Worst Itch Numeric Rating Scale for Prurigo Nodularis

Key Points

Question

Is the Worst Itch Numeric Rating Scale (WI-NRS) an appropriate tool for quantifying the intensity of pruritus associated with prurigo nodularis?

Findings

This secondary analysis of pooled data from the PRIME and PRIME2 clinical trials found that the WI-NRS was a valid and fit-for-purpose instrument to support the efficacy end points of clinical trials in patients with prurigo nodularis. WI-NRS demonstrated good evidence of validity, reliability, and sensitivity to change. The clinically meaningful within-patient improvement threshold was found to be 4 points.

Meaning

The WI-NRS may be a valid instrument for assessing pruritus in future clinical trials of prurigo nodularis.

This secondary analysis of 2 clinical trials examines the validity and fit-for-purpose of the Worst Itch Numeric Scale for assessing patient-reported outcomes of pruritus following treatment for prurigo nodularis.

Abstract

Importance

Prurigo nodularis (PN) is a debilitating skin disease characterized by the hallmark symptom of chronic itch; the intensity of itch in PN was assessed using the Worst Itch Numeric Rating Scale (WI-NRS) to evaluate the primary efficacy end point of 2 recent phase 3 studies of dupilumab treatment for PN.

Objective

To validate the psychometric properties and to determine the clinically meaningful improvement threshold for WI-NRS in patients with moderate to severe PN.

Design, Setting, and Participants

In this secondary analysis of the PRIME and PRIME2 trials, content validity of WI-NRS was assessed through in-depth patient interviews. Psychometric assessments used pooled data from masked, intention-to-treat (ITT) patients with PN from randomized, double-masked, and placebo-controlled studies. Psychometric assessments included test-retest reliability, construct validity, known-groups validity, and sensitivity to change in adult patients with moderate-to-severe PN. Thresholds for meaningful within-patient improvement in the WI-NRS score were determined using anchor and distribution-based approaches. Data were analyzed after completion of each study, December 2019 to November 2021 for PRIME and January 2020 to August 2021 for PRIME2.

Exposures

Dupilumab (300 mg) or placebo subcutaneously every 2 weeks for 24 weeks.

Main outcomes and measures

WI-NRS score at specified time points up to 24 weeks after randomization.

Results

A total of 20 patients were included across the 2 studies (mean [SD] age, 49.3 [17.2] years; 11 female [55%]); 311 patients were included in the pooled intention-to-treat analysis (mean [SD] age, 49.5 [16.1] years; 203 female [65.3%]). The WI-NRS questions (20 of 20 patients), recall period (19 of 20 patients), and response scale (20 of 20 patients) were easy to understand and relevant for patients with PN. Adequate test-retest reliability was observed between screening and baseline (intraclass correlation coefficient = 0.72, using Patient Global Impression of Severity [PGIS] to define stable patients). Convergent and discriminant validity was supported by moderate to strong correlations (absolute r range = 0.34-0.73) with other conceptually related measures and weaker correlations (absolute r range = 0.06-0.32) with less-related measures, respectively. WI-NRS was sensitive to change, as demonstrated by differences in change from baseline among groups (per PGIS change and PGI of Change [PGIC]). Using anchor-based approach with PGIS and PGIC, the clinically meaningful improvement threshold was 4 points (range, 3.0-4.5), which was also supported by distribution-based methods.

Conclusion and Relevance

This study found that WI-NRS may be a fit-for-purpose instrument to support efficacy end points measuring the intensity of itching in adults with PN.

Trial Registration

NCT04183335 (PRIME) and NCT04202679 (PRIME2)

Introduction

Prurigo nodularis (PN)—also called chronic nodular prurigo—is a debilitating, chronic, inflammatory skin disease characterized by itchy and hyperkeratotic lesions that commonly appear on the trunk, arms, and legs.^1,2,3 Itching is the most dominant and burdensome symptom in patients with PN^4,5 and is often severe and uncontrollable. Patients tend to scratch themselves to an extent of causing pain and bleeding. The chronic itching and repeated scratching (termed itch-scratch cycle) results in the development of erythematous, crusted, eroding nodular skin lesions.² The bothersome nature of itch and related pain imposes a substantial disease burden on the patients due to a reduced quality of life (QOL), disturbed sleep patterns, and low mental functioning.^6,7 The impact of PN on QOL has been reported to be higher than that in other common skin disorders, such as atopic dermatitis (AD) and psoriasis.^8,9

The primary goal in the management of PN is to reduce itch and control symptoms.^2,10 Itch is a subjective measure and would be most accurately assessed by directly eliciting the patient’s experience.¹¹ The need for the itch assessment for the comparison of treatment effect in clinical trials was emphasized by the International Forum for the Study of Itch.¹² Therefore, a patient-reported instrument serves as the most relevant tool to generate evidence that supports the assessment of treatment effects in PN clinical trials. Several patient-reported outcome measures to assess itch in pruritic diseases are currently available.^13,14,15

The Worst Itch Numeric Rating Scale (WI-NRS) is a single-item, patient-reported questionnaire designed to measure an individual’s “worst itch” (ie, intensity of itch) in the past 24 hours on an 11-point rating scale (with 0 representing “no itch” and 10 “worst itching imaginable”).¹⁶ The psychometric properties of WI-NRS for dermatologic conditions, such as psoriasis and AD, have been well documented^16,17,18,19; however, evidence supporting its validation in PN is limited.^20,21 This study aimed to evaluate the content validity of WI-NRS through qualitative interviews. Assessment of the psychometric properties and a confirmation of the within-patient improvement thresholds for the WI-NRS instrument was done using data from recent phase 3 PN trials. This will support the use of WI-NRS as the primary end point in phase 3 PN trials in patients either uncontrolled on topical prescription therapies or when those therapies are not advisable.

Methods

PRIME and PRIME2 were performed in line with the Declaration of Helsinki and Good Clinical Practice guidelines (study protocols available in Supplement 1 and Supplement 2).²² Local institutional review boards or ethics committees at each study center oversaw and approved the study. All patients provided written informed consent before participating in the trial.

Content Validity of WI-NRS

In-depth, 1-on-1 qualitative interviews were conducted for 20 patients with PN in the US (3 in New Orleans, Louisiana; 7 in St Louis, Missouri; and 10 in Los Angeles, California) using concept elicitation and cognitive debriefing methods (eAppendix 1 in Supplement 3). Patients participating in the content validity interviews met the following eligibility criteria: aged 18 years or older, confirmed PN diagnosis with or without mild AD (defined as Investigator’s Global Assessment [IGA] Scale score of 2) for at least 3 months, and willing to provide informed consent.

The concept elicitation interviews were carried out to understand whether patients’ experiences of PN (sign, symptoms, and impact) were in line with the WI-NRS content. These data helped in confirming whether the WI-NRS is a relevant instrument to measure itch in PN. After the concept elicitation interviews, patients self-administered the WI-NRS questionnaire and underwent cognitive cognitive debriefing interviews, which evaluated whether the WI-NRS wording, recall period, and response options were appropriate, clear, relevant, and well understood by patients. The interviews followed a semistructured guide and were conducted by skilled researchers in 3 waves of 6 to 7 patients each in English (eAppendix 2 in Supplement 3). Each interview lasted approximately 90 minutes and was audio recorded, transcribed verbatim, anonymized, coded by more than 2 independent researchers trained in qualitative research, and subjected to thematic analysis using ATLAS.ti version 7.5 (ATLAS.ti Scientific Software Development).

Psychometric Measurement Properties of WI-NRS

Psychometric properties of WI-NRS were evaluated in accordance with the regulatory guidance^23,24,25 and best practices in the field.²⁶ Psychometric analyses were performed using the pooled data from masked, intention-to-treat (ITT) patients with PN from PRIME and PRIME2 studies. Both studies were randomized, double-masked and placebo-controlled, designed to assess the efficacy and safety of dupilumab over a 24-week treatment period in adult patients with PN. Eligible patients were those with a clinically confirmed diagnosis of PN with an average worst itch score of 7 or higher on the WI-NRS in the 7 days prior to baseline and those with at least 20 PN lesions on both legs, both arms, and/or trunk at screening visit and baseline. The primary end point of these studies was the percentage of patients with improvement (ie, reduction) in WI-NRS score from baseline to week 12 (PRIME) and week 24 (PRIME2) of 4 points or higher.²⁷ For all time points and analyses, a weekly average WI-NRS score was defined as the average score over a 7-day period (a minimum of 4 daily scores out of the 7 days was required to calculate the baseline average score).

The psychometric evaluation of the WI-NRS included assessment of test-retest reliability, construct validity, known-groups validity, and responsiveness or sensitivity to change in other disease outcomes. Several clinician-reported outcome (CRO) and PRO measures were used to support this analysis (eTable 1 in Supplement 3).

Test-Retest Reliability

Test-retest reliability of the WI-NRS was determined by calculating the intraclass correlation coefficient (ICC) in patients with stable PN, defined as patients reporting no change on (1) the Patient Global Impression of Severity (PGIS) from baseline to week 4, (2) the PGIS from week 8 to week 12, and (3) the Patient Global Impression of Change (PGIC) at week 4 (note the PGIC measures change since starting the study medication). An ICC of 0.70 or higher indicated adequate test-retest reliability.

Construct Validity

Construct validity was assessed by examining convergent or discriminant and known-groups validity at baseline and week 12. Convergent validity, referring to how well constructs that theoretically should be related to each other are observed to be related, was evaluated by Spearman rank and polyserial correlation coefficients between WI-NRS scores and other PRO instruments. The objective was to demonstrate a moderate to strong correlation of WI-NRS (defined as absolute correlation 0.30 or higher) with other measures addressing similar concepts (Skin Pain-NRS, Sleep-NRS, Dermatology Life Quality Index [DLQI] total score, PGIS, EuroQol 5-Dimensions [EQ-5D] Pain/Discomfort Item, and DLQI item 1 [assessing disease symptoms, itch, pain, sore, and stinging]). It was expected that some measures (Investigator’s Global Assessment for Prurigo Nodularis Activity [IGA PN-A], Investigator’s Global Assessment for Prurigo Nodularis Stage [IGA PN-S], Prurigo Activity Score [PAS] items 4 [number of pruriginous lesions] and 5b [proportion of healed prurigo lesions], anxiety subscale of the Hospital Anxiety and Depression Scale [HADS-A] and depression subscale of the Hospital Anxiety and Depression Scale [HADS-D]) would be less related than others to the construct measured by the WI-NRS (low correlations defined as absolute correlation below 0.30 were expected), as some are rated by clinicians and HADS measures anxiety and depression.

Known-groups validity was evaluated to assess the ability of WI-NRS to distinguish between groups based on PN disease severity level (per PGIS and IGA PN-S) and health-related QOL (per DLQI total score). For this, separate analysis of variance (ANOVA) were used to test for group differences in the mean WI-NRS scores at baseline and week 12.

Responsiveness

Responsiveness, or sensitivity to change of WI-NRS, was assessed using the Spearman correlation coefficients between changes in the WI-NRS scores and changes in the Skin Pain-NRS, Sleep-NRS, PGIS, and DLQI total scores from baseline to week 12 and week 24 and the PGIC values at weeks 12 and 24. Furthermore, an analysis of covariance (ANCOVA) on change in WI-NRS, using PGIS and PGIC groups (deterioration, stable, improvement) as covariates, was performed in a separate analyses at each visit controlling for baseline scores. Concurrent improvement in PRO measures was expected. Least squares (LS) mean change at each visit, as well as the standardized response mean (SRM)—calculated as mean change from baseline divided by the standard deviation (SD) of the change—was also provided.

Threshold for Meaningful Change for WI-NRS

Anchor- and distribution-based analyses were conducted to define the thresholds for within-patient meaningful change on the WI-NRS from baseline to week 12 and week 24. Anchor-based analysis is the preferred method for establishing meaningful change thresholds recommended by regulatory authorities.^24,28,29 An anchor should exhibit a minimum correlation of 0.371 with the PRO of interest to be considered appropriate for clinically meaningful thresholds estimation.³⁰ In this analysis, PGIS and PGIC were identified as the suitable anchors, exhibiting correlations higher than 0.50 with WI-NRS. Improvements in 1 category on the PGIS and in the moderately better category on the PGIC at week 24 were selected as the target anchor change categories for deriving a range of thresholds for meaningful within-patient improvement. The empirical cumulative distribution function (eCDF) plots by anchor categories were examined, and distribution-based estimates were also examined as supportive. The distribution-based methods included 0.5 SD of baseline WI-NRS and standard error of measurement (SEM) (SD × √1 − ICC, using the lowest ICC from test-retest reliability) for the baseline WI-NRS score.

The threshold point estimate was determined based on the data at week 24, and the week 12 and week 24 data were used to assess the threshold range. Median values of the change from baseline in WI-NRS at weeks 12 and 24 for each of the anchor response groups were calculated. A mixed-model repeated measures approach, controlled for baseline values, was also used to estimate LS mean change in WI-NRS and 95% CI for each of the anchor response improvement groups. All P values are nominal.

Results

Content Validity (Qualitative Interviews)

Between December 2019 and February 2022, interviews were conducted for 20 patients with PN in St Louis, Missouri (7 patients); New Orleans, Louisiana (3 patients); and Los Angeles, California (10 patients). The mean age (SD) of patients was 49.3 (17.2) years, and 11 patients (55%) were female (Table 1). Among the 20 adults who participated in the interview, 9 had comorbid AD (45%).

Table 1. Demographics and Baseline Characteristics.

Characteristic	Patients, No. (%)
Demographics and baseline characteristics of patients included in content validity analysis
Total patients, No.	20
Age, mean (SD), y	49.3 (17.2)
Sex
Female	11 (55)
Male	9 (45)
Race or ethnicity
Black or African American	3 (15)
Hispanic or Latino	7 (35)
White	9 (45)
Othera	<3 (<15)
Hispanic heritage
Not Spanish, Hispanic, or Latino	13 (65)
Mexican, Mexican American, or Chicano	6 (30)
Puerto Rican	<3 (<15)
Time since diagnosis, mean (SD), y	5.5 (7.1)
Psychometric analysis pooled ITT cohort
Total patients, No.	311
Age, mean (SD), y	49.5 (16.1)
Sex
Female	203 (65.3)
Male	108 (34.7)
Region
Western countries	130 (41.8)
Asian	93 (29.9)
Latin America	55 (17.7)
East Europe	33 (10.6)
Race
American Indian or Alaska Native	6 (1.9)
Asian	106 (34.3)
Black or African American	19 (6.2)
Native Hawaiian or Other Pacific Islander	<3 (<1)
White	176 (57.0)
Multiple	<3 (<1)
Ethnicity
Hispanic or Latino	60 (19.3)
Not Hispanic or Latino	251 (80.7)
Hispanic or Latino	60 (19.3)
History of atopyb
Atopic	135 (43.4)
Nonatopic	176 (56.6)
WI-NRS scorec
Mean (SD)	8.50 (1.01)
Median (range)	8.50 (2.4-10.0)

Abbreviations: ITT, intention to treat; WI-NRS, Worst Itch Numeric Rating Scale.

^a Participant chose other and indicated they were Eastern European in their response.

^b Medical history of atopic dermatitis, allergic rhinitis, rhinoconjunctivitis, asthma, or food allergy.

^c There was a total of 309 patients.

All 20 participants reported itch as the most common and bothersome symptom associated with PN. Other frequently reported symptoms were bleeding (19 of 20 [95%]), scratching (19 of 20 [95%]), scarring (15 of 20 [75%]), burning sensation (15 of 20 [75%]) and redness (14 of 20 [70%]). The WI-NRS instrument was considered relevant by all 20 patients’ experiences of PN, and the items (20 of 20), recall period (19 of 20), and response scale (20 of 20) were well understood by the patients and provided a comprehensive assessment of itch severity. Across the first 2 waves, 2 patients reported that they were not only thinking about itching related to PN but also considering their comorbid AD when answering. All participants with comorbid AD in wave 3 (4 of 4 patients) reported thinking only of their PN after adding additional instructions to remind participants to think of their PN.

Psychometric Assessment (PRIME and PRIME2 Studies)

Patient Characteristics

A total of 311 patients with PN from PRIME and PRIME2 trials were included in the psychometric analysis. Patients had a mean (SD) age of 49.5 (16.1) years, 203 (65.3%) were female, and 135 (43.4%) had a history of atopy. The mean (SD) weekly WI-NRS score at baseline was 8.5 (1.0) (Table 1).

Test-Retest Reliability

Adequate test-retest reliability was reported in stable patients (defined as per PGIS) from screening to baseline (224 patients; ICC, 0.72; 95% CI, 0.63-0.79) and week 8 to week 12 (177 patients; ICC, 0.88; 95% CI, 0.82-0.92). Lower reliability was observed when stable patients were defined with PGIS from baseline to week 4 (163 patients; ICC, 0.44; 95% CI, 0.15-0.63) as well as PGIC at week 4 (56 patients; ICC, 0.49; 95% CI, 0.21-0.68) (eTable 2 in Supplement 3).

Construct Validity

As expected, moderate to strong correlations were observed between the WI-NRS scores and scores of similar constructs (baseline range: absolute r = 0.41 [Sleep-NRS] to 0.60 [Skin Pain-NRS]; week 12 range: absolute r = 0.46 [Sleep-NRS] to 0.73 [Skin Pain-NRS]) (Table 2). Low correlations were observed between the WI-NRS score and the measures expected to be less related (baseline range: absolute r = 0.06 [PAS item 5b] to 0.27 [IGA PN-S]; week 12 range: 0.23 [HADS-A] to 0.49 [IGA PN-S]) (Table 2).

Table 2. Psychometric Evaluation of WI-NRS in Stable Patients (per PGIS and PGIC) in the Pooled ITT Population.

Scale	Outcome source	Spearman rank correlation of WI-NRS with other instruments, r
		Baseline (n = 309)	Week 12 (n = 292)
Convergent and discriminant validity of WI-NRS
Convergent validity
Skin Pain-NRSa	PRO	0.60	0.73
Sleep-NRSb	PRO	−0.41	−0.46
DLQI total scorec	PRO	0.34	0.49
PGISd	PRO	0.41	0.65
EQ-5D-5L Pain/Discomfort	PRO	0.36	0.42
DLQI item 1e	PRO	0.54	0.71
Discriminant validity
IGA PN-A	CRO	0.17	0.49
IGA PN-S	CRO	0.27	0.49
PAS item 4f	CRO	0.12	0.40
PAS item 5bg	CRO	−0.06	−0.32
HADS-A	PRO	0.23	0.23
HADS-D	PRO	0.12	0.24

Abbreviations: CRO, clinician-reported outcome; EQ-5D-5L, EuroQol 5-Dimensions 5-Level; DLQI, Dermatology Life Quality Index; HADS-A, Hospital Anxiety and Depression Scale; HADS-D, Hospital Anxiety and Depression Scale; IGA PN-A, Investigator’s Global Assessment for Prurigo Nodularis Activity; IGA PN-S, Investigator’s Global Assessment for Prurigo Nodularis Stage; ITT, intention-to-treat; NRS, numeric rating scale; PAS, Prurigo Activity Score; PGIC, Patient Global Impression of Change; PGIS, Patient Global Impression of Severity; PRO, patient-reported outcome; WI-NRS, Worst Itch-Numeric Rating Scale.

^a Skin Pain-NRS is a single-item scale designed to assess worst skin pain daily, rated from 0 (no pain) to 10 (worst pain possible).

^b Sleep-NRS is a single-item, generic instrument developed to assess sleep quality daily upon awakening: rated from 0 (worst possible sleep) to 10 (best possible sleep).

^c The DLQI total score is derived from the sum of all 10 items, ranging from 0 to 30 (with a high score indicative of poor health-related quality of life).

^d PGIS is a daily 11-category scale to assess severity over the previous 24 hours, ranging from 0 (not present) to 10 (extremely severe).

^e DLQI item 1 assesses disease symptoms, itch, pain, sore and stinging.

^f PAS item 4 assesses exact number of pruriginous lesions in the representative area (excluding scars).

^g PAS item 5b assesses proportion of healed prurigo lesions.

Known-Groups Validity

The WI-NRS was able to distinguish between patients with different disease severity levels and QOL effects (Table 3). Significant differences in the mean WI-NRS scores between groups defined by PGIS, IGA PN-S, and DLQI scores were observed at baseline and week 12. The LS means were ordered in the expected direction; lower (better) WI-NRS scores were observed for patients with lower (better) PGIS responses. The difference in LS means at week 12 between adjacent stratification groups generally exceeded the SEM at 0.75, with a range between 0.52 to 1.96, and were closer to the 0.5-SD baseline (between 0.32 and 0.79).

Table 3. Psychometric Evaluation of WI-NRS in Stable Patients (per PGIS and PGIC) Using Known Groups at Week 12 in the Pooled ITT Population.

Measures	Groupa	Patients, No.	WI-NRS
			LS means (SE) [95% CI]	P valueb	Effect sizec
PGIS
Baseline	None, mild, or moderate	95	7.95 (0.10) [7.76-8.14]	<.001	NA
	Severe	214	8.74 (0.06) [8.62-8.87]		0.84
Week 12	None or mild	120	4.22 (0.18) [3.86-4.59]	<.001	NA
	Moderate	104	6.07 (0.20) [5.68-6.46]		0.87
	Severe	68	8.03 (0.25) [7.54-8.51]		1.05
IGA PN-S
Baseline	Severe	104	8.77 (0.10) [8.58-8.96]	<.001	NA
	Moderate	205	8.36 (0.07) [8.22-8.49]		0.42
Week 12	Clear or almost clear	65	4.20 (0.27) [3.66-4.74]	<.001	NA
	Mild	65	4.81 (0.27) [4.27-5.35]		0.26
	Moderate	118	6.49 (0.20) [6.09-6.89]		0.75
	Severe	44	7.57 (0.33) [6.92-8.23]		0.52
DLQI total scored
Baseline	No, small or moderate effect	56	8.03 (0.13) [7.78-8.28]	<.001	NA
	Very large effect	142	8.35 (0.08) [8.20-8.51]		0.33
	Extremely large effect	111	8.92 (0.09) [8.74-9.10]		0.68
Week 12	No effect	28	3.49 (0.42) [2.66-4.31]	<.001	NA
	Small effect	95	5.08 (0.23) [4.63-5.53]		0.68
	Moderate effect	68	5.60 (0.27) [5.07-6.13]		0.23
	Very large effect	75	6.87 (0.26) [6.37-7.38]		0.58
	Extremely large effect (21-30 score)	26	7.99 (0.43) [7.13-8.84]		0.54

Abbreviations: ANOVA, analysis of variance; DLQI, Dermatology Life Quality Index; IGA PN-S, Investigator’s Global Assessment for Prurigo Nodularis Stage; ITT, intention to treat; LS, least squares; NA, not applicable; NRS, numeric rating scale; PGIC, Patient Global Impression of Change; PGIS, Patient Global Impression of Severity; WI-NRS, Worst Itch-Numeric Rating Scale.

^a Adjacent category groups were combined if they had a low sample size (ie, fewer than 30 patients), except for the DLQI no effect and extremely large effect groups at week 12 (28 and 26 patients, respectively).

^b P value is based on ANOVA comparisons with α = .05 level. An ANOVA was used to test for group differences in scores at baseline and week 12.

^c Effect sizes were calculated as the mean difference in WI-NRS scores between groups divided by the pooled SD of those groups.

^d DLQI group classification: No effect (0 score), small effect (2 to 5 score), moderate effect (6 to 10 score), and very large effect (11 to 20 score).³¹

Responsiveness

Moderate to strong correlations were observed for the change in the WI-NRS scores with changes in Skin Pain-NRS, Sleep-NRS, DLQI total score and PGIS, and with PGIC from baseline to week 12 (absolute correlation coefficient r between 0.40 [Sleep-NRS] and 0.76 [Skin Pain-NRS]) and week 24 (absolute r between 0.46 [Sleep-NRS] and 0.76 [Skin Pain-NRS]) (Table 4). Furthermore, statistically significant differences were observed at both week 12 and week 24 among groups (deteriorated, stable, and improved) defined based on change in PGIS and PGIC (eFigure 1 in Supplement 3). Large standardized response means (defined as greater than 0.8) were observed for patients reporting improvement in the PGIS and PGIC at week 12 and week 24, vs small to moderate (defined as from above 0.2 to 0.5 and above 0.5 to 0.8) standardized response means for patients reporting stability or deterioration on the PGIS and PGIC at week 12 and week 24 (eTable 3 in Supplement 3).

Table 4. Responsiveness to Change of WI-NRS in the Pooled ITT Population.

Scores	Change from baseline to week 12		Change from baseline to week 24
	No. of patients	Correlation coefficient, ra	No. of patients	Correlation coefficient, ra
Skin Pain-NRS	294	0.76	266	0.76
Sleep-NRS	294	−0.40	266	−0.46
DLQI total score	292	0.46	265	0.47
PGISb	292	0.60	265	0.60
PGICb,c	292	0.57	265	0.59

Abbreviations: DLQI, Dermatology Life Quality Index; NRS, numeric rating scale; PGIC, Patient Global Impression of Change; PGIS, Patient Global Impression of Severity; PRO, patient-reported outcome; WI-NRS, Worst Itch-Numeric Rating Scale.

^a Spearman rank correlation coefficients between changes in WI-NRS scores and changes in other PRO scores are shown for WI-NRS with Skin Pain-NRS, Sleep-NRS and DLQI total score.

^b Polyserial correlations are shown for WI-NRS with PGIS and PGIC.

^c The actual PGIC response was considered, not the change from baseline.

Meaningful Change Threshold

Sufficient correlations were observed between the change in WI-NRS scores and the change in PGIS (r = 0.60 at week 12 and week 24) and PGIC (r = 0.57 at week 12 and r = 0.59 at week 24) anchors at both week 12 and week 24, supporting their use as anchors. At week 12, the median change from baseline in the WI-NRS score was −2.79 (95% CI, −3.39 to −2.56) in patients with 1-category improvement on PGIS and −3.32 (95% CI, −3.86 to −2.78) in patients reporting moderately better results on PGIC. At week 24, the median change from baseline was −3.88 (95% CI, −4.34 to −3.42) on PGIS with improved outcomes on 1 category and −4.19 (95% CI, −4.34 to −3.42) on the PGIC moderately better category. On rounding up the values to the nearest 0.5, the proposed within-patient threshold point estimate was 4.0 (range, 3.0-4.5) (Table 5; eFigures 2 and in Supplement 3), confirming the prespecified threshold used in both studies. This value exceeds the distribution-based results (0.5-SD = 0.50; SEM = 0.75), as well as the lower bound of the 95% CI for the mean of the change from baseline in the no change group (for PGIS at week 24, the lower bound was −2.22). Furthermore, the eCDF curves indicated a clear separation between the anchor categories when observing a 4-point change on the WI-NRS (eFigure 4 in Supplement 3).

Table 5. Estimates of Within-Patient Meaningful Improvement Thresholds for WI-NRS in the Pooled ITT Population.

Anchor/time point	Anchor categories	Raw change				Model results, LS means (95% CI)
		Patients, No.	Mean	Median	95% CI
PGIS
Week 12	Improved 3 categories	3	−5.94	−7.17	−12.12 to 0.25	−4.35 (−5.78 to −2.92)
	Improved 2 categories	80	−4.63	−5.30	−5.16 to −4.11	−3.70 (−4.04 to −3.36)
	Improved 1 category	97	−2.97	−2.79	−3.39 to −2.56	−2.99 (−3.30 to −2.68)
	No change	103	−1.21	−0.76	−1.59 to −0.82	−1.77 (−2.09 to −1.46)
	Moderate or large improvement	83	−4.68	−5.32	−5.19 to −4.16	NA
	Any improvement	180	−3.76	−3.27	−4.11 to −3.41	NA
	Small or moderate improvement	177	−3.72	−3.26	−4.07 to −3.38	NA
Week 24	Improved 3 categories	16	−7.19	−7.42	−8.23 to −6.15	−5.07 (−5.87 to −4.27)
	Improved 2 categories	85	−4.99	−5.43	−5.56 to −4.42	−4.72 (−5.11 to −4.33)
	Improved 1 category	99	−3.88	−3.67	−4.34 to −3.42	−3.69 (−4.05 to −3.32)
	No change	57	−1.66	−1.02	−2.22 to −1.10	−2.24 (−2.70 to −1.79)
	Moderate or large improvement	101	−5.34	−6.00	−5.87 to −4.81	NA
	Any improvement	200	−4.62	−4.86	−4.98 to −4.25	NA
	Small or moderate improvement	184	−4.39	−4.75	−4.76 to −4.02	NA
PGIC
Week 12	Very much better	92	−4.26	−4.21	−4.72 to −3.80	−3.59 (−3.91 to −3.26)
	Moderately better	80	−3.32	−2.86	−3.86 to −2.78	−3.10 (−3.43 to −2.77)
	A little better	62	−1.83	−1.41	−2.38 to −1.29	−2.40 (−2.77 to −2.04)
	No change	31	−0.68	−0.43	−1.24 to −0.12	−1.42 (−1.91 to −0.94)
	Moderate or large improvement	172	−3.82	−3.57	−4.18 to −3.47	NA
	Any improvement	234	−3.30	−2.90	−3.61 to −2.98	NA
	Small or moderate improvement	142	−2.67	−2.37	−3.07 to −2.27	NA
Week 24	Very much better	116	−5.20	−5.83	−5.70 to −4.70	−4.65 (−5.01 to −4.30)
	Moderately better	62	−4.19	−4.10	−4.77 to −3.62	−4.04 (−4.48 to −3.61)
	A little better	39	−2.38	−1.71	−3.08 to −1.67	−2.95 (−3.48 to −2.42)
	No change	23	−1.68	−1.14	−2.36 to −1.01	−2.02 (−2.70 to −1.35)
	Moderate or large improvement	178	−4.85	−5.26	−5.23 to −4.46	NA
	Any improvement	217	−4.40	−4.80	−4.76 to −4.04	NA
	Small or moderate improvement	101	−3.49	−3.29	−3.96 to −3.02	NA

Abbreviations: ITT, intention to treat; LS, least squares; NA, not applicable; PGIC, Patient Global Impression of Change; PGIS, Patient Global Impression of Severity; WI-NRS, Worst Itch-Numeric Rating Scale.

Discussion

This study confirms the psychometric and content validity of WI-NRS, thereby establishing it to be a clear, comprehensive, and relevant tool to measure itch in PN. The study also confirms the prespecified value of the meaningful change threshold for WI-NRS, which can aid in the interpretation of results in clinical practice.

Test-retest reliability for WI-NRS was adequate between the test and retest time points with ICC values falling above the recommended threshold of 0.70 for the PRO instrument (ICC range, 0.72-0.88),³² indicating stability of the scores in patients. However, lower test-retest reliability was observed with PGIS from baseline to week 4 and with PGIC at week 4, respectively. This might be due to a large incremental weekly benefit expected early in the treatment. A 2022 study³³ including 42 patients with PN also reported the WI-NRS to have adequate test-retest reliability, confirming the current study results. In another psychometric analysis, an ICC of 0.78 was noted for WI-NRS in patients with PN.³⁴ Consistently, excellent reliability (ICC, 0.99) for WI-NRS has also been demonstrated in patients with chronic prurigo.³⁵ Furthermore, the results of this study were in line with the ICC values reported for other PROs in patients with chronic itch (ICC between 0.74 and 0.87).^13,36

Results from the construct validity assessment were as expected, and the pattern of correlations with other measures confirmed the convergent validity of WI-NRS. Correlations with the Skin Pain-NRS and DLQI item 1 were strong in magnitude at baseline and week 12, indicating an association between itch, pain, and QOL. Furthermore, WI-NRS could easily distinguish between groups of extreme disease severity levels. Similarly, WI-NRS exhibited a strong correlation with the verbal rating scale for itch (r between 0.65 and 0.92) and moderate correlation with the pruritus-specific QOL rating instrument and DLQI (r between 0.36 and 0.46) in PN. Also, WI-NRS demonstrated an ability to differentiate between different levels of disease severity (itch intensity and proportion of prurigo lesions) and itch-related QOL.³⁴ A 2021 study by Storck et al³⁷ demonstrated a similar pattern, with evidence of an association between WI-NRS and DLQI total scores in patients with different pruritic dermatoses. Moderate to strong correlations were also noted between changes in the WI-NRS scores and changes in other PROs, providing sufficient evidence of sensitivity or responsiveness of the WI-NRS to the changes in other disease outcomes. Similarly, WI-NRS displayed sensitivity toward changes in disease and QOL-related outcomes in another psychometric analysis.³⁴ Overall, these findings corroborate the sound ability of WI-NRS in the measurement of itch in patients with PN. Data were consistent with previous psychometric analyses of numeric rating scales (NRSs) measuring the worst itch and average itch in the psoriasis and AD clinical trials.^17,34,38

As per the anchor-based analysis, a 4-point improvement on the WI-NRS can be considered as a within-patient meaningful change in the patients with PN. These data support the primary end point used in PRIME and PRIME2 clinical studies.²⁷ A 2022 study³³ reported a clinically meaningful threshold of 3.8-point improvement in WI-NRS in PN. Furthermore, an improvement threshold of 2.3 to 4.5 points, corresponding to a 30% to 60% improvement in WI-NRS scores was reported in a previous psychometric analysis in patients with PN.³⁴ In a recent study, utilizing phase 2 clinical trial data (67 patients), the within-patient meaningful change threshold for peak pruritus NRS (PP-NRS) was reported to range from 2 to 5 points.²¹ The difference in the threshold range for PP-NRS and WI-NRS could be attributed to the difference in the terminologies used to define the instruments (itch “at its most intense” in the WI-NRS vs “at the worst moment” in the PP-NRS). Similar findings were also reported in other studies for NRS instruments measuring pruritus in PN.^21,34,39

Limitations

This study had several limitations. Data from clinical trials with stringent patient eligibility criteria were used for psychometric analyses. Therefore, patients may have experienced changes in condition, such as worsening, that may not have been fully captured. Furthermore, a few patients reported worsening in the WI-NRS scores; therefore, only the improvement threshold could be estimated. Using observational studies might provide a more accurate insight of the thresholds for meaningful within-patient worsening.

Conclusions

The WI-NRS was found to be a valid and reliable instrument to capture itch data of patients with PN, and a 4-point improvement on WI-NRS may be used to represent a within-patient meaningful improvement in patients with PN. WI-NRS is a fit-for-purpose instrument to support efficacy end points measuring the intensity of itching in adults with PN.

Supplement 1.

PRIME Protocol and Statistical Analysis Plan

Supplement 2.

PRIME2 Protocol and Statistical Analysis Plan

Supplement 3.

eTable 1. PRO and ClinRO measures used in psychometric analyses and meaningful threshold evaluation of WI-NRS in patients with PN

eTable 2. Test-Retest Reliability of WI-NRS—Pooled ITT Population

eTable 3. Sensitivity to Change of WI-NRS from Baseline to Weeks 12 and 24- including Standardized Response Mean – pooled ITT population

eFigure 1. Sensitivity to change of WI-NRS using mean changes from baseline to Weeks 12 and 24 in WI-NRS scores by PGIC and PGIS groups - pooled ITT population

eFigure 2. Within-patient change thresholds for WI-NRS using anchor-based approaches—pooled ITT population

eFigure 3. Within-patient change thresholds for WI-NRS: LS mean change from baseline—pooled ITT population

eFigure 4. eCDF for change from baseline in WI-NRS - pooled ITT population: eCDF for WI-NRS scores at Week 12 and Week 24 per PGIS and PGIC change categories

eAppendix 1. Study Manual/Interview Guide: Hybrid Concept Eliciation/Cognitive Debriefing Patient Interviews in Prurigo Nodularis

eAppendix 2. Interviewer Form

Supplement 4.

Data Sharing Statement