Worsening sleep linked with later anxiety and depression in children

Children whose sleep-related problems were more severe at the start of a six-month study were more likely to screen positive for anxiety, depression or both at follow-up, according to an observational cohort study of pupils in an urban district of Chongqing, China. After researchers adjusted for demographic, family, health and behavioral factors, as well as children’s baseline emotional symptoms, each additional point on the five-item sleep index was associated with 14.4% higher odds of screening-positive anxiety, 9.7% higher odds of screening-positive depression and 14.6% higher odds of coexisting anxiety and depressive symptoms. These findings identify an association, not proof that sleep problems caused the emotional symptoms or that sleep scores can diagnose them.

The analysis included 1,529 children who completed both surveys in the broader child and adolescent growth and development cohort. Participants were recruited from grades 1 to 3 at three primary schools in one urban Chongqing district. The baseline assessment took place in May and June 2024, while follow-up surveys were conducted in November and December 2024; no intervention was introduced between the two points. The children had a mean age of 8.00 years, with a standard deviation of 1.0, and included 782 boys and 747 girls. The sample was concentrated geographically, and the investigators cautioned that results from these schools may not generalize to rural areas, less affluent communities or populations with different cultural backgrounds.

Sleep was measured with a five-item adaptation of the Pittsburgh Sleep Quality Index, called PSQI-5. Children reported subjective sleep quality, sleep-onset time, sleep disturbances, use of sleep medication and daytime functional impairment. Each dimension was scored from 0 to 3, producing a total from 0 to 15, with higher scores indicating more sleep-related problems. This abbreviated index was designed for young pupils completing questionnaires in a classroom setting, where asking them to recall detailed sleep times and efficiency could introduce errors. It is not equivalent to the full Pittsburgh Sleep Quality Index or its conventional clinical thresholds, and its internal consistency was modest: Cronbach’s alpha was 0.495 at baseline and 0.540 at follow-up.

Anxiety symptoms were assessed with the seven-item Generalized Anxiety Disorder scale, or GAD-7, covering the preceding two weeks. A score of at least 5 was classified as screening-positive, rather than diagnostic. Depressive symptoms were measured with the 10-item Short Form of the Children’s Depression Inventory, or CDI-S, also covering two weeks. A score of at least 7 was considered screening-positive. Children were classified as having coexisting symptoms when they crossed both thresholds at the same survey. Across the full baseline sample, 38.9% screened positive for anxiety, 22.3% for depression and 16.5% for both; at follow-up, the corresponding proportions were 35.8%, 19.1% and 13.1%. Wilcoxon signed-rank tests indicated lower aggregate PSQI-5, GAD-7 and CDI-S scores at follow-up, all with p values below 0.001.

Most of the unadjusted sleep–emotion associations became smaller after the statistical models accounted for potential confounding and children’s initial emotional symptoms, but they remained statistically significant. In the fully adjusted logistic regressions, the odds ratio was 1.144 for follow-up anxiety, with a 95% confidence interval of 1.071 to 1.222, and 1.097 for depression, with a 95% confidence interval of 1.019 to 1.180. For coexisting anxiety and depression, the odds ratio was 1.146, with a 95% confidence interval of 1.049 to 1.251. The corresponding p values were below 0.001, 0.013 and 0.003. When baseline PSQI-5 scores were divided into quartiles, the highest group remained associated with coexisting symptoms after full adjustment, with an odds ratio of 2.415 compared with the lowest group and a 95% confidence interval of 1.370 to 4.258.

The researchers also examined how sleep changed over the six-month interval by calculating a change score, ΔPSQI, as the follow-up PSQI-5 value minus the baseline value. A positive value therefore represented worsening sleep-related problems. After adjustment for baseline sleep, baseline anxiety and depression scores and other covariates, each one-point increase in ΔPSQI was associated with an odds ratio of 1.486 for coexisting anxiety and depression at follow-up, with a 95% confidence interval of 1.359 to 1.626 and p below 0.001. Baseline PSQI-5 also remained associated with co-occurrence in the same model, with an odds ratio of 1.499. Linear regressions also linked increases in ΔPSQI with follow-up anxiety and depression scores and with changes in both symptoms over the study period.

Dimension-specific analyses suggested that the more persistent associations involved subjective sleep quality and difficulty falling asleep. Associations for sleep disturbances, use of hypnotic medication and daytime dysfunction became weaker or were no longer significant after baseline emotional symptoms were included. Because a change score contains the follow-up measurement and the emotional outcome was also recorded at the end of that interval, the researchers described the ΔPSQI findings as concomitant associations within the same observation period. They could not determine the direction of influence, and the study did not model reverse effects in which anxiety or depression might worsen sleep. Screen time, physical activity, academic stress and parent–child relationships were not included as covariates, leaving possible residual confounding.

A sensitivity analysis restricted the longitudinal analysis to the 1,405 children with complete covariate data, 91.9% of the main sample. In this group, each additional baseline PSQI-5 point was associated with adjusted odds ratios of 1.215 for follow-up anxiety, 1.170 for depression and 1.147 for coexisting symptoms, with all p values below 0.05. Associations between changes in sleep-related problems and follow-up emotional outcomes remained stable, with p values below 0.001. The researchers had handled missing covariates in the main analysis by generating 20 imputed datasets with a fully conditional specification method. Baseline age, sex and PSQI-5, GAD-7 and CDI-S scores did not differ significantly between children with complete and incomplete covariate data.

The results point toward sleep-related experiences as potentially useful supplementary information in school-based mental health observation, but the authors explicitly rejected treating the index as an independent screening or diagnostic tool. Their objective was epidemiological risk assessment rather than individual clinical prediction. Screening scales identified symptom patterns but were not replaced by clinical diagnostic interviews, and the modest internal consistency of PSQI-5 adds measurement uncertainty. Although the authors observed that recurrent sleep difficulties can reflect changes in family routines, school pressures, peer relationships or electronic media use, those mechanisms were not directly tested. The study instead offers preliminary evidence from a short, two-wave cohort that both persistent sleep problems and their worsening may accompany a greater later burden of internalizing symptoms among young primary-school children.

Subject of Research: Longitudinal associations between sleep-related problems and internalizing symptoms in primary-school children

Article Title: Study on the association between sleep quality and anxiety–depression symptoms in school-aged children

Article References: Chen, J., Liu, Y., Liu, Q., & Zhang, F. (2026). Study on the association between sleep quality and anxiety–depression symptoms in school-aged children. Frontiers in Public Health, 14, Article 1927475. https://doi.org/10.3389/fpubh.2026.1927475

Image Credits: AI Generated

DOI: 10.3389/fpubh.2026.1927475

Keywords: child sleep, anxiety symptoms, depressive symptoms, school-aged children, longitudinal cohort, Chongqing, mental health screening, sleep quality, association, between, sleep, quality