Talk to the Veterans Crisis Line now
U.S. flag
An official website of the United States government

Health Services Research & Development

Veterans Crisis Line Badge
Go to the ORD website
Go to the QUERI website
HSRD Conference Logo



2019 HSR&D/QUERI National Conference Abstract

Printable View

4092 — Measuring Pain Care Quality (PCQ) in the Veterans Health Administration (VHA)

Lead/Presenter: Stephen Luther,  James A Haley Veterans Hospital, Tampa FL
All Authors: Luther SL (James A. Haley Veterans Hospital, Tampa FL,), Finch DK (James A Haley Veterans Hospital, Tampa FL), Han L (Pain, Research, Informatics, Medical comorbidities, and Education (PRIME) Center, West Haven) Brandt C (Pain, Research, Informatics, Medical comorbidities, and Education (PRIME) Center, West Haven) Goulet J (Pain, Research, Informatics, Medical comorbidities, and Education (PRIME) Center, West Haven) Bouayad L (James A Haley Veterans Hospital, Tampa FL) Fodeh S (Pain, Research, Informatics, Medical comorbidities, and Education (PRIME) Center, West Haven) Hahm B (James A Haley Veterans Hospital, Tampa FL) Lee A (Pain, Research, Informatics, Medical comorbidities, and Education (PRIME) Center, West Haven) Kerns RD (Pain, Research, Informatics, Medical comorbidities, and Education (PRIME) Center, West Haven Haven)

Objectives:
To identify and quantify empirically-derived, key indicators of PCQ in Veterans with musculoskeletal pain.

Methods:
Natural language processing (NLP) algorithms were developed to extract PCQ indicators from text notes of primary care providers in the VHA electronic health record. Approximately 2,500 documents were independently reviewed by two clinical annotators and adjudicated by a third expert to develop a reference set from which reliable (F-measure, 0.77- 0.99) rule-based NLP extraction algorithms were developed. The algorithms identify whether primary care providers documented quality indicators of: 1) Assessment of Pain (the presence, site, etiology, persistence, sensation, and pain intensity, whether physical diagnostic tests were performed, ordered or interpreted, and whether pain impacted function); 2) Reassessment of Pain; and 3) Treatments for Pain (education, medications, complimentary and integrative health (CIH), other interventions). The NLP algorithm was then applied to all outpatient primary care provider progress notes for Veterans newly diagnosed with musculoskeletal disorders and pain intensity ratings > = 4/10 in FY 2013.

Results:
Documents from 130 VHA facilities, 64,940 Veterans and 126,395 unique visits were analyzed. The cohort had a mean age of 53.0 + 15.6 years, was primarily male (89.6%) and white (64.8%) with half (50.3%) being married. The most commonly documented Assessment indicators were presence (97%), etiology (94%) and site (90%) of pain, while least commonly documented were persistence (41.8%) and sensation (31.2%) and pain impact on function (16.7%). The most commonly documented Treatments were medications (95.3%) followed by education (59.9%), and recommendation or use of assistive devices such as braces or canes (23.8%). The most commonly documented CIH were chiropractic procedure/care (4.84%) and massage (1.30%). Evidence of the Reassessment of pain was found in 79.4% of the notes.

Implications:
Results document the feasibility of using NLP to identify key indicators of PCQ with very good to excellent reliability. Estimates of the frequency of documentation of PCQ indicators have face validity and encourage further evaluation of the validity and utility of the measure.

Impacts:
Availability of a reliable measure of PCQ could provide a foundation for measurement-based quality improvement initiatives to improve outcomes and reduce exposure to treatment-related harms such as those associated with opioid therapy for Veterans with chronic pain.