Suicide is a leading cause of mortality among young people. This study examined temporal trends in suicide rates and methods among individuals under 30 years in Spain between 2000 and 2023, compared with adults.
MethodsA time-series analysis of population-level suicide indicators based on data from the Spanish National Statistics Institute. A total of 85,299 suicides were classified by age group (<15, 15–29, ≥30 years), sex, and method. Joinpoint regression was used to evaluate trends. Annual Percent Change (APC) represents change within segments, whereas Average Annual Percent Change (AAPC) summarizes the overall period.
ResultsAmong those under 15, rates increased from 2011 onward (APC = 12.1%; 95% CI: 1.6 to 75.9), with defenestration the predominant method (41.7%) and showing a marked upward trend (AAPC = 25.4%; 95% CI: 10.2 to 42.5). In individuals aged 15–29, rates declined among males between 2000 and 2010 but increased among females from 2010 (APC = 4.5%; 95% CI: 1.7 to 17.6); and non-violent methods (poisoning or drug overdose) also became more frequent. In adults aged ≥30, overall rates were relatively stable, although non-violent methods increased from 2018 (APC = 8.1%; 95% CI: 2.0 to 23.6).
ConclusionAlthough these estimates should be interpreted cautiously given the small number of events, the results indicate a recent increase in suicides among children under 15 and young women, as well as age- and sex-specific changes in the methods used. Defenestration and the use of non-violent methods showed increases among younger population groups.
Suicide is a public health problem worldwide and accounts for approximately 800,000 deaths per year, roughly 1% of mortality [1]. For every completed suicide, it is estimated that there are between 20 and 25 attempts [2], highlighting that suicidal behavior extends far beyond mortality figures and underscoring the need to also address non-lethal attempts, which are the strongest predictor of future suicide [3]. Suicide causes profound individual and collective suffering, affecting family members and close acquaintances, who may themselves be at increased risk and show poorer social adjustment [4]. Mental disorders—particularly mood disorders, anxiety disorders, and eating disorders such as anorexia nervosa—are closely associated with suicide, increasing the risk up to tenfold in youths with psychiatric conditions compared to the general population [5,6]. Among children and adolescents, a previous suicide attempt increases the risk up to 30 times in males and threefold in females [6,7].
Suicide rates among young people have risen over the past 50 years, yet our ability to predict suicidal behavior has not significantly improved during this time [8]. In Europe, suicide among individuals under the age of 25 has gained increasing prominence as a public health concern [9]. In Spain, it is the leading cause of unnatural death among young people aged 15 to 29 [10], with a sharp 250% increase in suicide attempts recorded in 2020 compared to previous years [11], coinciding with the COVID-19 pandemic and its psychosocial impact [12]. Although suicide is rare in childhood, its prevalence rises markedly during adolescence [13]. Since the 2000s, European trends have varied: the United Kingdom has seen a sustained increase—particularly among females—whereas rates have remained relatively low in Spain and Italy [13]. Understanding the methods used is key for prevention, given their direct relationship with lethality: among individuals aged 10 to 24, the most common methods are hanging, strangulation, and suffocation, with prevalence rates varying by country and sex [13].
Despite growing interest, few studies have systematically compared trends in suicide methods across age groups over an extended period [14,15]. In particular, children under the age of 15 have received little attention, as have specific methods such as defenestration, suffocation, or hanging, whose recent appearances in official records raise important questions. This emphasizes the need for a more specific, longitudinal prevention approach that addresses these behavioral transitions [16,7]. While previous Spanish and European studies have described overall suicide trends, fewer have jointly examined age-, sex-, and method-specific patterns over long observation periods, and children under 15 years remain particularly understudied. In this context, Joinpoint regression offers a useful approach to detect when relevant changes in direction or magnitude occur.
This study aims to analyze the evolution of suicide rates in Spain between 2000 and 2023, with a particular focus on the methods employed among children, adolescents, and adults. The analysis includes stratification by sex and age group, and Joinpoint regression models are applied to identify significant trend changes.
MethodsStudy designThis is a time series analysis study of population-level suicide indicators based on records that analyzes temporal trends in suicide frequency and methods used among children (<15 years) and youth (15–29 years), compared to adults (>30 years), in Spain from 2000 to 2023. The study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [17].
Data sources and variablesSuicide deaths were obtained from the Specific Causes of Death Statistics of the Spanish National Institute of Statistics (INE, for its Spanish acronym) (INE, n.d.) [18]. These official mortality records include all suicide deaths registered in Spain during the study period [19]. The variables analyzed included age at death, sex, year of death, and method of suicide. Age was categorized into three groups (<15 years, 15–29 years, and ≥30 years) based on the predefined groupings provided by the INE, which did not allow for further disaggregation. Suicidal deaths data range from 2000 to 2023 and method-specific information is available for the period 2005–2023.
Suicide methods were classified according to the International Classification of Diseases, 10th Revision (ICD-10), as provided by the INE [20]. We categorized the methods as violent or non-violent based on their lethality, following Åsberg's criteria [21,22]. This dichotomy is particularly relevant in the context of youth suicide, as violent and non-violent methods have been associated with distinct psychological profiles — impulsivity and limited planning in the former, and greater premeditation in the latter — with direct implications for prevention strategies [28,31]. Violent methods were further divided into three categories due to high prevalence of hanging and defenestration. Specifically, hanging (X70) and defenestration (X80) were analyzed separately. The remaining violent methods included high-impact physical actions such as firearms (X72, X73, X74), self-immolation (X75, X76), drowning (X71), cuts or sharp object injuries (X78), car-related suicides (X82), and other unspecified methods (X79, X81, X83, X84). Non-violent methods encompassed cases related to substance ingestion, including sedative drugs (X61, X62, X65), non-sedative drugs (X60), ingestion of harmful substances such as corrosives and pesticides (X66, X67, X68, X69, X77), and other drug-related poisonings (X63, X64).
Statistical analysisSuicide mortality ratesCrude suicide mortality rates were calculated by dividing the number of suicide deaths per year by the corresponding population in the same year. All rates are expressed per 100,000 population and stratified by sex. Confidence intervals at 95% (95% CI) were estimated. Since the study aimed to compare age-stratified groups within the same population rather than across different populations, age-adjusted rates were not calculated. Age standardization is primarily required when comparing populations with different demographic structures; therefore, adjustment was not expected to substantially alter interpretation. For graphical representation, we applied local regression smoothing using the LOESS (locally estimated scatterplot smoothing), which fits smooth curves to the data points to better visualize trends over time [23]. These smoothed curves were used for descriptive visualization only and were not employed for statistical inference, which relied on Joinpoint regression models.
Comparison of sex and suicide methods by age groupTo compare differences in the distribution of sex and suicide methods across age groups, two statistical tests were used depending on cell frequencies. The Chi-square test (χ²) was applied when all expected frequencies were ≥5; otherwise, Fisher’s exact test was used. In addition to global comparisons, pairwise comparisons between age groups were conducted to identify specific differences. To control for multiple testing, p-values were adjusted using the Bonferroni correction. These analyses provide a cumulative descriptive overview for the whole period and do not account for temporal variation, which is examined using Joinpoint regression. They describe the proportional distribution of methods among deaths rather than population risk.
Temporal trend analysisTo analyze temporal trends in suicide methods and identify periods where significant changes occurred between 2000 and 2023, Joinpoint regression models were constructed [24,25]. Joinpoint regression fits a series of log-linear segments to the observed rates, allowing the slope to change at statistically identified time points. This approach enables the identification of potential inflection years and the estimation of trends using the average annual percentage change (AAPC) for the overall period and the annual percentage change (APC) within each segment [26], with corresponding 95% confidence intervals to quantify uncertainty.
Joinpoints were selected using the Bayesian Information Criterion, specifically the Weighted BIC (WBIC), which balances model fit and parsimony and reduces the likelihood of spurious change points. The number of joinpoints was restricted according to program defaults and data availability. Rates were modeled on the logarithmic scale. Years with zero counts were assigned a constant rate of 0.01 per 100,000 to allow log-transformation within the Joinpoint model, and are acknowledged as carrying greater uncertainty.
As with any segmented regression, estimates—particularly in strata with small numbers—may be sensitive to random fluctuation. Therefore, interpretation emphasized the magnitude and direction of trends and their confidence intervals rather than statistical significance alone. Identified joinpoints should be understood as statistical approximations rather than exact moments of change.
Statistical analysesData cleaning, harmonization, and group comparison analyses were conducted using RStudio (R version 4.2.2). Group comparisons for specific categorical variables (i.e., age group, sex, country of birth, socioeconomic level, medical comorbidities, and smoking status) were performed using the chi-square test. Given the large sample size, Cramér's V was used to assess the strength of associations, with the following interpretations: small (V = 0.1), moderate (V = 0.3), and large (V = 0.5) [27]. To account for multiple pairwise comparisons, p-values were adjusted using the Bonferroni correction. Statistical significance after adjustment was set at p < 0.00278 (0.05 / 18 comparisons). Joinpoint regression analyses were conducted using the Joinpoint Regression Program developed by the Surveillance Research Program of the U.S. National Cancer Institute (Statistical Methodology and Applications Branch, 2024). Statistical significance was set at p < 0.05.
ResultsSuicide distribution by age and sexBetween 2000 and 2023, a total of 85,299 suicides were recorded in Spain. Of these, 76,729 cases (89.9%) occurred among individuals aged 30 years or older, 8354 cases (9.8%) among those aged 15 to 29 years, and 216 cases (0.25%) among individuals under 15 years of age (Table 1).
Distribution of Suicide Methods by Age Group.
Note: df, degrees of freedom; V, Cramer's V. ‡While suicidal deaths data range from 2000 to 2023, method-specific information is available only for the period 2005–2023. ◊Percentages in this row represent the proportion of suicides within the entire dataset, calculated within each age group. All other percentages indicate the distribution within each age group. * Indicates p < 0.05. Effect size (Cramér’s V) should be considered when assessing the magnitude of associations.
The distribution of sex across age groups differed between groups, although the association was weak (χ² = 55.41, p < 0.001; V = 0.025). Women represented a smaller proportion of cases in the 15–29 (21.6%) and ≥30 (24.7%) age groups, but accounted for a notably higher percentage in the under-15 group (36.1%). Pairwise comparisons detected differences between groups; however, effect sizes were small to negligible (V ranging from 0.014 to 0.055) (Table S1).
Suicide method distribution by age groupHanging was the most commonly reported method across all age groups, accounting for approximately 47–48% of suicides, with very similar proportions across age groups (Table 1).
In contrast, defenestration was significantly more prevalent among individuals under 15 (41.7%), compared to 29.1% in the 15–29 group and 23.4% in the ≥30 group, although the strength of association was small (χ² = 129.18, p < 0.001; V = 0.043). Pairwise contrasts pointed in the same direction, although effect sizes remained small.
Other violent methods were less frequently used among individuals under 15 (9.5%) compared to those aged 15–29 (16.2%) and ≥30 (16.8%) (χ² = 7.61, p = 0.022; V = 0.010). However, none of the pairwise comparisons remained significant after Bonferroni correction.
Non-violent methods were the least frequent among individuals under 15 (1.8%) but more common in older groups (8.1% in 15–29 and 11.9% in ≥30) (χ² = 93.79, p < 0.001; V = 0.037). Significant differences were observed between the <15 and ≥30 groups, and between the two older groups (p < 0.001 for both) with small effect sizes, while the comparison between <15 and 15–29 did not survive correction (p = 0.004). (Table 1).
Trends in suicide rates over timeThe Joinpoint analysis revealed changes in suicide rate trends across different age groups and by sex in the period studied (Table 2, Fig. 1).
Age- and Sex-Specific Rates and Estimates from Joinpoint Analysis in Spain, 2000–2023.
Note: AAPC: average annual percentage change for the entire study period; APC: annual percentage change within each segment defined by joinpoints. * Indicates p < 0.05. Estimates for age groups with small numbers, particularly those younger than 15 years, may be unstable and should be interpreted with caution.
Joinpoint Analysis of Age- and Sex-Specific Suicide Rate Trends in Spain, 2000–2023.
Legend: Joinpoint regression analysis of annual suicide rates per 100,000 inhabitants, stratified by age group and sex. (A) Individuals aged ≥30 years; (B) individuals aged 15–29 years; (C) individuals under 15 years of age. Solid lines represent fitted joinpoint models for each subgroup. Annual percentage changes (APC) and time periods are shown in the legend. Sex categories: blue = males, grey = total, purple = females.
In the overall population, suicide rates among individuals under 15 years showed a decrease until 2011, followed by a significant increase from 2011 to 2023 (APC = 12.1%; 95% CI: 1.6 to 75.9); however, the exceptionally wide confidence interval reflects the small number of events in this age group, and the point estimate should therefore be interpreted with considerable caution. Among individuals aged 15–29 years, there was an overall decline (AAPC = –1.3%; 95% CI: –2.2 to –0.3), primarily driven by a sharp decrease between 2000 and 2010 (APC = –4.7%; 95% CI: –10.9 to –2.3). For those aged 30 and older, no clear overall trend was observed, although two periods of significant change were detected: a decline from 2000 to 2009 (APC = –1.5%; 95% CI: –5.3 to –0.4) followed by a modest increase from 2009 to 2023 (APC = 0.8%; 95% CI: 0.2 to 2.9)
Among females, an increase was estimated in the 15–29 age group, where an increase in suicide rates was detected from 2010 to 2023 (APC: 4.5%; 95% CI: 1.7 to 17.6). No significant trends were identified in the <15 or ≥30 age groups.
Among males, the 15–29 age group showed an overall decrease in suicide rates (AAPC = –1.8%; 95% CI: –2.7 to –0.9), with a marked decline from 2000 to 2010 (APC = –4.8%; 95% CI: –11.0 to –2.7). In the ≥30 group, suicide rates declined from 2000 to 2009 (APC = –1.4%; 95% CI: –5.6 to –0.2), followed by a small increase from 2009 to 2023 (APC = 0.7%; 95% CI: 0.0 to 3.8). No significant trends were found among males under 15.
Trends in suicide methods over timeThe Joinpoint analysis identified different temporal patterns in suicide methods across age groups during the study period (Table 3).
Age- and Method-Specific Rates and Estimates from Joinpoint Analysis in Spain, 2000–2023.
Note: AAPC: average annual percentage change for the entire study period; APC: annual percentage change within each segment defined by joinpoints. * Indicates p < 0.05. Estimates for age groups with small numbers, particularly those younger than 15 years, may be unstable and should be interpreted with caution.
For hanging, no joinpoints were identified in any age group; therefore, a single trend was estimated for the entire period. These trends remained generally stable over time, with no clear increases or decreases.
For defenestration, joinpoints were detected in some age groups, allowing period-specific estimates. Among individuals under 15 years old, an overall increase was estimated (AAPC: 25.4%; 95% CI: 10.2 to 42.5), corresponding to a sustained upward trend from 2005 to 2023 (APC: 25.4%; 95% CI: 10.2 to 42.5). In the 15–29 age group, an increase was observed from 2010 to 2023 (APC: 3.1%; 95% CI: 1.2 to 12.9). No joinpoints or clear trends were identified in the ≥30 age group.
For other violent methods, no joinpoints were detected in the <15 or ≥30 groups, and patterns remained broadly stable. In contrast, in the 15–29 age group, an overall decline was estimated (AAPC: –2.9%; 95% CI: –5.4 to –0.4), reflecting a consistent decrease from 2005 to 2023 (APC: –2.9%; 95% CI: –5.4 to –0.4).
For non-violent methods, joinpoints were identified in some strata. In the 15–29 age group, an overall increase was estimated (AAPC: 3.6%; 95% CI: 1.6 to 5.7), corresponding to a rise from 2005 to 2023 (APC: 3.6%; 95% CI: 1.6 to 5.7). Among individuals aged ≥30 years, an increase was also detected (AAPC: 3.8%; 95% CI: 2.1 to 5.1), with a steeper rise from 2018 to 2023 (APC: 8.1%; 95% CI: 2.0 to 23.6). No joinpoints or clear trends were observed in the <15 age group.
DiscussionIn this nationwide registry-based study, we analyzed suicide trends in Spain from 2000 to 2023, focusing on minors and young adults: (1) Among individuals under 15 years suicide rates increased from 2011, with defenestration more prevalent than in older groups (2005–2023).(2) In the 15–29 age group, overall rates decreased, especially among males (2000–2010), while female rates increased after 2010. Defenestration rose, other violent methods declined, and non-violent methods became more common. (3) Among adults aged 30 years or older, trends were more stable, with a decline from 2000 to 2009 followed by a moderate increase (2009–2023), particularly among males. Non-violent suicides rose from 2018 onward. No significant trends were observed for hanging or other violent methods.
Among those under 15, our findings reveal a rising trend in suicide since 2011—particularly in defenestration—consistent with data from the Spanish National Statistics Institute (INE) [18] and studies on impulsivity and lower planning in child suicides [16,28]. Nevertheless, these estimates are derived from a small number of deaths and may therefore be unstable; even large relative increases should be interpreted with caution. The increase in suicide rates in minors since 2011 may be linked to parental factors [29], consistent with literature on the impact of socioeconomic context on children's mental health [30]. It is also important to consider other potential triggers for child suicide identified in recent research, including the presence of mental disorders, previous suicidal ideation or behaviors, traumatic experiences, family conflicts, school-related issues, and immediate stressors like punishment or arguments on the day of the suicide [16]. In children, a lower prevalence of psychopathology is observed compared to adolescents, along with a higher frequency of hanging and lower substance use, suggesting high impulsivity and limited planning [31].
Although no direct impact of the economic crisis on children's mental health has been identified [32], the observed increase may reflect a delayed effect of the financial downturn, as suggested by previous studies [33]. Among those aged 15–29, a decline in suicide rates was observed among males, while females experienced a sustained increase starting in 2010. This pattern has been linked to rising rates of affective disorders and suicidal ideation among young women [30]. The growing use of non-violent methods by females also aligns with studies on gender differences in suicide patterns. In contrast, other violent methods have decreased in this group, consistent with findings by Ramalle-Gómara et al. [34], who suggest that factors such as the impact of the economic crisis, gender roles, and social media influence may be linked to changes in method lethality.
Hanging remains the most frequent method across all age groups, including children, as reported by previous studies [35,36,31]. However, the increase in defenestration among minors since 2005 stands out, potentially related to impulsivity and easy accessibility [16]. This pattern is also seen in adolescents and young adults, with greater use of defenestration and non-violent methods among females, consistent with prior research [30,37]. Overall, suicide method patterns in children and adolescents reflect lethality, availability, and the social acceptability of means in ways that mirror national patterns across all age groups. Means restriction holds great potential for youth suicide prevention [31].
Among adults aged 30 or older, method trends appear relatively stable, except for an increase in non-violent methods from 2018 onward, potentially linked to social or cultural changes or increased medicalization, as recent reviews suggest [38]. The sex disparity in suicide—males having rates two to five times higher than females—persists in Spain [39,40]. While men still show higher suicide rates overall, a significant increase among young women has been detected since 2010, in line with Global Health Estimates 2019 [41]. Globally, WHO data show a general decline in suicide rates in Europe since the 1990s, though some countries have recently reported increases [13]. Our results partially align with this trend: we observed a decrease in suicide rates among the 15–29 population between 2000 and 2010, followed by a rise among females, consistent with findings from other countries [31]. Studies such as that by Cayuela et al. [38] identified key inflection points in 2000 and 2010. Our findings also highlight these years as critical moments—first a decline, followed by a resurgence among adults over 30, particularly men, in the context of the 2008 financial crisis [42,43]. This crisis led to a marked deterioration in mental health, particularly among men, for whom unemployment and the loss of the traditional provider role were key factors. This situation exacerbated socioeconomic inequalities and contributed to rising psychological distress and suicide risk among working-age males. These effects were measured at the population level using the GHQ-12 questionnaire, first included in the Spanish National Health Survey in 2006–2007, which revealed a significant increase in psychological distress during the recession years [42].
Limitations and strengths of the studyThis study has several limitations. The analysis relied solely on mortality registry data, without access to sociodemographic or clinical information about the deceased, their prior contact with mental health services, substance use at the time of death, or psychological autopsy data, all of which would be needed to understand the personal and social context of each suicide. Additionally, the small number of events in the under-15 age group limits the reliability of the estimates in this stratum; the detected trend change showed sensitivity to the choice of model selection criterion in the Joinpoint analysis and should therefore be interpreted as a preliminary signal rather than a definitive finding.
Strengths include the use of official, public, and updated data from the Spanish National Statistics Institute (INE), covering a long period (2000–2023) that allows identification of patterns and turning points, some coinciding with major events such as the 2007–2014 economic recession [44] and the SARS-CoV-2 pandemic (2020–2022) [45]. Data quality improved progressively over the study period: in 2007, judicial bulletins were replaced by standardized death certificates following international standards, and the subsequent integration of the Legal Medicine and Forensic Sciences Institutes (IMLCF) further enhanced cause-of-death coding specificity [46–48]. These changes likely reduced underreporting, though they may also partly account for observed increases in registered cases in the earlier part of the study period.
ConclusionThis study identified a significant increase in suicide among children under 15 years old over the past two decades, with high use of defenestration. In the 15–29 age group, overall suicide rates declined, though a concerning rise among young women has been observed since 2010. Among adults aged 30 and over, rates remained relatively stable, with a recent increase in the use of non-violent methods. These findings underscore the need for age- and sex-specific preventive interventions, as well as integrated strategies that address the clinical, social, and structural factors associated with suicide. Studies like the present one are fundamental to supporting health policies grounded in outcome-based evidence.
Ethical considerationsThis work complies with the ethical criteria for research.
FundingM.I.O. is supported by an AGA fellowship from the Intramural Research Grant Program (IREP), funded by IRB Lleida- CERCA centre / Generalitat de Catalunya.
The author declares no conflict of interest and states that no funding was received from any public or private entity for the conduct of this study.
