Identifying Multisite Chronic Pain with Electronic Health Records Data

Background. Multisite chronic pain (MSCP) is associated with increased chronic pain impact, but methods for identifying MSCP for epidemiological research have not been evaluated. Objective. We assessed the validity of identifying MSCP using electronic health care data compared with survey questionnaires. Methods. Stratified random samples of adults served by Kaiser Permanente Northwest and Washington (N = 2,059) were drawn for a survey, oversampling persons with frequent use of health care for pain. MSCP and single-site chronic pain were identified by two methods, with electronic health care data and with self-report of common chronic pain conditions by survey questionnaire. Analyses were weighted to adjust for stratified sampling. Results. MSCP was somewhat less common when ascertained by electronic health records (14.7% weighted prevalence) than by survey questionnaire (25.9% weighted prevalence). Agreement of the two MSCP classifications was low (kappa agreement statistic of 0.21). Ascertainment of MSCP with electronic health records was 30.9% sensitive, 91.0% specific, and had a positive predictive value of 54.5% relative to MSCP identified by self-report as the standard. After adjusting for age and gender, patients with MSCP identified by either electronic health records or self-report showed higher levels of pain-related disability, pain severity, depressive symptoms, and long-term opioid use than persons with single-site chronic pain identified by the same method. Conclusions. Identification of MSCP with electronic health care data was insufficiently accurate to be used as a surrogate or screener for MSCP identified by self-report, but both methods identified persons with heightened chronic pain impact.