Assessing the effectiveness of parent–child interaction therapy with language delayed children: A clinical investigation

I Introduction

Clinicians who work with children have frequently found it useful to study a child’s relationship with their parents and to modify this during therapy. This approach, commonly referred to as parent–child interaction therapy (PCIT), is widely used with children with behavioural problems and with children and their families who have psychological difficulties. It is now increasingly used in speech and language therapy (SLT), especially in the treatment of children with developmental delays in the acquisition of language and children who are dysfluent. A common set of assumptions underlies the use of PCIT in these different contexts. First, that work with the child, particularly the young child, will be facilitated by the presence of a parent. Second, that the dynamics of the interaction between parent and child will reveal parental behaviours that, if modified, will benefit the child. Finally, that the continuing contact and relationship between parent and child will allow these benefits to generalize beyond the clinical setting.

Joint working with parents and their children with delayed language has taken two major forms. In North America the Hanen parent programme (Manolson, 1992) remains popular while in the UK the term PCIT is used to describe a variety of broadly similar programmes developed by different clinical services. Hanen is the more prescriptive approach and its consistent methodology has encouraged investigations of its effectiveness. Early studies by Girolametto (1988) and Tannock et al. (1992) assigned children and their parents randomly to treated and untreated groups and demonstrated effects of the treatment on parent–child interactions. In the first study, parents talked less and there was more turn taking in their interactions with the children after treatment and the children were more responsive and used a broader vocabulary. The second study found that treated parents became more child centred in their approach and developed better strategies for maintaining interactions.

Although the treated children in these studies interacted more appropriately with their parents, there was no direct evidence that their language had advanced. In short, the programme had improved the nature of the interaction but the impact of therapy on the quality of the children’s language remained unclear. This raises issues about the objectives of therapy programmes that are designed for children and parents. Should they try directly to change the children’s language or will changes in the interactions alone promote language development? That few changes in language itself were seen suggests the former; however, changes in language may require longer to emerge and the finding in Tannock et al. (1992) that parents had maintained their improved interaction skills when assessed again four months later favours the view that longer term change may result.

Proponents of the Hanen programme took the view that a more direct approach was needed and revised the programme accordingly (for details, see Pepper and Weitzman, 2004). The new version taught parents to use focused stimulation (Ellis Weismer and Robertson, 2006), which encourages them to provide numerous repetitions and models of a language target. In a randomized trial of the new approach, Girolametto et al. (1996) found that treated parents used focused stimulation more frequently as well as changing their interaction styles, and that the children used more multiword utterances than controls. Similar results were reported in Girolametto et al. (1998) in a study using the focused stimulation version of Hanen with children with Down syndrome.

Initial enthusiasm for the Hanen approach in the UK led to a number of pilot studies and descriptions of its use (see Awcock and Habgood, 1998; Coulter and Gallagher, 1998). In practice this enthusiasm has not been maintained. In a survey (Joffe et al., in preparation) which asked clinicians what therapies they use with children, few mentioned Hanen specifically whereas PCIT was among the most popular approaches. One reason for this may be that the full Hanen programme is demanding both for parents who are asked to attend for up to 30 hours and for the resources of therapy services. Baxendale and Hesketh (2003) compared the Hanen programme with traditional one to one therapy for receptive and expressive language conducted in clinics or in the children’s nurseries and with a parent attending. Seventy one percent of the children improved but no difference was found between the therapies. However, Hanen required twice as much clinician time as the traditional approach.

Given these results it is understandable that SLT services have looked for less demanding methods that preserve the advantages of working with parents. PCIT met this need. Kelman and Schneider (1994) were early advocates and describe the approach. The advantages of using video in therapy were described by Cummings and Hulme (1997). The use of video to show parents where their interactions can be improved has become central to the intervention. Further evidence for the approach is available from Gibbard et al. (2004). Children with expressive language delay were allocated to two groups. One received therapy following current practice; the other a parent based intervention (PBI). In the latter parents received training in interacting with their children and in stimulating their language development. The PBI approach had significantly better outcomes despite only marginally increasing the demands on the SLT service. Though not PCIT as currently practised, PBI had similar objectives and a common theoretical background and demonstrated the value of training parents to offer indirect therapy.

PCIT aims to improve the interactions between parents and children in ways thought to be more conducive to language development. These objectives have been developed in part from clinical experience and from research in the 1980s and 1990s, which examined differences in the interactions between parents with typically developing children and children with delayed language (see, for example, Conti-Ramsden and Freil-Patti, 1986; Laskey and Klopp, 1982; Peterson and Sherrod, 1982). The direction of causality is unclear in these studies; nevertheless they support attempts to change the parents’ style of interaction. PCIT does this by persuading parents to become less directive, to follow the child’s lead and to allow them to initiate. Parents are encouraged to offer praise and to comment, extend and repeat what the child says rather than to question.

Typically PCIT programmes video parents with their children so that therapists and parents can analyse interactions and jointly set therapy goals. Therapy sessions develop parental insight and confidence and practise these goals. Parents are asked to reflect on the changes in their child’s communication and to link this to changes in their own strategies. Additionally, but crucially, parents are asked to set aside daily ‘special times’ to interact with their children at home. They are asked to do so throughout the therapy and beyond with the intention that this will become a model for their future interactions with their children.

The use of PCIT with children who stammer has been examined in single case and case series studies (see Mathews et al., 1997; Millard et al., 2008, 2009) and a recent paper (Allen and Marshall, 2011) has reported its use with 8–10-year-old children with specific language impairment (SLI). The former studies conducted within the Michael Palin Centre, London investigated the style of PCIT developed there. Allen and Marshall (2011) randomly assigned 16 children to treated and delayed treated (control) groups. Five outcome measures were used. In four a significant time by group interaction indicated that treated children and parents had improved more than controls. This study suggests that PCIT may be useful with a wider range of clients than has previously been thought.

Research on PCIT with children who stutter or have SLI has served to highlight the lack of evidence for its use as an approach to treating younger children with delayed language. Despite its popularity no experimental studies of its impact are available and it is not included among the evidence based therapies on the ‘what works’ website (The Communication Trust, no date; Law et al., 2010). The present research assessed PCIT in a clinical situation within a large London Community Healthcare NHS trust. Most referrals of pre-school children in the trust are for delayed language, and PCIT is routinely offered to them as the first package of care. The study used a within-participants design. Children were assessed over a baseline period prior to receiving therapy and again after completing the therapy. Outcome measures in the study were a parent rating scale (see Appendix 1) used to measure the parents’ interaction skills, a measure of the child’s mean length of utterance (MLU) and the ratio of time that the child spoke to the time the parent spoke. Each measure was expected to remain stable during the baseline and improve after therapy.

II Method

The research took place within a Community Healthcare NHS Trust. Clinics within the Trust were asked to participate and to identify possible participants. A working party of senior therapists experienced in PCIT was set up to monitor the research.

III Results

Each of the outcome measures – parent rating score, child’s mean length of utterance and the ratio of child to parental speech – were analysed using one factor within-participant analyses of variance followed by planned comparisons of the individual means. In the case of the parent rating scale the advice of Clark-Carter (1997) that ordinal scales of measurement with more than 20 values may be analysed using parametric tests was taken.

Table 1 gives the means and standard deviations of each outcome measure at the three assessment points. A significant difference was found between the assessments of the parent performance (F(1.33, 22.56) = 26.66, p < .001, η² = 0.61, Greenhouse–Geisser correction applied due to lack of sphericity). Planned comparisons showed that no difference existed between the pre therapy measures but that the difference between the second pre therapy and the post therapy assessments were significant (p < .01, η² = 0.55).

OPEN IN VIEWER

Table 1.

Mean scores at each time point for parent ratings and child’s mean length of utterance (MLU) (s.d. given in parentheses).

		Pre therapy 1	Pre therapy 2	Post therapy
Parent rating (n = 18)	mean	14.33 (3.22)	15.33 (2.87)	20.89 (3.60)
Child’s MLU (n = 17)	mean	1.04 (0.61)	1.17 (0.70)	1.65 (0.93)

MLU data from one video was not reported by the scorer. The table reports the data for the remaining 17 participants. Again the differences between the three assessments were found to be significant (F(2, 32) = 5.96, p < 0.01, η² = 0.27), and planned comparisons found no difference between the two pre therapy assessments and a significant difference (p < .05, η² = 0.22) between the second pre therapy and the post therapy assessments.

Table 2 reports the data on the times of child and parent speech. As it shows parents reduced and children increased the time they spoke. The analysis of variance carried out on the ratio of child to adult speech showed a significant difference between the assessments (F(2, 34) = 9.77, p < .001, η² = 0.36). Planned comparisons again found no difference between the pre therapy assessments and a significant difference (p < .01, η² = 0.37) between the second pre therapy and the post therapy assessments.

OPEN IN VIEWER

Table 2.

Mean number of seconds for which children and parents spoke and the ratio of child/parent speech (s.d. given in parentheses).

		Pre therapy 1	Pre therapy 2	Post therapy
Child	mean	25.29 (21.29)	31.27 (19.57)	42.15 (27.12)
Parent	mean	113.05 (56.89)	98.15 (36.24)	70.72 (42.46)
Child/parent ratio (n = 18)	mean	0.22 (0.24)	0.32 (0.30)	0.59 (0.38)

In summary, consistent results were found for each of the outcome measures. Each improved slightly but not significantly during the baseline period. The changes over the treatment period were significant for each measure and the effect sizes were large in each case (Cohen, 1988).

IV Discussion

In assessing the clinical significance of these results we need to consider the extent and the nature of the children’s progress. Interactions are complex phenomena that lend themselves to various forms of analysis. Often a compromise must be struck between measures that are reliable and readily obtained and the insight they offer about the interaction. The parent rating scale was used because it tries to assess a range of behaviours and because it is extensively used both by the present SLT service and by other services in the UK. MLU and the ratio of child to adult speech are crude measures of the child’s language development but they are objective, and the latter in particular reflects a change in the balance of an interaction that is a primary objective of the therapy.

Each measure showed the same pattern; the videos at the end of the baseline period showed that neither practice at being videoed nor any insight that parents may gain from this are sufficient to change their interactions with their children. In contrast the post treatment videos reveal significant changes and large effect sizes for each of the measures. Clearly the intervention has changed the interactions in ways that are consistent with the aims of PCIT. Parents now talk less and the children talk more, and parents demonstrate some or all of the behaviours on the parent rating scale more frequently.

The potential longer term benefits of the therapy need to be judged more cautiously, however. Will a comparatively short period of treatment be sufficient to compensate for the children’s previous experience of interactions with their parents (or the lack of them) and lead to improvements in communication? Three things need to be considered here. First, is the change brought about by the therapy sufficient in itself to improve the children’s communication? This can only be answered by following the children over time. Ideally such research might be larger in scale and would employ a control group. The need to withhold treatment would be inappropriate given that the present research was conducted in a clinical setting with referred children. A second issue concerns the ability or the willingness of parents to maintain their newly acquired skills. The use of ‘special time’ is helpful in imposing structure, and parents are encouraged to continue it. Ideally parents will generalize their skills to other interactions with their children. The finding in Tannock et al. (1992) that parents maintained their skills four months after completing the Hanen programme is encouraging. Finally, there is the issue previously confronted by the Hanen approach (Girolametto et al., 1996, 1998), that the intervention might be more effective if it directly targeted changes in the children’s language through techniques such as focused stimulation. PCIT has adopted a more flexible approach on this issue. Current practice is to use elements of this approach in cases where the child’s language is sufficiently developed.

Evidence based practice normally proceeds by conducting efficacy studies in controlled conditions before carrying out effectiveness studies in clinical conditions. No efficacy study of PCIT has been undertaken reflecting a broader failure of researchers to assess the methods that clinicians use. Bernstein-Ratner (2006) has argued for greater involvement of clinicians in research and pointed out the current problems of translating evidence into practice. These include the well-known failure of the academic literature to report interventions that do not work, as well as difficulties in interpreting those that do. Given the heterogeneity of clinical case loads, clinicians want to know what works and who it works with. Intervention studies may report positive results but fail to identify a minority of participants who do not respond. The gap between evidence and practice is well illustrated by the finding (Nail-Chiwetalu and Bernstein-Ratner, 2007) that clinicians make little use of evidence and are more likely to consult colleagues for advice.

The clinical nature of the study also influenced its design. A within-participants design was used because it was inappropriate to delay the intervention for some children and because it was consistent with routine clinical procedure. Though less rigorous than the standards of experimental research, this approach has other advantages. Most apparent is that the findings support a therapy approach that is widely used. The positive outcome also means that the current use of the therapy as the first option for children with delayed language is appropriate. The involvement of a number of clinicians in conducting the therapy also gives reasonable grounds for generalizing the findings to other clinicians using similar methods.

The exclusion of children with EAL from the research prevents us from generalizing the findings to them. The importance of this issue is illustrated by the fact that more than half of the children referred to the SLT service during the research had EAL. Awareness of cultural differences in child rearing and in the way parents interact with their children is important in treating these children and may impact on the conduct and the success of the therapy. A discussion of this issue with particular reference to Asian families is available in Awde (2009).

The children’s scores on the PLS4 were surprisingly variable. Some children had severely delayed language further confirmed by their low scores on the measures used in the research while others were weak but within the normal range. Nevertheless the interactions of the latter children with their parents showed weaknesses similar to those of children with poorer language skills. This confirms the point above that the use of PCIT in clinical practice is correctly based more on observation of interactions than on the children’s language level on a formal assessment. It also shows that weak language skills are not always the cause of poor interactions and that changing parents’ behaviour can benefit children with varying levels of delay. For those children in the low normal range PCIT may be sufficient for their needs; normal practice is to keep the more severely delayed children on the case load so that further therapy can be offered if necessary.

That so many SLT services use versions of the PCIT approach suggests that it is within their resources to do so and that they perceive it to be useful. The approach owes much to Hanen but is less demanding (Baxendale and Hesketh, 2003). Other alternatives are available. Buschmann et al. (2009) report positive effects of training parents to interact with and support their children’s language development. This approach avoids the need to see parents with their children and study videos of their interactions. Training is in small groups over a three month period. The authors show that the approach has a considerable cost advantage over individual therapy although they base their calculations of the latter on the cost of 43 individual sessions. A further approach to parent training has been developed by van Balkom et al. (2010). Parent based video home training involves visits by a speech and language therapist and a psychologist who watch and discuss videos of successful interactions with parents. Children and their interactions with a parent were assessed before, after and three months after the training and compared with children directly treated by speech and language therapists. The training programme was successful in improving the grammaticality of the children’s speech and the conversational coherence of their interactions with parents. Future research may need to explore the relative effectiveness and resource demands of these different approaches as well as examining their long term effects on children’s language development.

PCIT is widely used in speech and language therapy in the UK. Results of its use in dysfluency (Mathews et al., 1997; Millard et al., 2008, 2009) and recently with older children with language disorders (Allen and Marshall, 2011) have been published. It is commonly used with young children with delayed language but evidence for its effectiveness is lacking. The present article is a first step towards acquiring this evidence.