Our study investigated the suicidal transition rate in 133,661 French adults living in mainland France, of whom 6,456 reported past 12-month suicidal ideation. It represents one of the largest studies published on the topic. We found mean suicidal transition rates of 7.7% (95% CI (6.8–8.6)) in the French adult general population across seven independent samples interviewed between 2000 and 2021. These rates varied over 21 years between a minimum of 4.5% in 2005 and a maximum of 11.9% in 2014. Of note, a non-significant increase in suicidal transition rates was found after the Covid-19 pandemic started in early 2020, from 6.5% in 2017 to 7.9% in 2020 and 9.3% in 2021. Several factors were significantly associated with the transition rate. The factor associated with the highest risk of suicidal transition was a previous suicide attempt (aOR = 11.1). Younger age (18–25-year-olds) showed the highest transition rates across age groups with a mean transition rate of 13.8%, while older ages (56–75-year-olds) showed the lowest (mean = 4.8%). In contrast, there was no gender difference. Additional independent sociodemographic factors were lower income and lower and intermediate professional categories.

The study had several limitations. First, main outcomes (suicide ideation and attempt) were based on self-reports in a phone interview. Memory and desirability bias may have affected reports of these variables. Moreover, no definition of suicidal ideation and suicide attempt was given during the interview, which may have led to heterogeneity in measures. Second, the design of the study was retrospective. A longitudinal study may best capture the temporal sequence from suicidal ideation to an act although our findings appear to be within range of previous longitudinal cohort studies as discussed below. Third, while the number of participants was high, the number of people who attempted suicide was relatively limited (overall N = 537), which may have reduced the necessary statistical power to identify particular risk factors. Fourth, the Baromètre Santé does not include adolescents, a population with a high rate of self-harm [13] as well as people currently hospitalized in psychiatry and inmates. Fifth, many relevant variables are not available including family history, exhaustive mental disorder history, recent life events, and treatments. Sixth, 0.1% of the interviewed people refused to answer about 12-months suicidal ideation. They appeared to be less socially favored than those who answered in terms of incomes, education, work and marital status. Among these four variables, only lower income was associated with higher transition rates. Therefore, our findings related to transition rates may be slightly underestimated.

The mean suicidal transition rate found here is within the range (2.6–37%) of a recent review of 18 longitudinal studies in non-clinical populations [5]. In a comparable study in the USA (adult general population with 12-month follow-up using mailed questionnaires), the transition rate was higher, reaching 15.7% (10.6–20.8) [3]. In a Dutch adult population followed for 36 months, the transition rate was 12.2% (11.2–13.2) [14]. In a US adult cohort followed for 156 months, the transition rate was 10.3% (3.6–17) [15]. Much higher rates (above 20%) were rare and found in specific populations, including adolescents and army veterans [5]. Overall, our study adds to an unequivocal literature showing that a large majority (above 80% and usually 90%) of people with suicidal ideation in the general population do not attempt suicide.

In our study, we compared transition rates across different years in independent samples. Mean transition rates varied across surveys between a minimum of 4.5% in 2005 and a maximum of 11.9% in 2014. However, an explanation may be that small changes in 12-month self-reported rates of suicidal ideation and suicide attempt may have a large impact on the ideation/attempt ratio. For instance, a significant increase in transition rates was observed in 2010 and a significant decrease in 2017. The 2010 increase (vs 2005) was related to a combination of an increase of 0.2% in suicide attempt rate and a decrease of 1.2% in suicidal ideation rate while the 2017 decrease (vs 2014) was related to a decrease of 0.4% in suicide attempt rate along a limited decrease of 0.3% in suicidal ideation rate. In these conditions, it is difficult to conclude that some periods in time (here, somewhere between 2005 and 2017) may have had an impact of the suicidal transition process per se. While variations in suicidal rates seem to be related to variations in self-reported measures of suicidal ideation and attempts, no clear changes in suicidal transition rates appeared over the last 21 years and no clear temporal trends. Therefore, following the temporal changes in suicidal transition rates may usually not be relevant.

The strongest risk factor of suicidal transition was by far a previous suicide attempt with an increased risk multiplied by 11 as compared to people with no such history. This is in line with an important literature [16, 17]. Two non-exhaustive mechanisms may be in play. A past attempt may reveal the existence of long-term vulnerability factors enabling acting out in conditions of suicidal ideation and stress. The experience of a past suicide attempt may, in some individuals, also facilitate the realization of another act when suicidal following a process described as “acquired capability” or increased facilitation [18].

While gender does not seem to affect transition rates as previously reported [3, 14, 15], age contributed to significant differences. In our study, young adults below 25 were notably associated with higher rates over the seven surveys and overall, with an aOR of 1.8 as compared to middle-aged adults. However, these findings are conflicting. Cohorts with adolescents have shown low [19, 20], to very high [21, 22] suicidal transition rates. Borges et al. [3] found a trend toward higher rates in younger adults, but their choice of a larger age range (18–44-year-old) may have masked any effect. However, an age effect was not found in a subsequent study with a narrower age range (15–24) [16]. Similarly, no age effect was found in a Dutch adult cohort [14]. In contrast, the Baltimore cohort, which focused on incident suicidal ideation and attempt, reported higher transition rates in younger adults (18–29) [15]. Younger age is associated with higher impulsivity, peaking around 16 years-old (with some interindividual heterogeneity) [23] and impulsivity is associated with a higher risk of suicide attempt in young people [24]. Additional individual and societal factors may contribute to higher acting out in young suicidal people including contagion and media effects [25]. A potential higher transition rate in younger individuals needs more investigation.

We furthermore report that lower socioeconomic status (including specific professional categories such as manual workers, employees, intermediate professions, and low income) was independently associated with higher suicidal transition rates. This association has previously been reported [3, 15, 26]. The causal mechanisms are likely multifactorial, including, but not limited to, higher stress, more developmental difficulties and mental issues, lower access to care and cascade effects [27]. Reversing socioeconomic disadvantages may help to prevent suicidal acts.

These findings may have scientific, clinical and policy implications. First, our findings support the distinction between suicidal ideation and act, and the avoidance of the generic term “suicidality”. Therefore, research should continue to explore the social, clinical, psychological, and biological mechanisms of the suicidal transition and distinguish them from those facilitating the emergence of suicidal ideation. Current models of suicidal behavior, such as the “ideation to action” theories [28, 29], make a distinction between processes leading to ideation (e.g., pain, hopelessness) and those leading to action (e.g., access to lethal means, personal history of suicidal act). Although we did not examine risk factors of suicidal ideation here, our study reported factors more specifically associated to action. Second, the low rate of suicidal transition explains the limited value of suicidal ideation to predict a suicidal act. Identifying the factors and mechanisms of the suicidal transition may improve prediction, follow-up, and organization of care with a patient in crisis. Third, preventative strategies targeting all people with suicidal ideation, e.g. specific hotlines for people in suicidal crisis, may have limited effects on the prevention of suicide attempts and suicide death as most suicidal people will not “transition”. People in suicidal crisis may more effectively benefit from an easy access to primary care, combined with a good psychological and psychiatric evaluation. Studies have shown that up to 60% of people who died from suicide had seen their physician or presented to an emergency room the month before their death [30], suggesting potential improvement possibly through training. In contrast, focusing on people who have attempted suicide as currently promoted by some preventative strategies such as “VigilanS” in France [31], may be relevant. Meta-analyses have shown that the risk of repeated self-harm within one year of self-harm hospitalization is 16.3% while the risk of suicide is 1.6% [32]. Finally, improvement in socioeconomic status may have significant effect on the prevention of suicidal acts. All these suggestions are speculative, and a formal assessment of national strategies (both suicide-related and socioeconomic) is needed.

Finally, no clinical condition was statistically associated with higher suicidal transition rates in multivariable analyses, although a trend toward significance was found for daily smoking. Contrary to expectations, we did not find a higher transition rate in heavy drinkers after controlling for smoking. Similarly, a current depressive episode was not related to higher suicidal transition as previously reported by some [3, 7], but not all studies [14, 15].

In conclusion, we report a mean 7.7% suicidal transition rate, with yearly fluctuations, in a large sample of French adults aged 18 to 75. Nine out of ten suicidal people, therefore, do not attempt suicide in the general population. Variations across time may not be related to temporal trends in transition. Higher suicidal transition was independently associated with a history of previous suicide attempt, younger age and lower socioeconomic status. Clinicians examining suicidal patients should be aware of these findings when organizing care and follow-up. Extra vigilance for those with a personal history of suicide attempt is highly recommended.