0.3% is the percentage of participants, not the confidence in the diagnosis.
Every result reported in the paper follows that same pattern: "[N] participants had [result] ([% of participants], [95% confidence interval])".
Every result reported in the paper follows that same pattern: "[N] participants had [result] ([% of participants], [95% confidence interval])".
And note that your own link said that the .3% of participants were those that could possibly show signs of myocarditis, they didn't actually contract that disease. So what does your paper show that is damning evidence to you?