Abstract
Objective:
Present a systematic analysis of the outcome research concerning video feedback (VF) programs.
Method:
Twenty-nine studies published between 1990 and 2014 were examined. They focused on children 0–12 years old and had at least one control group.
Results:
VF programs were similar in design, with interventions focusing on parental and child behavior, parental sensitivity and attachment. In 41% of the measurements, the program effects were moderate or large in favor of the intervention groups, particularly regarding maternal sensitivity and children’s behavior. Components of the VF programs were also examined. Two general problems emerged: lack of transparency/specificity of many programs with respect to the components, and heterogeneity among instruments used for measurement.
Conclusions:
Future research should focus on articulating intervention components and a standardized approach to measurement. This would facilitate comparisons of approaches and increase the possibility of implementing VF programs with fidelity in different professional settings.
Video feedback (VF), as a means to support parenting and healthy child development, has become increasingly popular and widely used. Common to all VF programs is that parents are filmed taking part in interactions with their children and are then invited to watch and reflect on the video recording under the guidance of a therapist. The therapist’s reflections will differ depending on the theoretical base of the program. Some programs will focus more on behavior (parents and/or children; Fukkink, 2008) while others focus on parental sensitivity and attachment (Kennedy, Landor, & Todd, 2011; Klein Velderman, Bakermans-Kranenburg, Juffer, & van IJzendoorn, 2006).
Interventions have been developed that are designed as stand-alone approaches; other multicomponent interventions include VF as one of a number of therapeutic strategies. There is an emerging body of scientific knowledge about the effectiveness of VF from evaluation studies and one meta-analysis focusing mainly on behavioral effects from VF intervention (Fukkink, 2008). The current review aims to be an update research in this field. While it covers some of the same areas as Fukkink’s study, caregiver sensitivity and parent–child attachment are also examined (for convenience, the word “parent” is used throughout this article, even when other caregivers may be involved). This review includes only experimental studies having at least one control group.
VF originates from divergent perspectives and often crosses theoretical boundaries. The practice of VF can take place in a wide range of contexts, such as pediatrics, child psychiatry, social services, nurseries, and schools. The specificity of the VF programs described here also varies, with some described sparingly and others based on detailed manuals. Programs also vary with respect to intensity, duration, and content of training, supervision, and certification procedures. Despite the differences, the common factors in most VF programs are that video recording occurs in everyday situations, and that video recordings are primarily used in order to strengthen the supportive interaction between parents and children.
VF
Filming is an ideal means for observing interaction and was used as far back as the 1940s (van der Horst, 2011). In the 1970s, microanalysis (the detailed analysis of very short film sequences) was introduced as a tool for understanding mother–infant interaction (Brazelton, Koslowski, & Main, 1974; Bullowa, 1979; Stern, 1971; Trevarthen, 1979). The theory of child development at that time began to shift from a model of “one-way influence” to that of a transactional exchange model in which parent and child reciprocally influence each other (Hedenbro, Shapiro, & Gottman, 2006). As analog and digital video recording became more widely accessible, VF programs have developed and become more sophisticated (Juffer, Bakermans-Kranenburg, & van Ijzendoorn, 2008; Juffer & Steele, 2014; Kennedy et al., 2011; Steele et al., 2014; Wirtberg, Petitt, & Axberg, 2013).
The general purpose of VF is to help parents increase confidence and competence in their parenting role by providing specific and constructive feedback (Wirtberg et al., 2013). This allows them to become better equipped to meet the needs of the child in everyday interaction. No matter which specific VF program is used, the therapist will normally video record in everyday situations such as feeding, playing, doing homework, and so on (Note: we use the term therapist to denote those in studies we reviewed who led the VF intervention, even though both training and context varied in these studies). Common foci for therapists in VF programs are parental skills, parents’ and children’s behaviors, parental sensitivity (a parents’ sensitive responsiveness to the child’s needs), and attachment.
The VF session is often carefully prepared, and the video recording can be used in different ways, from showing global video sequences to using a microanalytic approach (Steele et al., 2014). The video sequences used in the VF session can vary from 15 to 30 s (Schechter et al., 2006) to 15 min or more (Benoit, Madigan, Lecce, Shea, & Goldberg, 2001). In preparation for the VF session, the therapist typically has identified examples of interaction that support the child’s development. The video or video sequences provide the therapist with the means to be very specific and concrete while presenting information, and the video recording can be stopped at any point in order to discuss and reflect on what has been seen and heard. Different techniques can be used during the VF session; for example, the parents’ attention can be directed to a specific sequence in which the child takes an initiative followed by the parents’ response to it (Fukkink, 2008). Other possibilities include using still photos and slow motion (Fisher, Frenkel, Noll, Berry, & Yockelson, 2016; Fukkink, 2008; Juffer & Steele, 2014; Steele et al., 2014).
Viewing the video sequences together provides a simultaneous, common reference point for parents and therapists, since they are seeing and describing the same course of events “in real time.” The therapist serves as an intermediary between the professional view of the video recorded interaction and the parent’s own views (Häggman-Laitila, Pietilä, Friis, & Vehviläinen-Julkunen, 2003; Stein et al., 2006). It is expected that watching and reflecting on the video recording will influence and alter not only the parents’ internalized idea of their child, but also of themselves and their interaction with their child. This might encourage and stimulate parents to reflect on their own parenting behavior, and increase their understanding of and sensitivity to the needs of the child (Dowrick, 1999; Fukkink, 2008; Juffer & Steele, 2014; Steele et al., 2014).
Previous Research
There is one meta-analysis that focused specifically on the effect of VF interventions (Fukkink, 2008). This work included studies having designs both with and without control groups, and others in which the focus was particularly on parents’ or children’s changes of behavior or on parental attitudes (Fukkink, 2008). The meta-analysis focused on 29 VF intervention studies (N = 1,844 families) and presented aggregated effects for the domain of parental behavior (Cohen’s d = .47), for parental attitude (Cohen’s d = .37), and for the effect on children’s behavior (Cohen’s d = .33; Fukkink, 2008). The inclusion age was from 0 to 8 years old (average age was 2.3 years). Five studies concerned a clinical parent population, while the children came from a clinical child population in 15 of 29 studies. Risk factors were often present at the parental level (Fukkink, 2008).
Fukkink’s study affirmed previous research, showing that VF interventions that were effective at the behavioral level did not necessarily lead to changes in attachment patterns or changes in insecure mental representations of the parents involved (Fukkink, 2008; Juffer et al., 2008; van Ijzendoorn, 1995; van Ijzendoorn, Juffer, & Duyvesteyn, 1995). The meta-analysis also suggested that programs with shorter duration (in terms of weeks) were more effective at improving parenting skills, a finding that affirmed earlier studies (Bakermans-Kranenburg, van Ijzendoorn, & Juffer, 2003; van Ijzendoorn, 1995). One conclusion Fukkink drew from his study was a “short but powerful” hypothesis: Interventions of a short duration are, on average, more effective. Another conclusion was that because VF often was combined with other components, it could not be demonstrated that VF was the crucial component.
Another relevant meta-analysis focused on interventions for parental sensitivity and attachment. In this meta-analysis, VF was an intervention in 15 out of the 70 studies reviewed (Bakermans-Kranenburg et al., 2003). The most effective interventions consisted of a short number of sessions, following the idea of “less is more.” Typical interventions consisted of six sessions during 6–8 weeks, and had a behavioral focus. Some of the families had multiple problems, including parenting difficulties, parental mental health issues, and economic or housing problems. The meta-analysis indicated that 54 of the 70 studies were successful in enhancing positive parenting behavior and that those programs that contained VF interventions were more effective (Cohen’s d = .44) than interventions without them (Cohen’s d = .31). The results indicated only small effects in changing attachment patterns (Cohen’s d = .20); thus the findings indicated that the enhancement of parental sensitivity might lead to positive changes in child behavior other than attachment security (Bakermans-Kranenburg et al., 2003).
Additional information about the effectiveness of VF can be derived from a review of early interventions (Stewart-Brown & Schrader McMillan, 2010), which stressed parenting support in infancy and early years. This review concluded that short sensitivity focused interventions starting at around 6 months of age in high-risk infants was recommended (Stewart-Brown & Schrader McMillan, 2010). It was reported that these kinds of programs, which often include VF, offer interactional guidance, enhance parental observation skills, and increase positive interchanges and enjoyment.
Purpose
The purpose of this article is to describe and examine the findings of trials evaluating VF programs in studies published between 1990 and 2014. This review includes only studies using experimental or quasi-experimental methodology, having randomized controlled trials (RCT) as the dominant design. The aim is to provide an overview of the components in VF programs as well as to synopsize the outcomes of the studies. The measures used in the studies will be described, as they indicate the focus of the various studies. Compared to earlier, similar studies, this study is more restrictive, requiring rigorous methods. The starting point is from a broader perspective than prior VF review papers as it also focuses on supplemental components in the VF programs, as well as on the measures used for the expected outcome.
The central questions are: What aspects were focused upon for measurement and what outcomes were presented from the included VF programs? What were the components of these VF programs?
Method
Search Strategies
The relevant published studies were collected systematically, using two various strategies: first by using electronic databases, and second by studying the references in the included studies (snowball method). In order to identify studies that were published in peer-reviewed journals, the main search was carried out using three databases (PubMed, Psych NET, and Social Sciences Citation Index [Institute for Scientific Information]). To ensure that no studies were missed, the more inclusive Google Scholar was used as well in order to search directly for program names that were not captured by the databases. Additional sources were gained from the publications of Bakermans-Kranenburg, van Ijzendoorn, and Juffer (2003), Fukkink (2008), Juffer, Bakermans-Kranenburg, and van Ijzendoorn (2008), and Kennedy, Landor, and Todd (2011). The final search was performed in August 2014. One researcher was mainly responsible for searching the databases (S.B.) and an additional researcher (I.W.) thereafter read the studies independently. To qualify the studies for this systematic review, inclusion criteria were used. For this review, the inclusion criteria are described below.
Inclusion Criteria
The study includes individual VF intervention that concerns parent and child interaction, having one specific child in focus. The study is published in English between the years 1990 and 2014. The child’s age in the study is between 1 month and 12 years old. The study reports at least one quantitative measure, a rating scale, or an observation method before and after treatment. The study includes at least two groups of participants (experiment group and control group).
Search Result
Included studies
In the search procedure, relevant terms were used and combined in the search profile (e.g., VF, intervention, children, maternal sensitivity, attachment, behavior, childcare, caregiver, teacher, school based). Studies were chosen if they included an evaluation of an intervention that made use of VF, in which one specific child was the focus. The search resulted in 29 studies, 21 RCT, and 8 quasi-experimental studies (all together 32 publications), which in turn included various VF programs, several of which were untitled. The programs are presented in Table 1.
Presentation of the VF Programs.
Notes. VF = video feedback; S = sensitivity; A = attachment; B = behavior; [—] = information is missing; VLBW = very low birth weight; number of VF sessions and session length comprise sessions where VF was used; focus of study = the research focus of the articles; VIPP = video feedback Intervention to promote positive parenting; VIPP-SD = video feedback Intervention to promote positive parenting–sensitive discipline; VIPP-R = video feedback intervention to promote parenting–representationally focused; VIPP-AUTI = video feedback intervention to promote positive parenting–adapted to autism; VIPP-TM = video feedback intervention to promote positive parenting–Turkish minorities; PALS = playing and learning strategies; VIP = video interaction project; IG = interaction guidance; VHT/Orion = video home training; VIG = video interaction guidance; RFI = relationship focused intervention.
Excluded studies
VF studies published in peer-reviewed journals were excluded from this review for a number of reasons. Nonexperimental studies were excluded due to, for example, lack of a control group with which to compare the treatment results, or having a qualitative approach (Beebe, 2003, 2010; Cramer et al., 1990; Damen, Kef, Worm, Janssen, & Schuengel, 2011; Elder, Valcante, Yarandi, White, & Elder, 2005; Häggman-Laitila et al., 2003; Moss et al., 2014; Neander & Engström, 2009; Phaneuf & McIntyre, 2011; Schechter et al., 2006; Sharry, Guerin, Griffin, & Drumm, 2005; Vik & Hafting, 2006, 2009; Zelenko & Benham, 2000). One study was excluded since VF had a less prominent role compared to other strategies in the intervention (Bernard et al., 2012). One study was unclear in terms of design and was excluded for that reason (Janssens & Kemper, 1996), and one study had a control group, but presented a measure point for only one of the groups (Kenny, Conroy, Pariante, Seneviratne, & Pawlby, 2013). Some studies did not meet the inclusion criteria for other reasons, for example, pilot studies (Lawrence, Davies, & Ramchandani, 2013; Santos, Feliciano, & Agra, 2011) and studies focusing on group interventions (Egeland, Weinfield, Bosquet, & Cheng, 2000; Fukkink & Tavecchio, 2010; Glanemann, Reichmuth, Matulat, & am Zehnhoff-Dinnesen, 2013; Groeneveld, Vermeer, van Ijzendoorn, & Linting, 2011; Hoffman, Marvin, Cooper, & Powell, 2006; Kemenoff, Worchel, Prevatt, & Willson, 1995; Marvin, Cooper, Hoffman, & Powell, 2002).
A few studies were followed by publications not relevant in this context—for example, focusing on design of component or on process or timing of intervention—and were excluded for that reason (Benzies, Magill-Evans, Harrison, MacPhail, & Kimak, 2008; Hodes, Meppelder, Schuengel, & Kef, 2014; Landry, Smith, Swank, & Guttentag, 2008; Meade, Dozier, & Bernard, 2014; Mendelsohn et al., 2007; Stolk et al., 2008). A number of other studies or publications were also excluded—mainly descriptive and based on consumer satisfaction, published as work reports, academic theses, or published in nonscientific journals.
Coding the Studies
For coding the studies that were included in this review, a number of characteristics that were of interest were carefully examined, that is research designs, measures, the parent and child populations, components, foci, and details about the performance of the interventions (Tables 1 and 2). For each of these studies, the outcome variables were coded into three categories: maternal sensitivity (S), parents’ behavior and/or children’s behavior (B), and attachment representations (A). There are different dimensions of attachment, for example, secure/insecure attachment and organized/disorganized attachment, and it is therefore urgent to emphasize the rough measure of this subdivision. This categorization was made according to what measures were chosen in the various studies. The process of categorization was somewhat difficult as there was not always a clear distinction between the three categories. Nor was it possible for all measures to fit into these categories (e.g., cortisol levels) which is why the categories does not include all the results even if they are presented in Table S1 of the Online Appendix A. For further information about the measures and outcome variables, see Table S2 in the Online Appendix B. It was sometimes difficult to interpret the detailed content of the studies, and hence we apologize for possible errors.
Components of VF Programs.
Notes. VF = video feedback; VIPP = video feedback intervention to promote positive parenting; PALS = playing and learning strategies; VIP = video interaction project; VHT/VIG = video home training/video interaction guidance; IG = interaction guidance; RFI = relationship focused intervention.
Statistical Methods
To compare the outcomes, we calculated the effect size (Cohen’s d) which is a well-established measure of the pooled standardized difference between two means (Cohen, 1988). First, we calculated a time effect (the reported mean in one outcome before and after the intervention, using two measure points). Second, we calculated the program effect (the difference between the mean outcome from the experimental and the control intervention at the second measurement point). According to the guidelines of Cohen (1988), effect sizes was defined as small, d = .20, moderate, d = .50, or large, d = .80. Outcomes are compiled in Table S2 of the Online Appendix B. Both the time effect and the program effect are presented in the result (if available), since not all the studies used randomization to equalize the first measure point before the intervention. Third, a χ2 test was used to examine if the magnitude of the time effects between the intervention group and the control group were different.
Results
Included Studies
First, the focal population is presented in Table 1, followed by information about the number of sessions, session length, and duration, as well as the different foci of the studies. The 29 studies included are numbered 1–29 in tables; number 1–21 was RCTs, and number 22–29 was quasi-experimental studies. A total sample of 2,713 families was examined. Since the video feedback intervention to promote positive parenting (VIPP) program was the most frequent program, those studies are presented first, ranked by sample size. Thereafter, VF programs with various names and VF programs with no program name are presented, also ranked by sample size. Altogether, eight 1 specific titled VF programs and eleven untitled VF programs met the inclusion criteria. The included studies varied in design, sample groups, targeting, and training for therapists. Populations and components in the VF programs are presented in Tables 1 and 2 and in the text below.
Focal Population
The children and parents in the focal population differed in categorization, either in terms of problems or civil status, linked to the child or parent or both (Table 1). A clinical population—for example, children exhibiting externalizing behavior, very low birth weight, or developmental disabilities—was examined in 14 of the 29 studies. Some populations consisted of parents from a clinical population—for example, parents suffering from eating disorders or depression (3 studies)—while in other studies, parents were chosen from different risk groups—for example, mothers who suffered from poverty or severely deprived families, families reported for maltreatment, teenage mothers, insecurely attached parents, or parents having a lack of sensitivity (10 studies). The studies examined the mothers’ interaction with their children, apart from three studies also examining the fathers’ skills in interactions with their infants (Axberg, Hansson, Broberg, & Wirtberg, 2006; Magill-Evans, Harrison, Benzies, Gierl, & Kimak, 2007; Moss et al., 2011). In one study, both parents and teachers were video recorded interacting with a child (Axberg et al., 2006). In about half of the studies, the included children were infants, between 1 month and 1 year old (15 studies). Some studies included toddlers, mostly between 1 and 3 years old (11 studies), while some included preadolescents, between 4 and 12 years old (3 studies).
Sessions and Treatment Duration
The numbers of sessions varied between 1 and 15 sessions (with a mean of 6 sessions and median of 6 sessions) and were mostly predetermined in quantity. The length of the sessions varied from 40 to 180 min (with an average of 79 min and a median of 90 min) and was over the course of 1–40 weeks (with an average of 13 weeks and a median of 12 weeks, consecutively).
Focus of the Studies
All included VF programs used video recordings of the interactions between a parent and one specific child to promote improvement in their relationship. Three categories were identified in the included studies, which served as a base for coding. To enhance maternal sensitivity (S). To improve parents’ or children’s behavior (B). To strengthen secure attachment representations, from insecure attachment to secure or from disorganized attachment to organized (A).
The categorization, made from the measurement instruments used in the studies and the outcome variables, indicated that there was more than one focus in more than half of the studies. The most common focus was on children’s and/or parent’s behavior, and was found in 12 studies. A combination of parental sensitivity and behavior was present in nine studies. A mix of behavioral, sensitivity, and attachment foci was found in four studies, while a focus on sensitivity and attachment was found in three studies. Having attachment as the only focus occurred in one study (Table 1).
Components in VF Programs
Since the particular details of the VF programs were occasionally not described in the published studies, it was not always clear which components were specific to each VF program. Further, in some articles the structure and method of the particular program was not disclosed; it was nonspecific concerning video recording analysis and how the feedback sessions were structured. In some publications, information was lacking concerning how or if the video recording was edited, and whether or not the themes of the video or video sequences that were used for feedback were predetermined. However, in some articles, this was clearly described. In this attempt to examine similarities between the VF programs, this lack of transparency must be kept in mind—nevertheless some patterns appeared regarding the components (Table 2).
Common to all VF programs was that the parent–child interaction was video recorded. In all VF programs, a typical day-to-day situation was video recorded, and most often nonstandardized video recordings were used (compared to standardized video recordings where the situation is predetermined—for example, to play with a specific toy for 2 min). The VF programs had clear targets for change (parental sensitivity, attachment, behavior, parental skills), which were fairly similar between the programs even if design and target groups differed. In three of the programs (Axberg et al., 2006; Juffer et al., 2008; Magill-Evans et al., 2007), a manual was used. In the VF sessions, either unedited or edited sequences from the video recordings were presented. A majority of the VF programs consisted of 1–7 VF sessions, while some consisted of 8–14 VF sessions. Most of the VF programs were home-based programs, which means that video recording and feedback were taking place in the family’s home. In sum, the results indicate that the described components are common in VF programs.
Theoretical Foundations
The theoretical framework, both in terms of content and clarity, varied across studies; sometimes a study referred to more than one theory, while in some publications this was neither clearly described nor referred to. Nor was there always a pronounced link between program theory and the outcome measures. The dominant theories were attachment theory (Bowlby, 1969, 1988), in more than two-thirds of the publications, and social learning theory (Bandura, 1977) or coercion theory (Patterson, 1982), in almost one-third of the publications. Since a common goal was behavior change, social learning theory and coercion theory were often referred to in combination with attachment theory. Other theories referred to were developmental, psychodynamic, neurobiological, communication, stress regulation, and systemic theory. The theoretical approach seemed to vary according to the focal population and the kind of problem being addressed.
Designs of the included studies
VF was presented as the main intervention in most of the studies, and is labeled the experimental intervention in this review (Table S1, Online Appendix A). However, there were six exceptions: VF as one of a four-part program (Brisch, Bechinger, Betzler, & Heinemann, 2003); VF combined with another component or intervention such as group instructions (Kim & Mahoney, 2005); a book (Juffer, Bakermans-Kranenburg, & van IJzendoorn, 2005; Juffer, Hoksbergen, Riksen-Walraven, & Kohnstamm, 1997); coordination meetings (Axberg et al., 2006); cognitive behavior self-help for eating disorders (Stein et al., 2006); baby massage, cognitive restricting, or modeling (van Doesum, Riksen-Walraven, Hosman, & Hoefnagels, 2008); and discussion (Bakermans-Kranenburg, Juffer, & van Ijzendoorn, 1998; Klein Velderman, Bakermans-Kranenburg, Juffer, & van IJzendoorn, 2006; Klein Velderman, Bakermans-Kranenburg, Juffer, van Ijzendoorn, Mangelsdorf, et al., 2006; Moss et al., 2011).
VF was most often compared to treatment as usual (TAU). Sometimes TAU was briefly explained, for example, telephone calls or home visits, but usually there was no more information. In other cases, VF was compared with other interventions such as “psychodynamic therapy” (Robert-Tissot et al., 1996), “feeding focused intervention” (Benoit et al., 2001), “supportive counseling” (Stein et al., 2006), “developmental assessment screening” (DAS; Landry, Smith, & Swank, 2006), and “attachment discussions” (Bakermans-Kranenburg et al., 1998; Klein Velderman, Bakermans-Kranenburg, Juffer, & van IJzendoorn, 2006; Klein Velderman, Bakermans-Kranenburg, Juffer, van Ijzendoorn, Mangelsdorf, et al., 2006). In one study, the intervention Family Check Up (FCU) was compared with FCU plus VF (Smith, Dishion, Moore, Shaw, & Wilson, 2013). A follow-up after posttest was made in 11 of the 29 studies (For more information, please refer to the individual studies). Some were performed 1–6 months after posttest, while some studies had follow-ups 1–3 years after posttest (Axberg et al., 2006; Bakermans-Kranenburg, van IJzendoorn, Pijlman, Mesman, & Juffer, 2008; Brisch et al., 2003; Eliëns, 2005; Klein Velderman., Bakermans-Kranenburg, Juffer, & van IJzendoorn, 2006; Klein Velderman, Bakermans-Kranenburg, Juffer, van Ijzendoorn, Mangelsdorf, et al., 2006; Mendelsohn et al., 2007).
The studies in this review all varied regarding, for example, sample size, attrition, and treatment integrity. Sample sizes ranged from 18 to 264, in which 7 of the 29 studies had a large total sample (more than 120), and 7 of the 29 studies had samples of fewer than 50 children. Furthermore, the sample was divided differently: into one experimental group and one control group; into one experimental group and two control groups; and sometimes into two or three experimental groups.
Measures
The studies all together presented outcomes from a broad selection of more than 50 structured measures (Table S2, Online Appendix B), although results were not always presented as numerical data. Numerical data was available that made it possible to calculate or present effect differences in 25 of the 29 studies (Table S2, Online Appendix B). The most common measures were Ainsworth’s Rating Scales for Sensitivity (Ainsworth, Bell, & Stayton, 1974) that focused on parental sensitivity, the Child Behavior Checklist (Achenbach & Rescorla, 2000, 2001) that focused on children’s behavioral and/or emotional problems, strange situation procedure (Salter Ainsworth, Blehar, Waters, & Wall, 1982) that focused on attachment, and Parent Emotional Availability Scales (Biringen, Robinson, & Emde, 2000) that focused on parental emotional responsiveness.
Measures were employed in a variety of different ways across studies. When focusing on the parents, measures were used to examine sensitivity or behavior. Measures focusing on the child emphasized mainly behavior. When the study groups involved preadolescents (ages 4–12 years), measures that focused on behavior were used exclusively. Furthermore, the measures sometimes covered several aspects. For example, “externalizing behavior” could include oppositional, aggressive, and overactive behavior (van Zeijl et al., 2006), “sensitivity” could include maternal responsiveness and the child’s social cooperation (Landry et al., 2006), and “attachment” could include daily hassles, depression and attachment security (Kalinauskiene et al., 2009).
General Outcomes From the Studies
A general conclusion from this systematic review is that the VF programs we examined do enhance positive parent–child interaction. The outcomes of the studies are presented in summary below, and in detail in Table S2 in the Online Appendix B. All together the studies present a disparate effect difference from small to large. Generally, the effect differences were small to moderate, though some studies presented somewhat larger effect differences (Bakermans-Kranenburg et al., 1998; Benoit et al., 2001; Juffer et al., 2005, 1997; Kalinauskiene et al., 2009; Kim & Mahoney, 2005; Koniak-Griffin, Vezemnieks, & Cahill, 1992; Landry et al., 2006; Weiner, Kuppermintz, & Guttmann, 1994).
The largest changes were observed in the domains of maternal sensitivity as well as in parents’ and children’s behavior. Effect differences in maternal sensitivity were small to large, with an emphasis on moderate effect differences (Bakermans-Kranenburg et al., 1998; Juffer et al., 2005, 1997; Kalinauskiene et al., 2009; Landry et al., 2006; Magill-Evans et al., 2007; Negrão, Pereira, Soares, & Mesman, 2014; Poslawsky et al., 2014; Robert-Tissot et al., 1996; van Doesum et al., 2008; Yagmur, Mesman, Malda, Bakermans-Kranenburg, & Ekmekci, 2014). Effect differences in the parents’ and/or the child’s behavior showed small to large effect difference with an emphasis on small and moderate effect differences (Axberg et al., 2006; Benoit et al., 2001; Bilszta, Buist, Wang, & Zulkefli, 2012; Eliëns, 2005; Juffer et al., 1997; Kim & Mahoney, 2005; Koniak-Griffin et al., 1992; Mendelsohn et al., 2007; Negrão et al., 2014; Poslawsky et al., 2014; Seifer, Clark, & Sameroff, 1991; Svanberg, Mennet, & Spieker, 2010; van Doesum et al., 2008; van Zeijl et al., 2006; Weiner et al., 1994; Yagmur et al., 2014). Effect differences concerning attachment (e.g., the rate of disorganized attachment, or increase in attachment security) were less well documented and were found in VIPP studies only, presenting small to large effect differences (Juffer et al., 2005; Kalinauskiene et al., 2009; Klein Velderman, Bakermans-Kranenburg, Juffer, & van IJzendoorn, 2006).
The results indicate that most of the included VF programs did present an overall positive outcome regarding both parents’ behavior and children’s behavior (except for interaction guidance [IG] presenting no results for children’s behavior, and Marte Meo presenting no results for parents’ behavior). Additionally, some studies presented a positive outcome regarding parental sensitivity and attachment. Outcomes regarding attachment were presented only in the VIPP studies, as VIPP was the only attachment-based program. VIPP studies also presented positive outcomes regarding maternal sensitivity. However, outcomes regarding maternal sensitivity—often connected to attachment—were also presented in other VF programs. The playing and learning strategies study presented outcomes on maternal sensitivity as well as parents’ and children’s behavior. The IG studies presented outcomes in maternal sensitivity and in parents’ behavior. The studies of untitled VF programs presented outcomes in both maternal sensitivity and in parents’ and children’s behavior, generally with more focus on behavior than on sensitivity. The VIP study presented outcomes from children’s behavior, but no outcome concerning either maternal sensitivity or attachment was presented. The video home training/video interaction guidance studies presented outcomes in children’s behavior. The Marte Meo study presented outcomes in children’s behavior and the relationship focused intervention study presented outcomes both in parents’ and children’s behavior.
To see the distribution of time effects and program effects presented in the studies, these are presented here in summary. According to the time effect (the reported mean in one outcome before and after the intervention, using two measure points), just over half (51%) of the measurements in the intervention groups show moderate or large effect, while almost half (47%) of the measurements in the control groups show no effect at all. The differences in time effect between the two groups are significant (χ2 = 23.82; df = 3; p ≤ .001) in favor of the intervention groups. In terms of program effect (the difference between the intervention group and the control group at the second measurement point, see Table S2 in the Online Appendix B), the measurements show moderate or large effect differences in 41% in favor of the intervention groups. Altogether, the results indicate that VF enhances parent–child interaction.
Discussion and Applications to Practice
In the VF programs we reviewed, the interaction between parent and child was addressed and common themes were parental skills, parents’ and children’s behavior, parental sensitivity, and attachment patterns (Table S2, Online Appendix B). Despite disparate program theories, measures, and target groups, the results seem consistent in terms of the overall program effect of VF. The largest changes were found within maternal sensitivity as well as in parents’ and children’s behavior, which suggests that VF programs can provide changes within these domains. However, owing to a lack of detail in the studies we reviewed, we cannot draw any conclusion as to which groups or for which objectives VF is found to be most effective.
The results from the studies also raise questions regarding research design. Studies employed disparate designs, measures, and procedures, and also varied in terms of problems and vulnerabilities in the populations. This may have affected the results. Whether the study was efficacy or a clinical study may also have affected the results. The results might also be affected if it is the child or the parent who are the subject (Fukkink, 2008).
The large variety of measures presented in the various studies point to a need for further attention and discussion concerning the choice of measures. Transparency would be preferable regarding reasons why a particular measure is chosen. The variety of measures might complicate comparison between different VF programs.
When examining the various included VF publications, there was generally a distinct lack of information on the specifics of the VF program, such as how recording, analysis, editing, and feedback were performed. In three of the VF programs, a manual was attached that made the procedures accessible and transparent. For VF programs to become more transparent and accessible, it is important to clarify the purpose of video recording, and the various steps in the program. Thus, some important questions emerge regarding VF programs: (a) What is the aim of video recording? (b) Is the video recording standardized or not? (c) Is the video recording edited into brief video sequences or does it consist of longer sequences of unedited footage? (d) Is there specific criteria for analysis of the video recording? (e) Are there criteria for how feedback to parent should be delivered? and (f) Is there a difference in outcome of treatment in regard to the variation of training, supervision, and certification procedures? These comments are consistent with Fukkink’s (2008) conclusion that manualizing VF interventions is an important goal for the field.
As indicated in Table 2, there are similar components in the VF programs, but there are also variations. Duration is one of the components, and a subject of ongoing discussion (Bakermans-Kranenburg et al., 2003). Short programs (around six sessions) have been reported to be more effective in improving parenting skills than longer programs (Bakermans-Kranenburg et al., 2003; Fukkink, 2008; Stewart-Brown & Schrader McMillan, 2010; van Ijzendoorn et al., 1995). This review supports that conclusion. However, it cannot be ruled out that the number of sessions and duration of treatment, and whether these were predetermined, could have influenced the results. In the most vulnerable families, the VF programs can be effective in some domains (behavioral), but may need to be complemented with other types of support in order to meet the needs of these families (Fukkink, 2008; Stewart-Brown & Schrader McMillan, 2010).
Given the range of techniques and approaches that are grouped within the general area of VF programs, a number of directions for future research may help to achieve progress in the field. Many of the VF programs that have been described in the literature, though grounded in theories such as attachment theory and social learning theory, are not based on a well-defined theory of change (i.e., a measurable conceptual model). As such, it is not clear what underlying specific processes in the parent, child, or relationship are hypothesized to be mediators of change in specific outcomes observed within the studies. Parenting behavior is assumed to be the primary determinants of child behavioral outcomes, but the specific targets of parenting are often both unspecified and not measured in the VF programs and evaluations. Moreover, insufficient attention has been given to understanding variations in response to specific VF approaches. A focus on moderators of intervention effects (such as level of family adversity, maternal depression, and child cognitive ability) will help to increase the extent to which VF interventions could effectively reach more individuals.
There is clearly a need for further research regarding the different strategies and their effects in VF programs, for example, location (home-based treatment vs. clinic-based treatment), nature of home practice assignments, dosage, edited video or raw footage of recordings, and different models used for video analysis. Knowledge about the impact of watching oneself on video recording is limited (the main component). Watching video recordings of oneself engaged in parenting might increase key underlying cognitive skills such as the areas of executive functioning that include inhibitory control and working memory (Fisher, Mannering, van Scoyoc, & Graham, 2013; Steele et al., 2014).
It is interesting to note that a majority of the publications reflect a long tradition of studying mothers as the main attachment person. There are only two exceptions (Axberg et al., 2006; Magill-Evans et al., 2007). This is somewhat surprising, as research has indicated that children have the ability to attach to more than one person and also benefit from that (Bretherton, 2010; Hedenbro et al., 2006; McHale & Fivaz-Depeursinge, 1999), a position also stressed by Bowlby (1969). In the future, there will hopefully be more research involving fathers’ interaction with their children, which, of course, presupposes fathers being involved in interventions concerning their children.
Another area for future research has to do with the ability of VF programs to be scaled. As is true in many areas of evidence-based practice, the utility of VF programs—even if controlled research has shown them to be effective—is only as good as their ability to be widely implemented with fidelity in real-world community settings (Fixsen, Naoom, Blase, Friedman, & Wallace, 2005; Mihalic & Irwin, 2003).
One question of importance to the topic of VF programs is what research in this area might add to the landscape of effective approaches to support children (families). VF, although currently popular, would need to demonstrate results equivalent to or better than existing interventions in order for its use to be justified. VF might achieve more impact with high-risk or vulnerable families. The families with the lowest skilled caregivers and the greatest number of risk factors (drug abuse, low income, domestic violence) may have difficulty on integrating the material being presented in traditional intervention programs that focus on providing information about effective parenting skills but do not involve an active skill-building component. VF may increase the saliency of the parenting behavior being addressed, having access to both visual and audial information simultaneously. Although only future research will help to clarify whether the advantages of VF are robust, we may see that public policy toward high-risk children and families increasingly embraces VF programs.
Methodological Considerations
There were methodological issues among the studies included in this review that may have influenced the results. Searching the studies from databases using specific keywords always involves a risk of missing important studies. In addition, the search was completed in August 2014 and new studies might have been added. The inclusion criteria could have contributed to interesting studies within the field being excluded, and we are aware that this could have led to a limitation since important findings might have been missed. The aim was to capture the causal effect of VF programs, which are best captured by RCT’s. In this review, experimental studies have been focused on. This does not preclude nonexperimental studies being significant, even if they were not included.
The studies were often based on different theories and hence different methodologies. The studies had VF as the main intervention, although VF was sometimes combined with another program or intervention. This could imply that VF could be an extraneous variable that in itself is not necessary or sufficient to produce the obtained effects. This made it difficult to know for certain what specifically influenced the results. The samples in the studies differed from small to large, which could also influence the conclusions. In a few publications, it was not clearly described how pre- and posttest were performed, which sometimes made it difficult to calculate effect differences. As the outcomes in this review were presented in Cohen’s d (Cohen, 1988) outcomes from publications presenting results in other outcome measures were therefore referred to in text. The studies will be found in the list of references.
In intervention studies, it is important to pay attention to the concept of “TAU.” In the included studies, control groups were often defined as TAU. One methodological consideration is that knowledge about the content of TAU is limited. Since the contents of TAU can differ considerably, this in turn may have affected the results (Löfholm, Olsson, Hansson, & Brännström, 2013). The interpretation of the results was influenced by the current categorization of different foci in the studies, which means that other ways of categorizing might have led to other conclusions. Finally, there are additional factors, which were not included in the studies reviewed, that could have influenced the result, for example, attrition and treatment fidelity/integrity in the included studies.
Conclusions
Overall, this systematic review of 29 experimental studies suggests that VF is effective in stimulating parents’ sensitivity and behavior in a positive manner, as well as causing a positive change in children’s behavior. It is hoped that changes in parent–child interaction will support the development of their children, and this systematic review has examined a number of experimental studies presenting different effects in VF programs. The results indicate that VF is effective at various ages, and in different settings, problem areas, and cultures. The current studies—and this systematic review—are important since they accentuate knowledge, reveal the shortcomings of current research, and illuminate the possibilities for future studies. Although the theoretical bases for VF programs differ, they appear to be fairly similar in overall design and content, for example, using VF to enhance existing positive parent–child interaction, and to highlight strengths and resources (Steele et al., 2014). However, questions remain about which components are most effective and what the overall common factors in VF programs are.
Supplemental Material
Supplemental Material, Table_3_Appendix_A_Revised_7_October - Video Feedback Intervention With Children: A Systematic Review
Supplemental Material, Table_3_Appendix_A_Revised_7_October for Video Feedback Intervention With Children: A Systematic Review by Stina Balldin, Philip A. Fisher, and Ingegerd Wirtberg in Research on Social Work Practice
Supplemental Material
Supplemental Material, Table_4_Appendix_B_Revised_7_October - Video Feedback Intervention With Children: A Systematic Review
Supplemental Material, Table_4_Appendix_B_Revised_7_October for Video Feedback Intervention With Children: A Systematic Review by Stina Balldin, Philip A. Fisher, and Ingegerd Wirtberg in Research on Social Work Practice
Footnotes
Acknowledgement
We would like to express our special gratitude in memoriam to Kjell Hansson, Professor Emeritus at School of Social Work, Lund University in Sweden, who was one of the initiators and co-authors of this review until he unexpectedly left us in 2013.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Supplemental Material
Supplemental material for this article is available online.
Note
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
