CONTEXT: Despite the availability of evidence-based guidelines for the management of pediatric asthma, health care utilization remains high.
OBJECTIVE: Systematically review the inpatient literature on asthma quality improvement (QI) and synthesize impact on subsequent health care utilization.
DATA SOURCES: Medline and Cumulative Index to Nursing and Allied Health Literature (January 1, 1991–November 16, 2016) and bibliographies of retrieved articles.
STUDY SELECTION: Interventional studies in English of inpatient-initiated asthma QI work.
DATA EXTRACTION: Studies were categorized by intervention type and outcome. Random-effects models were used to generate pooled risk ratios for health care utilization outcomes after inpatient QI interventions.
RESULTS: Thirty articles met inclusion criteria and 12 provided data on health care reutilization outcomes. Risk ratios for emergency department revisits were: 0.97 (95% confidence interval [CI]: 0.06–14.47) <30 days, 1.70 (95% CI: 0.67–4.29) for 30 days to 6 months, and 1.22 (95% CI: 0.52–2.85) for 6 months to 1 year. Risk ratios for readmissions were: 2.02 (95% CI: 0.73–5.61) for <30 days, 1.68 (95% CI: 0.88–3.19) for 30 days to 6 months, and 1.27 (95% CI 0.85–1.90) for 6 months to 1 year. Subanalysis of multimodal interventions suggested lower readmission rates (risk ratio: 1.49 [95% CI: 1.17–1.89] over a period of 30 days to 1 year after the index admission). Subanalysis of education and discharge planning interventions did not show effect.
LIMITATIONS: Linkages between intervention and outcome are complicated by the multimodal approach to QI in most studies.
CONCLUSIONS: We did not identify any inpatient strategies impacting health care reutilization within 30 days of index hospitalization. Multimodal interventions demonstrated impact over the longer interval.
- CAC —
- Children’s Asthma Care
- CI —
- confidence interval
- ED —
- emergency department
- LOS —
- length of stay
- QI —
- quality improvement
- RCT —
- randomized controlled trial
Asthma is the most common pediatric chronic disease, affecting more than 10% of American children (>7 million children younger than 18 years).1 In addition, asthma ranks as the leading nonsurgical diagnosis for pediatric hospital admission, totaling costs of $570 million annually.2 Despite an evidence-based guideline created by the National Institutes of Health,3 overall morbidity among youth with asthma, whether measured by attack rates, emergency department (ED) visits, or hospitalizations, has not significantly decreased in the last decade.4 An asthma-related ED revisit is common after hospitalization, with more than 40% of children returning to the ED within 12 months, based on findings from an urban pediatric facility.5 In addition, based on administrative data from over 40 children’s hospitals in the country, asthma readmission rates approach 25% within 12 months of index hospitalization.6
In an effort to address the high health care use associated with asthma in children, numerous quality improvement (QI) interventions have been studied. Recently, significant reductions in health care use were demonstrated by the authors of comprehensive, longitudinal QI projects over the continuum of outpatient to inpatient care.7,8 However, such comprehensive projects are still rare and much pediatric asthma QI is focused on a particular point of care, such as the ED or the hospital ward, or only a single intervention type, such as patient education. Although systematic reviews exist on outpatient asthma interventions9 as well as ED-based asthma interventions,10 the inpatient-initiated QI efforts for asthma have not been systematically described. The goal of this study was (1) to systematically review the literature for asthma interventions initiated in the pediatric inpatient setting, and (2) to assess the impact of these interventions on subsequent health care use, specifically ED revisit and readmission rates.
Search and Study Eligibility Criteria
We sought articles describing interventions to improve care for children with asthma, which specifically enrolled patients at an index hospitalization and were initiated in the inpatient setting. This systematic review was conducted and reported by following recommendations outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement.11 In collaboration with a medical librarian, we developed a search strategy by using search terms such as asthma, hospitalization, discharge, and readmission; this strategy was adapted from previous work12,13 (see Supplemental Information). We searched 2 databases, PubMed (January 1, 1991–November 16, 2016) and the Cumulative Index to Nursing and Allied Health Literature (January 1, 1991–November 16, 2016), and created a review protocol to identify interventional studies initiated in the inpatient setting and intended to improve asthma care. Bibliographies of included articles were also reviewed to include additional relevant articles.
The following inclusion criteria were established: (1) improvement effort initiated in the inpatient setting focused on children hospitalized with asthma exacerbations; (2) interventional study design; (3) reported outcomes included either clinical status (ie, symptom free days), functional status (ie, days missed), health services (ie, readmissions), and adherence to guidelines and/or recommendations (ie, Children’s Asthma Care [CAC] performance). Briefly, the CAC measures were introduced by The Joint Commission to improve hospital asthma care: CAC-1 measures use of short-acting relievers, CAC-2 measures use of systemic corticosteroids, and CAC-3 measures use of home management plan for patients and caregivers. We also limited our search to abstracts available in English and articles published after 1990 because this was the year of the original National Institutes of Health guidelines for asthma management. Studies failing to separate pediatric data from adult data were excluded. Studies that focused on work in countries with low or lower-middle incomes were excluded because interventions in these countries may not be generalizable to the United States.14
Study Selection and Data Collection
Two reviewers (K.P. and S.R.) split all eligible abstracts and screened by using title and abstracts to exclude topically irrelevant studies. The 2 reviewers independently reviewed each full-text article identified as potentially relevant after reading the title and abstract and a κ statistic was calculated to assess screener agreement. Discrepancies were resolved by re-review and consensus. The reviewers independently extracted data from each included study by using a standardized data collection form.
All studies reporting interventions for asthma care improvement initiated at an index hospitalization were included in the summary and qualitative analysis. We grouped interventions qualitatively by the primary type of QI strategy employed, specifically standardization of care, education, discharge planning, or multimodal. In addition, we also categorized reported outcomes to better describe the scope of QI work in the area of inpatient asthma care. Similar methodology was employed in a systematic review of outpatient asthma QI efforts.9
We focused the meta-analytic component of this review on ED revisit rates and hospital readmission rates for several reasons: (1) ED revisits and readmissions were the most commonly reported of the diverse outcomes and thus meta-analysis was feasible, (2) government legislation has designated reduction in avoidable rehospitalization as a target for health care cost savings with penalties for hospitals with higher rates of readmission, and (3) ED use for chronic disease is often considered potentially preventable and frequently appears in pay for performance contracts. Numerators (number of patients with either ED visit or readmission) and denominators (total number patients in sample) for both control and intervention populations were extracted from all articles where these data were reported. The 2 reviewers independently extracted this information and discrepancies were cross-checked and then resolved based on consensus.
Meta-analysis of the different reported outcomes was not considered feasible because other outcomes were not reported consistently across enough studies, except for length of stay (LOS). Although LOS is an important quality measure in the care of inpatients, analysis of this outcome is beyond the scope of the current review because ED interventions have been previously linked to shortened inpatient LOS15 and have already been systematically reviewed.10 (See Supplemental Table 4 for all reasons for study exclusion from the meta-analysis).
The interval of time between index admission and representation for care (either ED visit or readmission) was grouped into the following 3 categories for analysis: short (<30 days), medium (30 days–6 months), and long (6–12 months). The short time interval was based on the Centers for Medicare and Medicaid 30-day readmission metric, and the medium time and long time intervals were based on reported time increments in the available data. If data could not be extracted separately for each outcome (ie, ED revisit and hospital readmission were treated as a composite outcome), the study was not included in the meta-analysis. Also, if the reuse data for ED revisits and/or readmissions were included as a balancing measure and decreasing reuse was not an intended goal of the QI intervention, the study was not included in the meta-analysis. In addition, some studies reported data for reuse outcomes at more than 1 time interval or more than 1 setting; such data were analyzed separately for each outcome and time interval that could be extracted.
For the initial meta-analyses, we took the approach of grouping studies by targeted outcomes, assuming that the QI strategies employed were similar enough to be considered together. However, we also undertook subanalyses based on more granular groupings of the interventions where feasible, specifically those categorized as multimodal, education, and discharge planning. The subanalysis was not conducted on the standardize care category because no studies in this group targeted the outcomes chosen for meta-analysis. In our subanalyses, studies with medium and long time intervals were grouped together to increase the sample size.
Random effects models were used to generate summary statistics by interval and study heterogeneity was assessed by using the I2 statistic. Analyses were performed by using Comprehensive Meta-Analysis 2.0.
Assessment of Methodological Quality
We developed a tool for methodological quality assessment adapted from a tool used in a previous systematic review of pediatric QI interventions,16 along with elements derived from the Standards for Quality Improvement Reporting Excellence guidelines, which is a published rubric for describing and reporting QI studies.11 This tool was used to assess the quality of the studies included in the quantitative analysis in regards to the ability to draw meaningful conclusions regarding the subsequent health care use outcomes of interest for this review. Sensitivity analysis was performed by removing studies that failed to meet minimum quality standards in at least 4 of the 6 domains.
Our initial search identified 1040 unique articles. Approximately 10% of abstracts were screened for inter-rater reliability and Cohen’s κ statistic was 0.9. After screening abstracts, the foremost reasons for exclusion was that the study was not initiated in the inpatient setting or that the article did not have an interventional design. Subsequently, 113 articles underwent full-text review. During the full-text review process, the leading reason for exclusion was that the article focused on an ED-based (not inpatient) cohort. A total of 27 articles were found using our search strategies, and an additional 3 articles were included based on bibliography review of those 27 articles. Therefore, in total, we included 30 studies in our primary review (Table 1) and a subset of these (12 studies) in our quantitative (meta-analysis) review. A study flow diagram is presented as Fig 1.
Qualitative Assessment: Intervention Types, Outcome Measures
Our qualitative assessment included 30 studies reporting on an inpatient initiated QI intervention for asthma. Interventions were categorized as follows based on primary QI modality (Table 2): standardize care17–25,43 (n = 9), education26–34 (n = 9), discharge planning35–38,46 (n = 5), or multimodal39–42,44,45 (if there were more than 1 primary QI modality; n = 7). Nine studies used standardization of care to improve outcomes, specifically introducing or revising a clinical pathway or initiating criteria-led discharge. Nine studies used education as the primary QI modality, for example, focusing on asthma self-management education and/or assigning a specific educator role for discharge education. Five studies focused on tools for discharge planning, for example, facilitating medications in-hand at the time of discharge, assigning a case manager to navigate barriers to care, or facilitating communication with outpatient providers. There were 7 studies that used multimodal QI efforts, combining standardizing care, education, and/or discharge planning.
Measured outcomes in the included studies varied as well and were categorized in broad groups. These groups included the following: acute asthma management, hospital resources, discharge management, patient and parent outcomes, and postdischarge outcomes. See Table 3 for a detailed summary of included outcome measures in these included studies.
Quantitative Assessment: Risk of ED Revisit or Hospital Readmission
There were 8 unique studies that targeted the outcome of ED revisit for asthma after QI intervention. A forest plot for ED revisit rates by short, medium, and long time intervals are provided in Fig 2. Two studies (Bergert et al43 and Wesseldine et al41) provided data for ED revisit at more than 1 time interval, so these studies appear more than once on the forest plot. The risk ratio for ED revisit after QI intervention for the short time interval was 0.97 (95% confidence interval [CI]: 0.06–14.47; I2 = 0%), the medium time interval was 1.68 (95% CI: 0.67–4.29; I2 = 15%), and the long time interval was 1.22 (95% CI: 0.52–2.85; I2 = 18%). We did not find a significant summary effect for QI interventions on ED revisit rates in any follow-up interval. We do not present an overall summary statistic because of excessive study heterogeneity driven by follow-up interval; the I2 within specified intervals was within acceptable thresholds.
There were 12 unique studies that targeted the outcome of subsequent hospitalization for asthma after QI intervention. A forest plot of readmission rates by follow-up interval is provided in Fig 3. Two studies (Bergert et al43 and Wesseldine et al41) provided data for readmissions at more than 1 time interval, so these studies appear more than once in the forest plot. One study (Nkoy et al44) provided data from 2 different hospital types (free-standing children’s hospital and community hospital), so this study appears more than once on the forest plot. The risk ratio for readmission rates after QI intervention for the short time interval was 2.02 (95% CI: 0.73–5.61; I2 = 0%), the medium time interval was 1.68 (95% CI: 0.88–3.19; I2 = 0%), and the long time interval was 1.27 (95% CI: 0.85–1.90; I2 = 22%). We did not find a significant effect for QI interventions on readmission rates in any follow-up interval. We do not present an overall summary statistic for all readmissions because of excessive study heterogeneity driven by follow-up interval; the I2 within specified intervals are within acceptable thresholds.
Three studies (Chong et al,24 Kelly et al,17 and Jassal et al25) reported data on our chosen reuse outcomes but were excluded from the meta-analysis because lowering reuse rates was not an intended goal of the intervention, but rather served a balancing measure. Including data from these studies would bias the meta-analysis to the null hypothesis. The primary outcome of interest in these studies was hospital LOS.
Quantitative Assessment: Subgroup Analyses by Intervention
Multimodal QI Interventions
There were 4 studies employing multimodal interventions to decrease ED revisit rates. The risk ratio for this subgroup was 1.59 (95% CI: 0.67–3.75; I2 = 0%) for the combined medium and long time interval (Fig 4A). The risk ratio for the subanalysis of the 2 studies employing multimodal interventions to decrease ED revisit rate for the short time interval was 0.97 (95% CI: 0.06–14.47; I2 = 0%). We did not find a significant summary effect in subanalyzing by multimodal interventions on ED revisit rates in any follow-up interval.
There were 6 studies employing multimodal interventions to decrease readmissions for the combined medium and long time intervals. The risk ratio for this subanalysis was 1.49 (95% CI: 1.17–1.89; I2 = 32%) (Fig 4B). The risk ratio for the meta-analysis of the 2 studies employing multimodal interventions to reduce readmissions for the <30-day interval was 2.02 (95% CI: 0.73–5.61; I2 = 0%). In this subanalysis, we found a significant summary effect for multimodal interventions on readmissions for the combined medium and long time intervals, but not the short time intervals.
Education and Discharge Planning Interventions
Subanalyses were also conducted for the education interventions (n = 3) for both ED revisit and readmission outcomes, and for discharge planning interventions (n = 2) for readmissions only. For both of these intervention groups, there was no subanalysis for the short time interval because these studies did not provide those data. For the education interventions over the combined medium and long time intervals, risk ratio for ED revisits was 1.10 (95% CI: 0.54–2.27; I2 = 0%) and risk ratio of readmissions was 0.96 (95% CI: 0.36–2.54; I2 = 90%). The discharge planning interventions over the combined medium and long time intervals risk ratio for readmissions was 1.51 (95% CI: 0.63–3.61; I2 = 0%). Thus, we found no significant summary effect when subanalyzing by education or discharge planning interventions, although these results should be interpreted with caution as there were few studies in these subgroups.
Quality Assessment of Studies in Meta-analysis
As noted above, our meta-analysis included 12 studies that reported data on subsequent health care use, and we applied the study quality assessment tool to this cohort (see Supplemental Table 5). Of the 12 studies, the majority of articles were randomized controlled trials (RCTs) (n = 5), followed by pre- and postdesign (n = 3), retrospective observational design (n = 1), matched cohort design (n = 1), non-RCT (n = 1), and quasi-experimental (n = 1). Two characteristics of high-quality QI efforts are reproducibility, and clear rationale for processes and outcomes studied; all 12 of the included articles described the intervention in sufficient detail to reproduce and provided rationale for the process or outcomes that were studied. Only 1 study (Poling36) failed to meet our minimum quality standards, and sensitivity analysis removing this study did not substantially alter the pooled risk ratios.
The focus of this review was to describe asthma QI efforts that are initiated in the inpatient setting. The inpatient literature reports on a diversity of outcomes, and we further focused the meta-analytic component of this review on studies that worked to decrease subsequent health care use after the index asthma hospitalization because these measures are increasingly part of federal reporting requirements and pay for performance contracts. Although we found variation in inpatient QI strategies, the overall systematic assessment of these studies does not clearly demonstrate that such interventions are effective to reduce ED revisits or readmissions. However, with the subanalysis of the multimodal studies, we detected a reduction in readmissions over the medium and long time intervals (30 days to 1 year) but not the short time interval (<30-day).
Overall, in our review, we suggest that isolated interventions in the hospital setting may not have much impact on improving subsequent health care use in children with asthma. Regulatory quality measures for asthma most often include reuse outcomes and failure to link interventions with these outcomes may indicate that resources are misdirected. For instance, although patient education is obviously important, it has yet to be clearly and consistently linked to reuse outcomes. Specifically, asthma action plans provided in the inpatient setting, despite extensive uptake and evaluation, do not necessarily impact future health care use as they are intended.47 More comprehensive efforts are likely to be required for reuse outcomes.
Of the individual studies reported in this review demonstrating an impact on postdischarge reuse, 1 promising commonality was a strategy of providing or facilitating access to asthma medications. Specifically, providing medications in-hand at the time of discharge, either steroids for current exacerbation, controller medications, and/or steroids for future exacerbation, is an intervention that deserves further evaluation for reducing future ED visits or admissions. Two studies in our meta-analysis cohort (Madge et al39 and Wesseldine et al41) incorporated this strategy into their interventions with decreased health care use in their respective study cohorts; specifically, Madge et al39 provided families with oral steroids and guidance on when to start them, and Wesseldine et al41 provided oral steroids if previously used by the patient and if acceptable to parents. Our qualitative review highlighted 2 other studies focused on increasing medication access in this cohort. Sauers-Ford et al37 successfully leveraged partnerships with local pharmacies to improve the percentage of patients with medications in-hand at discharge, with subsequent reductions in 90-day ED revisits or readmissions. They were able to demonstrate a drop from 18% to 11% in 90-day health care use associated with their meds in-hand interventions that focused on relabeling inpatient inhalers for home use, as well as filling the patient’s oral steroid, rescue inhaler, and controller medication prescriptions.37 Hatoun et al38 also incorporated an outpatient pharmacist delivery service to increase rate of medications in-hand at the time of discharge for patients with asthma. By using insurance claims data, they found that patients discharged with medications in-hand had lower odds of ED revisit within 30 days of discharge.38
Although this current review includes articles from 1991 to 2016, there are 2 studies published even more recently which further highlight the role of medications in-hand at discharge in asthma QI. A multimodal intervention by Krupp et al7 undertook the goal of reducing the 30-day inpatient asthma readmission rate by 50% within 2 years at a single center. Their interventions identified multiple specific targets, including improving access to follow-up care, bedside education, relabeling inpatient inhalers to be used post discharge, and home visits. Although the authors note that no single intervention was solely responsible for the decrease in readmissions, incorporating medications in-hand is a notable commonality with other successful projects. A November 2017 article by Kercsmar et al48 highlighted an integrated multimodal approach in a population of Medicaid-insured children, and again this team noted in-hand medications as a key driver of improvement. This team focused on a number of different interventions in different settings, including standardized history and physical for inpatients, standardized tools for identifying asthma risk factors, embedded decision supports to ensure appropriate referrals, in-hand medication at discharge from the hospital, and home visits. They were able to demonstrate a 41.8% relative reduction in asthma-related hospitalizations and 42.4% relative reduction in ED revisits after the index asthma hospitalization, and this improvement was sustained for 12 months. The study’s authors suggest that providing the supply of daily asthma medications at discharge was key to the observed improvements in their study.
This systematic review and meta-analysis has limitations. The studies included in this review included a range of interventions and outcomes and the linkages between the 2 are potentially confounded. We limited the studies in our meta-analysis to include only studies that intended to change reuse rates with their intervention(s). Nevertheless, many of the interventions included multiple components, so it is often not clear which singular component, if any, may or may not have been effective. Also, the follow-up intervals that we selected are subjective, but given the chronic nature of asthma, we felt it was important to capture reuse across a wider period. Next, some of the included studies are conducted in different countries with different health care models, although we excluded studies in countries with low or lower-middle incomes to improve generalizability to a US setting.
Current inpatient asthma QI literature is diverse. Although there was no overall effect on decreasing reuse across the included studies, we noted the importance of multimodal studies that could impact subsequent health care use. After evaluating multimodal interventions both qualitatively and quantitatively, we suggest that in-hand medications as a possible high-value strategy to improve care for children hospitalized with an asthma exacerbation.
- Accepted January 16, 2018.
- Address correspondence to Kavita Parikh, MD, MSHS, Division of Hospital Medicine, Department of Pediatrics, Children’s National Health System, George Washington School of Medicine and Health Sciences, 111 Michigan Ave Northwest, Washington, DC 20010. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: Dr Parikh was supported by grant K08HS024554 from the Agency for Healthcare Research and Quality. The content is solely the responsibility of the authors and does not necessarily represent the official views of the Agency for Healthcare Research and Quality.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2018-0420.
- Moorman JE,
- Akinbami LJ,
- Bailey CM, et al
- National Asthma Education and Prevention Program
- Standards for Quality Improvement Reporting Excellence
- World Development Indicators
- Ralston S,
- Comick A,
- Nichols E,
- Parker D,
- Lanter P
- Johnson KB,
- Blaisdell CJ,
- Walker A,
- Eggleston P
- Beck AF,
- Sauers HS,
- Kahn RS,
- Yau C,
- Weiser J,
- Simmons JM
- Kuhlmann S,
- Mason B,
- Ahlers-Schmidt CR
- Chong C,
- Samson J,
- Di Lucia A, et al
- Jassal MS,
- Sadreameli SC,
- Pereira I, et al
- Stevens CA,
- Wesseldine LJ,
- Couriel JM,
- Dyer AJ,
- Osman LM,
- Silverman M
- Sauers-Ford HS,
- Moore JL,
- Guiot AB, et al
- Hatoun J,
- Bair-Merritt M,
- Cabral H,
- Moses J
- Madge P,
- McColl J,
- Paton J
- Wesseldine LJ,
- McCarthy P,
- Silverman M
- Fassl BA,
- Nkoy FL,
- Stone BL, et al
- Beck AF,
- Simmons JM,
- Sauers HS, et al
- Copyright © 2018 by the American Academy of Pediatrics