This is an interesting and informative manuscript on a topic that has been neglected thus far. The simple calculation for NAFLD is of value, as it enables others to attempt to replicate/extend these findings. The sample size is large, the methods generally well described, and the article is well written and well organized. The major limitations are acknowledged as such. I have just a few comments:
1. Given the large sample, the entry criterion for variables in the multivariable model, p<0.2 for univariate analyses, is liberal. What was the rationale for this threshold? “based on their clinical relevance” was another criterion for entry, however these variables were not specified in the methods. How was this determination made? There are far more variables that met the criterion than are reported in the multivariable model. Please clarify how variables were excluded/removed following entry.
2. How was age of BD onset defined?
3. Other variables that warrant evaluation are history of childhood abuse, psychosis, and family history of BD.
4. Regarding the medication findings, a comment should be added regarding the possibility of confounding by indication.
5. The clinical recommendations (“mandatory”) are overly prescriptive, as this is a single cross-sectional study. The manuscript should briefly address the value of screening, ie what is the treatment for NAFLD? Relatedly, consider integrating the recent ISBD Vascular Task Force report in relation to the future clinical and research implications of this study.