Probably to get any good advice you would have to explain how it can be that "the outcome variable is continuous" and that "the data nature is count" and what reason you have to say it's "appropriate to employ a Poisson regression".

And explain, if you initially wanted to estimate it with a linear model, why now you do not (since if you are worried about the residuals' distribution other things can be done). And sample size, research question, etc. In all, probably worth deleting this question and starting over.

Thank you for your feedback, dear scholars.

To get a well-targeted answer, I have modified the question.

The outcome variable is the number of food served in a day. I have read that there are times that we can use linear regression. However, the occasion is rare as the count data distribution is usually not normally distributed even after transforming the outcome. There are a plethora of articles that used linear regression for similar data. But, after considering some reports, the best method would be Poisson regression, which I believe is right. Therefore, I want to apply Poisson regression. Nevertheless, there are some decimals. Thus, this is to get your advice on whether rounding decimals to fit a Poisson model is possible.

Thank you!