Risk factors for acute asthma in tropical America: a case–control study in the City of Esmeraldas, Ecuador

Abstract Background Despite the high asthma rates described in Latin America, asthma risk factors in poor urban settings are not well established. We investigated risk factors for acute asthma among Ecuadorian children. Methods A matched case–control study was carried out in a public hospital serving a coastal city. Children with acute asthma were age‐ and sex‐matched to non‐asthmatics. A questionnaire was administered, and blood, as well as stool, and nasopharyngeal swabs were collected. Results Sixty cases and 119 controls aged 5–15 were evaluated. High proportions of cases were atopic with population‐attributable fractions for atopy of 68.5% for sIgE and 57.2% for SPT. Acute asthma risk increased with greater titers of mite IgE (3.51–50 kU/l vs. <0.70kU/l – OR 4.56, 95% CI 1.48–14.06, p = 0.008; >50kU/l vs. <0.70kU/l – OR 41.98, 95% CI: 8.97–196.39, p < 0.001). Asthma risk was significantly independently associated with bronchiolitis (adj. OR: 38.9, 95% CI 3.26–465), parental educational level (adj. OR 1.26, 95% CI: 1.08–1.46), and presence of sIgE (adj. OR: 36.7, 95% CI: 4.00–337), while a reduced risk was associated with current contact with pets (adj. OR: 0.07, 95% CI: 0.01–0.56). Rhinovirus infection was more frequent in cases (cases 35.6% vs. controls 7.8%, p = 0.002). None of the cases were on maintenance therapy with inhaled corticosteroids and most relied on emergency department for control. Conclusions A high proportion of children presenting to a public hospital with acute asthma were allergic to mite, particularly at high IgE titer. Poor asthma control resulted in overuse of emergency care.

Asthma affects an estimated 300 million people worldwide (1) and is the most common chronic disease in children (2). Asthma is now as common in some Latin American countries as in known high prevalence countries such as the United Kingdom (3), mainly in poor urban settings (4,5), where it is a significant public health problem (5). There are limited published data on asthma prevalence from poor populations in Latin American cities (1,6). In Ecuador, 16% of the population (1) and 0.8-10.1% of rural schoolchildren (7,8) have been estimated to have current asthma symptoms.
Asthma is a complex disease caused by interactions between environmental exposures and host genetics. Atopy, measured by skin test reactivity (SPT) or the presence of specific IgE (sIgE) to aeroallergens, is an important risk factor for asthma, but the population fraction of asthma attributable to atopy (PAF) varies greatly between regions (9).
The hygiene hypothesis has been put forward to explain temporal trends of increased asthma prevalence in industrialized countries (10) and a reduced prevalence among rural populations among whom exposure to farming and other factors associated with a more traditional rural lifestyle is considered to provide protection (11). It has been suggested that the hygiene hypothesis may not apply to urban Latin America (12) where prevalence rates are particularly high among the poor who live in unhygienic and overcrowded conditions (13).
Inadequate asthma management and control is frequent in Latin America (14), where asthmatic children often rely on emergency care for management of their disease (15). Acute asthma exacerbations are episodes of increasing shortness of breath, accompanied by cough, wheezing, and chest tightness, although there is disagreement on a precise definition (16). Severe acute asthma has been defined as that requiring systemic corticosteroids and emergency department attendance or hospital admission (17).
The aim of this study was to identify risk factors for asthma among children presenting with an acute exacerbation to the emergency department at the public hospital serving a small coastal city in Ecuador.

Study design and population
A matched case-control study was carried out between October and December 2012 at the public hospital, Hospital 'Delfina Torres de Concha' (HDTC) in the City of Esmeraldas in northeastern coastal Ecuador. Children aged 5-15 attending the emergency department at HDTC with a clinical presentation consistent with acute asthma (defined as respiratory distress and wheezing with or without coughing that improved on administration of bronchodilators) (17) were eligible as cases. Controls were children attending the emergency department for reasons other than asthma. Two controls were matched to each case by age (AE2 years) and sex. Exclusion criteria were no parental consent and children who had other chronic respiratory (e.g., cystic fibrosis and pulmonary hypertension), cardiac (e.g., congenital heart disease) or neurological (e.g., cerebral palsy and Down syndrome) diseases.

Study procedures
A questionnaire based on International Study of Asthma and Allergies in Childhood (ISAAC) phase II questionnaire (18) was administered in Spanish to the child's parent or guardian to collect information on socio-demographic factors, core allergy and asthma symptoms, risk factors, and previous treatment and management of asthma. This questionnaire has been widely field-tested in Ecuador (7,19). Bronchiolitis was defined as an episode of wheeze and difficulty breathing during the first two years of life (20). Weight and height were measured. A blood sample was taken to measure hematocrit, total white cell and differential count, and total and specific IgE.

Allergen skin prick test
Skin prick testing was carried out with eight allergen extracts (Greer laboratories, Lenoir, NC, USA): house dust mite (Dermatophagoides pteronyssinus/D.farinae mix), tropical mite (Blomia tropicalis, Leti, Spain), grass pollen, American cockroach, fungi, Alternaria tenuis, cat, dog, and positive histamine and negative saline controls. A positive reaction was defined as a mean wheal diameter ≥ 3 mm than the saline control 15 min after pricking the allergen onto the forearm. The test was undertaken at the moment of recruitment during the acute attack. Children taking antihistamines were excluded from SPT analyses.

Stool examinations
Single stool samples were collected up to 2 weeks after recruitment and analyzed for geohelminth eggs and larvae by direct wet mount and formol-ethyl acetate concentration methods (21).

Airways function and inflammation
Forced expiratory volume in the first second and FVC were measured before and after administering a b 2 agonist (2 puffs of Ventolin, GlaxoSmithKline, UK) using a portable spirometer (MicroMedical, UK). Fraction of exhaled nitric oxide (FeNO) was measured using an adapted NO breath instrument (Bedfont, UK) at flow rates of 50, 83, and 100 ml/sec, as described (22). These results were used to estimate FeNO parameters (alveolar concentration and bronchial flux) using Hogman method (23).

PCR for respiratory viral infections
A nasopharyngeal swab was collected into universal transport medium (Diagnostic Hybrids, Athens, OH, USA) and stored at À80C. RNA was extracted using commercial kits (Qiagen), and specific RNA for rhinovirus, respiratory syncytial virus (RSV), human metapneumovirus (HMPV) and parainfluenza viruses (PIV) 1-4 was converted to cDNA and amplified by reverse transcriptase real-time PCR (AgPath-IDTM One-Step RT-PCR kit, Invitrogen) using primer/probe combinations as described (24) on a 7500 Fast machine (Applied Biosystems). A positive test was defined as Ct values ≤38.

Statistical analysis
We estimated that 60 asthma cases and 120 non-asthmatic controls would give 90% power at p < 0.05 to detect OR ≥2.2 or ≥1.8 for asthma risk factors with 10% or 30% prevalence, respectively. Fisher's exact test and Mann-Whitney U-test were used to compare groups, as appropriate. The associations between acute asthma and risk factors were explored using multivariate conditional logistic regression. The population fraction of acute asthma attributable to atopy (PAF%) was estimated [(OR-1/OR) 9 prevalence of atopy among cases]. Statistical analysis was carried out using SPSS Statistics version 20.0 (Armonk, New York, United States).

Ethical Approval
The study protocol was approved by the Bioethics Committees of the Universidad San Francisco de Quito and Pontificia Universidad Cat olica del Ecuador, Quito, Ecuador. Informed written consent was obtained from the child's parent or guardian and minor's assent from the child.

Results
A total of 179 children (60 cases and 119 matched controls) were recruited. Differences between cases and controls for study variables are shown in Table 1. Atopy was much more common in cases than controls and cases had higher levels of total IgE. Cases had higher levels of bronchial NO but not alveolar NO. Few of the children were infected with intestinal helminths: 4 had Ascaris lumbricoides and 3 had Trichuris  (10), monthly income (1), birthweight (70), breastfeeding (1), contact with farm animals 1st year (3), humidity in household 1st year (3), bronchiolitis (1), day care attendance (1), parental asthma (1), birth order (2), contact with farm animals (5), humid house (1), intestinal helminths (60), BMI (17), hematocrit/eosinophil counts (5), SPT (2), asIgE/total IgE (6) trichiura. A history of bronchiolitis during infancy was associated with acute asthma and did not differ significantly by age and sex (Table S1). The results of multivariate analyses are provided in Table 2. Significant risk factors for acute asthma were bronchiolitis in early childhood, atopy (any positive sIgE), and higher parental educational level. Current contact with domestic pets was associated with reduced risk of asthma.
The associations between markers of atopy to specific aeroallergens and acute asthma are shown in Table 3. For SPT, we observed strong associations between reactivity to mites and asthma, and a weaker association with SPT to American cockroach. Few children were reactive to other aeroallergens: grass pollen (0% vs. 2%, p = 0.509), Alternaria (2% vs. 0%, p = 0.582), cat dander (2% vs. 4%, p < 0.001) and none for dog or fungi mix. The population fraction of acute asthma attributable to SPT to any aeroallergen was 57.2%.
We explored the effect of rhinovirus infection on the association between acute asthma and mite (any) IgE titer (Table S2). There was some evidence of a stronger association between mite IgE and acute asthma among children with rhinovirus infection compared to those without infection, but the small number of infections made the estimates of effect extremely imprecise.
Of the 60 acute asthma cases evaluated, 77% had been previously diagnosed with asthma, 20% had visited a doctor during the last year for regular asthma control, and 86% had attended the emergency department at least once in this same period with an acute exacerbation. Treatments received in the past year are shown in Fig. 2. Almost all children had used oral bronchodilators (91%) during acute attacks. Five children had been taking antileukotriene agents (montelukast) for long-term treatment as monotherapy, and none were receiving inhaled corticosteroids.

Discussion
In the present study, we have investigated risk factors for acute asthma among children aged 5 to 15 attending the emergency room of a public hospital in a small coastal city in tropical Ecuador. Our data show that the majority of children (82%) with acute bronchospasm in our population are atopic with a population fraction of acute asthma attributable to sIgE of 68.5%. Mites were by far the most important sensitizing aeroallergens. The odds for acute asthma increased with mite sIgE titer: such that at a titer of greater than 3.5 kU/l, the risk of acute asthma was approximately fivefold greater, and at titers greater than 50 kU/l, the odds of asthma were almost 40fold greater than controls.
The ISAAC phase II study estimated population-attributable fractions (PAFs) of wheeze/asthma among children aged 8-12 attributable to sIgE of 18.3% in non-affluent compared to 45.6% in affluent countries (9). The two Latin American study centers in ISAAC phase II estimated PAFs of 11% for wheeze/ SPT (9). Previous estimates of PAFs in Esmeraldas Province in case-control studies of current wheeze/SPT have estimated 10.1% in rural and 23.5% in urban schoolchildren (25). In the present study of acute hospital-based asthma, the PAF for SPT to any aeroallergen was much higher (57.2%) and most of the effect was explained by mite sensitization.
The higher PAF related to atopy found in our study, particularly in comparison with schoolchildren with wheezing illness from the City of Esmeraldas (25) may be explained by the severity of illness: Children in our study were recruited at a hospital emergency department during an acute asthma attack, while wheezing illness during the previous year was measured by parental questionnaire (18). Atopic asthmatic children tend to have more severe and difficult-to-treat disease than nonatopic asthmatics (26) and may be more likely to require emergency care. Studies from developed countries suggest that atopy, particularly to mite, is a risk factor for hospital admission among asthmatic children (27) and is more common among acute asthmatics compared to asthmatics not requiring hospital re-admission (28). There are very limited data available from developing countries on the role of atopy as a risk factor for acute asthma. In a study from South Africa, 74.0% of children admitted to a hospital emergency room with acute asthma had a positive SPT to any aeroallergen (29) and is consistent with our own findings of 68% with SPT.
Atopy in our asthmatic population was dominated by mite sensitization: All children with positive sIgE or SPT to any non-mite aeroallergen were mite sensitized, and the risk of acute asthma was associated with mite IgE titer. A recent study in another more developed tropical urban environment, Singapore, showed that over 70% of 206 individuals sampled from two population-based cohorts were atopic to mites, that sIgE to other aeroallergens was relatively infrequent (<30%), and that few individuals without mite atopy were sensitized to other aeroallergens (30). It was suggested that mite monosensitization with high-titer IgE observed in this population contributed to the high prevalence of asthma (30). In our study population, a lower proportion of non-asthmatic controls than the Singapore study (30) had mite atopy (39%): It is possible that with increasing development, the prevalence of mite atopy may increase translating into a greater prevalence of clinically significant allergic airways disease. The increase in PAF for wheeze/atopy observed previously in urban vs. rural children (7) supports this, and such an effect may be more strongly determined by environmental differences (e.g., potable water and sanitation) than relative affluence measured by monthly income. †No further stratification of IgE levels for American cockroach and Ascaris, given the small number of patients and therefore lack of power. The proportions were as follows: American cockroach: cases: 3.51-50 kU/l: 10 (17%), >50kU/l: 0; controls: 3.51-50 kU/l: 6 (5%), >50kU/l: 1 (1%). Ascaris: cases: 3.51-50 kU/l: 17 (28%), >50kU/l: 1 (2%); controls: 3.51-50 kU/l: 17 (15%), >50kU/l: 0.
The prevalence of rhinovirus infection among the acute asthmatics was high. Previous studies have associated acute respiratory viral infections with acute bronchospasm, particularly rhinovirus infections (31). This association has been described to be stronger among atopic children with high-titer IgE to mite (32), a trend that was also observed here. The risk of acute asthma may also be linked to high levels of mite exposure in the environment (28). The inflammation induced in the airways in the context of high-level exposure to mite allergens may modify risk of rhinovirus-associated wheeze and the development of acute exacerbations.
The role of intestinal helminths in the development of asthma and acute exacerbations remains unclear (13,26,33). Allergic sensitization to Ascaris has been associated with wheeze and atopy prevalence (34), an effect that may be mediated by a higher degree of atopy among asthmatic children exposed to ascariasis (35), or by a direct effect on airways reactivity by inflammation caused by the host immune response to Ascaris larvae migrating through the lungs (36). However, few studies have adjusted these associations for mite specific IgE, given the cross-reactivity which has been previously described between mite and Ascaris allergens (37). In the present study, in a population of low helminth prevalence, the association between Ascaris IgE and acute asthma disappeared after controlling for mite IgE, indicating confounding by this factor and suggesting that allergic sensitization to Ascaris does not explain acute asthma in our study population.
It has been suggested previously that the hygiene hypothesis may not be valid in Latin America because of the high prevalence of allergy and allergic diseases among the urban poor who live in conditions of poor hygiene (33). Our data indicate that respiratory infections and morbidity, either through a history in early childhood of bronchiolitis or current infections with rhinovirus, are risk factors for acute asthma,   while current pet exposures are protective. These observations are consistent with the findings of other studies (4,38). The presence of Ascaris IgE, which in a population of low prevalence can be used as a marker for exposure (39), was associated with a reduced risk of SPT and is consistent with findings of previous studies (8,33). Even in Latin America, there is strong evidence that common childhood infections including intestinal worms strongly attenuate atopy but seem to have less effect on asthma symptoms (40). Together, our findings reflect the complexity of the component exposures that make up poor hygiene. Such exposures may affect the risk of acute asthma in different wayssome exposures may protect against asthma, while others may be risk factors. Further, the effect of individual exposures may vary under different circumstances, depending on determinants such as age and intensity of exposure.
Underdiagnosis of asthma (77% of our cases had a previous diagnosis of asthma) likely contributes to poor asthma control -86% of asthmatics had attended the emergency department the previous year. This may explain the high proportion (45.6% in Ecuador and 73.2% in Latin America (15)) of annual costs of asthma-related health care due to unscheduled healthcare use. None of our asthma cases were receiving longterm inhaled corticosteroids compared to <6% of asthmatic patients in other regions of Latin America (14), and most relied on emergency care for asthma control.
Our study has several potential limitations. The use of maternal questionnaires to measure risk factors may be subject to recall bias, and we were not able to measure some important potential risk factors in an urban environment such as air pollution. The study had limited power to detect small effects.
In conclusion, acute asthma in a poor urban population in coastal Ecuador was strongly associated with mite atopy, particularly at high titer. Our data highlight the need for further studies in this and similar populations in Latin America to understand better the causes of acute asthma, particularly the role of the interaction between atopy, aeroallergen exposure, and respiratory viral infections that may allow the identification of novel preventive strategies to reduce disease burden.

Supporting Information
Additional Supporting Information may be found in the online version of this article: Table S1. Infant bronchiolitis among acute-asthmatic patients visiting the emergency room. Table S2. Effect of rhinovirus infection on association between acute asthma and IgE to any mite.