Despite efforts to provide high-quality healthcare, Americans die from medical errors each year and many patients do not receive recommended medical care. Risk is particularly acute during times of hospitalization.1-4 In response, the Institute of Medicine (IOM, now the Academy of Medicine) has released “Crossing the Quality Chasm: A New Health System for the 21st Century,” providing a framework to guide delivery and measurement of high-quality healthcare.5
Although the IOM framework has motivated the development of quality improvement (QI) and quality measurement initiatives, relatively few resources have been allocated to improving the quality of pediatric inpatient care.6,7 The resultant gap in our knowledge of quality and safety of pediatric hospital-based care is further widened by the variability of settings in which children are hospitalized. These settings include freestanding children’s hospitals, children’s hospitals nested within larger hospitals, and community hospitals, defined as general, nonuniversity, and nonchildren’s hospitals.8
Although almost three-quarters of children needing hospitalization are cared for outside of freestanding children’s hospitals, we know particularly little about the quality and safety of pediatric hospital-based care outside of these settings.6,9 Therefore, our scoping review aims to summarize literature regarding the quality and safety of pediatric inpatient care within community hospitals.
We used a scoping review approach because this methodology, by design, is utilized to synthesize evidence and map existing literature and is particularly useful when a body of literature is heterogeneous, rendering a more targeted systematic review approach infeasible.10 This methodology thereby provided an organized approach to answer our broad research question, “What evidence exists regarding the quality and safety of pediatric inpatient care in United States community hospitals?“ We followed the scoping review guidelines put forth by the Joanna Briggs Institute and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guideline Extension for Scoping Reviews.10,11
Data Sources and Search Strategies
We searched Medline, Medline-In-Process, Embase, the Cochrane Database of Systematic Reviews, the Cumulative Index to Nursing and Allied Health Literature (CINAHL), PsycINFO, and Scopus for studies that reported at least one outcome related to healthcare quality or patient safety and involved pediatric patients (aged <18 years) receiving inpatient care at a community hospital. Outcomes included measures from the IOM-defined aims of quality healthcare: (1) safety, (2) effectiveness, (3) efficiency, (4) timeliness, (5) patient-centeredness, and (6) equity (Appendix Table 1). Terms were searched as controlled vocabulary in applicable databases (Medline, Embase, CINAHL, PsycINFO) and as keywords in all databases. Search strategies tailored to each database were developed, tested, and refined in collaboration with a reference librarian. Date parameters for retrieval were set from 1989 to the search date, with the start date chosen to correspond with the establishment of the Agency for Health Care Policy and Research, currently known as the Agency for Healthcare Research and Quality (AHRQ), targeting literature produced in the wake of the AHRQ emphasis on quality in healthcare. Results were limited to articles published in English. Searches were conducted on October 24 and 25, 2016. Complete search strategies can be found in Appendix Methods.
Independent authors performed handsearches of Academic Pediatrics, BMJ Quality & Safety, Hospital Pediatrics, JAMA Pediatrics, and Pediatrics Quality Reports for the five years preceding the search date (July 2011-October 2016).
Study Selection and Definitions
To identify studies conducted in community hospitals, we operationalized the definition of community hospital proposed by Percelay.8 We used “community”, “general”, “nonuniversity”, and “nonchildren’s” as search terms and operationalized these through the addition of qualifying descriptors (Table 1).
Studies were excluded if they (1) were performed outside of the United States, as community hospital definitions may differ by country; (2) included participants aged >18 years and did not report any pediatric-specific results; (3) were performed exclusively in tertiary hospitals; (4) evaluated only outpatient or emergency department care; (5) did not report any results specific to community hospitals; (6) did not report any data-driven or parent/patient-reported measures of safety, effectiveness, efficiency, timeliness, patient-centeredness, or equity in their results; and (7) were case reports, case series, editorials, or abstracts without an associated full-text article. We read commentaries and literature reviews related to our objectives and reviewed their references to identify additional articles, but did not include these manuscripts.
Two authors independently reviewed each abstract, and full-text articles were reviewed if one or more authors determined that the abstract met the inclusion criteria. Two authors then independently reviewed each full-text article to determine whether the article met the criteria for the final review. Disagreements were resolved through consensus after discussion and review with the entire research team, and reasons for exclusion were recorded.
Charting the Results and Data Synthesis
We used a standardized charting form to collect information regarding study design, community hospital terms, population descriptors, IOM aims of quality healthcare, and outcome measures. To minimize bias in data collection, information from each full-text article was extracted independently by two authors, and differences in extraction were resolved through discussion and re-review among the same two authors. To evaluate the quality of included evidence, two authors (JCL and JKL) independently assessed the risk of bias for each study using modified Newcastle-Ottawa Quality Assessment Scales (NOS), with disagreements resolved through discussion and re-review among the same two authors. For cohort studies, five of the eight NOS domains were relevant and applied to all included studies (maximum score 6). Using the NOS adapted for cross-sectional studies, six of the seven domains were relevant and applied (maximum score 9). For cross-sectional studies, we defined the risk of bias to be low for scores ≥8, moderate for scores 5-7, and high for scores ≤4, consistent with previous work.12 For cohort studies, we similarly defined these strata by scores ≥5, 3-4, and ≤2, given the lower maximum score for this study type.
We categorized studies as either observational, defined as cohort or cross-sectional studies of usual healthcare delivery, or interventional, defined as studies evaluating the development and/or implementation of an intervention designed to improve healthcare quality. We further categorized the studies into the following four overarching medical domains: (1) neonatal, (2) pediatric medicine, (3) surgery, or (4) radiology.
After removal of duplicates, our search identified 2,068 abstracts for screening (Figure). Of these, 1,777 did not meet the inclusion criteria, leaving 291 articles for full-text review. Of these, 43 articles met all the inclusion criteria, and one additional article was included from search of references, resulting in a total of 44 articles.
Study designs, patient populations, and quality outcome measures were heterogeneous. We identified only one randomized controlled trial (RCT). A total of 30 articles were observational studies, 27 of which used retrospective cohort or cross-sectional designs; the remaining three used prospective cohort designs (Table 2). Of these studies, 20 involved multiple hospitals, whereas 10 were conducted at a single community hospital. Sample sizes ranged from 29 (single-site) to 107,727 (multisite) patients. Twenty-two studies aimed to compare quality outcomes at community hospitals with other hospital types, of which 16 performed risk-adjusted analyses (detailed findings of observational studies are summarized in Appendix Table 2). The remaining 14 articles were interventional studies (Table 2). Of these, 12 (86%) reported improvement in quality outcomes after implementation (detailed findings of interventional studies are summarized in Appendix Table 3).
The included studies evaluated quality outcomes addressing all six of the IOM aims of quality healthcare, with safety, effectiveness, and efficiency being the most predominant (Table 3). Patient-centeredness and timeliness were infrequently addressed, and only one study assessed equity.
Risk of bias was moderate or high for 27 (69%) of the observational and interventional cohort studies and four (100%) of the included cross-sectional studies (Table 2), with the median NOS score being 4 (range: 0-6) for cohort studies (Appendix Table 4) and 4 (range: 3-6) for cross-sectional studies (Appendix Table 5). The higher risk of bias was largely driven by low comparability scores due to inadequate risk adjustment or statistical reporting. Of the 12 studies with low risk of bias, 11 (92%) were multisite, 9 (75%) used large regional or national databases, and half reported quality outcomes limited to hospital charges and/or mortality.
Five multisite studies focused on neonatal care,13,14,16,19,20 of which four examined outcomes associated with the transfer of neonates to or from community hospitals to tertiary care hospitals, such as neonatal morbidity, readmission, completion of preventative health measures and screening, parent satisfaction, and hospital charges.14,16,19,20 For example, in a study of extremely premature very low birth weight infants born in Hawaii, Kuo et al. demonstrated that the odds of retinopathy of prematurity was 2.9 times higher for infants born at a community hospital and transported to their tertiary center compared with those inborn at the tertiary center (P = .02).16 The fifth study examined neonatal mortality by site of birth, demonstrating that, among infants with birth weights less than 2,000 g, birth at a hospital with a community neonatal intensive care unit (NICU) was associated with 1.4 times higher odds of risk-adjusted mortality compared with birth at a regional NICU (P < .001).13
Three studies evaluated the quality of neonatal care at a single community hospital.15,17,18 Quality outcomes were heterogeneous, including utility of rebound bilirubin levels for infants with jaundice, morbidity of neonates requiring mechanical ventilation, and provision of breastfeeding advice to mothers of breastfeeding infants. For instance, Meadow et al. attempted to determine the quality of care for ventilated neonates at one community NICU compared with a tertiary hospital.18 They found no difference in days on ventilation or need for home oxygen therapy between the community hospital and the tertiary center, although P values and effect sizes were not reported for these outcomes and analyses were not adjusted beyond matching on birthdate and birth weight.
Nine multisite studies explored the quality and safety of pediatric medical care across hospital types.21-23,25-30 Of these, four were conducted using the Kids’ Inpatient Database (KID)28-30 or the National Inpatient Sample (NIS),27 two were conducted using other national databases,21,25 and two were conducted using electronic medical record data.22,26 All of the KID and NIS studies examined hospital charges, either alone or in conjunction with mortality. Two of these studies found no differences in risk-adjusted hospital charges between children’s hospitals and community hospitals (for burn injuries),28,29 whereas two found that hospitalization at community hospitals was associated with lower risk-adjusted charges (for asthma and sepsis).27,30 The remaining five studies evaluated diverse quality outcomes such as medication errors, therapeutic drug monitoring, practice guideline compliance, antibiotic prescribing, or hospital-to-home transition summary scores.21-23,25,26
Three single-site studies examined quality outcomes for pediatric medical patients, including mortality, outcome ratings for dehydration, and measures of peripherally inserted central catheter (PICC) safety and effectiveness.24,31,32 For example, Frank et al. evaluated safety of care in one community hospital pediatric intensive care unit (PICU) compared with a tertiary hospital using the Pediatric Risk of Mortality (PRISM) score.24 They reported that the observed number of deaths in their community PICU did not differ significantly from the number of deaths predicted in a tertiary center (23 vs 33, respectively, P > .2).
Three multisite studies examined quality outcomes among children with surgical conditions, including surgical complications, readmission, and hospital charges.34,35,37 For example, Kelley-Quon et al. examined outcomes following surgery in infants with hypertrophic pyloric stenosis and found that infants who received their surgery at community hospitals had twice the odds of a surgical complication compared with those at children’s hospitals (P = .027).34 They also examined how the risk of appendiceal perforation differed by hospital type, finding that black children who received their surgery at children’s hospitals had twice the odds of appendiceal perforation compared with those who received care at community hospitals.35
In addition, two studies evaluated quality and safety outcomes for pediatric surgical care in a single community hospital.33,36 For example, Beaty et al. prospectively evaluated the incidence of missed injuries in hospitalized pediatric trauma patients, reporting that the incidence of missed injury was 33% when admission evaluation was performed by a trauma surgeon alone compared to 11% when performed by a pediatric doctor or a trauma surgeon and a pediatric doctor together (P < .001).33
Three multisite studies examined quality and safety outcomes associated with radiographic imaging in community hospitals, including radiation dosing and frequency of preoperative imaging modalities and accuracy.38,39,41 For instance, Marin et al. found substantial variation in radiation dose across hospital types, with children’s hospitals delivering lower median radiation doses than academic and community hospitals.39 Similarly, Saito et al. demonstrated increased use of radiating modalities when evaluation was performed at community hospitals, with four times higher odds of computed tomography (CT) and five times lower odds of ultrasound use compared to a children’s hospital.41
Two single-site studies also evaluated quality and safety outcomes associated with the use of radiographic imaging for pediatric appendicitis in a community hospital.40,42 For example, York et al. demonstrated that, compared with nonimaged patients, patients who underwent diagnostic imaging for appendicitis experienced a significant time delay from initial evaluation to surgery and incurred significantly higher hospital charges, whereas there were no significant differences in intraoperative findings, antibiotic requirements, and surgical complications between the groups.42
We identified six studies that evaluated interventions to improve healthcare quality for neonates in community hospitals; two involved telemedicine.43-48 Hall et al. described “Telenursery,” a program linking regional perinatal centers with a large academic neonatal practice through real-time teleconferencing, in addition to providing weekly educational conferences.45 After its implementation, there was an increase in the state-recommended delivery of very low birth weight infants at the regional perinatal center from 24% to 33% (P < .05); clinical outcomes were not discussed. In a similar study, Sable et al. found that a videoconferencing system for cardiologists from an academic center to guide care in a community setting provided diagnostic services more quickly (28 minutes vs 12 hours) and had high diagnostic accuracy.47 The remaining four studies evaluated heterogeneous interventions such as a maternal education program to reduce shaking injuries to infants or implementation of evidence-based order sets to reduce early onset group B streptococcal disease in neonates.43,44,46,48 Five of the six neonatal studies demonstrated improved quality outcomes after intervention.43-47
Seven studies evaluated QI interventions for children at community hospitals,49-55 three of which described interventions to improve quality of management of respiratory diseases.49,51,53 Dayal et al. evaluated the implementation of respiratory illness order sets and an asthma pathway, demonstrating a 41% reduction in asthma hospitalization cost per patient (P < .05), reduced bronchodilator use for all respiratory illnesses, and no change in readmission rates.49 Similarly, Nkoy et al. evaluated an asthma “Evidence-Based Care Practice Model” implemented at seven community hospitals and demonstrated a nonsignificant reduction in readmissions (P = .12), as well as lower hospitalization costs (P = .05).53 Using QI methods, Kuhlmann et al. demonstrated improved compliance from 43% to 97% with the Asthma Home Management Plan of Care measure.51 The remaining four studies evaluated heterogeneous interventions, including telemedicine critical care consultations, use of I-PASS, and consolidation of pediatric care onto one hospital unit.50,52,54,55 Six of the seven pediatric studies (86%) demonstrated improved inpatient quality outcomes after intervention,49,51-55 but only one study was multisite.53
We identified only one study evaluating a surgical intervention aimed at improving the quality of pediatric care.56 Kelley-Quon evaluated the impact of a community hospital partnering with an Academic Medical Pediatric Trauma Center to become a Level II Pediatric Trauma Center (PTC). After achieving Level II PTC designation, they reported that children treated at the community hospital had reduced rates of CT use, transfers, and in-hospital mortality (from 81% to 51%, 8.5% to 2.5%, and 2% to 0.4%, respectively, P < .05 for all) compared to those treated predesignation.
Within the domain of radiology, our review identified no interventional studies.
In this scoping review of the quality and safety of pediatric inpatient care in community hospitals, we identified 44 studies applying heterogeneous study designs and evaluating diverse patient populations and quality outcomes. We identified only one RCT in our search; all the remaining studies applied observational designs.
We found only three clinical areas that were explored in multiple studies, with consistent directionality of results: (1) perinatal regionalization, (2) telemedicine, and (3) imaging radiation. The limited evidence identified in our review suggests that delivery of early premature infants at community hospitals, rather than at tertiary hospitals, may increase risk of neonatal morbidity and/or mortality,13,16,46 that use of telemedicine may improve the effectiveness and efficiency of intensive or specialized care in community settings,45,47,52,55 and that CT use and radiation doses may be higher in community hospitals compared with other settings.38,39,41 However, even within these clinical domains, the literature was limited in amount and heterogeneous; additional research is needed to systematically review the effect of individual interventions or particular community hospital quality outcomes compared with other hospital types.
Our search identified only 14 studies evaluating QI interventions within community hospitals over the almost 30-year review period. Although limited in number, 86% of these demonstrated improvements in healthcare quality and safety, providing a positive “proof of concept” that pediatric care in community hospitals can be improved by multidisciplinary efforts and be sustained over time.43-47,49,51-56 As pediatric departments within community hospitals may have limited resources, aligning pediatric QI efforts with adult initiatives within the same hospitals could prove advantageous. However, we did not identify any studies meeting our inclusion criteria that took this approach. Alternatively, QI collaboratives across structurally diverse hospitals may provide valuable infrastructure for QI in community hospitals. For example, the Value in Inpatient Pediatrics Network has conducted several multisite QI initiatives that have engaged both children’s and community hospitals.57-60 However, none of these studies have reported community hospital-specific quality outcomes, resulting in their exclusion from this review. In future study, researchers may consider separating community hospital results from those of pediatric hospitals to highlight the effect of community hospital-specific QI efforts and to allow valuable direct comparisons between hospital types.
Many studies identified in our scoping review were conducted at single hospitals. Sample sizes were often very small and power calculations were rarely reported, raising questions about the validity of the “nonsignificant” differences reported by some. Although such single-center studies provide valuable information for local improvement of inpatient care, by design the findings from these studies are unlikely generalizable to other community hospital systems. Without inclusion of another hospital type and/or quality measures with clear national benchmark for comparison, the additional conclusions that can be drawn from these studies are limited. In comparison, many of the multicenter studies had large study samples, but more than half used data registries with limited data to evaluate outcomes, clinical context, or important covariates.13,20,21,25,27-30,34,35,37,39 The most frequently reported outcomes in these data sets—hospital charges and lengths of stay—are of limited utility in understanding healthcare quality without clear benchmarks. As a result, the evidence-base from which to draw conclusions about quality and safety in community hospitals is very limited.
Therefore, our review highlights a great need for additional research in community hospital medicine and the need for high-quality evidence generation. Risk of bias was moderate or high for the majority of included studies because of inadequate risk adjustment or statistical analysis. In the future, multicenter collaborations may help to connect research methodologists with community hospital teams to aid in the application of robust study designs and analytic techniques. Multisite collaboration may also overcome the limitation of small sample sizes that are a reality at many community hospitals.
Our study findings must be considered in the setting of several methodological limitations. The lack of a standard definition for a community hospital has led to inconsistent terms and hospital definitions used in the literature. It is possible that, while following our systematic approach to define a community hospital, we inadvertently missed relevant studies that used different terms. We also excluded unpublished articles. Given that publication bias tends to favor studies with significant associations, it is possible that some studies with insignificant changes in quality outcomes were missed. Finally, in our exclusion of all non-US studies, we may have unknowingly missed literature from countries with community hospital definitions similar to those in the United States.
Recognizing that more than half of all children admitted to hospitals in the US receive their care at community hospitals, understanding healthcare quality in community hospitals is important. This scoping review underscores the need for additional research and higher quality evidence to determine the quality of pediatric inpatient care in these settings and identifies some particularly wide gaps that could be targeted in future research. Acknowledging that further research is necessary to address all aims of quality healthcare, markedly few studies have examined timeliness, equity, or patient-centeredness. Collaborations between academic medical centers and community hospitals may be an effective means to connect researchers with community hospital clinical teams to facilitate the application of robust study designs and analytic approaches and to facilitate multisite investigations. Research in this field would benefit from a standardized definition of a community hospital that could be consistently applied in research and QI endeavors.
The authors have no potential conflicts of interest to disclose.
Jana Leary was supported by the National Center for Advancing Translational Sciences, National Institutes of Health, Grant Number 5TL1TR0001062-03. JoAnna Leyenaar was supported by the Agency for Healthcare Research and Quality (K08HS024133).
The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or the AHRQ.