Banner Health Data Access

Contact CB2 to help you with the tasks shown in blue in the summary flowchart below:

Legend for flowchart above: 

• BH/gray = Banner Health
• UA = U of A
• IIT/IIS = Investigator-Initiated Study/Trial
• HBs = honest (data) brokers
• DUA = Data Use Agreement
• Sky blue = contact CB2 for help if needed
• Light brown = emails

Check your @arizona.edu email regularly

U of A and Banner only email @arizona.edu addresses throughout this process

• Gray (step 1) = using TriNetX is optional
• U. of Arizona blue (steps 2 and 4) = required by U of A
• Azurite blue (steps 3 and 5-6) = required by Banner

Already have a list of patients?

Do you have patient IDs, such as medical record numbers (MRNs), for which you need EHR data from Banner, or are you already working with someone to obtain a list?

Yes: proceed to step 2

No: it is highly recommended you:

 Define and assess cohort(s) with TriNetX 

Using TriNetX saves time by allowing you to:

1. Assess and confirm the feasibility of your study, i.e. to perform a reality check to see if there are enough Banner patients meeting your study criteria: TriNetX displays numbers of patients meeting the criteria you specify
2. Share your patient cohort(s)' definition(s) with Banner honest brokers so they will know for which patients to extract data, to get you data more quickly 

Contact CB2 to work with you on both of these tasks, and/or if you cannot find patients you need.

Start step 2 once determine feasibility, then finish step 1 in parallel

Once you have determined your study is feasible, step 1 (this step) can be finished, by refining your patient cohort(s), in parallel with step 2 (below) to save more time.

Unless you only need summary statistics such as for a grant application, which you can obtain from TriNetX, regulations usually require approval from an IRB when accessing patient level data: 

Initiate U of A IRB approval

…including drafting and gathering IRB materials to complete the next step (RAP form)

If using AI, before completing the RAP form (step 3):

1. Contact CB2 if you plan to use AI in a manner which might need approval from Banner
2. Specify what, where and how AI will be used, with which Banner data, in your IRB protocol

After you have determined your study is feasible and initiated U of A approval, and if needed, have AI approval from Banner, you must submit your data request to Banner via the U of A Research Administration Portal (RAP): Review RAP instructions. Banner can then confirm if your study is feasible and approve.

Complete appropriate RAP form

Which form and fields to complete

Select the appropriate form for your study as listed below and complete all required fields, including those listed below.

If you are only doing a retrospective study, fill out the Retrospective Chart Review (RCR) Submission / CRDW Request form

1. For the question "Is this study an Investigator Initiated Study/Trial (IIS/IIT)", if applicable, select Yes (else select No), then attach your IIS/IIT approval
2. For the question “Do you need Cerner CRDW data extracted by the Honest Broker team?”, select Yes…
3. ...then detail the following, read EHR FAQs for more information and example(s):
   1. Inclusion criteria:  
      1. If you have a list of patient IDs, note this and attach them at the bottom of the form 
      2. If you used TriNetX, list your TriNetX study name, and the criteria you used
      3. If you were not able to specify (all or any) criteria in TriNetX, list all criteria needed in the EHR to find patients you need, including codes, time ranges, encounter and clinician types, etc.
   2. Exclusion criteria: Most studies need to exclude some patients from their cohorts, to ensure they are actually selecting the correct patients
   3. Data requested: 
      1. if you only need a few data elements, list them
      2. else attach a Data Abstraction Sheet at the bottom of the form, to ensure you receive data in the format you need and to make it faster for the honest brokers to get you data 
   4. Start and End Dates for use of requested data: date range specified in your IRB protocol for which you need EHR data

If you are doing a prospective study, with or without using retrospective data, fill out the New Study Submission/Feasibility Resubmission Form, and complete the above steps as applicable

Once your submit your RAP form, the following is completed which typically takes at least 3 weeks:

1. U of A Health Sciences (UAHS) will confirm you provided the minimum necessary data, and email you questions if needed
2. UAHS will then forward your request to Banner Health
3. Banner will review your application to confirm your study is feasible and meets their requirements, and email you if AI approval is needed and/or if they have questions

Once your request on the RAP is approved by Banner (after the above steps are completed):

Submit to U of A eIRB

Submit your IRB application with your:

• Banner approval from the RAP
• Banner approved IRB protocol

Study team complete COI forms and training

In order for the IRB to approve, all study team members must complete the conflict of interest (COI) forms in their eIRB dashboard after submission, and be up to date with their CITI and any other required regulatory training.

Please note:

Banner cannot start to pull data for you until your final IRB protocol is approved by the U of A IRB.

When the Banner honest brokers receive your email with your approved IRB protocol, they put your request into their queue of data requests and prioritize your study. Once they verify the data you requested matches your protocol approved by the IRB, they extract the EHR data according to their queue.

The amount of time it then takes the Banner honest (data) brokers to extract data for you depends on:

• If you used TriNetX to define your patient cohort(s) or already had a list of patient IDs (less time if have either)
• How clearly written and complete your IRB protocol and data request is
• How many data elements you need (less elements = less time)
• How complex that data is, including if need PHI de-identified (more complex = more time)
• Other studies in their queue

Once the Banner honest brokers pull data for you:

PI must reply to DUA email from Banner

1. Banner will create a ticket for your data request in their system
2. Banner's legal / contracts department and UAHS contracts will begin the process for your study's PI to agree and acknowledge the necessary legal terms in the existing DUA between U of A and Banner
3. The PI for your study will be sent an email with the DUA: only after the PI acknowledges the terms of the DUA by signing and replying to the email, can you receive data

To check the status of the DUA for your study, search under the "Contracts" section of U of A's CRA (Clinical Research Administration) Project Status Report. This part of the process (all of step 6, including the PI signature) typically takes no more than 6 weeks.

Soon after you reply to Banner's email acknowledging the DUA, and thus all of the above steps are complete, then Banner honest data brokers will retrieve the EHR data they pulled from the Banner Clinical Research Data Warehouse (CRDW) and deliver that data to the PI of your study via Microsoft OneDrive.

You can just use TriNetX to find numbers of patients who meet your criteria, gather other summary statistics and even perform statistical analyses.

If you cannot find the criteria you need or the statistics are not what you expect, Contact CB2 for help. 

Read instructions for QI and other Banner collaborations

You will also need to submit a CRDW/RCR RAP form

Check your @arizona.edu email regularly 
U of A and Banner only email @arizona.edu addresses for this process

The time it takes varies widely, and can take months, depending on the reasons listed under the ‘…then Banner honest data brokers extract data listed’ step after step 5 above, and: 

• If you are using AI; in particular, if the AI system is storing EHR data, especially PHI, or being used to make clinical decisions (more time for separate approval if so)
• If requesting PHI, could take longer to approve

Not currently, but there are efforts to make this data available.  In the meantime, Banner Health honest brokers can:

• extract the following so you can then review charts to extract this data:
  • medical record numbers (MRNs)
  • dates of the following by type
    • visits
    • imaging
    • other tests
• use PowerShare to pull images for you to browse (but can’t export images)

Review next FAQ for other discrete data which is available 

The following list is not comprehensive

• Answers to questions from surveys such as:
  • Personal Health Questionnaires (PHQs)
  • Screenings for depression and anxiety
• Obstetric (OB) data such as:
  • Last menstrual period (LMP) and Estimated Due Date  (EDD)
  • gravitas and parity

Contact CB2 if need discrete data not listed here 

Answer from Banner honest broker: 

"Banner only has a deceased date and time for patients that passed away within their facility. On rare occasions, this information is reported to Banner and added to the patient’s chart. However, this is very few and far between. ... [such as,] for patients that are regularly being seen at clinics and such. ... it cannot be relied on to provide accurate deceased information for those that may have passed elsewhere."

Medications administered or prescribed, including:

• Names, Brand or generic
• Route, i.e. form such as oral, IV, etc.
• Strength, i.e doses
• Frequency: how often prescribed to take (number of times daily, weekly, etc.)
• When prescribed/ordered; and if inpatient, when administered*

*meds administered, such as in a clinic or hospital, also have date and time of administration, in addition to order date.  Medications patients take themselves on an outpatient basis will only have order date.

Types of visits, include, but are not limited to:

• ambulatory / outpatient
• inpatient, inc. "non-acute"
• emergency department
• observation
• short stay
• laboratory
• virtual
• home health

... by facility / clinic

Available history includes:

• patient medical history, inc. self-reported history
• social history such as smoking status
• family medical history

Banner’s clinical research data warehouse (CRDW) is a subset of data from Banner’s Cerner EHR DB (database): Currently, only structured data (not text documents, yet) is available for researchers from the CRDW.

Banner’s CRDW does not have data from Epic EHR (2014-17)

There is encounter billing data, from the following administrative DBs, in Cerner:

• NextGen: ambulatory encounters, such as “wellness” exams
• MS4 (Siemens MedSeries4):  registration and scheduling inc. referrals