Information obtained at around age 8 years was used to construct a model that predicted persistence of, and recovery from, stuttering several years later. A logistic regression model that classified children as persistent or recovered at the teenage years using stuttering history and symptom information obtained at around age 8 years was constructed and validated.
A longitudinal study of 222 children who stuttered was conducted. The children were followed up from around age 8 until the teenage years. Persistence and recovery outcomes were established at the teenage years for 206 of the children, based on agreement across 3 standardized instruments. The data from 132 children were used to develop the model, and the data from the remaining 74 children were used to validate the model. Risk factors assessed at the beginning of the study were head injury, age at stuttering onset, family history of stuttering, handedness, whether a second language was spoken in the home, gender, and scores from the Stuttering Severity Instrument Version 3. The information about risk factors was obtained at around age 8 years by interview, except for the severity estimate, which was obtained by analysis of recordings and observations of physical concomitants associated with stuttering. The model was developed using logistic regression procedures.
The only factor to predict the persistence of, and recovery from, stuttering at the teenage years was stuttering severity at around age 8 years (none of the other factors being significant). For the initial model, the sensitivity (percentage of the group that was classified as persistent) was 84.1% and specificity (the percentage of the group that was classified as recovered) was 78.3%. For the validation, sensitivity was 76.3% and specificity was 72.2%.
Persistence and recovery at teenage can be predicted from information that can be collected at around age 8 years with sensitivity and specificity of ∼80%.
From the Division of Psychology and Language Sciences, Cognition, Perception and Brain Research Department, University College London, London, United Kingdom.
Received July 2010; accepted January 2011.
This study was supported by the Wellcome Trust, London, United Kingdom.
Address for reprints: Peter Howell, PhD, Division of Psychology and Language Sciences, University College London, 26 Bedford Way, London WC1H OAP, United Kingdom; e-mail: firstname.lastname@example.org.