Locations of zip codes inhabited by at least 1 PEER study enrollee.
A Kaplan-Meier curve demonstrating the proportion of individuals with at least 5 years of follow-up at a given age who ever reported a 6-month symptom-free and medication-free period (A) and the proportion of enrollees at a given age who reported no symptoms of AD and medication use in the previous 6 months (B).
Margolis JS, Abuabara K, Bilker W, Hoffstad O, Margolis DJ. Persistence of Mild to Moderate Atopic Dermatitis. JAMA Dermatol. 2014;150(6):593-600. doi:10.1001/jamadermatol.2013.10271
Copyright 2014 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
Atopic dermatitis (AD) is a common illness of childhood.
To evaluate the natural history of AD and determine the persistence of symptoms over time.
Design, Setting, and Participants
A cross-sectional and cohort study of a nation-wide long-term registry of children with AD enrolled in the Pediatric Eczema Elective Registry (PEER).
Main Outcomes and Measures
Self-reported outcome of whether a child’s skin was AD symptom–free for 6 months at 6-month intervals.
A total of 7157 patients were enrolled in the PEER study for a total of 22 550 person-years. At least 2 years of follow-up were observed for 4248 children and at least 5 years of follow-up were observed for 2416 children. Multiple demographic and exposure variables were associated with more persistent AD. At every age (ie, 2-26 years), more than 80% of PEER participants had symptoms of AD and/or were using medication to treat their AD. It was not until age 20 years that 50% of patients had at least 1 lifetime 6-month symptom- and treatment-free period.
Conclusions and Relevance
Based on this large longitudinal cohort study, symptoms associated with AD seem to persist well into the second decade of a child’s life and likely longer. Atopic dermatitis is probably a life-long illness.
Atopic dermatitis (AD) or eczema is a common skin disease that often begins in early childhood.1- 7 It begins before the second year of life, has a waxing and waning clinical course, and is often reported to resolve soon after the first decade of life.2 The etiology of AD is likely multifactorial resulting from a complex interaction between genetic and environmental factors.2- 5 Little has been reported about the natural history of AD and how genetic and environmental factors are associated with its waxing and waning nature.7- 11
About a decade ago, the US Food and Drug Administration (FDA) approved topical calcineurin inhibitors like pimecrolimus and tacrolimus for the treatment of AD. Because of potential safety concerns of the novel topical use of these products, the FDA and the European Medicines Agency required the manufacturers of these drugs to each conduct long-term postmarketing safety studies.12,13 The pimecrolimus postmarketing study is called the Pediatric Eczema Elective Registry (PEER). The PEER study is an ideal resource to examine the natural history of children with mild to moderate AD. The goal of our current study was to evaluate the natural history of AD in a group of children who were enrolled in the PEER study. Specifically, we evaluated the prevalence of symptoms of AD overtime and several risk factors that are associated with the persistence of AD.
The PEER study is an ongoing prospective observational registry and cohort that began enrollment in 2004 and will follow participants for 10 years. The enrollment criteria and goals of the PEER study have been described in detail elsewhere.10,14,15 Briefly, eligible patients were 2 to 17 years of age at the date of enrollment. Additional inclusion criteria included a diagnosis of AD by a treating physician; the application of pimecrolimus cream for at least 42 days of the preceding 180 days prior to enrollment; and written informed consent by the patient or caregiver. The diagnosis of AD for each child was made by the enrolling physicians, most of whom were pediatricians, allergists, or dermatologists.10,14,15 Enrolling physicians had no other role in this study. The physician’s diagnosis was confirmed based on the patient’s self-report and the UK working party criteria.14 Once enrolled, the patient or caregiver completed an enrollment survey as well as 6-month follow-up surveys via the mail. Missing information was obtained via telephone queries. Children were not required to continue to use pimecrolimus once they were enrolled in the PEER study, and many did not.15 Written informed consent was obtained for each of the study participants, and our research protocol was approved by the institutional review board at the University of Pennsylvania.
We investigated the self-reported outcome of whether a child’s skin, without requiring the use of topical medication (eg, steroids or calcineurin inhibitors to treat their AD), was AD symptom–free during the previous 6 months.10 This was determined by the response to the question: “Has the rash cleared completely at any time during the past 6 months?” The patient’s self-described outcome of disease control is a well-validated measure of eczema severity that correlates with Eczema Area and Severity Index (EASI) scores.8,14- 18 Because individuals in this study were followed longitudinally and surveyed every 6 months, this outcome was reported on more than 1 occasion.5 A child’s AD was thought to be intermittent (ie, not persistent) if, in addition to having complete disease control, he or she answered “no” to the treatment use questions. Information about this outcome was collected longitudinally every 6 months by survey, thereby capturing the waxing and waning nature of AD.
The initial enrollment questionnaire contained 39 questions. These included information on the age of onset of AD; the age of enrollment into the PEER study; sex; US census categories for race/ethnicity; annual household income; baseline eczema symptoms, such as the question noted in the previous subsection, age of disease onset; pruritic regions of the body; visits to health professionals; questions on the child’s history of asthma and wheezing, seasonal allergies, food allergies, and other environmental allergies, such as to pets; and questions on family history of AD and other atopic illnesses. Many of the atopic disease–based questions were modeled after the ISAAC (International Study of Asthma and Allergies in Childhood) questionnaire (http://isaac.auckland.ac.nz/).6 Follow-up questionnaires were sent to enrollees every 6 months.
Data were evaluated using 2 different epidemiologic designs. First, using a cross-sectional design we evaluated the demographic information and covariates described herein from the full PEER population at baseline. For these analyses, means with 95% CIs or standard deviations and medians (as appropriate) were estimated. Statistical comparisons were made using t test, analysis of variance, or χ2 tests as appropriate.
Next, we analyzed our data set as a longitudinal cohort study in order to examine “risk factors” for any given 6-month symptom- and treatment-free period. For this analysis statistical, comparisons were assessed using random-intercept logistic regression (XTLOGIT) that allowed for repeated measures of our outcome within each child. For these analyses, we present unadjusted and adjusted odds ratios (ORs) with 95% CIs. All analyses were repeated using a variant of a mixed logistic regression model called generalized linear latent and mixed models (gllamm) with nearly identical results (not reported). The variables used in the adjusted models were selected based on a priori knowledge of their potential association with the persistence of AD and included age at onset, age at enrollment, sex, any family history of atopy, personal history of atopy, and self-reported severity at enrollment. We also estimated a Kaplan-Meier plot to graphically depict the proportion of individuals at any given age who ever reported a 6-month symptom- and treatment-free period for those who had been enrolled for at least 5 years. Finally, we used this cohort to create prognostic models. For this evaluation, we created an outcome based on whether a child had any symptom-free and medication-free period during his or her first 2 years or their first 5 years of observation. In this setting, we used logistic regression and assessed our prediction models using the area under the receiver operating curve (AROC). All statistical analyses were conducted using Stata software (version 12.1; Stata Corp).
At the time of analysis, 7157 patients were enrolled in the PEER study. At least 2 years of follow-up were observed for 4248 children, and at least 5 years of follow-up were observed for 2416 children. Quiz Ref IDThe mean age at the onset of their AD was 1.7 years (95% CI, 1.7-1.8 years), and the average age at enrollment into the PEER study was 7.4 years (95% CI, 7.3-7.5 years). Since their enrollment into the PEER study, a median of 5 surveys (range, 3-9 surveys [25%-75%]) at 6-month intervals were completed per patient, most often by a parent. With respect to those eligible to respond to a survey, approximately 70.0% responded to any given query, and 1.9% withdrew from the study. With slight variation during the study period of observation, 85.3% to 89.6% of children reported that they were seen by a health care provider for a well-child visit at least once per year. Patients were followed for 22 550 person-years of observation. As noted in Table 1, 46.9% of the children enrolled in the PEER study were male, 41.8% were white, and many had other findings of atopic illnesses. Slightly more than half of PEER participants lived in the South Atlantic or the East South Central US Census regions (Table 1). However, at least 1 PEER participant resided in 47 of the 48 contiguous states in the United States (Figure 1). At the time of enrollment, 14.2%, 20.8%, and 37.5% of the children were directly seeking care for their AD from a dermatologist, allergist, or pediatrician, respectively. Quiz Ref IDAt enrollment, self-reported disease activity in the previous 6 months was complete disease control in 5.2%, good disease control in 48.9%, limited disease control in 36.2%, and uncontrolled disease in 9.8%. Self-reported disease activity at baseline was associated with more persistent disease (complete disease control: reference [ie, 1.00], good disease control: OR,0.52 [95% CI, 0.37-0.74]; limited disease control: OR, 0.29 [95% CI, 0.20-0.40]; or uncontrolled disease: OR, 0.18 [95% CI, 0.12-0.27]). At enrollment, all had used topical medications in the past 6 months: pimecrolimus, 100% of patients; tacrolimus, 12.5%; and topical steroids, 60.9%. By the fifth year of follow-up, topical medications used were pimecrolimus, 46.3%; tacrolimus, 16.7%; and topical steroids, 52.9% (medications could be used in combination during the 6-month period). Quiz Ref IDDuring 5 years of follow-up, 64.0% of patients never reported a 6-month period when their skin was symptom free while they were not using topical medications, and less than 1.5% of the study population reported that they had achieved this outcome more than 60% of the time.
We conducted both unadjusted and adjusted analyses (age at onset, sex, race/ethnicity, age at enrollment, and any history of seasonal allergies or asthma) for each question item on the enrollment survey with respect to whether the child achieved the 6-month symptom- and medication-free outcome longitudinally by survey completed. The questions listed in Table 1 were grouped in categories roughly related to basic demographics, atopic and allergic illnesses and responses, environmental exposures, family atopic illnesses, and disease-specific categories. Many of the variables within a category likely have overlapping pathophysiologic characteristics and/or are on the causal pathway with respect to AD symptoms. However, most are associated with a decreased odds of reporting a symptom-free and medication-free state overtime (eg, more persistent AD) (Table 1). Our adjusted and unadjusted effect estimates were often nearly identical, with the exception of the estimate for income, which was primarily confounded by race/ethnicity (or vice versa). A random-effects model using location as a marker of center-based effects did not change our estimates.
Most children and young adults still reported symptoms of AD that required the use of medication. But, as children got older they were more likely to have experienced at least 1 symptom-free and medication-free period. We depict this graphically in 2 ways (Figure 2). Based on a Kaplan-Meier survival estimate of those with at least 5 years of follow-up, by age 20 years about 50% reported at least one 6-month symptom- and medication-free period (Figure 2A). However, at every age, more than 80% of PEER participants had symptoms of AD and/or were using medication to treat their AD (Figure 2B). The odds that a child reported no symptoms of AD while not using medication was increased 30% for every survey answered (OR, 1.30 [95% CI, 1.28-1.31]).
Models were created to predict the likelihood that a child would have persistent disease after up to 2 years and up to 5 years of follow-up; Table 2 contains complex multivariate prognostic models as well as simple models. For example, a prediction model containing information on age at onset, age at enrollment (eg, age of the visit), race/ethnicity (white), any associated atopy (food, asthma, and/or seasonal allergy), AD at the time of examination reported to be under limited control or to be uncontrolled, and a household income greater than $50 000 per year can differentiate between a child with or without persistent AD during his or her first 2 years of care (eg, at least one 6-month period that was symptom- and topical medication–free) 67.3% of the time (AROC = 0.673). A simple count model created based on logistic regression coefficients and several dichotomized variables also had a similar AROC of 0.644. This model is a simple summation of 1 point for household with an income of greater than $50 000 per year, 2 points if the child reported complete or good disease control at the first meeting, age at onset of eczema, and a reduction of 1 point if the child had any other atopic illness and a reduction of one-half his or her age at the visit (ie, a reduction of 2 points for a 4-year-old).
This study of more than 7000 children is one of the largest prospective longitudinal reports of children with mild to moderate AD. Most the children enrolled in this study developed AD before 2 years of age. An advantage of this study was regular 6-month measurements of disease activity. Unlike past teaching, most children enrolled in our study continued to report symptoms and used medications well into the second and even the third decade of life (Figure 2).2,7,8 In our PEER cohort, during the second decade of life most enrollees were very likely to have had at least 1 period when their skin was clear while not requiring medications (Figure 2A). However, this finding did not persist and should not be confused with a “permanent” remission in that at most ages the majority of enrollees had symptoms and were using medications (Figure 2B). Quiz Ref IDPast teaching that nearly 50% to 70% of children with AD will achieve a resolution of their AD by age 12 years was not achieved in our study.2,3,8,11 Interestingly, the persistence of AD symptoms noted in our study is consistent with findings in recent investigations that have shown that in the United States the yearly prevalence of AD or eczema in childhood and adulthood does not differ dramatically, as would be expected if few with childhood AD continue to have AD in adulthood.19 It is also important to realize that many previous studies evaluated the time to when a child was first symptom free, and these studies were not designed to reevaluate patients longitudinally.4,5,9,20,21 In contrast, we have many years of follow-up concerning both symptoms and the use of topical medications.
Furthermore, nearly every environmental exposure we queried at enrollment, as well as history of other atopic illnesses, race/ethnicity, and household income less than $50 000, were associated with more persistent AD. This was also noted with respect to household pet allergies and allergies to common foods. Many of these exposures were also described by Langan et al9 and by others22,23 as factors that were associated with flares of AD. The difference between the outcome used by Langan et al9 and the one we used may be semantic, or, in the extreme case, the outcome measured by Langan et al9 was also included within the outcome we measured.9 Even if our outcomes were different, the study by Langan et al9 and ours both measured longitudinal outcomes related to disease activity. However, it is important to note that our study, as well as that by Langan et al,9 did not evaluate factors that cause incident AD.9 Factors that cause the onset of a disease and contribute to either its persistence or exacerbation are not always the same. This is an important difference because many previous reports on factors that “cause” AD do not consistently attribute all of the factors that Langan et al9 or we report to causation.5,24
A recent study by Ballardini et al8 evaluated a birth cohort in Sweden at ages 1, 2, 4, 8 and 12 years for allergy-related disease. As part of this study they determined the yearly prevalence of eczema at these 5 time points. Childhood eczema prevalence ranged from 15% to 20%, with the maximum prevalence occurring at age 4 years and the minimum prevalence at age 12 years. They also measured “remissions” or periods where a child with previously active AD noted no AD activity in the preceding year. Remissions occurred in 5% to 15% of the population with active AD at a given time point.8 The overall change in AD prevalence between ages 4 and 12 years in the study by Ballardini et al8 was a 25% decrease and in our study was a decrease of approximately 15%. Another birth cohort study of close to 1200 initially healthy children from the Isle of Wight in the United Kingdom evaluated the presence of eczema (defined by an itchy erythematous eruption) in the previous year also showed little variation in the prevalence of eczema in those 1, 4, 10, and 18 years of age.25 However, very few patients from the subcohort of 160 children who had eczema at the study’s inception had persistent eczema, defined as the presence of disease at all time points.25
There are several potential explanations for our findings. It is possible that PEER participants have more severe disease and therefore have more persistent AD. It is also possible that many previous studies lacked sufficiently long follow-up to measure disease persistence and ultimately followed too few with eczema to precisely estimate persistence. Perhaps individuals with AD since childhood are not bothered sufficiently to continue to seek care by adolescence or young adulthood, thereby making it seem to health care providers that a young patient’s AD had resolved. And, it is possible that health care providers focus more on the other atopic illnesses, such as asthma and allergic rhinitis, than the concomitant skin disease. Quiz Ref IDHowever, our report of the lack of resolution of AD lends support to recent findings that demonstrate genetic associations with AD and filaggrin (FLG) protein loss-of-function mutations.26 About 16.2% of a subset of patients in the PEER cohort have these mutations and they do have more persistent AD.10 These mutations result in the inability of an individual to make FLG protein, resulting in skin barrier dysfunction.26 Because these are germline mutations, the ability to produce FLG does not improve with age. Finally, a recent study by Silverberg and Hanifin19 on the prevalence of eczema in US adults revealed that the yearly prevalence is about 10% which when compared with the childhood prevalence rates of about 10% to 15% would seem to argue that childhood AD often persists into adulthood.8,27
We also created prognostic models. Our models had an AROC of about 60% to 65%, indicating that the prognostic ability of these models was fair to good. The prognostic models could be very helpful in 2 settings. First, when planning clinical trials, if the goal is to treat children with more persistent AD, then the models can be used to define who is more likely to have more persistent disease and thereby will modestly improve the likelihood that patients with more persistent disease will be enrolled. The second setting is during the physician’s initial visit with the child and his or her family. Often parents inquire about how long their child will have symptoms of AD. In this case, the models can be helpful in informing the parents that their child will likely have more persistent symptoms of AD.
As with all epidemiologic studies, there are limitations to this report. It is possible that our results do not generalize to all children with AD. Individuals enrolled in the PEER cohort had mild to moderate AD at the time of enrollment that required the use of topical medications (specifically, pimecrolimus) and may have had more severe disease at baseline than those seen in general practice. We were not able to capture incident disease and have only self-reported data on the time of onset. We do not know about disease persistence prior to enrollment, although we do know that the children had persistent disease at the start of the study and if they had it during the study. We do not know why local health care providers decided to use therapy with pimecrolimus or stop using pimecrolimus or any other topical medication, although we know that medications were used, and therefore the disease was classified as persistent. It is also possible that our definition of persistent disease is different from what is used clinically to determine that AD is in remission. It may be possible that those who answered surveys had more persistent disease or were more likely to report their disease as being persistent. However, our definition is based on self-reported eczema severity and is likely a useful measure for following disease activity over time. It might be possible that children in our study were not adequately treated. However, this is highly unlikely in that many were initially enrolled from specialists’ offices (allergists and dermatologists), all patients enrolled were receiving topical therapies, most had at least yearly follow-up with a pediatrician, and the definition of persistent disease included the use of a topical agent. We do not know if their treatment had an effect on AD persistence. Finally, our prognostic models are based on the PEER cohort and, like all prognostic models, need to be validated in other cohorts. However, it is important to remember that this is one of the largest longitudinal studies of children with diagnosed AD, and children were enrolled from nearly every US state.
In conclusion, symptoms associated with AD seem to persist well into the second decade of a child’s life and likely longer. As previously reported, we note that many factors are associated with more persistent disease. While some of these factors may not be associated with the onset of AD, there is at least a belief by PEER participants that they make their symptoms worse. It is important to realize that these statements may not be mutually exclusive. Because of the broad geographic distribution of children enrolled in the PEER study and because they were enrolled from several different types of health care providers’ offices, it is likely that these results generalize to children with mild to moderate AD. Based on our findings, it is probable that AD does not fully resolve in most children with mild to moderate symptoms. Physicians who treat children with mild to moderate AD should tell children and their caregivers that AD is a lifelong illness with periods of waxing and waning skin problems.
Accepted for Publication: December 8, 2013.
Corresponding Author: David J. Margolis, MD, PhD, Departments of Biostatistics and Epidemiology and Dermatology, University of Pennsylvania Perelman School of Medicine, 901 Blockley Hall, 423 Guardian Dr, Philadelphia, PA 19104 (firstname.lastname@example.org).
Published Online: April 2, 2014. doi:10.1001/jamadermatol.2013.10271.
Author Contributions: Mr Margolis and Dr Margolis had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: J. S. Margolis, D. J. Margolis.
Acquisition of data: J. S. Margolis, Hoffstad, D. J. Margolis.
Analysis and interpretation of data: J. S. Margolis, Abuabara, Bilker, D. J. Margolis.
Drafting of the manuscript: J. S. Margolis, D. J. Margolis.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: J. S. Margolis, Bilker, Hoffstad, D. J. Margolis.
Obtained funding: D. J. Margolis.
Administrative, technical, and material support: D. J. Margolis.
Study supervision: D. J. Margolis.
Conflict of Interest Disclosures: The PEER study is funded by a grant from Valeant Pharmaceuticals, a company that makes pimecrolimus, a drug used to treat AD. The company was notified that the study was submitted for review and publication. The PEER study is an FDA-mandated study as part of the FDA approval process. No other disclosures are reported.
Funding/Support: This study was supported in part by R01-AR0056755 from the National Institute of Arthritis Musculoskeletal and Skin Diseases (NIAMS).
Role of the Sponsors: The NIAMS and Valeant Pharmaceuticals had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.