Estimated cumulative probability of atopic dermatitis diagnosis based on Kaplan-Meier survival analysis, using 411 subjects in the Copenhagen Prospective Study on Asthma in Childhood, up to age 3 years. Fifty-five subjects who withdrew from the study before age 3 years were right-censored at the time of withdrawal. Estimates across ages were not significantly different between boys and girls (P = .82).
Atopic dermatitis prevalence rates, by age and sex, up to age 3 years. Subjects were included in calculations until time of withdrawal from the study. Prevalence rates began to decline slightly before age 3 years for both boys and girls.
Affected skin regions at scheduled (6-, 12-, and 18-month) visits compared with visits for acute exacerbations of skin symptoms. The percentages are based on 139 infants with skin lesions by 18 months and atopic dermatitis diagnosis by age 3 years. Ext indicates extensor surface; flex, flexure.
Halkjær LB, Loland L, Buchvald FF, Agner T, Skov L, Strand M, Bisgaard H. Development of Atopic Dermatitis During the First 3 Years of LifeThe Copenhagen Prospective Study on Asthma in Childhood Cohort Study in High-Risk Children. Arch Dermatol. 2006;142(5):561-566. doi:10.1001/archderm.142.5.561
To describe the development of atopic dermatitis (AD) during the first 3 years of life and identify the localization of the early skin lesions that predicts the development of AD.
Prospective, longitudinal, birth cohort study of children born to mothers with a history of asthma, followed up for 3 years with scheduled visits every 6 months as well as visits for onset or acute exacerbations of skin symptoms.
The cohort was recruited from greater Copenhagen, Denmark, and followed up at a clinical research unit, which controlled all diagnoses and treatment of skin diseases.
A total of 411 infants were enrolled in the cohort; 55 had incomplete follow-up and were excluded from certain analyses.
Main Outcome Measures
Atopic dermatitis was defined based on the criteria of Hanifin and Rajka, and severity was assessed by the SCORAD (Scoring Atopic Dermatitis) index. Predictive odds ratios of early skin lesions for those who developed AD vs those who did not were calculated.
The cumulative incidence of AD by age 3 years was 44% (155/356). The prevalence rate peaked at age 2 years for boys and at age 2.5 years for girls, but there were no other sex differences in the proportion of children developing AD. Skin involvement in infants with AD was found to begin at the scalp, forehead, ear, and neck in a balaclava-like pattern and continue to the extensor sides and trunk, finally affecting the flexor sides of the extremities. Early skin lesions of arms and joints best predicted AD at age 3 years.
Atopic dermatitis begins at the scalp, forehead, ear, and neck in a balaclava-like pattern. Eczema at the arms and joints provides the highest predictive value for the development of AD at age 3 years. This may be used for early prediction and intervention of AD.
Atopic dermatitis (AD) is a chronically relapsing inflammatory skin disease, which usually presents in the first years of life1- 3 and is often associated with a family history of atopy.1,4,5 The cumulative incidence of AD has increased in the previous 4 decades, especially in countries with a Western lifestyle.6- 8
The diagnosis of AD is based on a syndrome of clinical criteria in which the major features are pruritus, typical morphologic features and distribution of the lesions, chronic relapsing course, and personal or family history of atopy. The diagnosis is particularly imprecise in early infancy, partly because itching, as one of the cardinal symptoms, is difficult to recognize. The morphologic features are therefore instrumental to the diagnosis at this young age. The predilection sites in the first years of life have been reported to be the head, the trunk, and the extensor surfaces of the extremities,9- 12 although, to our knowledge, the early presentation and progression of eczema lesions have not been described prospectively in detail.
The primary aim of this study was to describe the progressively changing predilection skin lesion pattern of AD during the first 3 years of life in a cohort of high-risk infants. Second, we aimed to identify an early localization pattern of skin lesions that may predict AD within 3 years of life.
The Copenhagen Prospective Study on Asthma in Childhood (COPSAC) is a prospective longitudinal birth cohort study of 411 children (208 female) born of mothers with a history of asthma as previously described in detail.13 The study was conducted in accordance with all applicable regulatory requirements and the guiding principles of the Declaration of Helsinki. The study was approved by the Copenhagen and Frederiksberg Ethics Committee, with informed written consent obtained from the caregivers. In brief, mothers with a history of asthma were invited during pregnancy, and the infants were enrolled during the first month of age. Infants born before 36 gestational weeks or with known severe congenital diseases were excluded. The main catchment area of the cohort was greater Copenhagen, Denmark. The children were born between August 1998 and December 2001, and 397 were white. All mothers had a history of asthma and 48%, a history of AD. Of the fathers, 16% had a history of asthma and 13%, a history of AD.
The participants were seen in the COPSAC Clinical Research Unit (CRU) at age 1 month and at scheduled visits every 6 months thereafter as well as for any acute onset or exacerbation of symptoms from skin or airways. Skin examinations, diagnoses, and treatment of eczema were handled solely by trained medical physicians employed for this purpose in the CRU under the supervision of dermatologists (ie, the population was not treated by the family practitioner). At each visit to the CRU, a history of the symptoms and medication used in the past 6 months were recorded in an online database according to predefined questions and response categories.
Skin lesions with typical dermatitis morphologic features were described online according to predefined morphologic categories and localization. The skin morphologic categories of lesions were registered as erythema, edema, excoriations, lichenification, crusts, dryness, vesicles, and squamatization and cracks on a scale from 0 to 3 (none, mild, moderate, and severe). The localization of the lesions was characterized in 35 predefined skin areas covering the whole body. Skin lesions caused by chicken pox, human parvovirus B19, Gianotti-Chrosti syndrome, herpes, seborrheic dermatitis capitis, and virus exanthema were excluded from this recording. Seborrheic dermatitis capitis was defined as greasy scaling solely in the scalp with no other skin symptoms.
Atopic dermatitis was defined based on the criteria of Hanifin and Rajka.14 This diagnosis requires the presence of 3 of 4 major criteria and at least 3 of 23 minor signs. Major criteria are defined as (1) pruritus; (2) typical morphologic features and distribution (erythema with vesicles/papules and/or scaling/squamatization in a minimum of 2 regions); (3) chronic dermatitis (physician-verified dermatitis for a minimum of 6 weeks or recurrent dermatitis within 6 months); and (4) atopic history (all mothers included had a history of asthma). Of the minor signs in the criteria, the following 4 were excluded: keratoconus and anterior subcapsular cataracts (because they required identification by an ophthalmologist), delayed blanch (because it required an injection of methacholine), and impaired cell-mediated immunity.
The severity of AD was scored using the SCORAD (Scoring Atopic Dermatitis) index, ranging from 0 to 103 points,15 which includes the assessment of (1) extent, (2) intensity (erythema, edema/papules, oozing/crust, excoriation, lichenification, and dryness on a scale from 0 to 3), and (3) subjective symptoms (pruritus and sleeplessness [using a scale from 0-10 of the parents' subjective opinion of the child's degree of pruritus and sleeplessness]). The severity of AD was subsequently categorized into mild (<15 SCORAD points), moderate (15-40 SCORAD points), and severe (>40 SCORAD points) according to the objective components of the index (clinical signs and disease extent), ranging from 0 to 83 points16 (ie, excluding subjective components pruritus and sleeplessness from the SCORAD index).
Standard operation procedures at the CRU outlined treatment recommendations as skin moisturizers and topical corticosteroids, mainly mild (hydrocortisone) for eczema located to the head and genitals and mid-strength (hydrocortisone butyrate) for eczema located at the body and extremities. The amount of treatment used was quantified as number, amount, and duration of treatment courses.
Data were entered online into the cohort database. The central tendency and dispersion are reported as the arithmetic mean ± SD.
Kaplan-Meier survival analysis methods were used to compute cumulative probabilities of AD diagnosis by age up to 3 years. Those censored in this analysis were children not completing the study to age 3 years (censored at time of withdrawal) and those completing visits through age 3 years without an AD diagnosis (censored at age 3 years). This analysis was stratified by sex, and the log-rank test was used to compare probabilities between sex across age.
The odds of AD diagnosis by age 3 years were computed for those with and without a skin lesion observed at the 18-month visit for each of 10 main body regions. Sample odds ratios (ORs) were calculated, including 95% confidence intervals (CIs) based on asymptotic theory. For this analysis, subjects were restricted to those who had had a skin lesion in at least 1 body region by the 18-month visit and completed the study through age 3 years.
Fisher exact and χ2 tests were used to compare proportions. P<.05 was considered significant.
The COPSAC study enrolled 411 neonates at age 1 month, 356 of whom provided complete follow-up data by age 3 years. The earliest sign of dermatitis leading to the AD diagnosis was recorded at age 1 month, with the highest incidence rate occurring during the second half-year of life. For those with complete follow-up, the cumulative incidence of AD was 31% (109/356) at age 1 year, 41% (147/356) at age 2 years, and 44% (155/356) at age 3 years (Figure 1). The prevalence of AD peaked at age 2 years for boys and at age 2.5 years for girls (Figure 2). By age 18 months, the cumulative incidence of seborrheic dermatitis capitis was 24% in children with AD, 28% in children with skin lesions but not fulfilling AD diagnosis, and 21% in children without other dermatitis skin lesions. This mitigates against seborrheic dermatitis capitis being mistakenly diagnosed as AD.
Most infants presented with mild AD, and only single cases presented with severe AD. The severity of AD declined with age, with an increased fraction of mild cases and a reduced fraction of moderate severity and no obvious sex difference (Table 1). Topical corticosteroid (mild, mid-strength, and potent) was prescribed in 7.6 (2.9, 4.5, and 0.2, respectively) courses per child during the first 3 years of life, with a mean ± SD of 15 ± 14 days per treatment period. The non-AD lesions group received a total of 4 courses (mean, 16.3 days). Systemic corticosteroid courses were given for respiratory symptoms on 8 occasions in 7 children with AD.
For those who completed 3 years of the study, the cumulative incidence of AD was 43% (78/183) for girls and 45% (77/173) for boys (P = .70 for difference). Kaplan-Meier estimates of the cumulative probability of AD diagnosis by age 3 years were slightly lower (40% for both boys and girls). However, these estimates incorporated all subjects in the cohort until they dropped out of the study. Differences in cumulative AD probabilities across age (0-3 years) between boys and girls were not significant (P = .80). Girls and boys did not differ in AD severity measured by the SCORAD index (objective components) (Figure 1 and Table 1).
Skin involvement in infants who developed AD started at the scalp, forehead, ear, neck, and cheek and later spread to the extensor side of the extremities and the rest of the face and trunk, finally affecting the flexor sides of the extremities (Table 2). The most commonly involved regions in children with AD seen at the scheduled visits were the cheeks, the flexures of the knee, and the chin, occurring in at least 1 in 4 infants, whereas the eye area, palm of hand, foot sole, and nose regions were very rarely affected. During the visits for acute exacerbations, the forehead, ear, neck, and eye area were more commonly affected (Figure 3).
Table 3 shows the pattern of skin lesions that may predict AD. Of the 356 subjects who completed 3 years of the study, 139 (39%) had had a skin lesion in at least 1 body region by the 18-month visit and AD by age 3 years; 100 (28%) had a skin lesion by the 18-month visit but no AD diagnosis by age 3 years; and 117 (33%) had not had any skin lesions at any visit during these first 18 months of life. The risk of AD at age 3 years was estimated for 10 body regions, which were aggregated based on anatomy and the progression seen in the children with AD (Table 2). Early lesions presenting on the arms and joints showed a strong correlation to later development of AD (OR, 7.5-11.8; maximum 95% CI, 2.7-50.9), whereas diaper area demonstrated no correlation to development of AD (OR, 1.9; 95% CI, 0.9-4.1). The cheeks were affected in almost 80% of children with AD, but cheek involvement was likewise observed in more than 40% of children without AD, indicating that skin lesions on the cheek are not specific for AD. Lesions in the children without AD were itchy, eczematous, and indistinguishable from those in the children with AD.
The COPSAC prospective cohort study of high-risk infants provides new and precise information on the progression of AD during the first 3 years of life. This gives new insight into early prediction of AD and improves the ability to diagnose and intervene early.
The first sign of AD was diagnosed from age 1 month. The highest dermatitis incidence rate was found during the second half-year of life, with a declining incidence rate thereafter (Figure 1). Such early onset concurs with previous reports.1,17,18
Using survival analysis methods, the cumulative incidence of AD was 40% by 3 years of life in this cohort study of children born to mothers with a history of asthma. A recent study based on posted questionnaires in the same geographic region reported that 38% of children with single parental atopic history developed AD by age 4 years.5
We used the diagnostic criteria defined by Hanifin and Rajka14 to identify children with AD. This classification has been a matter of debate, especially with respect to the significance of some minor features in younger children.19- 23 A recent study24 has demonstrated good agreement between Hanifin-Rajka criteria and other criteria used to diagnose AD at this young age.
The specificity of the AD diagnosis is probably high in our cohort because the diagnosis, detailed phenotyping, and management of skin lesions was controlled solely by the CRU physicians. This reduces the risk of misclassification. This reduced risk of misclassification is of particular importance in the clinical evaluation of AD, for which interobserver variation is a problem.25
The cumulative incidence of seborrheic dermatitis capitis was not significantly different in children with AD and in children with skin lesions not fulfilling AD diagnosis. This suggests that misclassification between AD and seborrheic dermatitis capitis was not a major confounder. The sensitivity of the AD diagnoses in the present study was recently supported by comparing the COSPAC database with the records of the family practitioner, which revealed no diagnoses of AD that was not recorded in the COPSAC database.
The severity of the disease was mild to moderate in most cases (Table 1). This is in line with a recent study of a community birth cohort in the same region, which showed that the mild cases constituted 45%.18 Data from other epidemiological studies have equally shown that most AD cases are characterized as mild and moderate and only 1% to 2% as severe.26- 30 Even if the present cohort is at high risk for development of AD, the severity of symptoms should not necessarily be expected to be increased compared with children with AD from a nonrisk population.
Use of topical corticosteroids in the study was almost limited to the AD group. Registration of treatment courses and number of days was performed carefully by the parents, which in this study was preferred to attempts of quantitative measurement of use of topical corticosteroids.
In this prospective cohort study, there was no difference in cumulated incidence between sexes, although boys tended to peak 0.5 years earlier than girls. Nor did boys have more severe disease. Previous studies have provided conflicting evidence with respect to the importance of sex in the incidence of AD in young children. No sex effect on AD was found in children younger than 3.5 years who were studied in a retrospective population-based postal questionnaire study in the United Kingdom31 and in children younger than 2 years who were studied in a prospective population-based case-control study in Sweden.18 In contrast, Harris et al32 found boys to be more affected than girls in a cohort of children recruited before birth and followed up to age 2 years.
Progression of skin region involvement in infants who develop skin lesions before age 18 months and AD by age 3 years was found to begin at the scalp, forehead, ear, and neck in a balaclava-like pattern and continue to the extensor sides of the extremities, rest of the face, and trunk, finally affecting the flexor sides of the extremities. Involvement of the palm of the hand and foot sole was seldom seen and the latest to occur (Table 2).
Cheek, chin, and flexures of the knee were found to be the most commonly affected skin regions in children who developed AD (Figure 3). Exacerbation of AD, as evaluated at visits for acute exacerbations, had almost similar skin region involvement as AD at scheduled visits (Figure 3), but the presence of skin lesions at the eye area, forehead, ear, scalp, and neck proportionally led most frequently to visits for acute exacerbations. Possible explanations for this may be that these skin regions attract more attention from the parents compared with lesions on the extremities and trunk and also that the forehead, ear, scalp, and neck front were the first areas to be affected and therefore caused more concern leading to the first visit for acute exacerbation (Table 2).
Comparing skin lesions during the first 1.5 years of life between children who eventually did or did not become diagnosed as having AD at age 3 years showed that the presence of eczema at the arms and around the joints (Table 3) had the highest predictive value for AD. A skin lesion in the diaper region is generally considered not to represent AD, and in this study, diaper area involvement was confirmed not to predict AD (OR, 1.9; 95% CI, 0.9-4.1) (Table 3). While the cheeks were the most commonly involved region in children who later developed AD, involvement of this region was also common in children who did not develop AD. These observations may be useful for early prediction of AD.
The COPSAC cohort is a birth cohort of children at high risk for the development of atopic diseases including AD. This selection restricts how the findings may be generalized, and validation in unselected populations may be needed. Adherence to the study program was good, with a high retention rate. The COPSAC study provides prospective, longitudinal, follow-up of phenotyping and severity of AD in early infancy. The most important distinguishing factor of the COPSAC study was the emphasis on regular objective skin lesion phenotyping and control of AD diagnosis and treatment solely by the CRU.
In conclusion, the present study shows that the progression of skin region involvement in infants who develop AD begins at the scalp, forehead, ear, and neck in a balaclava-like pattern. A high predictive OR was found for eczema at the arms and around the joints, while the predictive OR was lower for involvement of the head and neck and the predictive OR for involvement of the diaper region was not significant. This improved description of the progression of skin lesions facilitates early diagnosis of AD in infancy and allows studies examining early intervention and prevention strategies.
Correspondence: Hans Bisgaard, MD, DMSci, Danish Pediatric Asthma Centre, Department of Pediatrics, Copenhagen University Hospital, Gentofte, Niels Andersens Vej 65, DK-2900 Hellerup, Denmark (Bisgaard@copsac.dk).
Financial Disclosure: Dr Bisgaard has been a consultant to, paid for lectures by, and holds sponsored grants from Aerocrine, AstraZeneca, Altana, GlaxoSmithKline, MedImmune, and Merck. He does not hold stock or options in any pharmaceutical company in the respiratory field.
Accepted for Publication: August 30, 2005.
Author Contributions:Study concept and design: Bisgaard. Acquisition of data: Brydensholt Halkjær, Loland, and Buchvald. Analysis and interpretation of data: Brydensholt Halkjær, Agner, Skov, Strand, and Bisgaard. Drafting of the manuscript: Brydensholt Halkjær and Bisgaard. Critical revision of the manuscript for important intellectual content: Brydensholt Halkjær, Agner, Skov, and Bisgaard. Statistical analysis: Strand. Obtained funding: Brydensholt Halkjær, Agner, and Bisgaard. Administrative, technical, and material support: Bisgaard. Study supervision: Agner, Skov, and Bisgaard.
Funding/Support: The COPSAC cohort study is sponsored by the following funds: the Pharmacy Foundation of 1991; the Lundbeck Foundation; Ronald McDonald House Charities; the Danish Medical Research Council; The Danish Pediatric Asthma Center; The Danish Research Center of Allergy; Direktør, Cand Pharm K Gad Andersen og Hustrus Familiefond; Aage Bangs Fond; Danish Lung Association; Kai Lange og Gunhild Kai Langes Fond; Direktør Ib Henriksens Fond; Gerda og Aage Hensch's Fond; Rosalie Petersens Fond; Hans og Nora Buchards Fond; Dagmar Marshalls Fond; Foundation of Queen Louise’ Children Hospital; the Danish Hospital Foundation for Medical Research, Region of Copenhagen, the Faroe Island, and Greenland; Gangsted Fond; Højmosegård-Legatet; Fonden til Lægevidenskabens Fremme; A. P. Møller og Hustru Chastine McKinney Møllers Fond til almene Formaal; The Danish Ministry of the Interior and Health's Research Centre for Environmental Health; AstraZenaca; LEOpharma; Yamanouchi Pharma; and Pharmacia-Pfizer.
Acknowledgment: We thank all the families participating in the COPSAC cohort and members of the COPSAC Clinical Research Unit.