Missing data on the web interface and extraction xls files

Dear all,
We setup an integrated eCRF using last version of OC. We recently realised that some data entered on the interface turn to disappear from the interface as well as from the extraction files, while present in the database (checked with pgadmin).
Indeed, CRFs marked as complete after double checked data entry are partial. We could not figure out any strict data type relation between present vs. missing information.
Just few words about our design: it is a single visit with 28 CRFs altogether collecting > 5000 items per patient.
Does anyone has an idea on what may be the issue for the interface not showing all the data?
All the best
Encarnita

Comments

  • GerbenRienkGerbenRienk Posts: 792 ✭✭✭
    Hi Encarnita,
    Do the items also disappear when extracting in odm-xml format?
    If they are available in odm-xml, but not in for example tab/xls, can you give some examples?
    Many thanks,
    Gerben Rienk
  • kristiakkristiak Posts: 1,313 ✭✭✭
    Hi Encarnita,

    How do you extract the data?. Have you tried to divide the data in smaller extracts. such as by CRF or any other logical division!

    Best

    Krister
  • encarnitaencarnita Posts: 13
    edited February 26
    Hi,
    @kristiak, I created 4 datasets of similar item numbers. We can of course try to subdivide even more, but although the extraction part is a concern we could try to deal with the odm_xml files, which are apparently complete, we also have web display issues, which avoid monitoring!!!.

    @GerbenRienk You are totally right. As I said, we have two issues:
    - 1: items displayed on the web interface and on the xml, not in xls
    - 2: items not displayed on the web nor on the xls, but yes on the xml for some patients only. This is very critical since we cannot do the monitoring as data are note displayed....

    I tried to put examples of the 2 issues in the attached document. Do not hesitate to ask further information if needed.

    Thanks for any help, suggestion, hint!
    Encarnita
  • encarnitaencarnita Posts: 13
    edited February 26
    sorry... this is an empty post cause I can't delete it... :-(
  • kristiakkristiak Posts: 1,313 ✭✭✭
    I would try to work with the XML files a and use the excellent MS ACCESS tools that Gerben has created that you will find on his Web site Trail Data Solutions.

    Best Regards

    Krister
  • GerbenRienkGerbenRienk Posts: 792 ✭✭✭
    Hi Encarnita,
    Thank you for detailed comments. This reminds a bit of https://jira.openclinica.com/browse/OC-409 and OC-422 and OC-2476 but I think you're working with a clean install of oc3.14, correct? Not one that started before oc3.1?
    Just to be 100% sure: can you confirm using pgAdmin that table item_data has a constraint called pk_item_data_new, consisting of item_id, event_crf_id, ordinal?
    And could you maybe download the studymetadata-file of your study? Then we can have a look if there's anything unusual there.
    Many thanks,
    Gerben Rienk
  • encarnitaencarnita Posts: 13
    Dear Gerben Rienk,

    I could not get access to the tickets, but if it wasabout the version, we are using OC 3.13.
    As for the item_data table, yes we have a constraint called pk_item_data_new, consisting of item_id, event_crf_id, ordinal.
    Can I pm you the studymetatda file? Which email can I use?

    All the best
    Encarnita
  • encarnitaencarnita Posts: 13
    @kristiak Thanks for the suggestion.

    I used datalisting tool from Trial Data Solutions with a small xml file from the CRFs having issues with the extraction of some items on excel (example 1 from the file dowloaded on feb 26th). But it failed when trying to get the metadata with the following error message:
    "Exexution error '5': Argument or call procedure invalid.

    When getting onto the debugging I got the following highligthed lines in Virtual Basics:
    In ModeRomovedHTML file:
    intPositionSecondOpeningTag = InStr(intPositionFirstOpeningTag + 1, strContent, "<")

    I have no idea of what this means...
Sign In or Register to comment.