Atopic dermatitis (AD) is a common allergic disease that often develops in early childhood, beginning in the first year of life in half of the cases (1). According to the nationwide International Study of Asthma and Allergies in Childhood (ISAAC) survey in Korea, the life time prevalence of physician-diagnosed AD increased from 1995 to 2000 in 6-12 yr olds (16.6% in 1995 and 24.9% in 2000) and in 12-15 yr olds (7.3% in 1995 and 12.8% in 2000) (2). This increasing prevalence of AD is a consistent trend in the developing countries shown by ISAAC Phase One and Phase Three Study Groups (3).

Recently, loss-of-function mutations of the epidermal barrier protein, filaggrin, have been demonstrated to cause skin barrier dysfunction and predisposed to AD (4). These mutations were reported in European Caucasians as well as Asians (5, 6), and may be associated with early onset, severe and persistent AD (7, 8). However, the mutations of filaggrin gene represent a subset of AD and the clinical symptoms in a number of AD patients appear to improve as they grow older with varying rates (9-13).

Understanding natural history of AD would be helpful to predict the prognosis and to establish appropriate management strategy for each patient. Considering the AD is caused by genetic susceptibility, environmental factors or in combination, the natural history might depend upon the pathogenesis, ethnicity, geographical differences and so on. The analysis of this variation requires long-term studies during the course of AD in each region with presumably similar environment or in each group of susceptible individuals. However, there is a lack of reports about the course of AD in Korean children, although several reports are available from other countries (14-16). In this study, we aimed at analyzing the natural history of AD in Korean children occurring in their first year of life as well as the risk factors associated with persistent clinical symptoms.

Among the patients with AD who visited Department of Pediatrics, Samsung Medical Center, Seoul, Korea between September 1995 and March 2006, those who manifested skin symptoms in the first year of life were selected in this study. The age at their first visit to our clinic was less than 36 months, and the diagnosis of AD was based on the criteria of Hanifin and Rajka (17). For this study, their medical records were reviewed and telephone survey using preformed questionnaire was done if the patients did not visit our clinic for more than one year. Out of 834 children 237 patients were excluded because of loss of follow-up after first visit. A total of 597 children who were regularly followed at least once a year at the clinic or by telephone were finally enrolled in this study (Fig. 1).

All children were divided into 3 groups: a group with complete remission, intermittent and persistent AD, by medical records and parental telephone interviews. The "complete remission group" was comprised of children who showed no AD symptoms for over 1 yr without any medication including topical steroid. The "persistent group" included children whose skin symptoms persisted and required medical management at least once a month. The remaining patients whose symptoms were between "complete remission group" and "persistent group" were classified as "intermittent group". In the "complete remission" group, the average healing time of AD was calculated, and the expected healing time of AD, i.e., the expected duration of disease from the disease onset to remission in the case of AD occurring in the first year of life, was also estimated by Kaplan-Meier survival analysis.

Of the 597 children with AD, we analyzed the risk factors of 200 patients whose information was completely obtained by their medical records, parental telephone interviews and blood tests. Those risk factors were gender, family history of allergic diseases, parental AD, feeding patterns during infancy, the age at introduction of solid foods, pet ownership, parental smoking, hospitalization history before 1 yr of age, serum specific IgE level to food allergens before 1 yr of age, and the disease severity at the first examination.

The family history of allergic diseases was determined as positive when at least one family member has been diagnosed with AD, asthma or allergic rhinitis by a physician. Parental AD was considered positive when either mother or father or both have AD. For the feeding patterns during infancy, duration of breast milk feeding (BMF) and/or formula feeding were evaluated. In case of BMF, maternal restriction of allergenic diet during lactation such as egg white, dairy products or peanuts was assessed. The presence of hospitalization history was determined when the patients were admitted due to serious febrile illnesses before 1 yr of age. The level of serum specific IgE against two common food allergens, egg white and cow's milk were measured by ImmunoCAP (Phadia, Uppsala, Sweden), and concentrations above 0.7 kU/L were regarded as positive. The severity of AD was evaluated on the basis of the SASSAD (Six Area, Six Sign Atopic Dermatitis) score (18), and "moderate-to-severe AD" was arbitrarily defined when the disease score at the first visit was over 20.

Kaplan-Meier survival analysis was used to analyze the natural history of AD and the expected healing time of AD. Breslow and Log Rank tests were performed to analyze the risk factors associated with AD remission, and Cox's proportional hazard regression was used for multivariate analysis. The SAS software 9.1.3 program (SAS Institute Inc., Cary, NC, USA) was used for all statistical analysis. A P value < 0.05 was considered significant.

In clinical practice, one of the misunderstandings of the parents of children with AD is that AD is a life-long persistent disease, although this is true only in a small number of patients. As a consequence, they tend to try something specific for cure, instead of maintaining conventional treatment such as bathing, application of moisturizer, avoidance of aggravating factors, and appropriate use of topical corticosteroid. Therefore, reassurance about the natural history of the disease is very important to keep this proper management, as spontaneous remission of AD over time has been observed in a majority of patients (11, 12). The prognosis of AD may be different between the countries, because clinical characteristics could be affected by both genetic and environmental factors. In this regard, our study could be a useful data in Korea since there are few reports about natural history of AD in Korean children.

Illi et al. (11) found that in children with AD occurring in the first 2 yr of life, 43.2% were in complete remission by age 3 yr, 38.3% had an intermittent pattern of disease, and 18.7% had symptoms of AD every year during 7 yr follow-up in their prospective birth cohort. In a long-term, retrospective follow-up study of 252 children with AD occurring before 3 yr of life (12), AD had completely disappeared in 124 cases (60.5%), and a longer period is required in moderate-to-severe AD. Our study showed that 70.6% were in complete remission, 25% had an intermittent pattern of disease, and 4.4% had persistent AD symptoms. This result indicates a higher remission rate and better prognosis of AD, compared with previous studies (11, 12). This might be explained by several factors including difference in genetic predisposition, selection of study population, and duration of follow-up. It has been reported that filaggrin gene (FLG) mutations are associated with persistent and more severe eczema (19, 20). Although FLG null alleles were identified in 8.8% and 12.0% in approximately 7,000 English birth cohort (19) and in the Danish birth cohort (4), respectively, there is no report on the prevalence of FLG mutations in Korea, and might be lower, like in Japanese population (5). In addition, by selection of AD patients whose clinical symptoms occurred in the first year of life, a large number of mild cases with a good prognosis could be included in our study. Furthermore, the length of follow-up might influence on the results. In this study, the total follow-up duration was relatively short, so we might miss the cases of potential relapse or late-onset AD.

Significant predictors for a poor prognosis of AD occurring during the first year of life were found to be severity of disease, atopic sensitization (11, 12). In a population-based cohort study where 2,857 children with 9 to 11 yr of age were followed up to 8 yr prospectively to investigate the course of AD over puberty, high socioeconomic status, female sex, atopic asthma, parental history of allergic diseases and having worked in a high-risk job seem to be significant predictors for the course of disease (21). Our study showed that breast milk feeding, maternal restriction of allergenic food during lactation, pet ownership and severity of AD were associated with prognosis in univariate analysis, but this significance was lost in multivariate analysis. However, we found that prognostic factors are different according to the severity of the disease. In other words, in mild cases no prognostic factors were found, while good prognosis was expected in those who were not sensitized to cow's milk and whose mother restrict ingestion of allergenic food during lactation in moderate-tosevere AD, although the patients whose mother restricted their diet during lactation seem to be included more in our risk factor analysis. Our result is consistent with the previous studies (11, 12) and suggests that food allergy might be associated with the prognosis in Korean children with moderate-to-severe AD. In our study, there was no significant difference in severity, partly because severity was assessed at one time point when the patients were first brought to our clinic. Our study also failed to show that family history of allergic diseases is a predictor for prognosis. There is concern of diagnostic accuracy as data on family history were not confirmed by objective methods such physical examination or laboratory tests.

Some limitations have to be taken into account. First, this study was done in a single institute, so selection bias might occur. In our study, the majority of patients had mild AD, which was related to good prognosis. Second, assessment of related risk factors was relied on questionnaires. Risk factor analysis was not very well controlled, and could be affected by recall bias. Third, the remission state was determined by telephone survey in some patients, not by physical examination. There is a possibility that severe persistent patients were selectively lost, although telephone survey was done in every parent or caretaker who could be contacted. Fourth, we classified our patients into mild AD and moderate-to-severe AD according to the severity at first visit. However, as the AD severity changes from time to time according to the various environmental factors, it is clear that SASSAD at first visit does not represent the AD severity of each patient during the disease course. Lastly, unintended selection bias due to retrospective study design might exist. In other words, we analyzed the risk factors only in 200 patients, not in 597 patients, so missing data of each risk factor might be distributed differently between 200 patients group and 597 patients group.

In conclusion, AD occurring in the first year of life disappears or improves in a significant proportion of children in Korea as they grow older. These data will support the parents or caretakers of infants with AD in reassuring spontaneous regression over time in most cases. In addition, in Korean children with moderate-to-severe AD, maternal diet restriction of allergenic food during lactation and sensitization to cow's milk seem to be predictive of prognosis of AD.