Abstract Details

Title Retrospective Diagnosis of Painful Diabetic Polyneuropathy from Electronic Abstraction of Medical Records

Topic Pain

Presentation(s) P2 - Poster Session 2 (5:30 PM-6:30 PM)

Poster/Presentation
Number 7-064

Objective Electronically identify painful diabetic polyneuropathy (DPN+P) from the electronic health record (EHR) to study the natural history of pain in diabetic polyneuropathy (DPN), and aid in recruitment to clinical trials.

Background Treatment of pain in DPN is an unmet need. Only a minority of patients respond to individual medications recommended by guidelines.

Design/Methods Charts of 400 randomly selected type 2 diabetic patients followed within our institution were manually reviewed by a neurologist for diagnoses of DPN and DPN+P. DPN was inferred if diagnosed by a treating physician, or if notes recorded symptoms and signs of polyneuropathy. DPN+P was concluded if neuropathic pain warranting treatment was also recorded. Accuracy of EHR abstraction (eDPN and eDPN+P) was examined. A separate cohort of 200 randomly selected patients with eDPN was employed to test the accuracy of eDPN+P.

Results Of 400 charts, 151 (37.7%) had DPN and 70 had DPN+P. Based solely upon diagnostic coding, eDPN sensitivity and specificity were 80.8% and 91.2% respectively, and Youden was 0.72. Informed by the chart review process, definitions of eDPN+P using pain indicators, medications, conditions, and demographics were created and compared using predictive modeling and clustering. Of 200 eDPN patients in the test cohort, 97 (48.5%) had DPN+P. The best accuracy (sensitivity 89.7%, specificity 60.2%, Youden 0.50) was achieved by either a notation of leg pain or any use of a neuropathic pain medication. Medications vary in their predictive value for DPN+P. Gabapentin is the most likely to be used to treat DPN+P, a reflection of prevailing practice.

Conclusions Diagnostic codes are accurate for DPN, but identification of DPN+P from electronic chart abstraction is challenging. It remains to be determined if DPN+P is a distinctive cluster within DPN, or just represents a range on a continuum. Nonetheless, it is important to prospectively capture pain in DPN given its effect on quality-of-life.

Authors/Disclosures
Brittany Lapin PRESENTER	No disclosure on file
	No disclosure on file
	No disclosure on file
	No disclosure on file
Kristen Johnson, PhD (Novartis)	No disclosure on file
	No disclosure on file
Nimish Thakore, MD (Cleveland Clinic)	No disclosure on file

��ɫ��

��ɫ��