Logo: to the web site of Uppsala University

Children who do not develop their language as expected are at risk of difficulties in school and with peer relations. The Swedish Child Health Services (CHS) offer preventive health surveillance, including language screening, at 2.5-3 years. However, the screening has only been validated for monolingual children. 

The aims of the thesis were to explore CHS nurses’ experiences of language screening of bilingual children, to evaluate the 3-year screening for language disorder for use at 2.5 years, and to assess the stability of the children’s diagnoses and the classification accuracy of the screening after 6 months.  

In Study I, 863 nurses answered a web-based survey. Half of the nurses simplified the screening processes for bilingual children, and 74% postponed referrals to speech and language services. About 80% believed that language development was slower in bilingual children which was the strongest predictor of simplified screening practices. In Studies II and III, 105 monolingual and 111 bilingual children were screened and clinically assessed at age 2.5. The screening classification accuracy for monolinguals vs bilinguals screened in both their languages was: sensitivity 91% vs 88%, specificity 91% vs 82%, PPV 56% vs 67%, and NPV 99% vs 94%, respectively. Study IV reassessed 141 children (48 mono- and 93 bilinguals) at age 3, i.e. all children with a positive outcome at age 2.5 from studies II and III, and matched children from the same studies with negative outcomes. The number of new cases with language disorder (n=4) was about the same as the number who had recovered (n=5) at age 3. These few changes were statistically significant, but not considered to be of clinical relevance since the classification accuracy of the screening was still within 95 CI after 6 months. The screening classification accuracy for monolinguals vs bilinguals was: sensitivity 71% vs 81%, specificity 93% vs 82%, PPV 39% vs 67%, and NPV 98% vs 90%. 

In conclusion, the modified language screening can be recommended for use at age 2.5. Bilingual children who don’t pass the screening in Swedish, should be screened in their mother tongue using a standard procedure. The wait-and-see strategy reported by the nurses in Study I was not supported as screening results remained stable between 2.5 and 3 years. There is a need to implement new evidence-based routines in child health services to equitably screen both monolingual and bilingual children for language disorder.