Question

I am working on a cancer database and I have one column with the date of a patient’s local recurrence (if they had one), and another column with the date of a patient’s distant recurrence (if they had one). I want to create another column that consists of the date of FIRST recurrence, regardless of whether it was local or distant. I’m not sure how to proceed because some patients only had local or only had distant, and thus many fields are “NULL”. Here’s an example of what I’m looking to do:

Date_Local     Date_Distant     Date_Any
2010-08-01     2009-05-25       2009-05-25
NULL           2001-01-07       2001-01-07
1999-12-12     NULL             1999-12-12
NULL           NULL             NULL

Answer 1

Assuming your already added the column with an ALTER TABLE statement, you would populate it using a CASE statement in the UPDATE query like:

ALTER TABLE cancertable ADD Date_Any DATE AFTER Date_Distant;

UPDATE cancertable SET Date_Any = 
  CASE
    WHEN Date_Local IS NOT NULL AND Date_Local <= Date_Distant THEN Date_Local
    WHEN Date_Distant IS NOT NULL AND Date_Distant < Date_Local THEN Date_Distant
    ELSE NULL
  END

The above statement will update all rows since it has no WHERE clause. Before doing the UPDATE, test it out with a SELECT to make sure it looks like you expect:

SELECT *,
  CASE
    WHEN Date_Local IS NOT NULL AND Date_Local <= Date_Distant THEN Date_Local
    WHEN Date_Distant IS NOT NULL AND Date_Distant < Date_Local THEN Date_Distant
    ELSE NULL
  END AS Date_Any
FROM cancertable

Consider, however, whether you really even need another column added though. Depending on how often you intend to query this, you might just create a view using the SELECT statement above instead, so the Date_Any value will be “live”.