Introduction

Family systems interventions in pediatric chronic illness are used across pediatric populations (e.g., cancer, T1DM, TBI, asthma, obesity, and sickle cell, Eccleston, Fisher, Law, Bartlett, & Palermo, 2015; Mullins, Gillapsy, Molzon, & Chaney, 2014). These interventions are of critical importance because of the prevalence of pediatric chronic illness (e.g., Van Cleave, Gortmaker, & Perrin, 2010) and the wide-reaching impact of many pediatric illnesses on the entire family system. For example, within the field of pediatric oncology, several studies highlight the interconnectedness between family members in terms of psychosocial functioning and outcomes (Gutiérrez-Colina, Lee, VanDellen, Mertens, & Gilleland Marchak, 2017; Pai et al., 2007; Van Schoors et al., 2017). Moreover, research shows that parent, child, and family-level functioning are closely linked, and that a positive family environment supports better physical and psychosocial outcomes for children with chronic illnesses (Alderfer, Kazak, & Canter, in press). Family systems interventions include interventions for parents (e.g., Sahler et al., 2005, 2013) or multiple members of the family (e.g., Kazak, Simms, & Rourke, 2002; Palermo et al., 2009). While the advantages of family systems interventions are well documented in the field, logistical difficulties related to participant recruitment, scheduling, and retention often interfere with intervention delivery, particularly during times of acute crisis (Devine et al., 2016; Lutz Stehl et al., 2009; Tercyak, Donze, Prahlad, Mosher, & Shad, 2006).

eHealth (electronic) and mHealth (mobile) interventions may help overcome some of these challenges by offering alternative and potentially more accessible modalities of intervention delivery when compared to in-person interventions. Broadly speaking, eHealth interventions use technology (e.g., telehealth) to enhance delivery of psychosocial or medical care, and may supplement or replace in-person intervention (Cushing & Steele, 2010; Palermo, 2008). mHealth generally refers to the use of mobile technology, including smartphones, wearable devices, and text messaging (Borrelli & Ritterband, 2015). With regard to family systems interventions, eHealth and mHealth interventions present a unique opportunity to overcome several barriers common to in-person interventions. For example, eHealth and mHealth interventions are generally more flexible than traditional face-to-face interventions in terms of timing and location of intervention delivery, which may make it easier to recruit and retain multiple family members. A recent manuscript from Doss, Feinberg, Rothman, Roddy, and Comer (2017) highlights some of the potential advantages of using technology to deliver family interventions, including reduced barriers to care and potential engagement of multiple family members, some of whom may be unwilling to commit to an in-person intervention. Additionally, given the high estimated costs associated with several pediatric chronic illnesses (e.g., Miller, Coffield, Leroy, & Wallin, 2016) and the increased focus on cost-containment and cost-effectiveness within healthcare (McGrady, 2014; Orszag & Emanuel, 2010), eHealth interventions that increase access to and efficiency of evidence-based interventions may be appealing to insurers and healthcare systems from a financial perspective.

A recent review of the literature found that the internet is commonly used by caregivers to seek out health-related information and communicate with health professionals, as well as to build support networks (Park, Kim, & Steinhoff, 2016). Most households in the United States have access to a computer (83.8%) and/or smartphone (77%) and 74.4% have internet access (File & Camille, 2014; Pew Research Center, 2017), indicating that web-based interventions are a potentially important mechanism for overcoming barriers to face-to-face intervention services. To that end, eHealth and mHealth interventions have been utilized increasingly by psychologists for delivery of family interventions in cystic fibrosis (e.g., Stark et al., 2016), injury (Marsac et al., 2013), encopresis (e.g., Ritterband et al., 2008), chronic pain (e.g., Long & Palermo, 2009; Palermo et al., 2009), and cancer (e.g., Mayer et al., 2010).

Several meta-analytic and systematic reviews have previously been conducted regarding use of eHealth/mHealth in child psychology (Brannon & Cushing, 2015; Cushing & Steele, 2010; Fedele, Cushing, Fritz, Amaro, & Ortiga, 2017; Stinson, Wilson, Gill, Yamada, & Holt, 2009), and a recent review summarized the literature related to technology-based interventions for youth with mental/behavioral health concerns and their parents (Watson McDonell & Prinz, 2017). However, to the best of our knowledge, there has not been a prior review specifically focused upon technology-focused family interventions in pediatric chronic illness. As these interventions increase in frequency and popularity, there is an important need to summarize the existent literature and understand the potential impact of these interventions on family-centered processes (e.g., conflict, communication). This primary purpose of this review is to describe characteristics of current technology-focused interventions for families of children with chronic illnesses, including any evidence for effectiveness with regard to targeted family-centered processes. A secondary aim is to generate recommendations for future family systems work in the field of eHealth/mHealth based upon the present state of the literature.

Methods

Search Process

Procedures for this systematic review were guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) model (Moher, Liberati, Tetzlaff, Altman, & The PRISMA Group, 2009). Initial literature searches were performed in PubMed, CINAHL, and PsycInfo for all articles published from database inception through August 2016. A second search was conducted in July 2017 to capture any additional published studies. Two medical research librarians assisted with the selection of search terms and search process. No attempts were made to locate or contact researchers for unpublished studies such as dissertations or conference presentations.

Search terms included the following words: “family system,” “family function,” “family,” “parent,” “eHealth,” “mHealth,” “internet,” “telehealth,” “web-based,” “pediatric,” “child.” A wildcard symbol (*) was used for certain terms (i.e., family function*, family system*, child*) to ensure that variations of the keywords were retrieved. Reference lists of seminal papers in the field and papers identified for inclusion through the initial literature search were also manually searched to ensure that all potentially eligible studies were identified. The search yielded 5048 peer-reviewed articles, with 3262 remaining after duplicates were removed.

Study Selection

To be eligible for inclusion in the systematic review, articles had to meet the following criteria: (1) written in English; (2) published in a peer-reviewed journal; (3) randomized control trial; (4) pediatric chronic illness focus (pediatric defined as ≤ age 21); (5) Family focused, defined as either focusing on a family systems outcome (e.g., direct impact of a pediatric health condition on a parent behavior), a family process (e.g., problem solving), or a family relationship (e.g., parent–child relationship). (6) eHealth or mHealth intervention.

Articles were excluded if they failed to meet any of the inclusion criteria. For the purpose of tracking reasons for exclusion, articles that were excluded for multiple reasons (e.g., not pediatric, not eHealth) were only tracked as excluded for the first inclusion criteria that they failed to meet. Interventions that were primarily in-person interventions (e.g., in-person intervention with two phone-based check-ins) were excluded. Although mHealth studies were eligible for inclusion, no studies that met all inclusion criteria featured mHealth interventions.

Review Process

Twenty percent of all initially identified articles were screened independently for inclusion/exclusion by three coders (K.C., J.C., and M.S.) to establish reliability. All discrepancies were reviewed and discussed until consensus was achieved. Following the establishment of reliability, remaining articles were evenly divided between the three coders and titles and abstracts were screened to determine eligibility for full-text review. The coders then conducted full-text review of eligible articles, and met again to review the process for data abstraction. Data abstracted from the articles included: sample information (e.g., population, age range), intervention information (e.g., type of eHealth, length of intervention), outcome information (e.g., target outcomes, data collection time points, measures used), and family-specific results (e.g., results related to family conflict or communication). Final eligibility for inclusion in the systematic review was established by consensus of all coders.

Study quality was assessed using the CONSORT checklist for reporting trials of non-pharmacologic treatments. All included studies (n = 23) were independently assessed by two study authors (K.C. and M.S.). Agreement was 93% following the first search and all discrepancies were discussed until consensus was achieved. Three additional studies were added to the review following the conclusion of the second search, and agreement was 100% for these new studies. Using a descriptive system created by Stinson et al. (2009) to group studies based on quality according to the 23-item CONSORT checklist for non-pharmacological trials, studies were rated “poor” if scored ≤ 8, “fair” if rated between 9 and 16, and “good” if rated ≥ 17. According to this system, articles were awarded one point for each item on the CONSORT checklist that was addressed by the authors. A copy of the most recent CONSORT checklist is available for download from the CONSORT statement website (http://www.consort-statement.org/extensions?ContentWidgetId=558; Boutron et al., 2017).

Results

A final search yielded 3262 unique records. Of these, 3123 were removed after an initial title/abstract screen. Full-text records (n = 139) were assessed for eligibility, and 116 were excluded for reasons detailed in Fig. 1 (e.g., not pediatric health focused). Twenty-three articles were included in the systematic review.

Fig. 1
figure 1

Reproduced with permission from Moher et al. (2009). http://www.prisma-statement.org

PRISMA diagram for article selection.

Full size image

Study Quality

There was some variability with regard to study quality for the included studies. One study (4%) was rated as “poor,” 7 studies (30%) were rated as “fair,” and 15 (65%) studies were rated as “good.” The only study rated as poor did not report anticipated results due to high attrition, so zeros were awarded for many categories in the rating system by default. It appeared that the majority of the “fair” studies and all of the “good” quality studies were reported with specific attention to include the elements of the CONSORT checklist, as they frequently reported on specific items covered by the checklist such as randomization procedures.

Characteristics of Study Participants

Detailed demographic data for each study are shown in Table 1. InterventionsFootnote 1 targeted families of children with asthma, burns, cancer, chronic kidney disease, chronic pain, epilepsy, headache, obesity, type 1 diabetes, sickle cell disease, and traumatic brain injury (TBI). Interventions included a wide range of number of participants, ranging from 6 individuals to 320 individuals. Most interventions included multiple members of the family; however, 6 of the 23 interventions were delivered to only one family member (4 to caregivers only, 2 to children only). About one-third of the studies (8 out of 23) were from the same research group and focused on TBI. 17 studies provided some measure of family socioeconomic status or income, 14 studies included some measure of parent educational status, and only 4 studies provided any information about parental employment status.

Table 1 Characteristics of participants
Full size table

Characteristics of Interventions

Interventions varied in specific characteristics (Table 2). Control group conditions included no-treatment waitlists, usual care, specialized care, and access to information-only websites. Intervention length ranged from 3 to 36 sessions, and duration ranged from 2 weeks to 2 yearsFootnote 2. 17 of the 23 interventions were delivered via multiple modalities, such as through a website with a teleconference follow-up component. 20 of the 23 interventions included some form of contact with an interventionist, ranging from online messaging with a coach to videoconferencing to review session content and learn new skills.

Table 2 Characteristics of included studies
Full size table

Family-Centered Processes

Several common categories of family-centered processes were evident across studies and are discussed below. Refer to Table 3 for detailed information about the processes measured in each study. It was rare for the family-centered process to be the sole or primary outcome. The majority of included studies investigated a range of measured outcomes, including feasibility and psychosocial outcomes (e.g., internalizing symptoms) for a specific member of the family.

Table 3 Family-specific results for included studies
Full size table

Family Conflict/Problems/Communication

Eight studies (Glueckauf et al., 2002; Grey et al., 2013; Ketchen et al., 2006; Lehmkuhl et al., 2010; Narad et al., 2015; Sveen et al., 2017; Wade et al., 2011; Whittemore et al., 2012) targeted conflict/problems/communication between members of the family and assessed this as an intervention outcome, with six of the eight studies reporting a significant reduction in some form of conflict following the intervention for at least some participants. Whittemore et al. (2012) also reported a reduction in conflict for all participants, although this reduction was not significant. Several studies also reported significant results in the opposite direction than was predicted. Lehmkuhl et al. (2010) found that parents in the treatment arm reported a significant increase in unsupportive behaviors and youth in the treatment arm reported a significant decrease in warmth and caring parenting behaviors, while parents and youth in the waitlist group reported the opposite effects. Narad et al. (2015) reported mixed effectiveness with regard to parent–child conflict, but also found significant declines in observed effective parent–child communication for participants with severe TBI in the intervention group. Conflict was measured differently across studies, ranging from illness-specific measures of family conflict (Grey et al., 2013) to single items about perceived relationships between family members (Wade et al., 2011). Although it is difficult to draw summative conclusions due to different definitions and measurement of conflict/problems, results from included studies suggest that eHealth interventions are generally effective with regard to decreasing family conflict.

Protective Parenting Behaviors

Three studies (Law et al., 2015; Palermo et al., 2009, 2016) focused on chronic pain and headache with a goal of reducing protective parenting behaviors. One study by Palermo et al. (2016) reported a greater decline in protective behaviors and miscarried helping for the intervention group when compared to the internet education only group; the two remaining studies (Law et al., 2015; Palermo et al., 2009) found declines in both the intervention group and the control group. This may suggest that eHealth interventions are as effective, but not more effective, than other treatment options (e.g., specialized in-person care) for changing parent behavior related to pediatric chronic pain.

Caregiver Self-Efficacy

Caregiver self-efficacy was an intervention target in three studies from two research groups (Petranovich et al., 2015; Swallow et al., 2014; Wade et al., 2014). All studies found improvement in measures of self-efficacy for perceived self-efficacy and competence in management of their child’s condition, although self-efficacy was defined and measured differently by all investigators. One group (Petranovich et al., 2015) concluded that their results likely suggested preexisting differences between parents in the intervention and control group rather than an effect of the intervention, and another study from this same research group (Wade et al., 2014) found a significant effect only for parents who were non-frequent computer users. Mixed results, such as improvements for very specific subgroups of users, do not make a strong argument for the general effectiveness of eHealth interventions on caregiver self-efficacy. However, there is some evidence that self-efficacy can be improved for certain subgroups of caregivers.

Other Family-Specific Outcomes

Other outcomes included observed parenting behaviors (Antonini et al., 2014), mealtime behavior (Davis et al., 2013, 2016), problem-solving skills (Narad et al., 2015; Wade et al., 2006, 2012; Wakefield et al., 2016), parenting stress (Sveen et al., 2017), overall family functioning (Wakefield et al., 2016), family life satisfaction (White et al., 2004), general parent–child relationship (Ketchen et al., 2006), and collaboration with parents (Whittemore et al., 2012). Results were mixed with regard to effectiveness; however, due to great variability among the processes selected by authors, specific results are described in Table 2. Only two studies focused on a disease-specific outcome. Results were not reported for asthma parenting behaviors (Clarke et al., 2014) because of high study attrition, and a significant reduction for certain domains within parent pain-related impact was reported by Palermo et al. (2016).

Discussion

The current systematic review sought to synthesize the eHealth/mHealth literature with a specific focus on family-centered processes. Family-centered care has long been recognized in the literature as an important component of optimal psychosocial care for children with a chronic illness, as the impact of childhood chronic illness extends beyond the affected child and often presents as a major stressor requiring a number of systemic changes with wide-reaching implications for all members of a family system. For the purposes of this systematic review, we use the term “family-centered” to describe a specific focus on a family process or relationship. Although eHealth and mHealth studies were eligible for inclusion, no studies utilized an app-only (mHealth) delivery modality.

Family-centered processes were not the primary measured outcome for many included studies; in fact, many studies aimed to establish feasibility/acceptability of a novel intervention design, measuring psychosocial outcomes (including family variables) as secondary outcomes. Results highlighted the broad range of work being done in this area; given the variability, it proved difficult to draw conclusions about the “best” intervention format or mode of delivery for family-centered eHealth interventions. For example, some included studies used a self-directed format with no live interventionist, while other interventions were delivered entirely via telehealth. The overwhelming majority of studies featured some type of hybrid design, with only three studies relying on a “pure” internet delivery modality. This highlights the wide range of formats that are captured under the eHealth umbrella and may be logical given the relative newness of this line of research. However, this variability with regard to design and methods also presents unique challenges when attempting to summarize the literature and draw conclusions about overall effectiveness, which was a primary aim at the outset of this study. Future researchers might consider using standardized intervention formats or delivery modalities in order to allow for easier comparisons between interventions and the ability to more readily evaluate effectiveness for a certain type of eHealth intervention. Careful attention to the CONSORT checklist may also help ensure that published literature in this field is high quality, which will be helpful for future researchers hoping to draw conclusions about the literature.

The majority of interventions measured family processes as one of the many outcomes (e.g., internalizing symptoms), and often were directed to a specific family member (e.g., skills for parents). In the future, researchers and clinicians may want to explore options for including multiple members of the family at one time, as this more closely mirrors traditional family therapy and may lead to more robust results regarding family systems variables. For example, the ability to include multiple members of the family from different locations in an intervention via eHealth may lead to improvements in family cohesion or decreases in family conflict. This is particularly important given the interconnectedness between members of a family system, and the impact that one member’s psychosocial functioning can have on other members of the family (e.g., the impact of parental depression on child functioning). Engagement of multiple family members may also be higher for eHealth interventions, as individuals who are unwilling to attend in-person mental health sessions due to stigma may also be more willing to engage in eHealth interventions (Doss et al., 2017). Additionally, future work should move beyond feasibility/acceptability and focus on family processes and outcomes as primary intervention targets in order to better understand the potential impact of these novel interventions.

Additionally, while many of the included studies reported a positive effect on targeted family-centered processes, this positive effect was not always stronger or different than the effect reported by participants in the control group. Control group conditions also varied greatly among included studies, ranging from no-treatment waitlists to provision of highly specialized in-person care. Further complicating matters is the fact that many study teams found positive intervention effects for one component of family functioning while finding non-significant or negative results for other family-focused outcomes. This is consistent, to some extent, with previous literature related to family functioning in the context of childhood illness. For example, Eccleston et al. (2015) found some evidence for the effectiveness of family problem-solving therapy on parent adaptive behavior, but limited overall evidence for the ability of psychological intervention to impact parental behavior or mental health.

Similar to findings reported by Stinson et al. (2009), most studies also included some form of contact with an interventionist. With an interventionist component, even if the role is greatly reduced, some of the challenges with in-person interventions likely still exist (e.g., scheduling based on family and clinician availability). This was apparent in several of the included studies where high attrition or recruitment challenges occurred, even preventing data analysis for one study team (i.e., Clarke et al., 2014). There is some suggestion in the family systems literature that the presence of an interventionist/coach positively impacts intervention retention and outcomes (Roddy, Nowlan, & Doss, 2017), although this remains an understudied area within pediatric psychology and eHealth. If eHealth interventions are not, in fact, improving accessibility to family-centered psychosocial care, this lends further support to the need to investigate other advantages of this intervention modality (e.g., cost effectiveness) and is also a call to action to work harder to eliminate barriers to accessibility.

The overwhelming inclusion of interventionist contact could also potentially speak to a broader discomfort in the field with removing all of the human interaction from a therapeutic intervention. While this hesitation is not inherently negative, it will be important for future studies to compare “pure” eHealth interventions with no interventionist to interventions with some “live” component in order to begin to understand the potential role of interventionists in eHealth intervention delivery. This type of comparison will lead to a deeper understanding of the potential impact of intervention design on important psychosocial outcomes. By better understanding the potential role of the interventionist as a potential “active ingredient” in effective eHealth interventions, future researchers will be able to develop interventions with maximal impact on family functioning and clinical care.

In addition to future work identifying the core components of effective eHealth interventions, future researchers should consider including additional outcomes, such as cost effectiveness, which may make a strong case for eHealth intervention delivery even if family-level outcomes do not differ significantly from those reported in effective in-person interventions. With the identification of “active ingredients” and a clearer understanding of factors that predict success in eHealth interventions, novel intervention designs such as SMART trials may provide a structured model for using decision points to triage families into eHealth vs. in-person interventions (Collins, Collins, Murphy, & Strecher, 2007; Noser, Cushing, McGrady, Amaro, & Huffhines, 2017). Another potential area for future research is person preference studies, where participants would select whether they would like to receive an in-person or eHealth/mHealth intervention. This would help to determine whether these interventions are effective for individuals who select this modality and, in theory, might not otherwise receive a desired or necessary service. eHealth/in-person “hybrid” designs, where technology might be utilized to include family members who are geographically or otherwise unable to participate in person, present another interesting possibility for future research. By actively increasing the flexibility of eHealth/mHealth interventions, as opposed to assuming that they are inherently more flexible by nature of relying on technology, the impact on family-centered processes might become more robust.

Like all studies, this systematic review should be considered in the context of its limitations. Given the variability among selected target outcomes and measurement of these outcomes, this review is primarily descriptive and does not include a meta-analytic component. Unpublished studies were not included, which raises concern for publication bias. In addition to potential non-publication of null findings, it is possible that new and exciting work that could change conclusions exists but has not yet been published (e.g., doctoral dissertations). This review was also limited to studies with a chronic-illness focus; family-centered studies with broader focuses (e.g., health promotion or prevention programs in the general population) may also provide important guidance for future work in pediatric psychology.

The current study provides a synthesis of the literature related to family systems eHealth interventions for children with chronic illnesses and their families. While many of the included studies reported encouraging results in at least one target outcome, there was great variability between studies with regard to selected outcomes and robustness of findings. As society continues to move rapidly towards greater reliance on technology in all areas, it will be important for the field of family psychology to gain a better understanding of the processes underlying the development of successful eHealth interventions to ensure the design and deployment of effective and efficacious family systems interventions.