Long-term prognosis of low language proficiency in children.

backgroundLow language proficiency (LLP) affects up to 10% of children, depending on how it is defined, and is a known risk factor for academic and social difficulties. Despite growing awareness, there is limited consensus on the long-term consequences of LLP across quality of life domains. This review synthesises evidence from longitudinal cohort studies to evaluate long-term risks associated with LLP identified between ages four and eight, with follow-up from age 12 onwards.

objectivesTo examine the long-term outcomes of children aged four to eight years with low language proficiency (LLP), in terms of language, literacy, and quality of life across five WHO domains (physical, psychological, independence, social relationships, and environment). SEARCH

methodsCENTRAL, MEDLINE, Embase, PsycINFO, ERIC, and 17 other databases were searched without language or date restrictions. In addition, reference lists were checked, citation searching was conducted, and study authors were contacted to identify additional studies. The most recent search was completed in March 2025. SELECTION CRITERIA: We included prospective or retrospective longitudinal cohort studies that identified children with LLP between ages four and eight years and assessed outcomes at age 12 or older. Studies could include comparison groups or single-cohort designs. LLP was defined either using standardised diagnostic criteria (≤ -1 SD on normed language measures, encompassing ~10-15% of the population) or clinical judgement by qualified professionals. DATA COLLECTION AND ANALYSIS: We extracted data on language, literacy, and World Health Organization (WHO) quality of life domains and used robust variance estimation (RVE) meta-analysis models, including hierarchical and correlated effects models. Effect sizes were reported as Hedges' g or risk ratios (RR), with Grading of Recommendations, Assessment, Development and Evaluation (GRADE) applied to each outcome domain. We conducted sensitivity and moderator analyses and assessed small-study effects. MAIN

resultsWe included 80 studies; 72 contributed to meta-analyses across 15 independent cohorts (≈ 28,828 participants). Studies were published between 1982 and 2024 and conducted in Europe and North America. Risk of bias, assessed across study participation, attrition, and outcome measurement, varied; only three studies were at low risk across all assessed domains, with study attrition most frequently rated as moderate or high. Language outcomes showed large adverse associations in long-term follow-up (145 effect sizes; 11 cohorts; Hedges' g = -1.36, 95% Confidence Interval (CI) -1.80 to -0.92; moderate certainty). Literacy outcomes were similarly affected (86 effect sizes; 7 cohorts; g = -1.12, 95% CI -1.28 to -0.96; high certainty). Physical outcomes were reported in only one study and could not be meta-analysed; evidence was insufficient to draw conclusions. Psychological outcomes showed moderate adverse associations (91 effect sizes; 7 cohorts; g = -0.50, 95% CI -0.80 to -0.19; moderate certainty), with risk ratio analyses indicating lower probability of favourable psychological outcomes (RR = 0.53, 95% CI 0.42 to 0.67). Independence outcomes showed smaller and more uncertain associations (17 effect sizes; 2 cohorts; g = -0.59, 95% CI -1.35 to 0.18; low certainty). Social relationship outcomes showed moderate adverse associations (35 effect sizes; 6 cohorts; g = -0.64, 95% CI -1.35 to 0.07; moderate certainty). Environmental outcomes showed large but imprecise adverse associations (23 effect sizes; 5 cohorts; g = -1.10, 95% CI -1.47 to -0.73; moderate certainty). Across domains, sensitivity analyses supported the robustness of findings, although degrees of freedom were often low due to the limited number of independent cohorts. Where binary outcome data were available, risk ratio estimates were directionally consistent with continuous outcomes but frequently imprecise. AUTHORS'

conclusionsEarly low language proficiency (LLP) is consistently associated with substantial long-term language and literacy difficulties. LLP is also associated with poorer psychological well-being and social relationship outcomes later in life, with moderate certainty of evidence. Associations with independence outcomes are smaller and more uncertain, reflecting limited data and low certainty. Environmental outcomes, including education, employment, and societal participation, show large adverse associations, although estimates are imprecise due to the small number of contributing cohorts. Physical health outcomes remain under-researched. Taken together, these findings indicate that childhood LLP is not a transient delay but a marker of enduring developmental vulnerability, underscoring the need for sustained identification and support across developmental stages. Further high-quality longitudinal studies are needed, particularly in under-represented outcome domains.

fundingInternal sources Department of Special Needs Education, University of Oslo, Norway, supported the preparation of the protocol and review to be carried out during office hours for ÅMH, KR, and MM-L. Department of Education, University of Oslo, Norway, supported the preparation of the protocol and review to be carried out during office hours for AL. External sources The Research Council of Norway, Norway This review is part of the project, Better Equipped, which is funded by the Research Council of Norway, Grant 324207. Lervåg and Melby-Lervågs participation was founded by The Research Council of Norway, Centres of Excellence, Grant 331640. The funder had no role in the study design, conduct, methods, data analysis, reporting, or publication of this protocol for the review. REGISTRATION: https://discovery.ucl.ac.uk/id/eprint/10165057/1/Hagen_et_al-2023-Cochrane_Database_of_Systematic_Reviews.pdf.