Anxiety disorders are prevalent in the US, with almost 1 in 5 Americans having an anxiety disorder in the previous 12 months (NIMH 2023). Anxiety disorders are associated with suicide attempts (Nepon et al., 2010; Sareen et al., 2005) and appeared to be more common during and since the COVID-19 pandemic. Overall, the U.S. suicide rate increased by 36 % between 2000 and 2021 (Centers for Disease Control 2023). Most patients who die by suicide see a healthcare provider within 30 days of their death (Ahmedani et al., 2019). Pre-marketing drug randomized trials are underpowered to detect changes in suicide attempt rates but have led to an FDA mandated black box warning about suicide risk for all anxiolytic medications. Non-randomized studies provide an opportunity to improve our understanding of associations between medication exposures and this rare adverse event.

Alprazolam is a short-acting triazolobenzodiazepine anxiolytic used in the treatment of anxiety disorders including panic, generalized anxiety, and social anxiety disorders (van Marwijk et al., 2012). It acts on the gamma-amino-butyric acid (GABA)A receptor in the brain, enhancing GABA affinity and thereby reducing anxiety (Dawson et al., 1984). A Cochrane review reported superiority of alprazolam relative to placebo in the treatment of major depression, with an effect similar to tricyclic antidepressants (van Marwijk et al., 2012). It is the 37th most widely prescribed drug in the United States with an estimated almost 17 million prescriptions in 2020 to over 3.5 million patients (Kane, 2022). It has side effects similar to other benzodiazepines. These include tiredness, impaired concentration, balance, and coordination. Memory may be affected (Medication Guide 2023).

Few studies have examined the association between alprazolam and suicidal behavior. Most studies are under-powered to detect an effect with a low base rate outcome event, and the results are mixed, with several finding an association with suicide risk (Darke et al., 2014; Gardner and Cowdry, 1985; Youssef, 1990; Dodds, 2017) and others not finding this association (Jonas and Arthur, 1996; Kravitz et al., 1993; Park et al., 2015). In 2018, alprazolam-related fatalities in the United Kingdom resulted in its removal from the National Health Service formulary, making it available only via private prescription (Corkery et al., 2022). A Bayesian screening study was completed for risk of suicidal behavior among patients prescribed selective serotonin reuptake inhibitors (SSRIs) and any one of 53 common adjunctive central nervous system therapies in 262,721 patients in Sweden. Of the 52 adjunctive therapies, alprazolam (IRR = 1.39) was one of only two associated with higher suicide risk (Lagerberg et al., 2022). More recently, a nationwide French case-crossover study estimated the risk of suicide attempt (n = 111,550) and suicide (n = 12,312) in patients taking benzodiazepines. Risk was highest in the 30-day period (prior to suicide event) compared with two matched reference periods (−120 to −91 days and −90 to −61 days), in patients with or without recent psychiatric history (Tournier et al., 2023).

Shorter acting benzodiazepines (e.g., alprazolam and lorazepam), place patients at greater risk for withdrawal symptoms, because, compared with longer acting benzodiazepines (e.g., diazepam), they have a more abrupt offset of action, that may lead to distressing rebound anxiety and potentially enhance suicide risk. Among the shorter acting benzodiazepines, alprazolam has been particularly associated with increased risk of seizures (Warner et al., 1990), indicating more severe withdrawal reaction, and therefore potentially more suicide risk than long-acting benzodiazepines.

The iDEAS algorithm (High Dimensional Empirical Bayes Screening) generates drug safety signals for all drugs on the market and a single adverse event (Gibbons et al., 2019). To illustrate its use, we applied it to 922 drugs on the U.S. market, to generate drug safety signals for suicide attempt (Gibbons et al., 2019). Ten drugs associated with higher suicide attempt risk and 44 drugs associated with lower suicide attempt risk were identified. Alprazolam, butalbital, hydrocodone, and codeine/promethazine generated the strongest signals for higher suicide attempt risk, and folic acid, mirtazapine, hydroxyzine, disulfiram, and naltrexone generated the strongest signals for lower suicide attempt risk (Gibbons et al., 2019).

The 72 % higher risk of suicidal events in patients taking alprazolam was not anticipated (Gibbons et al., 2019). It remained statistically significant after excluding overdoses and was comparable for males and females and younger and older patients (Gibbons et al., 2019). A similar, but smaller association with higher risk of suicidal events was also found for diazepam (28 %) (Gibbons et al., 2019).

We proposed (Gibbons et al., 2019) that “signals related to individual drugs identified by iDEAS should be tested using more rigorous pharmacoepidemiologic studies”, which is the focus of this paper. This is the third confirmation study of signals generated by the iDEAS study, each using the same medical claims database and similar methodology. The first confirmed the signal for folic acid and lower suicide attempt risk (Gibbons et al., 2022) and the second confirmed the signal for benztropine and lower suicide attempt risk (Gibbons et al., 2023). This is the first study designed to confirm a drug safety signal for increased suicide attempt risk identified in the original iDEAS study.