Professional Documents
Culture Documents
Review Article
Purpose: To present a systematic review of single-case or augmentative and alternative communication (n = 2). Most
experimental treatment studies for childhood apraxia of participants responded positively to treatment, but only 7 of
speech (CAS). 13 approaches in SCED studies reported maintenance and/or
Method: A search of 9 databases was used to find generalization of treatment effects. Three approaches had
peer-reviewed treatment articles from 1970 to 2012 of all preponderant evidence (Smith, 1981). IRD effect sizes were
levels of evidence with published communication outcomes calculated for Integral Stimulation/Dynamic Temporal and
for children with CAS. Improvement rate differences (IRDs) Tactile Cueing, Rapid Syllable Transition Treatment, and
were calculated for articles with replicated (n > 1), statistically Integrated Phonological Awareness Intervention.
compared treatment and generalization evidence. Conclusions: At least 3 treatments have sufficient evidence
Results: Forty-two articles representing Phase I and II for Phase III trials and interim clinical practice. In the future,
single-case experimental designs (SCEDs; n = 23) or case efficacy needs to be established via maintenance and
series or description studies (n = 19) were analyzed. generalization measures.
Six articles showed high CAS diagnosis confidence. Of the
13 approaches within the 23 SCED articles, treatments were Key Words: dyspraxia, intervention, efficacy,
primarily for speech motor skills (n = 6), linguistic skills (n = 5), methodological rigor, generalization
C
hildhood apraxia of speech (CAS) is a developmen- especially in the realization of lexical or phrasal stress” (ASHA,
tal disorder of speech motor planning and/or pro- 2007, p. 4).
gramming (American Speech-Language-Hearing Although impaired movement planning and program-
Association [ASHA], 2007). It is also known as develop- ming are considered to underlie CAS, there are also reports of
mental verbal dyspraxia in the United Kingdom (excluding disrupted development of speech perception, language, and
being the result of any known neurological disorder) and phonology (including phonological awareness) in children
has previously been called developmental apraxia of speech with CAS (Groenen, Maassen, Crul, & Thoonen, 1996; Lewis
and dyspraxia. CAS causes reduced speech intelligibility et al., 2004; Maassen, Groenen, & Crul, 2003). It is unclear
because of a hypothesized impairment in the “transformation whether these are primary deficits or flow-on effects from
of an abstract phonological code into motor speech com- CAS, comorbid impairments, or perhaps compensatory be-
mands” (Terband, Maassen, Guenther, & Brumberg, 2009, haviors, as children with CAS develop their linguistic, pho-
p. 1598). Such impairment leads to the current consensus- nological, and motor skills concurrently (Alcock, Passingham,
based core CAS features of “(a) inconsistent errors on con- Watkins, & Vargha-Khadem, 2000; Marion, Sussman, &
sonants and vowels in repeated productions of syllables or Marquardt, 1993; Ozanne, 2005). Children with CAS can
words, (b) lengthened and disrupted coarticulatory transitions therefore present with a range of difficulties requiring therapy
between sounds and syllables, and (c) inappropriate prosody, from speech-language pathologists (SLPs; Royal College of
Speech and Language Therapists, 2011).
a
The long-term functioning of people with CAS is
University of Sydney, New South Wales, Australia largely unreported. The available longitudinal studies suggest
Correspondence to Elizabeth Murray: that CAS is a persistent disorder that requires therapy (Hall,
Elizabeth.murray@sydney.edu.au Jordan, & Robin, 1993; Jacks, Marquardt, & Davis, 2006;
Editor: Carol Scheffner Hammer Stackhouse & Snowling, 1992). Children with CAS, like others
Associate Editor: Ken Bleile with persistent speech sound disorder, are also at risk for
Received March 27, 2013
Revision received July 7, 2013
Accepted December 1, 2013 Disclosure: The authors have declared that no competing interests existed at the
DOI: 10.1044/2014_AJSLP-13-0035 time of publication.
486 American Journal of Speech-Language Pathology • Vol. 23 • 486–504 • August 2014 • A American Speech-Language-Hearing Association
literacy, academic, social, and vocational difficulties (e.g., Smith’s (1981) level of certainty hierarchy considers the
Lewis et al., 2004; Moriarty & Gillon, 2006). research design and the possible effects of the intervention
A Cochrane systematic review, a subsequent journal to provide an overall judgment on how likely the results are to
article, and a treatment review have reported no published be true. For the CAS literature, this can also be extended
randomized controlled trials (RCTs) or nonrandomized to include confidence in diagnosis. SCEDs can show early
controlled trials (NRCTs) for any intervention for CAS evidence for an intervention, with results from such studies
(Morgan & Vogel, 2008, 2009; Watts, 2009). Despite this being classed as suggestive or “possibly true.” However,
lack of high-level evidence, many published articles on the statistically compared outcomes from SCEDs with confident
treatment of CAS could facilitate practice and could help CAS diagnoses, replication (n > 1), and evidence of both
identify potential lines of further research. Narrative reviews treatment effects and generalization of treatment effects could
have identified a range of treatment methods for children be considered preponderant evidence or “probably true.”
reported to have CAS, likely reflecting the diversity of symp- SCED designs are usually designated as Phase II and Level IIb
toms seen in these children and potentially the research and evidence (but see Hegde, 2007; Kearns & de Riesthal, 2013).
clinical interests of the authors (ASHA, 2007; Strand & Currently, Phase III studies (Robey, 2004) are typically RCTs
Skinder, 1999). They encompass motor treatments (including and NRCTs. These generate Level IIa evidence using groups
electropalatography), linguistic approaches, augmentative of participants to reduce bias and to eliminate individual
and alternative communication (AAC), or some combina- variance as a factor in treatment success. It is only through
tion thereof (ASHA, 2007; Gillon & Moriarty, 2007; Hall, meta-analyses and systematic reviews of several Phase III
2000; Morgan & Vogel, 2008). Few of these lower level studies of a given treatment approach, coupled with Phase IV
treatment studies have been examined rigorously, as they effectiveness studies in real-world clinical situations, that
were excluded on the basis of quality in previous systematic results can be defined as conclusive or “undoubtedly true”
reviews (Morgan & Vogel, 2008; Watts, 2009). Provision (Smith, 1981). As all studies of CAS treatments to date are
of recommendations regarding which treatments have sup- classified as Phase I and Phase II, the goal of this review is
portive evidence has therefore not been possible (ASHA, to identify treatment approaches with suggestive or prepon-
2007; Morgan & Vogel, 2008; Pannbacker, 1988). This article derant evidence.
presents a systematic review of all levels of evidence that Central to this review is treatment efficacy. Efficacy
may be critical to inform clinical practice until high-level considers clinical cause-and-effect relationships between
evidence becomes available. the provision of intervention and change in participant be-
Two primary challenges face a systematic review of havior (e.g., McReynolds & Kearns, 1983; Olswang & Bain,
intervention for CAS. The first challenge is in identifying the 2013). Demonstration of efficacy extends beyond treatment
rigor of each study in terms of research phase (Robey, 2004), effects, requiring assessment of maintenance and generaliza-
the research design, the level of evidence generated (ASHA, tion of treatment effects that signify instrumental change.
2004; Perdices et al., 2006), as well as the level of certainty that Response generalization evaluates a child’s performance on
the effects reported for a given treatment approach are real untrained items that are somehow related to trained items,
(Smith, 1981). Treatment research often follows a develop- to determine whether more widespread change is occurring
mental pathway that is associated with increasing research (Olswang & Bain, 1994). Stimulus generalization assesses
rigor and different research questions. Robey (2004) defined performance on untrained materials, people, or settings/
five phases in a research program. Phase I and Phase II environments (Olswang & Bain, 1994). Such change is neces-
studies represent pilot or feasibility studies seeking to deter- sary to meet the overall goals of an intervention.
mine whether effects justify more rigorous study. These can
generate Level III or Level IIb evidence (ASHA, 2004;
Perdices et al., 2006). Level III evidence constitutes quasi-
Aims
experimental group (case series) or single case reports with This systematic review evaluated studies of interven-
pre- to posttreatment measurement and no within-subject tion, published between 1970 and October 2012, that state
comparison or control conditions. Level IIb evidence comes an intention to treat children with CAS. The aims fall into
from more rigorous single-case experimental designs (SCEDs) four broad areas:
that systematically apply and withdraw treatment and es- 1. Study quality: to describe for each identified study the
tablish control using a stable baseline phase or limited change research phase, the level of evidence, and the level of
in control conditions (Byiers, Reichle, & Symons, 2012; confidence in CAS diagnosis;
Olswang & Bain, 1994; Perdices & Tate, 2009).
The second challenge is that there is not yet a validated 2. Treatment procedures: to define the behavioral goals
assessment tool for diagnosing CAS. Thus, before evaluation and structure of treatment (e.g., intensity and dosage
of the treatment in any intervention study, the descriptions according to Warren, Fey, & Yoder, 2007) for each
of participants must be scrutinized to determine the level of SCED study (at/above Level IIb evidence) and to
confidence in the authors’ diagnosis. This involves determin- group similar treatment types to facilitate treatment
ing to what extent the participants are described as meeting outcome analysis;
the three consensus-based core features of CAS listed earlier 3. Treatment outcomes: to examine reported treatment,
(ASHA, 2007). maintenance, and generalization outcomes; and
impaired speech motor planning and/or programming. Non- CAS+, in which CAS was the primary diagnosis, but other
discriminative features were those shared with other dis- disorders were present. A rating of 1 indicated high con-
orders, such as poor intelligibility, slow progress, or delayed fidence in CAS diagnosis, and a rating of 5 indicated no
language (ASHA, 2007; McCabe et al., 1998). Clear cases confidence. Intrarater reliability (first author) on four judg-
of comorbid disorders were also noted, such as receptive ments (presence of each of the three primary features and
language impairment or dysarthria. On the basis of this CAS diagnosis) for 83/83 children studied was 94%. Inter-
analysis, participants were classified either as CAS only or as rater reliability between the first and second authors for a
Level 1 All the primary characteristics were CAS was the primary diagnosis CAS without another comorbid disorder
described as follows: described. Any characteristics that was reported (excluding expressive
• Inconsistency and were attributable to other disorders language delay).
• Lengthened and disrupted may have been described but
coarticulatory transitions between were not used to diagnose CAS.
sounds and syllables and
• Inappropriate prosody.
Level 2 All the primary characteristics were CAS was the primary diagnosis This includes the following:
described as follows: described. Other characteristics • CAS without another comorbid
• Inconsistency and attributable to other disorders (e.g., disorder (excluding expressive
• Lengthened and disrupted dysarthria) were described and may language delay) or
coarticulatory transitions between have been used to diagnose CAS. • Clear cases of comorbid CAS,
sounds and syllables and in which CAS was the primary
• Inappropriate prosody. diagnosis.
Level 3 Two of the three primary characteristics CAS was described. Other This includes the following:
were described: characteristics that were attributable • CAS without another comorbid
• Inconsistency and/or to other disorders (e.g., dysarthria) disorder or
• Lengthened and disrupted were described and may have been • Clear cases of comorbid CAS, in
coarticulatory transitions between used to diagnose CAS. which CAS was the primary
sounds and syllables and/or diagnosis or
• Inappropriate prosody. • Cases of CAS in which another
comorbid disorder had the same
severity (e.g., language delay,
dysarthria).
Level 4 Only one of the three primary Other characteristics that were Unclear whether CAS was the primary
characteristics was reported, or attributable to other disorders may diagnosis.
incomplete/inadequate description of have been described, and it is unclear
the primary characteristics of CAS whether these were used to diagnose
was provided. CAS.
Level 5 Diagnosis of CAS was reported or Unclear whether CAS diagnosis was Other comorbid disorders may be
implied, but no primary characteristics likely and/or whether CAS was the present.
were described. primary diagnosis.
random 33/83 children was 91%. Discrepancies were resolved Aim 3: Treatment Outcomes
by consensus. Reported treatment, maintenance, and generalization
Exclusions. Articles that lacked experimental control outcomes for each intervention were analyzed for (a) number
(Level III evidence; n = 19) were excluded, as they could of participants with a treatment gain (change immediately
not be used to determine treatment outcomes. The remain- after treatment compared with baseline); (b) assessment
ing 23 Level IIb articles were analyzed to address Aims 2, 3, measures and statistics used in determining treatment effects;
and 4. No articles were excluded because of confidence in (c) maintenance of treatment gains at least 2 weeks post-
CAS diagnosis; however, confidence in CAS diagnosis was treatment, from report or by comparing treatment data with
a factor in determining certainty of evidence—see Aim 4 performance in maintenance probes; (d) response generaliza-
below. tion data, when statistical analysis was used; and (e) stimulus
generalization data.
Aim 2: Treatment Procedures
Articles designed with adequate experimental control Aim 4: Certainty of Evidence
(see Aim 1 above) were analyzed descriptively regarding the Smith’s (1981) three levels of certainty were applied, on
nature of the treatment. Using the stated treatment goals, the basis of design (i.e., level of evidence, research design,
selected stimuli, and specific cueing strategies reported, treat- confidence in CAS diagnosis, and statistical comparison) and
ments were categorized as primarily (a) motor, (b) linguistic/ possible effects of the intervention/outcomes. A treatment
phonological (including literacy), or (c) AAC (ASHA, approach was categorized as having preponderant evidence
2007; Gillon & Moriarty, 2007; Hall, 2000; Martikainen & when it showed Level IIb or better evidence (SCEDs), rep-
Korpilahti, 2011). The structure of treatment delivery was licated cases, diagnostic confidence ratings of 1–3, statistically
also determined (Warren et al., 2007), including dose (trials significant treatment and generalization effects (or at least
per session), dose frequency (number of times a dose is moderate effect sizes), and clear maintenance of treatment
provided over days or weeks), and total intervention time gains at least 2 weeks posttreatment. Any Level IIb or better
(number of sessions). When reported, home practice and evidence that did not meet all the above criteria received the
service delivery model were documented. lowest rating of suggestive evidence.
Table 2. Research design, level of evidence, participant description, and diagnosis analysis of the 23 Level IIb (SCED) articles.
Aided AAC modeling Binger and Light (2007) Multiple baseline across IIb 2/5 (4;2 [years;months] Severe CAS+ (GDD) 4 (dysprosody and
3 participants SCED and 4;4, male, had sequencing NR)
previous SLP)
Binger, Kent-Walsh, Multiple baseline IIb 1/3 (3;4, female, previous Severe CAS+ 4 (dysprosody and
Berens, Del Campo, across probes SCED SLP NR) (suspected VCFS inconsistency NR)
and Rivera (2008) with profound VPI)
Binger, Maguire- Multiple baseline IIb 1/3 (6 years, female, Severe CAS+ (receptive 5
Marshall, and across 3 participants previous SLP NR) and expressive LD)
Kent-Walsh (2011) SCED
Articulation with Stokes and Griffiths (2010) ABA single case IIb 1 (7 years, male, Mild SSD (Hx of CAS) 4 (dysprosody and
facilitative vowel design 1 year previous SLP) nconsistency NR)
contexts
Combined intraoral Lundeborg and ABABABABA single IIb 1 (5;1, female, 1.50 years Severe CAS 3 (dysprosody NR)
stimulation, McAllister (2007) case design previous SLP)
Electropalatography
(EPG) with NDP
Combined melodic Martikainen and Multiple baseline IIb 1 (4;7, female, Severe CAS 1
intonation therapy Korpilahti (2011) across participants 1 year previous SLP)
(MIT) and touch cue SCED
method (TCM)
Combined stimulability Iuzzini and Forrest (2010) Multiple baseline IIb 4 (3;7–6;10, 2 males, Severe CAS 4 (dysprosody and
(STP) and modified SCED 2 females, previous sequencing NR)
core vocabulary (mCVT) SLP for 3/4)
Computer-based Harris, Doyle, Multiple baseline IIb 1 (5 years, male, Severe CAS+ 3 (dysprosody NR)
and Haaf (1996) across discourse approximately 3 years (Hx OME, receptive
contexts SCED previous SLP) and expressive LD)
(table continues)
Table 2 (Continued).
Integral Stimulation/ Strand and Debertine Multiple baseline SCED IIb 1 (5 years, female, Severe CAS (Hx of VPI) 1
Dynamic Temporal (2000) 4 years previous SLP)
and Tactile Cueing Strand, Stoeckel, Multiple baseline SCED IIb 4 (5;5–6;1, all male, Severe CAS+ 4 (for all cases;
(DTTC) and Baas (2006) 2–4 years (2 with mild spastic dysprosody
previous SLP) and/or ataxic and inconsistency
dysarthria, 1 with mild NR, clearly comorbid)
intellectual disability,
and 1 with OME)
Baas, Strand, Elmer, Multiple baseline SCED IIb 1 (12;8, male, 10 years Severe CAS+ (CHARGE 5
and Barbaresi (2008) previous SLP) syndrome intellectual
disability)
Edeal and Gildersleeve- AB—alternating IIb 2 (6;2 and 3;4, male, 6;2—severe CAS+ 1
Neumann (2011) treatments single 1–4 years (repaired CLP, severe
design (with three previous SLP) receptive LD),
stable baselines) 3;4—severe CAS
Maas and Farinella Multiple baseline SCED IIb 4 (5;0–7;9, 2 females, CAS001—moderate– 3 (by consensus);
(2012) 2 males, previous severe CAS; CAS001 = 1, CAS002 = 3
Murray et al.: Systematic Review of Treatment Outcomes for CAS
Table 2 (Continued).
Integrated Phonological Moriarty and Gillon Multiple baseline SCED IIb 3 (6;3–6;10, 2 males, Male 1—Severe CAS+ 3 (Male 1 = 4 inconsistency
Awareness Intervention (2006) 1 female, up to (receptive and and dysprosody NR;
(N = 5) 2 years previous expressive LD); Male 2 and Female = 3)
SLP) Male 2—Severe
CAS+ (receptive
and expressive LD);
Female—mild–
moderate CAS
McNeill, Gillon, and Multiple baseline SCED IIb 12 (4;2–7;6, 3 females, Mild–moderate to 3 (dysprosody NR)
Dodd (2009a) 9 males, previous severe CAS
SLP NR) (no other diagnoses
reported)
McNeill, Gillon, and Multiple baseline SCED IIb 2 (identical twins also CAS (small interstitial 3 (dysprosody NR)
Dodd (2009b) in McNeill et al., deletion on
2009a) chromosome
(4;5, male, moderate– 10 (deletion at
severe) 10q21.2–22.1)
McNeill, Gillon, and Quasi-experimental group IIb Same participants as Mild–moderate 3 (dysprosody NR)
Dodd (2010) (following SCED) McNeill et al. (2009a) to severe CAS
(no other diagnoses
reported)
Rate control therapy Rosenthal (1994) ABAB single case design IIb 4 (10–14 years, 3 males, CAS (severity not 4 (dysprosody and
(with alternating 1 female, all had stated) inconsistency NR)
treatments) previous SLP)
MIT Krauss and Galloway ABAA single case design IIb 2 (6 and 5 years, male, CAS (severity not 4 (dysprosody and
(1982) had previous SLP) stated) inconsistency NR)
Partners in augmentative Culp (1989) ABA single case design IIb 1 (8 years, female, Severe CAS+ 2 (comorbid CAS)
communication 5 years previous SLP) (intellectual disability,
training (PACT) Hx tube insertion,
congenital
heart defect)
Rapid Syllable Transition Ballard, Robin, McCabe, Multiple baseline IIb 3 (7;8–10;10, 2 males, Mild or mild– 1
Treatment (ReST) and McDonald (2010) across behaviors and 1 female, 1–5 years moderate CAS
participants design previous SLP)
Voice output devices Bornman, Alant, and ABA single case design IIb 1 (6;6, male, 2.50 years CAS+ (anoxia causing 5
Meiring (2001) previous SLP) slight left hemiplegia,
grand mal fits)
Note. Please see the online supplemental materials, Supplemental Table 3, for the 19 Level III articles. SLP = speech-language pathology; CAS+ = comorbid childhood apraxia of speech;
GDD = global developmental delay; NR = not reported; VCFS = velocardiofacial syndrome; VPI = velopharyngeal incompetence; LD = language delay/disorder; A = baseline/withdrawal
phase; B = treatment/intervention phase; SSD = speech sound disorder; Hx = history of; NDP = Nuffield Dyspraxia Programme; OME = otitis media with effusion/glue ear; CHARGE
syndrome = coloboma, heart disease, atresia of the choanae, retarded growth and mental development, genital anomalies, and ear malformations and hearing loss; CLP = cleft lip
and palate.
Table 3. Treatment outcomes for the 23 SCED articles.
Integral Stimulation/ 11/13 Rx accuracy Yes—9/13 Yes for 5/7 (6 NR) 2–4 weeks 6/7 (6 NR) NR Preponderant
DTTCd moderate–large post
effect sizes
ReSTe 3/3 Perceptual stress Yes—significant Yes for 2/3 4 weeks 3/3 NR Preponderant
matches effects post
Rate Control 4/4 Rx accuracy No NR NR NR 0/4 to discourse Suggestive
Therapyf
Linguistic Combined STP 4/4 (only 3/4 (1) PCC, No NR NR NR NR Suggestive
with and mCVTg for CSIP)
some
motor
aspects
(2) phones added to
inventory, (3) ↓
inconsistency
(CSIP),
(4) ↓ inconsistency
(ISP)
Integrated 11/15 (1) % suppression of Yes—significant Yes—as group 6 months 11/15 NR Preponderant
Phonological process usage, effects of 12 (3 NR) post
Awareness (PA) (2) PA accuracy
Interventionh
MIT with traditional 2/2 (1) Porch Index of Yes—significant NR NR NR NR Suggestive
therapyi Communicative verbal naming and
Ability in imitation
Children, (2) MLU
(table continues)
495
496 American Journal of Speech-Language Pathology • Vol. 23 • 486–504 • August 2014
Table 3 (Continued).
Linguistic Aided AAC 4/4 Rx frequency Yes—moderate– Yes for all 2, 4, and NR (1) 3/3, Suggestive
with modeling (with large effect sizes 8 weeks
some communication post
AAC board or voice
output devices)j (1) multisymbol (2) NR
messages
(2) morpheme ↑ speech and
accuracy participation,
↓ frustration
Computer-based 1/1 Rx accuracy No NR NR NR NR Suggestive
AACk (1) book reading
(2) discourse
AAC Voice output 1/1 No. of appropriate No Yes 4 weeks NR Spoke intelligibly Suggestive
devices— responses post after 1 year
Macawl
PACTm 1/1 Communicative No NR NR NR ↑ participation Suggestive
effectiveness
(frequency
of turns)
Note. Rx = medical prescription; PCC = percentage of consonants correct; PPC = percentage of phonemes correct; PWC = percentage of words correct; PVC = percentage of vowels
correct; PMLU = phonological mean length of utterance; PWP = proportion of whole-word proximity; CSIP = consonant substitute inconsistency percentage; ISP = inconsistency severity
percentage; MLU = mean length of utterance.
a
Stokes and Griffiths (2010). bLundeborg and McAllister (2007). cMartikainen and Korpilahti (2011). dStrand and Debertine (2000), Strand et al. (2006), Baas et al. (2008), Edeal and
Gildersleeve-Neumann (2011), Maas and Farinella (2012), and Maas et al. (2012). eBallard et al. (2010). fRosenthal (1994). gIuzzini and Forrest (2010). hMoriarty and Gillon (2006) and McNeill
et al. (2009a, 2009b, 2010). iKrauss and Galloway (1982). jBinger and Light (2007) and Binger et al. (2008, 2011). kHarris et al. (1996). lBornman et al. (2001). mCulp (1989).
†
MIT completed in the first block, and TCM completed in the second block.
or traditional articulation therapy (Krauss & Galloway, 1982). and Farinella (2012) and Maas and Farinella (2012) for three
The first was classified as a motor approach because of the of four participants following Dynamic Temporal and Tactile
goals and PML incorporated; the second was classified as a Cueing (DTTC; in any PML condition). The combined
linguistic approach, as MIT was the primary experimental MIT/TCM treatment showed significantly improved percentage
approach, and linguistic outcomes were primarily sought and of vowels correct; however, it significantly reduced percent-
reported. Finally, for one motor treatment, the participant age of consonants correct for the participant immediately
continued regular AAC therapy during the research (Edeal post the first block of MIT. The greatest gains were noted
& Gildersleeve-Neumann, 2011). after withdrawal of treatment. Despite the authors suggesting
this to be due to the treatments given, this equally may be due
Structure of Treatment Delivery to maturation or improvement after withdrawal of treatment
All 23 treatments were delivered individually, with providing unclear evidence to the effect of these treatments.
22 delivered in a clinic and one delivered at the participant’s Six studies of linguistic-based treatment reported a
home (Lundeborg & McAllister, 2007). Caregiver and child treatment effect for speech measures for 17 of 21 participants,
training sessions were utilized in the two AAC studies within a with 16 of 21 participants supported by statistical compar-
consultative-collaboration service delivery model (Bornman ison (Iuzzini & Forrest, 2010; Krauss & Galloway, 1982;
et al., 2001; Culp, 1989). Inclusion of parent training and McNeill et al., 2009a, 2009b, 2010; Moriarty & Gillon, 2006).
home practice protocols or activities was more prevalent in With Integrated Phonological Awareness Intervention, 11 of
AAC-based treatments and was used in six of 23 articles. 15 participants were reported to reduce phonological pro-
For motor treatments reporting dose frequency, the cesses and to improve phonological awareness skills. Another
median was three times a week, with a maximum of once a four articles reported that five of five participants increased
day and minimum of twice a week. Sessions were between use of multisymbol messages (phrases or morphemes) with
20 and 60 min long. Most linguistic and AAC approaches linguistic-based treatment utilizing AAC (Binger et al., 2008;
provided treatment two to three times a week for between Binger & Light, 2007; Binger et al., 2011; Harris et al., 1996).
15- and 60-min sessions. A small number gave intensive daily, Finally, AAC treatment studies focusing on communi-
short-term training. Dose in terms of treatment trials com- cative effectiveness reported treatment effects for two of two
pleted within sessions was adequately described in five of participants, with no statistical analyses conducted (Bornman
23 articles or three of 13 approaches and ranged from 60 to et al., 2001; Culp, 1989). The children reportedly increased
120 trials for motor approaches and from 10 to 30 trials appropriate responses and frequency of turns in conversation.
for linguistic and AAC approaches (Ballard, Robin, McCabe,
& McDonald, 2010; Binger et al., 2008; Binger & Light, Generalization
2007; Binger et al., 2011; Stokes & Griffiths, 2010). Further Seven articles considered response generalization, and
details of the treatment procedure analyses for each study another five considered stimulus generalization. No article
and approach are provided in the online supplemental mate- measured both response and stimulus generalization.
rials, Supplemental Table 3. Response generalization. All treatments measuring
generalization used statistical analysis. For motor-based
treatments, significant improvement in articulation accuracy
Aim 3: Treatment Outcomes for untrained responses was noted for one participant after
Analyses of treatment, maintenance, and generalization facilitative vowel treatment (Stokes & Griffiths, 2010) and for
outcomes for the 23 SCED articles are reported in Table 3. four of seven participants across any behavior/condition in
All studies used baseline phases, and 91% incorporated un- three studies applying Integral Stimulation/DTTC (Edeal
trained control items intended to demonstrate some exper- & Gildersleeve-Neumann, 2011; Maas et al., 2012; Maas
imental control. & Farinella, 2012). Significant generalization was reported
for three of three participants for lexical stress accuracy
Treatment and Maintenance Data in untrained three syllable pseudowords and for one of
All articles reported treatment effects for the majority three participants in untrained real word production for ReST
of the participants, despite a range of goals and measures treatment (Ballard et al., 2010).
being used. Statistical comparison of at least one key outcome Only one linguistic-based approach, the Integrated
was provided for 16 of 23 studies. Phonological Awareness Intervention, reported response
Of the 23 participants given motor-based treatment, 21 generalization. The same 11 of 15 children who demonstrated
were reported to demonstrate positive treatment effects, and a treatment gains also showed significant improvement in
statistical analysis of effects was reported for 17 (see Table 3). speech intelligibility, mean length of utterance, and phono-
Not all participants showed significant changes in all mea- logical awareness skills (phoneme segregation, manipulation,
sures assessed. The majority of measures consisted of percent- nonword reading, reading accuracy, and letter–sound cor-
age of accuracy on treated items or percentage of consonant, respondences; McNeill et al., 2009a, 2009b, 2010; Moriarty
vowel, phonemes, or words correct. Three studies demon- & Gillon, 2006).
strated improvement for treated prosodic accuracy: Ballard Stimulus generalization. Only five articles (22%) re-
et al. (2010) for three of three participants using the Rapid ported stimulus generalization, with four of these utiliz-
Syllable Transition Treatment (ReST) and Maas, Butalla, ing AAC treatments. Three participants, who increased
Motor Integral Stimulation/ Strand and Debertine (2000) 1 1 1.00 [0.97, 1.00] Very large
DTTCb Edeal and Gildersleeve- 2 1 0.98 [0.88, 1.00] Very large
Neumann (2011)
Maas and Farinella (2012) 4c 3 0.18 [0.03, 0.33] Small or questionable
Maas et al. (2012) 4c 2 0.22 [0.08, 0.36] Small or questionable
Overall 0.60 [0.53, 0.67] Moderate
ReST Ballard et al. (2010) 3 1 0.78 [0.54, 1.00] Large (prosody—
PVI duration)
Linguistic Integrated Phonological Moriarty and Gillon (2006) 3 3 1.00 [0.89, 1.00] Very large
Awareness McNeill et al. (2009a) 12 3 0.10 [–0.06, 0.24] Small or questionable
Intervention Overall 0.51 [0.39, 0.58] Moderate
consensus-based diagnostic features (ASHA, 2007) were ap- primarily conveyed on vowels, we counted mention of vowel
plied, only 16.6% achieved a rating of high confidence. How- errors in participants as possible evidence of dysprosody to
ever, most of the studies reviewed were published prior to accommodate older articles. Despite the chance that dys-
2007 and were using common descriptors for their time, now prosody would be overestimated because of this decision (as
considered by many to be nondiscriminative (e.g., ASHA, only a subset of vowel errors or distortions would be indi-
2007; McCabe et al., 1998). The most commonly overlooked cative of stress errors), only 50% of the articles (21 of 42)
CAS characteristic across studies was dysprosody, which reported vowel errors. Furthermore, the high rate of comor-
was not considered a core feature of CAS in many checklists bidity in CAS (41% in this sample) complicates diagnosis and
prior to 2007. As prosody (e.g., lexical or phrasal stress) is could partially account for lower confidence ratings.
Confidence
Therapy Therapy No. of in CAS Dx Generalization to Interpretation
type approach Article cases (1 = highest) untrained itemsa IRD 95% CI of effect
Note. NA = not applicable; MFF = moderate frequency feedback; HFF = high frequency feedback.
a
Generalization items were individualized for each participant. bBaas et al. (2008) and Strand et al. (2006) were excluded from IRD because of
diagnostic confidence ratings of 4 and 5. cThree participants in common across two studies.