FormalPara Infographic
figure a
FormalPara Key Summary Points

Why carry out this study?

The severity, burden, and impact of atopic dermatitis (AD) can be assessed using various clinician- and patient-reported outcomes, but these do not necessarily correlate well and there are limited data in Asian patients.

The current study assessed how well clinician- and patient-reported outcomes correlate in Asian patients with AD.

What was learned from the study?

Although correlations between clinician-reported outcomes were high/very high and those between dermatological patient-reported outcomes were moderate, correlations between clinician-reported and dermatological patient-reported outcomes were low/moderate.

This highlights the importance of including patient-reported outcomes when assessing disease severity, burden, and impact to guide treatment and assess response to treatment.

Digital Features

This article is published with digital features, including an infographic, to facilitate understanding of the article. To view digital features for this article, go to https://doi.org/10.6084/m9.figshare.24885015.

Introduction

Atopic dermatitis (AD) is a chronic relapsing condition that causes itching and impacts health-related quality of life (HRQoL) [1]. Various instruments are available for clinicians to assess disease severity and extent (e.g., Investigator’s Global Assessment [IGA], Eczema Area and Severity Index [EASI], and body surface area [BSA] affected) and for patients to report their AD severity (e.g., Patient-Oriented Eczema Measure [POEM] and Worst Itch Numerical Rating Scale [WI-NRS]) and the impact of AD on their HRQoL (e.g., Dermatology Life Quality Index [DLQI]).

Various studies have reported that although clinician-reported outcomes generally correlate well with each other [2,3,4], they tend to correlate less well with patient-reported outcomes [2, 3, 5,6,7]. For example, some patients with low clinician-reported disease severity may report a high burden, highlighting the importance of assessing patient-reported outcomes. As skin color can affect AD assessments [5, 6] and most of these studies included predominantly White patients [2, 4,5,6,7] or Japanese patients with exclusively moderate-to-severe AD [3], the objective of the current study was to assess correlations between clinician- and patient-reported outcomes in Japanese patients with a wider range of AD severity levels.

Methods

Study Design

ADDRESS-J was a prospective, non-interventional, longitudinal study that evaluated the real-world effectiveness and safety of AD treatments in adults (aged 20–59 years) with moderate-to-severe AD (IGA 3 or 4) [3]. ADDRESS-J was conducted from 2016 to 2019, and each patient was followed for 2 years or less. Patients must have been prescribed a new AD medication or been switched to a higher potency/dose at baseline [3]. There were no restrictions on the use of concomitant medications/procedures after enrollment.

The current study analyzed anonymized data from the ADDRESS-J study. The original study was carried out at 30 Japanese medical institutions in accordance with the ethical principles derived from the 1964 Declaration of Helsinki plus subsequent amendments and the “Ethical Guidelines for Medical and Health Research Involving Human Subjects,” which was established in 2014. The protocol and other documents were approved by institutional review boards/ethics committees, details of which can be found in the Ethical Approval section. All patients provided written informed consent before any study procedures [3].

Outcomes

Clinician- and patient-reported outcomes were collected at baseline and every 3 months throughout the 24-month observation period. Clinician-reported outcomes were IGA (0–4), EASI (0–72), and BSA (0–100%), and dermatological patient-reported outcomes were POEM (0–28), DLQI (0–30), and WI-NRS (0–10); higher scores indicate more severe disease. We also studied two general HRQoL outcomes: the 5-dimension EuroQoL questionnaire (EQ-5D) (0–1) and EuroQol Visual Analog Scale (EQ-VAS) (0–100); higher scores indicate better health. Various biomarkers were analyzed when these were available in routine clinical practice: serum levels of thymus and activation-regulated chemokine (TARC), lactate dehydrogenase (LDH), total immunoglobulin E (IgE), and peripheral blood eosinophil counts (PB EOS).

Statistical Analysis

Analyses of correlation between outcomes were performed, with scores ranked from least to most severe disease, regardless of the direction of the numeric scores. Spearman’s correlation coefficients were calculated using all available pooled data from baseline through 24 months. Values of 0.90–1.00 were interpreted to show very high correlations; 0.70–0.89, high; 0.50–0.69, moderate; 0.30–0.49, low; and 0.00–0.29, negligible [8]. The statistical package SAS® (version 9.2 or later; SAS Institute, Cary, NC, USA) was used for all analyses.

Results

Of the 300 patients included in ADDRESS-J (60.7% male; median age 34 years) [3], 288 had at least one post-baseline evaluation [9]. At baseline, most of these 288 patients (80.2%) were only receiving topical AD treatment (corticosteroid and/or calcineurin inhibitor), 14.6% systemic treatment, and 5.6% ultraviolet phototherapy [9]. By 24 months, the corresponding figures were 59.7%, 29.9%, and 12.2%, respectively, and 2.1% had received biologics [9].

Mean clinician- and patient-reported outcomes improved from baseline to 6 months and were then maintained for the remainder of the study (Table 1). The proportion of patients with IGA 3 or 4 fell from 100% at baseline to 50.4%, 44.6%, and 42.0% at 6, 12, and 24 months, respectively, providing data for a range of AD severities.

Table 1 Scores at baseline and 6, 12, and 24 months; and all pooled (baseline through 24 months)
Full size table

Correlations between the three clinician-reported outcomes (IGA, EASI, BSA) were high/very high (Spearman’s correlation coefficients 0.76–0.92), while those between the three dermatological patient-reported outcomes (POEM, DLQI, WI-NRS) were moderate (0.53–0.64), and those between clinician- and patient-reported outcomes were mainly low (0.37–0.51) (Fig. 1). Correlations between the general patient-reported HRQoL outcomes (EQ-5D, EQ-VAS) and the other six outcomes were negligible to moderate (0.26–0.60) (Fig. 1).

Fig. 1
figure 1

Spearman’s correlation coefficients between outcomes using the pooled data. BSA body surface area, DLQI Dermatology Life Quality Index, EASI Eczema Area and Severity Index; EQ-5D 5-dimension EuroQoL questionnaire, EQ-VAS EuroQol Visual Analog Scale, HRQoL health-related quality of life, IGA Investigator’s Global Assessment, IgE immunoglobulin E, LDH lactate dehydrogenase, PB EOS peripheral blood eosinophil counts, POEM Patient-Oriented Eczema Measure, TARC thymus and activation-regulated chemokine, WI-NRS Worst Itch Numerical Rating Scale. *Cutoffs from Mukaka [8]

Full size image

Correlations between clinician- or dermatological patient-reported outcomes were low/moderate for TARC or LDH (Spearman’s correlation coefficients 0.44–0.63), low for PB EOS (0.33–0.41), and generally negligible for total IgE (0.01–0.32), with better correlations with biomarkers for clinician- vs. dermatological patient-reported outcomes overall (Fig. 1). Correlations between biomarkers were low/moderate (0.37–0.63) (Fig. 1).

Figure 2 shows how the individual scores were correlated. For example, the EASI/BSA correlation shows a strong linear relationship, but the EASI/POEM and EASI/WI-NRS correlations had many datapoints with low EASI but high POEM or WI-NRS, indicating a potential discrepancy between clinician- and patient-reported outcomes.

Fig. 2
figure 2

Correlations for clinician- and patient-reported outcomes using the pooled data. BSA body surface area, DLQI Dermatology Life Quality Index, EASI Eczema Area and Severity Index, EQ-5D 5-dimension EuroQoL questionnaire, EQ-VAS EuroQol Visual Analog Scale, IGA Investigator’s Global Assessment, POEM Patient-Oriented Eczema Measure, WI-NRS Worst Itch Numerical Rating Scale

Full size image

Discussion

In this analysis of approximately 2000 datapoints, correlations between clinician-reported outcomes were high/very high (Spearman’s correlation coefficients 0.76–0.92), with moderate correlations (0.53–0.64) between dermatological patient-reported outcomes. However, correlations between clinician- and dermatological patient-reported outcomes were mainly low (0.37–0.51), indicating a discrepancy in disease assessment and perception between clinicians and patients. This highlights the importance of including patient-reported outcomes when assessing disease severity for treatment selection and response [10, 11] (as is recommended for dupilumab in the UK [12]). In Japan, newer systemic treatment options have recently been approved for patients with AD [1, 13]. These are only indicated for patients who have IGA ≥ 3 and BSA ≥ 10% and EASI ≥ 16 (or head and neck EASI ≥ 2.4) [1, 13], which are all clinician-reported. The addition of patient-reported outcomes might be clinically meaningful for identifying patients not otherwise considered to have high disease burden.

The current results are well aligned with correlation analyses performed in predominantly White patients [2, 4,5,6,7]. Overall, POEM and WI-NRS appear to be suitable for assessing disease improvement, whereas EASI and DLQI seem to be appropriate for assessing more severe disease. This is because lower POEM or WI-NRS scores are associated with milder skin lesions and better HRQoL, while higher scores may not always indicate severe skin lesions or poor HRQoL.

Although there was a moderate correlation between DLQI and EQ-5D (Spearman’s correlation coefficient 0.60), correlations between the other outcomes and EQ-5D or EQ-VAS were negligible/low (0.26–0.49), showing the limited clinical relevance of these outcomes for Japanese patients with AD. The limitation (ceiling effect) of EQ-5D as a tool for HRQoL assessment has previously been demonstrated [14].

Although the biomarker results should be interpreted with caution because of the limited data available, correlations between PB EOS or total IgE and clinician-/patient-reported outcomes were negligible/low (0.01–0.41), showing that these are poor indicators of AD severity, although total IgE is generally higher in patients with atopic predisposition [1]. TARC and LDH showed moderate correlations with clinician-reported outcomes (Spearman’s correlation coefficients 0.58–0.63) and low/moderate correlations with dermatological patient-reported outcomes (0.44–0.53). Japanese guidelines recognize that serum TARC levels correlate more strongly with disease severity and progression than with the other three biomarkers [1], although we found similar correlations for TARC and LDH. Recent data have suggested that interpretation of the levels of each biomarker depends on the treatments used [15].

Strengths and Limitations

To our knowledge, this is the largest Japanese dataset that has been analyzed for correlations between outcomes among patients with a range of AD severities. It expands upon our previous correlation analysis (which included only data at baseline, when all patients had IGA 3 or 4) [3]. However, there are some limitations: some data were missing for some patients (particularly biomarker data), the number of patients diminished over time, and there were fewer patients with low IGA scores. The low number of patients with biomarker data was due to the observational nature of ADDRESS-J and available data tended to be in patients with more severe disease. Hence, the biomarker correlation results should be interpreted with caution. Also, although we included three clinician-reported outcomes and five patient-reported outcomes, we did not include other scores, e.g., Atopic Dermatitis Control Test or Recap of Atopic Eczema, which have recently been recommended for the evaluation of long-term disease control [11].

Conclusions

The results of this study confirm that clinician- and patient-reported outcomes are not necessarily correlated among Japanese adults with AD. This highlights the importance of including patient-reported outcomes when assessing disease severity/impact, planning treatment, and assessing treatment response.