Assessment of Egyptian children who stutter using the standardized Arabic form of the Test of Childhood Stuttering

Introduction Stuttering is one of the most common speech disorders affecting children. The proper assessment of stuttering has been a point of interest for researchers over the years. The aim of this study was to adapt the Test of Childhood Stuttering (TOCS) to suit the Egyptian culture and to test the validity and reliability of application of this test for the assessment of Egyptian children who stutter, for proper management of this ailment. Patients This study included 130 children. The ages ranged between 4 years and 12 years 11 months among both boys and girls. The children were divided into two groups: group 1 included 30 children who stuttered; group 2 included 100 typically developing children as the control group, who did not suffer from any language or speech disorder. Methods A pilot study was conducted on eight children who stuttered, after Arabic translation of the TOCS, and a few modifications were made to the pictures of the picture book of the TOCS to suit the Egyptian culture. Reliability was assessed by the test-retest method, and test validity was established on the basis of content description validity, internal consistency validity, convergent validity, and known group comparisons. Results The Arabic version of the TOCS is a valid and reliable test for the evaluation of Arabic speaking children who stutter and to determine the severity of a child′s stuttering. The results were highly significant and were capable of discriminating between children with normal disfluency and those who stutter.


Introduction
Stuttering is a multifactorial speech disorder characterized by repetition of sounds or syllables, sound prolongations, interjection, broken words, blocking of sounds, word substitutions, or excessive physical tension during speech production [1]. Other coexistent symptoms may include facial grimacing, tremors of muscles involved in speech, and eye blinks, as well as avoidance of words or situations that exacerbate stuttering episodes [2,3]. A total of 80-90% of developmental stuttering begins by 6 years of age and aff ects about 5% of children [4,5]. Spontaneous recovery occurs in about 75% of individuals [6].
Development of a comprehensive tool in the Arabic language is needed to tackle more detailed parts of the problem of stuttering. Th e Test of Childhood Stuttering (TOCS) [7] serves as an objective and standardized measure for identifying children who stutter, and it has four principal uses: (i) To identify children who stutter, (ii) To determine the severity of a child's stuttering, (iii) To document changes in a child's fl uency functioning over time, and (iv) For research on childhood stuttering.
Th e basis of our study was to use a new scale that could elicit speech disfl uency in children under certain situations such as time pressure, by using sentences that contain developmentally advanced syntactic structures, by utilizing a context that is similar to the types of conversational exchanges that the children are likely to encounter in school settings, and by producing narratives. Th e study is designed to test the validity and reliability of this scale.

Introduction
Stuttering is one of the most common speech disorders affecting children. The proper assessment of stuttering has been a point of interest for researchers over the years. The aim of this study was to adapt the Test of Childhood Stuttering ( TOCS) to suit the Egyptian culture and to test the validity and reliability of application of this test for the assessment of Egyptian children who stutter, for proper management of this ailment.

Patients
This study included 130 children. The ages ranged between 4 years and 12 years 11 months among both boys and girls. The children were divided into two groups: group 1 included 30 children who stuttered; group 2 included 100 typically developing children as the control group, who did not suffer from any language or speech disorder.

Methods
A pilot study was conducted on eight children who stuttered, after Arabic translation of the TOCS, and a few modi cations were made to the pictures of the picture book of the TOCS to suit the Egyptian culture. Reliability was assessed by the test-retest method, and test validity was established on the basis of content description validity, internal consistency validity, convergent validity, and known group comparisons.
Alexandria Main University Hospital, during the year 2013.
All the children were physically fi t. Children with acquired stuttering, brain damage, mental retardation, hearing impairment, highly unintelligible speech, psychiatric problems, a history of clinical and subclinical fi ts, or with any medical condition that seemed to interfere with proper analysis of the patient were excluded. Th eir age ranged between 4 years and 12 years 11 months. Th ey were divided into two groups: group 1 included 30 children who stuttered; group 2 included 100 typically developing children as the control group, who were not suff ering from any language or speech disorder. Th e pilot study included eight children. Th e proposed Arabic version of the TOCS was applied to the children of the pilot group to validate the materials used in the test (clear or not, suitable or not, valid to illustrate the item or not).

Methods
All studied individuals were subjected to the protocol of assessment of stuttering, which included assessment of stuttering severity using the Stuttering Severity Instrument-3 ( SSI-3) [8].
Th e components of TOCS [7] include the following: (1) Th e standardized speech fl uency measure.
Th e reliability of the test was assessed by test-retest after 8-12 days. Th e validity of the test was assessed on the basis of content description validity, internal consistency validity, convergent validity, and known group comparisons.
Informed written consent for participation in the study was obtained from the parents and/or the legally caring surrogates of the children; in addition, the child's assent was also obtained.
Data were analyzed using IBM SPSS software package version 20.0 (SPSS Inc., Chicago, Illinois, USA). Comparison between diff erent groups with regard to categorical variables was performed using the c2-test, Monte Carlo correction, or Fisher's exact test. For normally distributed data, two independent populations were compared using an independent t-test. For abnormally distributed data, the Mann-Whitney test was used to analyze stuttering and normal groups. Correlations between two quantitative and ordinal variables were assessed using Spearman's coeffi cient. Sensitivity, specifi city, positive predictive value, negative predictive value and accuracy were calculated. A receiver operating characteristic (ROC) curve was plotted to analyze the recommended cutoff ; the area under the ROC curve (AUC) denotes the diagnostic performance of the test. An area greater than 50% represents acceptable performance and an area about 100% represents the best performance of the test. Reliability statistics were assessed using the testretest method (Pearson's coeffi cient). Signifi cant test results are quoted as two-tailed probabilities. Signifi cance of the obtained results was judged at the 5% level.

Pilot study
Th e pilot study was conducted on eight children who stuttered, after translation of the TOCS to the Arabic language and its modifi cation to suit the Egyptian culture.
(1) Th e modifi cations made to some of the pictures of the speech fl uency task 1, Rapid Picture Naming, are as follows: (a) In practice items, we substituted lawn mower with car. (b) In the test items, we substituted the following: (i) Crab with fi sh.

Age
Th e controls and cases were classifi ed into two age groups: (1) Age group A: from 4 years 0 months to 8 years 5 months.
Statistically, there was an insignifi cant diff erence between the case and control groups (Table 1).

Sex
Nearly 80% of the studied population was male. No signifi cant diff erence was found between cases and controls.

Test of Childhood Stuttering
(1) Th e standardized speech fl uency measure: Table 2 shows the distribution of the studied groups with regard to the results of the standardized speech fl uency measure of the TOCS [6,7].
Statistically, there was a highly signifi cant diff erence between the case and control groups, whereas there was an insignifi cant diff erence between groups A and B in the case and control groups. (2) Observational rating scales: Table 3 shows the distribution of the studied cases with regard to the results of the observational rating scales of the TOCS [7]. Statistically, there was a highly signifi cant diff erence between the case and control groups, according to the Mann-Whitney test (P<0.001), whereas there was an insignifi cant diff erence between groups A and B in both case and control groups. (3) Supplemental clinical assessment activities: Table 4 shows an insignifi cant diff erence between groups A and B in the case and control groups in speech naturalness, average number of units per repetition, speech rate, disfl uency duration, stuttering frequency, and associated behaviors.

Reliability
Error due to time sampling refers to the extent to which a Pearson's test performance is constant over time. It was estimated by the test-retest method. Th e time laps between testing varied from 6 to 12 days. Test-retest reliability was measured in the pilot study group of eight children. Table 5 shows that all items of the TOCS are highly signifi cant.    It is a measure of test homogeneity. It is measured by correlating each composite with the total test score using Pearson's correlation. Table 6 shows correlation between each subscale and the total scale. Th e test proved to have good internal consistency, as there is signifi cant positive correlation between each subtest score and the TOCS score. (3) Convergent validity: It examines the degree to which the test is similar to (converges on) other tests that it theoretically should be similar to. Th is was done by comparing TOCS with SSI-3: Figure 1 shows that there was a signifi cant positive correlation between the total TOCS score and SSI-3 in the stuttering group. (4) Known group comparison: Comparison of the mean and SD of the control and case groups showed that the score among cases was signifi cantly higher than that among controls on all test items. Table 7 shows the performance of the two groups on the standardized speech fl uency measure of the TOCS in terms of mean and SD. Table 8 shows the performance of the two groups on the observational rating scales of the TOCS in terms of mean and SD.

Sensitivity and speci city
Th e ROC curves with the AUCs and their statistical signifi cance were used as an indicator for scale and subscale performances. Statistically signifi cant AUCs denoted performances better than chance (AUC = 0.50), and the greater the AUC, the better was the performance, with a maximum AUC of 1.00 denoting a gold standard like performance. Th e cutoff values for diagnosis were identifi ed just for statistically signifi cant AUCs, where the value that maximized both sensitivity and specifi city of the scale was chosen. Values above or the same as the identifi ed cutoff values denote case values. Th e ROC curve was used to obtain cutoff values for the proposed test (Fig. 2). Table 9 shows the cutoff scores of the total score of speech fl uency measure of the TOCS. Th e test proved to have high sensitivity and specifi city at the identifi ed cutoff points ranging up to 7.

Discussion
About two-third of the patients in this sample were between 4 and 8 years, as the primary symptoms of developmental stuttering are fi rst observed between the ages of 2 and 4 years, following a period of relatively fl uent speech production [6,9]. Data obtained from the University of Illinois Stuttering Research Program revealed that in 65% of the participating children, stuttering onset occurred before the age of 3 years; the fi gure rose to 85% by 3.5 years of age [9].
In the present study, boys constituted 76.7% of the cases and 80% of the control group. Th ere was an insignifi cant diff erence between the case and control groups and between the two age groups in both case and control groups. Th is corresponds to several studies reporting male predominance. A strong association is seen between sex and stuttering, especially in terms of a higher incidence and prevalence among male than among female individuals on the one hand [10] and a higher percentage of natural recovery on the other hand [6]. Previous research on various stuttering parameters, for example, disfl uency characteristics,   has reported few or no statistically signifi cant sex diff erences [6].
Th e SSI-3 is the most commonly used measure of stuttering severity. Th e SSI-3 measures the frequency of symptoms, the duration of the three longest symptoms, and physical concomitant behavior and combines them into an overall score [8]. In the current study, there was a highly signifi cant diff erence between the case and control groups, whereas there was an insignifi cant diff erence between groups A and B of both case and control groups. Riley [8] indicated that SSI-3 can be used as part of diagnostic evaluations, can assist in tracking changes in severity during and after treatments, and can be used to validate other assessment instruments. Lewis [11] did not support Riley's reliability procedures and found that two of the three parameters of SSI-3 failed to satisfy the fundamental requirement for measures of judge agreement. He concluded that SSI-3 scores do not accurately refl ect observations of stuttering behaviors. Th e SSI-3 has the advantage of being the only available standardized measure of stuttering severity.
Reliability of the TOCS was proven to be high by the test-retest method. We investigated test-retest reliability using a group of eight children. Th ese children had all been previously diagnosed with stutters. Th e time lapse between testing varied from 6 to 12 days. After the testing was completed, the raw scores for each test were correlated. Th e results indicated highly signifi cant correlations between initial and retest scores, proven statistically by the correlation coeffi cient (r), which ranged from 0. In the case of TOCS, a measure of speech fl uency, a rating scale for speech fl uency, and a rating scale for disfl uency and its related consequences, one would expect that individuals with identifi ed stuttering would do worse than children with typical fl uency. Th e means for the control group were much lower than those for cases, and all attained a highly signifi cant probability.
In the current study, the sensitivity index refl ects the ability of a test to correctly identify individuals who were previously identifi ed as stutters. Th e specifi city index refl ects the ability of a test to correctly identify individuals who were not identifi ed as stutterers. Th e positive predictive value refl ects the proportion of individuals who were identifi ed as stutters among all those identifi ed by the measure as having stutteringrelated problems.
Th e results of the TOCS were highly sensitive and specifi c. Th e percentage of sensitivity and specifi city ranged from 95 to 100% for cutoff scores ( 7) of the total score of speech fl uency measure of the TOCS, whereas in the original test of TOCS, the percentage of sensitivity and specifi city ranged from 79 to 90% for cutoff scores ( 9) of the total score of speech fl uency measure.
Th e ROC curve was plotted and the AUC was calculated for the TOCS total score, with an SSI-3 value of 0.986, indicating good discriminatory potential between typical fl uency children and children with a stutter. Th e cutoff score of the total score of speech fl uency measure of the TOCS was established at 7 or lower, showing a sensitivity of 100% and a specifi city of 95.24%. Th e cutoff scores at which the sensitivity and specifi city were maximum were 7 or lower. A raw score of 7 or higher by any rater is considered to be a high rating for a child to b e categorized as a stutterer, and ratings of 6 or lower are considered to represent children with typical fl uency.

Conclusion
(1) TOCS off er clinicians straightforward and effi cient fl uency assessment techniques for children between 4 and 12 years of age.
(2) Th e Arabic version of the TOCS is a valid and reliable test for evaluation of Arabic speaking children with stutters and to determine the severity of a child's stuttering.
(3) Th e results were highly signifi cant and were capable of discriminating between children with normal disfl uency and those who stutter.