Data submitted by KU Hospital to the University HealthSystem Consortium (UHC) is used as the source for hospital billing diagnoses and also provides quality, visit, and additional procedure details. Key observations include:

  • Length of Stay
  • Comorbidity, MS-DRG
  • Service Line
  • Agency for Healthcare Research and Quality (AHRQ) measures such as Central Venous Catheter Blood Stream Infections

Integration of UHC data from KUH has been released to the HERON user community:

We presented this work at AMIA CRI in March 2013:

  1. Nair M, Connolly DW, Whittkopp C, Waitman LR. Combining Quality and Accountability Study Metrics from University Health System Consortium in an i2b2 based Clinical Data Repository with EMR data for Patient Centered Outcome Research (PCOR). AMIA Clinical Research Informatics Joint Summit. San Franciso, CA, USA; 2013.

Source code and development notes for UHC ETL

While the code is not (yet) packaged for use independent of the KUMC HeronLoad environment, you are welcome to study it as-is:

not yet published:

Development notes include:

heron-eldorado-update milestone tickets:

query by MSDRG, length of stay, service line from UHC 1-year snapshot

heron-cimarron-update milestone tickets:

Load UHC Core measures observations into HERON
Load UHC physician specialty and roles

heron-waconda-update milestone tickets:

search for diagnosis by source: encounter, problem list, UKP IDX billing, UHC

heron-neosho-update milestone tickets:

get the Low Socioeconomic Status (low SES) flag from UHC into HERON

CDB/RM File Specification

A copy of the specification was provided by KUH staff to the HERON team:

  • CDBRM_Dwnloads_File_Specification (4).xlsx, Pilarz, Thu 4/16/2015 3:42 PM

Requirements gathering and design notes

from Feb 23, 2012 meeting

Discussion of where to start

  • load things in a draft separate ontology first ala the NAACCR TumorRegistry
  • focus on on query first that would involve length of stay, service line and msdrg

Dan: what's the story?

  1. How many patients do I have length of stay and service line is kidney/pancreas transplant? How about "hip fracture"? figure out where "hip fracture is in UHC"
  • of those, how many have a LOS < 3 days? LOS > 7 days?
  • then explore comorbidities and adverse events that may explain variance (long run, in R) but early exploration with i2b2 like:
    • look at Epic/IDX diagnoses and see if there are more LOS > 7 days if the patient has diabetes and/or congestive heart failure. Or weight's impact/BMI. And likely labs related.
    • can also explore mortality ala SSDI and Hospital.
  • do this with data post Nov 2007 of course we just have a year of uhc data. 

Technical notes with Dan:

  • technically, LOS should be done in the visit dimension.  But, we haven't done that yet. 
    • this is unprecedented for our team but let's see if we can make it work.
  • First pass, we don't care if the comorbidity was documented during the visit.

Sketching some SQL:

-- Check that we're in the KUMC sid.
select  recordid from UHC.encounter where 1=0;

-- Service Line facts
select 'UHC|ServiceLine:' || e.uhcserviceline as concept_cd
  , e.patientid
  , e.admissiondate as startdate
  , e.dischargedate as enddate
  , e.encounterid
from UHC.encounter e;

-- Note: the concept_cd should perhaps be 'UHC|SERVICELINE:' for consistency

Last modified 5 years ago Last modified on Apr 22, 2016 9:10:42 AM