This paper investigates the longitudinal association between ADHD with or without ODD at age 16 and substance use disorder at age 33 in a large population cohort. Generally, I find the writing style pleasant. While the paper is not unique (also see some refs below that might be worthwhile for your paper), I do think it is a contribution to the field. I have some comments (none of them major) below that I would like to see clarified.

1. Why did you choose SWAN scores and not diagnoses? It seems like the data is available? Similarly so, I am assuming that you have data on medication use in this group. Since this is one of the major questions concerning substance use in ADHD I think it is wise to include this in your paper.
2. I think your CI indicate a relatively reliable estimate, which is a strong point

Introduction
3. It would be great if you, in your introduction, highlight the discussion about ODD and whether ODD is uniquely associated to SUD or not in your introduction
4. I have trouble understanding why your study is unique and what exactly the scientific relevance is. You argue this: therefore there is a gap in knowledge concerning persons with elevated ADHD symptoms but who may not have been picked up by clinical registers. But perhaps you want to add here something about ADHD being a continuum of symptoms and something on the elevated right with individual symptoms?
5. Our meta analyses (Groenman et al. 2017) did not find a moderating effect of percentage of ODD/CD on the association between ADHD and SUD, however, be careful not to interpret absence of evidence as evidence for absence of an effect. In your introduction you are quite careful with this statement but in the discussion, it does feel like you interpret this slightly different. Perhaps if you are interested in showing evidence for the null hypothesis you could perform Bayesian analyses?
6. In the introduction I was a bit confused about the numbers of the cohort. Is it ~9000 individuals or ~6000 like I read in the abstract? This is also not the clearest in the start of your method section
Method
7. I see that you already have data on medication reimbursement, is it possible to use this data for your analyses? Those prescribed medication, and those who did not received medication?
8. You use the 95% percentile on the SWAN as a cutoff for ADHD. When I saw the percentage of those scoring above this percentile, I thought this was a reasonable approach. However, I was wondering if your choice to classify everyone <95% as control influenced your results? I could also image doing a sensitivity analyses where you use for example <80% as a cutoff.
9. I wondered why you chose for model 2 and not to stratify your results according to sex, parental psychiatric disorder, and urbanicity. I think it would be neater to conduct separate analyses for males and females, with and without parental psychopathology etc. Covariate analyses cannot "regress" this variance out, so it would be more informative to do stratified analyses, so the effect that these variables have on your outcome become clear. Especially since for example female sex appears to be associated with a reduction in SUDs in your sample (is this in the whole sample?) while in ADHD there have been reports that show an increased risk in women with ADHD.
10. The reasoning for correcting for “Frequent alcohol intoxications past year” in model 2 is unclear to me. It seems very logical that those with SUDs are more frequently intoxicated? Can you add the reasoning for this somewhere?
11. Can you make explicit which data were censored?
Results
12. Does parental psychiatric disorder also include SUDs in parents?
13. I noticed that your supplementary Figures appear to not have a heading. (which Figure is Figure 2 etc?)
14. Was the difference between ODD+ and ODD- significant?
15. In your Figure 3, it appears that your ADHD-ODD+ group was very small (big steps in the plot), was this correct? If so I would mention something about this in your paper, since it appears to be a limitation. This made me realize that I cannot find anything on these 4 groups anywhere in your paper. Perhaps make a descriptive Table with these 4 groups?
Discussion
16. We used the same sample in Ilbegi et al. 2018 describes (but a later FU) as in Groenman 2013. Both are clinical samples.
17. I would add a bit more discussion on medication treatment.
18. Did you really mean to highlight that you used an all white homogeneous ethnic sample as a strength of your study? I see this as a very clear limitation.
19. I was wondering if you had any reasoning to assume that the estimate of 3.1% SUDS is an underestimate. Can you substantiate this claim?
20. Is impairment itself a predictor of SUDs?
Since you did not look at medication treatment, I suggest not making any conclusions on medication treatment as your final statement, since your data does not show this (i.e., Our findings carry a public health message that early identification of these disorders is important to mitigate the trajectories of SUD.)

Some very minor points, but there are some inconsistencies in your writing. For example sometimes you use a space between 95% CI, sometimes you do not. Sometimes you capitalize Figure/Table, sometimes you do not. I recommend checking your manuscript for these things.
https://pubmed.ncbi.nlm.nih.gov/32237241/
https://pubmed.ncbi.nlm.nih.gov/31182215/

DFear authors,

Thank you for your careful consideration of my previous comments. I have read through your response letter, and a few minor points remain:

in your discussion you conclude: Thus, our findings suggest that all individuals with ADHD are at risk ... etc. I think you should omit the "all" here, since you are doing group based analyses.

c10, I see that you clarify this, but I still think this is a substantial bias that you are imposing onto your data. Similiarly so for C12.

c15: you mention that you added a new table (great!!) but I think the new table is Table 3, not 2.
c17: I cannot seem to locate your discussion on medication use

c18: I suggest reconsidering mentioning a homogenic ethnic sample as a strength. I understand that this might be a strength in genetic research, but this is not that. I think considering diversity a confound is very much not where the field is now.

Dear Authors,

Thank you very much for your careful consideration of my previous comments. While I indeed have a different view on comment 2 (bias caused my excluding frequent alcohol use, because I think you are excluding some individuals with alcohol use disorders with that, and thus underestimating how big the problem of SUDs are in ADHD with ODD). While I think it would have been better to at least perform sensitivity analyses where you include those individuals with frequent alcohol intoxication, I do think that you are being very transparant that you made to choice of excluding these individuals from your manuscript, and the readers can make their own decisions on the value of your results.
Further, I am happy with how you addressed my other comments.