Same question, different answer: A comparison of global health assessments using vas scales

Background: Visual analogue scales (VAS) are simple and easy to use by patients, and are easily analysed. However, VAS valuations can differ according to their presentation. We compared patient responses to 2 VAS scales from 2 different instruments, each asking “how good or bad is your health today, in your opinion”, ranging from 100 “Best imaginable health state” to 0 “Worst imaginable health state.” Methods: Patients were participants in a randomised controlled trial of aggressive vs. symptomatic control of inflammation in patients with established (>5 years duration) stable rheumatoid arthritis followed for 3 years. Patients completed the EuroQol (EQ-VAS), patient global assessment VAS (PG-VAS), HAQ and DAS28 at baseline and a 3 year follow-up assessment. The EQ-VAS is a 20cm vertical scale, with 100 at the top, with markers at each increment of 10. The PG-VAS is a horizontal 10 cm scale from left (0) to right (100) with no incremental markers. Cross-sectional correlations were assessed using Spearman's rank, and longitudinal correlations assessed using Pearson correlation. Differences between VAS scales were calculated and compared across the range of values using a locally weighted regression line. Results: PG-VAS and EQ-VAS assessments were completed by 449 patients with a mean (s.d.) age of 61 (11), disease duration of 13 (7) years. 68% were female. Median (IQR) baseline HAQ and DAS28 scores were 1.5 (0.9, 2.0) and 4.0 (3.1, 4.9) respectively; self-reported health scores were EQ-VAS 65 (50, 80) and PG-VAS 66 (51, 77). Despite similar median scores, the correlation of VAS scores was only moderate (0.58). PG-VAS scores were evenly distributed; EQ-VAS scores clustered at scale increments of 10. There was a systematic trend in the difference between scores. EQ-VAS scores<50 tended to be lower than the PG-VAS, whilst EQ-VAS>50 were higher. 3-year change was calculated for 375 patients; the mean change for the PG-VAS and EQ-VAS was -4.9 (s.d. 22.1) and -3.8 (s.d. 20.8) respectively. The correlation of change scores was 0.51. Within patient differences between VAS scores exceeded 10, the minimum important difference for these global health VAS scales, in 176 (39%) patients at baseline and in the change scores of 210 (46%) patients. Rounding PG-VAS and EQ-VAS scores to the nearest increment of 10 to account for clustering around markers did not improve cross-sectional (0.57) or longitudinal (0.49) correlations. Conclusions: Even when asking the same question, scores on global health VAS scales appear sensitive to presentation and/or the context in which they are asked. Differences between scores were not explained by the presence of markers on the scale. Differences may be due to the horizontal/vertical presentation, length, or questions preceding the scale. The EQ-VAS followed general quality of life questions and the PG-VAS followed pain and fatigue scales; preceding questions may prime patients to respond differently to global health VAS scales.