Selective mutism (SM) is a relatively rare anxiety disorder (prevalence rates varying from 0.2% to 1.9% [1
]) that is characterized by consistent failure to speak in various specific social situations where speaking is expected (e.g., at school), whereas the child does speak in other situations (e.g., at home) [3
]. SM typically manifests between the ages of 3 and 5, coinciding with the start of (pre)school and cannot be attributed to a language or speaking disorder [4
]. Parents may not always realize there is a problem, as the children do speak freely at home. In such cases, the consistent failure to speak needs to be noticed by professionals in the life of the child, for example at school [6
]. A major problem in recognizing SM, is that there is a lack of validated instruments that assess the different responses associated with SM, and are able to distinguish children with SM from those with other anxiety disorders. Without adequate instruments, risk increases that SM is overlooked or not recognized as such [9
]. If as a result no timely intervention is started, this can lead to chronic and complex anxiety and mood issues. Currently the selective mutism questionnaire (SMQ) is the most widely used screening and assessment tool with good psychometric properties, that is able to distinguish children with SM from other anxiety disorders [10
]. Besides the SMQ, there are few other instruments investigating SM symptomatology; such as the Frankfurt Scale of Selective Mutism (FSSM [11
]) or the Speech Situations Questionnaire (SpSQ [12
]), however limited research has been conducted investigating their psychometric properties.
The SMQ assesses speaking behavior across different situations. It can be used by healthcare professionals in the first line of care (general practitioners, municipal child health clinics, school doctors and counselors). The psychometric studies into this instrument [10
] demonstrate a three factor structure consisting of school, home/family and public/social settings. Bergman et al. [10
] first described the SMQ and initially studied the instrument in an internet sample of 589 participants (3–11 years) where parents identified their child as having difficulties in speaking in some settings. In their second study the psychometric properties of the SMQ were investigated in a group of 48 children (3–10 years) with SM and 18 anxious children without SM. Bergman et al. report internal consistency of Cronbach’s α = 0.84. In addition, other groups investigated the SMQ: The study of Letamendi et al. [13
] included 102 parents of children with SM and 43 parents of children without SM (3 to 11 years), reporting Cronbach’s α = 0.783. In addition, the questionnaire was translated to Norwegian [14
] and Spanish [15
] showing good psychometric properties within these languages. Oerbeck et al. [14
] investigated the psychometric properties of the SMQ in Norway, in 32 children with SM and 32 typically developing children (3–9 years), reporting Cronbach’s α = 0.96 in the total sample. Additionally, a study into a Spanish version of the SMQ by Olivares-Olivares et al. [15
] included 110 children with SM (3–10 years), reporting Cronbach’s α = 0.90. Data from their Spanish sample fitted the factorial model of Bergman et al. [10
], and their data on the reliability and validity of the Spanish SMQ were robust. Authors concluded that the Spanish SMQ is a good instrument for assessing SM in Spanish-speaking children.
The aim of this study is to facilitate early screening of SM in children by conducting a validation study of the translated version of the SMQ in the Netherlands. In contrast with previous studies investigating the psychometric studies of the SMQ, in our academic center for child and adolescent psychiatry we see a broad age range (3–18 years vs. 3–11 years in previous studies) of children and adolescents being referred for diagnostics and treatment due to presumable SM. This seems in line with a general increase of referral of older children to clinical practice, and growing attention for SM in older children. Therefore we decided to also include older participants as to provide a realistic representation of our population.
The validation of the Dutch SMQ has two purposes: firstly the ability of the instrument to screen for SM, secondly the ability to assess symptom level severity (which for example can be monitored following treatment).
Among the children who participated in this study (n = 303), 54% were female and 46% were male. The age of the children ranged from 3 to 17 years with a mean age of 7.94 (SD = 3.84). Of the total group, 27.3% was bilingual. The clinical group consisted of 106 children (age range 3–16 years, mean = 6.33, SD = 2.98). 42.5% of the children in this group were boys, 57.5% were girls. In addition, 45.2% of the clinical group was bilingual or multilingual. The control group consisted of 197 children with no SM classification or SM related problems (age range 3–17 years, mean = 8.8, SD = 3.98). 48% of the children in this group were boys and 52% girls. Furthermore, 17.8% of the control group was bilingual or multilingual. The education level of the families in the clinical group and the control group is shown in Table 1
Despite our efforts to include a representative control group in terms of gender, age and bilingualism, especially parents of younger children (< 4 years) and bilingual/multilingual families were less likely to participate, therefore our clinical group was younger, and conducted more children with bilingual/multilingual background in comparison with our control group. Furthermore, the education level of parents in the control group was higher than in the clinical group (See Table 1
Within the clinical group, all children were referred to our academic center for child and adolescent psychiatry due to suspected SM and were referred for treatment and/or diagnostics. Some of them (N = 83), participated in an ongoing randomized controlled trial (RCT) study [16
]. The data collection of the clinical group was part of care as usual and was also performed in the context of the baseline assessment in the RCT [16
]. All assessments were performed by psychologists of the SM expertise team of our institution. To recruit a control group with similar age and gender distribution, (pre)elementary schools, secondary schools, sports and recreation clubs in Amsterdam and surrounding areas were contacted to distribute information about the research project. The participants in the control group responded to posters, over 6500 folders and information that was distributed in (newsletters of) over 100 (pre)schools, around 70 sports clubs and other recreation associations such as music schools in the same regional areas as the clinical group. Parents of 270 children were interested in participating and filled out a contact form and received further information and informed consent forms. After receiving information, N = 197 families decided to participate in the control group. Parents and children from age 12 and older filled out a written informed consent, questionnaires were sent out through a secured online program and a telephone interview for additional measures similar to our clinical group was performed by research psychologists and master students under supervision of a psychologist of the SM expertise team. In both groups, demographic data were collected with use of a semi structured interview (e.g., gender, age, bilingualism of the child, parental educational and occupational status, nationality of child and parents and languages being spoken at home).
The study was approved by the Medical Ethical Committee of the Amsterdam University Medical Center.
The SMQ [10
] is a parent reported questionnaire, assessing the child’s speaking behavior and SM symptoms in various situations. The SMQ consists of two scales: the symptom scale (17 items) stating different situations in which a child is expected to speak, covering three domains: school (6 items, e.g., “When appropriate, my child asks his or her teacher questions.”), family (6 items, e.g., “When appropriate, my child talks to family members while in unfamiliar places.”) and social/public situations (5 items, e.g., “When appropriate, my child speaks with his or her doctor and/or dentist.”). Parents rate the frequency of speaking behavior on each item using a 4-point scale (3 = always, 2 = often, 1 = seldom and 0 = never). An SMQ symptom scale score was calculated as an individual’s average item score multiplied by the number of items (17), equaling the sum score in case no data are missing. SMQ symptom scale scores thus ranged from a minimum of 0 to a maximum of 51; the lower the score on the SMQ symptom scale the more problems with daring to speak and the less speaking behavior. In order to realize a cutoff score for the screening of SM, the SMQ scoring in this study was converted so that a higher score indicates more problems.
The interference scale includes 6 items (e.g., “How much does not talking interfere with school for your child?”, response categories: “no”, “slightly”, “moderately” and “extremely”). An SMQ interference scale score was again calculated as an individual’s average item score multiplied by the number of items (6). A higher score on the SMQ interference scale indicates higher impact on child and family functioning.
The Anxiety Disorders Interview Schedule for Children for DSM-IV [17
] is a semi structured diagnostic interview to assess anxiety and mood disorders according to DSM-IV criteria in children and adolescents. The SM segment of the ADIS-C was conducted in parents, and the child version was conducted in children from ages 8 and up, if they agreed to answer the questions. The SM segment consists of 8 items, covering the speaking behavior and school functioning of the child. Parents rate the interference of the symptoms with the child’s daily life on a 9 point scale (0–8). The interviewer rates the interference on the same 9 point scale for the Clinician Severity Rating (CSR). A cutoff of 4 on the CSR indicates a classification. The ADIS-C was part of the care as usual in the clinical sample, and the SM segment was administered through a phone interview in the control group. If the parent interference rating and the CSR differed, the CSR was decisive.
Children’s Internalizing and Externalizing Problems
The Child Behavior Checklist (CBCL) [18
], preschool and school-age versions, is a parent report questionnaire assessing behavioral and emotional problems in children. The Youth Self Report (YSR) [18
] is the parallel questionnaire of the CBCL, formulated for the child from ages 11 and older. The CBCL and the YSR have different subscales, combining in an internalizing scale (includes Withdrawn, Somatic Complaints, and Anxiety/Depressed Problems) and externalizing scale (includes Delinquent and Aggressive Behaviors). T-scores of 65 and higher are in the clinical range.
Statistical Analysis for Validation
As mentioned in the introduction, the validation of the SMQ consisted of two parts. The first part concerned an investigation of the descriptive statistics, followed by the psychometric properties of the SMQ when using the instrument as a screening device. To this end, we carried out a reliability analysis (obtaining Cronbach’s α), a receiver operating characteristic (ROC) curve analysis, and a (confirmatory) factor analysis on the correlational structure among the 17 items of the SMQ in the subjects sample as whole, i.e., the combined clinical and control group. Following the factor analysis, the discriminant validity between the SMQ and CBCL was established. The second part concerned an investigation of the SMQ as a clinical measuring instrument. To this end, we repeated the reliability and factor analysis within the clinical subsample only, while also considering the reliability and validity of the interference scale.
The results of the current study show the validity of the Dutch SMQ both as a screening tool and as a clinical instrument. The psychometric properties of the SMQ can be considered as good. First of all its consistency and reliability is high. Secondly, the instrument is able to distinguish well between individuals that fulfill the diagnostic criteria of SM and those that do not. Thirdly, the factorial structure of the SMQ was confirmed providing additional evidence for the validity of three subscales (Home/Family, School and Social situations), as already found by Bergman et al. [10
]. Lastly, in line with findings from previous research, as mentioned in Bergman et al. [10
], the SMQ shows that SM is a construct that differs from other variables, including externalizing and internalizing behavior as assessed by the CBCL.
Our results support earlier findings by Letamendi et al. [13
] showing correlations between the internalizing scale of the CBCL and the SMQ. The correlations are not so high that both instruments seem to measure the same construct, thus both instruments can be regarded as measuring distinct constructs [21
]. In addition, since the CBCL does not measure SM specifically, there is still a need for an instrument aimed to the speaking behavior of the child and the interference of SM in the daily life. This result corroborates again previous empirical evidence in support of the interpretation of SM as a distinct anxiety disorder [22
In previous psychometric studies on the SMQ, no cutoff score was established. To enable low threshold access to care, we decided to investigate the possibility of using a cutoff score in our sample. To be able to calculate the cutoff score, we reversed the scoring of Bergman et al. [10
] so that a higher score indicates more SM symptomatology to facilitate the use and interpretability. If a cutoff score of 13 would be used for the SMQ in the current sample, all children with SM would be screened as positive. Using such a sensitive instrument would ensure early detection. We recommend that for screening sensitivity ought to be prioritized over specificity, since the risk of missing a false negative and thus not referring for adequate treatment should be minimized in order to promote early detection and intervention. Since we know that early intervention improves prognosis, this would be preferable [7
]. In future research, this cutoff, based on reversed scoring, can be investigated in other samples to improve the generalizability of this finding and its use in clinical practice. If the original scoring of Bergman et al. [10
] (higher score indicates less SM symptomatology and more speaking behavior) is maintained, we expect a cutoff score of 38 (51; maximum total score—13; cutoff with reversed scoring) where it is to be interpreted that a score of 38 or less is an indication for SM.
Strengths and Limitations
A strength of this study was that we made an effort to recruit participants for the control group widely, spreading over 6500 folders, and involving over 100 (pre)schools and 70 sports and leisure clubs in different neighborhoods and zip codes of Amsterdam and its suburbs. By reaching out through social media, a radio interview and school newsletters, we intended to reach a diverse group. Furthermore a strength was that all children in our clinical group were referred to our specialized care center, thereby reflecting our regular clinical population. At our institution, we see children and adolescents from ages up to 18 and in some cases 23. To ensure that the instrument was applicable to our broad clinical population, another strength was that we were able to include also adolescents in our study. As previous psychometric studies included younger children [10
], the current broad age range in the sample is rather unique. Since SM is often underrecognized, which can result in misdiagnoses and delay in start of treatment [23
], it was of surplus value that this study encompassed a broader age range both for our clinical population and the control group. This decreased the possibility that there were children with ‘latent’ SM in the control group, where the problem might not have been recognized at a younger age. An important innovative strength was that this study established cut-offs for screening purposes.
Among the limitations of this study is that the control group consisted of families that wanted to participate in the study intrinsically, and thereby possibly differed (selection bias) from the complete general population and the group that is referred to our institution for SM. The need to actively reach out to participate can lead to inevitable selection bias. Despite our efforts to include a representative typically developing control group, both groups differed significantly in terms of age and bilingualism. Even though we did actively distribute information in preschools and early childhood health centers, the participants in the control group were on average significantly older than the clinical group, which can have influenced the results. Noteworthy, however, both the clinical and the control group covered children in the preschool, elementary school and secondary school age groups. For future research, we recommend that the validity of the SMQ is further studied in older children and adolescents with SM.
In the current study we only included parent report measures (SMQ, ADIS and CBCL). Due to the small numbers of patients completing self-reports (YSR), due to the age-range of this instrument, no warranted statistical analyses on YSR data could be performed. For future research we recommend to study SMQ and YSR data for a larger group of clinical patients. In addition, we recommend also including data on the SMQ and also the TRF from teachers and/or schools as to include the perspective of the school where the not speaking is most present [24
Now that the SMQ is validated for use in the Dutch population, it can be easily accessed by health care professionals. We conclude that the SMQ can be used as a reliable and valid screening tool to assess the need for further diagnostics. With high sensitivity and specificity it is a suited tool for clinical practice and research purposes. In the future the SMQ may be added to the student tracking system as an optional questionnaire when the school suspects the student might not talk as much in the school setting. The SMQ can be embedded in the Dutch system of first line care, through early childhood health centers and/or school and youth doctors. This would enhance the early detection of SM, thus improving the timely diagnosis and early start of treatment, leading to a better prognosis.
Selective mutism (SM) is an anxiety disorder in children/adolescents, characterized by the absence of speaking in specific social situations, mostly at school. The Selective Mutism Questionnaire (SMQ) is a parent report, internationally used to assess SM symptomatology and treatment outcomes [10
]. The present study investigated the psychometric properties of the Dutch translation of the SMQ, through reliability, confirmatory factor, and ROC analyses, using data of 106 clinical children with SM and 197 control children without SM. Results showed that the Dutch SMQ is highly reliable (α = 0.96 in the combined sample; 0.81 within the clinical group) and followed the expected factor structure. We conclude that the Dutch SMQ is a reliable and valid tool both as a screening and clinical instrument. Now that the SMQ is validated for use in the Dutch population, it can be easily accessed by health care professionals. The SMQ can be embedded in the Dutch system of first line care, through early childhood health centers and/or school and youth doctors. This would enhance the early detection of SM, thus improving the timely diagnosis and early start of treatment, leading to a better prognosis.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.