How to compare same column values rows?

You're attempting to order the data in your query. You need to define something to order it by. From the example above, can you order by the first column? It looks to be ordered in the fashion you need. The query would look something like this (but with your column and table names, since I don't know what they are, I made something up).

SELECT a.Description
FROM dbo.TableA AS a
ORDER BY a.ID ASC

This query will retrieve the description column from the table, but it will order it by the first column (ID). It will do this in an ascending manner (ASC), counting up.

Here's an alternative approach.

If you've got a unique value to track a particular patient through this system (eg a PatientID), then you might consider this approach:

-- set up your data in a table variable for me to show the concept
-- note the additional "PatID" field.
declare @VitalData table (PatID integer, LineID integer, StrangeCode varchar(20), Description varchar(50), EntryDate datetime)
insert into @VitalData values
(1, 316, '0005B7DCB817', 'VitalsAckByEmr', '2012-03-20 12:07:16.547'), 
(1, 376, '0005B7DCBB3C', 'PatientSelected', '2012-03-20 12:28:26.723'), 
(1, 388, '0005B7DCB9A2', 'VitalsAcquired', '2012-03-20 12:29:51.263')

-- Use a PIVOT table to build a resultset that can be worked against
SELECT PatID, VitalsAckByEmr, PatientSelected, VitalsAcquired
FROM
(
    SELECT PatID, Description, EntryDate
    FROM @VitalData
) AS src
PIVOT
(
    MAX(EntryDate) 
    FOR Description IN ([VitalsAckByEmr], [PatientSelected], [VitalsAcquired])
) as pvt

That brings back the following results:

PatID  VitalsAckByEmr           PatientSelected          VitalsAcquired
1      2012-03-20 12:07:16.547  2012-03-20 12:28:26.723  2012-03-20 12:29:51.263

If you want to specifically select those PatientID records that don't match the above process, I've added the following records to the @VitalData table:

(99, 990, 'foo', 'VitalsAckByEmr', '2012-06-01 01:01:01'),
(99, 991, 'bar', 'PatientSelected', '2012-06-01 01:02:03'),
(99, 992, 'baz', 'VitalsAcquired', '2012-06-01 00:00')

and you can add a WHERE clause to the outer SELECT, for example:

where (PatientSelected < VitalsAckByEmr) OR (PatientSelected > VitalsAcquired)

and this time, we get:

PatID  VitalsAckByEmr           PatientSelected          VitalsAcquired
99     2012-06-01 01:01:01.000  2012-06-01 01:02:03.000  2012-06-01 00:00:00.000

Just an idea, mind...

This might work but would need testing to make sure it scales past 2 records...

IF OBJECT_ID('Patients') IS NOT NULL 
DROP TABLE Patients
GO

CREATE TABLE Patients
(
PatID INTEGER ,
LineID INTEGER ,
StrangeCode VARCHAR(20) ,
Description VARCHAR(50) ,
EntryDate DATETIME
)

-- set up your data in a table variable for me to show the concept
-- note the additional "PatID" field. * - borrowed from Thomas' answer ;)
INSERT INTO Patients
VALUES -- Good Row
( 1, 316, '0005B7DCB817', 'VitalsAckByEmr', '2012-03-20 12:07:16.547' ),
( 1, 376, '0005B7DCBB3C', 'PatientSelected', '2012-03-20 12:28:26.723' ),
( 1, 388, '0005B7DCB9A2', 'VitalsAcquired', '2012-03-20 12:29:51.263' ),
-- Bad Row
( 2, 396, '0005B7DCB817', 'VitalsAckByEmr', '2012-03-20 12:07:16.547' ),
( 2, 476, '0005B7DCBB3C', 'VitalsAcquired', '2012-03-20 12:28:26.723' ),
( 2, 588, '0005B7DCB9A2', 'PatientSelected', '2012-03-20 12:29:51.263' )


;WITH datas
  AS ( SELECT lineid ,
[p].[Description] ,
ROW_NUMBER() OVER ( PARTITION BY p.[PatID] ORDER BY [p].[EntryDate] ) AS r_n
   FROM [dbo].[Patients] AS p
)
SELECT [p].[PatID] ,
[d].[Description] ,
[d].[r_n]
FROM [datas] AS d
INNER JOIN [dbo].[Patients] AS p ON [d].lineid = p.[LineID]
WHERE ( [d].[Description] != 'VitalsAckByEmr'
  AND [d].r_n = 1
)
OR ( [d].[Description] != 'PatientSelected'
AND [d].r_n = 2
   )
OR ( [d].[Description] != 'VitalsAcquired'
AND [d].r_n = 3
   )