To systematically review the evidence on screening for speech and language delay and disorders in children age 5 years or younger.
PubMed/MEDLINE, the Cochrane Library, APA PsycInfo, ERIC, Linguistic and Language Behavior Abstracts (ProQuest), and trial registries through January 17, 2023; reference lists of retrieved articles; outside experts; and reviewers, with surveillance of the literature through November 24, 2023.
Two investigators independently selected English language studies using a priori criteria. Eligible studies included cohort studies or trials directly comparing screening versus no screening, as well as studies of screening test accuracy for speech and language delay or disorders among children age 5 years or younger. Randomized, controlled trials (RCTs) of interventions for speech and language delay or disorders enrolling children age 6 years or younger reporting on the benefits and harms of interventions were also eligible.
One investigator extracted data and a second checked accuracy. Two reviewers independently rated quality for all included studies using predefined criteria.
Thirty-eight studies reported in 41 articles (N=9,006) were included. No study evaluated the direct benefits of screening compared with no screening. Twenty-one studies (23 articles; N=7,489) assessed the accuracy of 23 instruments for detecting speech and language delay and disorders in young children. The sensitivity and specificity varied widely across included studies, and no more than one or two studies reported on the accuracy of each instrument. Ten instruments, described in 10 studies (11 articles), used parent reports to detect speech and language delay and disorders, and 13 instruments, described in 14 studies, required a trained examiner to administer the instrument to children. Most included instruments were designed to screen for global language problems (provide an overall score for “language”) and nine provided scores for specific aspects of language (e.g., expressive language skills only). Sensitivity and specificity of the three parent-reported instruments of emerging expressive language skills were consistent; median sensitivity was 91 percent, (range, 88% to 93%) and specificity was 88% (range, 88% to 85%). The accuracy of global language instruments based on parent reports was inconsistent, with a median sensitivity of 74 percent (range, 55% to 93%). Accuracy of provider-reported global and specific language problems varied significantly across tools. Seventeen RCTs (18 articles; N=1,517) compared an intervention for a speech and language delay or disorder with an inactive control. Eight RCTs of treatment were limited to children with language delay and no obvious speech-sound or fluency disorder. Three assessed parent-delivered, group training interventions. Of these, two that evaluated longer, more intensive interventions (11 bimonthly 60- to 75-minute sessions, and 11 weekly 2.5-hour sessions followed by 3 weekly home visits) found benefit on different measures of expressive language outcomes, and one RCT of a shorter parental group training intervention (6 weekly 2-hour sessions) found no statistically significant difference between groups for any language outcome measure. Other RCTs of interventions for language delay that enrolled heterogeneous populations and assessed different interventions showed mixed results. Two RCTs delivered interventions featuring school-based, whole-class curriculum components (or Tier 1 interventions) designed to advance language and literary skills over the course of an academic year. Both demonstrated improved receptive and expressive language outcomes in favor of the intervention; however, one found improvement for some measures but not others. Two RCTs that assessed fluency treatment in young children focused on the Lidcombe Program of Early Stuttering Intervention delivered by speech-language pathologists (SLPs) and featured parent training to provide verbal contingencies for stutter-free speech (e.g., “that was smooth talking”) and stuttering (e.g., “that was a bit bumpy”). Both found benefit for stuttering fluency associated with the intervention at 9 months. One RCT, which delivered the intervention face-to-face in a clinic setting, showed a 2.3 percent lower proportion of syllables stuttered (95% confidence interval [CI], 0.8 to 3.9) compared with the control group, whereas the second RCT, which was delivered via telehealth, showed a larger reduction from the baseline mean number of syllables stuttered in the intervention group than in controls (-3.0; p=0.02). Three RCTs assessed interventions for three different types of speech-sound disorders and reported on various measures of speech-sound; results were generally inconsistent across different measures of speech. Two RCTs that evaluated treatment for children newly referred from primary care for any speech or language problem found inconsistent results, with improvement on some domains of speech and language but not others and no consistent benefit for a similar outcome domain. Eight RCTs (N= 1,239) reported on one or more outcomes specific to school performance or early literacy, health-related quality of life, function, behavior, or socialization. No studies assessing the same type of intervention among similar groups of children reported on similar outcomes, and most studies found no difference between groups for measures of early literacy, function, and quality of life. No RCTs reported on the harms of interventions.
No studies reported on the benefits and harms of screening vs. no screening, or on the potential harms of interventions. Studies of screening test accuracy and interventions for children with speech and language problems were heterogeneous in terms of the enrolled population and specific type of speech or language disorders targeted. Very few studies of screening test accuracy evaluated the same instrument. Similarly, few studies of interventions for speech and language delay or disorder enrolled similar populations and evaluated similar types of interventions. Two RCTs of treatment enrolled children who were newly referred from primary care; however, it is not clear whether children were identified via routine screening and if the studies differed in terms of setting, mean age of enrolled children, and other factors.
We found no eligible studies that reported on benefits directly arising from screening when compared with usual care or no screening. Parent-reported screening tools of emerging expressive language skills had reasonable accuracy for detecting expressive language delay; however, the accuracy of global language instruments based on parent reports was inconsistent. Accuracy of examiner-administered instruments was also variable, especially for examiner-administered instruments of specific language skills. Existing evidence supports the benefit of group parent training programs for speech delay that provide at least 11 parental training sessions for improving receptive language skills, as well as the Lidcombe Program of Early Stuttering Intervention delivered by SLPs for reducing stuttering frequency.
