Why Clinicians Choose Their Language Intervention Approach: An International Perspective on Intervention for Children with Developmental Language Disorder Free

Children with developmental language disorder (DLD) have impaired expressive and/or receptive language skills [1‒3] with certain aspects of neurodevelopment selectively impaired [4]. Intervention for DLD has been proven to be valid and effective [5], but scientific evidence-based practice recommendations, while frequently published, are not always implemented [6]. A variety of different models of service delivery for DLD intervention have been proposed, but there has been limited agreement on the best clinical practice for this client group [7].

There are undoubtedly challenges to the efficient implementation of evidence-based practice, and studies suggest that evidence-based interventions for children with speech, language and communication needs are tacit, poorly researched and underspecified [8]. Other barriers preventing successful implementation of evidence-based practice by speech and language therapy (SLT) practitioners include insufficient time to read research and implement new ideas, and reluctance to adopt new ideas because of the perceived value of existing practice which has not been scientifically tested [9, 10]. While the evidence does exist, Law et al. [11] found that of the 58 interventions identified for children with speech, language and communication needs, only 3% were found to have a strong level of evidence, with the remaining having only a moderate to indicative level [11]. This suggests that the evidence base may not be strong enough to directly inform practice.

SLT practitioners use external scientific evidence as one component of decision-making [12], combining this with practitioner and parental judgement about what is appropriate and feasible [13]. This is where client-centred and practitioner-centred factors can have an impact on the decision-making process for the type of intervention that is used with a child who has DLD. When categorising the types of intervention that are used, there are thought to be 3 styles of intervention that are commonly used to facilitate communication, language, and speech development: practitioner-centred, client-centred, and a hybrid mixture of both of these factors. Practitioner-centred rationales draw heavily on the practitioner’s knowledge and experience [14]. Client-centred rationales can be parent-oriented or child centric, for example, basing intervention on the parent’s main area of concern or basing the intervention approach on the individual child’s functional communication barriers. The “mixed/hybrid” approach draws on both practitioner and client-centred factors.

Decisions about language intervention are likely, to some extent, to be context-specific in the sense that external evidence may support an intervention but local issues may influence whether it is acceptable or feasible. Yet it is of particular interest to understand whether decisions about the application of evidence to practice are made in the same way across countries [15]. And, indeed, this raises the question of whether the research priorities are articulated at all (https://www.rcslt.org/members/research/research-priorities#section-4) and if so whether they are similar across countries. By focusing too much on a few dominant cultures, we might be missing out on a richer understanding of what facilitates language growth across languages and environments. Few international surveys of SLT practice have been conducted. Those that exist often do not include a wide range of countries and predominantly include English-speaking countries.

Salomone et al. [16] found considerable variation in early language interventions used for language disorders associated with autism spectrum disorder, and that in many countries in Europe, SLT services were simply not available [16]. This may be a function of availability but may also combine with socio-economic factors which can lead to children who are the most in need of services, such as language intervention, being the least able to access them [17]. Results of an international survey carried out across 13 countries to investigate SLT intervention for bilingual children suggests that clinical practice is not always based on research findings and available literature [18]. A comparison of practice in the USA and Taiwan found that the majority of SLT practitioners used established or emerging evidence-based practice [19], suggesting that the use of scientific evidence could reduce the variability and enhance the effectiveness of interventions used across countries and cultures.

In a national survey about the management of children with receptive language impairments in the UK [20] practitioners were found to use a wide range of activities, materials, and underlying theoretical rationales for their interventions. The authors suggest that theories underpinning the deficit are often much better developed than those underpinning the intervention. When considering the traditional pillars of evidence-based practice, more is known about external evidence than the practitioner-centred factors which affect the clinical judgement of the expert practitioner or the view of the fully informed patient [11, 21]. One study which focused on a single country [22] reported the rationales given by SLT practitioners for vocabulary intervention approaches. This included the client’s functioning, school-based needs and the client’s personal needs [22]. This was consistent with findings from Law et al. [20] who also identified individual needs, functional requirements and academic factors. Practitioners report that time, effort and collaboration with other professionals is required to optimise therapy potential [22]. Parental involvement is also identified as key by many [23‒26] although other factors affecting decision making may be more functional in origin, for example, affordability, availability, the age of the client, time available and the ease of administration [24].

Of course, this all has to be seen in the context of the origin of the evidence underpinning interventions and whether this is perceived to be generalisable across countries. For example, the majority of the literature is usually first reported in English, and most interventions have yet to be tested in different languages. It is essential as the intervention science in this field develops to acquire data on practice and evidence for interventions used in non-English-speaking countries.

The current study aims to explore the characteristics of the practitioner and client and their bearing on the application to decisions related to interventions for children with DLD.

This paper seeks to address 3 specific questions.

Research Question 1: Do client characteristics affect the rationales used for intervention with children with DLD?

This question concerns whether the severity of the child’s difficulty or their age dictated intervention selection. Specifically we anticipated that children who were older and with more severe difficulties would be more likely to receive a client-centred rather than a family style of intervention, the latter being more commonly a feature of interventions offered to younger children.

Research Question 2: Do practitioner characteristics affect the intervention approaches that are used?

Here we were interested in exploring whether the age or the educational level of the practitioner affected their intervention decisions, the argument being that greater experience may lead to a more person-centred approach.

Research Question 3: What are the characteristics of practitioners who report using scientific evidence when delivering intervention?

Again, we were interested in whether the experience and educational level of the practitioner informed the extent to which they relied on external evidence in their decision making. We anticipated that more recently qualified and thus the younger professionals would have been more likely to have been trained in evidence-based principles.

Data collection was completed in 2017 by members of COST Action IS1406, an international research network comprising representation from 39 countries funded by the EU’s COST Action programme. Survey questionnaires were developed, translated, back-translated and distributed online by professional bodies and national teams for each COST Action country in the appropriate language or languages, participants also shared the questionnaire online which increased the distribution of the survey. The web links were released in July 2017 and closed on October 13, 2017. Participants answered closed and open-ended questions online to provide information about their management of children with DLD. Full details of the survey are provided in Law et al. [27]. The survey itself and the relevant accompanying documents are publicly available in the Newcastle University data repository (available at: https://data.ncl.ac.uk/articles/COST_Action_IS1406_Practitioner_Survey/9802880). In addition members of the Action have been involved in a series of separate analyses of the survey data with regard to the role of parents across the respondent countries [28], cultural and linguistic factors [29], the use of direct and indirect therapy in different countries [30] and the ways that these services are funded [31].

The survey questionnaire comprised 4 sections:

1 Some information about you

2 Issues regarding intervention delivery

3 Theoretical considerations

4 The social and cultural context of intervention for children with language impairment (referring to the practitioner’s overall caseload)

Section 1 gained information about the practitioner’s characteristics:

1a Level of “professional” qualification (closed question: non-university diploma, non-university: other, university: undergraduate/bachelor degree, university: masters, university: Dr. (PhD), university: other (e.g., diploma))

1b Years of experience with children with language impairment since qualification

Section 2 and 3 questions were linked to a reference child from the practitioner’s caseload. Section 2 gained information about the client’s characteristics:

2a How old is the child?

2b Using your clinical judgement, what is the severity of the child’s language impairment? (closed question: mild, moderate, severe)

The majority of responses were in quantifiable Likert format but a subset was of a more qualitative, open-ended format, requiring the respondents to add their own text in the space provided.

The primary analysis for this paper is based on the 2 open-ended questions from section 3, which investigate the language intervention approach and rationale behind the choice of intervention:

3a Name the main intervention approach (ranging from client-centred, to hybrid, to clinician-centred) you used for his/her language difficulty

3b Why did you choose this intervention approach?

The section 1 responses about the practitioner’s characteristics and the section 2 responses about the client’s characteristics were also used in this paper to explore relationships with the open-ended intervention responses from section 3.

Respondents needed to have responsibility for the management of children with language disorders and make key decisions about intervention and service delivery. The questionnaire contained a definition of “language impairment” and specified that the questionnaire was based on primary language impairment, which is now known as DLD [2]. This definition excluded language difficulties associated with bilingualism, secondary language impairments and use of alternative or augmentative language systems. Six thousand and 3 responses were collected from 65 countries; 979 of these responses included participant details only, so were excluded. This resulted in 5,024 responses, which were reduced further. For the purposes of the present analysis only the 39 countries that were included in the development of the survey were included in the analyses (Fig. 1). Finally, many respondents did not answer the targeted questions about intervention approaches and rationales (questions 3a and 3b) and so were excluded from this study, resulting in a total of 2,408 viable participants.

Data were collected, anonymised and translated into English from 34 different languages by national teams who were members of COST Action IS1406, an EU-funded research network which included representation from the 39 countries. Answers to the closed-ended questions were coded by an author of this paper and another COST Action member, using the coding manual that was developed by themselves. Responses to closed-ended questions were analysed in a previous study [27]. The open-ended questions were then re-coded for the present study as follows.

Survey Question 3a: Name the Main Intervention Approach You Used. There were a variety of general and specific intervention approaches. Figure 2 illustrates the intervention codes; headings show the general categories and subheadings allow for more specific responses. There is the potential for considerable crossover between the responses to questions 3a and 3b. Nevertheless there should be a clear distinction between whether an intervention itself is classified as child centred, clinician directed or hybrid (on Fey’s continuum [32]) in question 3a compared with the rationale/thinking of the clinician as they decide to use a particular intervention in question 3b. Of particular interest is to what extent respondents report that they rely on their own experience/beliefs about intervention choice, whether they respond to parent/carer expectations or whether they rely on scientific evidence to guide them in their decision making.

Some responses to question 3a included specific language interventions. Three hundred and sixty specific interventions were recorded. Interventions were listed alongside the number of participants reporting each intervention, in order to identify the interventions that were frequently used, and to allow for further investigation of a few frequently stated interventions. A total of 161 interventions were indicated by multiple participants. See electronic online supplementary material (see www.karger.com/doi/10.1159/000513242) for a full list of specific interventions.

The practitioner’s country of practice was recorded in order to explore possible variation or consistency across countries. Interventions were grouped into categories related to the targeted area of difficulty or disorder, derived from supporting information on the front of the questionnaire and relevant literature [2] (Table 1).

Survey Question 3b: Why Did You Choose This Intervention Approach?With the data on theoretical rationales for choice of intervention, general headings and subheadings were also used. The rationales identified by the previous studies included in the literature review were also considered during the coding process. Figure 3 illustrates the headings and subheadings for these data.

To examine the inter-rater reliability of the coding process for questions 3a and 3b [33], 2 members of the research team independently coded and then compared responses for 9.2% of the participants, 222/2,408 participants. The researchers created codes independently and then compared which participants they coded as “client-centred,” “practitioner-centred” or “mixed” (when participants included both of these codes) across questions 3a and 3b. For question 3a an additional code for “specific interventions” was used, and for question 3b an additional code for “scientific evidence” was used. The assessed percentage agreement of the codes was then calculated. Any differences were discussed before proceeding. Reliability of coding was 96.4% for question 3a and 94.1% for question 3b. With consultation 100% consistency was achieved.

After coding the qualitative data, the frequencies of codes generated from responses to questions 3a and 3b were initially analysed in IBM SPSS [34]. χ² analysis (with no expected frequencies below 5) was used to relate the responses to one another and to test associations. The predictor variables were the rationales from question 3b, and outcome variables were the intervention approaches from question 3a. But these variables, being initially qualitative were then converted to binary variables for the purposes of quantitative analysis and because we were interested in the outcome being present or absent. In this format (i.e., binary data) the variable allows the use of logistic regression to measure the effect of the dependent variable on the outcome [35]. Therefore, a participant either did or did not include a specific code in their response (1 or 0). In most cases this was either coded as present (1) or not present (0). For client age and years of practitioner experience, these data were not coded into binary data because this was the independent variable and therefore did not need to be re-coded. There were 3 levels of severity (mild, moderate, severe), and these categories were used in separate analyses creating dummy variables, for example, coding severe as (1) and mild as (0) for one analysis, and then coding moderate as (1) and mild as (0) in a separate analysis. Associations were then explored between the theoretical or intervention data (dependent variable) and the practitioner characteristics or client characteristics (covariates). These variables were then added into logistic regression models and the outputs expressed as Exp (B) which is effectively an odds ratio.

Predictors entered as covariates were the severity of the impairment, the age of the client, practitioner experience (in years) and level of practitioner education (diploma, degree, postgraduate). The education data were split into 2 categories by joining and re-coding responses. This allowed categorisation of participants who did or did not have a university qualification. The Wald χ² test was used to assess whether explanatory variables in the model were significant. “Significant” means that they add something to the model; variables that add nothing can be deleted without affecting the model. The Hosmer and Lemeshow goodness of fit test was used to assess whether there was evidence for lack of fit in the logistic regression model. Simply put, this test assesses how well the data fit the model [29].

Findings are presented below for each of the 3 research questions together with some specific subquestions about relationships between respondent characteristics and the outcomes. Figures 4 and 5 demonstrate the spread of all responses related both to the intervention approaches (Fig. 4) and the rationales behind the choice of intervention (Fig. 5).

Figure 4 indicates that client-centred interventions are favoured by nearly half the respondents and relatively speaking the number of practitioner-centred responses is relatively small. The reference to specific interventions refers to the interventions listed in the electronic online supplementary material. Generally they are relevant to the child’s needs but might not be truly child centric because they will commonly follow protocols determined by the intervention not the child. The focus on the model service delivery driving the choice of intervention suggests that this may be prescribed as far as some practitioners are concerned. Turning to Figure 5 the rationales underpinning the chosen intervention are predominately children centred, reflecting the child’s needs. Of particular interest here is the fact that the reason for choosing an intervention is rarely driven by the underlying evidence.

The frequencies of intervention approaches and rationales are summarised in Tables 2 and 3. In accordance with Figures 4 and 5, most participants included a client-centred element in their response although this was more common for the “rationale” question (survey question 3b), over responses to the “intervention” question (survey question 3a).

Table 3 demonstrates the frequency of specific responses under the client-centred heading. The percentages in the initial analysis tables are not cumulative because many participants included multiple codes in their response. There was a range of subheadings for client-centred approaches (survey question 3a) and rationales (survey question 3b). The frequencies for each response can be compared across the “intervention” and “rationale” questions below (Tables 3-6).

We then examine the role played by the age of the child and the severity of their condition to address the first question.

Examining the choice of client-centred and family-centred interventions in relation to age, age had relatively little bearing on whether a client-centred approach was adopted but it did have an effect on whether the intervention was a family-centred approach which was significantly less likely to be employed with older children (Exp [B] = 0.988 [95% CI 0.980–0.996]; p = 0.003, p < 0.05), although the effect size was relatively small. By contrast, reported severity of the child’s difficulties had bearing on the rationale, if the level of difficulty was considered severe (Exp [B] = 0.790 [95% CI 0.658–0.95]; p = 0.12, p < 0.05), but not if the difficulty was considered to be moderate (Exp [B] = 0.848 [95% CI 0.601–1.198]; p = 0.349, p < 0.05). Thus, if a participant reported that the client had a more severe impairment, then they were more likely to use the client’s level of functioning as a rationale for the intervention approach they used. This suggests that if the child has a severe difficulty it will be a rationale, for the more severe the child’s difficulty, the more likely severity will be a criterion for intervention choice, but it will not be a criterion if the difficulties are more moderate.

Table 4 indicates that practitioner-centred intervention approaches are much less common than those that are client centred. The relative frequencies of responses can be compared across questions 3a and 3b below.

As there were a limited number of responses deemed practitioner centred, relevant variables from survey questionnaire section 1 (practitioner information) of the questionnaire were used to explore associations between practitioner characteristics and the intervention approaches and rationales that they adopted. The experience of the respondent (in years) appeared to have no bearing on whether the respondent reported that they used personal experience as a rationale for intervention choice (Exp [B] = 0.999 [95% CI 0.995–1.002]; p = 0.436, p > 0.05) and they were no more likely to use practitioner-centred rather than client-centred rationales for their choice of intervention approach (Exp [B] = 0.999 [95% CI 0.997–1.0]; p = 0.172, p > 0.05). As discussed above, practitioners were asked to indicate their level of qualification, here split into above and below degree level. Interestingly those with a higher level of qualification were less likely to use a client-centred approach (Exp [B] = 0.631 [95% CI 0.477–0.834]; p = 0.001, p < 0.05), and they saw their educational level as a rationale for their choice of intervention (Exp [B] = 1.037 [95% CI 0.471–2.283]; p = 0.928, p > 0.05). Overall, these results suggest that it is the practitioner’s level of education rather than their years of experience, which is related to their choice of intervention rationales, and the clinical approaches adopted.

There were relatively few practitioners who reported scientific evidence as a rationale for their choice of DLD intervention (5.44%) but it was possible to ascertain that those that did exhibited certain characteristics (Table 5).

As indicated above the number of participants using scientific evidence was higher when reporting the rationale for intervention. Within these responses, the majority of participants stated “evidence-based” instead of naming a specific evidence base (Fig. 6). Results suggest that participants were more likely to use practitioner experience as a rationale than scientific evidence (9.47 vs. 5.44%).

It was predicted that participants who used scientific rationales would also use client-centred or practitioner-centred approaches as opposed to using scientific evidence alone and indeed analysis shows that there is a significant association between responses containing scientific evidence and client-centred responses (χ² = 33.98, p < 0.001) as well as practitioner-centred responses (χ² = 16.13, p < 0.001), which supports this prediction. Further analysis shows that there is a significant association between participants stating a specific intervention and giving an evidence-based rationale (χ² = 62.16, p < 0.001). This supports the finding that participants who stated a specific intervention (n = 131) also reported the evidence base behind the intervention (n = 90), which is 68.70% of these participants, although of course the proportion who report using an evidence-based approach was still relatively low.

The odds of practitioners including scientific evidence as a rationale depending on their education, training and experience were then explored further. Practitioners were twice as likely to use scientific evidence as a rationale if they had a university level qualification (Exp [B] = 0.530 [95% CI 0.245–1.15]; p = 1.08, p > 0.05) although this result is not statistically significant. Thus having a university degree is not, in itself, a significant predictor of the use of scientific evidence, and the same was found for years of clinical experience (Exp [B] = 1.002 [95% CI 1.0 to –1.03); p = 0.011, p < 0.05). In summary, these analyses suggest that the practitioner’s level of education is more strongly associated with the use of scientific evidence-based practice than the number of years of experience. However, there are no significant differences between practitioners who did, or did not, use scientific evidence as a rationale.

More than a third of the participants (34.46%) included one or more specific interventions in their response to question 3a. There are 1,676 instances of a specific intervention being named, which is greater than the number of participants reporting a specific intervention because many participants reported more than 1 intervention. Out of 360 different interventions named, 161 interventions were reported by one participant only and 199 interventions (55.28%) were reported by more than 1 participant. Practitioners from a wide range of countries included a specific intervention in their response.

χ² analyses were employed to explore the associations between the use of a specific intervention and responses in sections 1 (practitioner information) and 2 (client’s characteristics) in the questionnaire. A significant association was recorded between participants reporting a specific intervention and a service delivery model (χ² = 7.87, p = 0.005).

The interventions are detailed in the electronic online supplementary material, and the most commonly reported language interventions are summarised in Table 6 with the frequency of their use and the participants’ country of practice. Many general approaches such as recasting and behavioural interventions were included, alongside the more specific language interventions. The most frequently reported therapy, Zollinger therapy, was reported by the 3 German-speaking countries (Germany, Austria and Switzerland) only. The specific interventions reported by respondents were underpinned by different levels of evidence, ranging between having a strong evidence base (e.g., context optimisation) and having no readily accessible evidence base (e.g., Zollinger therapy).

This study has explored the associations between the client’s characteristics, the practitioner’s characteristics and the use of scientific evidence in relation to the practitioner-reported speech, language and communication intervention approaches and rationales. Key features of the analysis were the very large sample size (even allowing for non-response) and the use of open-ended questions to investigate specific interventions.

Most practitioners used at least one of the many client-centred rationales (Table 2), suggesting that intervention planning may involve many ways of taking the client’s individual qualities into account. The reason that there were many different client-centred rationales may relate to practitioners focusing on different priorities, depending on their training [36]. The dominance of client-centred responses is positive in practice as personalisation of intervention is important [7]. Previous research has highlighted the need for more individualised interventions [23], and this may have led to client-centred approaches being included in training programmes. Additionally, client-centred characteristics may be dominant as they are driven and promoted in policy guidelines and scope of practice documents emanating from governing bodies and policy makers such as the WHO [37] and CPLOL [38].

The child’s performance and their perceived needs function rather differently, and respondents appeared to base interventions on the client’s perceived specific needs rather than the severity of their difficulties. Previous studies have combined both of these terms [20] so could not separate the client’s needs from their abilities. In the current study, participants frequently cited the general belief that the practitioner knew what was best for the client, and the child’s motivation was often noted as a rationale behind using a specific intervention. The descriptive statistics also suggest that the client’s age and behaviour are significant rationales behind using a specific intervention.

Practitioner-centred factors were not as evident in the responses as client-centred factors, which suggests that the client’s characteristics have more of an influence on intervention choices. There were more practitioner-centred rationales than intervention approaches, which could be because a practitioner’s knowledge and training can have a significant influence on the rationale behind an intervention.

Within the specific practitioner factors, results suggest that ease of use is a significant aspect of clinical decision-making, although there have been limited reports of this previously. Time however has been reported by multiple studies [22, 24]. Reference to involvement with the multidisciplinary team was a common response, which rather contradicts earlier studies which have suggested that there was a lack of cooperative work between SLT practitioners and other professionals [39]. This could be because the current study was carried out across countries instead of within a country. Practitioner experience was a frequently used rationale, which suggests that it highly influences clinical decision making, and practitioner experience is an important aspect of evidence-based practice [40].

One of the most interesting findings is that practitioners with a higher level of education are less likely to use a client-centred intervention approach. This is surprising because individualisation of intervention has been included in practitioner training [7]. An explanation could be that practitioners with higher education have been trained to use a scientific theoretical approach, so have more knowledge about the scientific evidence behind intervention, and therefore are less likely to use an individualised approach to intervention. This coheres with findings of separate analyses conducted on other data from the questionnaire (reported in Saldana and Murphy [41]), where practitioners educated to a higher level were more likely to claim use of theory of language development in their decision making.

The results regarding scientific evidence as a rationale agree in part with the practitioner characteristic analysis, because education is more strongly associated with scientific evidence-based practice than years of experience but contradict evidence that recently graduated practitioners are more likely to use scientific evidence [32]. This could suggest that there is adequate continued professional development because practitioners with more experience use evidence as frequently as newly qualified practitioners. Additional higher education did not have a significant impact on the use of scientific evidence, contradicting the theory that university training would necessarily positively influence decisions associated with adopting an explicitly scientific theoretical approach. This may be relatively surprising but of course does beg the question of the content of the courses concerned.

When participants used scientific evidence, they infrequently named the specific evidence base. This could suggest that participants were unable to provide information about specific evidence when completing the questionnaire. However, practitioners who know that an intervention is evidence based do not need to remember a researcher’s name in order to carry out intervention appropriately. Consistent with previous findings [23], practitioners were more likely to use experience as a rationale than scientific evidence.

The lack of responses associated with scientific evidence suggests that it is not frequently used as a rationale for intervention. This could have contributed to the variation in clinical practice [14]. The fact that service delivery was more frequently cited than evidence-based practice highlights the reduced frequency of these responses. This is despite evidence that scientific evidence-based practice is required to carry out effective intervention, consistent with Michie et al. [6] and in the UK it is a fundamental element of SLT practitioner training [42].

Results could suggest that scientific evidence is not ingrained in the practitioner’s clinical knowledge, questioning training in evidence-based practice and/or access to evidence across countries. The association between reporting a specific intervention and giving a scientific rationale could support findings that interventions are often produced and developed based on sufficient scientific evidence and the theoretical knowledge around typical development but that the efficacy of the intervention in question is not formally tested [43]. For example, in the development of the What Works for SLCN database. Law et al. [44] used a survey format to identify what interventions were most commonly used in the UK by speech and language therapists. The results were then checked against the available external evidence published in the peer-reviewed and grey literature, that is, literature that is either not published or is available online and has not been peer reviewed. Although authors often claimed theoretical and face validity, very few had explicit external evidence supporting the value of the interventions themselves.

One potentially important barrier to the adoption of evidence is its availability across languages and separately, considering the population in question, children with DLD, the availability of evidence for different languages. The translation challenges, the possibility that some terms may not have been understood despite this, the varying theoretical approaches adopted in different countries and the fact that many interventions have not been tested across languages could all be considered. Additionally, preferences in terms of theoretical underpinnings in different countries may influence practitioners’ choice of intervention and what they believe works.

The frequently used interventions were used across countries and therefore across cultures, which is consistent with previous findings [19]. Some interventions had no available evidence or evidence that questions the effectiveness of the intervention. It is interesting that there were varying levels of evidence behind frequently stated interventions, because using interventions with limited evidence could suggest that clinical practice is based on factors other than scientific evidence, such as experience [12].

Results suggest that practitioners frequently gave additional service delivery information when reporting a specific intervention. This suggests that practitioners consider the best method of service delivery for the intervention that they are carrying out, even if they are not asked for this information. However, practitioners who report that they use a specific intervention might not be carrying it out in the most appropriate way for best clinical practice.

Many participants did not answer either of the open-ended questions, which suggests that participants did not necessarily understand the questions, and this is perhaps indicative of a level of sampling bias in our findings. The questions had been trialled with practitioners in all of the countries that were included in this study, which should have increased the odds that questions would be understood. The current study focuses on the reported interventions and rationales linked to one client, but these questions do not ask for every intervention or rationale used.

The survey questions analysed focus on a reference child from the practitioner’s caseload. Practitioners were able to choose any child, so could have chosen a client who benefited the most from intervention, instead of a client who showed limited progress, potentially introducing bias in the response. The term “intervention approaches” rather than “specific interventions” was adopted in the survey, which could have been interpreted in a number of ways and could be why only a third of participants stated a specific intervention. The survey format might not be the best way to assess specific interventions because it does not allow for follow-up questions to investigate reported interventions. Nevertheless, there are many other merits of the survey approach including the large sample size and wide range of countries involved, which increases the reliability of the results.

Finally, there is a possibility, as there is in any survey, that the ordering of questions may prime a response. The example in this study was the relationship between questions 3a and 3b. Thus question 3a could prime responses for 3b, which asked for the clinician’s rationale for choosing the intervention approach. Indeed, the fact that some albeit a small number of respondents indicated they did not know what was being asked in question 3b would support this. This could be explored further using a more open-ended interview approach using general topic prompts which would be another means of surfacing clinicians’ decision-making processes. However, as we have stated, having responses across such a large group of clinicians from different countries is valuable, and despite the possibility of a priming effect from question 3a to 3b there are still many responses which are unique and not related to the examples in question 3a.

Some of the findings reported here appear to contradict those in earlier studies. Further research should include the possibility of carrying out more detailed face-to-face interviews in the countries concerned, to tease out some of the implications of the present study’s findings. This would complement the information that was gathered by the current study because follow-up questions could be used to investigate intervention approaches and rationales. This would elicit more detailed information about the specific interventions that SLT practitioners included in their responses. Interviews would also provide an opportunity to gain more information about the separation of “speech” and “language” in certain countries. In addition, interviews could shed light onto the relationship between the external evidence (e.g., randomised controlled trials) and the perception of practitioners. Technically the external evidence should be replicable across national boundaries but whether an intervention is acceptable or indeed feasible given the funding remain fundamental questions.

Results suggest that there should be more initiatives in place to encourage the use of scientific evidence-based practice at an international level. Our findings highlight the variability in intervention rationale and practice. And this in turn could have a bearing on the training of SLT practitioners about client-centred care. These results highlight the current influence of client and practitioner characteristics, which could usefully be monitored in terms of the effect that they have on outcomes over time.

Our findings show that practitioners draw on a wide range of different rationales when making decisions about intervention approaches for children with speech, language and communication needs. Clearly practitioner experience is critical to the decision-making process, reflecting the importance of the practitioner perspective as one of the pillars of evidence-based practice. When it comes to specific intervention programmes, many different interventions are used, and of these many had common theoretical underpinnings and characteristics but often, as far as we could ascertain, with relatively little evidence of their value having been tested. There is a need for SLT practitioners across countries to pay due regard to major contextual differences. It would appear that external evidence, key to the principle of evidence-based practice, is not the main driver of intervention choice for most of the participants, despite there being a more robust evidence base now available. This suggests that practitioners are either not familiar with the available evidence or, if they are, are not convinced of the relevance of the data to their practice.

The data for this study were derived from the Practitioner Survey which was part of the work of Cost Action 1406, entitled “Enhancing children’s oral language skills across Europe and beyond,” a collaboration focusing on interventions for children with difficulties learning their first language. Cost Action is funded by the European Union. We are grateful to the over a hundred members of the Action who contributed to the development of the questionnaire and particularly to Prof. David Saldana for his contribution to the questions upon which this paper is based. Finally, we would like to thank Elizabeth Quinn, MSc Speech and Language therapy, University of Limerick, for her contribution to data analysis and reliability testing and for comments on an earlier draft.

An ethical opinion was sought from the University of Newcastle Research Ethics Committee. The approval was granted on January 18, 2017 (Ref: 11532/2016). A copy of the ethical opinion was circulated to all those involved in preparing the survey, and they were asked to communicate this with local bodies as necessary.

The authors have no conflicts of interest to declare.

The study was indirectly funded by the European Union’s Cost Action Programme.

Rachel Forsythe carried out the analysis for her University BSc dissertation under the supervision of Professor Law. She drafted the paper and commented on subsequent versions. Carol-Anne Murphy oversaw the development of the aspect of the questionnaire employed in this study. She supervised Elizabeth Quinn, a student at the University of Limerick, who contributed to the reliability checking. She commented on the paper and suggested edits as it developed. Josie Tulip helped develop the survey and co-ordinate its translations, assisted with dissemination of the survey, and coded and created the data sets from the survey data. They assisted Rachel Forsythe with the access to the cohort data, specifically the qualitative data used in her analysis which had not been used before. James Law oversaw the development of the original survey instrument, supervised Rachel Forsythe’s dissertation and contributed to the writing/editing of the paper and submitted the paper.