I extract records where the ICD-9 codes are between 520 and 529. Very few are the direct cause of death, most of the 20 records are the related records (I forget the terminology used on the death certificate, it’s been a couple of months since I looked at this). I have not been analyzing the data, just extracting it but my casual glance told me that a high percentage of the deaths were in very young infants, almost new born. In other words a large percentage were born with a condition that would cause death and this condition affected the mouth in some way.
The rate has remained approximately the same over the years though it does range from about 12 to 26 each year.
I need to stop procrastinating and finish my data extract. The last 10 years or so they moved to ICD-10 codes and the file layout is shuffled once more. More work for me