Abstract
Capsule Summary
Adults with atopic dermatitis commonly described their itch using pain-related sensory descriptors, such as stinging, tingling, burning, and painful.
The number of pain-related itch descriptors was associated with increasing severity of itch and atopic dermatitis.
The presence of pain-related itch descriptors was associated with increased comorbidities, including anxiety, depression, and sleep disturbance, and worse quality of life, independent of the severity of itch.
INTRODUCTION
Atopic dermatitis (AD) is a chronic inflammatory disease that commonly affects children and adults. Pruritus is the universal symptom of AD and the most bothersome symptom to patients. 1 Chronic pruritus leads to an itch-scratch cycle that can further exacerbate the AD and predispose toward cutaneous infections. AD is also associated with other symptoms, such as skin pain,2,3 sleep disturbance,4-6 anxiety and depression,7-9 fatigue, 10 and impaired quality of life (QOL) and wellbeing,1,11 which are largely secondary to uncontrolled pruritus.
There are many different potential triggers of itch in patients with AD. 12 Yet, little is known about the characteristics of itch in AD. Additionally, the impact of itch characteristics on AD severity and QOL are not well known. We hypothesize that patients with certain itch phenotypes are more likely to have severe AD and worse QOL, independent of itch severity. In this prospective study, we examined the association of itch characteristics and patterns of itch with AD severity and outcomes in adults with AD.
METHODS
Study Design
A prospective, dermatology practice-based study of adults was performed with AD as defined by Hanifin-Rajka diagnostic criteria. 13 All adult patients from the eczema clinic at an academic medical center were invited to participate. Exclusion criteria included those without a definite diagnosis of AD or being unwilling or unable to complete assessments. Virtually, all (>99%) patients who were invited agreed to participate. Patients received standard of care follow-up and treatment at time of enrollment.
Self-administered electronic surveys were completed prior to their encounter. Questionnaires were completed in the following order: patient-reported global assessment (PtGA) of AD severity (“Would you describe your AD or eczema as clear, almost clear, mild, moderate, or severe?”), 14 numeric rating scale (NRS) for worst-itch and average-itch in the past 7 days (1 question each, range: 0–10), 15 verbal rating scale (VRS) for worst-itch and average-itch in the past 7 days (1 question each, none/mild/moderate/severe/very severe), 15 frequency of itch, 15 characteristics of itch (biting, burning, crawling, painful, pinprick, sensitive, sharp, shooting, stinging, throbbing, tight, tingling, warm), Scoring AD (SCORAD) visual analog scale for itch and sleep (1 question each, range: 0–10), 16 Patient-Oriented Eczema Measure (7 questions, range: 0–28), 16 Dermatology Life Quality Index (DLQI; 10 questions, range: 0–30),17B18 -20 NRS for sleep disturbance in the past 3 days, Patient Health Questionnaire-9 (PHQ-9; 9 questions, range: 0–27), 21 and history of diagnosis of anxiety or depression.
Patients were assessed with a full body skin examination by a dermatologist (J.I.S.). Validated Investigator Global Assessment scale for AD (0 = clear, 1 = almost clear, 2 = mild, 3 = moderate, 4 = severe), 22 Eczema Area and Severity Index (EASI; 4 signs [erythema, excoriation, swelling, lichenification] on 4 body sites, range: 0–72),23-25 body surface area (BSA; range: 0–100%), 23 and objective component of SCORAD (objective-SCORAD; 6 signs [erythema, excoriation, swelling, oozing/crusting, lichenification, dryness] on 8 body sites, no symptoms; range: 0–83)23B24 -26 were the clinically reported outcomes examined. Gestalt assessment of severity for facial erythema, eyelid dermatitis, hand eczema, and foot eczema were also assessed.
Patients were enrolled sequentially between January 2016 and September 2019. The study was approved by the institutional review board of Northwestern University. Informed consent was obtained electronically.
Statistical Analysis
Descriptive statistics were calculated for baseline characteristics. Continuous variables were compared between groups with Kruskal–Wallis tests. Categorical variables were compared with χ 2 or Fisher’s exact tests. Patient and clinician reported outcome scales were stratified according to available severity classes.14,20,23
Multivariable Poisson regression models evaluated the association of the number of itch characteristics (predictor) with disease severity or QOL impact (outcome), adjusting for age (continuous), sex (binary), and race and ethnicity (categorical). Multivariable logistic regression models were constructed to assess the association of AD severity or itch severity by VRS average-itch (predictors) with endorsement of ≥1 itch characteristic or individual itch descriptors (binary outcomes). Models adjusted for age (continuous), sex (binary), and race and ethnicity (categorical). Additional models assessed the association of ≥1 itch characteristic or individual itch descriptors (predictors) with history of anxiety, history of depression, depression by PHQ-9 score, sleep disturbances, and severe QOL impact by DLQI total score (binary outcomes). Multivariable logistic regression models were constructed to assess the association of dermatitis affecting the face, eyelids, hands, or feet (binary predictors) with itch descriptors (binary outcomes). Models adjusted for age (continuous), sex (binary), race and ethnicity (categorical), and VRS average-itch (categorical). Logistic regression models assessed the association of itch and AD severity (predictors) with sleep disturbances, QOL impact, history of anxiety and depression, and symptoms of depression by PHQ-9 scores (outcomes), adjusting for age (continuous), sex (binary), race and ethnicity (categorical), and endorsement of ≥1 itch descriptor (binary).
Latent class analysis (LCA) was utilized to examine phenotypical patterns of 13 binary itch variables of AD. Conditional probabilities were estimated using maximum likelihood to characterize the latent classes by calculating the odds that a member of the class would have a certain response (yes or no) for the specific item. The number of latent classes and best fitting model were selected through minimization of the Bayesian information criterion and Akaike information criterion along with interpretability.
To examine the association of various outcomes with membership in the latent classes, χ2 tests were conducted. Multivariable logistic regression models investigated the relationship of membership in each latent class (independent variable) with patient- or clinician-reported disease severity (dependent variable), adjusting for age (continuous), sex (binary), and race and ethnicity (categorical). Additional models examined the relationship of latent class membership with mental health, sleep, and QOL outcomes and included age (continuous), sex (binary), race and ethnicity (categorical), objective-SCORAD (continuous), SCORAD-sleep (continuous), and SCORAD-itch (continuous) as covariates to mitigate potential confounding. For binary outcome variables, binary regression models were used. For ordinal outcome variables, binary regression models were used when the Score-test P value was <0.01; otherwise, ordinal regression models were used.
In all regressions, adjusted odds ratios (aORs) or adjusted regression coefficients along with their respective 95% confidence intervals (CIs) were estimated. Complete case analyses were performed. A 2-sided P < 0.05 was considered statistically significant.
RESULTS
Patient Characteristics
Overall, 456 adults (age: range, 18–98; mean ± standard deviation [SD], 46.0 ± 17.9; median [interquartile range (IQR)], 45.7 [30.6] years) were included in the study, of whom 266 (58.3%) were women and 312 (68.4%) self-identified as non-Hispanic White (Table 1). The mean ± SD (median [IQR]) BSA of involvement was 17.5% ± 24.3% (6.2% [22.0%]), NRS worst-itch was 5.1 ± 3.1 (5.0 [6.0]), objective-SCORAD was 25.4 ± 15.2 (24.6 [21.8]), SCORAD was 32.1 ± 18.8 (29.2 [27.0]), EASI was 9.1 ± 11.4 (4.3 [11.3]), and DLQI was 7.6 ± 7.2 (5.0 [9.0]).
Baseline Characteristics Among All Patients and in Analyses Stratified by Number of Itch Descriptors
Bolded values indicate P < 0.05.
Some outcomes were implemented during the study, resulting in different total frequencies for certain measures.
AD, atopic dermatitis; BSA, body surface area; DLQI, Dermatology Life Quality Index; EASI, Eczema Area and Severity Index; IGA, Investigator Global Assessment; NA, not applicable; objective-SCORAD, Objective SCORing AD; PHQ-9, Patient Health Questionnaire-9; PtGA, patient global assessment; QOL, quality of life; SCORAD, SCORing AD; VRS, verbal rating scale.
Characteristics of Itch in AD
At baseline, 283 (62.1%) patients described their itch with ≥1 descriptor, particularly stinging (29.4%), tingling (23.9%), burning (23.3%), painful (23.0%), sensitive (21.3%), crawling (20.6%), warm (14.7%), pinprick (14.0%), tight (12.3%), throbbing (10.5%), sharp (9.4%), biting (8.6%), and, least commonly, shooting (3.7%). The mean ± SD (median [minimum–maximum, IQR]) number of itch characteristics was 2.1 ± 2.7 (1 [0–13, 3]) with 133 (29.2%), 79 (17.3%), and 71 (15.6%) adults reporting 1–2, 3–4, and ≥5 itch descriptors, respectively.
Patients reporting ≥1 itch characteristic were more likely to have a history of food allergy (31.4% vs 20.1%) or hay fever (45.3% vs 33.3%) compared with adults without any itch descriptors (P < 0.05).
Patients who described their itch with ≥1 descriptor had significantly worse AD severity (SCORAD, objective-SCORAD, EASI, IGA, PtGA, BSA), itch severity (VRS worst-itch, VRS average-itch), itch frequency and itch intensity (all P < 0.01). In general, the proportions of patients who endorsed each of these itch characteristics rose with successive stepwise increases in AD severity. In contrast, no consistent patterns of associations were found for each itch descriptor with age, sex, educational attainment, or type of health insurance.
Relationship of Number of Itch Characteristics with AD Severity and QOL
In multivariable Poisson regression models, the number of itch characteristics was associated with AD severity (SCORAD, objective-SCORAD, EASI, IGA, PtGA, BSA), itch severity (VRS worst-itch, VRS average-itch), itch frequency, itch intensity, lesional distribution severity, QOL decrement, history of diagnosed anxiety or depression, current symptoms of depression, and sleep disturbance (Table 2). In general, patients with more severe AD, itch, and QOL impact described their itch with greater numbers of descriptors.
Association of Atopic Dermatitis Severity and Quality of Life Impact with Number of Itch Characteristics
Bolded values indicate P < 0.05.
Multivariable Poisson regression models evaluated the association of disease severity (independent variable) with the number of itch characteristics (dependent variable). Models included age (continuous), sex (binary), and race and ethnicity (categorical) as covariates.
Some outcomes were implemented during the study, resulting in different total frequencies for certain measures.
BSA, body surface area; CI, confidence interval; DLQI, Dermatology Life Quality Index; EASI, Eczema Area and Severity Index; IGA, Investigator Global Assessment; objective-SCORAD, Objective SCORing AD; PHQ-9, Patient Health Questionnaire-9; PtGA, patient global assessment; QOL, quality of life; SCORAD, SCORing AD; VRS, verbal rating scale.
Relationship of Itch Severity with Itch Descriptors
Of 81 (19.2%) patients with severe VRS average-itch scores, 71 (87.7%) endorsed ≥1 of the 13 itch descriptors. In comparison, of 219 (51.9%) patients with clear/mild itch, only 117 (53.4%) endorsed ≥1 of these itch characteristics.
When modeled individually in multivariable models controlling for age, sex, and race and ethnicity, endorsing ≥1 itch descriptor was associated with sleep disturbance (aOR [95% CI]: 3.50 [1.75–6.99]), QOL impact (5.47 [2.55–11.75]), history of diagnosed anxiety (3.96 [1.96–8.02]) and depression (2.15 [1.10–4.20]), and active symptoms of depression (PHQ-9) (2.33 [1.38–3.92]), whereas moderate-severe itch was associated with sleep disturbance (16.17 [8.19–31.92]), QOL impact (10.66 [5.67–20.04]), and active symptoms of depression (PHQ-9) (2.86 [1.81–4.53]) (Table 3). When modeled together in multivariable models, endorsing ≥1 itch descriptor remained significantly associated with sleep disturbance (3.45 [1.57–7.60]), QOL impact (6.37 [2.48–16.35]), history of diagnosed anxiety (3.75 [1.57–8.98]), and active symptoms of depression (2.22 [1.02–4.86]) independent of itch severity.
Impact of Itch Severity and Itch Descriptors on Mental Health, Sleep, and Quality of Life
Bolded values indicate P < 0.05.
Multivariable logistic regression models examined the association of itch severity and itch descriptors (predictors) with outcomes, adjusting for age (continuous), sex (binary), and race and ethnicity (categorical).
Odds ratios were not estimable due to low sample sizes in these subgroups.
CI, confidence interval; NA, not applicable; OR, odds ratio; QOL, quality of life.
Eight of the 13 individual itch descriptors (burning, crawling, sensitive, sharp, shooting, stinging, throbbing, warm) were associated with history of diagnosed anxiety (P < 0.05). Seven itch descriptors (biting, burning, crawling, painful, sharp, shooting, stinging) were associated with history of depression. Twelve itch characteristics (biting, burning, crawling, painful, pinprick, sensitive, sharp, shooting, stinging, throbbing, tight, tingling) were associated with symptoms of depression by PHQ-9. All 13 itch descriptors (biting, burning, crawling, painful, pinprick, sensitive, sharp, shooting, stinging, throbbing, tight, tingling, warm) were associated with sleep disturbance and QOL impact. History of diagnosed anxiety (3.21 [1.87–5.54]) was most strongly associated with itch described as stinging. History of diagnosed depression (3.44 [1.15–10.30]), active symptoms of depression by PHQ-9 (8.61 [1.87–39.64]), and QOL decrement (7.65 [1.84–31.90]) were most strongly associated with itch described as shooting. Sleep impairment (5.15 [2.83–9.35]) was most strongly associated with itch described as painful.
In multivariable models controlling for itch severity and patient demographics, many of the itch descriptors remained significantly associated with active symptoms of depression (PHQ-9), sleep disturbance and QOL impact.
Relationship of Itch Severity and Type with AD Distribution
We examined whether the distribution of AD lesions was associated with itch severity and characteristics. Facial erythema, eyelid dermatitis, hand eczema, and foot eczema were associated with moderate (3.05 [1.73–5.40], 3.52 [1.99–6.23], 3.12 [1.76–5.54], 3.31 [1.85–5.90], respectively) and severe/very severe (5.98 [2.81–12.73], 6.42 [3.04–13.57], 5.43 [2.54–11.60], 5.52 [2.56–11.90], respectively) average-itch (Table 4). The magnitude of the associations increased with rising itch severity.
Association of Atopic Dermatitis Distribution with Itch Severity and Itch Characteristics
Bolded values indicate P < 0.05.
Multivariable logistic regression models assessed the association of dermatitis affecting the face, eyelids, hands, or feet (binary predictors) with itch descriptors (binary outcomes), adjusting for age (continuous), sex (binary), race and ethnicity (categorical), and VRS average-itch (categorical).
CI, confidence interval; OR, odds ratio.
In addition, facial erythema (2.69 [1.63–4.42]), eyelid dermatitis (2.04 [1.15–3.62]), hand eczema (2.45 [1.49–4.02]), and foot eczema (2.50 [1.24–5.07]) were all associated with reporting ≥1 itch characteristic. In particular, facial erythema was associated with burning (3.21 [1.86–5.52]), crawling (1.91 [1.13–3.22]), pinprick (1.87 [1.02–3.41]), stinging (1.64 [1.01–2.66]), throbbing (2.45 [1.17–5.14]), and warm (2.54 [1.37–4.71]) itch. Eyelid dermatitis was associated with burning (2.01 [1.15–3.50]), throbbing (2.25 [1.10–4.59]), tight (1.90 [1.00–3.59]), and warm (1.95 [1.06–3.56]) itch. Hand eczema was associated with biting (2.57 [1.16–5.69]), burning (2.62 [1.54–4.46]), painful (2.77 [1.57–4.90]), sharp (2.87 [1.34–6.17]), and stinging (2.02 [1.25–3.26]) itch. Foot eczema was associated with biting (3.32 [1.49–7.43]), crawling (2.26 [1.25–4.07]), painful (2.33 [1.25–4.34]), sharp (2.85 [1.30–6.25]), stinging (2.00 [1.14–3.50]), and tingling (1.91 [1.09–3.35]) itch.
Patterns of Itch Descriptors
LCA was utilized to characterize patterns of itch associated with AD. The best fitting model contained 3 latent classes. Conditional probabilities of each latent class having specific itch characteristics are plotted in Supplementary Figure S1. Class A (n = 264 [57.9%]) had the lowest probabilities for all itch descriptors. Class B (n = 153 [33.6%]) had intermediate probabilities for burning, crawling, painful, sensitive, stinging, and tingling itch. Class C (n = 39 [8.6%]) had intermediate to high probabilities for every itch characteristic studied.
Latent class membership was associated with AD severity, comorbidities and QOL impact. In multivariable logistic regression models controlling for patient demographics, class C was associated with most severe AD (SCORAD, objective-SCORAD, EASI, IGA, BSA, PtGA) and itch (VRS worst-itch, VRS average-itch, itch intensity, itch frequency), whereas class B was associated with intermediately severe AD and itch (Table 5). Even with adjustments for AD severity, itch, and sleep, class C was associated with active symptoms of clinical depression by PHQ-9 (2.45 [1.04–5.78]) and greatest decrement in QOL (11.86 [4.62–30.44]) followed by class B (2.70 [1.59–4.59]). Similarly, class C remained associated with active symptoms of depression (3.36 [1.11–10.21]) and greatest QOL impact (26.52 [6.07–115.86]) followed by class B (2.97 [1.39–6.35]).
Relationship of Latent Classes with Atopic Dermatitis Severity, Itch Severity, Mental Health, Sleep, and Quality of Life
Bolded values indicate P < 0.05.
Some outcomes were implemented during the study, resulting in different total frequencies for certain measures.
Multivariable
Multivariable
AD, atopic dermatitis; BSA, body surface area; CI, confidence interval; DLQI, Dermatology Life Quality Index; EASI, Eczema Area and Severity Index; IGA, Investigator Global Assessment; objective-SCORAD, Objective SCORing AD; OR, odds ratio; PHQ-9, Patient Health Questionnaire-9; PtGA, patient global assessment; SCORAD, SCORing AD; VRS, verbal rating scale.
DISCUSSION
In this prospective observational study, we found that adult patients with AD describe their itch heterogeneously using various combinations of 13 different itch qualities. Stinging, tingling, burning, and painful were the most common qualities of itch reported. Patients with severe AD and severe itch were most likely to describe their itch as burning, crawling, painful, sensitive, and so forth. With increasing AD and itch severity, patients reported rising numbers of sensory-related perceptions of itch. Of note, even a single individual’s itch was quite heterogeneous; that is, each patient could experience several types of itch, the number of which generally corresponded with disease and itch severity. While approximately half of the patients with mild itch described their itch with one or more of these qualities, nearly all patients with severe itch used these descriptors. Previous studies found that itch severity (ie, NRS itch) is a strong predictor of worse QOL and health outcomes.16,27 We similarly found that itch severity was highly associated with sleep disturbance, QOL impact, anxiety and depression. However, even after controlling for itch severity, we found that patients who described their itch using one or more of these characteristics had significantly higher likelihood of sleep disturbances, worse QOL impact, more frequent diagnosis of anxiety and active symptoms of depression. All 13 of the individual itch characteristics examined were associated with sleep disturbance and QOL impact, even after adjusting for itch severity. However, patients who described their itch as burning, crawling, sharp, stinging, and particularly shooting were considerably more likely to have a diagnosis and symptoms of anxiety and depression. Taken together, the results of this study indicate that the quality of itch may independently be a predictor of poor outcomes in patients with AD.
Patients who described their itch using one or more of the descriptors examined were more likely to have more severe AD, more severe and persistent itch, as well as dermatitis on the face, eyelids, hands and feet. Using LCA, we found three broad patterns of these itch characteristics. Most patients were in the latent class with low probabilities of any of the itch characteristics. The fewest patients were in the latent class with moderate to high probabilities of all the itch characteristics. The remainder was in a latent class with moderate probability of itch being described as burning, painful, stinging, and sensitive. The latter group is particularly interesting because it encompasses a wide range of severity for AD and itch, including two-thirds with more limited BSA (<25%) and many having almost clear or mild disease. Some patients in this group appear to experience itch that is burning, painful, stinging, and sensitive, despite having fairly mild disease. Other important subsets included patients with dermatitis affecting the hands, feet, face, and eyelids. Those with hand and foot dermatitis were more likely to describe their itch as painful, sharp, and stinging, while patients with face and eyelid dermatitis were more likely to describe their itch as throbbing and warm. These itch characteristics or their underlying mechanisms may contribute to poor outcomes in patients with dermatitis affecting these specific anatomical locations.
The underlying mechanisms for the different characteristics of itch are unknown. Itch is a complex symptom with numerous converging immune and neuronal pathways. Itch characteristics may vary due to different profiles of cytokine or neuropeptide expression in skin,28-30 central mechanisms related to the perception of itch and pain,31,32 and behavioral patterns with respect to the itch-scratch cycle and how patients mechanically respond to their itch. Prior studies found elevated density33,34 and length 35 of dermal nerve fibers in patients with AD. In addition, patients with more severe AD and itch may develop neuronal sensitization to various mechanical and chemical stimuli. 36 Chronic peripheral and central neuronal alterations may rewire perception and processing of itch and pain.
Our findings highlight several clinically relevant considerations in the management of AD. First and foremost, clinicians should recognize that itch is complex and central to the presentation of AD. Inquiring about specific patterns of itch may identify patients who are more likely to experience greater AD severity and QOL burden. Second, patients with dermatitis on the hands, feet, face, and eyelids are more likely to describe their itch as painful, sharp, stinging, throbbing, or warm. These itch characteristics may in turn signal worse treatment outcomes. Additionally, patients with dermatitis affecting these regions who describe their itch using these descriptors may experience more sensitivity and application site reactions to personal care products and topical medications. Third, it remains unknown whether the itch phenotypes observed are associated with prognostic or treatment outcomes, which should be assessed in future studies.
Our study has several strengths, including the prospective design and use of more than a dozen validated patient- and clinician-reported assessments of AD signs, symptoms, mental health, sleep and QOL impact. LCA was utilized to identify clinical phenotypes of itch and was able to detect 3 unobservable groups of patients with AD. Multivariable modeling included additional covariates to adjust for potential confounders. However, this study has limitations. Itch characteristics were assessed by self-report, which were subject to recall and misclassification biases. It is possible that patients did not recognize all relevant characteristics of their AD-associated itch. We did not examine the impact of various treatments on itch qualities and did not control for patients with untreated versus treated disease. Given that this study was conducted in a single tertiary care academic setting, there was likely some degree of selection bias, leading to the inclusion of patients with more severe AD. Consequently, these results may not be fully generalizable to the broader AD patient population.
In conclusion, patients with AD describe their itch heterogeneously using multiple pain-related sensory descriptors. In particular, patients with severe AD and itch or dermatitis affecting the face, eyelids, hands and feet were more likely to report these itch characteristics. These results indicate that the quality of itch may independently be a predictor of worse QOL impact, poor sleep and mental health outcomes in patients with AD.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
