Systematic Review of Epidemiological Studies

The prevalence of atopic eczema has been found to have increased greatly in some parts of the world. Building on a systematic review of global disease trends in asthma, our objective was to study trends in incidence and prevalence of atopic eczema. Disease trends are important for health service planning and for generating hypotheses regarding the aetiology of chronic disorders. We conducted a systematic search for high quality reports of cohort, repeated cross-sectional and routine healthcare database-based studies in seven electronic databases. Studies were required to report on at least two measures of the incidence and/or prevalence of atopic eczema between 1990 and 2010 and needed to use comparable methods at all assessment points. We retrieved 2,464 citations, from which we included 69 reports. Assessing global trends was complicated by the use of a range of outcome measures across studies and possible changes in diagnostic criteria over time. Notwithstanding these difficulties, there was evidence suggesting that the prevalence of atopic eczema was increasing in Africa, eastern Asia, western Europe and parts of northern Europe (i.e. the UK). No clear trends were identified in other regions. There was inadequate study coverage worldwide, particularly for repeated measures of atopic eczema incidence. Further epidemiological work is needed to investigate trends in what is now one of the most common long-term disorders globally. A range of relevant measures of incidence and prevalence, careful use of definitions and description of diagnostic criteria, improved study design, more comprehensive reporting and appropriate interpretation of these data are all essential to ensure that this important field of epidemiological enquiry progresses in a scientifically robust manner.


Introduction
Atopic eczema is a very common inflammatory skin disorder [1]. Its prevalence appears to vary across the world as noted in key international epidemiological studies [2][3][4][5]. Such variation has been found in both children and adults and points to the likely importance of environmental risk factors. In addition, atopic eczema has been shown to cluster in families and there is growing evidence that it is an herald condition in many people who go on to develop allergic problems affecting other organ systems (e.g. food allergy) [6,7]. Genetics are important in the aetiology of atopic eczema: in particular, recent genetic epidemiological studies found a strong association between filaggrin gene defects (present in 1 in 10 Europeans and North Americans), and atopic eczema [7]. Filaggrin plays a role in maintaining the epidermal skin barrier function, whereby it helps to retain moisture in the skin and limits penetration by allergens. These functions can be impaired in filaggrin loss-of-function mutations, this resulting in dry, scaly skin, which increases the risk of allergic sensitisation and disease [7][8][9].
Monitoring disease trends over time aids aetiological understanding and helps with the planning of health services nationally and internationally. Building on our previous work on asthma, we sought to describe international trends in the incidence and prevalence of atopic eczema [10]. We aimed to draw preferentially on high quality studies using appropriate study designs and, in particular, studies using validated instruments [such as the International Study of Asthma and Allergies in Childhood (ISAAC) or the European Community Respiratory Health Survey (ECRHS)] [11,12].

Methods
This review is reported using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement as a guide (see Appendix S1) [13]. The methods for this review were specified in advance and documented in a study protocol.
Our full search strategy is given in Appendix S2. In short, we searched seven electronic databases, namely Medline, CINAHL, Embase, Global Health, Global Health Library, Google Scholar and Web of Knowledge, from 1 January 1990 to 19 May 2010 (date of last search). We used both Medical Subject Headings (MeSH) and free text terms of the following concepts: (atopic eczema OR atopic dermatitis) AND (cohort studies OR crosssectional studies OR ISAAC OR ECRHS) AND (incidence OR prevalence OR trend). The searches were not limited by age, sex, ethnicity or language. Furthermore, bibliographies of key reports were scanned and a citation search was conducted for any additional papers of interest. We only included full-text reports of cohort studies, repeated cross-sectional surveys or analyses of routine healthcare datasets, as we considered these appropriate designs for the assessment of disease trends. Studies were required to present at least two estimates of atopic eczema incidence and/or prevalence within the period 1990 to 2010 and, at each assessment time point, they needed to use a similar approach and instrument (see Table 1). The screening of titles and abstracts and the eligibility assessment of full-text reports was independently performed by two reviewers. Disagreements were resolved by discussion or by a third reviewer if agreement could not be reached. Similarly, to establish the methodological quality of each study, the internal and external validity was examined using the Critical Appraisal Skills Programme (CASP) tool [14] and scored as 'good', 'moderate' or 'poor'. This methodological assessment included for example an appraisal of whether validated instruments were used [i.e. at least one of the ISAAC key questions (see Table 2)]. Reviewers were not masked when assessing study quality. Incidence and/or prevalence data as well as study and participant characteristics were extracted onto a customised data extraction sheet by one reviewer and thoroughly checked by the second reviewer.
To compare disease trends, our primary outcome measure was the lifetime prevalence of symptoms suggestive of atopic eczema or the incidence of atopic eczema (see Table 3). We also collected data on the secondary outcomes, such as the lifetime prevalence of physician-diagnosed eczema or 12-month prevalence measures. There was too much heterogeneity of populations studied and methods employed to undertake meta-analysis.

Results
Our searches retrieved 2,464 titles from which we identified 70 papers that satisfied our inclusion criteria (see Figure 1). We excluded one of these studies because the full-text paper was only available in Korean [15] and we were unable to procure a translation; there were therefore 69 papers in our final dataset. Data from included studies judged to be of moderate or good quality are summarised in Table 4 and explored descriptively by region (see Tables 5,6,7,8 and 9) [16]. Data from the primary outcomes are additionally represented on a map (see Figure 2). Data from studies judged to be at greater risk of bias are available from the corresponding author [17][18][19][20][21][22]. Nearly all studies had prevalence data, while incidence data were only reported in three European studies [23][24][25]. Prevalence data are described using lifetime prevalence of atopic eczema symptoms.

Africa
As presented in Table 4, we found four studies on atopic eczema trends for Africa [26][27][28][29]. Incidence was not measured in any of these studies. Prevalence was measured based on parental-or selfreport as assessed by ISAAC-based questions (see Table 5). Data were mainly from [13][14] year old children and in these children the general trend in Africa (Kenya, Morocco and South Africa) for the prevalence of atopic eczema was increasing [25,27,28] [ 26,27,29]. In these children, an approximate doubling of the lifetime prevalence of atopic eczema symptoms was found for Morocco [e.g. flexural rash in Marrakech, Morocco -from 9.9% (1995) to 20.9% (2001-02)] [26], for South Africa [e.g. flexural rash -from 10.2% (1995) to 16.5% (2002)] [29] and for Kenya [e.g. itchy recurrent rash in flexural areas -from 11.4% (1995) to 19.8% (2001)] [27]. In Nigeria in children of this age group, the lifetime prevalence of itchy rash decreased from a high baseline prevalence [from 26.1% (1995) to 18.0% (2001-02)] [28]. Prevalence estimates in 2001-02 were, however, comparable for all countries. An approximate doubling was also seen in the lifetime prevalence of physician-diagnosed atopic eczema in [13][14] year olds in South Africa and Kenya [27,29]. In contrast, the prevalence of physician-diagnosed atopic eczema in Nigeria considerably decreased over a 5-year period from 1995 to 2001 in 6-7 year olds [from 9.4% to 6.8%] and in 13-14 year olds [from 38.4% to 19.4%] [28]. The baseline estimate for 13-14 year  Table 2. Key question for atopic eczema from the ISAAC questionnaire.
Have you ever had an itchy rash which was coming and going for at least six months?
Have you had this itchy rash at any time in the last 12 months?
Has this itchy rash at any time affected any of the following places: the folds of the elbows, behind the knees, in front of the ankles, under the buttocks, or around the neck, ears or eyes?
At what age did this itchy rash first occur; under 2 years, age 2-4 years or age 5 or more?
Has this rash cleared completely at any time during the last 12 months?
In the last 12 months, how often, on average, have you been kept awake at night by this itchy rash; never in the last 12 months, less than one night per week or one or more nights per week?
Have you ever had eczema? doi:10.1371/journal.pone.0039803.t002 olds was again extremely high. In other African countries, single estimates of atopic eczema prevalence may have been reported, but we were unable to locate any serial data on trends.
In eastern Asia, the general trend for atopic eczema prevalence was mainly increasing across different age groups. Lee et al. (2007) reported an increase in the sex-and age-standardised lifetime prevalence of ISAAC-based parental-report of atopic eczema symptoms in Taiwan [37] and in Japan in a wider-ranged age group 7-15 [from 10.1% (1996) to 13.6% (2006)] [39]. Moreover, baseline prevalences were low, but considerably higher in Korea and Japan, compared to Taiwan and China. In a slightly younger age group 6-12 in Korea, atopic eczema symptoms showed a modest increase from a substantially higher baseline prevalence [from 15.3% (1995) to 17.0% (2000)] [36]. In the youngest children aged 6-7, the prevalence of atopic eczema symptoms was stable in Hong Kong [e.g. chronic rashfrom 5.7% (1995) to 5.4% (2001)] [31], whilst a modest increase was seen in a later study in Taiwan in a similar age group 6-8 [e.g. chronic rash -from 5.8% (2002) to 7.7% (2007)] [34]. Trends in the lifetime prevalence of physician-diagnosed atopic eczema followed nearly the same pattern as the lifetime prevalence of atopic eczema symptoms; trends were increasing in most countries among different age groups with only few exceptions.
In south-eastern Asia, the prevalence of different atopic eczema symptoms showed mixed trends. For chronic rash, the lifetime prevalence was stable in 6-7 year olds in Singapore [10.5% (1994) and 12.5% (2001)] [42] and in north-eastern Thailand [18.0%  [41]. Moreover, this prevalence also remained stable in older children (aged 12-15) in Singapore and, even though the baseline prevalence was appreciably lower, in [13][14] year olds in north-eastern Thailand. For chronic rash with a typical distribution, however, the lifetime prevalence was increasing in Singapore in children of both age groups [e.g. in 6-7 year olds -from 6.1% (1994) to 9.8% (2001)] [43]. In Malaysia and two specific geographical areas in Thailand (Chiang Mai and Bangkok) only data regarding the 12-month prevalence of atopic eczema symptoms were available [40,41]. In Malaysia and Chiang Mai, the 12-month prevalence of atopic eczema symptoms increased in 6-7 year olds, but was stable in 13-14 year olds, whereas the opposite was seen in Bangkok. In western Asia, data were found for Georgia, Kuwait, Turkey and Israel [44][45][46][47][48][49]. In Georgia, the lifetime prevalence of atopic eczema symptoms was found to be decreasing in two different geographical areas among 6-7 year olds: in Tbilisi [from 4.5% (1996) [44]. This trend was additionally apparent in the lifetime prevalence of physician-diagnosed atopic eczema in these children. There was also a decrease in the prevalence of atopic eczema symptoms in 13-14 year old children from Kuwait [from 17.5% (1995-96) to 10.6% (2001-02)], but the baseline prevalence was much higher [48]. In Israel, the lifetime prevalence of itchy rash in a distribution suggestive of atopic eczema was found to be increasing [from 5.9% (1997) to 8.7% (2003)] [49]. In Turkey, two measures of the prevalence of physician-diagnosed atopic eczema were reported. The lifetime prevalence was stable in 6-13 year old children [6.1% (1992)

The Americas
We found no studies on atopic eczema trends for North America, one study for Central America [50] and four studies for South America [51][52][53][54] (see Table 4). No studies reported an incidence trend. The study from Central America, which was conducted in Mexico in 6-8 and 11-14 year old children (see Table 7) [50]. This study showed a sharply decreasing lifetime (and 12-month) prevalence of itchy rash in both age groups [e.g. in 6-8 year olds -from 15.0% (1995) to 7.3% (2002)] and, conversely, a from low baseline increasing lifetime prevalence of physician-diagnosed atopic eczema in both age groups [e.g. in 6-8 year olds -from 3.9% (1995) to 6.1% (2002)].

Europe
The largest set of reports (n = 31) on atopic eczema trends is for Europe. The majority of all trends were increasing, although decreasing and stable trends were found in some areas (see Table 4).
Incidence. Three studies reported on incidence trends in atopic eczema in Europe [23][24][25]. In Denmark, the adjusted cumulative incidence of the UK Working Party-based parentalreport of physician-diagnosed atopic eczema in 7 year olds was 18.9% in 1993 and 19.6% in 1998 (see Table 8). Compared to the survey of 1993 the sample size was over nine times larger in the survey of 1998 [23]. Further, the cumulative incidence of parental-    [24]. Finally, the age-and sexstandardised incidence of physicians' recorded atopic eczema diagnosis as based on secondary analysis of QRESEARCH, a large primary care dataset (n = 333,294) in England, increased from 9.6% (2001) to 13.6% (2005) per 1000 patient-years [25].

Oceania
For Oceania, we found three papers from Australia with prevalence data on atopic eczema trends (see Table 4) [83][84][85]. As shown in Table 10, the lifetime prevalence of atopic eczema symptoms was measured in Melbourne in 6-7 year olds, where it markedly increased from 22.6% in 1993 to 32.3% in 2002 [84]. Two other studies measured trends in lifetime prevalence of atopic eczema diagnosis. In one study this was increasing in 4-6 year olds, even though the baseline prevalence was high [from 31.0% (2000) to 37.0% (2005)] [83] and in another study, using a nonvalidated questionnaire, it was stable in 8-11 year olds [85].

Discussion
The considerable body of international literature identified by this systematic review was heterogeneous in many respects rendering it difficult to directly compare different regions. That said, there was no obvious consistent overall global trend in the incidence or prevalence of atopic eczema symptoms and diagnosis. Nevertheless, in Africa and eastern Asia there was an increasing trend for both the lifetime prevalence of parental-and selfreported atopic eczema symptoms and physician-diagnosed atopic eczema. In western Europe and parts of northern Europe (i.e. the UK), these trends were also mainly increasing. There were extremely diverse trends among different age groups and countries in south-eastern Asia, western Asia and southern Europe. In addition, data for the Americas, eastern Europe and Oceania were limited. The heterogeneous findings in some regions and the limited data available for other regions have precluded conclusions regarding a global atopic eczema trend and atopic eczema trends in major parts of the world.
We found that many outcome measures are used across studies to determine changes in atopic eczema prevalence. Although we found that trends of all outcomes generally pointed in the same direction, we considered the lifetime prevalence of parental-or self-report of atopic eczema symptoms the optimal outcome for the purpose of comparing disease trends between regions within our highly heterogeneous dataset. As atopic eczema occurs in episodes and may be season-related it is particularly difficult to compare studies measuring current or 12-month symptomatology or if patient-and/or study-characteristics, such as age group, do not match. Furthermore, there are marked differences per region in current medical practice, including prevention strategies, national guidelines and physician's awareness of the problem, that make prevalence estimates and trends of physician-diagnosed atopic eczema difficult to compare across the globe. Even though the diagnostic process of a physician is overall likely to be standardised, there is no objective gold standard. This is highlighted in the ENRIECO project which shows that different countries use    Table 9. Good and moderate quality studies reporting the incidence and prevalence of parental-or self-report of atopic eczema between 1990 and 2010 in Europe.           different terms to describe atopic eczema [86]. In addition, not every language has disease labels, nor are they understood in the same way. This means that a diagnostic label may be influenced by region-specific guidelines for the diagnosis of atopic eczema and this may therefore render it difficult to compare estimates of physician-diagnosed atopic eczema prevalence between regions. We thus judged that the lifetime prevalence of atopic eczema symptoms was most likely to prove useful in relation to yielding comparative data on trends in atopic eczema. Strengths and limitations. To our knowledge, no systematic review on international disease trends in the incidence and prevalence of atopic eczema has been published. We searched a large amount of potential relevant literature using seven electronic databases and included 69 papers which reported on trends in atopic eczema. These should represent a good coverage of published literature. Furthermore, we searched systematically, according to a protocol and used stipulated inclusion criteria. To ensure that included studies are above a specific quality threshold, the studies were independently quality-filtered by two reviewers. Where a consensus could not be obtained a third reviewer provided arbitration guidance. In contrast with earlier work into this field, we included all reports on atopic eczema trends, whereas previously papers have limited themselves to single estimates of atopic eczema [4,5], or to original data from the ISAAC study [2,87].

Study
There are gaps in the literature. We could include particularly few reports from the Americas, eastern Europe and Oceania. In general, studies are available on the prevalence of atopic eczema in these regions. However, information from these studies will not be relevant until they are repeated over time. This perhaps somewhat surprising gap for North America is likely to be, at least in part, due to the fact that the ISAAC programme had difficulty identifying a regional coordinator for this region [88]. We were unable to obtain the full-text translation of one Korean paper. Nevertheless, we are reasonably confident that this report or any other additional reports would be unlikely to undermine our overall findings -that there is no clear trend in the worldwide incidence and prevalence of atopic eczema. For nearly all regions information on atopic eczema is questionnaire-based. Questionnaires are non-specific and the measured symptoms suggestive of atopic eczema may overlap with symptoms of other conditions, such as contact dermatitis. The ISAAC questionnaire stipulates the typical distribution and the onset of the itchy rash (see Table 2), which helps to enhance its specificity. At the population level and particularly for the purpose of between-population comparison, ISAAC questions are therefore likely to provide adequate symptom-derived prevalence estimates [89]. That said there is inevitably some loss of ability to differentiate between atopic eczema and other differential diagnoses such as allergic contact dermatitis. This problem may have been more pronounced had we also identified studies using the ECRHS; in the event however, no such relevant studies were found to be eligible.
Future work. Further research in this area should firstly address methodological issues to help inform the optimum design, execution and reporting of future epidemiological studies of trends in atopic eczema. In our dataset various outcome measures were reported and various assessment tools were used, data were analysed differently across studies and results were reported in  different formats (e.g. with and without confidence intervals (CI)), age groups did not match and studies were inconsistently stratified for sex. All of these factors enhance the incomparability of studies. In view of the above, we suggest full and elaborate reporting of the results (including CI) of all of the outcomes obtained. We recommend that the above gaps be addressed using the complete ISAAC tool (and, where possible, also include detailed clinical assessment to allow atopic eczema to be differentiated from allergic contact dermatitis) and be reported according to a standardised format, so that comparisons to other reports on trends are possible. However, even if studies are comparable the prevalence of atopic eczema may still be difficult to compare across countries, without a universal definition. Thus, we need a range of relevant measures of incidence and prevalence as well as a careful description of the diagnostic criteria used together with appropriate interpretation of these data in order to ensure that this important field of epidemiological enquiry progresses in a scientifically robust manner.
Interpretation. Although there is no consistent overall global trend in atopic eczema incidence and prevalence, there are some specific trends which are worth remarking upon further, as they may be of interest for research into the aetiology of atopic eczema. Firstly, there was a stable incidence of atopic eczema in 5-6 year olds in West Germany [12.5% (1991) and 12.8% (1997)] and a sharply increasing incidence in East Germany [from 9.6% (1991) to 23.4% (1997)] [24]. This coincides with the adoption of a ''Western'' lifestyle in East Germany as a result of political change. A consequence of changed lifestyle and increased socio-economic wealth may be an increased frequency of bathing and a greater availability of soaps and bubble baths, which may remove the skin's natural barrier oils and make atopic eczema more prevalent [90]. This is a biologically credible mechanism to explain an increase in incidence -in particular of mild disease. Nevertheless, Schafer et al. (2000) found that, after adjustment for potential confounders, including socio-economic status, the difference in incidence between East and West Germany remained [24]. Other factors, such as nutritional factors, allergens and irritants or infections may therefore be important in the aetiology of atopic eczema. Political changes resulting in, for example, improved access to physicians in East Germany after reunification or due to changes in disease labelling could also have impacted on measures of the incidence of atopic eczema, as has been noted in relation to asthma diagnosis and prescribing [91]. If this were the case, this would reflect an increase in reporting behaviour rather than any true change in the epidemiology of eczema.
Other trends of interest regarding aetiological hypotheses include diverging trends between neighbouring regions. For example, there are marked increases in lifetime prevalence of atopic eczema symptoms in most countries in Africa [e.g. in South Africa -from 15.5% (1995) to 26.2% (2002)] [29], whereas there is a large decrease in Nigeria [from 26.1% (1995) to 18.0% (2001-02)] [27]. This anomalous decrease is most likely a consequence of the extremely high baseline prevalence, as prevalence estimates in 2001-02 are largely comparable for all African countries. Rather than a true prevalence, this high baseline estimate may be a reflection of the presence of another skin condition, such as another rash, perhaps caused by parasites, which are common in these regions. In addition, in our dataset there were also marked baseline differences between neighbouring countries. This is indicated by the low baseline prevalence of atopic eczema symptoms in [12][13][14][15] year olds in Taiwan [2.4% (1995-96)] [32] and the much higher baseline prevalence in Korea [7.2% (1995)] [36]. In these countries too, cultural, social and diagnostic differences may potentially explain this pattern. In contrast, large changes in prevalence estimates within one country in a short space of time are of interest as such changes are likely to represent a true change. For example the doubling in lifetime prevalence of both atopic eczema diagnosis [from 13.9% (1995-96) [80] is likely to represent a true change and we must consider environmental explanations for this.
In conclusion, we have found no overall trend for the incidence or prevalence of atopic eczema worldwide. However, in Africa, eastern Asia, western Europe and parts of northern Europe (i.e. the UK) trends in atopic eczema prevalence were mainly increasing. There are gaps in the literature, particularly in the Americas and Oceania and for measures of atopic eczema incidence. Future research should investigate trends in what is now one of the most prevalent disorders in Europe and other regions in a scientifically robust manner. In order to do so, the careful use of key definitions, improved study design and more comprehensive reporting are essential.