05 - NIH BD2K Think Tank. Session 2: Multiple providers / EHRs for single participant; multiple other data sources

NIH BD2K Think Tank

Session 2: Multiple Providers/EHRs

for Single Participant; Multiple Other

Data Sources

Jeffery Talbert, PhD

University of Kentucky

Agenda

KDOC Project Example

Background

Strategy

Lessons Learned

Conclusions

The

K

entucky

D

iabetes and

O

besity

C

ollaborative

(KDOC)

Vision:

Develop a healthcare data repository that will be used to improve the health of Kentuckians via QI activities, comparative effectiveness, and research. The KDOC data repository will bring together

up-to-date clinical data from multiple primary care safety-net providers, plus Medicaid claims data, all linked at patient level. Secure web-based will facilitate use while maintaining appropriate levels of privacy and security.

The

K

entucky

D

iabetes and

O

besity

C

ollaborative

(KDOC)

KDOC partnering organizations

•

Kentucky Primary Care Association (KPCA)

• Individual FQHCs

•

Kentucky Medicaid: Medical & KHIE Directors

•

University of Kentucky

• Academic Health Center

• Center for Clinical and Translational Science • Center for Health Services Research

Strategy: acquire data to support

innovative research and education

Acquire Data

• Identify collaborators with potential research data • And, with a need for data services

Provide Services

• Data collection, management, integration, analysis • Design analytics and QI reporting

New Research

• Use data model for research and training

• Cohort identification, registries, HSR, outcomes

Activate new collaborations to improve quality and create opportunities for research

Payers QI Cost-Effectiveness KPCA QI Support IPA Gain-sharing ACA Expansion UK Develop Community Based Translational Research Network

Participating Federally-Qualified

Community Health Centers

Eight Community Health Centers (FQHCs) serving 39 mostly rural Kentucky counties; diabetes prevalence as high as 17% and obesity prevalence as high as 51%

2 to 15 clinic sites / FQHC with 6 to 31 providers / FQHC 19,900 to 143,000 annual patient visits per FQHC

Approx 124,000 patients served by the 8 FQHCs, total Five different EMR brands across the 8 FQHCs; time since EMR implementation: from < 1 year to several years

vendors: eClinicalWorks, Nexgen, Meditab, Greenway, Allscripts

Technical Workflow

Complete regulatory documents

Select data extraction process Load into ETL staging area

ETL process to standardize data models, link to Medicaid

Load into KDOC data

Lessons Learned

Diverse site technical infrastructure

Very vendor dependent (new EMR users)

Limited site IT staff (contractors, part time,

busy with day job)

Limited site database capacity (required

flexible after hours connectivity, multiple

small reports)

3 Processes:

1)Special KDOC data extracts

2)Core database access

Lessons Learned: Labs

None of the systems use any standard

for Labs

We did the standardization ourselves

using RELMA, but still had to manually

verify that codes didn't get left out

Issues with the linking data within the

site

There would be a Medical Record Number

(MRN) specified in the Lab file that didn't

exist in the Patient file

Lessons Learned: MRN,

IDs, Encounters

Some systems tied everything

together with an Encounter Id, some

did not

In some multi-site set-ups, Providers

were given multiple ID’s across sites

Quality of data related to technical

knowledge of staff

Conclusions

Interfacing with EMRs for data transfer into a

shared repository cannot yet be standardized.

EMR vendor characteristics and practice-based

concerns must be addressed one-by-one

Expert technical assistance is required for

practices to share clinical data for QI or research

EMR Vendors should be at the table from the

So how do we get clean

data into the warehouse?

Iterative process

Load data

Generate reports

Review results with clinic director

Drill down in a real time process to

discover data issues