PrevPrev Go to previous topic
NextNext Go to next topic
Last Post 15 Jan 2013 12:00 AM by  SuperUser Account
Error Checks that I am not sure are right
 5 Replies
Sort:
You are not authorized to post a reply.
Author Messages


Basic Member


Posts:289
Basic Member


--
08 Jan 2013 12:00 AM


    Hello all

    I am currently spending my time making sure that the Quality Assurance Checks in the HIV DDM tool reflect the HICDEP 1.60 error-codes.

    In this more detailed process of going through the tables I noticed some odd things.

    for tblLAB_VIRO

    Error Code LVW001 for tblLAB_VIRO is in fact NOT a "Within-Table" check as it is defined as "Patient has no record in BAS" - I would suggest renaming this check "LVC001" and moving it to the "Cross-Table" section of the qa checks.


    I am not sure if Error Code LVW008 and LVW009 are suitable for as a standard HICDEP check. I see no reason that only one positive and one negative HIV-test result should be allowed. In my understanding this table should contain all the VIRO/SERO results. It seems arbitrary to insist on only two HIV test -- one positive and one negative.
    I suggest that these are marked as study specific - or reformulated to search for AT LEAST 1 with negative result and AT LEAST one with positive result. This would allow you to know that indeed the patient has HIV - and by the negative results you should be able to tell from the latest date approximately when seroconversion occured.

    for tblLAB

    Error codes LW006 and LW007 reference fields that are not required fields in the tblLAB (LAB_LLN and LAB_ULN) - at the very least I am not sure that the optional fields should have the column HICDEP=YES, if they should even be included in the list.

    tblOVERLAP

    the Error Checks OW001 (invalid other cohort) and OC002 (patient not found in BAS for other cohort) are not feasible to code in the HIVDDM tool, as you would not neccesarily have access to the patientlists submitted by other cohorts in the study upon making your own data submissions.


    Basic Member


    Posts:289
    Basic Member


    --
    10 Jan 2013 12:00 AM


    "tblLAB_VIRO

    Error Code LVW001 for tblLAB_VIRO is in fact NOT a "Within-Table" check as it is defined as "Patient has no record in BAS" - I would suggest renaming this check "LVC001" and moving it to the "Cross-Table" section of the qa checks."

    I agree this proposal.


    Basic Member


    Posts:289
    Basic Member


    --
    11 Jan 2013 12:00 AM
    "for tbl LAB_VIRO

    I am not sure if Error Code LVW008 and LVW009 are suitable for as a standard HICDEP check. I see no reason that only one positive and one negative HIV-test result should be allowed. In my understanding this table should contain all the VIRO/SERO results. It seems arbitrary to insist on only two HIV test -- one positive and one negative. I suggest that these are marked as study specific - or reformulated to search for AT LEAST 1 with negative result and AT LEAST one with positive result. This would allow you to know that indeed the patient has HIV - and by the negative results you should be able to tell from the latest date approximately when seroconversion occured."

    I think QA check "LVW008" is correct, you wouldn't expect to find more than one or less than one HIV positive test for a patient.

    However you might find more than one HIV negative test result for a patient or no negative test results at all, therefore "LW009" is probably a study specific test (CASCADE?).


    Basic Member


    Posts:289
    Basic Member


    --
    12 Jan 2013 12:00 AM


    "for tblLAB

    Error codes LW006 and LW007 reference fields that are not required fields in the tblLAB (LAB_LLN and LAB_ULN) - at the very least I am not sure that the optional fields should have the column HICDEP=YES, if they should even be included in the list."

    As the variables LAB_LLN and LAB_ULN are not present in the current HICDEP table LAB, I agree to exclude these QA checks from the list (or to add the variables LAB_LLN and LAB_ULN to the table LAB).


    Basic Member


    Posts:289
    Basic Member


    --
    14 Jan 2013 12:00 AM


    for tblREFILL there is another crosstable check that is wrongly placed

    tblREFILL - WithinTable - PRW001 - Patient doesn't have a record in BAS

    Once again I suggest a renaming to "PRC001" instead of the "PRW001"

    Regards, Rikke


    Basic Member


    Posts:289
    Basic Member


    --
    15 Jan 2013 12:00 AM


    tblBAS - CrossTable - BC001 - "AIDS_Y=1 but no records in tblART "


    Should the check not be "RECART_Y=1 but no records in tblART" ??
    You are not authorized to post a reply.