This is a VERY bad database design indeed.
For ex: what if you decide to remove '22' from all rows that contain '22'?
Instead, create a new table that holds one row for every one of these values, and link them to the 'current' records through their ID.
That way you can add new values by just inserting a new record in the second table, and putting in the ID of the record that the value belongs to, and ofcourse the actual value itself
original table:
id, whateverelse
1, ...
2, ...
3, ...
new table:
originalid, value
1,12
1,3
1,4
1,22
2,7
3,8
2,66
1,543
now you can add '35' to record '1' by inserting a record in the second table:
INSERT INTO secondtable (originalid,value) VALUES (1,35);
and you can get all values for record 1:
SELECT value FROM secondtable WHERE originalid=1;