This paper aims: (1) to describe the prevalence of psychiatric disorders (assessed during hospitalisations) among men registered in an outpatient alcohol rehabilitation center ; (2) to compare these prevalences (AUD included) to the one observed with matched controls; (3) to assess the temporality between AUD and the other psychiatric disorders.

I thank the authors for their valuable work: this paper is of interest, and could provide interesting findings (see below for comments)
Study strengths include the use of two large cohort studies and the assessment of different psychiatric disorders.
Here are some comments that aim to improve the clarity and the quality of the manuscript

Main issues:
- Although it may already be described in previous papers based on the same cohorts, it is not clear to me how patients diagnoses were made: were the made by psychiatrists, by physicians,… ? were the psychiatric disorders systematically assessed altogether (i.e., using a clinical interview) or was it based on the main or secondary psychiatric disorder for which the patients sought treatment for (this latter option would mean an underestimation of the prevalence of psychiatric disorders) ? This should be better specified because it could have important implications for the results and the conclusions
- An important strength is the use of large cohorts, but an important limitation is the use of different classifications, not only because diagnostic criteria for a given disorder changed between classifications, but also because new diagnostic categories appeared in the ICD-10, and finally because disorders changed in their definition (i.e., “AUD” ICD-8 – in fact “alcoholism” – included episodic excessive drinking, habitual excessive drinking, and “alcohol addiction” that was characterized by the compulsion to drink and by withdrawal symptoms when drinking was stopped; while in the ICD-10, the AUD category refered to a more reliable approach based on harmful use and dependence). ICD-10 AUD diagnosis was based on dependence and less on craving or alcohol-related harms, and this may thus underestimate the prevalence of AUD (when compared to current DSM-5 or future ICD-11). It is thus debatable that the “alcohol use disorder” phenotype assessed may be the same in the ICD-8 and ICD-10. This comment also applies to the ICD-8 “neuroses” and ICD-10 “anxiety disorders” categories that do not assess the same phenotypes. Use of different diagnostic categories with different names and different criteria renders it difficult to use a generic term for a given diagnostic category (as proposed in Table 1 for example) and more difficult to interpret the findings
- If the authors wished to stay with the current database, it would be helpful if the authors could provide information about the differences between the ICD-8 and the ICD-10 (in terms of evolution in diagnostic categories and diagnostic criteria for a given category) and better justify why it is relevant to merge different diagnostic categories with different diagnostic criteria. This would provide the reader a better understanding of what was assessed and how it was assessed. Another option would be to limit themselves to a specific classification, and this would provide much more specific and reliable results (but it would need to rerun statistical analyses based on a more specific sample).
- (in line with previous comments: in addition the “F10” mentioned in Table 1 means that patients with alcohol intoxication but no alcohol abuse or dependence were included in the study but it does not seem to be the case on ICD-8 diagnoses (was alcohol intoxication F10.0 included ?).
- Statistical analyses: although we can understand the need to first undergo explorative univariate analyses without correction for multiple tests, use of multivariable analyses with adjustment on the variables associated with dependent variables (in univariate analyses) could improve the quality of the manuscript by taking into account important bias (i.e. but not limited to age, maybe also intelligence, …). The sample size allows them to do such adjustment (e.g., multiple logistic regressions).
- Although the third objective seems interesting (to assess temporality between AUD and other psychiatric disorders), I am afraid that the current study plan does not allow a proper assessment of such temporality for the following reasons: use of different classifications; “only” psychiatric disorders seems to be based on disorders during hospitalization and thus exclude psychiatric disorders diagnosed during outpatient treatment either before or after the hospitalization; no “psychiatric hospital diagnosis” does not mean that there was no psychiatric diagnosis at all ; change in diagnostic criteria and diagnostic categories.

Minor points:
- The terms “Psychiatric disorders” and “psychiatric diagnoses” are used in the manuscript, but the ICD and DSM refers to “psychiatric disorders”
- The term “psychiatric hospital diagnosis” is not clear to me; in fact, it seems as if the authors assessed “psychiatric disorders” in a specific context (when the patient was hospitalized); this could probably be better specified for a better understanding of what is really assessed
- Current literature refers to “dual disorders” or “co-occurring disorders” or “psychiatric comorbidities” when assessing the link between AUD and other psychiatric disorders, but it seems as if these terms were not in the manuscript.
- The statement “psychiatric hospital disorders are unlikely to be the primary factor for development of AUD in this population” in th abstract and throughout the manuscript seems overstated for the following reasons: the current study design and methodology do not allow to infer causation, “only half” is a huge percentage ! (I am not sure we could state for example that “tobacco is unlikely to be the primary factor for development of lung cancer because only half of these patients smoked”), there might be interindividual differences between some patients with “primary AUD” and “secondary AUD”, the current classifications rely much more on the existence of a comorbidity (AUD is associated with other psychiatric disorders” than on the determination of a specific temporal link which is difficult to demonstrate, even when AUD seemed to happen before the other psychiatric disorders). See for example Castillo-Carniglia et al. Lancet Psychiatry 2019 for possible explanations for these associations. This comment also apply for the whole manuscript (e.g., but not limited to, “significant outcomes” third bullet point, introduction ,methods, results and discussion).
- Limitations: use of different classifications with different definitions of alcoholism / AUD is an important limitation; results are based on psychiatric disorders diagnosed once the patient was hospitalized and not during outpatient treatment care.
- P.5: statement “we have only identified two studies assessing psychiatric disorders among individuals receiving outpatients AUD treatment” seems overstated: I believe there is much more studies in this field than proposed; see for example Castillo-Carniglia et al. Lancet 2019 https://www.thelancet.com/journals/lanpsy/article/PIIS2215-0366(19)30222-6/fulltext ; Mendes Oliveira et al. 2018
- https://pubmed.ncbi.nlm.nih.gov/30236640/
-
- Methods:
o study type should be presented more clearly at the beginning of the methods.
o Matching variables: please indicate p-value for the difference in date of birth/age
o Lifespan in AUD is lower than in the general population, so matching on lifespan may include patients with lower life expectancy than the general population ?
o Is there indicators of reliability for the BPP test for intelligence ? Is there indicators of mean / SD of this test in the general population to assess how this population differed from the general population ?
o Psychiatric diagnoses: as indicated before, please specificy how psychiatric disorders were assessed
o Statistical analyses: use of multivariable analyses in addition to current analyses would provide much more reliable results with less bias.
o One limitation is also the lack of assessment of ADHD, which is highly comorbid with AUD; but it was not assessed in the ICD adults classifications so it is understandable ; it could explain a lower rate than expected and could be added as a limitation in the discussion.
Results:
- P. 11: although OR is lower than the other disorders, I do not agree with “the OR for psychotic disorders were relatively low” (95%CI 1.98 – 2.65) is not a small effect size.
Discussion :
- Please be careful in the interpretation of any temporal causality (see before) and tone down the statement of the low prevalence of psychiatric disorders in patients with AUD (e.g. but not limited to p. 12 “less than half of the men had been registered… this number is remarkably low); this could be explained by the important treatment gap in AUD (around 10% AUD patients seek for adequate treatment), which is much lower than most other psychiatric disorders. See Rehm et al. works for example or Tuithof et al. 2016 https://www.karger.com/Article/Abstract/446822
- P. 12: admission to a psychiatric department : does it include emergency treatment care ? outpatient treatment are ?
- P. 15 : please revise “suggesting that other psychiatric diagnosis are unlikely to be the primary factor for AUD” (see before)
- P. 15: results may also be explained by the important treatment gap in patients with AUD and psychiatric disorders.
- P. 15 : please revise the sentence “another explanation … the co-occurrence of AUD and other psychiatric disorders reflect mental characteristic such as neuroticism and low intelligence” (interindividual differences: does not apply to all people; link between variables and psychiatric disorders may be bidirectional; as it is currently stated, may be perceived as stigmatizing for persons with psychiatric disorders / AUD)
- “SUD and BSPDF may reflect personality disorders” ; current classifications are much more based on comorbidity than interpretation of causality between the considered disorders; please revise
- P. 16 : please revise “we included virtually the full spectrum of treated lifetime psychiatric hospital diagnoses” (debatable considering the DSM-5 or ICD-11 that include more diagnoses)
- Last sentence: seems overstated : “only 51.8%” is important and does not allow the determination of causation mechanisms between AUD and psychiatric disorders.

I thank the authors for their detailed reply and for their careful revision of the manuscript.
The authors answered to most of my comments; here are however four points for which I believe the authors answered to my comments but only partly, and for which I suggest some modifications to improve the manuscript’s quality:
- The authors acknowledge the important differences between the two different classification systems used (ICD-8 and ICD-10) and it is stated in the point by point response to the comments. However, this point does not appear in the limitation section nor in the abstract. It is thus not clear for the reader that the authors are referring to two different classifications to diagnose psychiatric disorders, and I believe it is an important point to highlight. I would therefore suggest the authors to clearly specify in the abstract that the psychiatric disorders diagnoses were based on two different classification systems (ICD-8 and ICD-10). It would also be helpful to include this point in the Limitation Section at the beginning of the manuscript as well including it and discussing it in the limitation subsection of the Discussion.
- Ex-Point 25: I proposed to revise the sentence “Another possible explanation is that there exist common causes including mental characteristics such as the personality trait neuroticism (46, 47) and low intelligence (9) that increase the risk of both AUD and other psychiatric disorders” because I thought that linking low intelligence and AUD and psychiatric disorders may be perceived as stigmatizing. In addition, the reference mentioned by the authors is based on men only, and on the risk of hospital admission (for psychiatric disorders) rather that on the prevalence for psychiatric disorders. In addition, the literature on this subject is more complex that proposed: the same author (Gale et al. ) demonstrated that there was a ‘reversed-J’ shaped association: men with the lowest intelligence had the greatest risk of being admitted with pure bipolar disorder, but risk was also elevated among men with the highest intelligence (https://www.nature.com/articles/mp201226) ; it is also not demonstrated for all psychiatric disorders (ADHD, bipolar disorder, …). Such a formulation may lead the reader to think that because low intelligence may be a risk factor for psychiatric disorders, high intelligence may be protective and this is in fact the opposite (see for example Karpinski et al. 2018: High intelligence : a risk factor for psychological and physiological overexcitabilities ;https://www.sciencedirect.com/science/article/pii/S0160289616303324). High intelligence is also a risk factor for ADHD https://www.sciencedirect.com/science/article/pii/S0149763415303213). There is a strong association between high intelligence, ADHD and bipolar disorder, as well as between these disorders and AUD. In line with all of these points, I would therefore suggest the authors to revise the fact that low intelligence is ”a common cause … that increases the risk of both AUD and other psychiatric disorders” (both low intelligence and high intelligence are risk factors for some but not all psychiatric disorders).
- Ex-Point 27: the authors did not fully answer the comment: it was proposed to revise the sentence “we included virtually the full spectrum of treated lifetime psychiatric hospital diagnoses” because the authors did not included the full spectrum of treated lifetime psychiatric hospital diagnoses (true if you refer to ICD-8 or ICD-10 diagnoses assessed here, but not if you refer to the current DSM-5 and future ICD-11 classifications, that include more/different diagnoses); this should be specified
- Reviewer 2, ex point 2: I would suggest to add in the text the p-value for the comparison between the two populations (“Among men registered in an outpatient alcohol clinic, the mean lifespan of those with a psychiatric hospital diagnosis was 58.46 years, while the mean lifespan of those without a psychiatric hospital diagnosis was 58.94 years. This difference is significant (p=0.034), but quite small relative to a standard deviation of 10.39.”)

Thank you very much for your valuable work