Student Portal
Course Director Portal
Student Portal
Course Director Portal

QI Data

Obtaining data in the process of defining the current condition or measuring the effect of a countermeasure is critical to the process of quality improvement – one cannot improve what one cannot measure. Below is a list of data resources available within Penn Medicine and CHOP.

Data Resources and Information at Penn Medicine

You can find these tools in your myPennDataStore Analytics Storefront*

  • SlicerDicer (PennChart/Epic Tools)
  • Reporting Workbench Reports (PennChart/Epic Tools)
  • Radar Dashboards (PennChart/Epic Tools)
  • Microsoft PowerBI dashboards
  • PennDiver Reports and Dashboards
  • Tableau Dashboards

*Must be logged onto the Penn network to access.

Penn Medicine Data Analytic Center (DAC)* – The hub of healthcare data at Penn Medicine

TIPS for DAC Data Requests:

  • The DAC is available to help you create and to fulfill requests for Penn Medicine Data
  • The Penn Medicine IRB may need to review requests for protected health information
  • Focus on 2-3 key elements/metrics
  • Your request will likely involve some element of negotiation
  • For longitudinal requests, remember PennChart applications as well as ICD-9/10/CPT codes change over time
  • Unstructured data available via PennSeek or Linguamatics (natural language processing)

DAC Open Office Hours

  • All questions from all levels of learners welcome
  • Every Tuesday from 1-4PM
  • 3600 Civic Center Blvd., 7th floor
  • Contact Yuliya Borovskiy at yuliya.borovskiy@pennmedicine.upenn.edu two weeks in advance to schedule
  • New submissions can be made here: https://uphsnet.uphs.upenn.edu/DACReportRequest/#/dac*

*Must be logged onto the Penn network to access.

Data Warehouses:

Penn Data Store

  • Sunrise (inpatient pre-March 2017)
  • Theradoc
  • Cerner
  • Clindoc
  • Emtrac

Clarity

  • Epic/PennChart data
    Updated every morning at 2AM from production data

Data Resources and Information at CHOP

Reporting Workbench is an analytic tool provided by Epic which provides a user interface to identify and retrieve information about a small number of patients.

  • There are considerable limits on how many patients can be searched, what information can be searched, and what information can be obtained.

Barriers:

  • Relatively simple pre-defined patient identification criteria
  • Limited pre-defined data that can be retrieved
  • More complicated identification or data retrieval requires substantial specialized knowledge and access

CHOP’s Clinical Data Warehouse includes data from many sources including, but not limited to Kronos, Clinical Data Registries, Lawson, GE Healthcare, Salfesforce, Allscripts, Press Ganey, and Strata Decision Technology.

Barriers:

  • Getting approval to access CDW
  • Requires understanding of Structured Query Language (SQL)
  • The data structure is very complex (days of training is often necessary for a full understanding)
  • Understanding the pitfalls of the data model requires experience

Arcus is a major hospital funded strategic project with dozens of employees.

  • A data bridge between clinical data and research data and can act as an honest broker to connect datasets
  • Optional: Using deidentified data means no IRB protocol

Why use the Arcus Initiative?

  • Sometimes you just need to know if a
    potential project is even feasible before you
    invest the effort to develop a human subjects
    research protocol.
  • You can do exploratory analysis without performing human subjects research.

Barriers:

  • Cohort discovery – almost none
  • Required knowledge of SQL to perform queries to the Arcus Data Repository
  • Labs require some willingness to use some combination of R, SQL, Python
  • Data sharing with Arcus by the investigator is expected

How to get started:

Account Management (you have to sign TOU and have CITI training)
https://arcus.chop.edu/

Cohort Discovery
https://arcus.chop.edu/cohort-discovery

Data Repository Query
https://arcus.chop.edu/data-repository-query

Data Model Browser:

https://adr-mdm.arcus.chop.edu/browser/arcus-public/adr_coded

Arcus Labs:
Contact Arcus support (arcus-support@email.chop.edu) to get started

On @CHOP, search for “data request portal” and you’ll end up at https://at.chop.edu/communities/enterprise/analytics-exchange/Pages/Data-Request-Portal.aspx

  • This will be the first step in an iterative conversation to get the right data for you.
  • There may (for research requests, which go to the CRU) be a cost associated.

QI Data Tips

Refine your question

  • What is your time period?
  • What is your base population?
  • Identify the key elements you need.
  • Identify what each row in your potential spreadsheet will represent (encounter? patient? lab result? ICD code?)
  • Sketch out column names and think about aggregation.
  • Be specific.

‣Data and systems evolve over time.

  • There may be multiple sources that need to be queried for the complete answer
  • Example: ICD9 to ICD10, consolidation of systems to Epic

‣Proxy measures are creative ways to get an answer out of the EHR

  • Perhaps there was previously poor or no documentation in the EHR. Could you use another method to answer your question?
  • Example: Trying to measure nausea post-op, look at the ordering of anti-emetics

Consider the following:

1.Procedure data corresponding to revenue codes

2.Lab results

3.Orders, medication administration, flowsheets, vital signs

4.Admission/discharge/transfer events

5.Meta data – alerts triggered, BPA triggered, SmartTool usage

6.Diagnosis codes in history, problem lists, encounters?

‣Write down a few MRN’s of patients you recently saw that would fit into your cohort.  Are they in the final data set?

‣Allot the time and effort for validation as part of your project plan

  • This will lead to more than one round of iteration with the data analyst

‣What will be unique about each row of the dataset?  the patient, the encounter, the lab result

  • Don’t be afraid to state this in your request.
  • If something can happen multiple times in a visit, like a weight, specify that you want the first weight upon admission or last weight before discharge.
Address
  • Hospital of the University of Pennsylvania
  • Philadelphia, PA 19104
Links
Newsletter
  • Join our newsletter for CHIPS news, research, and events!
Sign Up

Copyright © 2024 The Trustees of the University of Pennsylvania