Pediatric patients account for a significant proportion of all emergency department (ED) visits (30% in Canada, 31.2% in Korea, and 20.5% in the United States) worldwide [1-3]. Previous studies and reports have reported that 58% [4] to 87% [5] of all pediatric patients visited the ED with nonurgent conditions, meaning that most are discharged home, where parents are expected to manage care. As a result, the discussions that ED staff have with patients and caregivers during the discharge process regarding what care is required after the ED visit is a significant component of safe practice and quality patient care [6].

Discharge communication among providers, parents, and patients occurs at multiple points during an ED visit, and sharing information related to diagnosis, prognosis, treatment plans, and anticipated course of illness is critical for successful discharge to home [7]. Poor compliance and lack of comprehension of discharge instructions have significant clinical implications, including unfinished treatment, poor pain management, and possible progression of illness [8]. Deficits in the understanding of discharge instructions are widely documented, ranging from 24% of discharged patients with poor understanding of their follow-up plan [9] to patients correctly identifying only 59% of instructions [10]. A review of 48 pediatric ED studies determined that one-third to almost half of the parents who had visited the ED with their child made medication dosing errors during post-ED home care [11]. Within the complex, fast-paced, highly stressful, and highly distracting ED environment, discharge communication can take place in as little as 76 seconds [12]. As a result, interventions to improve communication must balance reliability with flexibility across a wide range of clinical presentations.

The use of information and communication technologies (ICTs) in pediatric health care institutions is increasing, as are the multiple ways in which different technologies are deployed. Examples include computer kiosks, mobile apps, interactive television and whiteboards, electronic health records, videos, websites, and automated email [13,14]. Technologies create new opportunities for communication and dynamic updates for patient care; however, at the same time, they can also introduce potential interruptions or changes in clinical workflow [15]. Greater emphasis on the interplay between the social (people, values, and norms), technical (tools, hardware, equipment, and processes), and behavioral (routines, roles, and tasks) aspects of ICT implementation in discharge communication could help address some of these barriers [16].

To improve the experience of care during and after an ED visit, there is a great need to better leverage the strengths of technologies to support efficient discharge processes, particularly for nonurgent visits. However, few guidelines exist to support health care institutions in decision-making and implementation planning for such technologies. Research on the use of ICT to support care transitions is predicted to grow rapidly as patients and clients increasingly demonstrate preferences for the use of these technologies in their care [17]. Health care providers also recommend better and more appropriate use of ICT to support families in self-managing care at home [18]. Despite the communication challenges faced by families during this transition point, strain on existing ED resources and the lack of standards and implementation guidelines remain significant barriers to the widespread adoption of electronic discharge communication tools (EDCTs) in pediatric emergency contexts. Systematic reviews of traditional pediatric discharge communication practices [19,20] and computer technology have enabled discharge communication outside the ED [21]; however, to our knowledge, there has not been a comprehensive review of how EDCTs are being used to support and guide pediatric emergency discharge communication.

This systematic review of academic literature was undertaken to identify, appraise, and describe the use of EDCTs in pediatric emergency contexts. Our goal is to advance the knowledge base for researchers, technology designers, and decision makers to anticipate the impact of their communication tools on the clinical workflow and the optimal ways to measure impact (Textbox 1).

The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) [22] guidelines were followed, and the review was registered with PROSPERO CRD42020157500.

A comprehensive search strategy using the Population Intervention Comparator Outcome framework [23] was codeveloped with an experienced information technician. The search terms were intentionally broad to capture the range of EDCTs. Namely, terms included technology (eg, electronic documents or web-forms, mobile device apps, patient portals, notification systems, text messages or SMS notifications, interactive online decision trees, automated email, and video-based programs) used to prompt communication between caregivers/patients and ED staff about the ED visit, and structure the exchange of information, or promote compliance, education, and information sharing about what care should be given after the ED visit is over. A total of 7 databases of publisher-controlled and gray literature were searched: EBSCO, MEDLINE, CINAHL, PsycINFO, EMBASE, Scopus, and Web of Science. The original search was conducted in June 2019 and was updated in August 2021 to capture current evidence. Records from 1989 onward were included. Multimedia Appendix 1 presents the sample search strategy. The reference lists of systematic reviews were also hand searched for primary studies.

We used a broad definition of EDCTs, including tools that prompt communication between caregivers and patients and ED staff about the ED visit and structure the exchange of information and promote compliance, education, and information sharing about what care should be given after the ED visit is over. We did not limit the search to a particular technology modality; therefore, tools including web-based documents or web-forms, mobile device apps, patient portals, notification systems, text messages or SMS notifications, interactive web-based decision trees, automated email, and video were eligible for inclusion. As telephone-based services are part of Health Canada’s definition of eHealth, we included phone-based services under the broad umbrella of electronic tools.

Specific inclusion and exclusion criteria are presented in Textbox 2.

Eligibility screening was performed using Covidence software [24]. All titles and abstracts were independently reviewed by 2 reviewers. Discrepancies regarding which studies to include in full-text reviews were resolved by discussion. A total of 2 reviewers independently assessed the full texts for inclusion. Discrepant classifications were resolved through discussion.

The team co-designed and piloted a structured data extraction table with the 4 studies included in the review. The form included sections on (1) study identification (eg, type of publication, year, and author); (2) methods (eg, study design and sample size); (3) delivery settings and demographics (eg, ED features, age, setting characteristics, and computer proficiency); (4) intervention design (eg, design framework, frequency and duration of interaction, tailoring, bidirectional functionality, content, tool, and primary technology modality); (5) outcomes (eg, category of outcome measure, follow-up schedule, and covariates); (6) implementation (eg, who administered the tool, training requirements, interoperability, and cost); and (7) practice, policy, and research implications extracted verbatim from the Discussion and Conclusions sections.

As a broad range of study designs was anticipated, the Mixed Method Appraisal Tool (MMAT) version 2018 [25] was used for methodological quality appraisal. The MMAT is a 21-item checklist with 5 research designs. Each research design category has 5 quality criteria that are appraised as yes (criterion met) and no (criterion not met or cannot tell [unable to tell from text if the criterion was met or not]). Assigning studies an overall numerical score based on the ratings of each criterion is discouraged, because a single number cannot provide insight into which aspects of the study methodology are problematic [26]. Instead, we classified studies as having lower methodological quality when they met ≤60% of the MMAT criteria and higher quality when they met >60% of the criteria. This is consistent with the approaches outlined by the MMAT authors [26].

A reviewer independently conducted data extraction and MMAT scoring for all full-text articles. As a quality assurance measure and to ensure the accuracy of extraction, a second reviewer independently extracted data from a randomly selected subset of 30% of full texts. The results were compared, disagreements were resolved by discussion, and additional instructions for the coder were updated.

Following standard practices for systematic reviews–included [27] studies were synthesized using several approaches: (1) structured tabulation to explore patterns in the raw data, (2) vote counting of raw data (eg, reporting on the frequency of different study features), (3) constructing a common rubric to transform qualitative data (eg, lengthy descriptions of the technology features) into a simplified quantitative form (eg, assigning tools to a modality category), (4) descriptive statistics (eg, range, mean, or median) to summarize quantitative data, (5) inductive thematic analysis (eg, hierarchical coding of verbatim policy, practice, and research implications), and (6) visual depiction of summary data.

Duplicates were excluded, and 17,827 potential reports were returned. Hand searching of the reference lists of 15 related systematic reviews produced no additional eligible full-text reports. A total of 231 reports were read in full, with 182 (78.8%) excluded, leaving 49 (21.2%) reports detailing findings for 55 unique EDCTs. A flowchart of the process is shown in Figure 1.

The studies were conducted in 8 countries between 1989 and 2021 (Table 1). The intervention group sample size ranged from 3 to 4091 participants or events (median 95). In all, 62% (34/55) of the studies were conducted in the United States and 20% (11/55) in Canada, with the remainder (10/55, 18%) conducted in Australia, China, the Netherlands, South Korea, and the United Kingdom. A study did not report the country of origin. Interventions were evaluated using randomized controlled trial designs in 58% (32/55) of the cases, nonrandomized trials and cohort designs in 22% (12/55), quantitative descriptive studies in 20% (10/55), and mixed methods at an instance 2% (1/55).

MMAT appraisal was conducted on 37 studies (abstracts for which no full text was available were excluded). Overall, the methodological quality of the studies varied: 30% (11/37) of the studies met ≤60% of the criteria outlined by the MMAT (lower methodological quality), and 70% (26/37) of the studies met >60% of the criteria (higher methodological quality) [26]. Reviewers’ ratings for each methodological quality criterion are presented in Multimedia Appendix 2 [28-33,35-39,41-46,49-51,54-66,68,73,74,76].

In all, 40% (22/55) of the EDCTs were designed for use after the ED visit when families were already at home. Over half of the tools targeted a single specific presenting complaint with asthma (15/55, 27%), fever (6/55, 11%), fractures (3/55, 6%), head injury (3/55, 6%), and otitis media (3/55, 6%), being the most frequently cited. In 13% (7/55) of studies, the discharge communication tool could be used for multiple presenting complaints (eg, patients with fever or head injury). Finally, 20% (11/55) of the tools were designed for use in any illness presentation. Some tools focused on a specific task or a narrow aspect of discharge communication (eg, medication regimen adherence) [70], whereas other tools were multi-focused with broader education, symptom monitoring, and care plan elements [57].

EDCTs support diverse communication pathways among providers, caregivers, patients, and other health care providers. Most of the tools targeted communication between an ED health care provider and the parent and caregiver (52/55, 94%) with a smaller number (6/55, 11%) also including communication with other health care providers (eg, family physician). One study of the Texas Emergency Department Asthma Surveillance programs [68] was an example of a multi-audience tool. In the study, the ED asthma educator used a Microsoft-based platform to individualize an education package for the caregiver (eg, select relevant video segments, figures and graphs, skills training, and motivational messaging). The plan was shared and discussed with the caregiver and then printed and sent to the family’s primary care provider. The educator could also generate and print a child-friendly version of the tailored written action plan for elementary-aged patients.

The primary technology modalities used were videos (20/55, 36%), kiosks (11/55, 20%), telephone calls (7/55, 13%), and text messaging (6/55, 11%). The remaining modalities include a wide range of offline stand-alone interactive computer programs and web platforms, mobile apps, interactive websites, and web-based games with multiple audiovisual elements. For example, a private multiplayer web-based social game called iCare Adventure uses noncompetitive gameplay for children and parents to explore therapeutic content on an iPad while in the ED waiting room [74].

A density map of presenting complaints targeted and primary technology modalities used to deliver the EDCT was generated (Tables S1 and S2 in Multimedia Appendix 3). Darker cells indicate where the largest number of studies have been conducted. Kiosks and videos are the 2 predominant modalities used as stand-alone asthma tools. Videos are the most studied modality for less frequently investigated medical concerns (eg, vomiting and pain).

There was substantial heterogeneity between the studies in the amount of time and effort required by patients and caregivers to use the tool. In all, 42% (23/55) of the tools required single-use, time-limited interaction (eg, watched one video once or entered information at a kiosk once). A total of 3 studies involved web platforms or interactive computer programs with larger educational components that allowed unlimited access (4/34, 12%). A program provided access over a specified follow-up period (eg, 120 hours after discharge) [43]. Multiple planned interactions with a tool typically involved a level of automation (eg, 2 automated text messages twice a week for 4 weeks) [51] or chronologically sequenced learning modules. All text messaging interventions were automated 1-way messaging of reminders with no option of bidirectional texting directly with a health care provider.

Duration of contact with the EDCT (ie, how long it took end users to complete expected tasks) was reported in 31% (17/55) of the studies. Among those that did report, the length of contact time for the patient and caregiver ranged from 110 seconds at a kiosk [37] to 80 minutes (where the latter measured the time to complete 5 web-based modules) [63]. A total of 44% (7/17) of those reporting took ≤5 minutes to complete, (4/17, 24%) took between 6 and 10 minutes, and 24% (4/17) took >10 minutes. The interventions (3/17, 18%) that took >12 minutes all specifically targeted asthma. Caregiver perceptions of frequency and duration were explored in a study of 243 families where 66 (27.2%) reported they had “had no time” to enter the website [64].

There was significant heterogeneity in the reported purpose of deploying the EDCT and subsequent outcomes measured. Tables S1 and S2 in Multimedia Appendix 4 show a matrix of the outcomes measured per mode of EDCT technology delivered. The intensity of shading shows clusters (darker) versus gaps (lighter) within technologies.

The highest density of evidence was from the study of changes in caregiver knowledge after using video-based EDCTs (16 instances). The most assessed category of outcomes overall (including both primary and secondary) were caregiver and patient beliefs and attitudes (eg, confidence in managing at home and level of anxiety; 36 instances), knowledge and comprehension (eg, knowledge about symptoms; 29 instances), and health service use (eg, return visits to the ED; 25 instances). Health care provider satisfaction (5 instances) and cost (2 instances) were the least measured outcomes across all technology modalities.

Text message interventions were more likely to be measured on behavioral outcomes (eg, compliance with medication regime and follow-up appointment with primary care), whereas studies of video-based EDCTs typically used knowledge acquisition–related measures. A randomized controlled trial by Jové-Blanco et al [46] comparing video discharge instructions and standard verbal instructions for gastroenteritis showed that 49% of the intervention group and 18.6% of the control group answered all knowledge acquisition questions correctly (P<.001) [48]. However, EDCTs with greater technological sophistication do not always produce better knowledge outcomes. In a head-to-head trial of a static website and an interactive website about fever, Hart et al [65] unexpectedly found that both modalities had comparable knowledge gains, although caregivers were significantly more satisfied with the interactive version.

Measurement of knowledge outcomes occurred largely through bespoke self-report questionnaires that assessed general knowledge about symptoms, treatment options, medication and activity adherence, and service use [29]. Validated measures were most often cited in relation to patient health status (eg, Faces Pain Scale-Revised and Acute Otitis Media Severity of Symptom) and functioning (ie, Acute Asthma Behavioral Capability Questionnaire; Integrated Therapeutics Group Child Asthma Short Form). No adverse events were reported in any of these studies.

The directionality of primary outcomes pointed to positive effects for the primary measure (44/55, 80%) or no significant difference (10/55, 18%). Only one study reported negative findings with an increase in return visits to the ED after receiving the intervention compared with the control group (P<.03) [59]. Often, the authors reported positive primary outcomes but mixed results across secondary measures. For example, a study by Baker et al [39] showed increases in parental knowledge about fever but no significant differences in subsequent health service use. Similarly, a study by Zorc [56] showed significant changes in beliefs about the benefits of follow-up, but medication adherence and ED visits did not significantly differ at follow-up. Parental satisfaction with EDCTs was consistently moderate to high across all technology modalities. However, in some instances, respondents in the control condition, typically verbal or written discharge instructions, also reported high levels of satisfaction [46].

The ability to tailor information via the EDCT was particularly well received by parents when this option was available. For example, tailored mental health recommendations facilitated by electronic screening were perceived by parents as more useful (69.5% vs 30.5%) and more practical (71.8% vs 28.2%) compared with verbal instructions [34]. In another study, 23% of caregivers’ free text entries in the EDCT provided data that were not contained in the official electronic medical record [35].

Patient age [72], gender of caregiver [62], and parent education level [37] were the most frequently reported, statistically significant covariates vis-à-vis the primary outcome. Of note, only 3 studies reported collecting baseline data on the level of computer proficiency [28] and none in the past decade.

In all, 42% (23/55) of the EDCTs were evaluated in at least one explicitly stated urban community. The majority were evaluated in pediatric EDs (37/55, 67%) or mixed ED settings (ie, both adult and pediatric populations, 13/55, 24%); the rest provided insufficient information to decide. English, Spanish, and Dutch were the only languages in which interventions were available and evaluated. No other culturally specific content or culturally adaptive features of the interventions were reported. The interventions (12/55, 22%) included baseline racial demographic factors, with most participants being African American or White. The EDCTs were most frequently delivered by research study staff (19/55, 35%), ED health care providers (15/55, 27%), or by computers or automated systems (8/55, 15%).

Very few interventions (3/55, 6%) were tested in studies that provided remuneration to participants. No studies have reported interoperability with other ICT systems within the ED or hospitals. The authors of 2 interventions (2/55, 4%) briefly mentioned sustainability planning, and 33% (18/55) stated that due consideration should be given to the technical performance of the system. Only 2 interventions (2/55, 4%) included details of direct costs; a study reported that per patient mean cost for videos was US $61 (SD US $36) versus US $31 (SD US $20) for phones; P<.001 [50]. Another study estimated the operating budget for the tool in “hundreds of dollars” [74]. Privacy and security were highlighted as necessary implementation context considerations in 11% (6/55) of the instances.

No direct policy or decision-making implications were explicitly discussed by the primary authors. High-level theming of future research directions posited by primary authors revealed three main directions: (1) more diverse sample populations that reflect a wider view of social determinants of health, (2) triangulation of data from sources outside of self-report (eg, primary care follow-up data and hospital administrative data), and (3) isolating the functionality of the tools to test the impact on engagement (eg, increase uptake). Practically, the authors generally endorsed the use of EDCTs, even if statistically significant findings were mixed or effect sizes were modest.

The primary aim of this review was to describe and assess evidence based on the EDCTs used in pediatric EDs. The evidence base included the principal features, measured outcomes, and implication contexts under which they were studied.

First, an important and promising finding of this review is that although the contextual complexity of EDs poses communicative challenges and risks, there is a growing body of evidence that EDCTS have been successfully integrated. Our review found at least five studies in each of the 4 major modality categories (ie, videos, kiosks, text messaging, and phone-based) and numerous presenting complaints that are among the most frequent reasons for ED visits reported in the literature (asthma, fever, head injury, fractures, pain, mental health, etc) [77]. In other words, there is growing breadth and depth of positive evidence.

The evidence base for newer technology modalities, kiosks, text messaging, and web-based games and apps is still maturing, with just under a third of all studies being conducted in the last 5 years. It is vital to monitor this evidence base as more automated and ambient technologies (eg, chat bots, wearables, and artificial intelligence) become normalized. Indeed, they are already being studied in ED communication for the adult population [78,79]. Our review adds to this dialogue by showing that technological sophistication may not necessarily result in clinically meaningful improvements. Videos and phone calls also produced positive changes. In fact, most EDCTs in this review reported at least some positive impact in 80% of cases and no adverse events. There is a need to move beyond demonstrating the known value of EDCTs and focus on how to optimize which tools for which populations, under which circumstances. This is supported by caregivers reporting high satisfaction regardless of modality or presenting concern. In other words, the technology modality used to support caregivers in discharge planning may be less crucial than the opportunity to engage with them.

Second, our review has shown that EDCTs have been largely assessed for changes in cognition (knowledge and beliefs), meaning that we know less about their impact on behavior (adherence to treatment regime), therapeutic relationship (caregiver-provider rapport), or service use. Our findings and overall methodological quality appraisal results point to the need for future meta-analyses to explore the magnitude and direction of effects within specific modalities. Such an analysis could support decision makers in determining which tools are fit for different primary purposes, reduction in nonurgent visits versus improved experiences of care. Caregivers may be highly satisfied with a tool and experience improved recall and comprehension, but this may not translate into fewer nonurgent visits to the ED in the future. The lack of description provided in primary studies related to implementation and environmental context features contributes to gaps in knowledge about the sustainability of these tools, particularly the costs associated with setup and ongoing operations.

Another significant finding of this review is that outcomes related to caregiver-provider rapport were understudied across all modalities and for all clinical presentations. This gap in the evidence is exacerbated by the few studies that assessed health care provider satisfaction with the tools in general. Assessment of their expectations and experiences with EDCTs may help illuminate barriers and enablers to uptake, as well as predictors of positive and negative client experiences. Recent work on quality pediatric communication in EDs [80] points to gaps in measures of care experiences in a complex, high-stress environment. Given the diverse implementation contexts for EDCTS found in this review, the development of quality standards for discharge communication should consider the role of electronic tools, which will undoubtedly continue to mediate and moderate care experiences in the future.

Finally, research designs for EDCTs need to incorporate mediators and moderators related to technological functions (eg, synchronicity, automation, visual aesthetics, and gamification) to determine the minimum viable functions. Our findings suggest that technological complexity is not necessarily better. Augmenting quantitative self-report survey data with observational, qualitative, and administrative data could help make sense of the aspects of these tools (ie, mechanisms of change) that drive the desired change. For example, there was some evidence that tools take >5 minutes for caregivers to complete (impact on workflow) and were administered by research team members rather than health care providers, giving us a slightly skewed view of real-world implementation. More work is needed to understand how the duration and frequency of interaction with tools (both provider and caregiver or patient) could be optimized for busy ED workflows without adding unnecessary complexity to the clinical pathways. Our review showed that over half of the EDCTs studied to date target a specific illness, but this could add burden to health care providers and caregivers who might then need to access and navigate a different tool for each presenting condition.

The findings of this review point to several high-impact future lines of research to address gaps, including (1) exploring how computer-mediated communication in pediatric emergency contexts impacts the quality dimensions of communication and rapport building (eg, sense of shared decision-making, empathy, and active listening), (2) meta-analysis of data subsets within a particular presenting illness field (eg, asthma) or within a single well-defined technology modality (eg, kiosks), (3) developing taxonomies for electronic discharge communication interventions that capture complex person-to-person and person-to-technology pathways, and (4) use of A or B (ie, split) testing to isolate specific technology features that may be driving outcomes so that the least intensive interventions necessary to achieve desired outcomes are pursued by developers and decision makers.

This study had several limitations. First, mapping the broad relevant literature parameters of EDCTs lacked clarity before the literature search. Terms related to technology, digital devices, and electronic communication were ambiguous in the literature, and our criteria were subject to significant revision during the initial search execution. This resulted in a less-focused initial title and abstract screening process. Second, the review included several study abstracts that were not published as full articles, limiting what data could be abstracted and fully analyzed. Finally, no taxonomies for presenting complaints have been validated or published in the literature; likewise, no taxonomies for electronic communication modalities are commonly used. Thus, our heat-map categorizations were based more on practical considerations and, to a lesser degree, on theoretically validated distinctions.

To our knowledge, there has been no other systematic review of the broad evidence related to EDCTs in pediatric EDs. The findings demonstrate that a range of technologies are being used successfully. However, it is essential that trials of emerging technologies use robust and consistent measures of quality patient-provider communication, clinician experience, cost-effectiveness, and health service use so that influential evidence on these outcomes can accumulate.