EBP THERAPY ANALYSIS for
Single Case Designs
NOTE: The summary of the intervention procedure(s) can be viewed by scrolling about two-thirds of the way down on this page.
C = Clinician
CAS = childhood apraxia of speech
EBP = evidence-based practice
NA = not applicable
P = Patient or Participant
pmh = Patricia Hargrove, blog developer
SLP = speech–language pathologist
SOURCE: Preston, J. L., Leece, M. C., McNamara, K., & Maas, E. (2017). Variable practice to enhance speech learning in ultrasound biofeedback treatment of childhood apraxia of speech: A single case experimental study. American Journal of Speech-Language Pathology, 26, 840-852.
DATE: February 12, 2020
ASSIGNED OVERALL GRADE: B+ The highest possible grade, based on the design of the investigation is A-. The Assigned Grade for Overall Quality represents a judgment about the level of evidence supporting the intervention. It is not a judgment about the quality of the intervention.
TAKE AWAY: This single-case experimental design investigation involved 6 children between the ages 8 and 16 years who had been diagnosed with childhood apraxia of speech (CAS). The results indicate that variable practice in the form of modulated prosody can increase performance compared to interventions without prosodic variability, particularly when the metric was mean effect size rather than raw data. However, both experimental (Prosody variation) and the control (No Prosody variation) yielded positive results.
- What was the focus of the research? Clinical Research
- What type of evidence was identified?
- Whattype of single subject design was used? Single Subject Experimental Design with Specific Client – Alternating Treatment
- What was the level of support associated with the type of evidence? Level = A-
- Was type of treatment concealed?
- from participants? No
- from clinicians? No
- from data analyzers? Yes
- Were the participants (Ps) adequately described? Yes
– How many Ps were involved in the study? 6 What the P characteristics/variables controlled or described?
– CONTROLLED CHARACTERISTICS
- cognitive skills:no more than 1.33 standard deviations below the mean on the Wechsler Abbreviated Scales of Intelligence Matrix Reasoning subtest
- receptive language:no more than 1.33 standard deviations below the mean on the Peabody Picture Vocabulary
- speech skills:below 7th %ile on Goldman-Fristoe and produced speech sounds in conversation
- pretreatment generalization probes:below 25% correct on 2 probes
- hearing:passed pure-tone screen at 1, 2, 4 kHz.
- diagnosis:Childhood Apraxia of Speech (CAS)
– DESCRIBED CHARACTERISTICS
- age:8;2 to 16;8
- gender: 2f; 4m
- sentence repetition:
– 50%- 91% for /s/
– 0% -4% for “r”
- speech sounds:Linguisystems Articulation Test
– Standard Score = less than 58 to 62
– Inconsistency (total = 12) = 1 to 6
- Multisyllablic word repetition task:
– percent consonants correct – 71% to 85%
– percent lexical stress correct = 35% to 80%
– ‘r’ onset percent correct: 0% to 33%
– ‘r’ rhyme percent correct: 0% to 92%
– /s/ rhyme percent correct: 33% (only one scoure
- Dysarthria score:0 to 2 (out of 2)
- Apraxia score:all 2 (out of 2)
- Percent Consonants Correct:71% to 85%
- Syllable Repetition Task:
– Percent Consonants Correct: 76% to 96%
– Percent of Words with Additions: 6% to 28%
- Emphatic Stress Task:38% to 100%
- Inconsistency Task:1.5 to 2.9 (average novel productions_
– Elision Scaled Score: 4 to 10
– Blending Scaled Score: 6 to 12
– Phoneme Isolation Scaled Score: 6 to 10
- Nonword Repetition Task:69% to 86% (Percent Consonants Correct)
– Recalling Sentences Scaled Score: 5 to 13
– Formulating Sentences Scaled Score: 7 to 11
- Severity of CAS:mild to moderately severe
– Were the communication problems adequately described? Yes
– The disorder type: Childhood Apraxia of Speech
– Other aspects of communication that were described: See Described Characteristics (above)
- Was membership in treatment maintained throughout the study? Yes, but not all of the Ps completed treatment during the duration of the investigation.
– If there was more than one participant, did at least 80% of the participants remain in the study? Yes
– Were any data removed from the study? No, even the data from the P who did not progress out of the Prepractice was reported.
- Did the design include appropriate controls? Yes
- Were baseline/preintervention data collected on all behaviors?Yes
- Did probes/intervention data include untrained stimuli?Yes
- Did probes/intervention data include trained stimuli?No
- Was the data collection continuous? Yes
- Were different treatment counterbalanced or randomized? Yes
- Describe here how it was counterbalanced or randomized.
– Each P was assigned 2 speech sound targets to be worked on throughout the investigation.
– The PROS treatment condition was randomly assigned to one of the speech sound targets; the other speech sound was assigned the No-PROS treatment condition. These pairing were constant throughout the investigation.
– For the first session of each week, the order of the speech sounds to be treated was randomly selected. To counterbalance the order of treatment, for the second session of the week the order of the speech sound targets was reversed.
- Was the outcome measure appropriate and meaningful? Yes
- OUTCOME: Accuracy of target sounds in read words with no models or feedback
- Was the outcome subjective?YES
- Was the outcome objective? NO
- Reliability Data associated with the outcome measure:
– OUTCOME: Accuracy of target sounds in read words with no models or feedback—95% agreement among 3 judges who were blinded to the Ps’ status
— Did the target behavior improve when treated? Yes, for the most part_
- OUTCOME #1: Accuracy of target sounds in read words with no models or feedback
– PROS Condition: strong positive response; 79.2 %
– No-PROS Condition: strong positive response; 83.2 %
– PROS Condition: strong positive response; 92.4 %
– No-PROS Condition: strong positive response; 72 %
– PROS Condition: fair positive response; 16.7%
– No-PROS Condition: small positive response; 7.2%
– PROS Condition: fair positive response; 14.4%
– No-PROS Condition: fair positive response; 22.7%
– PROS Condition: fair positive response; 12.1%
– No-PROS Condition: negative response; 4.2%
– PROS Condition: small positive response; 14.2%
– No-PROS Condition: negative response; -3.6%
- Description of baseline:
– Were baseline data provided? Yes
– Was baseline low (or high, as appropriate) and stable?
- DANICA: low and stable
- ETHAN: stable
- FINN: low and stable
- GREG low and stable
- HANNAH: low and stable
- ISAAC: Pros was low and stable; No-Pros was unstable
– Was the percentage of nonoverlapping data (PND) provided? No
- What is the clinical significance?
– OUTCOME: Accuracy of target sounds in read words with no models or feedback
- magnitude of effect:PROS = 46.35; No-PROS = 36.63
- measure calculated:d2
- interpretation: Although both conditions yielded strong positive improvement, the raw data (% correct) indicated that the No-PROS condition was superior. However, the investigators preferred the effect size data which indicated that PROS was superior because there was marked variability in Danica’s performance.
- magnitude of effect:PROS = 27.5; No-PROS = 8.71
- measure calculated:d2
- interpretation:Both metrics resulted in strong improvement but the PROS condition was superior to the No-PROS condition.
- magnitude of effect:PROS = 27.5; No-PROS = 8.71
- measure calculated:d2
- interpretation:The results indicated fair to moderate improvement. Yhe PROS condition was superior to No-PROS.
- magnitude of effect:PROS = 3.15; No-PROS = 1.42
- measure calculated:d2
- interpretation:Both conditions yielded fair positive improvement. Referencing the raw data (% correct), No-PROS condition was superior. However, the investigators preferred the effect size data which indicated that PROS was superior because there was marked variability in Greg’s performance.
- magnitude of effect:PROS = 2.27; No-PROS = 1.51
- measure calculated:d2
- interpretation: The metrics resulted in fair to small improvement. The metrics also revealed that PROS condition was superior to the No-PROS condition.
- magnitude of effect:Pros = 4.00; No-Pros = -0.19
- measure calculated: d2
- interpretation:The metrics revealed a small positive increase for the PROS condition and a small decrease for the No-PROS condition using the raw data and d2. Isaac never progressed out of the Prepractice phase of treatment.
- Was information about treatment fidelity adequate? Yes. The investigators defined treatment fidelity as the percentage of the time appropriate verbal feedback was provided to the P.
- A research assistant listened to 2 randomly selected audiotapes of sessions for each of the Ps.
- The clinicians (Cs) provided correct feedback 98.6% of the time.
- The investigators also measured the interrater reliability between the Cs and the research assistant regarding the correctness of the Ps’ performances during treatment. The overall interrater reliability was 94.1%.
- Were maintenance data reported? Yes
- Although maintenance data were provided, the investigators’ discussion was brief.
- Four of the 6 Ps returned 2 months after the termination of intervention for post-intervention probes. All 4 Ps showed evidence that they maintained their progress.
- Were generalization data reported?Yes
- The probes used to track progress were generalization probes. That is, Ps read aloud a list of words that had not been part of the training protocol and received neither models nor feedback. Accordingly, all the probe data can be considered to be generalization data.
- The probes were administered
– before and after the course of treatment
– at the beginning of every other treatment
– 2 months follow-up (see Item 12)
- The results of the probing indicate that the Ps’ progress varied from strong to small and that the PROS condition tended to result in more progress than the No-PROS condition.
- The investigators provided reliability data for the judgments of correct/incorrect on the generalization probes for 3 judges who were blinded to the treatment condition and the session from which the data were extracted. Overall, interrater reliability was 0.56 using the Fleiss kappa.
- Brief description of the design:
- Six children with CAS were involved in this investigation. Several speech and language assessments were administered to the Ps prior to the interventions documenting the Ps’ speech and language functioning.
- Using an alternating treatment single-case experimental design, the investigators administered the intervention controlling for dosage and order of treatment.
- Progress was measured by probing the Ps behaviors pre intervention, during intervention, post intervention, and following the termination of intervention. The number of pre and post probing sessions varied from 2 to 5 sessions. In addition, probes were administered every other session during treatment and 2 months after the post intervention sessions.
ASSIGNED OVERALL GRADE OF THE QUALITY OF SUPPORT FOR THE INTERVENTION: B+
SUMMARY OF INTERVENTION
PURPOSE: to investigate the use of prosodic variation during a biofeedback intervention for speech sound errors
POPULATION: Childhood Apraxia of Speech; Children, Adolescence
MODALITY TARGETED: Expression
ELEMENTS OF PROSODY USED AS INTERVENTION: terminal contour, rate, loudness
OTHER ASPECTS OF LANGUAGE/COMMUNICATION TARGETED: speech sounds
DOSAGE: individual sessions, 2 one- hour sessions per week for 7 weeks (each treatment received ½ hour treatment per week)
ADMINISTRATOR: Speech-language pathologist
- The targets were
- /r/ before the vowel in the syllable, including blends (onset) and
- /s/ or /r/ including s-blends at the end of the syllable (coda/rhyme) rhyme
Each treatment session involved targets from 1 and 2 with the specific targets being chosen specifically for each P.
- During the course of treatment, P’s targets progressed from easier to more difficult:
– vowels and consonant cluster syllables (e.g., /-st/, /br-/) or consonant and vowel syllables (/ri/, /is/)
– single syllable words (e.g., read)
– multisyllable words (e.g., reading)
– in a phrase short phrase (e.g., I’m reading)
– in self-generated sentences (e.g., P produces a sentence with the short phrase as a core and spontaneously extends it “I’m reading a new book today.)
- During each production portion of each treatment session, Ps received feedback with Ultrasound for part of the section for the other part of the section, they did not receive Ultrasound feedback.
- The were 2 variation treatments or conditions under investigation:PROS and No-PROS. The treatment protocols were similar for these variation conditions except that the PROS treatment/condition included modeling of prosodic variations paired with the speech sound targets and No-PROS involved only the modeling of the speech sound target. (Prosodic variations are described more fully below.)
- For each of the Ps, one of the conditions (PROS) was randomly assigned to one of the targets (e.g., /r/ onset) throughout the course of treatment. The other condition (No-PROS) then was assigned to the remaining target (e.g., /s/ or /r/ coda/rhyme). The pairings of targets and conditions remained constant throughout the intervention. Each day the Ps received 20 minutes of a target paired with the PROS condition and 20 minutes of a target paired with the No-PROS condition.
- The time schedule for each session included 2 sections with each section pairing a target (onset target or coda/rhyme target) and a treatment/condition (i.e., PROS or No-PROS.) The time schedules for the 2 sections of a session were identical:
– Auditory Perception Training (8 minutes, 50 trials)
– Production Training for 1 pairing of target and treatment/condition
∞ Ultrasound used (10 minutes)
∞ No Ultrasound used (10 minutes)
- The investigators randomly assigned the order of target/condition for the first session of each week and then they alternated to the other target/condition as the initial part of the second session of each week.
- The treatment protocols were complex and the summary below is incomplete. However, the investigators provide supplemental information online to assist in the interpretation of the protocols.
AUDITORY PERCEPTION TRAINING
- Both sections of a treatment began with Auditory Perception Training which lasted approximately 8 minutes.
- The clinician (C) played recordings on a computer of different speakers producing correct and incorrect 50 versions of the target sound. P judged the accuracy of the productions and received feedback from the computer.
- There were multiple modules of speakers and production to be used in different sessions.
- ULTRASOUND VISUAL FEEDBACK TRAINING: This occurred during the first 10 minutes of each section.
– C provided visual and verbal feedback to Ps regarding the acceptability of the production of the target.
– C used an Echo Blaster to provide visual feedback to Ps. This allowed the Ps to see a visual representation of their production of a target allowing them to make changes to as appropriate.
– C provided verbal descriptions and drawings of the ideal tongue shapes. C also provided transparencies of the ideal shape to facilitate Ps’ comparing the production with the ideal.
– For a production to be judged as correct, it need only be acoustically correct. However, the entire syllable needed to be correct for the production to be considered correct.
– There were 2 steps in Production Training which occurred first paired with Ultrasound feedback and then without Ultrasound feedback. (See next section for No Ultrasound Production Training. The Production Training should last 10 minutes
- PREPRACTICE with Ultrasound Feedback: Ps were required to imitate their C’s modeling of 4 different examples of the target (e.g., for /r/ — /ri/, /ro/, /bru/, and /tru/) correctly, 3 times each. Ps received verbal and visual feedback including Ultrasound Feedback after each production. Once Ps met the criterion, they moved to the second step (Structured Practice). The Prepractice generally ccould be completed in 2 minutes. However, Prepractice could continue for the entire 10 minutes, if necessary. See note at the bottom of Structured Practice with Ultrasound Feedback, if criterion was not achieved during the first 10 minutes of practice on a speech sound target.
- STRUCTURED PRACTICE with Ultrasound Feedback: For the remaining first 10 minutes, Ps still received feedback from C using verbal and visual feedback (including Ultrasound Feedback) and
∞ C modeled the targets in blocks of 6 attempts of the target, starting with syllable levels targets (e.g., /ri/). If Ps met the criterion of 5 out of 6 correct productions, the C changed the target to the monosyllabic word level, then the multisyllabic word level, and so forth each time Ps met the criterion
∞ If Ps did not reach the criterion of 5 out of 6 attempts, C cycled to a different target sound at the syllable level.
NOTE: If Ps did not reach Prepractice criterion during the first 10 minutes , Prepractice continued into the second 10 minutes; however, during the second 10 minutes; however, Ultrasound feedback as not was not used.)
- NO ULTRASOUND VISUAL FEEDBACK TRAINING:This was scheduled to occur during the final 10 minutes of each section.
– This involved the Structured Practice that mirrored the Structured Practice in the Ultrasound visual feedback training except
∞ feedback did not include visual feedback training
∞ in each block of 6 repetitions, P self-evaluated accuracy 3 times.
PROSODY VERSUS NO-PROSODY CONDITIONS
- Recalling that one of the 2 speech sound targets for each of the Ps was permanently paired with either the PROS or No-PROS condition, the differences in the 2 conditions were at the Structured Practice step.
- For the PROS condition during Structured Practice, C explained to P that they would be practicing speech sound production with different voices (i.e., prosodic cues). During the PROS condition, C modeled the target speech sound at the appropriate linguistic level (e.g., monosyllabic words, multisyllabic word, etc.) using a predetermined prosodic variation of the target:
– fast, or
- The number of trials including modeling of prosodic cues varied based on the linguistic complexity of the target:
– monosyllabic words = 2 prosodic cues in the block of 6 trials
– multisyllabic words = 3 prosodic cues in the block of 6 trials
– phrases = 6 prosodic cues in the block of 6 trials
– self generated sentences = 6 prosodic cues in the block of 6 trials
- A correct response required only the accurate production of target consonants and vowels (not prosody).
- The No-PROS condition involved C modeling the target using a neutral prosody.
- The investigatorsprovided a scoresheet in the Appendix as well as sample treatment videos in Supplemental Materials