We thank Chick et al, and Witt et al, for their welcome responses to our article. ^{Reference Chitty, Dobbins, Dawson, Isbister and Buckley1} We agree that both improving access and facilitating referral to alcohol services are essential strategies with regard to reducing deliberate self-poisoning that may be a product of harmful use of alcohol.

We share the concerns of Chick and colleagues – it is dangerous to make causal assertions from cross-sectional data, especially if preliminary analyses and author interpretations are introduced into clinical practice from the abstract alone. We agree that people prescribed tricyclic antidepressants and typical antipsychotics are different from those on other drugs – that they are less likely to co-ingest alcohol during intentional self-poisoning is one such example. As highlighted by Chick et al, the underlying nature of this relationship (whether it is causal or correlated because of shared factors) has many possibilities, for which we presented three interpretative and non-mutually exclusive speculations. We agree with the further interpretation put forth in their letter – individuals with increased access to higher-toxicity medications may negate any perceived role of alcohol in the poisoning. Of course, this is only relevant in cases when alcohol is used as a tool to facilitate the self-harm (i.e. to ‘numb fears’) as opposed to the person being intoxicated before the desire to self-harm arises. It is noteworthy that a recent study found that over 70% of people interviewed after a suicide attempt that involved acute alcohol use reported they did not use alcohol to facilitate the action. ^{Reference Bagge, Conner, Reed, Dawkins and Murray2} However, we recognise that the methods of suicide attempts in this aforementioned small sample size study were heterogeneous and that self-poisoning is more likely to involve alcohol as a substance perceived to increase the toxicity of the poison or mask the taste of the co-ingested substances. We are currently conducting a study to investigate patient self-reported reasons for use of alcohol before and during deliberate self-poisoning, which will further shed light on this.

We are pleased that our analysis prompted Witt and colleagues to investigate a similar line of enquiry within their own cohort. The similarities between the data analysis conducted by Witt et al and our findings are notable – those prescribed antipsychotics, anticonvulsants and stimulants were less likely to co-ingest alcohol during a non-fatal self-poisoning.

Compared with the Japanese study cited by Witt et al, in which nearly half of individuals are not transported to hospital after suicide attempts or episodes of self-harm, our experience specifically for deliberate self-poisoning (via toxicology services and Poison Information Centres) tells us this is not the case in Australia. Indeed, the Australian study conducted by the authors themselves shows that over 90% of the time individuals who have taken overdoses with unknown intent attended by paramedics are transported to hospital. ^{Reference Lloyd, Gao, Heilbronn and Lubman3} Thus, the rationale for adjusting for medical severity is not clear to us in this context; however, we do agree that engaging a breadth of services for a holistic response to self-harm is essential, especially drug and alcohol services when alcohol may be a driving factor.