Research Data Browser October 2015 Release Notes

CHANGES TO THE RDB DASHBOARD

We have made the following changes to the Research Data Browser dashboard with the October 2015 release:

  • RDB will now support queries and reporting on ICD10, including a full hierarchy of categories, within less than 1 month of ICD10 being used in patient care
  • The pre-load feature will make a huge improvement in ease and speed of use for our investigators.  In DEV, the first screen came up in just 1-2 seconds!
  • In the "All Selections" tab, Users can now query by the following additional data elements:
    • Death Date
    • Birth Date
    • Patient ID (de-identified - for trouble-shooting)
    • Encounter ID (de-identified)
    • Billing charge HCPCS code
    • ICD10 codes, descriptors and categories (as above)
  • A set of other changes won't be noticed by QlikView users, but will make the de-identified data much cleaner and more comprehensible to the users of our RDB extracts. These people, including many investigators in Atul Butte's group and several Data Scientists collaborating with the Chief of ED, Chris Barton, will be using machine learning to extract new clinical knowledge from our APeX data.

CHANGES TO THE RDB EXPORT AND FLOWSHEET FLAT FILES

We have made the following changes to the de-identified flat files with the October 2015 release:

  • ICD10 has been added to the DIAGNOSES table.  Where available, each diagnosis has an ICD10 code and ICD10 descriptive name associated with that code.  In additional, six levels of categories (ICD10 level 1 through 6) are provided for each code.
  • Death Date has been added to "Patients.txt".  Note that this has been date-shifted as have been all other dates.

  • The APeX "Medication ID" has been added to "MEDICATION ORDERS.txt".

  • The APeX Lab result component ID has been added to "LAB.txt"

  • The Billing Procedure HCPCS code has been added to "BILLING.txt"

  • All Age groups (e.g., 11-20 years old) have been removed from all tables; Age is still present, in addition to date of birth.

  • All (de-identified) dates and years of events have been retained, but derived fields including "day of the week", "day" (of the month), "week" (of the year) and month (of the year) have been removed to make the files smaller.

  • The "IsCurrent" field in PATIENTS.txt has been removed.  All patient demographics are current values as of the time of the latest load.

  • The "%PatientKey" field in PATIENTS.txt has been removed.  "Patient ID" is unique to identify a patient and link to other tables.

  • Visit Type is now correctly populated in each table where it is defined.  This is entered only in outpatient appointment scheduling, to reflect the expected duration of the outpatient visit, so it will be null in many records.

  • There is now one and only one row in "PATIENTS.TXT" for each real patient with a unique "Patient ID".

  • The patient race has been moved to its own file, "PATIENT RACE.txt", because there can be more than one race defined per patient.

  • There are no longer any NULL values for "PATIENT ID" in "BILLING.txt" or "LAB.txt".  The records that used to have NULL values for patient id have been removed because these records were from the pre-APeX era.

  • In the "DEPARTMENTS.txt", the field "%EncounterKey" has been removed, and the field "Encounter ID" has been added.

  • The data dictionary spreadsheet, "File and Field dictionary 2015-10-21.xlsx" has been revised and contains all the corrections described above.