Past suicide attempts

Within our cohort, 43% of patients reported a history of suicide attempts. Similar rates, from 36.6 to 43%, were observed in other AUD populations, ^{Reference Wojnar, Ilgen, Jakubczyk, Wnorowska, Klimkiewicz and Brower20–Reference Preuss, Schuckit, Smith, Danko, Bucholz and Hesselbrock22} although another study reported a slightly lower rate of 27%. ^{Reference Pandey, Bolstad, Lien, Walby, Myhre and Bramness23}

Univariate analyses comparing patients with and without prior suicide attempts revealed markers suggestive of a more severe AUD course among those with a history of suicide attempts, including earlier onset of AUD symptoms and greater maximum quantities of alcohol consumed per occasion or within 24 h. Interestingly, patients with past suicide attempts also reported significantly more days of abstinence in the preceding 6 months. The combination of higher maximum alcohol consumption per occasion – reflecting greater severity – and a higher number of alcohol-free days during the preceding 6 months suggests an irregular drinking pattern characterised by intermittent heavy consumption interspersed with abstinence. This result implies that, besides the severity of the disorder, which has been linked to increased prevalence of past suicide attempts, ^{Reference Preuss, Schuckit, Smith, Danko, Buckman and Bierut7,Reference Pandey, Bolstad, Lien, Walby, Myhre and Bramness23} consumption patterns may also heighten the risk of suicidal behaviour. Indeed, acute alcohol use is also a risk factor for suicidal acts. ^{Reference Conner and Bagge24} High rates of acute alcohol use have also been noted among individuals with AUD who died by suicide. ^{Reference Pfeifer, Bartsch, Hemmer and Reisch25} However, none of these factors remained significant in the multivariate analysis.

Univariate comparisons also showed that patients with a history of suicide attempts were more likely to have a history of sedative or cannabis use disorders and a history of MDD. They also exhibited higher levels of hopelessness, suicidal ideation, depression, anxiety and symptoms consistent with ADHD, both currently and during childhood. These findings align with prior research documenting higher rates of psychiatric and substance use comorbidities among AUD patients with prior suicide attempts. ^{Reference Preuss, Schuckit, Smith, Danko, Buckman and Bierut7} More specifically, current depressive symptoms, hopelessness and suicidal ideation, substance use disorder and a history of MDD were found to be associated with a history of suicide attempts. ^{Reference Pandey, Bolstad, Lien, Walby, Myhre and Bramness23,Reference Surrence, Miranda, Marroquín and Chan26–Reference Guvendeger Doksat, Zahmacioglu, Ciftci Demirci, Kocaman and Erdogan28} Multivariate analyses further confirmed that a history of MDD, suicidal ideation and hopelessness was independently associated with past suicide attempts.

The final significant factor in our multivariate model was female gender, a finding that is consistent with previous studies that have identified a higher prevalence of lifetime suicide attempts and suicidal ideation among females with AUD. ^{Reference Barr, Neale, Chatzinakos, Schulman, Mullins and Zhang9,Reference Hsu, Chang, Chang, Chiu, Lin and Lane29} This gender difference has also been observed in general adolescent and young adult populations. ^{Reference Miranda-Mendizabal, Castellví, Parés-Badell, Alayo, Almenara and Alonso30} However, some studies in AUD populations have found no significant gender differences in suicide attempt prevalence or incidence. ^{Reference Preuss, Schuckit, Smith, Danko, Bucholz and Hesselbrock22,Reference Pandey, Bolstad, Lien, Walby, Myhre and Bramness23}

Prospective risk of suicidal act

Over the 24-month follow-up, 11.5% of patients engaged in a suicidal act. This incidence is considerably higher than the 4.5% reported in a major prospective study of suicide attempts in patients with AUD. ^{Reference Preuss, Schuckit, Smith, Danko, Bucholz and Hesselbrock22} This discrepancy may reflect differences in AUD severity, as our sample comprised in-patients, indicating more severe or complicated cases.

Univariate survival analyses identified 11 factors significantly associated with the occurrence of a suicidal act during follow-up; however, following Holm–Bonferroni correction, only two of these remained significant. First, patients who engaged in a suicidal act were more likely to have a history of severe alcohol withdrawal symptoms (delirium tremens or seizures). This factor is an indicator of AUD severity, and severe AUD is well documented as being associated with suicide attempts. ^{Reference Preuss, Schuckit, Smith, Danko, Bucholz and Hesselbrock22,Reference Flensborg-Madsen, Knop, Mortensen, Becker, Sher and Grønbæk31} Second, a history of sedative use disorder was linked to an increased risk of suicidal acts, concurring with previous findings that substance use disorders predict suicidal behaviour in patients with AUD. ^{Reference Preuss, Schuckit, Smith, Danko, Bucholz and Hesselbrock22,Reference Kõlves, Draper, Snowdon and De Leo32}

In the multivariate survival analysis, three factors retained their significance; prior suicide attempts was one of these, consistent with data from the general population ^{Reference Jha, Chan and Orji33} and among patients with AUD. ^{Reference Preuss, Schuckit, Smith, Danko, Bucholz and Hesselbrock22,Reference Kõlves, Draper, Snowdon and De Leo32} Interestingly, it was not the predominant factor in the present analysis. Instead, a history of severe alcohol withdrawal symptoms (delirium tremens or seizures), already highlighted in univariate analyses, showed the strongest association with subsequent suicidal acts. Interestingly, this variable also emerged as significant in the alternative multivariate analysis including only variables with statistically significant corrected P-values (Supplementary Table 2). While research on alcohol withdrawal seizures is limited, two recent studies have reported an association between a history of suicide attempts and delirium tremens in patients with AUD. The first found that a history of suicide attempts was more prevalent among patients with a history of delirium tremens. ^{Reference Bramness, Pandey, Moe, Toft, Lien and Bolstad34} The second, conducted by the same research team, reported that the association was significant only in males, which may be due to the limited number of females with a history of delirium tremens. ^{Reference Pandey, Bolstad, Lien, Walby, Myhre and Bramness23} The present study lends further support to the notion that delirium tremens influences suicidal behaviour in patients with AUD. Importantly, the evidence comes, this time, from a prospective approach, demonstrating that a history of delirium tremens or alcohol withdrawal seizures is associated with an elevated risk of future suicidal acts. AUD severity is more pronounced in patients with a history of delirium tremens. ^{Reference Bramness, Pandey, Moe, Toft, Lien and Bolstad34} Concordantly, our additional exploratory analysis revealed a positive association between a history of severe alcohol-withdrawal symptoms and markers of AUD severity, i.e. the maximal lifetime quantity of alcohol consumed in 24 h and the number of AUD symptoms. Thus, the impact of this factor may be mediated by a higher severity of AUD. Because the history of severe alcohol withdrawal symptoms is a readily obtainable piece of information, it has the potential to serve as an effective indicator of patients at risk of suicidal behaviour.

Finally, the last factor that retained its significance in the multivariate analysis was living alone, which was identified as a protective factor. This finding is surprising, because being single and loneliness are commonly cited as risk factors: being separated or divorced was identified as a significant risk factor in univariate analyses from a previous study. ^{Reference Preuss, Schuckit, Smith, Danko, Bucholz and Hesselbrock22} However, there is also evidence that partner relationship disruptions are strongly associated with suicidal behaviour in people with AUD. ^{Reference Sher35} Likewise, recent serious arguments with a spouse or family were found to be involved in suicides of patients with AUD. ^{Reference Kõlves, Draper, Snowdon and De Leo32} Therefore, our finding may result from the fact that individuals living alone are less exposed to such arguments.

Our study used a 2-year follow-up period, balancing informativeness with feasibility. Indeed, this duration allowed for the detection of both proximal and distal suicidal acts. Extending follow-up further to capture more distal suicidal acts would increase the risk of higher attrition. Conversely, shorter follow-up periods are easier to implement. We conducted additional survival analyses for suicidal acts occurring within the first 3 months of follow-up, to identify factors associated with immediate suicidal risk. These analyses confirmed the significance of a history of severe alcohol withdrawal symptoms and previous suicide attempts as risk factors.

Limitations

Our study has several limitations. First, despite dedicated catch-up efforts, 15.7% of patients were lost to follow-up. This is common in AUD populations, particularly among those with severe illness requiring in-patient care. ^{Reference Homman, Smart, O’Neill and MacCabe36,Reference Nadkarni, Bhat, Ebrahim and Patel37}

Second, the study design limited the collection of prospective data to the occurrence of suicidal acts, thereby losing the ability to capture the impact of events occurring during the follow-up, such as relapse into drinking, changes in medication or substance use, or the onset of MDD. While this approach probably minimised attrition, it constrained the range of risk factors identified because these were based solely on baseline measurements.

Third, data on the lethality of suicidal acts are currently unavailable. This gap will be addressed in a subsequent study with an extended follow-up period, which will enable a more detailed assessment of the lethality of suicidal behaviour and overall mortality within this population, many of whom remain outside established care pathways.

Fourth, although it could be argued that histories of delirium tremens and alcohol withdrawal seizures warrant separate analysis, we combined these two variables given that they represent the most severe manifestations of alcohol withdrawal syndrome and frequently co-occur. ^{Reference Canver, Newman and Gomez38,Reference Kádár, Gajdics, Pribék, Andó and Lázár39}

Fifth, our sample is relatively old. While we acknowledge that societal and clinical contexts may have evolved since data collection, the core features of AUD and associations, such as the one we observed between suicidal acts and severe alcohol withdrawal symptoms, are likely to remain stable over time.

Sixth, suicidal ideation was measured using a single item from MADRS rather than a comprehensive instrument, which may limit the granularity of data on this risk factor.

Seventh, we did not employ the Severity of Alcohol Dependence Questionnaire, which could have facilitated comparison of AUD severity with other clinical populations.

Finally, as our sample was drawn exclusively from in-patient settings, representing individuals with severe AUD, the findings may not generalise to community populations or to those with less severe forms of AUD. Future research should seek to replicate these findings in broader and less severe clinical samples.

Taking into consideration these different limitations, the finding that the strongest predictor of suicidal acts within the 2 years following admission for alcohol withdrawal management was a history of severe alcohol withdrawal symptoms might be particularly relevant for clinicians. Indeed, the association between suicidal acts and a history of severe alcohol withdrawal symptoms is apparent even within the first 3 months. Given that this clinical information is readily obtainable, it offers a valuable indicator for identifying patients at elevated risk. These results underscore the need for vigilant monitoring and integrated care strategies addressing both addiction and psychiatric risk factors in this vulnerable population.