in adherence to process measures suggest that pediatric hospitalists may more consistently use evidence-based therapies than nonhospitalists,16 and hospitalists vary less in their management of common inpatient illnesses, especially with regard to use of proven therapies.29 However, these data are limited to national surveys and may not reflect actual practice behavior. This study represents an initial step in developing metrics for evaluating the quality of care provided to children hospitalized with bronchiolitis. The results from this study demonstrate that hospitalists more often meet the identified quality indicators than nonhospitalist pediatricians and therefore provide higher quality care to children admitted for bronchiolitis.
Children hospitalized for bronchiolitis experienced similar clinical outcomes regardless of whether they were treated by a pediatric hospitalist, in terms of LOS and readmission rates. In terms of adherence to proposed quality indicators, both hospitalist and nonhospitalist pediatricians routinely discontinued unneeded racemic epinephrine but insufficiently stopped ineffective albuterol therapy. However, hospitalists more frequently stopped unnecessary corticosteroid and antibiotic therapy, suggesting hospitalists better adhere to bronchiolitis guidelines and provide higher quality care. Quality indicators used in this study can distinguish physician performance in the inpatient management of bronchiolitis and may prove useful in future efforts to reduce overutilization of unnecessary medical interventions for bronchiolitis management.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
RESEARCH SUPPORT: Research support was provided by the Department of Pediatrics, Hasbro Children’s Hospital, Providence, Rhode Island.
- American Academy of Pediatrics
- acute otitis media
- length of stay
- reactive airway disease
- urinary tract infection
- Dwight P,
- MacArthur C,
- Friedman JN,
- Parkin PC
- Landrigan CP,
- Srivastava R,
- Muret-Wagstaff S,
- et al
- Srivastava R,
- Landrigan CP,
- Ross-Degnan D,
- et al
- Bellet PS,
- Whitaker RC
- Boyd J III.,
- Samaddar K,
- Parra-Roide L,
- Allen EP,
- White B
- Landrigan CP,
- Conway PH,
- Edwards S,
- Srivastava R
- Christakis DA,
- Cowan CA,
- Garrison MM,
- Molteni R,
- Marcuse E,
- Zerr DM
- 11.↵American Academy of Pediatrics Subcommittee on Bronchiolitis. Diagnosis and management of bronchiolitis. Pediatrics. 2006;118(4):1774–1793.
- Barben J,
- Kuehni CE,
- Trachsel D,
- Hammer J
- Davidoff F,
- Batalden P,
- Stevens D,
- Ogrinc G,
- Mooney S
- Conway PH,
- Edwards S,
- Stucky ER,
- Chiang VW,
- Ottolini MC,
- Landrigan CP
- Seid MQ,
- Kurtin PS
- 18.↵American Academy of Pediatrics Steering Committee on Quality Improvement and Management. Classifying recommendations for clinical practice guidelines. Pediatrics. 2004;114(3):874–877.
- Walshe K,
- Freeman T
- Pawson R,
- Greenhalgh T,
- Harvey G,
- Walshe K
- Bradley JP,
- Bacharier LB,
- Bonfiglio J,
- et al
- Wells RD,
- Dahl B,
- Wilson SD
- Copyright © 2012 by the American Academy of Pediatrics