Primary Health Care System
RESEARCH AND KNOWLEDGE TRANSFER

SEED FUNDING REPORTS

EMR Data Quality Management Methods: A Feasibility Study
Michelle Greiver; Jan Barnsley; Babak Aliarzadeh; Paul Krueger; Rahim Moineddin ; Debra Butt; Neil Drummond; Lisa Jaakkimainen; Karim Keshavjee; David White; David Kaplan

Abstract :

Objective:  To evaluate the feasibility of an aspect of data quality (data reliability) improvement in four specific areas of the electronic medical record (EMR) chart: diagnostic coding for a chronic disease [chronic obstructive pulmonary disease (COPD)]; structured data entry for a risk factor (smoking); structured specialist referral designation (meta-data) and interprofessional encounter designation.

Design:  Feasibility project incorporating an evaluation of data quality with a quasi-experimental before and after evaluation of a data quality management intervention.

Participants:  Thirteen community-based family physicians that are members of an interdisciplinary Family Health Team (the North York FHT).

Intervention:  Once baseline measures were recorded, a data clerk was then tasked with correcting the data as follows: a) changing all unstructured smoking data into structured categories;  b) with physician permission, adding the ICD9 COPD code “496” in the cumulative patient profile (CPP) instead of free text; c) changing referral designations in the address book of the EMR to conform with College of Physicians and Surgeons of Ontario specialist designation ; d) adding interprofessional encounter headers to the EMR if not present and inform Allied Health Professionals (AHPs).  The participating physicians and AHPs  were provided with “data manuals” suggesting improved methods of data entry, along with screen shots. 

Main outcome measure:  EMR data quality was measured as outlined above at baseline and six months.  Evaluation of acceptability by physicians will be measured through a questionnaire incorporating usefulness and usability at 6 months (in progress).

Results:  Unstructured smoking data decreased from 29% to 2% at 6 months after data entry.  Coded COPD entries in the CPP increased from 56% to 96%.  Referrals with specialist designations increased from 50% to 71%.  Identifiable interprofessional headers increased from 28% to 37%.  A statistical analysis is underway.  The data entry clerks spent 53 hours restructuring the smoking data, 3  hours on recoding COPD, 70 hours on adding specialist designations to the address book.  Interprofessional headers were added in less than an hour.

Conclusion:  This intervention (using a data clerk to restructure data in the four areas of the EMR we indentified) led to measureable improvements in the aspects of data reliability we measured, with a reasonable use of resources such as time and personnel.  The results could be used to plan a phase 2 study.

Key Messages:

• Data in the EMR for smoking, COPD, referral designations and interprofessional encounters are frequently missing or entered using free text
• Unstructured free text data can be variable across practices and over time; this can present challenges in terms of using data for quality improvement, research and epidemiology
• A Data Manager can program queries to discover unstructured data in the EMR
• Trained data entry clerks can re-enter data in a structured form.  We have outlined the time spent by the clerk on each data element.
• Several aspects of data quality in the areas of the EMR we studied appears to be quantifiable
• Our methods were associated with an increase in the elements of data quality we measured in all four areas
• There may be deterioration of data structure over time.  Efforts may be needed to maintain data quality over time.

For further information, please contact:
Dr. Michelle Greiver
mgreiver@rogers.com

Back to Seed Funding Projects