Study Sample Characteristics
A total of 673 patients from 64 centers across seven countries were included in the study. The majority of enrolled patients were female (72.7%) and White/Caucasian (90.4%). The mean (standard deviation [SD]) age of patients at enrollment was 48.8 years (15.47).
Medical records were abstracted for all 673 patients, of whom 614 (91.2%) returned the 7-day UAS questionnaire. Of these 614 patients, 605 (98.5%) had non-missing values for UAS7TD, and 504 (82.1%) had non-missing values for UAS7OD1MAX and UAS7OD2AVG. The concordance analyses were restricted to 501 (81.6%) patients with non-missing UAS7 values for all three scoring methods (i.e. UAS7TD, UAS7OD1MAX, and UAS7OD2AVG).
All Versions of the UAS Yielded Similar UAS7 Score Values
All three versions of the UAS7, i.e. UAS7TD, UAS7OD1MAX, and UAS7OD2AVG, yielded very similar results, with a mean (SD) score value of 17.3 (10.49), 17.7 (8.90), and 16.2 (8.68), respectively (Fig. 1a, b). As expected, the overall mean (SD) weekly itch score values based on the daily maximum of twice-daily ratings [9.9 (5.28)] was higher than that based on the daily average of twice-daily ratings [8.4 (4.97)]. The mean (SD) weekly hive score values based on the daily average of twice-daily ratings was slightly higher [9.0 (6.58)] than that based on exact 24-h hive counts [7.7 (5.04)].
Pairwise Correlation among UAS7TD, UAS7OD1MAX, and UAS7OD2AVG
Of the 501 patients with complete data for the three scores, the correlation coefficients between UAS7TD and UAS7OD1MAX, UAS7TD and UAS7OD2AVG, and UAS7OD1MAX and UAS7OD2AVG were 0.94, 0.95, and 0.99, respectively, indicating very high positive pairwise correlation (Table 4). The pattern of correlation between each UAS7 version and other PRO measures was also consistent. For example, the correlation coefficient of UAS7TD, UAS7OD1MAX and UAS7OD2AVG to CU-Q2oL total score was 0.44, 0.45, and 0.44, respectively, and 0.45, 0.47 and 0.46, respectively, to DLQI total score (Table 4).
Table 4 Correlation coefficients between PRO scores across UAS7 scoring methods and other PRO measures
Concordance and Agreement Across UAS7 Scoring Methods
UAS7TD Versus UAS7OD1MAX
Of the patients with complete data for all three scores (N = 501), 388 (77.4%) were assigned to the same disease activity score band for UAS7TD versus UAS7OD1MAX (electronic supplementary Table S1). Additionally, of the 113 (22.6%) patients with different disease activity score band assignments based on UAS7TD versus UAS7OD1MAX, the majority (111/113 [98.2%]) were classified in adjacent score bands very close to the respective thresholds. Among these patients, the most commonly observed mismatches between UAS7TD and UAS7OD1MAX disease activity bands were seen in 36 (7.2%) patients classified in the 7–15 band based on UAS7TD, but with the UAS7OD1MAX in the higher activity band (16–27), and in 34 (6.8%) patients classified in the 28–42 band with UAS7TD, but with UAS7OD1MAX in the lower disease activity band (16–27). Results from scatterplots indicated that the UAS7TD and UAS7OD1MAX values for these patients tended to cluster near the threshold values for each score band range, with stronger clustering observed in the 7–15 score band (Fig. 2a) than in the 28–42 score band (Fig. 2b).
UAS7TD Versus UAS7OD2AVG
Among the 501 patients with complete data, 406 (81.0%) patients were grouped in the same band for UAS7TD versus UAS7OD2AVG (electronic supplementary Table S2). Similar to the previous analysis of 95 (19.0%) patients with different score band assignments based on UAS7TD versus UAS7OD2AVG, the majority (93/95 [97.9%]) were classified in adjacent score bands (electronic supplementary Table S2). Of all 501 patients, the most commonly observed mismatch between UAS7TD and UAS7OD2AVG disease activity bands was observed in 51 (10.2%) patients classified based on UAS7TD in the 28–42 band, but on UAS7OD2AVG in the 16–27 band. Similar to the scatterplots for UAS7TD versus UAS7OD1MAX, Fig. 3 shows that the UAS7TD and UAS7OD2AVG values for these patients are clustered toward the threshold values of the two score bands.
The coefficient of agreement [κ (95% confidence interval)] was 0.78 (0.75–0.82) for UAS7TD versus UAS7OD1MAX, and 0.82 (0.78–0.85) for UAS7TD versus UAS7OD2AVG, demonstrating substantial to almost perfect agreement between the scores [18].