Test-retest reliability and measurement error of the numerical rating scale and visual analogue scale in people with low back pain

Abstract

The 0–10 numerical rating scale (NRS) and 0–100 visual analogue scale (VAS) are commonly used to assess pain intensity in low back pain (LBP) trials, but their measurement properties remain unclear. AIMS: We aimed to determine the reliability and measurement error of the NRS and VAS in non-specific LBP. We used a test-retest design with online questionnaire administration. Adults proficient in English with acute (<6 weeks), subacute (6–12 weeks), or chronic (>3 months) non-specific LBP were recruited. Pain intensities were recorded using the NRS and VAS at baseline, as well as 20 min and 24 h after baseline. Intraclass correlation coefficients (ICC), standard error of measurement (SEM), and smallest detectable change (SDC) were estimated for acute and chronic LBP over two recall periods: 24 h and 7 days. A total of 733 began the survey; 298 (100 acute, 198 chronic) and 165 (62 acute, 103 chronic) completed 20-minute and 24-hour follow-ups respectively. Reliability was good to excellent (ICC 0.75 to 0.94) for the NRS and moderate to good (ICC 0.68 to 0.89) for the VAS across both intervals and for both LBP types. The SDC was 1.6 to 2.8 of 10 for the NRS and 20.9 to 36.5 of 100 for the VAS. Our results suggest the reliability of the NRS and VAS is acceptable in people with low back pain, however their measurement errors are higher than generally accepted minimal important change scores, which needs to be considered when interpreting trial results and measuring pain in clinical practice.

Perspective

This study examines the reliability and measurement error of two commonly used pain scales in people with low back pain. The findings suggest that small changes in pain intensity scores should be interpreted cautiously and add to growing evidence on the challenges of measuring fluctuating pain.