Agreement between electronic medical record-based and self-administered pain numeric rating scale: clinical and research implications

Abstract

Background: Pain screening may improve the quality of care by identifying patients in need of further assessment and management. Many health care systems use the numeric rating scale (NRS) for pain screening, and record the score in the patients' electronic medical record (EMR).

Objective: Determine the level of agreement between EMR and patient survey NRS, and whether discrepancies vary by demographic and clinical characteristics.

Methods: We linked survey data from a sample of veterans receiving care in 8 Veterans Affairs medical facilities, to EMR data including an NRS collected on the day of the survey to compare responses to the NRS question from these 2 sources. We assessed correlation, agreement on clinical cut-points (eg, severe), and, using the survey as the gold standard, whether patient characteristics were associated with a discrepancy on moderate-severe pain.

Results: A total of 1643 participants had a survey and EMR NRS score on the same day. The correlation was 0.56 (95% confidence interval, 0.52-0.59), but the mean EMR score was significantly lower than the survey score (1.72 vs. 2.79; P<0.0001). Agreement was moderate (κ=0.35). Characteristics associated with an increased odds of a discrepancy included: diabetes [adjusted odds ratio (AOR)=1.48], posttraumatic stress disorder (AOR=1.59), major depressive disorder (AOR=1.81), other race versus white (AOR=2.29), and facility in which care was received.

Conclusions: The underestimation of pain using EMR data, especially clinically actionable levels of pain, has important clinical and research implications. Improving the quality of pain care may require better screening.

Figure 1

Number of VACS participants reporting pain scores, by source of NRS