Then again, it is a problem of dates and datetimes.
SAS dates are stored in number of days since January 1st, 1960. So substracting one date from another gives durations in days without needing functions.
SAS datetimes (and most date columns imported into SAS Guide from Oracle, DB2, Excel) are stored in number of seconds since midnight on January 1st, 1960. So substracting one datetime from another will give delays expressed in seconds (that should be the result that you currently get).
The DATEPART function is transform a datetime into a date, by removing the time part and computing the right number of days from 01/01/1960.
So your delay computations should look like :
DATEPART(column1) - DATEPART(column2)
The last question is to guess if a given column is date or datetime : check its format. If the format is something like DATETIME., DATEAMPM., then it is datetime. If the format is DATE., DDMMYY., MMYY., YEAR., then it is a SAS date.