I've just finished searching a batch of 74 infant deaths (age=0) from the old GRO index (i.e. FreeBMD data) using the new GRO data. These cover the timespan 1868-1954.
Of the 74 a total of 22 still have an age of 0, the remainder have a number in the range 1 - 11.
Although this is a fairly small sample, I wonder whether the absence of any numbers more than 11 means the number represents months, and anything under one month is being shown as a '0'?
The majority of the 74 records are pre-1900, but the post-1900 ones show a similar pattern to the earlier ones.