Electronic Patient Record data as proxy of GPs' thoughts.

Area of Science:

• Health Informatics
• Primary Care Research
• Quality Improvement Science

Background:

• Electronic Patient Records (EPR) in primary care offer potential for quality of care studies.
• Assessing the extent to which EPR data reflect general practitioners' (GPs) thoughts is crucial for quality improvement cycles.

Purpose of the Study:

• To evaluate the fidelity of data extracted from primary care EPR systems as a proxy for GPs' clinical reasoning and knowledge.
• To identify the quality and completeness of specific data points (diagnoses, drugs, parameters, referrals) extracted from EPRs.

Main Methods:

• Utilized consolidated data from three software systems across 26 practices, involving 1,554 hypertensive patients and 1,977 contacts.
• Employed an electronic questionnaire as the "gold standard" for GPs' thoughts post-contact, measuring missing and incoherent values.
• Assessed extracted EPR data quality using sensitivity, positive predictive values (PPV), correctness, and global completeness.

Main Results:

• The "gold standard" showed low missing (1.88%) and incoherent (3.92%) values.
• High PPV (>95%) was observed for extracted drugs and parameters (excluding smoking status) in most practices.
• Variable PPV for diagnoses (42.1%-94.9%) and low sensitivity (<67%) for diagnoses and drugs were noted. Parameter completeness for height and weight was below 76%, and referrals were poorly recorded.

Conclusions:

• Current EPR data in Belgium, without calibration, are insufficient proxies for GPs' thoughts and knowledge.
• Implementing tools like electronic questionnaires is necessary to calibrate EPR data for reliable quality assessment.
• Priorities for improvement include enhancing extraction procedures, software interfaces, quality control, and GP training to boost data accuracy and completeness.