CONTEXT: Successful care transitions between emergency departments (EDs) and outpatient settings have implications for quality, safety, and cost of care.
OBJECTIVE: To investigate the effectiveness of ED-based care transition interventions in achieving outpatient follow-up among pediatric patients.
DATA SOURCES: Medline, Embase, CINAHL, Cochrane Library, trial registers, and reference lists of relevant articles.
STUDY SELECTION AND DATA EXTRACTION: Eligible studies included randomized controlled trials of ED-based care transition interventions involving pediatric patients (aged ≤18 years). Study selection, data extraction, and risk of bias assessment were performed in duplicate and independent manner. Study results were pooled for meta-analysis by using a random effects model.
RESULTS: Sixteen randomized controlled trials, comprising 3760 patients, were included in the study. Most interventions were single-site (n = 14), multicomponent (n = 12), and focused on patients with asthma (n = 8). Pooling data from 10 studies (n = 1965 patients) found moderate-quality evidence for a relative increase of 29% in outpatient follow-up with interventions compared with routine care (odds ratio, 1.58 [95% confidence interval, 1.08–2.31]). Successful interventions included structured telephone reminders, educational counseling on follow-up, and appointment scheduling assistance. There was low-quality evidence when pooling data from 5 studies (n = 1609 participants) that exhibited little or no beneficial effect of interventions on ED readmissions (risk ratio, 1.02 [95% confidence interval, 0.91–1.15]).
LIMITATIONS: All studies were conducted in urban US hospitals which makes generalization of the results to rural settings and other countries difficult.
CONCLUSIONS: ED-based care transition interventions are effective in increasing follow-up but do not seem to reduce ED readmissions. Further research is required to investigate the mechanisms that affect the success of these interventions.
- CI —
- confidence interval
- ED —
- emergency department
- GRADE —
- Grading of Recommendations Assessment, Development and Evaluation
- OR —
- odds ratio
- PCP —
- primary care physician
- RCT —
- randomized controlled trial
- RR —
- risk ratio
Emergency departments (EDs) care for ∼25 million children every year.1 ED discharges without appropriate linkages to outpatient settings have been associated with increased health care utilization costs2 and adverse events.3,4 With the goal of minimizing such costs and streamlining the process of care transitions of patients from the ED to outpatient settings, federal agencies have implemented several models to improve care coordination effectiveness.5 In addition, the Patient Protection and Affordable Care Act has provided opportunities (as well as financial incentives and penalties) to promote care coordination through accountable care organizations.6 The common theme underpinning the varying perspectives/models of care coordination is predicated on care professionals working collaboratively to meet patient care needs for ensuring the right care at the right time.7
Transitions of care are a key aspect of care coordination, and the role of the ED in facilitating such efforts needs to be determined. As highlighted by Katz et al,8 it has to be established whether the ED plays a central (ie, ED-centric care coordination) or a peripheral (ie, care coordination programs managed by longitudinal care workers outside the ED) role. Formulating the role of the ED within the care coordination model can only be ascertained through a comprehensive understanding of the impact of ED-based interventions.
The present article offers a systematic review and meta-analysis regarding the effectiveness of ED-based interventions in supporting care transitions to outpatient settings for pediatric patients. Such synthesized evidence can provide insights on the nature and characteristics of effective ED-based interventions for outpatient follow-up encounters and potentially fewer ED readmissions among pediatric patients.
Previous systematic reviews on the effectiveness of ED-based interventions for care transitions have been limited. Schatz et al9 investigated ED-based interventions that promoted outpatient follow-up rates for patients with asthma exacerbations; similarly, Katz et al8 reviewed evidence on the effectiveness of ED-based interventions for improving care coordination in adult and pediatric patients. Our review differs in scope and purpose: first, this review is the first systematic analysis to incorporate a quantitative synthesis of results on the effectiveness of care transition interventions for outpatient follow-up rates and ED readmissions. Second, this review serves as an update to the last review on care coordination and transitional interventions conducted in 2010.8 Finally, it specifically focuses on the pediatric patient population, whose care coordination and management needs differ from those of adults.10
A systematic search of Medline, Embase, CINAHL, and the Cochrane Central Register of Controlled Trials was undertaken on February 16, 2015, to identify relevant studies for inclusion. The search strategy combined controlled vocabulary (eg, Medical Subject Heading terms) and free text terms with Cochrane clinical queries for randomized controlled trials (RCTs) (Appendix 1). No language or publication type restrictions were applied. Searches were supplemented by hand-searching reference lists of relevant studies and the clinicaltrials.gov database, contacting experts in the field, and reviewing references of previously published systematic reviews of ED-based care transition interventions.8,11
Study Screening and Selection
Two reviewers (J.A. and T.K.) independently examined the title and abstract of all retrieved references. Citations that did not meet the inclusion criteria were excluded. The full-text article of references that seemed to be relevant was retrieved and independently assessed by 3 reviewers (J.A., T.K., and S.K.) based on the inclusion and exclusion criteria. Disagreements were resolved through group discussion.
Eligible studies included RCTs of ED-based care transition interventions (ie, interventions that were initiated in the ED for promoting care transitions to outpatient settings) involving pediatric patients (≤18 years of age) of either sex, any ethnic group, or clinical condition. Studies were excluded that included adult (or a combination of adult and pediatric) patients, were not ED-based, and were not aimed at promoting care transitions. The main outcomes of interest were outpatient follow-up rates (either primary care physician [PCP] or specific specialty physician), and ED readmissions.
Data Abstraction and Management
Two reviewers (J.A. and T.K.) independently extracted data pertaining to the population, intervention, comparison group, outcomes, and clinical setting characteristics, and recorded them by using a standard data extraction form (additional details regarding study design, and clinical setting were also recorded). Differences were resolved through discussion.
Assessment of Risk of Bias in Included Studies
Two reviewers (J.A. and S.K.) independently assessed risk of bias for each included study using the Cochrane Collaboration criteria and recommendations described in the Cochrane Handbook for Systematic Reviews of Interventions.12 The following risk of bias domains were assessed: sequence generation (selection bias); allocation concealment (selection bias); blinding of participants and personnel (performance bias); blinding of outcome assessment (detection bias); whether incompleteness of outcome data were adequately addressed (attrition bias); and whether there was selective outcome reporting (reporting bias). Disagreements were resolved through discussion.
Data Analysis and Synthesis
Raw data extracted from the included studies were used for ascertaining the sample size, the characteristics, and the effects of the interventions. A continuity of care framework was used to characterize the purpose of the reported care transition interventions. The continuity of care framework considered 3 functions for ensuring effective care transitions (adapted from Hesselink et al11): (1) information, focused on the quality of information shared between providers or between providers and patients who potentially improved the completeness, accuracy, or clarity of instructions or education (eg, templates, structured discharge instructions); (2) communication, focused on transmission of messages between patients and providers or between providers with an intended purpose (eg, text, e-mail, telephonic reminders); and (3) coordination of care, focused on explicit management of interdependencies between ED and outpatient settings, planning and meeting patient transitional needs to avoid fragmentation of care (eg, discharge planning, early assessment of needs, referrals, scheduling appointments, care provider notifications, case management). Two researchers (J.A. and T.K.) coded the continuity of care framework with 97% agreement. Disagreements were resolved through discussion.
For investigating the effect of the interventions, we computed the odds ratios (ORs) for outpatient follow-up rates and risk ratios (RRs) for ED readmissions. All analyses were performed by using Review Manager 5.13 In case of missing data in the included studies, study authors were contacted via e-mail. We also followed up with the study authors, if no response was received for the initial e-mails. For studies that used both self-reported outcome data (eg, based on recall) and data corroborated by using medical record audits, the latter was used for analysis. In addition, if adjusted and unadjusted estimates were reported, the unadjusted estimate was used. For studies that used multiple intervention groups (with a single control), we selected one pair of interventions (the intervention most relevant to our research question) excluding the other, as suggested in the Cochrane Handbook for Systematic Reviews of Interventions.14
Forest plots were created to visually inspect and represent the direction and magnitude of intervention effects. The heterogeneity of results between trials was assessed by using a combination of visual inspection of the graphs, the χ2 test statistic (with statistical significance set at P < .10), and the I2 statistic.15 Publication and reporting bias were assessed by using funnel plots. Owing to the anticipated differences in participant populations, intervention characteristics, and length of follow-up, results from the trials were pooled by using the Mantel-Haenszel statistical method, as well as the random effects model in Review Manager 5.15,16
Quality of evidence was assessed by outcome of interest using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) system.17 GRADE is used for rating the confidence in the estimates of intervention effects and relies on 5 factors: (1) risk of bias (limitations in the design and execution of the trials)18; (2) inconsistency (statistical heterogeneity between study results)19; (3) indirectness (applicability of participants, interventions, and outcomes to the clinical question under consideration)20; (4) imprecision (impact of random error as reflected by the relative confidence interval (CI) of the pooled effect estimate)21; and (5) publication bias (publication or nonpublication of research findings depending on the direction and statistical significance of study results).22
Our search yielded a total of 8095 articles. After removal of 2354 duplicates, 5741 articles were included for initial screening. After screening titles and abstracts, 60 full-text articles were retrieved for further review and assessment. The references of these articles were manually screened to capture any missing publications. Four additional articles were identified through this process. Based on a full-text review, 23 articles met the eligibility criteria (41 articles were excluded). Full citations and primary reasons for the exclusion of these articles are provided in Appendix 2. The selected articles (n = 23) were part of 16 RCTs23–45 (Fig 1).
The general characteristics regarding the population, interventions, comparison group, and primary outcomes for the 16 selected studies are provided in Table 1. All studies were published between 1991 and 2014, with a majority being published on or after 2006 (n = 11). All studies were published in peer-reviewed journals except 2 studies that were published as abstracts.27,34 Most studies were single-site (n = 14). All included studies were conducted in urban hospitals in the United States. Seven studies were federally funded,24,29,31,37,38,42,43 and 1 study was partially funded through an academic award.45 The remaining studies did not provide details of funding or had no funding support.
The characteristics of the patient population varied across the selected studies. Age varied between 0 and 18 years (one study34 did not mention the age range but specified the average age as 2.2 [SD, 2.8]). One-half of the selected studies (n = 8) included patients with asthma (or acute asthma)28,31,36–38,42,43,45; 2 studies focused on suicide attempts or suicide ideation24,30; and 1 study focused on mental health problems.29 The remaining studies lacked specificity in the clinical conditions: 3 described minor health concerns26,27,40; 1 studied acute infections33; and 1 did not specify the clinical condition that was considered.34
The interventions could be generally classified into the following categories: educational support (face-to-face, video-based, or telephonic),24,27,28,31,37,38,42,43 appointment scheduling,27–30,42,45 telephonic reminders,27–30,42,45 ED-based discharge instructions,28,36,40,45 monetary incentives,37,38 text message reminders,34 providing PCP information,40 nurse support line,33 mailed reminders,43 therapeutic sessions,24,29 and case management program28 (Table 2).
Interventions in the included studies differed, varying from a single component (eg, a telephone call) to multicomponent (eg, patient education, in conjunction with a follow-up telephone call). Four studies used single-component interventions,26,31,34,36 whereas the rest (n = 12) used multicomponent interventions.24,27–30,33,37,38,40,42,43,45
In terms of the care continuity framework, 14 interventions were identified that focused on the quality of information,24,27–31,33,35,37,38,40,42,43,45 11 interventions that supported communication,26,29–31,33,34,37,38,40,43,45 and 10 interventions that promoted coordination of care.24,27–30,37,38,42,43,45 Only 6 interventions supported all 3 functions of care continuity.29,30,37,38,43,45
We also investigated the time of administration of the interventions (ie, during the visit, after the visit [eg, within 24–48 hours]). Interventions were administered during the ED visit in 3 studies27,28,36 and after the ED visit in 3 others.26,33,34 In the remaining 10 studies, interventions were administered both during and after the ED visit (ie, involving multicomponent intervention).24,29–31,37,38,40,42,43,45
Finally, 9 studies identified 1 of the following theoretical frameworks as the basis of their intervention design: health belief model and promotion of self-efficacy36,42,43; cognitive behavioral theory24; social cognitive theory31; health promotion model27; and trans-theoretical model of behavior change (decisional-balance).38
All studies had a control group that received usual (routine) or an enhanced usual care. In most studies, the control group received routine care (n = 13).26–29,31,33,34,36–38,40,43,45 Three studies included an enhanced usual care: Teach et al42 (usual care with an asthma educational packet), Asarnow et al24 (usual care with ED staff trained on linking patients with outpatient care), and Grupp-Phelan et al30 (usual care with standard referral).
The sample size ranged from 13 to 263 participants for the intervention groups, and from 11 to 264 for the control groups.
Risk of Bias in Included Studies
The domain judged to have the lowest risk of bias was incomplete outcome data and selective reporting, whereas the domain with the highest risk of bias was blinding of participants and personnel. Blinding of personnel and participants is not always feasible in studies of care transition interventions. Hence, there is a high risk of performance bias in the included studies. Knowledge regarding the intervention received could also potentially affect care delivery performance and behavioral or subjective outcomes. Consequently, blinding of outcome assessors and use of robust measurement methods become critical to avoiding potential biases.
Assessment of outpatient follow-up rates and ED readmissions were judged to have a low risk of bias in 10 of the 16 studies (Fig 2). Two studies were at high risk of bias, as they relied only on patient reported measures and did not use other objective means to corroborate patient reporting (eg, medical record/chart reviewing). Six studies had loss to follow-up. However, risk of bias for outcome assessment was not considered high because attrition rates were similar between groups. None of the studies had selective outcome reporting bias for the 2 outcomes of interest (outpatient follow-up, ED readmissions).
Figure 2 shows the risk of bias assessment for each individual study; detailed information about the reviewer judgments for each study is provided in the appendix. Figure 3 shows a summary graph of the risk of bias by each considered domain.
Effects of Interventions
Outpatient Follow-up Rates
Ten studies (n = 1965 participants) evaluated the effect of ED-based care transition interventions on outpatient follow-up rates (Fig 4). Of these, 6 found beneficial effects with the intervention.24,27,30,36,37,45 Three studies included patients with asthma, 2 included patients with mental health issues, and 1 involved patients with minor illnesses. Effect sizes varied between trials ranging from an OR of 1.20 to 9.63. In 5 of the 6 trials that demonstrated effective interventions, results were statistically conclusive24,27,30,36,45; in the sixth, the significant effects observed within the first 2 weeks (OR, 2.38 [95% CI, 1.60–3.54]) were not sustained during the 6-month reporting period (OR, 0.86 [95% CI, 0.56–1.32]).37 Strategies found to improve the rate of outpatient follow-up included structured telephone contact, postdischarge reminders,24,30 assistance with scheduling an outpatient follow-up appointment during the ED visit,27,45 and verbal/written instructions or education about the importance of outpatient follow-up.27 One study involving patients with asthma also found that providing parents with documentation of their child’s allergic status improved adherence to scheduled continuity appointments.36
Pooling of data from the 10 studies showed that the odds of outpatient follow-up visits were increased by one-half with the aforementioned care transition strategies versus routine care (OR, 1.58 [95% CI, 1.08–2.31]; P = .002). However, statistical heterogeneity was high between study results (I2 = 64%), suggesting that the magnitude of effect in a randomly chosen trial may vary considerably depending on the characteristics of the study. Heterogeneity could not be further explored or explained by means of subgroup analysis due to the small number of included studies. Similarly, the presence or absence of publication bias could not be properly assessed through funnel plots and statistical tests because of the small number of included studies.
Overall, based on the GRADE assessment, there is evidence that ED-based care transition interventions increased outpatient follow-up rates. However, the quality of evidence is moderate due to risk of bias in the included studies.
Seven studies evaluated the effect of ED-based care transition interventions on the rate of ED readmissions, but 1 trial did not provide sufficient data for meta-analysis,42 and one40 reported only disaggregated data based on the acuity level of ED visits (low versus high). In the former, there was a significantly higher cumulative risk in the control group of making ≥1 ED visit during the 6-month follow up period (adjusted RR, 1.74 [95% CI, 1.31–2.31]). In the second study, the intervention group had a significantly lower rate of nonurgent pediatric ED utilization (RR, 0.79 [95% CI, 0.63–0.99]), and the magnitude of effects was similar between the 6- and 12-month follow-up periods (RRs of 0.69 and 0.79, respectively).
In the remaining 5 studies (4 of which included patients with asthma and 1 with minor illnesses),27,28,37,43,45 ED-based care transition interventions had little or no effect on ED readmissions (RR, 1.02 [95% CI, 1.02–1.15]) (Fig 5). Results were generally consistent across the included studies (I2 = 0). Publication bias could not be assessed due to the small number of trials. Based on the GRADE assessment, the quality of evidence for ED readmission was low due to risk of bias and imprecision (ie, optimal information size not met).
To the best of our knowledge, this is the first systematic review of RCTs evaluating the effectiveness of ED-based interventions for improving care transitions of pediatric patients to outpatient settings and reducing return visits to the ED. We identified 16 RCTs, including 3760 patients, conducted at urban US hospitals. Most interventions were multicomponent, focused on patients with asthma, and were evaluated at a single site. Based on moderate-quality evidence from 10 RCTs (n = 1965 participants), we found that on average, ED-based care transition interventions for pediatric patients increase the relative probability of follow-up visits with primary care providers by 29% (OR, 1.58 [95% CI, 1.08–2.31]). The absolute effect is 11%: 113 more people (95% CI from 18 to 206 more) per 1000 had a follow-up visit with an outpatient provider when exposed to care transition interventions, equating to a number-needed-to-treat of 9.
Effective outpatient follow-up interventions included verbal/written instructions at discharge, education about the importance of outpatient follow-up, structured telephone contact, and postdischarge reminders and/or assistance with scheduling an outpatient follow-up appointment during/after the index ED visit. However, the mechanisms underlying the success of these interventions are still unknown. The magnitude of effects, nature of interventions, and patient populations varied significantly across the included studies, making it challenging to determine which components or configurations are more effective, for which types of patients, and under what circumstances.
With respect to ED readmissions, there is low-quality evidence when pooling data from 5 studies (n = 1609 patients) that showed little or no beneficial effect with ED-based care transition interventions (RR, 1.02 [95% CI, 0.91–1.15]). Two of the studies that found significant improvements in outpatient follow-up rates reported no reductions in ED readmissions with the same intervention.37,45 Our results are similar to the results presented by Katz et al,8 which suggests that primary care follow-up alone may be insufficient for improving longer term outcomes and that more intensive, multilevel interventions may be required. Furthermore, the fact that all studies were conducted in urban US hospitals makes generalization of the results to other countries or rural settings difficult.
Insights drawn from our review can have implications for both research and practice. First, in terms of the care continuity framework, most successful interventions were multicomponent and were focused on explicit strategies for coordination of care (eg, ED-based appointment scheduling or reminder), and information sharing (verbal/written discharge instructions), similar to those that have been suggested among adult populations.46 Second, all interventions were socio-technical, independent of the clinical condition being addressed. Third, although the focus of care coordination efforts of the Patient Protection and Affordable Care Act were reliant on integrating electronic health records, it was surprising that none of the interventions (published in 2011 or later [n = 5]) included components that were integrated (or could be integrated) into electronic health records or health information technology (HIT) systems. Fourth, time or cost analyses were not included in any of the studies, and thus we were unable to assess the economic viability of any of the suggested interventions. The intervention effects found with ED-based care transition interventions would be relevant if improved outcomes could be achieved and sustained across pediatric patients. However, if these interventions require additional resources and support, future studies should consider not only the cost-effectiveness but also the cost savings, as well as potential long-term benefits of interventions in promoting care transitions. Fifth, all of the interventions relied on episodic follow-up (eg, education at the ED, a follow-up telephone call). With the availability of remote and ubiquitous monitoring applications, sustainable low-cost applications relying on mobile technology may be viable solutions (eg, enhanced text-based or mobile application–based follow-up). Finally, similar to studies in adults,47–49 we found no evidence of association between outpatient follow-up rates and ED readmissions (for the same interventions). This outcome points to the need for developing interventions that engage clinicians from outpatient settings. Such collaborative endeavors can potentially help in sustaining long-term patient care management and patient–provider relationships.50
Some of the significant challenges of conducting care transition research should also be acknowledged. As highlighted in our risk of bias assessment, it is difficult to blind the personnel and participants during the study (only 1 study reported on blinding). Similarly, some of the studies relied on patient self-report as the basis of ascertaining outpatient follow-up. With the fragmented nature of patient health care utilization, it can often be difficult to ascertain actual rates of actual outpatient follow-up.
This systematic review adhered to the Cochrane methodology and Preferred Reporting Items for Systematic Reviews and Meta-analyses guidelines for minimizing the impact of bias arising from the included studies and the review process itself. We searched a range of databases and did not limit our search by date, language, or type of publication. We also ensured that 2 independent reviewers performed the study selection, data extraction, and risk of bias assessment tasks. We eliminated double counting of the evidence by merging multiple publications of the same study. In addition, meta-analytic methods were used to present and contrast study results. Although we were not successful in obtaining missing information from study authors, this outcome was not surprising given the elapsed time since the publication of some of the included studies. Also, we were unable to explore potential causes for the observed heterogeneity in outpatient follow-up rates or conduct publication bias assessment due to the small number of included studies. Systematic reviews are inevitably constrained by the quality and reporting characteristics of the included studies. The risk of bias assessment indicated that performance or other types of bias could have affected study results. To this end, we applied the GRADE criteria to rate the quality of evidence according to outcome of interest and formulated conclusions accordingly, taking into consideration the limitations of the included studies.
Our review highlights the relative success of ED-based interventions in improving outpatient follow-up, while having limited success in reducing ED readmissions. Future research should account for the feasibility, economic viability, and health information technology–based integration of care transition interventions.
The authors thank Bianca Rad, Shirley Burton, and Carla DeSisto for their help with data extraction and organization.
- Accepted April 29, 2016.
- Address correspondence to Joanna Abraham, University of Illinois at Chicago, PhD, 1919 West Taylor St, Chicago, IL 60612. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- Wier LM,
- Yu H,
- Owens PL,
- Washington R
- Miller ME; US Senate, Medicare Payment Advisory Commission (MedPAC)
- Agency for Healthcare Research and Quality
- Centers for Medicare & Medicaid Services.
- Centers for Medicare & Medicaid Services.
- Higgins JPT,
- Altman DG
- ↵Review Manager (RevMan). 5.0. Copenhagen, Denmark: The Nordic Cochrane Centre, The Cochrane Collaboration; 2008
- Higgins JPT,
- Deeks JJ,
- Altman DG
- Deeks JJ,
- Higgins JPT,
- Altman DG
- Borenstein M,
- Hedges LV,
- Higgins JP,
- Rothstein HR
- Atkins D,
- Best D,
- Briss PA, et al; GRADE Working Group
- Chande VT,
- Exum V
- Chande VT,
- Kimes D
- Salinero EA,
- Cramm K,
- Papa L
- Scarfi CA,
- Cunningham SJ,
- Wiznia A,
- Serebrisky D,
- Crain EF
- Sturm JJ,
- Hirsh D,
- Weselman B,
- Simon HK
- Sturm JJ,
- Simon HK,
- Hirsh DA
- Zorc JJ,
- Chew A,
- Allen JL,
- Shaw K
- Zorc JJ,
- Scarfone RJ,
- Li Y, et al; Randomized trial
- Lieu TA,
- Quesenberry CP Jr,
- Capra AM,
- Sorel ME,
- Martin KE,
- Mendoza GR
- Copyright © 2016 by the American Academy of Pediatrics