Pathways guide clinicians through evidence-based care of specific conditions. Pathways have been demonstrated to improve inpatient asthma care but mainly in studies at large, tertiary children’s hospitals. It remains unclear if these effects are generalizable across diverse hospital settings. Our objective was to improve inpatient asthma care by implementing pathways in a diverse, national sample of hospitals.
We used a learning collaborative model. Pathway implementation strategies included local champions, external facilitators and/or mentors, educational seminars, quality improvement methods, and audit and feedback. Outcomes included length of stay (LOS) (primary), early administration of metered-dose inhalers, screening for secondhand tobacco exposure and referral to cessation resources, and 7-day hospital readmissions or emergency revisits (balancing). Hospitals reviewed a sample of up to 20 charts per month of children ages 2 to 17 years who were admitted with a primary diagnosis of asthma (12 months before and 15 months after implementation). Analyses were done by using multilevel regression models with an interrupted time series approach, adjusting for patient characteristics.
Eighty-five hospitals enrolled (40 children’s and 45 community); 68 (80%) completed the study (n = 12 013 admissions). Pathways were associated with increases in early administration of metered-dose inhalers (odds ratio: 1.18; 95% confidence interval [CI]: 1.14–1.22) and referral to smoking cessation resources (odds ratio: 1.93; 95% CI: 1.27–2.91) but no statistically significant changes in other outcomes, including LOS (rate ratio: 1.00; 95% CI: 0.96–1.06). Most hospitals (65%) improved in at least 1 outcome.
Pathways did not significantly impact LOS but did improve quality of asthma care for children in a diverse, national group of hospitals.
Pathways have been shown to improve quality of care for children who are hospitalized with asthma, but previous research has primarily consisted of single-center studies in large, tertiary children’s hospitals.
This multicenter study study of pediatric inpatient asthma pathways included hospitals that varied widely in size, type, location, and structure. We found that pathways were not associated with statistically significant changes in length of stay but were associated with improvements in care quality.
Childhood asthma is a leading cause of hospitalization, with total estimated direct costs of ∼$1.6 billion annually in the United States.1,2 Evidence-based guidelines for asthma management are widely available.3,4 However, clinicians face many challenges in adhering to guidelines.5 This poor adherence contributes to suboptimal health outcomes for children who are hospitalized with asthma, including longer recovery time and length of stay (LOS), higher risk of transfer to intensive care, and higher risk of hospital readmission.6–8
Pathways are a potential tool for improving guideline adherence and health outcomes. They are succinct versions of evidence-based guidelines that visually guide clinicians step by step through the timing, indications, and details of recommended tests and treatments for managing a specific condition. Pathways have been shown to improve quality of care for children who are hospitalized with asthma by increasing the use of recommended medications such as bronchodilators and corticosteroids, reducing use of antibiotics, decreasing LOS and associated health care costs, and improving health care disparities.9–14 This previous research has primarily consisted of single-center studies in large, tertiary children’s hospitals9 ; however, <30% of childhood hospitalizations in the United States occur in these settings.15 Thus, the purpose in our initiative was to improve inpatient asthma care for children by implementing pathways in a diverse, national sample of hospitals.
Methods
Setting and Population
This study was done by the Value in Inpatient Pediatrics network, the inpatient pediatric quality improvement (QI) network of the American Academy of Pediatrics (AAP).16 Recruitment occurred via the Value in Inpatient Pediatrics electronic mailing list (Listserv). The Listserv includes clinicians from >250 hospitals around the country that range widely in size, type (eg, children’s or community), ownership model (eg, private or nonprofit), and location (eg, rural or urban). The study took place from January 2018 to March 2019 and was approved by the AAP’s institutional review board.
Study Planning and Aims
Our global aim was “to improve the value of hospital care for children with asthma.” We used the positive deviance theory in planning the intervention. This theory proposes that examination of higher- and lower-performing hospitals can facilitate the discovery and wide dissemination of strategies to improve care.17 In a previous analysis, hospitals with higher and lower performance in pediatric asthma pathway implementation were identified.11 These hospitals were then studied to identify strategies to improve pathway implementation and asthma care: (1) using QI methodology, (2) getting teams to commit to shared goals, (3) integrating pathway content into the electronic health record (EHR), (4) developing standardized instructions and/or protocols for nurses and/or respiratory therapists to titrate bronchodilators, and (5) engaging with hospital leaders.18 A panel of experts assembled by the AAP met regularly throughout 2017 to finalize study aims (Table 1), specific interventions (Table 2, column 1), implementation strategies (incorporating these previous findings above18 ), and study design. Members of this panel worked in diverse hospital settings and had expertise in pediatric hospital medicine, pediatric emergency medicine, asthma, quality measurement, QI, and health services research. Additionally, 2 nurses and 1 respiratory therapist reviewed and provided feedback on the intervention.
Global and Specific Aims
| Baseline Performance | Aim or Target | |
|---|---|---|
| Global aim: to improve the value of hospital care for children with asthma | ||
| Specific aims | ||
| Decrease inpatient hospital LOS by 10%, h | 36 | 32 |
| Increase early transition to administering bronchodilator via MDI by 50%, % | 43 | 65 |
| Achieve 90% compliance with screening for secondhand smoke exposure, % | 78 | 90 |
| Achieve a 50% increase in documentation of caregiver referral to smoking cessation resources for eligible patients, % | 28 | 42 |
| Baseline Performance | Aim or Target | |
|---|---|---|
| Global aim: to improve the value of hospital care for children with asthma | ||
| Specific aims | ||
| Decrease inpatient hospital LOS by 10%, h | 36 | 32 |
| Increase early transition to administering bronchodilator via MDI by 50%, % | 43 | 65 |
| Achieve 90% compliance with screening for secondhand smoke exposure, % | 78 | 90 |
| Achieve a 50% increase in documentation of caregiver referral to smoking cessation resources for eligible patients, % | 28 | 42 |
Effects of Specific Interventions (N = 68 Hospitals)
| Specific Intervention | Hospitals That Reported Implementing the Intervention, n (%) | Effect Outcome, Ratio (95% CI) |
|---|---|---|
| Guidance on dosing of bronchodilators via MDI | 60 (88) | Early administration of MDI: OR 4.11 (2.16–7.84) |
| Pathway and/or instructions for standardized bronchodilator titration by nurses or respiratory therapists | 51 (75) | LOS: RR 0.95 (0.83–1.09) |
| Hospital discharge criteria | 57 (84) | LOS: RR 1.02 (0.93–1.13) |
| Reminders to screen for secondhand tobacco exposure | 63 (93) | Screening for secondhand tobacco: OR 1.12 (0.74–1.7) |
| Reminders to refer caretakers who screen positive for tobacco use to smoking cessation resources | 62 (91) | Smoking cessation resource referral: OR 3.04 (1.64–5.64) |
| Specific Intervention | Hospitals That Reported Implementing the Intervention, n (%) | Effect Outcome, Ratio (95% CI) |
|---|---|---|
| Guidance on dosing of bronchodilators via MDI | 60 (88) | Early administration of MDI: OR 4.11 (2.16–7.84) |
| Pathway and/or instructions for standardized bronchodilator titration by nurses or respiratory therapists | 51 (75) | LOS: RR 0.95 (0.83–1.09) |
| Hospital discharge criteria | 57 (84) | LOS: RR 1.02 (0.93–1.13) |
| Reminders to screen for secondhand tobacco exposure | 63 (93) | Screening for secondhand tobacco: OR 1.12 (0.74–1.7) |
| Reminders to refer caretakers who screen positive for tobacco use to smoking cessation resources | 62 (91) | Smoking cessation resource referral: OR 3.04 (1.64–5.64) |
Local implementation leaders were surveyed to collect these data, and 68 of 68 responded (100%). Implementation rates were summarized by using descriptive statistics. Implementation effects were analyzed by using multilevel models with an ITS approach. OR, odds ratio; RR, rate ratio.
Intervention
Participating hospitals were provided with sample evidence-based pathways and order sets based on pathway content. Specific interventions (Table 2, column 1) included guidance on selection and dosing of medications (eg, metered-dose inhaler [MDI]), standardized instructions for titrating bronchodilator therapy on the basis of asthma severity, reminders of evidence-based practices (eg, screening for secondhand tobacco exposure), and discharge criteria. Each hospital designated a local physician implementation leader. Local leaders recruited local multidisciplinary teams, including nurses, respiratory therapists, pharmacists, and/or administrators. Local teams adapted and implemented pathways to fit local needs and context. To support pathway implementation, we used a learning collaborative model.19 Hospital teams were provided with several resources for implementation support, including educational seminars, QI training, monthly video conferences for all local implementation leaders to facilitate peer learning, in-person meetings during national pediatric conferences, monthly audit and feedback (on local and national performance), and physician mentors with QI expertise to guide them in the learning collaborative.
These mentors were selected by the expert panel on the basis of experience in pediatric hospital medicine, QI (including experience with national learning collaboratives), and community hospitals. They were trained via 3 teleconference sessions, with topics including evidence-based asthma care, study details (design, time line, and data collection methods), and application of QI principles to asthma pathway implementation. Mentors were matched to sites on the basis of geographic proximity to facilitate better understanding of local health regulations and/or policies and meeting scheduling. They met with local implementation leaders monthly for 2 months before and 4 months after pathway implementation to provide support in engaging hospital leadership, garnering local clinician buy-in, planning improvement cycles,19 and addressing institutional barriers. Mentors met as a group with expert panel members quarterly via teleconference to share lessons learned and assist each other in QI planning.
Outcomes
Study outcomes (Table 3) were selected through a consensus process among the national expert panel. The panel selected study outcomes and/or measures on the basis of (1) recommendations of evidence-based guidelines,3,4 (2) potential to improve health outcomes, (3) variability and/or room for improvements in performance, and (4) feasibility of measurement. Hospital LOS12 was selected as the primary outcome. We also selected measures of adherence to evidence-based guidelines,3,4 including screening for secondhand tobacco exposure,20 referral of caregivers to smoking cessation resources,20 and early administration of MDIs. Early administration of MDI was selected because MDIs are more cost-effective and have fewer side effects than nebulizers,3,21 and MDI use promotes asthma education and better chronic-asthma control.22–24 Balancing measures included 7-day hospital readmissions and 7-day emergency revisits.25,26
Study Outcomes and Definitions
| Outcome | Definition | Goal |
|---|---|---|
| Inpatient LOS (primary) | LOS (h) | ↓ |
| Early administration of bronchodilator via MDI | MDI: (1) ordered at admission, (2) first ordered at 1-h frequency, or (3) first ordered at 2-h frequency | ↑ |
| Screening for secondhand tobacco smoke exposure | Documented screening for secondhand tobacco smoke exposure | ↑ |
| Referral of caregivers to smoking cessation resources | Documented referral to cessation resources for caretakers who screen positive for tobacco use | ↑ |
| Hospital readmission within 7 d of discharge (balancing) | Hospital readmission for any cause within 7 d of hospital discharge | — |
| ED revisit within 7 d of discharge (balancing) | ED revisit for any cause within 7 d of hospital discharge | — |
| Outcome | Definition | Goal |
|---|---|---|
| Inpatient LOS (primary) | LOS (h) | ↓ |
| Early administration of bronchodilator via MDI | MDI: (1) ordered at admission, (2) first ordered at 1-h frequency, or (3) first ordered at 2-h frequency | ↑ |
| Screening for secondhand tobacco smoke exposure | Documented screening for secondhand tobacco smoke exposure | ↑ |
| Referral of caregivers to smoking cessation resources | Documented referral to cessation resources for caretakers who screen positive for tobacco use | ↑ |
| Hospital readmission within 7 d of discharge (balancing) | Hospital readmission for any cause within 7 d of hospital discharge | — |
| ED revisit within 7 d of discharge (balancing) | ED revisit for any cause within 7 d of hospital discharge | — |
—, no change.
Data Collection
Data on hospital characteristics and the specific interventions implemented were collected via electronic survey of implementation leaders.20,27 Hospital characteristics (eg, type) were verified by using data from the American Hospital Association Annual Survey Database.28 Data on outcomes (Table 3) were collected via chart review of children ages 2 to 17 years who were hospitalized with a primary diagnosis of asthma. Chart review was performed at each site. Reviewers were provided with chart review manuals, trained via video conference by the central research team (eg, definitions of variables and where to find them in the EHR), and encouraged to ask this team questions at any time during chart review. Charts from January 2017 to December 2017 were reviewed retrospectively; charts from January 2018 to March 2019 were reviewed prospectively (during and after pathway implementation). Charts were selected in chronological order each month (all admissions up to a maximum of 20 per month per hospital). During manual chart review, charts were excluded if children had chronic medical conditions that precluded pathway use (eg, cystic fibrosis, restrictive lung disease, bronchopulmonary dysplasia, congenital or acquired heart disease, airway issues, immune disorders, sickle cell anemia, or neuromuscular disorders). Data quality was audited monthly by the central research team and by quality audit functions within Research Electronic Data Capture 8.5 (Nashville, TN).
Analyses
Hospital and patient characteristics were summarized by using descriptive statistics. Hospital characteristics for those who dropped out versus completed the study were compared by using χ2 tests for categorical variables and Mann-Whitney U tests for interval variables. Patient characteristics for those visits in the before versus after pathway period were compared by using logistic regression.
For our primary analysis, we used multilevel regression models with an interrupted time series (ITS) approach (levels: hospital and admission, random effects). ITS accounts for secular trends and evaluates (1) changes in the outcome at the time of implementation and (2) changes in the rate of change in an outcome after versus before implementation.29 Models were adjusted for patient characteristics, including age, sex, insurance type (a proxy for socioeconomic status), and previous prescription of inhaled corticosteroids (a proxy for chronic-asthma severity). Models were also adjusted for hospital type (eg, children’s or community). LOS was modeled by using γ regression, and a quadratic term was included to correct for seasonal trends. All other outcomes were modeled by using logistic regression. Data from the first 2 months of the intervention period were washed out to allow for partial pathway implementation. All data from all hospitals were analyzed regardless of whether the hospital completed the study. Random hospital intervention effects were added to the model to calculate hospital-level changes in outcomes.
Secondary Analysis: Hawthorne Effect
We conducted a secondary analysis to determine if pathway implementation was associated with improvements in quality of care compared with project enrollment alone (Hawthorne effect).30 Half of the participating hospitals started pathway implementation in January 2018, and half of the participating hospitals were in a control or waiting period until April 2018. We conducted a difference-in-differences analysis that compared projected values on the basis of outcomes from January 2017 to December 2017 to outcomes from January 2018 to March 2018 in the pathway implementation (intervention) versus waiting period (control) hospitals. This analysis was done by using multilevel regression models. Models included the same patient and hospital characteristics described above for the primary analysis.
Sensitivity Analysis: Effects of Specific Interventions
Implementation leaders from each hospital reported if they had implemented specific interventions (eg, MDI dosing guidance). We conducted analyses to determine if there were associations between leader-reported implementation of these specific interventions (at the hospital level) and study outcomes. These analyses were done by using multilevel regression models with an ITS approach. Models included the same patient and hospital characteristics described above for the primary analysis.
Sensitivity Analysis: Effects of Placing a Pathway Order
Developing electronic orders for standardized titration of bronchodilators has been described as a key element of pathway implementation.18,31 During chart review, we collected information on whether individual patients had a pathway order placed (for standardized bronchodilator titration). We conducted an analysis to determine if there was an association between electronic pathway orders being present in the patient’s chart and LOS. This analysis was done by using a multilevel γ regression model with an ITS approach. The model included the same patient and hospital characteristics described above for the primary analysis.
Of note, the sensitivity analysis of specific interventions (described above) determined associations between leader-reported implementation of a pathway and/or protocol for standardized bronchodilator titration and LOS. This sensitivity analysis determined associations between electronic pathway orders being present in the patient’s chart and LOS. So, this sensitivity analysis measures the effects of more successful, higher-fidelity implementation.
All analyses were performed with SAS 9.4 (SAS Institute, Inc, Cary, NC). P <0.05 was considered statistically significant.
Results
Hospitals and Study Population
A total of 85 hospitals enrolled, and 68 hospitals (80%) completed the study (participated in the learning collaborative and associated outcome monitoring for the full 15-month study duration). Hospitals that joined were diverse in terms of hospital size, type, and location (Table 4). Seventy-eight hospitals (92%) were teaching hospitals. Hospitals that dropped out (n = 17) did not statistically significantly differ from those that completed the study in any of the characteristics described in Table 4.
Hospitals in the Pathways for Improving Inpatient Pediatric Asthma Care Study
| Characteristic | Hospitals (n = 85) |
|---|---|
| Hospital type, n (%) | |
| Children’s | 40 (47) |
| Community | 45 (53) |
| Teaching hospital, n (%) | 78 (92) |
| Inpatient pediatric beds, mean (SD) | 39 (51) |
| Hospital size, total beds, n (%) | |
| Small (<100) | 6 (7) |
| Medium (100–249) | 26 (31) |
| Large (≥250) | 53 (63) |
| Geographic region, n (%) | |
| West | 18 (21) |
| South | 26 (31) |
| Northeast | 14 (16) |
| Midwest | 27 (32) |
| Location, n (%) | |
| Urban | 39 (46) |
| Suburban | 38 (45) |
| Rural | 8 (9) |
| Presence of pediatric ICU, n (%) | 52 (61) |
| Ownership model, n (%) | |
| Government | 9 (11) |
| Private, nonprofit | 68 (82) |
| Private, investor owned | 6 (7) |
| Presence of EHR, n (%) | 83 (98) |
| Total physicians on staff for inpatient pediatric unit, mean (SD) | 7.6 (6.4) |
| Characteristic | Hospitals (n = 85) |
|---|---|
| Hospital type, n (%) | |
| Children’s | 40 (47) |
| Community | 45 (53) |
| Teaching hospital, n (%) | 78 (92) |
| Inpatient pediatric beds, mean (SD) | 39 (51) |
| Hospital size, total beds, n (%) | |
| Small (<100) | 6 (7) |
| Medium (100–249) | 26 (31) |
| Large (≥250) | 53 (63) |
| Geographic region, n (%) | |
| West | 18 (21) |
| South | 26 (31) |
| Northeast | 14 (16) |
| Midwest | 27 (32) |
| Location, n (%) | |
| Urban | 39 (46) |
| Suburban | 38 (45) |
| Rural | 8 (9) |
| Presence of pediatric ICU, n (%) | 52 (61) |
| Ownership model, n (%) | |
| Government | 9 (11) |
| Private, nonprofit | 68 (82) |
| Private, investor owned | 6 (7) |
| Presence of EHR, n (%) | 83 (98) |
| Total physicians on staff for inpatient pediatric unit, mean (SD) | 7.6 (6.4) |
Characteristics of children admitted for asthma before and after pathway implementation are presented in Table 5 (n = 12 013). Children who were admitted before and after pathway implementation were clinically similar, and all characteristics listed in Table 5 were included in our multivariable regression models.
Characteristics of Children Admitted Before and After Pathway Implementation
| Characteristic | All Admissions (n = 12 013) | Before Pathway (n = 6506) | After Pathway (n = 5507) |
|---|---|---|---|
| Age, y, median (IQR) | 6 (3–9) | 6 (3–9) | 6 (3–9) |
| Male sex, n (%) | 7216 (60) | 3887 (60) | 3329 (60) |
| Previous prescription of inhaled corticosteroid, n (%) | 6163 (51) | 3426 (53) | 2737 (50) |
| Insurance type, n (%) | |||
| Public | 7574 (63) | 4077 (63) | 3497 (64) |
| Private | 3545 (30) | 1979 (30) | 1566 (28) |
| Tricare | 126 (1) | 64 (1) | 62 (1) |
| Other, self-pay, or unknown | 460 (4) | 210 (3) | 250 (5) |
| Missing | 308 (3) | 176 (3) | 132 (2) |
| Characteristic | All Admissions (n = 12 013) | Before Pathway (n = 6506) | After Pathway (n = 5507) |
|---|---|---|---|
| Age, y, median (IQR) | 6 (3–9) | 6 (3–9) | 6 (3–9) |
| Male sex, n (%) | 7216 (60) | 3887 (60) | 3329 (60) |
| Previous prescription of inhaled corticosteroid, n (%) | 6163 (51) | 3426 (53) | 2737 (50) |
| Insurance type, n (%) | |||
| Public | 7574 (63) | 4077 (63) | 3497 (64) |
| Private | 3545 (30) | 1979 (30) | 1566 (28) |
| Tricare | 126 (1) | 64 (1) | 62 (1) |
| Other, self-pay, or unknown | 460 (4) | 210 (3) | 250 (5) |
| Missing | 308 (3) | 176 (3) | 132 (2) |
Effects of Inpatient Pediatric Asthma Pathways: Aggregate Analysis of all Hospitals
Baseline rates and study targets are specified in Table 1. All study outcomes are summarized in Fig 1. Pathway implementation was tailored at each participating hospital. In aggregate, pathway implementation was not associated with statistically significant changes in LOS (mean 36 hours before implementation and 33 hours after; rate ratio: 1.00; 95% confidence interval [CI]: 0.94–1.06). However, pathway implementation was associated with statistically significant increases in odds of early administration of bronchodilators via MDI (projected versus actual rates: 49% vs 74%, respectively) and caretaker referral to smoking cessation resources (projected versus actual rates: 27% vs 65%, respectively). Pathway implementation was not associated with statistically significant changes in secondhand tobacco exposure screening or in 7-day hospital readmissions or ED revisits (balancing measures).
Effects of inpatient pediatric asthma pathways: aggregate analysis (all hospitals). ITS models analyzed changes in (1) the outcome at the time of implementation and (2) the rate of change in an outcome after versus before implementation. Purple, solid lines represent the trends before pathway implementation; purple, dashed lines represent projected future trends (if no pathway implementation occurred); blue, solid lines represent actual trends observed after pathway implementation; and gray lines represent 95% confidence limits. a Statistically significant. RR, rate ratio.
Effects of inpatient pediatric asthma pathways: aggregate analysis (all hospitals). ITS models analyzed changes in (1) the outcome at the time of implementation and (2) the rate of change in an outcome after versus before implementation. Purple, solid lines represent the trends before pathway implementation; purple, dashed lines represent projected future trends (if no pathway implementation occurred); blue, solid lines represent actual trends observed after pathway implementation; and gray lines represent 95% confidence limits. a Statistically significant. RR, rate ratio.
Effects of Inpatient Pediatric Asthma Pathways: Hospital-Level Analysis
Figure 2 details all study outcomes at the hospital level. The majority (44 hospitals, 65% of those completing the study) had significant improvements in at least 1 outcome, and 23 hospitals (34% of those completing the study) had significant improvements in more than 1 outcome. The most common outcomes improved were caretaker referral to smoking cessation resources (28 hospitals) and early bronchodilator administration via MDI (27 hospitals). Two hospitals (3% of those completing the study) had significant increases in 7-day hospital readmissions or emergency department (ED) revisits.
Effects of inpatient pediatric asthma pathways: hospital-level analysis. Open circles represent hospitals with an annual volume of ≥50 patients with asthma, and closed circles represent hospitals with an annual volume of <50 patients with asthma. Each circle is positioned on the basis of baseline performance and degree of improvement during the intervention period. Green, shaded areas indicate improvements in outcomes.
Effects of inpatient pediatric asthma pathways: hospital-level analysis. Open circles represent hospitals with an annual volume of ≥50 patients with asthma, and closed circles represent hospitals with an annual volume of <50 patients with asthma. Each circle is positioned on the basis of baseline performance and degree of improvement during the intervention period. Green, shaded areas indicate improvements in outcomes.
Effects of Pathway Implementation Versus Project Enrollment Only (Hawthorne Effect)
We compared changes in study outcomes from the baseline period to the period from January 2018 to March 2018 in hospitals that implemented pathways (intervention) during these months versus those that waited (control). This analysis included a total of 6837 admissions. We found that pathway implementation was associated with statistically significant increases in odds of early administration of bronchodilator via MDI (odds ratio [OR] 3.94; 95% CI: 2.85–5.44) and caretaker referral to smoking cessation resources (OR: 2.22; 95% CI: 1.38–3.57), which is in line with our primary analysis. Thus, improvements in these outcomes were associated with pathway implementation rather than project enrollment only (Hawthorne effect). None of the other study outcomes, including the primary outcome of LOS, were significantly associated with pathway implementation compared with project enrollment only.
Effects of Specific Interventions
Local leaders self-reported the implementation of specific interventions, and the proportion of hospitals that implemented these interventions varied from 75% to 93% (Table 2). Implementation of MDI dosing guidance was associated with improvements in early administration of MDI (OR: 4.11; 95% CI: 2.16–7.84), and implementation of reminders was associated with improvements in referral to smoking cessation resources (OR: 3.04; 95% CI: 1.64–5.64).
Effects of Placing a Pathway Order
Of all the children cared for in the period after pathway implementation, 63% had a pathway order placed. Pathway order placement was associated with an 8% decrease in LOS (rate ratio 0.92; 95% CI: 0.89–0.96).
Discussion
This multisite study of inpatient pediatric asthma pathways is the first (to our knowledge) to use a sample of hospitals that varied widely in size, type, location, and structure. Pathways were tailored and implemented at participating hospitals, and overall, pathway implementation was not associated with statistically significant changes in LOS. However, pathway implementation was associated with improvements in quality of care, including increased odds of early administration of bronchodilator via MDI and referral of caretakers to smoking cessation resources. We found no concerning effects on our balancing measures, 7-day hospital readmissions, or emergency revisits. These improvements were significant in our aggregate analysis of all participating hospitals, and we also found that 65% of the hospitals in this diverse sample had statistically significant improvements in at least 1 outcome and/or quality measure. Thus, pathways’ effects on quality of care were generalizable across this diverse sample of hospitals.
Previous studies of inpatient pediatric asthma pathways have taken place in single hospitals or within a single hospital network (Intermountain Healthcare, a not-for-profit network, provides hospital and other medical services in UT and ID and offers integrated managed care).9,12 Our study involved a diverse sample of hospitals from around the country that were not financially or structurally linked, but all hospitals had a physician who was a member of the AAP’s Value in Inpatient Pediatrics network. Our findings align with previous literature in demonstrating that pathways can improve guideline adherence, specifically the administration of recommended medications (eg, bronchodilators and corticosteroids).11,12,32
Our aggregate analysis of all participating hospitals demonstrated that pathway implementation was associated with improvements in care, and these improvements were even larger in hospitals that reported successful implementation of key specific interventions. In our aggregate analysis of all participating hospitals, pathway implementation was associated with increases in early MDI administration (OR: 1.18; 95% CI: 1.14–1.22), and these effects were larger in hospitals that reported successful implementation of guidance on MDI dosing (OR: 4.11; 95% CI: 2.16–7.84). Similarly, in our aggregate analysis of all participating hospitals, pathway implementation was associated with increases in referral of caretakers to smoking cessation resources (OR: 1.93; 95% CI: 1.27–2.91), but these effects were larger in hospitals that reported successful implementation of reminders to refer (OR: 3.04; 95% CI: 1.64–5.64). Reported implementation of protocols for standardized bronchodilator titration was not associated with LOS. However, when we examined a marker of even higher fidelity and/or better implementation (pathway orders for standardized titration placed in a child’s chart), we found that LOS was reduced by 8%. This finding aligns with and reinforces previous studies in demonstrating the importance of developing electronic orders and/or order sets that facilitate standardized bronchodilator titration.18,31–36 The learning collaborative provided supports to promote successful implementation of these specific interventions, including evidence-based interventions (pathways), opportunities for learning from and comparing performance with peer hospitals, and QI mentors (who can facilitate the identification of implementation barriers and problem-solving to address those barriers).37,38 However, some hospitals in our study were unable to implement specific interventions (Table 2) despite these supports. Important barriers to implementation may include a lack of leadership support and resources, including resources for timely modification of EHRs.18,37,39
Many proposed quality measures for inpatient asthma care have been abandoned because of a lack of association with health outcomes (eg, provision of asthma action plans) or a lack of variability in performance (eg, administration of systemic corticosteroids).40,41 This study used a novel quality measure of inpatient pediatric asthma care: early administration of bronchodilator via MDI. This was defined as an MDI being ordered immediately on hospital admission or at a 1- or 2-hour frequency any time during hospitalization. MDI use has been advocated by international evidence-based guidelines because it has fewer side effects and is more cost-effective than nebulizers.3,21 Also, earlier and more frequent use of MDI in the inpatient setting promotes asthma self-management education and training on proper use of masks and spacers, thereby improving chronic-asthma control.22–24 This study showed a 25% increase in the proportion of children being administered an MDI early in the hospital stay. To enable improvements in this outcome, we provided pathways that specified MDI dosing. Our study demonstrates that (1) there is wide variability and/or room for improvement in early use of MDI (Fig 2), (2) pathways can improve early use of MDI (Fig 1), and (3) measuring early use of MDI is feasible. We propose that policy makers, hospital leaders, and clinicians consider using this measure in future efforts to improve quality of care for children with asthma.
Previous analyses of QI interventions have been critiqued for using pre-post study designs42 because these designs are weaker in distinguishing whether improvements in care were driven by the specific intervention studied versus (1) beginning to observe and measure provider behavior (Hawthorne effect) or (2) other concomitant factors influencing care (secular trends). To address these concerns, we (1) performed a secondary analysis of potential Hawthorne effect and (2) used an ITS approach. Half of the hospitals participating in this study began implementing pathways in January 2018, whereas the other half of the hospitals collected prospective data on study outcomes without implementing pathways until April 2018 (control period). During this period, we found that pathway implementation was associated with significant improvements in outcomes compared with project enrollment alone (Hawthorne effect). We used multivariable regression models to address potential confounding bias from differences in patient characteristics before versus after pathway implementation and an ITS approach to account for any changes in care driven by secular trends.43,44 ITS has been demonstrated to perform similarly to cluster-randomized trials to account for the potential influence of secular trends.45,46 We encourage QI leaders and researchers to consider these approaches in future studies of interventions to improve quality of care.
This study had several limitations. Although we used a control group and ITS approach, our observational study findings may still be affected by the Hawthorne effect, secular trends, or other potential confounders. Also, we had limited ability to demonstrate improvements in several outcome measures due to high baseline performance among the participating hospitals. These included LOS (baseline 36 hours) and secondhand tobacco exposure screening (baseline 78%). Furthermore, recruitment for this study was done via an electronic Listserv of physician members the AAP’s Value in Inpatient Pediatrics network16 ; these members have a strong interest in finding opportunities to lead local pediatric QI efforts. Thus, we may have selected hospitals with physicians who are more open to recruitment via electronic mail and/or uniquely motivated by pediatric QI. Our findings may only be most directly applicable to similar settings, but the wide-ranging diversity in the characteristics of participating of hospitals likely promotes broader generalizability. Lastly, there was variation in implementation fidelity (eg, implementation of specific interventions), which may have decreased the effect sizes we observed in comparison with previous single-center and single–health system studies. However, our sensitivity analyses, which focused on hospitals with more successful implementation, showed effect sizes similar to previous literature.
Conclusions
In this national, multicenter study of inpatient pediatric asthma pathways, pathways were not associated with statistically significant changes in LOS but were associated with improvements in quality of care. Leaders of QI efforts should consider pathway implementation and promotion of early MDI use in their efforts to improve quality of care for children who are hospitalized with asthma.
Acknowledgments
We acknowledge the AAP staff and all the local Pathways for Improving Inpatient Pediatric Asthma Care implementation leaders who made this study possible.
Dr Kaiser conceptualized and designed the study, designed the data collection instruments, collected data, and drafted the initial manuscript; Mr Rodean conceptualized and designed the study, designed the data collection instruments, and conducted the initial analyses; Ms Jennings coordinated and supervised data collection; Drs Cabana, Garber, Ralston, Fassl, Quinonez, Mendoza, McCulloch, and Parikh conceptualized and designed the study and designed the data collection instruments; and all authors interpreted the data, reviewed and revised the manuscript, approved the final manuscript as submitted, and agree to be accountable for all aspects of the work.
FUNDING: Drs Kaiser and Parikh are supported by career development grants from the Agency for Healthcare Research and Quality (K08HS024592 and K08HS024554), but this agency played no role in the design, data collection, analysis, or reporting of this study.
