Suicidal thoughts during childhood and early adolescence do not follow a single developmental path. A new six-year longitudinal study of 2,607 South Korean students has identified three markedly different trajectories from fourth grade through ninth grade, showing that most children remained at consistently low levels while substantial minorities followed either a gradually worsening or an initially severe but improving pattern.
The peer-reviewed study, published on 28 September 2026 in Humanities and Social Sciences Communications, followed the same cohort annually between 2018 and 2023. Rather than treating suicidal thoughts as a single outcome measured at one point in time, researcher Changmin Yoo used latent class growth analysis to ask whether distinct developmental patterns could be detected across the transition from elementary to middle school.
The answer was clear. Nearly three quarters of the students belonged to a low-stable group, but about one in five followed a moderate and gradually increasing trajectory. A smaller group began the study with very high levels of suicidal thoughts, improved substantially, and then showed a modest rebound. The analysis also examined whether individual, family and school characteristics measured at baseline were associated with membership in these different trajectories.
Why trajectories matter more than a single snapshot
Research on suicidal thoughts in young people often relies on prevalence estimates at one age or one survey wave. Those estimates are important, but they can conceal very different developmental experiences. Two children who report similar thoughts at one point may be moving in opposite directions over time. One may be entering a period of increasing vulnerability while another may be recovering from an earlier period of severe distress.
This distinction is especially important during the transition from late childhood into adolescence. The years spanning elementary and middle school involve changes in peer relationships, academic expectations, family dynamics, sleep patterns, smartphone use and self-concept. A longitudinal design can therefore reveal patterns that a cross-sectional survey cannot.
The South Korean context adds particular relevance. The authors note that suicide is a major public-health concern in the country and is the leading cause of death among people aged 10 to 39. The study does not attempt to explain national suicide mortality, however. Its narrower purpose is to describe how self-reported suicidal thoughts develop across six school years and which baseline characteristics distinguish the resulting trajectory groups.
Following 2,607 students from fourth to ninth grade
The analysis used the Korea Children and Youth Panel Survey 2018, a nationally organised longitudinal dataset. The study followed 2,607 students from fourth grade in 2018 through ninth grade in 2023. Suicidal thoughts were assessed annually using a single-item measure.
At baseline, the researcher considered factors across several parts of children’s lives. Individual measures included gender, sleep duration, self-esteem, problematic smartphone use and academic performance. Family measures included household income and dimensions of parenting such as warmth, rejection and inconsistency. School-related measures included peer relationships and teacher-student relationships.
Latent class growth analysis was then used to identify groups of students whose repeated responses followed similar patterns over time. This approach does not assume that every participant follows the average trend. Instead, it searches for statistically distinguishable developmental trajectories within the sample. Baseline characteristics were subsequently examined as correlates of membership in those classes.
Three developmental paths emerged
The largest class contained 1,915 students, or 73.5% of the sample. These children formed the low-stable trajectory. Their reported suicidal thoughts remained consistently minimal across the six annual measurements.
A second group of 501 students, representing 19.2% of the sample, followed a moderate-increasing trajectory. Their levels were persistently elevated relative to the low-stable group and gradually increased over time. This pattern is important because it identifies a sizeable group whose difficulties did not simply appear as a short-lived spike. Instead, elevated thoughts persisted and worsened across a major developmental transition.
The third class contained 191 students, or 7.3%. This initially-high declining group began with extremely high levels of suicidal thoughts. Their reported levels then improved substantially before showing a modest rebound later in the observation period.
That third trajectory is particularly informative. A decline from a high starting point is encouraging, but it does not mean that risk has permanently disappeared. The later rebound illustrates why repeated monitoring can reveal information that would be lost if researchers observed children only at baseline and again several years later.
Self-esteem produced the strongest baseline distinction
Among the baseline characteristics examined, self-esteem emerged as the strongest correlate of trajectory membership. Compared with the moderate-increasing group, higher self-esteem was associated with more than twice the odds of belonging to the low-stable group. The reported odds ratio was 2.187, with a 95% confidence interval from 1.422 to 3.363.
The relationship looked very different when the initially-high declining group was compared with the moderate-increasing trajectory. Higher self-esteem was associated with substantially lower odds of membership in the initially-high declining class, with an odds ratio of 0.149 and a 95% confidence interval from 0.084 to 0.264.
These estimates should be interpreted carefully. They show that baseline self-esteem distinguished the statistical trajectory groups after the study’s modelling procedure. They do not establish that raising self-esteem by a particular amount would cause a child to move from one trajectory to another. Self-esteem may reflect or interact with broader emotional, family, social and developmental conditions that were not fully captured by the analysis.
Gender also differentiated trajectories. Male students had a higher likelihood of membership in the low-stable group. The study considered a broader set of individual, family and school factors, reinforcing the idea that suicidal thoughts develop within a wider psychosocial context rather than as an isolated individual characteristic.
A worsening group can be hidden by population averages
The practical importance of the analysis lies less in a single odds ratio than in the heterogeneity it exposes. If all 2,607 students were represented by one average trend, the low-stable majority would dominate the overall pattern. That could make the developmental picture appear more reassuring than it is for the 19.2% whose trajectory was persistently elevated and gradually increasing.
Conversely, the initially-high declining group shows why an early severe response does not necessarily imply an unchanging course. Most students in that class improved over time, although the later rebound suggests that continued support and monitoring may remain relevant even after symptoms have eased.
The findings therefore support a developmental view of prevention. Universal school and community programmes may remain valuable because most young people cannot be sorted perfectly into future risk groups. At the same time, repeated assessment may help identify students whose pattern is worsening or whose earlier improvement is beginning to reverse.
What the study can and cannot tell us
The study has several strengths. It followed a large cohort over six consecutive years, covering an important developmental transition. Its trajectory approach allowed different patterns to emerge instead of forcing all participants into one average curve. It also considered baseline characteristics from individual, family and school domains rather than focusing on a single explanatory factor.
There are also important limitations. Suicidal thoughts were measured with a single item, which cannot capture the clinical detail available from a multi-item instrument or diagnostic interview. The analysis concerns reported thoughts, not suicide attempts or deaths, and those outcomes should not be treated as interchangeable.
The trajectory classes are statistical summaries of patterns in this particular dataset. They should not be interpreted as fixed psychological types or used to label individual children. Class membership is probabilistic, and different modelling choices or populations could produce different groupings.
Most importantly, the baseline associations are not proof of causation. A factor associated with a lower-risk trajectory may be protective, may reflect other underlying advantages, or may be linked to the outcome through several pathways. The results therefore identify useful targets for further study rather than demonstrating that changing any one baseline factor will prevent suicidal thoughts.
Prevention may need to respond to direction as well as severity
The research adds a simple but consequential insight to youth mental-health monitoring: where a young person is heading may matter alongside where they are now. A moderate level that is rising over several years can carry different implications from a high level that is falling, even when the two trajectories cross at a particular survey wave.
For schools, families and health systems, that argues for continuity. One-off screening can identify current distress, but repeated, developmentally appropriate assessment is better suited to detecting persistent deterioration, recovery and relapse. The study’s findings do not prescribe a specific screening schedule or intervention, but they provide evidence that meaningful variation exists across time.
The results also caution against assuming that a low population average means risk is evenly low. In this cohort, roughly one in four students belonged to one of the two trajectories characterised by elevated suicidal thoughts at some stage. At the same time, the majority remained in the low-stable group, and even the initially-high group showed substantial improvement. Both realities matter when communicating youth mental-health research responsibly.
Source Information
Study: Yoo, C. “Psychosocial perspectives on suicidal thoughts: developmental trajectories and associated baseline factors among South Korean children.”
Journal: Humanities and Social Sciences Communications
Published: 28 September 2026
DOI: 10.1057/s41599-026-09241-1
Study design: Prospective longitudinal secondary analysis of the Korea Children and Youth Panel Survey 2018, following 2,607 students annually from fourth through ninth grade between 2018 and 2023.
Research Today note: This article reports associations and developmental trajectories from observational data. It does not establish that the measured baseline factors caused later suicidal thoughts or that trajectory membership can predict an individual child’s future course.
