The causes of asthma and other allergic diseases remain unknown. Genetic and epidemiological studies suggest that for these multifactorial diseases the expression of different phenotypes depend on complex interactions between susceptibility genes and the environment [1]. This is reflected in the wide differences in prevalence and natural history of allergic diseases around the world [2]. In many regions of Latin America asthma is a public health problem affecting children and adolescents in urban areas; wheezing, asthma and allergic rhinitis are very frequent in some regions [3], with rates similar or even higher than in industrialized countries [4, 5]. Interestingly, anticipated protective factors, such as low hygienic conditions, do not confer protection in poor and overcrowded communities, where a high prevalence of asthma and early infections concur [6–8]. In addition, there are disparities among some phenotypes, such as allergic sensitization and prevalence of atopic eczema, when compared to those observed in industrialized countries [9, 10]. In urban zones of Colombia, asthma is the most common chronic disease in children and IgE sensitization to mites is a hallmark in most patients [11–13]. The population of Cartagena, Colombia, has been previously studied to investigate genetic and environmental risk factors for asthma and allergy, not only because asthma is common [14] but particularly because the city is in a tropical region where a warm and humid environment, facilitate the growth of a diverse mite fauna and the perennial exposure to high concentrations of their allergens [15–17]. Moreover, most of the population is poor and exposed to parasites, generating an interesting setting to study the influence of environmental factors on the susceptibility to allergic diseases (Figure 1).

There are few Latin American birth-cohort focused on allergic diseases [18–20] and they have explored several phenotypes and risk factors using different study designs. For example, Lopez et al. detected sensitization to D. pteronyssinus in 30% of wheezing and 11% of asymptomatic infants at 12 months of age in a prospective study (n = 102) in Brazil. They also found a weak association between wheezing and specific IgE to mites at 12 months. Environmental exposures and socioeconomic status were not evaluated [18]. Rullo et al. studied the role of respiratory infections, exposure to mouse allergens and breastfeeding on wheezing in 104 children living in a socially deprived community of Brazil. Analysis at 30 months showed strong association of wheezing with "respiratory infections requiring antibiotics". No association was found with endotoxin exposure or mite sensitization [19]. Cooper et al. in a study protocol presented the strategies for investigating the impact of early life exposure to geohelminth infections on the development of vaccine immunity, allergic sensitization and allergic inflammatory diseases in 2,403 neonates followed up to 8 years of age [20]. Our study population is an urban low-income community of admixed genetic background [21, 22], living in the tropics under limited sanitary conditions and exposed to mites and helminth allergens. We hypothesize that, for children growing up under these particular genetic and environmental conditions, the prevalence of some allergic phenotypes, as well as the nature and effects of risk factors are different to those found in cohorts from industrialized countries.

The aims of this study were: 1. To create a community-based birth cohort study for analyzing the effects of allergen exposure, early parasite infections and poor living conditions on the inception of allergic diseases, specially asthma; 2. To evaluate the effects of prenatal and other risk factors on the prevalence of wheezing and eczema and 3. To prospectively collect biological samples of children living in poor neighborhoods of a tropical city for further immunological testing, molecular genetics and molecular microbiology screenings. Here we describe the study protocol, baseline characteristics, demographical observations and risk factors for wheezing up to 24 months, of the "Risk Factors for Asthma and Allergy in the Tropics" (FRAAT) study.

Birth cohort studies in different environments and populations are useful to understand the natural history of allergic diseases. Nowadays there are a number of such studies, most of them multicentric, with some common objectives and including a wide range of participants [33–36]. The results, mainly from Europe and US, have shown the importance of particular and common risk factors and their differential influences on allergic diseases, supporting the relevance of doing this type of investigations worldwide, to evaluate the great diversity of settings associated, not only with the climate, geographical location and genetic background, but also with low socio-economic conditions that generate particular risk factors and limit the access to health services. The prevalence of allergic diseases varies among regions and in Latin America, figures of asthma and other allergies in children are very high [37–42]. However, as mentioned, prospective birth cohorts in the region are still scarce, although a growing number are being reported [18–20].

A remarkable finding in our study is the high prevalence of wheezing ever and recurrent wheezing, higher than reported in some European cohorts [43–46], similar to data from the USA [47, 48] and lower than those found in some Latin American surveys [19, 41]. Considering that 12% of children of this cohort attended a hospital with bronchiolitis, one explanation of our findings is the effect of viral diseases, as has been shown in other studies from Latin America [19]. In fact, bronchiolitis was a high risk factor for wheezing in our cohort, with odds ratio similar to that found when laboratory diagnosis of viral infection was done [49]. Lower respiratory tract infections by Respiratory Syncytial Virus (RSV) are common in tropical developing countries causing 27 to 96% of all acute wheezing hospitalizations in children under 6 month of age [50]. In our children an important proportion of wheezing during this period might be caused by viruses and this related to the finding that only half of children wheezing before 6 months continued wheezing until 24 months. However, the existence of a group with recurrent wheezing (14.2%) suggests that other factors, including atopy, may be influencing this phenotype [18]. RSV infections can induce a change of the Th1/Th2 balance, expressed by an increase of the IL-4/IFN-gamma ratio and a persistent IgE response over the years [51, 52], and human rhinovirus-induced bronchiolitis carries a markedly risk of persistent wheezing and childhood asthma [53]. Besides, in the tropics, co-exposure to perennial high concentrations of mite allergens and helminth infections may induce an early allergic response with respiratory effects [54]. Also, living in socially deprived urban areas of underdeveloped countries lead to an early exposure to high levels of endotoxin, in turn associated with increased risk of wheezing during the first year of life [55, 56]. Allergy diagnosis by in vitro assays and skin prick tests will help to improve phenotype definition in our cohort and obtain data about risk factors for atopy associated-wheezing [57].

In this study, maternal asthma was consistently associated with wheezing and recurrent wheezing in the offspring, a finding already observed in different populations and considered an important risk factor for asthma in children before 5 years [58–61]. The mechanisms of this "maternal effect" and why it is more evident during early childhood are unknown. It has been reported that maternal asthma, as well as the child HLA-G genotype contribute to the presence of bronchial hyperreactivity [62]; in addition, several factors may act in combination with maternal cytokines, antibodies and other mediators that act in utero or during the perinatal period [63, 64]. Furthermore, recent works in animals suggest that epigenetic mechanisms may play a role [65, 66], although the transmission of the maternal susceptibility via epigenetic changes or parental imprinting remains to be demonstrated. Remarkably, this association was detected at 12, from 13 to 24 and 0 to 24 months, suggesting that the maternal effect is on children that start wheezing after 6 months. Moreover, it was also evident with recurrent wheezing, a more severe phenotype and always held up after adjustments for covariates. These results are in agreement with a study in Brazil, were parental asthma was associated with atopic and non-atopic wheezing [67] but in contrast with other where maternal atopy was a risk factor for wheezing in poor-environments [68] because we found no association between maternal sensitization alone and wheezing.

Additionally, and in concordance with other surveys [69–71], we found that male gender was associated with wheezing in cases between 0 and 6 months, but the effect disappeared with increasing age. As it has also been reported, the prevalence and severity of asthma are higher in boys than in girls in the first 10 years of life, but this pattern is reversed after puberty [72]. Although the reasons of this observation remain unknown, there are interesting analysis and hypothesis [73, 74].

Contrary to expected, our data show no association of wheezing with active or passive maternal smoking, and the low prevalence of self-reported maternal smoking during pregnancy precluded further association analysis. Therefore, we investigated association between passive intra-domiciliary exposure and wheezing at 6, 12 and 24 months but the results were the same. The prevalence of wheezers was always higher in the group of children with prenatal passive exposure: 6 months (41.1% vs. 50.0%), 12 months (40.4% vs. 47.1%), and 24 months (40.0% vs. 45.2%) but this was not statistically significant. In addition to the type of housing and cultural patterns, another explanation for these results may be a Type II error because the sample is underpowered to detect the effects of smoking. Increasing the sample size and using biological markers of tobacco exposure could help to obtain a more accurate analysis.

The mechanisms of atopic dermatitis are not clear but hereditary factors and environmental exposures are supposed to play important roles [75, 76]. A main finding of this study is that in a socially deprived population of the tropics, the natural history of atopic conditions seems to be different to expected according to the "atopic march", where eczema appears before the first two years followed by allergic respiratory symptoms [77, 78]. In fact, we found no cases of atopic eczema during the first two years of age, while wheezing was a frequent phenotype. Although we do not know yet if atopy is playing a role in the origin of wheezing in these children, it is clear that early respiratory symptoms predominate over cutaneous, atopy-associated, symptoms. In addition, asthma and allergic rhinitis were the most common allergic manifestations in mothers but there was no eczema among them. In a previous study we found a low frequency of atopic dermatitis in the general population of Colombia [9]. In developed countries, the prevalence of this disease is around 10 to 20% in children under 5 years [79, 80], but the ISAAC Phase III study in Latin America reports that, with some exceptions, the prevalence is less than 5% [81]. Under these rates we expected around 15 cases of eczema in our cohort.

Increasing evidence suggest that the "atopic march" includes only a subgroup of patients [82–84]. However, the reasons we found no cases of atopic eczema should be further analyzed. We hypothesized that the genetic background, with an important component of African ancestry and a different distribution of susceptibility alleles for eczema could be protective. Also, particular exposures of these children, with early infections as a hallmark, may confer protection, as has been suggested [56, 85]. The diagnostic criteria we used [30] are very important when aiming to study this disease in the tropics because infections and other medical conditions producing "itchy skin" are very frequent. Although cases of eczema may appear in our cohort after 2 years, it has been suggested that a great number begin within the first 12 months, disappearing around 3 years in a significant proportion of children [86]. Therefore, monitoring symptoms of this disorder, as well as other allergic diseases in this cohort is mandatory.

There are reports of associations between low income and asthma [87]. We found no relationship of allergy symptoms with socioeconomic strata, possibly because the study population is homogeneously poor and differences between strata are small. Potential risk factors related to poverty, such as housing conditions, access to tap water and sewage system, exposure to fume from firewood and trash burning, intradomiciliary contact with poultry and pigs and large family size were not associated with wheezing. For most of them the prevalence of exposure among non-affected was high enough to detect an effect. Since the study population is not representative of all the social strata of the city the generalization of these results is limited. Further research is needed to evaluate the potential interactions that lead to allergy phenotypes in socially deprived environments.

Since our study was based in a prospective cohort design that included physician assessment of a broad range of factors, we think that accurate information was obtained. This strategy may reduce the number of participants and requires more resources but it is necessary to increase accuracy avoiding the bias from poorly defined phenotypes, as may occur when the information is obtained only from self-administered questionnaires, especially in populations with low education level. This is especially important for atopic eczema and wheezing because they may not be easily evaluated by parents. Two sources of ascertainment bias should be mentioned: one from the type of housing and living conditions that make it difficult to identify some exposures (e.g. pet ownership, smoke exposure) and other from the detection of phenotypes. The diagnosis of wheezing was made only when patients were evaluated by a physician or had documented history of medications and hospital admissions, as has been recommended [88] and this might bias our attention to the most severe cases, lowering the sensitivity of the survey. However, considering that the biologic samples are intended for serologic and molecular screenings we chose for well-defined phenotypes to reduce heterogeneity. Among the limitations of this study, one is the small number of children recruited, if compared to multicenter surveys. Although, according to the formerly reported prevalence of wheezing in this population our analyses have good power to detect some associations, increasing the number of participants could provide better information. However, many obstacles must be overcome to organize a prospective cohort in countries with high degree of poverty. Low education precludes the use of self-reported questionnaires, limiting the number of enrolled families. Another problem affecting the response rate and the follow-up is the lack of well urbanized neighborhoods, which greatly increase the difficulty of the visits.

As in other birth-cohorts, a valuable selection of biological material has been done in this study [35]. Laboratory tests will help to assess genetic and environmental factors linked to asthma development, providing novel information about disease mechanisms in the tropics. An open question when studying allergies in this zone is the influence of A. lumbricoides infection on the process of IgE sensitization to common allergens, especially house dust mites [89]. Thus, IgE serology to Ascaris spp. and allergen extracts will be serially performed to explore primary sensitizers and how the infection status influences the frequency and strength of specific IgE response. Another aspect to be evaluated is the relationship between the immunological profile at birth and further wheezing in children. Immune- responsiveness gene expression will be monitored at different time-points. As differences in microbiota composition may influence immune response [90–92], stool samples has been collected from birth to 24 months, to identify gut microbiota composition and explore their potential relationship with the IgE responses and allergies.

Here we describe the demographic characteristics of the FRAAT birth cohort study, representative of socially deprived urban areas of underdeveloped tropical countries. At 24 months, the history of allergic symptoms is different to the "atopic march" described in some populations of industrialized countries. Although the rates of wheezing were higher than those reported in these countries, cases of eczema were not found. Wheezing is the most frequent phenotype during the first 24 months of life and is strongly associated with maternal asthma. The availability of cord blood serum and other biological samples, will allow to further study risk factors for asthma and atopy during early childhood, including the effects of particular factors related to poverty, such as early microbial and parasite exposure.