Depending on the base data volume you may consider one of the following:
Small volume
Each test may fetch all data from a table and check that you find what you expect
Relatively large volume
You can construct a reference database containing all the tables and data that should always be in your database. You can execute queries to check that reference information is there:
-- checks that used table is not missing any reference table information
-- if count > 1 then something is missing
SELECT COUNT(1)
FROM RefDb.TheTable RT
WHERE NOT EXISTS (SELECT 1 FROM ActualDb.TheTable AT
WHERE AT.PkId = RT.PkId
-- join can be extended for all interesting column values
)
This solution is much faster that the first one for larger volume of data, as information is not fetched into the application layer. It also, allows an isolation between reference and used data, thus more security.
Another option is to store the information in a separate schema, but this means that reference information will replicated along with actual data for all environments (test, preprod etc.)