Deborah Gibbard
21 Holland Park, Sarisbury Green
SOUTHAMPTON, SO31 7TR, HANTS. (Angleterre)
Tél. : 01489 575976
e-mail : steven.gibbard@btinternet.com

Research is also beginning to examine different intervention approaches and techniques for language-impaired children (for example direct versus indirect intervention, or individual versus group intervention). In particular, this research is beginning to look at the outcomes and treatment effects of these approaches, searching for greater specificity in the efficacy question. Pearson (1995) highlights an urgent need for evaluation of the effectiveness of intervention in children, stressing that Purchasers need to know the most effective way of delivering a service. Although in general, there is a positive feeling for a closer partnership between parents and professionals, there equally remains much uncertainty about what to do and where to start. Relatively few speech & language clinicians are actively engaged in parental training.

A child diagnosed with a speech & language delay may be offered any one of a number of different types of intervention. In Portsmouth HealthCare NHS Trust, current practice is for the pre-school child with a speech & language delay to receive "general care" (defined later). Indirect approaches, if able to demonstrate effectiveness, will be attractive to managers and service providers in that they have the potential of demonstrating economy in service delivery. Of further importance though is establishing the most effective and least costly methods of intervention.

The evaluation of treatment in speech and language therapy is a complex process (Law, 1992, 1997; Enderby & Emerson, 1995). However, the evidence, within the limits of methodological problems clearly suggests that children can learn more when parents and therapists participate together. Several studies in the literature have found substantial gains can be made to children’s language by working with their parents in different formats (eg. McConkey et al, 1979; Beveridge & Jerrams, 1981; Fey et al, 1993).

Very few studies though have examined the effects of different types of treatment for the same condition. Gibbard (1992, 1994) carried out work that demonstrated the effectiveness of parent-based intervention in comparison with both a no-treatment control group and an individual, direct model of intervention. An effective method of developing children’s syntactic skills was established.

Even fewer economic evaluations of language intervention have taken place. Law et al (1998) highlight a lack of cost-benefit analyses in the past. Not only the costs of implementing particular types of intervention are important, but the cost savings over time are also vital. Of the few studies that have examined economic considerations, Barnett et al (1988) investigated the economic efficiency of several types of intervention with pre-school language impaired children. They found that a home-based intervention programme was most efficient. Centre-based treatment (both individual and group) was found to be more expensive and less effective. When both types of intervention were combined, the costs increased significantly without increasing effectiveness.

Further, Eiserman et al. (1990) found that pre-schoolers on a home training programme made similar speech and language improvements to pre-schoolers on a clinic-based programme and that in addition, there were no differences in programme costs. Results from follow-up testing one year after the intervention ended corroborated the results immediately following intervention (Eiserman et al, 1992).

The objective of an economic analysis or evaluation is to determine if a clinical intervention or mode of service delivery is worth doing when compared with other possible uses of the same resources. It is only appropriate to consider cost data where different treatments or provision of service produce equivalent outcomes. If outcomes vary between alternative methods of treatment, then both economic costs and effectiveness or outcome data needs to be collected and combined. This enables a comparison of the relative cost of health care provisions and the resulting outcomes. Hence, an economic evaluation, if properly conducted, precludes any conclusions that cheaper health care alternatives of lesser quality are necessarily better.

There are a number of frameworks for undertaking an economic evaluation (cost-effectiveness analysis, cost-utility analysis and cost benefit analysis). Which is appropriate is dependent upon the type of outcome measure used (for example, physical units (eg. bed days saved or quality of life). For a full discussion of economic evaluation, see Drummond, Stoddart and Torrance (1996) or Jefferson, Demicheli, and Mugford (1996). In this study costs and consequences are presented separately.

Recent research (Whitehurst & Fischel, 1994) suggests that treatment decisions for children with developmental language delays should be considered in terms of the relative costs and benefits of the treatment.

It is hypothesised that parent-based intervention (Gibbard, 1998) will provide a cost-effective method of intervention.

Research has begun to identify the features of early language learning that are important for child language acquisition. The importance of the relationship between the child’s focus of attention and the linguistic input has been emphasised (eg. Tomasello & Todd, 1983; Akhtar, Dunham & Dunham, 1991). Bruner (1983) stressed the importance of turn-taking. The use of language in familiar and repetitive contexts has been found to be of importance (Harris, 1992). Parent intervention models are therefore based on the premise of providing a frequent, familiar, repetitive and salient input for the language impaired child.

However, many parent language intervention models using naturalistic intervention procedures have provided general language stimulation (eg. Tannock, Giralometto & Siegel, 1990), rather than aim to achieve specific language objectives or targets. These approaches may be suitable for children where the aim is to increase conversational competence, but may be less useful where the aim is to increase and develop linguistic structures. Indeed, evaluation of one interactive approach found no improvements in children’s linguistic abilities at the end of a course (Tannock, Giralometto & Siegel, 1992).

The setting of clear language targets, is an additional feature of the parental-based language intervention model (PBI Programme, Gibbard, 1998), evaluated in Gibbard (1992; 1994) and children were found to make significant improvements with their linguistic development.

An intervention study was used to compare outcomes for children receiving one of two different interventions. The main treatment objective for the six month period of intervention was to increase the child’s linguistic complexity from a single word level to that of using three to four word utterances.

Intervention 1: indirect parent-based group treatment ("Parent-based intervention programme, Gibbard, 1998: defined later). This condition will be referred to as PBI.

This group received PBI treatment over a 6 month period (ie. parental attendance at 11 group PBI sessions, each lasting approximately 90 minutes in length).

Two PBI groups took place. Two individual speech and language therapists carried out the PBI group treatment (intervention 1) and they:

(a) Read the "Parent-based Intervention Programme - a group approach for language-delayed children" (Gibbard, 1998).

(b) Attended an additional training session on implementing PBI.

Intervention 2: indirect individual general care treatment (Current practice: defined later). This condition will be referred to as General Care.

This group received General care over a 6 month period (ie. parent and child attendance at 2 individual General care sessions, each lasting approximately 60 minutes in length).

All speech & language therapists were briefed, describing and standardising the exact nature of general care (current practice) across the Trust.

This intervention programme was followed with the PBI group.

The intervention method employed in this programme uses a combination of the setting of linguistic objectives and an interactional approach. Several other studies have also advocated the setting of language targets as part of parental training programmes (McConkey et al, 1979; Warren & Kaiser, 1986).

The emphasis of this programme is on developing a child’s expressive language development through the use of daily routines and situations which are of importance to them. Over a period of eleven group sessions, fortnightly language objectives are set for the parents to work on at home with their child. A structured teaching demonstration takes place during the group sessions for each language objective set, in order to explain and clarify each objective to the parent, ensuring that their full understanding is achieved. Practice activities are also devised during the sessions, to encourage the parents to think about each language objective flexibly. However, each parent would then implement the language objectives set at home with their child in a different way, according to the particular routines and interests of the child. The parents are encouraged to follow their child’s lead at home and strong emphasis is placed on achieving the language objectives through the use of daily routines and situations.

All Portsmouth HealthCare NHS Trust speech & language therapists (paediatric) have provided information on three language-delayed children that they have treated over the last 12 months. From the compilation of this information it has been established that current practice is to see each parent and child individually, on a three monthly review and advice basis. During these sessions, the parent receives appropriate general (non-specific), verbal advice on techniques and methods to stimulate and develop the child’s language level. These techniques and suggestions are summarised in a written report which the family receive within two weeks of attending their appointment.

Following referral to the speech & language therapy department, each parent and child attends an initial interview and assessment (this is current practice).

During the session it was established whether the child met the following pre-determined set of selection criteria:

1) 22 -36 months of age.

2) Little or no expressive language (a vocabulary of 30 or less single words).

3) No formal evidence of verbal comprehension difficulties more than 2 SD from the norm, or no informal evidence of more than moderate verbal comprehension delay.

4) No previous speech & language therapy intervention

5) A primary speech & language therapy diagnosis of either: (a) language delay, or (b) speech & language delay.

6) No known aetiology.

The first 14 children meeting these criteria were allocated to PBI intervention and parents offered a place on a PBI course at one of two local clinics. Parents were given a choice of which group they wished to attend, as each of the two groups took place at different venues, on different days of the week and at different times.

The next 14 children meeting the above criteria were allocated to general care (current practice).

These numbers were realistic in that they were able to provide initial information over a relatively short time-scale (approximately 8 months) on the cost-effectiveness of parent intervention. Past experience has shown that around 8 participants in a PBI group is manageable and these figures also related to relevant current speech and language therapy referral rates.

Any parent/child offered PBI intervention, but unable or unwilling to attend a group, received general care. However, these children were not included in the study.

In total, 28 participants were offered intervention as part of the study. 22 participants completed the study (12 in the PBI intervention group; 10 in the general care intervention group). There was therefore a 78.6% compliance rate. Data for 21 participants is included in this study (11 in the PBI group; 10 in the general care group), as one clinician failed to collect all post-intervention assessment measures required.

Information was collected on the subjects to investigate whether the two intervention groups were evenly matched in terms of child age, parental age, ordinal position in the family and socio-economic status (Classification of Occupations). Table 1 provides a summary of this information.

Any differences between the groups were not found to be significant on statistical analysis using two-tailed independent t-tests, with the exception of child age which did show a significant difference (p<0.43), with the General Care group children being slightly younger at the beginning of the study.

Child language abilities were assessed pre- and post-intervention. Profiles of language ability were compiled, using a combination of standardised and other measures of each child’s expressive and receptive language abilities.

The following assessments were administered:

a) Parental report of child language

- word list

- estimate of phrase length

The parents were asked to compile a list of all the words that their child used spontaneously and meaningfully. They were also asked to estimate whether their child used single words only or combined words. If word combinations were used, the parent was asked to estimate the average length of utterance used.

b) Reynell Developmental Language Scales (Reynell, 1983) - comprehension section.

c) Language Sample

- transcriptions of spontaneous, meaningful words and phrases

- mean length of utterance (MLU)

Parents were asked to play with their child, whilst the language sample was recorded. Written transcriptions of all utterances were obtained and those utterances that were intelligible as spontaneous, meaningful words or phrases were noted. Each language sample consisted of a minimum of 50 child utterances. Parents were asked to confirm that the sample obtained was representative of their child’s expressive language ability.

The MLU was obtained for each child from this language sample.

d) Pre-school Language Scale-3UK (PLS-3UK) (UK adaptation, Boucher & Lewis, 1997) - expressive and comprehension sections.

Both the Reynell Scales and the Pre-school Language Scale are standardised assessments, providing measures of reliability and validity.

e) Portsmouth HealthCare NHS Trust speech & language therapy outcome code. This is a numerical grading system of severity of problem on a scale from 1-15, with 1 representing no problems and 15 representing the most severe problems (see Figure 1).

Data were collected for two visits for each patient. For the general care group this represented their entire treatment. For the treatment group the data represents only a partial sample. This was considered appropriate for the purposes of this pilot study in order to reduce the burden on the participating therapist and patients.

Direct treatment costs were estimated based on the standard resource cost assumptions from the decision makers perspective (ie. the Trust’s perspective). A record sheet was designed to enable the therapists to record the resources they used inclusive of their own time. Specifically, the therapists were asked to record time spent in consultation with the parent, writing up case notes and liasing with other professionals. Use of telephone, stationary and other miscellaneous items associated with a meeting but related to a specific patient were also recorded. Cost data for each setting were obtained from the Financial Directorate for the Trust. These costs included labour costs, administrative costs and overhead costs. Costs that were identical in the two groups (eg. report write up) were not estimated.

There were three grades of therapist in the study; generalist, specialist and chief. For consistency, an average staff cost (per hour) was applied equally to both forms of delivery as there were no a priori reason to assume that one form of delivery required a higher grade therapist than the other. Financial capital and overhead costs provided by the Trust represented the accounting costs of the whole clinic. Hence, it was necessary to apportion these costs. Inclusion of capital/overhead costs in the study implicitly assumes that the clinics were operating at full capacity.

No costs were estimated for parents who did not attend. A further assumption of the study was that resource use was recorded on a per patient basis with the exception of the time in consultation for the treatment group as this was inclusive of more than one patient. The estimation of marginal costs is problematic where indivisibilities are experienced (as in the case of the treatment group). Hence, in order to facilitate a comparison in resource use between the control and treatment group it is necessary to compare resource use of the delivery of the treatment (ie. consultation plus additional resource use) on an average per patient basis over the entire period.

An advantage of economic analysis is that it allows a wider perspective than simply the viewpoint of the Trust to be considered. The viewpoint of the Trust will determine the best mix of treatments that it provides subject to its own budget constraint. However, a broader societal perspective would necessitate the collection of costs borne by patients and their families. One category of such direct costs is patient out-of-pocket expenses. The delivery of the therapy to the parents in the treatment group meant that they were likely to bear greater additional out-of-pocket expenses compared with the control group since they were visiting the therapist more often. In order to try and estimate these costs a record sheet was designed for parents to record the time it had taken them to get to the therapy session, mode of transport and an estimation of cost if public transport or a taxi was used. A further category of costs borne by the patients and family in this study that could be assumed to differ between the two groups were time lost from work (an indirect cost from a societal perspective). From the data that parents were asked to record employment status is known. However, whilst parents were asked if they were missing work as a consequence of attending the session no salary information was collected.

The small sample size meant that parents attending the control group and the treatment group were not necessarily representative of the same patient population. Direct parent costs are a function of the parent characteristics in the sample rather than a function of the treatment. Hence the small sample would affect the average resources as recorded by either group. To reduce the small sample problem it was considered appropriate to pool the data recorded by parents to determine direct costs incurred by parents in accessing the therapy sessions.

Assuming the groups were homogenous the marginal cost of a visit should be equal between the two groups. Hence, it was considered appropriate to compare the total direct costs incurred by parents of attending either the control group or the treatment group. For this to be a valid comparison it was assumed that parents would attend all therapy meetings for either the control or treatment group. The data on time spent at the clinic by the respective groups was not pooled since this was directly related to the therapy rather than the parent’s characteristics.

Child language abilities were analysed by comparing pre and post intervention scores across the complete range of child language measures taken.

The mean scores were found to improve post-intervention for both the PBI intervention group and the general care intervention group, but with larger gains evident for the PBI group.

Independent two-tailed t-tests were conducted on the pre-intervention scores of the assessment measures to determine whether any significant differences existed between the two groups at the beginning of the study. Table 2 shows a summary of the pre-intervention mean scores for both groups across all assessment measures used, along with the t-test analysis. No significant differences existed between the groups pre-intervention (p<0.05) on any of the assessment measures.

Independent t-tests were conducted on the post-intervention scores of the assessment measures to look for any significant differences between the groups post-intervention. Table 3 shows a summary of the post-intervention mean scores for both groups across all assessment measures used, along with t-test analysis (significance at p<0.05). Table 4 shows a summary of an analysis of the differences in the gains between the groups using independent t-tests.

Across all measures but one (parental estimate of word total), statistical analysis demonstrated a high significance in the greater language gains of the PBI intervention group, pre- to post-intervention, in comparison with the general care group.

Although the post-intervention mean score for parent estimate of word total was considerably higher for the PBI group in comparison with the general care group, the standard deviation for this measure was very large.

Further statistical analysis was carried out using a two-way analysis of covariance (ANCOVA) design. The initial, pre-therapy scores formed one co-variate factor, in order to take account of any initial differences (even though these were found to be non-significant) that may have existed between subjects. The children’s ages formed a second co-variate factor. This analysis allowed examination of the interaction between the initial language levels of each child, the initial age of the child and which treatment was the most beneficial.

Table 5 contains a summary of the analysis of co-variance. A significant difference was still found to exist between the groups post-intervention after taking into account the initial differences between the groups in child age and in any slight pre-intervention scores differences.

In summary, the PBI group obtained greater language gains across all outcome measures and these were found to be statistically significant on all but one measure.

Direct therapy costs are presented in Table 6. Direct therapy costs are indicative of the average cost per patient from the perspective of the Trust. However, as discussed above these do not take account of the patients (parents) cost of attending the sessions.

The direct costs for the treatment group were derived based on a group of eight parents attending.

An attempt was made to collect information that was indicative of the parents’ costs of attending the sessions. This included the amount of time parents spent travelling and the distance travelled to the relevant clinic. These costs are reported in Table 7. Travel costs were imputed based on miles travel and average cost per mile. For consistency the same rate was used for patients as was for the therapist that incurred travel costs.

A comparison of simple costs reveals that the PBI group is the least cost of the two options. A useful analysis in this context, however, would be to compare the additional costs that the PBI group imposes over the general care group compared with the additional benefits it delivers. This is referred to as incremental analysis (Drummond, Stoddart and Torrance 1996).

The above analysis was based on the assumption of eight parents attended a group. While the Trust may have a policy on minimum group numbers this may not necessarily be eight. A sensitivity analysis was conducted to determine the minimum viable group size from the Trust perspective as compared with the control group (Table 8). The treatment group is the least cost option compared to the control as long as there are at least 7 or more members in a group session.

As would be expected the total travel cost and total travel time of the average parent attending a PBI group exceeded those of the parents attending general care.

An attempt was made to collect information on indirect patient costs. The number of parents in the total sample in employment was small. However, none of these recorded that they missed work as a consequence of attending the sessions. Whilst this implies the indirect costs in term of productivity losses were zero, from a societal perspective, the attempt made to estimate these was relatively unsophisticated so no robust conclusions can be drawn.

The aim of this study was to examine and evaluate the effects and costs of two different interventions for pre-school language delayed children.

The results indicate that children who received parent-based language intervention made significantly greater language gains than children who received general care intervention.

In addition, the results of this study which involved two independent clinicians running the PBI groups compare favourably with results obtained in previous research (Gibbard, 1992;1994).

An advantage of economic analysis is that it allows the costs of different delivery of treatment to be considered from different perspective. From the Trust perspective PBI is the least cost alternative. The next stage of this study is to combine cost and outcome data.

The collection of direct cost data from the parent allows a broader perspective to be taken. In this study, the results indicate that parent’s cost would increase significantly given the introduction of PBI as standard practice. However, from the parents’ viewpoint this is only one part of the equation. The benefits of the attending the therapy in terms of increased language skills of their child may compensate for the increased costs in accessing the therapy.

The results of this study indicate that parent-based intervention provides a cost-effective service for language-delayed children. Parent-based intervention was cheaper to provide than general care treatment, with the language gains made by the children being significantly greater. Previous research (Gibbard 1992;1994) showed that parent-based intervention was also as effective as direct intervention on an individual basis between the clinician and child. Parent-based intervention therefore also offers a more cost-effective method of intervention than individual, direct treatment.

A recent systematic review of the literature (Law et al., 1998) into developmental speech and language impairments concluded that in approximately 60% of children with purely expressive language delays, those delays will resolve spontaneously before the age of three. However, one of the key difficulties for speech and language therapists is identifying which are the children whose difficulties will be likely to resolve and which are the children whose difficulties are likely to persist.

Alongside this, over the years, there has been emphasis on providing early intervention (Law, 1997). But is early identification and intervention justified?

Increasingly, children are being referred for speech and language therapy assessments at an earlier age. Although the clinician is then able to make an early diagnosis of developmental language delay, there follow important clinical decisions to be made. Firstly, the clinician must decide if and when a child will benefit from intervention and secondly, the form of intervention that is likely to be most effective.

Past research has indicated that many children with early language delays will achieve normal language levels by the age of 5 to 6 years (eg. Bishop & Edmundson, 1987; Whitehurst & Fischel, 1994). However, there is evidence that although this may appear to be the case, longer-term outcomes may be much less favourable. Scarborough & Dobrich (1990) report research indicating that in later childhood, from 28% to 75% of children whose pre-school language was delayed, will exhibit residual speech/language problems, and from 52% to 95% will go on to have reading impairments.

Some researchers have argued that children with language delay should be diagnosed later, due to instability of early language delay (eg. Stark, Mellitis & Tallal, 1983). However, other researchers, such as Silva, Williams & McGee (1987) have concluded that children with early language delay are an exceptionally high-risk group, with high rates of later developmental disadvantage (eg. reading difficulties) and behavioural problems.

There is also debate in the literature over the long-term benefits of early language intervention. Paul & Ellwood (1991) recommend treatment for young children with specific expressive language delay (ELD), whereas Whitehurst & Fischel (1994) believe that ELD is self-correcting and argue against the provision of treatment.

There is evidence to suggest that pre-school children experiencing language difficulties are at risk for later difficulties, in areas such as social skills, reading and writing attainments and later language problems (eg. Padget, 1988). Enderby and Emerson (1995) estimate, based on the current research available, that 40 to 75% of pre-schoolers with language impairment will continue to experience difficulties related to this impairment, even beyond their school years.

Even Whitehurst & Fischel (1994) concede that children who have shown evidence of a specific language delay in the pre-school period, but whose language skills are in the normal range by 5 years, may continue to be weaker in language and reading than other areas of development.

As Fey, Catts & Larrivee (1995) comment, it may be more productive to view language impairment in pre-schoolers not only for what it is at present, but also for what it is likely to become as the child gets older.

In terms of the cost efficacy, Rosetti (1996) reports data in support that early intervention saves money, suggesting a 7:1 cost benefit ratio, with long-term benefits to not only the participant and family involved, but society in general. He summarises that those children identified early and those children whose caregivers are involved in the provision of intervention demonstrate a better overall outcome.

Whitehurst, Fischel, Arnold & Lonigan (1992) point out that even for children for whom expressive language delay is self-correcting, intervention may be justified. Some medical conditions are self-correcting (eg. a common cold), but treatment of the symptoms it produces are considered valuable in alleviating discomfort and stress.

Given that the developmental course of expressive language delay is variable and that there is current debate over the potential usefulness of expensive treatment intervention strategies for this condition, then clearly, a cost-effective service delivery model is required. The results of this preliminary study are supportive of the cost-effectiveness of parental intervention programmes employing the setting of specific linguistic targets within the child’s naturalistic setting (such as Gibbard, 1998). The implications of this study are therefore of importance to service providers, as well as service managers, speech and language therapists and other professionals working with pre-school language delayed children.

REFERENCES

Akhtar N, Dunham F & Dunham P (1991) Directive interactions and early vocabulary

development: the role of joint attentional focus. Journal of Child Language, 18, 41-49.

Barnett WS, Escobar CM & Ravsten MT (1988) Parent and clinic early intervention for children with language handicaps: a cost-effectiveness analysis. Journal of the Division for Early Childhood, 12, 290-298.

Beveridge M & Jerrams A (1981) Parental involvement in language development: An evaluation of a school-based parental assistance plan. British Journal of Educational Psychology, 51, 259-269.

Bishop DVM & Edmundson A (1987) Language-impaired 4 year olds: Distinguishing transient from persistent impairment. Journal of Speech and Hearing Disorders, 52, 156-173.

Boucher J & Lewis V (1997) Pre-school Language Scale-3 UK (PLS-3UK). Psychological Corporation, Kent.

Bruner J (1983) Child’s Talk: Learning to use language. Oxford University Press, Oxford.

Conti-Ramsden G (1994) Language interaction with atypical language learners, Gallaway C & Richards B (eds) Input and interaction in language acquisition. CUP, Cambridge.

Drummond MF, Stoddart GL & Torrance GW (1996) Methods for the Economic Evaluation of Health Care Programmes. Oxford: Oxford Medical Publications.

Eiserman WD, McCoun M & Escobar CM (1990) A cost-effectiveness analysis of two alternative program models for serving speech-disordered preschoolers. Journal of Early Intervention, 14, 297-317.

Eiserman WD, Weber C & McCoun M (1992) Two alternative program models for serving speech-disordered preschoolers: a second year follow-up. Journal of Communication Disorders, 25, 77-106.

Enderby P & Emerson J (1995) Does speech and language therapy work? A review of the literature. Whurr Publishers, London.

Fey ME, Catts HW & Larrivee LS (1995) Preparing pre-schoolers for the academic and social challenges of school. In Masland RL & Masland MW (eds) Preschool prevention of reading failure. York Press:Parkton, MD.

Fey ME, Cleave PL, Long SH & Hughes DL (1993) Two approaches to the facilitation of grammar in children with language impairment: An experimental evaluation. Journal of Speech and Hearing Research, 36, 141-157.

Galloway C & Richards B (eds) (1994) Input and interaction in language acquisition. CUP, Cambridge.

Gibbard DJ (1992) An evaluation of parental-based intervention with pre-school language-delayed children. Unpublished PhD Thesis, University of Portsmouth.

Gibbard DJ (1994) Parental-based intervention with pre-school language-delayed children. European Journal of Disorders of Communication, 29, 131-150.

Gibbard DJ (1998) Parent-based Intervention Programme - A group approach for language-delayed children. Winslow Press, Oxon.

Hall D (ed) (1996) Health for all children (3rd edition). Oxford University Press.

Harris M (1992) Language experience and early language development - from input to uptake. Lawrence Erlbaum, Hove.

Jefferson T, Demicheli V & Mugford M (1996) Elementary Economic Evaluation in Health Care. London: BMJ Publishing Group.

Law J (ed) (1992) The early identification of language impairment in children. Chapman & Hall, London.

Law J (1997) Evaluating intervention for language-impaired children: A review of the literature. European Journal of Disorders of Communication, 32, 1-14.

Law J, Boyle J, Harris F, Harkness A & Nye C (1998) Screening for speech and language delay: a systematic review of the literature. Health Technology Assessment 1998; 2 (9).

McConkey R, Jeffree D & Hewson S (1979) Involving parents in extending the language development of their young mentally handicapped children. British Journal of Disorders of Communication, 14, 203-218.

Padget SY (1988) Speech and language impaired three and four year olds: A five year follow-up study. In Masland RL & Masland MW (eds) Pre-school prevention of reading failure. York Press: Parkton, MD.

Paul R & Elwood T (1991) Maternal linguistic input to toddlers with slow expressive language development. Journal of Speech and Hearing Research, 34, 982-988.

Pearson, VAH (1995) Speech and language therapy: is it effective? Public Health, 109, 143-153.

Reynell JK (1983) Reynell Developmental Language Scales. NFER-Nelson, Windsor.

Rosetti LM (1996) Communication intervention - Birth to three. Singular Publishing Group, London.

Scarborough HS & Dobrich W (1990) Development of children with early language delay. Journal of Speech and Hearing Research, 33, 70-83.

Silva PA, Williams S & McGee R (1987) A longitudinal study of children with developmental language delay at age three: later intelligence, reading and behaviour problems. Developmental Medicine and Child Neurology, 29, 630-640.

Tannock R, Girolametto L & Siegel L (1990) Are the social-communicative and linguistic skills of developmentally delayed children enhanced by a conversational model of language intervention? In Olswang LB, Thompson CK, Warren SF & Minghetti NJ (eds) Treatment efficacy research in communication disorders. Rockville, MD:ASHA Foundation.

Tannock R, Girolametto L & Siegel L (1992) Language intervention with children who have developmental delays: effects of an interactive approach. American Journal on Mental Retardation, 97, 145-160.

Tomasello M & Todd J (1983) Joint attention and lexical acquisition style. First Language, 4, 197-212.

Warren SF & Kaiser AP (1986) Incidental language teaching: a critical review. Journal of Speech and Hearing Disorders, 51, 291-299.

Whitehurst GJ & Fischel JE (1994) Practitioner review: early developmental language delay: what, if anything, should the clinician do about it? Journal of Child Psychology and Psychiatry, 35, 613-648.

Whitehurst G, Fischel J, Arnold D & Lonigan C (1992) Evaluating outcomes with children with expressive language delay. In Warren S & Reichle J (eds) Causes and effects in communication and language intervention. Baltimore: Paul Brookes.

Based on NHS travel cost per mile to determine an approximate figure for the opportunity cost of travel.