In this study where self-reported information were linked with register data on an individual level, we found that approximately 1/3 of 18-year-olds had experienced thoughts of suicide, while 9% girls and 7% boys had more developed suicide thoughts in the form of planning by considering methods and making preparations. Additionally, almost 6% girls and 3% boys aged 18 years had had a suicide attempt. Our results further quantify the ‘hidden number’ of adolescents with suicide attempts, as for every girl and boy who presented to hospital with a suicide attempt, two girls and six boys, respectively, had reported a suicide attempt that had not resulted in hospital presentation. A social inequality was found for suicide attempts. The algorithm used to identify hospital-recorded suicide attempts aligned well with self-reported data, especially among girls.

A previous study indicated that among Danish high school students (the majority aged 15–20 years), 3·4% girls and 1·8% boys reported having ever attempted suicide [8]. This is lower than the estimates presented in this study, possibly because high school students represent a more well-functioning group than the DNBC-18 population that also includes adolescents who are not in high school. In an international meta-analysis, the life-time prevalence of suicide ideation, suicide plans and suicide attempt among children and young people aged 6–22 years was estimated to be 21%, 10% and 7%, respectively [11]. Thus, the prevalence of suicide ideation was higher in our study, while suicidal plans and attempts were less common.

Supporting previous research, we showed a substantial hidden number of adolescents with suicide attempts [5, 6]. Similar to our findings, a previous English study found the ‘hidden number’ to be higher in boys than girls in children aged 12–14 years although no sex-differences in the ratios were observed for adolescents aged 15–17 years [5]. In contrast, an Irish study found that the ‘hidden number’ was higher in girls than boys [6]. These diverging results could be explained by cultural and societal differences in treatment seeking and availabilities between countries and methodological differences between studies. Furthermore, the UK and Irish studies examined ‘self-harm’, which includes all intentional episodes of self-injury or self-poisoning irrespective of motive (including suicidal intent) [5], while in our study the adolescents were specifically asked whether they had tried to take their own life.

This is, to our knowledge, the first time that data on hospital- and self-reported suicide attempts, plans and ideation have been linked at an individual level. By integrating self-reported and hospital records for the same individuals, more exact estimates could be made. Additionally, previous literature was based on data from selected areas e.g., hospitals and cities, whereas this study used nationwide data and a well-defined background population of Danes born during the same period as participants with register data on suicide and hospital-recorded suicide attempts. Thereby we were able to calibrate findings to the background population of all children born in Denmark in 1996–2003.

Our findings support the perception that suicide attempts are more frequent in girls than boys [1, 5, 6]. However, a smaller sex-difference was found for self-reported suicide attempts compared to hospital-recorded suicide attempts, with a boy-to-girl ratio of 1:2 and 1:4 respectively. Thus, the sex-difference could be overestimated in studies where only hospital-recorded suicide attempts are included. Self-reported suicide ideation and suicide plans were likewise more common among girls, although the sex-difference was less substantial.

We demonstrated a social gradient, with the highest prevalence of suicide attempts among adolescents with low parental income. Danish residents have universal and free access to health care, which could explain the relatively constant ratio between suicide attempts with and without hospital contact across income groups. In countries with unequal access to health care, the ratio might vary more between income groups.

A part of the DNBC-18 data collection was conducted during the national lockdowns implemented to mitigate the COVID-19 pandemic. As we previously found the prevalence of suicidality in DNBC-18 to be similar before and during lockdown, we consider data collected in these periods to be comparable [22]. Even though healthcare-seeking at hospitals in general might have been compromised during lockdown, a Danish study comparing rates of hospital-recorded suicide attempt before and during the first and second lockdown did not observe this in adolescents [23].

The Danish algorithm used to identify hospital-recorded suicide attempt has previously been evaluated in all age-groups by researchers with a medical background reviewing medical records of 357 randomly identified cases of suicide attempt for descriptions of suicidal intentions [13]. The algorithm could reliably identify suicide attempts with suicidal intention in 42% men and 59% women. In our study among adolescents, we likewise found that the algorithm was more valid among girls. However, it is not possible to distinguish whether the algorithm is more likely to misclassify boys or boys are less likely to self-report suicide attempt. In the algorithm, suicidal intent is a prerequisite in one out of four definitions. The importance of suicidal intent and the distinction between non-suicidal self-injury and suicidal acts has been widely discussed [24, 25]. A psychiatric diagnosis in combination with injuries to the lower forearm or intoxication may reflect non-suicidal self-injury, misuse of drugs, or accidents rather than a suicide attempt. However, among individuals identified with a hospital recorded suicide attempt, without any confirmation of intent, i.e., by definition (b), (c), or (d), 86% reports having had suicidal ideation and 70% reported a suicide attempt. Despite a somewhat lower threshold than for records with a confirmed suicide attempt, this supports that the major share of people identified through definitions without confirmed intent did have a suicide attempt. The true number of suicide attempts with hospital presentation is expectedly in-between the number of suicide attempts identified by definition (a) and definitions (b), (c), and (d). The algorithm developed to identify suicide attempts in Danish register data can, thus, reliably be used for suicide research in Denmark, although it will not identify the large number of suicide attempts without hospital contact.

In average in every classroom of 30 pupils as many as 2–3 individuals aged 18 years have either seriously considered or attempted to take their own life, with girls and adolescent from lower income families being overrepresented. The high prevalence of suicide ideation among adolescents is especially concerning when taking the moderate to high transition rate to suicide attempt into consideration [26]. The substantial ‘hidden number’ of adolescents with suicide attempts is also concerning. Medical treatment following a suicide attempt is an opportunity for healthcare professionals to assess the underlying cause(s), which contributed to an individual’s suicidal behaviour, and refer them to appropriate treatment. Since many adolescents, especially boys, do not present to hospital following a suicide attempt or receive psychiatric treatment afterwards [27,28,29], offering help to individuals in the community is crucial for preventing repetition and addressing underlying mental health issues. Future research should address the mechanisms of community occurring suicide attempts and how these differ from attempts recorded at the hospital. A review has suggested that one out of three adolescents do not seek help after a suicide attempt, not even from informal sources, such as family members [7]. Secrecy could imply that appropriate help or treatment needed is not provided. This emphasizes the importance of ensuring that adolescents and their parents, who often serve as the first line of care takers, have knowledge and easy access to help, such as mental health hotlines and Suicide Prevention Clinics. The ‘hidden number’ of suicidality further underscores the need for community-based strategies for suicide prevention, preferably at an early age, as suicide ideation and behaviour may occur in childhood [30]. Several trials of school-based suicide prevention programs have yielded encouraging findings in reducing suicidality in school pupils [31]. Thus, a way forward could be implementing school-based suicide prevention as an important tool for early identification and intervention to prevent the development of suicidality [32]. This is particularly important among boys as they less commonly present to the hospital, especially considering that boys are more likely to die by suicide [2, 4].

Strengths of this study include the individual-level linkage of register-based and self-reported data, multiple measures of suicidality, a large well defined study population and utilization of sample weights. Pilot-testing ensured that questions on suicidality were interpreted as intended. Data on suicides from the Cause of Death register have previously been evaluated as reliable [33].

A limitation of this study is the possible underestimation of suicidality due to the fact that DNBC-18 constitute a selected population. Psychiatric studies have been suggested as particularly sensitive to attrition [34, 35]; however, when compared to individuals born in Denmark in the same years as the DNBC participants, the proportion having a psychiatric diagnosis was not markedly lower (11% vs. 12%). When sample weights were applied, the prevalence of suicidality increased, although this might still be somewhat underestimated. Hospital-recorded suicide attempts are under-recorded and the algorithm used to identify suicide attempts in the registers lacks both specificity and sensitivity; [12, 13] thus, implying a potential source of bias. Suicidality questions were not from a validated questionnaire and answers may be subject to reporting bias. The question about suicide ideation was vaguely phrased, with the formulation, ”even though you would not do it,” which might explain the very high frequency. It was not possible to estimate the extent of suicidality including suicide plans for the whole population, as data only was available for a sub-group. Information about contact with general practitioners regarding suicide ideations and attempts was not available but would have been valuable to this study. Given that information on frequency or timing (i.e., date) of the self-reported suicide attempt was not available, we could only estimate the proportion of the population with suicidality, whereas most research report rates of suicidality, i.e., based on more detailed information on time at risk. Further, adolescents with hospital-recorded suicide attempt may also have had other ‘hidden’ suicide attempts that did not lead to hospital contact.

In conclusion, a large proportion of adolescents experience suicidal ideation or behaviour, especially girls and adolescents in lower income families. In a classroom of 30 students, on average, 2–3 individuals will have either planned to or attempted to take their own life. Further, the substantial ‘hidden number’ indicate that the majority of adolescents do not seek hospital treatment immediately after a suicide attempt. Universal prevention strategies targeting adolescents with suicidality should be focused on the community, especially schools, and take account of social inequality.