Can you easily and efficiently copy or edit the INSERTED table in a trigger? - sql

Question

I'm writing a trigger in which I need to check the incoming data and potentially change it. Then later in the trigger I need to use that new data for further processing. A highly simplified version looks something like this:
ALTER TRIGGER [db].[trig_update]
ON [db].[table]
AFTER UPDATE
AS
BEGIN
DECLARE #thisprofileID int
IF (Inserted.profileID IS NULL)
BEGIN
SELECT #thisprofileID=profileID
FROM db.otherTable
WHERE userid = #thisuserID;
UPDATE db.table
SET profileID = #thisprofileID
WHERE userid = #thisuserID;
-- XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
-- XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
END
IF ({conditional})
BEGIN
UPDATE db.thirdTable
SET [profileID] = Inserted.profileID
...{20+ other fields}
FROM Inserted ...{a few joins}
WHERE {various criteria}
END
END
The problem that we're running into is that the update statement fails because Inserted.profileID is null, and thirdTable.profileID is set to not allow nulls. table.profileID will never stay null; if it is created as null then this trigger should catch it and set it to a value. But even though we're updated 'table', Inserted still has the null value. So far it makes sense to me why this is happening.
I'm unsure how to correct the problem. In the area with commented Xs I tried running an update query against the Inserted table to update profileID, but this resulted in an error because the pseudo-table apparently can't be updated. Am I incorrect in this presumption? That would be an easy solution.
The next most logical solution to me would be to INSERT INTO a table variable to make a copy of Inserted and then use that in the rest of the trigger, but that fails because the table variable is not defined. Defining that table variable would require more fields than I care to count, and will present a major maintenance nightmare any time that we need to make changes to the structure of 'table'. So assuming this is the best approach, is there an easy way to copy the data and structure of Inserted into a table variable without explicitly defining the structure?
I don't think that a temp table (which I could otherwise easily insert into) would be a good solution because my limited understanding is that they are far slower than a table variable that lives only inside the trigger. I assume that temp table also must be public, and cause problems if our trigger fires twice and both instances need the temp table.