Key words: asthma risk; intrauterine; labor complications; maternal age; siblings effect; upper and lower respiratory tract infections

Transcription

1 Risk of Physician-Diagnosed Asthma in the First 6 Years of Life* Natalia Dik, MSc; Robert B. Tate, PhD; Jure Manfreda, MD; and Nicholas R. Anthonisen, MD Objective: The objective of this cohort study was to determine if complications of pregnancy and labor, characteristics at birth, and exposure to infections influence the incidence of asthma in the first 6 years of life. Design: We identified all children born between 1980 and 1990 in the Province of Manitoba, Canada. We used records of physician contacts (inpatient and outpatient) and services of the universal provincial health insurance plan to follow up 170,960 children from birth to the age of 6 years to identify the first diagnosis of asthma. Information on mothers and siblings was also obtained to determine family history of disease and exposure to infections. Results: During the study period, a diagnosis of asthma was made in 14.1% of children by the age of 6 years. The incidence was higher in boys than in girls, in those with family history of allergic diseases. It was higher in urban than in rural areas, and lowest in those born in winter. Asthma was more likely in those with low birth weight and premature birth. Certain congenital abnormalities and complications of pregnancy and labor also increased the risk of asthma. The risk of asthma increased with maternal age. Both upper and lower respiratory infections increased the risk of subsequent asthma, and this effect was more important than exposure to familial respiratory infections, which also tended to increase asthma risk. The risk of asthma decreased with the number of siblings when siblings had a history of allergic disorders. Conclusions: In addition to genetic influences, intrauterine and labor conditions are determinants of asthma. Exposure to both upper and lower respiratory tract infections increases the risk; these infections do not explain the protective effect associated with the increasing number of siblings. (CHEST 2004; 126: ) Key words: asthma risk; intrauterine; labor complications; maternal age; siblings effect; upper and lower respiratory tract infections Abbreviations: CI confidence interval; HR hazard ratio; ICD International Classification of Diseases; LRI lower respiratory tract infection; MPHRR Manitoba Population Health Research Repository; URI upper respiratory tract infection The prevalence of asthma varies considerably worldwide, 1,2 and may be increasing. 3,4 In Canada, asthma is the leading cause of pediatric hospitalization. 5 The onset of asthma is most common in the first year of life, 6 when it is difficult to distinguish it from other wheezing syndromes. 7 Although in many children wheezing is transient, those who wheeze in the first 3 years of life have an increased *From the Departments of Community Health Sciences (Ms. Dik, and Drs. Tate and Manfreda) and Internal Medicine (Dr. Anthonisen), University of Manitoba, Winnipeg, MB, Canada. Support was provided by the Medical Research Council Canada. Manuscript received December 9, 2003; revision accepted May 19, Reproduction of this article is prohibited without written permission from the American College of Chest Physicians ( permissions@chestnet.org). Correspondence to: Nicholas Anthonisen, MD, University of Manitoba, Department of Medicine, 810 Sherbrook St, Winnipeg, MB, Canada R3A 1R8; nanthonisen@exchange.hsc.mb.ca risk of asthma at the age of 5 to 7 years. 8 The high incidence of asthma in early life has focused attention on environmental factors such as allergen exposures and infections that occur before and soon after birth as potential causes. 9 It is also possible that factors that interfere with normal pregnancy create an intrauterine cytokine environment favoring the development of allergy in those with a genetic predisposition. 10 The objective of this study was to identify factors that may increase the risk of physician-diagnosed asthma in children between birth and the age of 6 years. Database Materials and Methods The Province of Manitoba has a relatively stable population of approximately 1.2 million people, and a universally accessible CHEST / 126 / 4/ OCTOBER,

2 health-care system with fee for service as the method of physician payment. Payments depend on submission of claims that include the diagnosis, type of service, and the patient s identifier, date of birth, gender, and residential postal code. There is no record of ethnicity. Nonnominal records of medical services, including hospitalizations, are stored in the Manitoba Population Health Research Repository (MPHRR), and can be ascribed to individual residents through unique anonymous personal identifiers. Between 1980 and 1996, diagnoses were coded according to the ninth revision of the International Classification of Diseases (ICD). 11 The reliability and validity of the diagnoses and personal information recorded have been established, 12,13 although not specifically for asthma between birth and age of 6 years. Population We used the MPHRR to identify study subjects and to track their use of health-care services during the first 6 years of life. We selected all children born in Manitoba between January 1, 1980, and December 31, 1990, who were residents from their birth date and had a hospital delivery record (n 185,988). Children who were institutionalized or had 1 year of insurance coverage (n 8,973), as well as children with missing information (mother s record, birth weight, gestational age, area of residence) were excluded (n 6,055). After exclusions, there were 170,960 children available for the analysis. The size of annual cohorts was similar, increasing from 14,483 in 1980 to 16,303 in Outcome The onset of asthma was defined as follows: (1) the first claim submitted by a physician with a diagnosis of asthma (ICD 493), (2) the second claim (defining asthmatics as those having at least two physician claims), and (3) the first hospital discharge with asthma as the primary diagnosis. Because there were only minor differences in results, we used the first definition in this article. Risk Factors The MPHRR provided for each child the birth date, gender, and the postal code of the parental residence. Residences were classified as urban (metropolitan Winnipeg, approximately 650,000 or 54% of the total provincial population) or rural (outside Winnipeg, the largest town having 30,000 people). The Winnipeg postal codes were matched with contemporary Canadian census data to obtain the mean household income in that area, an indicator of socioeconomic status. Postal code areas were ranked and divided into quintiles. For each child, we determined the primary caregiver on the basis of physician office visits made Monday through Friday during the first year of life. The primary caregiver was classified as a general or family medicine practitioner; or pediatrician, if there were more visits to a pediatrician than to a general practitioner; or mixed, if there were more than two visits to both general practitioners and pediatricians. Few children (3,745 or 2.2%) did not visit a physician s office Monday through Friday in the first year of life. Children were classified into five groups according to gestational age and four groups according to birth weight, forming approximate quintiles, or quartiles. For each child, we identified selected congenital anomalies (ICD 277, ) and perinatal conditions (ICD ). In addition, we identified all physician visits (including the date) for asthma, acute bronchitis and bronchiolitis (ICD 466), bronchitis not otherwise specified (ICD 490), upper respiratory tract infection (URI) [ICD ] and lower respiratory tract infection (LRI) [ICD ], and selected allergic and related conditions: nasal polyps (ICD 471), allergic rhinitis (ICD 477), atopic dermatitis (ICD 691), contact dermatitis and other eczema (ICD 692), dermatitis due to substances taken internally (ICD 693), and urticaria (ICD 708). Visits for noninfectious gastroenteritis and colitis (ICD 558) were also identified. Using the population registry of the MPHRR, we identified each child s mother, her age at birth of the child, and siblings with their ages. For the mother and siblings, we abstracted the dates of all physician visits for asthma, the other respiratory and allergic disorders noted above, as well as URI and LRI and noninfectious gastroenteritis. For mothers, we also identified selected complications of pregnancy (ICD ) and childbirth (ICD , , 669). Using data from the mother and siblings, we determined if the child was exposed to URI or LRI in the family during the first and/or the second year of life. Analysis We used the Cox proportional hazard model 14 to identify characteristics of the child and of his/her mother and siblings that were significantly associated with the first diagnosis of asthma. Variables associated with pregnancy and birth, as well as characteristics of mothers and siblings, were modeled as time-fixed variables. Variables that changed during the course of the study (onset of respiratory and allergic conditions in infants) were modeled as time-dependent variables. Hazard ratios (HRs) with 95% confidence interval (CI) indicate relative risks of asthma. Statistical significance was defined as p 0.05 or 95% CI not overlapping the value of 1.0. Results There were 170,960 newborns followed up to their sixth birthday. Of these, 24,026 newborns (14.1%) were seen for asthma. The incidence was the highest in the first 2 years of life (2.6% and 2.9%, respectively) and declined to 2.0% in years 5 and 6. The Table 1 Adjusted HRs (With 95% CI) for Asthma: Associations With Infant Characteristics* Variables No. HR (95% CI) Gender Female 83, Male 87, ( ) Primary care Family physician (general 96, practitioner) Pediatrician 53, ( ) Mixed 17, ( ) None 3, ( ) Residence Rural 81, Urban Q1 20, ( ) Urban Q2 14, ( ) Urban Q3 16, ( ) Urban Q4 21, ( ) Urban Q5 17, ( ) *HRs are adjusted for all variables. Q1 to Q5 areas with the lowest, to areas with the highest mean household income Clinical Investigations

3 Table 2 Adjusted HRs (With 95% CI) for Asthma: Associations for Variables Measured at Birth* Variables ICD Code No. HR (95% CI) Birth weight, g 3,100 39, ( ) 3,100 3,490 50, ( ) 3,500 3,790 37, ,800 43, ( ) Gestational age, wk 36 5, ( ) , ( ) , ( ) 40 58, , ( ) Congenital/perinatal Cystic fibrosis ( ) Other congenital 747 1, ( ) anomalies of circulatory system Congenital anomalies of ( ) respiratory system Respiratory distress ( ) syndrome Pregnancy/delivery Hypertension/eclampsia , ( ) Multiple gestation 651 3, ( ) Long labor , ( ) Obstetric trauma ( ) *HRs are adjusted for all variables. cumulative incidence increased 2.3 times (95% CI, 2.2 to 2.5) from the cohort born in 1980 to that born in Tables 1 4 show variables with significant HRs for the incidence of asthma. These HRs were derived from the model that included all significant timefixed and time-dependent variables. Table 1 shows Table 4 HRs (With 95% CI) for Asthma: Associations With Time-Dependent Variables Representing the Occurrence of Respiratory Infections and Allergic Disorders in Children* Variables ICD Code HR (95% CI) URI ( ) Acute bronchitis/bronchiolitis ( ) Allergic rhinitis ( ) Pneumonia/influenza ( ) Bronchitis ( ) Noninfectious gastroenteritis ( ) Atopic dermatitis ( ) Contact dermatitis ( ) Dermatitis (food and drugs) ( ) Urticaria ( ) *HRs are adjusted for all variables. that male gender, pediatricians as primary care providers, and urban residence increased the likelihood of asthma. In Winnipeg, the risk did not vary significantly among neighborhoods with different mean income. Table 2 shows that risk of asthma was elevated in infants with lower birth weight and in those born prematurely. In addition, certain conditions associated with either pregnancy or delivery slightly increased the risk. Multiple gestations were associated with a reduced risk. Cystic fibrosis, respiratory distress syndrome, and congenital anomalies of the respiratory and circulatory system increased the risk of asthma. Table 3 shows that a maternal history of asthma was a strong predictor of asthma, while a maternal history of allergic rhinitis was less so. The risk of asthma increased with the mother s age. However, Table 3 Adjusted HRs (With 95% CI) for Asthma: Associations With Characteristics of the Mother and Siblings* Variables ICD Code No. HR (95% CI) Mother Age per year 1.01 ( ) Asthma , ( ) Allergic rhinitis 471,477 27, ( ) Has a sibling 133, ( ) Sibling with Asthma , ( ) Allergic rhinitis , ( ) Dermatitis due to food and drugs 693 5, ( ) Contact dermatitis , ( ) Noninfectious gastroenteritis , ( ) Mother or sibling with URI, 1st yr , ( ) URI, 2nd yr , ( ) LRI, 1st yr 480 7, 466, , ( ) LRI, 2nd yr 480 7, 466, , ( ) *HRs are adjusted for all variables. CHEST / 126 / 4/ OCTOBER,

4 this was the case only when time-dependent variables (Table 4) were added to the model; without them, the risk of asthma in infants was inversely related to maternal age. Table 3 also shows that having a sibling was protective. However, having a sibling with a history of allergic conditions increased the risk of asthma. Asthma risk was more strongly associated with asthma in siblings than with maternal asthma. The association with maternal or sibling allergic rhinitis was weaker than that with asthma. A sibling s history of noninfectious gastroenteritis and of contact dermatitis was protective. Table 4 shows significant time-dependent variables. A physician s diagnosis of several atopic conditions and noninfectious gastroenteritis increased the likelihood of a subsequent diagnosis of asthma. In addition, both LRI and URI increased the risk of asthma. Exposure of an infant during the first or second year of life to URI or LRI in the mother or siblings did not affect the risk of asthma when time-dependent variables were in the model (Table 3). Without time-dependent variables in the model, however, the significantly increased risk of asthma was associated with mother s and sibling s URI and LRI in both the first and the second year of life. Having a sibling was protective (Table 3). We explored this by substituting the number of siblings for the categorical sibling variable. When there were no time-dependent variables in the model, only children with four or more siblings had a significantly lower risk of asthma than those without siblings. When time-dependent variables (Table 4) were in the model, those with one sibling were protected (HR, 0.94; 95% CI, 0.91 to 0.97), and protection increased almost linearly so that the HR for those with five and more siblings was 0.74 (95% CI, 0.67 to 0.82). We classified children into those with at least one sibling with asthma, or allergic rhinitis, or atopic dermatitis, and those without (Fig 1). In the latter, having a single sibling was protective, and the effect did not increase with the number of siblings. Those with siblings with allergic disorders had a higher risk than those without siblings, but the effect decreased progressively with the number of siblings. We also explored the effect of the month of birth on the risk of asthma (Fig 2). The risk was significantly higher for those born between July and December than for those born between January and March when the risk was the lowest. Discussion There are several weaknesses associated with using administrative databases to study asthma. The Figure 1. HRs (95% CI) for asthma by the number of siblings for children (top, A) with siblings and (bottom, B) without siblings with allergy. Reference group is children without siblings. first is the diagnosis of asthma. There is no objective test for asthma in children, 15 and there may be labeling problems in relation to related diseases. We relied on the diagnosis submitted by the physician for payment of services. We defined the onset of asthma in three different ways and carried out analyses with each definition as the outcome. Because the same variables were identified as risk factors with similar HRs, we report results with onset of asthma based on the first claim. This gave us greater power to detect significant associations. We did not need to worry how to model the first claim for asthma if the onset was defined as the second claim, and we did not miss cases because the second claim occurred after the sixth birthday. Using hospitalization as the criterion for asthma diagnosis may be more specific, and may also reflect the severity of 1150 Clinical Investigations

5 Figure 2. HRs (95% CI) for asthma by month of birth. Children born between January and March are the reference group. the disease. However, the subset of children who were hospitalized for asthma was unlikely to be representative of the cohort as a whole. There are several advantages of using the MPHRR for asthma research. It includes the whole population with universal coverage and access to health care, and has substantial statistical power to detect associations. There is no reporting or recall bias in regard to either asthma or the risk factors examined. We were able to identify objectively the primary caregiver in the first year, and families (mother and siblings), and assess the relationship of their diseases to the onset of asthma in the children studied. The overall cumulative asthma incidence by age 6 years was 14.1%, similar to 18.2% for the British 1958 birth cohort at age 7 years. 16 Incidence increased 2.3 times between 1980 and 1990, and we controlled for the cohort effect in our analyses. The same risk factors were identified when individual cohorts were examined. Our finding that asthma was more likely to be diagnosed in boys is consistent with most studies. 3,16 18 Primary care pediatricians were more likely to diagnose asthma than family physicians. The increased sensitivity on the part of pediatricians may reflect greater access to guidelines and other educational efforts concerning asthma. We found that childhood asthma was less common in rural areas than in the urban area, as have others in Canada. 19 This cannot be ascribed to pediatrician supply since we controlled for this effect. There was no association between asthma and urban neighborhood mean household income. This is consistent with our previous studies 20 in contrast to the United States, where asthma in urban children was more common in lower socioeconomic groups than in higher socioeconomic groups. 21 As others, 17 we found a strong effect of a family history of asthma and allergic disorders on asthma risk. In the present study, childhood asthma was more strongly associated with asthma in siblings than in mothers, presumably because siblings could have inherited an asthmatic predisposition from the father as well as the mother. Our database did not identify the biological father. Independent of familial history, the occurrence of an allergic disorder (for example allergic rhinitis) in a child further increased likelihood of a subsequent diagnosis of asthma (Table 4). In the New Zealand cohort, low birth weight was not associated with an increased prevalence of asthma. 22,23 However, our results are consistent with studies showing the association of low birth weight with an increased risk of asthma. 17,21 In our study, low gestational age was an additional risk factor for asthma, independent of birth weight. This is in agreement with some other studies, 18,24 but not all. 21 Since maternal smoking is associated with an increased risk of both low birth weight and premature birth, maternal smoking may underlie the association we found between these factors and asthma. We found that some congenital anomalies of the respiratory and circulatory system apparently increased the risk of asthma. This was true for both cystic fibrosis and perinatal respiratory distress. This likely represents a labeling problem since these CHEST / 126 / 4/ OCTOBER,

6 conditions are associated with bronchial hyperresponsiveness and wheezing. 28,29 Certain conditions during pregnancy and labor further increased the risk of asthma in our study. This is consistent with literature indicating a wide variety of abnormalities of pregnancy and birth can be associated with subsequent asthma in the child. Though maternal asthma increases the risk of complications during pregnancy, 34,35 we controlled for this influence. Others 21,36 have found maternal age inversely related to wheezing lower respiratory illness. This may reflect a greater vulnerability of young mothers and higher exposures to risk factors such as smoking. In our study, asthma risk increased with maternal age after we considered time-dependent variables that reflected the child s own respiratory infections and allergic disorders. Children of younger mothers may be more likely to have URI and LRI, accounting for previous findings. Our finding that the risk of asthma was lower for infants born in winter months is consistent with findings on the Isle of Wight, 17 but not with those in Germany 37 or in New Zealand, 23 where only skin sensitivity to allergens was associated with season. Businco et al 38 found that children with skin reactivity to pollen were significantly more likely to be born during pollen season. Our study strongly suggests that infections during the first years of life increase the risk of asthma. This might have been expected in the case of LRI, which present major labeling problems. However, we found little difference between the effects of LRI and URI. Our findings are consistent with those of McKeever et al, 39 but others 40 have reported that URI had a protective effect in terms of subsequent asthma diagnoses. Respiratory infections in mothers and siblings in the first 2 years significantly increased the likelihood of asthma in study subjects when time-dependent variables were not in the model, but not when time-dependent variables, representing infections in children themselves, were added (Table 3). This suggests that the effects of exposure to respiratory infections were largely mediated by the acquisition of these infections by the children themselves. We found that previous URI predisposed to asthma, in contrast to the hygiene hypothesis, 41 which argues that exposure to URI drives the immune system in the T-helper type 1 direction thought to be protective against asthma. 40 However, children with siblings having noninfectious gastroenteritis and contact dermatitis were protected. These disorders are poorly defined and very common, and may be proxies for unknown exposures in early life. However, gastroenteritis and contact dermatitis increased asthma risk in the children themselves. The protective effect of siblings has been cited as a proxy for exposure to infections favoring the hygiene hypothesis. The fact that we found that siblings had a protective effect after considering previous URI, gastroenteritis, and contact dermatitis suggest that the effect was not due to exposure to infections. Finally, the progressive protective effect of sibling number seemed to be restricted to children in atopic families in our study. Celedón etal 42 reported that maternal history of asthma modified the relationship between day care exposure and asthma. However, contrary to their observation that the protective effect of day care exposures was greater in those without maternal asthma, in our study the siblings protective effect was greater in those with family history of asthma. References 1 The International Study of Asthma and Allergies in Childhood (ISAAC) Steering Committee. Worldwide variations in the prevalence of asthma symptoms: the International Study of Asthma and Allergies in Childhood (ISAAC). Eur Respir J 1998; 12: European Community Respiratory Health Survey. Variations in the prevalence of respiratory symptoms, self-reported asthma attacks, and use of asthma medication in the European Community Respiratory Health Survey (ECRHS). Eur Respir J 1996; 9: Sears MR. Epidemiology. In: Barnes PJ, Rodger IW, Thomson NC, eds. Asthma: basic mechanisms and clinical management. 3rd ed. San Diego, CA: Academic Press, 1998; Beasley R, Crane J, Lai CKW, et al. Prevalence and etiology of asthma. J Allergy Clin Immunol 2000; 105:S466 S472 5 Editorial Board. Respiratory disease in Canada. Ottawa, Canada: Health Canada, Yunginger JW, Reed CE, O Connell J, et al. A communitybased study of the epidemiology of asthma: incidence rates, Am Rev Respir Dis 1992; 146: Martinez FD. Development of wheezing disorders and asthma in preschool children. Pediatrics 2002; 109: Martinez FD, Wright AL, Taussig LM, et al. Asthma and wheezing in the first six years of life. N Engl J Med 1995; 332: Jones AC, Miles EA, Warner JO, et al. Fetal peripheral blood mononuclear cell proliferative responses to mitogenic and allergenic stimuli during gestation. Pediatr Allergy Immunol 1996; 7: Warner JA, Jones CA, Jones AC, et al. Prenatal origins of allergic disease. J Allergy Clin Immunol 2000; 105:S493 S International Classification of Diseases, Ninth Revision. Geneva, Switzerland: World Health Organization, Robinson JR, Young TK, Roos LL, et al. Estimating the burden of disease: comparing administrative data and selfreports. Med Care 1997; 35: Huzel L, Roos LL, Anthonisen NR, et al. Diagnosing asthma: the fit between survey and administrative database. Can Respir J 2002; 9: Cox DR. Regression models and life tables. J Royal Stat Soc 1972; B34: Strunk RC. Defining asthma in the preschool-aged child. Pediatrics 2002; 109: Clinical Investigations

7 16 Anderson HR, Pottier AC, Strachan DP. Asthma from birth to age 23: incidence and relation to prior and concurrent atopic disease. Thorax 1992; 47: Arshad SH, Stevens M, Hide DW. The effect of genetic and environmental factors on the prevalence of allergic disorders at the age of two years. Clin Exp Allergy 1993; 23: Miller JE. Predictors of asthma in young children: does reporting source affect our conclusions? Am J Epidemiol 2001; 154: Ernst P, Cormier Y. Relative scarcity of asthma and atopy among rural adolescents raised on a farm. Am J Respir Crit Care Med 2000; 161: Erzen D, Carriere KC, Dik N, et al. Income level and asthma prevalence and care patterns. Am J Respir Crit Care Med 1997; 155: Schwartz J, Gold D, Dockery DW, et al. Predictors of asthma and persistent wheeze in a national sample of children in the United States: association with social class, perinatal events, and race. Am Rev Respir Dis 1990; 142: Leadbitter P, Pearce N, Cheng S, et al. Relationship between fetal growth and the development of asthma and atopy in childhood. Thorax 1999; 54: Sears MR, Holdaway MD, Flannery EM, et al. Parental and neonatal risk factors for atopy, airway hyper-responsiveness, and asthma. Arch Dis Child 1996; 75: Yuan W, Fonager K, Olsen J, et al. Prenatal factors and use of anti-asthma medications in early childhood: a populationbased Danish birth cohort study. Eur J Epidemiol 2003; 18: Berkowitz GS, Papiernik E. Epidemiology of preterm birth. Epidemiol Rev 1993; 15: Sears MR. Epidemiology of childhood asthma. Lancet 1997; 350: Tough SC, Svenson LW, Johnston DW, et al. Characteristics of preterm delivery and low birthweight among 113,994 infants in Alberta: Can J Public Health 2001; 92: Morgan WJ, Butler SM, Johnson CA, et al. Epidemiologic study of cystic fibrosis: design and implementation of a prospective, multicenter, observational study of patients with cystic fibrosis in the U.S. and Canada. Pediatr Pulmonol 1999; 28: Balfour-Lynn IM, Elborn JS. CF asthma : what is it and what do we do about it? Thorax 2002; 57: Xu B, Pekkanen J, Jarvelin MR. Obstetric complications and asthma in childhood. J Asthma 2000; 37: Nafstad P, Magnus P, Jaakkola JJK. Risk of childhood asthma and allergic rhinitis in relation to pregnancy complications. J Allergy Clin Immunol 2000; 106: Nafstad P, Samuelsen SO, Irgens LM, et al. Pregnancy complications and the risk of asthma among Norwegians born between 1967 and Eur J Epidemiol 2003; 18: Annesi-Maesano I, Moreau D, Strachan DP. In utero and perinatal complications preceding asthma. Allergy 2001; 56: Alexander S, Dodds L, Armson BA. Perinatal outcomes in women with asthma during pregnancy. Obstet Gynecol 1998; 92: Palmer GW, Claman HN. Pregnancy and immunology: selected aspects. Ann Allergy Asthma Immunol 2002; 89: Martinez FD, Wright AL, Holberg CJ, et al. Maternal age as a risk factor for wheezing lower respiratory illnesses in the first year of life. Am J Epidemiol 1992; 136: Wjst M, Dold S, Reitmeir P, et al. Month of birth and allergic disease at the age of 10. Clin Exp Allergy 1992; 22: Businco L, Cantani A, Farinella F, et al. Month of birth and grass pollen or mite sensitization in children with respiratory allergy: a significant relationship. Clin Allergy 1988; 18: McKeever TM, Lewis SA, Smith C, et al. Early exposure to infections and antibiotics and the incidence of allergic disease: a birth cohort study with the West Midlands General Practice Research Database. J Allergy Clin Immunol 2002; 109: Illi S, von Mutius E, Lau S, et al. Early childhood infectious diseases and the development of asthma up to school age: a birth cohort study. BMJ 2001; 322: Strachan DP. Hay fever, hygiene and household size. BMJ 1989; 299: Celedón JC, Wright RJ, Litonjua AA, et al. Day care attendance in early life, maternal history of asthma, and asthma at the age of 6 years. Am J Respir Crit Care Med 2003; 167: CHEST / 126 / 4/ OCTOBER,