OBJECTIVE. The goal was to test the hypothesis that pediatric hospitalists use evidence-based therapies and tests more consistently in the care of inpatients and use therapies and tests of unproven benefit less often, compared with community pediatricians.
METHODS. A national survey was administered to hospitalists and a random sample of community pediatricians. Hospitalists and community pediatricians reported their frequency of use of diagnostic tests and therapies, on 5-point Likert scales (ranging from never to almost always), for common inpatient pediatric illnesses. Responses were compared in univariate and multivariable logistic regression analyses controlling for gender, race, years out of residency, days spent attending per year, hospital practice type, and completion of fellowship/postgraduate training.
RESULTS. Two hundred thirteen pediatric hospitalists and 352 community pediatricians responded. In multivariable regression analyses, hospitalists were significantly more likely to report often or almost always using the following evidence-based therapies for asthma: albuterol and ipratropium in the first 24 hours of hospitalization. After the first urinary tract infection, hospitalists were more likely to report obtaining the recommended renal ultrasound and voiding cystourethrogram. Hospitalists were significantly more likely than community pediatricians to report rarely or never using the following therapies of unproven benefit: levalbuterol, inhaled steroid therapy, and oral steroid therapy for bronchiolitis; stool culture and rotavirus testing for gastroenteritis; and ipratropium after 24 hours of hospitalization for asthma.
CONCLUSION. Overall, in comparison with community pediatricians, hospitalists reported greater adherence to evidence-based therapies and tests in the care of hospitalized patients and less use of therapies and tests of unproven benefit.
In 1996, Wachter and Goldman1 coined the term “hospitalist” for physicians focusing on the care of hospitalized patients. The Society of Hospital Medicine (SHM) membership grew from <100 to >4000 between 1997 and 2004,2 and the SHM estimates that nearly twice as many hospitalists may be in practice.3,4 Eight percent of SHM members, or ∼300, are pediatric hospitalists; a recent pediatric hospitalist conference attracted >200 attendees (D. Rauch, verbal communication, 2005). No formal counts have been made, but there are likely 500 to 2000 pediatric hospitalists in practice today.5
Pediatric hospitalist systems were found in 6 of 7 studies to decrease lengths of stay and inpatient costs, but data on the quality of care are lacking.6–12 Although 2 adult studies found decreased mortality rates in hospitalist systems,13,14 the processes of care that led to the mortality rate decreases were not studied. No studies in pediatric ward settings demonstrated mortality rate differences, although they all were underpowered to do so.6–9 Because death in pediatric inpatient ward settings is uncommon, more-sensitive indicators of care quality are required to measure the effectiveness of a care system and its providers. No studies have investigated potential differences in care provided by hospitalists and community pediatricians. Measuring the processes of care delivery, such as the consistency with which providers follow evidence-based practices and national guidelines, is one means of measuring quality more precisely. To compare the quality of care delivered in hospitalist and traditional systems, we undertook a nationwide survey to compare hospitalists' and community pediatricians' reported use of evidence-based therapies in the care of common inpatient conditions.
In autumn 2003, a survey was designed to collect data on variations in inpatient management of common pediatric conditions. The survey included questions on respondents' demographic features and training background, questions on hospital characteristics, and 48 questions on the management of common inpatient conditions, namely, pneumonia, gastroesophageal reflux, bronchiolitis, gastroenteritis, first urinary tract infection (UTI) for infants, and asthma. Respondents' reported their frequency of use of diagnostic tests or therapies on 5-point Likert scales (ranging from 1 = never to 5 = almost always). The original survey questions were pilot-tested with a small group of hospitalists and pediatricians; after incorporation of their feedback, a final survey instrument was developed for distribution.
The refined survey was administered to the entire population of pediatric hospitalists identified nationally through 3 known mechanisms to reach pediatric hospitalists, namely, (1) a national listserv for pediatric hospitalists sponsored by the SHM and the American Academy of Pediatrics (AAP) Section on Hospital Medicine, (2) the Pediatric Research in Inpatient Settings network listserv, and (3) the attendance list from the first national pediatric hospitalist conference, held in San Antonio, Texas, in November 2003. Hospitalists identified through >1 of these sources were counted only once; in total, there were 320 identified pediatric hospitalists. Respondents had the option of completing the survey online, using a secure, Internet-based interface, or on paper. Regular reminder notices and additional copies of surveys were sent to all nonresponders. Respondents were excluded if they did not practice primarily as hospitalists.
A comparison group of community pediatricians were administered the same questions in September 2004. From the American Medical Association (AMA) nationwide list of 41358 office-based primary care pediatricians, 2500 randomly selected pediatricians were invited to participate. The number of invitations was based on a power calculation that indicated that at least 200 respondents would be necessary to detect significant differences in the responses of pediatricians and hospitalists, with a target 8% response rate. The Medical Marketing Services listserv managers of the AMA reported participation rates of 2% to 5% in similar national surveys with their database (M. L. Walters, verbal communication, 2004), but we hoped to achieve a somewhat higher response rate by minimizing the length of the survey. Three weeks after the first mailing, the survey was mailed a second time to all nonresponders. Respondents had the option of completing the survey online, using a secure, Internet-based interface, or on paper. Survey respondents were included only if they worked primarily as primary care pediatricians and had served as the attending physician of record for ≥1 hospitalized pediatric patient (excluding well newborns) in the past 12 months. Community pediatricians were asked identical questions regarding demographic features, practice setting, training background, and management of common inpatient pediatric conditions. Community pediatricians were asked specifically to report their use of diagnostic tests and therapies for hospitalized patients only and not to report their outpatient management practices.
To determine differences in compliance with evidence-based practices between hospitalists and community pediatricians, the British Medical Journal (BMJ) publication Clinical Evidence was used to identify therapies proven to be beneficial.15 Three therapies with proven benefit in treating hospitalized asthmatic children were identified, namely, steroids, albuterol, and ipratropium in the first 24 hours. To determine adherence to AAP clinical practice guidelines, the AAP guideline recommending a renal ultrasound and a voiding cystourethrogram after the first UTI in infants was selected.16 We then defined the evidence-based response for use of these proven therapies and nationally recommended tests as respondents reporting use often or almost always.
Next, BMJ Clinical Evidence was searched to identify therapies of unproven benefit, on the basis of the BMJ designation of the treatment in the “unknown effectiveness” category. Four therapies of unproven benefit for bronchiolitis (levalbuterol, inhaled steroid therapy, oral steroid therapy, and albuterol) and 1 therapy of unproven benefit for asthma (ipratropium after 24 hours of hospitalization) were identified.15 AAP-endorsed clinical practice guidelines were searched to identify tests that were not recommended routinely because of unproven benefit. Gastroenteritis management was the only applicable AAP guideline, recommending against obtaining stool cultures or performing rotavirus testing for routine gastroenteritis.17,18 The evidence-based response for these 7 unproven tests and therapies was defined as reporting use never or rarely. The other 36 therapies and tests included on the questionnaire were not classified specifically as being either of proven benefit or of unknown effectiveness in the BMJ of Clinical Evidence, and explicitly defined recommendations in AAP practice guidelines did not exist. Therefore, there was no evidence-based response defined for these therapies and tests; variability between hospitalists and community pediatricians was measured, but no judgments about the evidence underlying these management choices were made.
Hospitalists' and community pediatricians' uses of tests and therapies were compared by using the Wilcoxon rank sum test for all 48 therapies and tests. For the 12 therapies and tests with an evidence base, univariate analyses (Fisher's exact tests) were conducted to compare the percentages of hospitalists and community pediatricians reporting compliance with the evidence-based response for each therapy. Univariate relationships with the following factors hypothesized a priori to correlate possibly with reported management decisions were also determined: completion of fellowship/postgraduate training, days attending per year, years out of residency, and hospital practice type. For days attending per year, high-volume practitioners (defined as those attending >90 days per year) were compared with low-volume practitioners. For years out of residency, pediatricians ≤10 years out of residency were compared with pediatricians ≥11 years out of residency. Hospital practice type compared pediatricians in an academic center setting with those in a community hospital setting. To test more thoroughly the principal hypothesis regarding hospitalists' use of evidence-based and unproven therapies, multivariable logistic regression analyses controlling for gender, race, days attending per year, years out of residency, hospital practice type, and completion of additional fellowship or graduate training were conducted for each therapy or diagnostic test of interest.
A total of 213 pediatric hospitalists and 352 community pediatricians responded to the survey (response rate for hospitalists: 67%; response rate for community pediatricians: 15% of 2425 correct addresses). Of these, 198 hospitalists (93%) and 228 community pediatricians (65%) met the inclusion criteria, completed the section on inpatient management decisions, and were included in the analyses. The demographic features of the participating community pediatricians (53% female; average time out of residency: 16 years) were very similar to those of all pediatricians in the AMA national database of office-based pediatricians (51% female; average time out of residency: 16 years). Hospitalists were more likely than community pediatricians to have completed fellowship/postgraduate training and to practice in an academic center; on average, hospitalists attended significantly more days per year and had completed residency more recently (Table 1).
Univariate Analyses of Differences in Use of Evidence-Based and Unproven Therapies
In univariate analyses, hospitalists reported often or almost always using 4 of 5 evidence-based therapies and nationally recommended tests significantly more often than did community pediatricians. Hospitalists were significantly more likely to report rarely or never using all 7 therapies and tests of unproven benefit, compared with community pediatricians. Univariate analyses of all respondents, both hospitalists and community pediatricians, demonstrated that physicians from an academic center were more likely than physicians from a community hospital to report often or almost always using 3 of 5 evidence based-therapies and recommended tests and were more likely to report never or rarely using 4 of 7 tests/therapies of unproven benefit. Practitioners who attended for >90 days per year reported greater adherence to the evidence-based therapy for asthma regarding ipratropium treatment in the first 24 hours and less use of 5 of 7 therapies/tests of unproven benefit (Table 2).
In multivariable regression analyses controlling for gender, race, days attending per year, years out of residency, hospital practice type, and completion of additional fellowship/graduate training, hospitalists were significantly more likely to report often or almost always using the following evidence-based therapies for asthma: albuterol (odds ratio [OR]: 3.42; 95% confidence interval [CI]: 1.08–10.85) and ipratropium in the first 24 hours (OR: 3.56; 95% CI: 1.82–6.97). Hospitalists were more likely to report often or almost always using steroids in the treatment of asthma, but the difference was of only borderline statistical significance (OR: 3.96; 95% CI: 0.93–16.91). With respect to the first UTI in infants, hospitalists were significantly more likely to report obtaining the AAP-recommended renal ultrasound (multivariate OR: 4.86; 95% CI: 1.26–18.79) and voiding cystourethrogram (OR: 3.27; 95% CI: 1.19–8.98), compared with community pediatricians (Table 3).
In addition, hospitalists were significantly more likely than community pediatricians to report rarely or never using tests and therapies of unproven benefit for bronchiolitis: namely, levalbuterol (OR: 5.33; 95% CI: 2.63–10.79), inhaled steroid therapy (OR: 5.25; 95% CI: 2.46–11.21), and oral steroid therapy (OR: 5.37; 95% CI: 2.67–10.79). In contrast to the univariate analysis, however, there was no statistically significant difference in the use of albuterol for bronchiolitis (OR: 1.30; 95% CI: 0.53–3.20). Hospitalists were also significantly more likely than community pediatricians to report rarely or never using the following diagnostic tests not proven typically to alter management for routine gastroenteritis: stool cultures (OR: 2.64; 95% CI: 1.27–5.48) and rotavirus testing (OR: 3.46; 95% CI: 1.17–10.23). Finally, for asthma, hospitalists were significantly more likely to report rarely or never using ipratropium after 24 hours (OR: 1.93; 95% CI: 1.01–3.68) (Table 4).
Variability in Care
In evaluation of all of the reported management decisions, including those without an evidence base to support a particular decision, hospitalists and community pediatricians made significantly different management decisions for 36 (75%) of the 48 tests and therapies evaluated (P < .05). The inpatient condition with the most variability was bronchiolitis, for which hospitalists and community pediatricians reported significantly different management decisions for 11 of 11 diagnostic and treatment modalities studied. Pneumonia had the least variability in management; hospitalists and community pediatricians reported significantly different management decisions for only 2 of the 6 studied tests and therapies.
Hospitalists reported using evidence-based therapies and national guideline-recommended tests in the care of hospitalized children significantly more often than did community pediatricians. In addition, hospitalists were less likely than community pediatricians to report using therapies and tests of unproven benefit and tests specifically not recommended in national AAP-endorsed guidelines. These differences persisted even after controlling for gender, race, years out of residency, days attending per year, hospital practice type, and completion of graduate/fellowship training.
To our knowledge, this is the first study to indicate that the quality-of-care processes in hospitalist systems may differ from those in traditional care systems. Two previous adult studies demonstrated decreased mortality rates in hospitalist systems, but the mechanisms through which such improvements were achieved were not clear.13,14 In pediatric ward settings, no studies have demonstrated differences in mortality rates or quality-of-care processes in hospitalist systems, compared with traditional systems.5–8 The differences in inpatient care processes we identified also suggested a mechanism through which hospitalists may achieve cost savings in hospital care, beyond those associated with decreased length of stay; hospitalists' decreased use of therapies and tests of unproven benefit likely yields cost savings because of reductions in unnecessary laboratory and pharmacy use.6,7,9,11
There are several limitations to this study. The first limitation is that the study is based on self-reporting of clinical decisions, which may not represent actual practice; true compliance with evidence-based guidelines may be less than that reported by providers. There is little reason to postulate, however, that hospitalists and community pediatricians would differ significantly in the accuracy of their self-reports. It is likely that, if evidence-based practices are known, then the 2 groups would have similar tendencies to overreport their compliance. The second limitation is that adherence to practice guidelines is only a proxy for quality of care and practice guidelines can change over time or can be called into question. Specifically, the utility of a renal ultrasound after the first UTI was called into question by Hoberman et al19 in 2003, and pediatricians' reports of not obtaining renal ultrasound scans might be based on this new evidence. However, we needed to define evidence-based practices on the basis of sources that summarize the literature, and AAP clinical practice guidelines and BMJ Clinical Evidence are generally accepted sources of evidence-based practice guidelines. A third limitation is the possibility that our sample of community pediatricians was not representative of pediatricians throughout the nation. Because pediatric hospitalists as a group have been highly motivated to participate in inpatient pediatric research, we anticipated and achieved a high response rate for that group. We did not expect to achieve such a high response rate from the random nationwide sample of community pediatricians, especially given the Medical Marketing Service historical response rates of 2% to 5% for national surveys administered to a random sample of providers.16 Although a response rate of 15% exceeded our expectations, selection bias might have occurred. We think, however, that the effect of such a bias would likely have been to minimize differences between the groups, because of selection for a cohort of community pediatricians interested in a study of inpatient care management and confident in their treatment of inpatients. In addition, although the proportion of pediatrician respondents was small, the respondents seemed to be similar demographically to the AMA nationwide list of pediatricians.
Previous studies demonstrated improved outcomes for patients cared for by subspecialists, with more experience and higher patient volume, in areas such as pediatric cardiac surgery, neonatal intensive care, and pediatric oncology.20–24 In addition, studies of specialist versus generalist care without a specific focus on patient volume demonstrated improved outcomes with specialists for diseases as diverse as depression, allergic rhinitis, and heart failure.25–27 Hospitalists both specialize in inpatient pediatrics and attend a greater number of inpatients per year, and this study suggests that this expertise leads to greater adherence to evidence-based therapies. The correlation between reported management decisions and outcomes needs to be demonstrated and should be a focus of additional research. The AAP Section on Hospital Medicine endorsed this need to measure outcomes in their recently published report.28
A fundamental problem in inpatient pediatric medicine is the lack of a robust evidence base regarding the optimal treatment of many common inpatient illnesses. Many of these conditions have been investigated only in relatively small studies without the necessary power to measure differences in mortality rates or subtler but clinically important effects of commonly used therapies. Research networks such as the Pediatric Research in Inpatient Settings Network have the potential to address these knowledge gaps through multicenter studies of inpatient care and coordination of research efforts. Hospitalist models have emerged as potentially beneficial models of care delivery,2–4,13 but extensive work is needed to more effectively measure and improve the quality of inpatient care.
This study was supported by the Children's Hospital Boston Resident Research Fund and the AAP Resident Research Fund. Dr Conway is the recipient of a training grant from the Robert Wood Johnson Foundation Clinical Scholars Program. Dr Landrigan is the recipient of a career development award from the Agency for Healthcare Research and Quality (grant K08 HS13333). None of the funding sources had any role in the design or conduct of the study.
We thank Dr Frederick Lovejoy for his support of this research and Daniel Kinnamon for his help with statistical analysis. We also acknowledge Mary Lou Walters of Medical Marketing Services for help in identifying relevant community pediatricians and Bruce Sprague for online survey design and management.
- Accepted March 22, 2006.
- Address correspondence to Patrick H. Conway, MD, Robert Wood Johnson Clinical Scholars Program, University of Pennsylvania, 423 Guardian Dr, Blockley Hall, 1303A, Philadelphia, PA 19104. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵Society of Hospital Medicine. Sixth annual meeting attracts largest number of hospitalists ever. Hospitalist. 2003;7 :9– 10. Available at: www.hospitalmedicine.org/Content/NavigationMenu/Media/PressReleases/Past_Press_Releases.htm. Accessed March 13, 2006
- ↵Adler J. Hospitalist specialty takes off. Herald News. December 27, 2005. Available at: www.northjersey.com/page.php?qstr_exJpcnk3ZjcxN2Y3dnF1ZUVFeXkyJmZnYmVsN2Y3dnF1ZUVFeXk2ODQ2ODM1. Accessed March 13, 2006
- ↵Lye PS, Rauch D, Ottolini M, et al. Pediatric hospitalists: report of a leadership conference. Pediatrics.2006;117 :1122– 1130
- ↵Bellet PS, Whitaker RC. Evaluation of a pediatric hospitalist service: impact on length of stay and hospital charges. Pediatrics.2000;105 :478– 484
- ↵Landrigan CP, Srivastava R, Muret-Wagstaff S, et al. Impact of a health maintenance organization hospitalist system in academic pediatrics. Pediatrics.2002;110 :720– 728
- ↵Ogershok PR, Li X, Palmer HC, Moore RS, Weisse ME, Ferrari ND. Restructuring an academic pediatric inpatient service using concepts developed by hospitalists. Clin Pediatr (Phila).2001;40 :653– 660
- ↵Wells RD, Dahl B, Wilson SD. Pediatric hospitalists: quality care for the underserved? Am J Med Qual.2001;16 :174– 180
- ↵Seid M, Quinn K, Kurtin PS. Hospital-based and community pediatricians: comparing outcomes for asthma and bronchiolitis. J Clin Outcomes Manage.1997;4 :21– 24
- ↵Tovey D, Bedford M, Brunnhuber K, et al. Clinical Evidence: The International Source of the Best Available Evidence for Effective Health Care. London, United Kingdom: BMJ Publishing Group; 2005;13 :709
- ↵American Academy of Pediatrics, Committee on Quality Improvement, Subcommittee on Urinary Tract Infection. Practice parameter: the diagnosis, treatment, and evaluation of the initial urinary tract infection in febrile infants and young children. Pediatrics.1999;103 :843– 852
- ↵Centers for Disease Control and Prevention. Managing acute gastroenteritis among children: oral rehydration, maintenance, and nutritional therapy. MMWR Recomm Rep.2003;52(RR-16) :1– 16
- ↵American Academy of Pediatrics. Statement of endorsement: managing acute gastroenteritis among children: oral rehydration, maintenance, and nutritional therapy. Pediatrics.2004;114 :507
- ↵Hannan EL, Racz M, Kavey RE, et al. Pediatric cardiac surgery: the effect of hospital and surgeon volume on in-hospital mortality. Pediatrics.1998;101 :963– 969
- Jenkins KJ, Newburger JW, Lock JE, Davis RB, Coffman GA, Iezzoni LI. In-hospital mortality for surgical repair of congenital heart defects: preliminary observations of variation by hospital caseload. Pediatrics.1995;95 :323– 330
- Stiller CA, Draper GJ. Treatment centre size, entry to trials, and survival in acute lymphoblastic leukaemia. Arch Dis Child.1989;64 :657– 661
- ↵Percelay JM, Strong GB. Guiding principles for pediatric hospitalist programs. Pediatrics.2005;115 :1101– 1102
- Copyright © 2006 by the American Academy of Pediatrics