Online Program

Using large datasets to identify and evaluate patients with chronic pain in a primary care setting

Wednesday, November 6, 2013 : 12:50 p.m. - 1:10 p.m.

Terrence Tian, MPH, Weitzman Quality Institute, Community Health Center, Inc., Middletown, CT
Daren Anderson, MD, Weitzman Institute, Community Health Center, Inc., Middletown, CT
Ianita Zlateva, MPH, Weitzman Institute, Community Health Center, Inc., Middletown, CT
Sissi Wang, Weitzman Quality Institute, Community Health Center, Inc., Middletown, CT
Background. Chronic pain is an extremely prevalent and costly condition that is often overlooked despite the fact that it affects an estimated 116 million Americans, with an annual cost of $635 billion in medical treatment and lost productivity. The increasing use of electronic health records (EHRs) provides an opportunity to learn more about chronic pain and to use large datasets to drive improvement. However, an ideal method does not exist to identify and track patients with chronic pain using electronic data and individual measures such as pain scores or ICD9 codes are not reliable or comprehensive. Purpose. Create and validate a practical method to accurately identify patients with chronic pain. Method. Using electronic data of a multisite community health center, we analyzed visit codes, patient pain scores, and medication records. We then derived a method to most accurately identify chronic pain using combinations of data from the EHR after determining the accuracy of each individual element in identifying chronic pain. This method was validated by review of 381 random patient charts for chronic pain to determine its accuracy. Using the developed method, chronic pain patient characteristics were populated for three years, 2010-2012. Results. The established algorithm consisted of pain scores, opioid prescription data, and ICD9 codes. The sensitivity and specificity of this algorithm was 84.8% and 97.7%, respectively. It was more accurate (95.0%) than pain scores (87.9%) or ICD9 codes (93.2%) alone. The Receiver Operating Characteristic was 0.981. Patient characteristics and service patterns remained consistent from 2010 through 2012. Conclusion. Using an iterative process of chart reviews and combinations of data elements captured from an EHR, we derived an algorithm able to identify chronic pain with a high degree of sensitivity and specificity. With this, we were able to better study the characteristics of patient suffering from chronic painful condition.

Learning Areas:

Administration, management, leadership
Chronic disease management and prevention

Learning Objectives:
Identify patients with chronic pain accurately using pain scores, opioid prescription data and ICD-9 codes. Assess the characteristics of people with chronic pain, including demographics, health care utilization, and referral patterns. Design quality improvement initiatives to improve the care of patients with pain.

Keyword(s): Primary Care, Chronic Illness

Presenting author's disclosure statement:

Qualified on the content I am responsible for because: I serve as the Research Director for the Weitzman Quality Institute at CHC, Inc. In this role I am responsible for managing a wide range of research activities and supporting the overall growth in primary care research that is relevant to the CHC’s mission to provide quality healthcare services to all. I have diverse educational background and an extensive range of research experiences, including planning analyses, data extraction/collection, data manipulation and analysis.
Any relevant financial relationships? No

I agree to comply with the American Public Health Association Conflict of Interest and Commercial Support Guidelines, and to disclose to the participants any off-label or experimental uses of a commercial product or service discussed in my presentation.