William S. Rosenthal
Department of Communicative Sciences and Disorders, California State University, Hayward, Hayward, California, 94542, USA
A parent rating procedure for young children who stutter is described. The procedure is subjective, requires no objective Counts of behavior, but corresponds well with both clinician assessments and objective SSI-3 scores. These parent ratings can be important sources of confirmation about the progress (or lack thereof) observed by clinicians. They are also useful during breaks in therapy, so that significant or alarming changes can be detected early and timely intervention provided. Graphic and statistical analysis of the data shows good correspondence between SSI-3 changes and parent rating changes over the same period of time.
Stuttering behavior is variable, waxing and waning over periods of weeks, months and years. Clinical measures of stuttering, that might otherwise be reliable, are always cross-sectiona1 slices of behavior that may or may not be representative of a child’s typical behavior. Therefore, the validity of periodic measures is open to doubt. For example, measurements taken in an environment that produces little fluency breakdown may seriously underestimate the severity of stuttering in other, more stressful situations. The subjective ratings that are made by clinicians are similarly plagued by validity issues, and their usefulness is limited by the frequency with which children are seen, and by the situations in which they are observed. Even when accurate counts are made of stuttering behaviors in the clinic, they are often at odds with the anecdotal reports of parents and teachers, who see the child for longer periods of time and in different contexts.
Parents, who have the most frequent and enduring contact with their children, are uniquely able to evaluate their children’s stuttering. Parent ratings have been used in previous research (Yairi et al., 1993), and the Lidcombe program for early stuttering, which requires daily parent measures, using a subjective 1-10 point scale (Harris et al., 2002). The reliability, validity, and utility of the latter procedure were investigated by Onslow, et.al. (1990) on four parent subjects. The authors found the procedure sound, but concluded that parents might need training to use such rating schemes successfully. I would add that the reliability of such ratings depends on the complexity of the parents’ assignment. To expect untrained parents to perform behavioral counts that are comparable to trained professionals is probably unwarranted. Still, parent ratings can be important sources of confirmation about the progress (or lack thereof) observed by clinicians. These ratings are also useful during breaks in therapy, so that significant or alarming changes can be detected early, and timely intervention provided.
Eight children (3 females and 5 males), ranging in age from 3 to 8 years old participated in the study. The median age was 8. Six of the children were at-some period of treatment in a group of up to four participants. For those children, group was the principal treatment modality. Two children received individual therapy only. Treatment duration varied from three months to two years. Two of the children had received stuttering treatment previously in a school setting briefly. One child had received prior language therapy in a pre-school setting and was discharged because he no longer required services.
Like the Lidcombe procedure (Harris, et al., 2002), the rating scheme used in this study was subjective and did not depend on accurate counts of behavior. During each child’s diagnostic evaluation the parents were asked to rate their child’s stuttering for a period that included that day and for the week preceding. They were provided with the following scale. A score of 10 represents their child’s worst stuttering, while a score of 1 represents their child’s best fluency, or the absence of stuttering. A score of 5 indicates stuttering that is average or typical for that child. The parents were asked to provide a number from 1 to 10 that best described their child’s stuttering. This is obviously a relative scale, and scores do not reflect a level of severity that permits the comparison of one child with another. The parents learned to use this scale easily when they were coached to recall the periods of stuttering that caused them most concern (10) and to compare those with the times that they felt most hopeful that stuttering had disappeared (1).
All eight children, were administered the SSI-3 during diagnostic evaluation, and at approximately three month intervals thereafter.
Parent ratings were obtained during the initial evaluation, weekly thereafter during therapy, and weekly during breaks from therapy. The rating periods covered from 21 to 83 weeks (as of June 6, 2003). A total of 40 data points were collected that matched SSI-3 scores and weekly Parent Rating scores. The number of matched pairs varied across participants, ranging from 2 to 8. A statistical analysis was performed, to determine the association between Parent Ratings, the SS1- 3 Total Scores, as well as with each of the SSI-3 subscales; frequency, duration, and concomitant behaviors. Bonferroni corrected correlation coefficients (7) were calculated for each comparison.
3.1 Statistical Analysis
The results are shown in Table l. Moderately high and statistically significant correlations were found between Parent Ratings and SSI-3 total scores, as well as with SSI-3 duration and frequency sub-scales. The parent ratings are more closely associated with the duration of stuttering events than with their frequency, and evidently not at all associated with concomitant behaviors, although those occurred among some participants in the study. The magnitude of these correlations accounts for between 46% (Frequency) and 53% (Duration) of the variance attributable to these relationships.
The large implied error variance terms may reflect the rater reliability problems of the SSI-3 (Lewis, 1995) and the untested reliability of the Parent Ratings.
Table 1. Correlation (r) of Parent Ratings and SSI-3 Total Scores, Frequency, Duration, and Concomitant Behaviors
Concerning validity, it is not clear what criteria parents use in arriving at their ratings, despite the instructions that are given to them. Nevertheless, the mean of associated parent ratings was 4.09 (Median = 4), just below the midpoint of the scale. At the same time, the mean of the SSI-3 total scores was 15.75, corresponding to the 24”â to 40th percentile for both preschoolers and school age children. Both measures, the Parent Ratings and SSI-3 Total scores, appear to have similar relative midpoints. However, it is the lack of exact correspondence between the SS1-3 and Parent Ratings that highlights the potential usefulness of parent ratings in clinical applications. This is best illustrated by viewing the results for the individual participants in the study.
3.3 Graphic Analyses of Parent Ratings and SSI-3 Scores
Each of the following graphs (Figures 1-8) shows the weekly parent ratings for each of the eight children in the study. Sometimes there are gaps when we did not make contact with the parent during that week. The line that courses through each graph is a linear trend line that smoothes the weekly variations. For each participant a companion graph shows the relationship between the parent ratings and SSI-3 total scores obtained during the same week.
Figure 1. JA 5 year old female Rated 10 at evaluation, therapy began the following quarter and the parent rating had dropped to 5. It reached 1 by the end of the term, and continued at that level over the summer: The child was reevaluated and discharged. Figure la. There is a clear correspondence between Parent Ratings and simultaneous SS1-3 scores taken at the beginning of treatment and at the end of the summer '
Figure 2. WT 8 year old male. Early fluctuations in parent ratings become more consistent and move steadily downward. The child was discharged from therapy. Figure 2a. This shows good correspondence between Parent Ratings and simultaneous SSI-3 scores. However, there is a fluctuation in the SSI scores that is not seen in the parent rating.
Figure 3. WM 8 year old female. Some initial parent ratings are higher than most of the other children observed, but these decline and very high levels are not seen again during treatment. Progress is slow, but ratings are gradually decreasing. Treatment continues. Figure 321. There is very good correspondence between Parent Ratings and associated SSI-3 scores. However, in this case, the parent ratings fluctuate more than the SSI scores.
Figure 4. LW 8 year old male: In this pattern of parent ratings, we see highly consistent ratings week after week, and then sudden drops. This may indicate the parent’s lack of sensitivity to subtle changes in this child ‘s speech until some threshold is reached. During the past quarter of therapy, there is evidence of increasing discrimination of week-to-week changes. Treatment continues. Figure 4a. Although there is overall correspondence between the Parent Ratings and SSI-3 scores, especially from the beginning of treatment to the present, the lack of variability in the parent ratings contrasts with the more gradual and regular decline in SSI scores.
Figure 5. FB 8-year-old male.” This shows considerable fluctuations in stuttering behavior as indicated by parent ratings. Although there is a slight, overall downward trend, it does not inspire confidence that this child is speech has stabilized. The anomalous low rating of 2 occurred while the child was on Christmas vacation with relatives in Idaho. We call it the “Idaho effect”. Treatment continues. Figure 5a. While there is general relative agreement between Parent Ratings and SSI-3 scores, especially from high point to low point, the SSI scores suggest a steady, downward trend that may be too optimistic. The corresponding parent ratings show an unstable pattern that may actually reflect a more recent increase in severity.
Figure 6. SA 4 year old male: This is an interesting case, because the child demonstrated dysfluencies that were related to emerging language, and others that were clearly not language related. The wide swings in the parent ratings reflect, in part, the mothers difficulty in distinguishing between these types of dysfluencies. As her discrimination improved, the ratings became more consistent. In addition, over the two quarters that this child was seen, his language developed substantially and the language related dysfluencies diminished significantly. He was reevaluated after a four-week break and discharged. Figure 6a. The comparison of Parent Ratings and SSI-3 scores shows a corresponding decline in both.
Figure 7. PJ 5 year old female.’ This series of parent ratings shows a stable pattern and the trend line is essentially flat. The pre therapy rating of 9 has never been revisited. Until recently, this child seemed locked in a range from 3 to 6. The lower ratings represent those taken during periods of therapy. The higher ratings are those taken during breaks in therapy, over the summer and between quarters. Figure 7a. The comparison of Parent Ratings and SSI-3 scores showed considerable discrepancy. This child ‘s mother saw no change over the first five measurement periods, while the SSI shows a downward trend that suggested improvement. Taped samples of this child ‘s speech recorded at home indicated that the parent ratings probably reflected rather accurately what was going on away from the clinic. This child is speech had improved in the therapy room, but not at home. Recently, particular attention to home intervention has produced improvement on both the child ‘s SS1 score and Parent Ratings. 230 Theory, research and therapy in fluency disorders
Figure 8. LJ 7 year old male.” This series shows high Parent Ratings early in treatment, with only slight improvement acknowledged by his parents so far Figure 8a. Here the change in SSI score is not reflected in the parent ratings, due to the parent’s failure to difierentiate stuttering versus language-based dysfluencies not counted by the SS1.
Thanks to the many therapists who have worked with the children over the course of this study, in both individual and group settings; Shannon Austermann, Erica Burdoin, Kathye Castaneda, April Clifton, Sandra Deane, Andrea Doermann, Mary Eberhard, Nikki Lallis-Natsis, Arezu Moshirian, Kelley Oschegger, Judy Ramos, Annick Rivera, Nancy Thomas, and Winnie Yip. Through their diligence, we were able to collect the test data and parent ratings that were necessary for this report.
Cordes, A.K. & Ingham, R.J. (1994) The reliability of observational data: II. Issues in the identification and measurement of stuttering events. Journal of Speech and Hearing Research, 37, 279-294.
Harris, V., Onslow, M., Packman, A., Harrison, E., & Menzies, R. (2002) An experimental investigation of the impact of the Lidcombe Program on early stuttering. Journal of Fluency Disorders, 27, 203-214
Lewis, K.E. (1995) Do SSI-3 scores adequately reflect observations of stuttering behaviors? American Journal of Speech-Language Pathology, 4, 46-59.
Onslow, M., Andrews, C. & Costa, L. (1990) Parental severity of early stuttered speech: Four case studies. Australian Journal of Human Communication Disorders, 18, 47-61.
Riley, G.D. (1994) Stuttering Severity Instrument for Children and Adults (3rd ed.). Austin TX: PRO-ED.
Yairi, E.; Ambrose, N.G. & Niermann, R. (1993) The early months of stuttering: A developmental study. Journal ofSpeech and Hearing Research, 36 (3), 521-528.
Yaruss, J . S. (1997) Clinical measurement of stuttering behaviors. Contemporary Issues in Communication Science and Disorders, 24, 33-44. 232