Early Respiratory Infections and the Development of Asthma
The results of our population-based, 20-year prospective cohort study show that the occurrence of respiratory tract infections during early childhood predicts the development of asthma up to young adulthood. We were able to utilize both questionnaire data that included information on the occurrence of respiratory infections during the 12 months preceding baseline data collection (part 1) and registry-based data that included information on respiratory infections leading to hospitalization (part 2). In part 1, persons with at least 1 early LRTI had a 2.8-fold incidence of asthma by the age of 12 years and a 2.1-fold incidence by the age of 27 years in comparison with persons without LRTIs. The risk of developing asthma up to young adulthood was also increased in relation to having at least 1 LRTI that required hospitalization by the age of 6 years (part 2), suggesting a 2-fold risk of asthma by the ages of 18 and 27 years in comparison with persons without LRTIs.
Although our results are consistent with the causal hypothesis, there are alternative explanations, such as the presence of one or several common causes for both the studied infections and asthma (genetic, environmental, or their joint effect), which could partially explain the observed association between early infections and the risk of asthma. For example, children with a genetic propensity to asthma could be more prone to frequent and/or severe respiratory infections. Even in this case, respiratory infections predict the development of asthma and are likely to be harmful in relation to asthma development.
The risk of asthma had the strongest association with infections occurring before the age of 3 years. The immunological and physiological development of the respiratory system has been found to be fastest during early childhood, which could make the respiratory tract more susceptible to the effects of infections at those ages. Respiratory infections at that time may cause permanent changes, which could have an influence on asthma development all the way to adulthood.
The strengths of our study include the prospective follow-up of a large population-based cohort from early childhood to the ages of 20–27 years. The response rate at baseline was high (80.3%). The present study population—including those who participated in either the 6-year follow-up or the 20-year follow-up or both—included 86.8% of the baseline population. There were no substantial differences between the baseline population and the present study population (Table 2), suggesting that any major selection bias was unlikely.
The definition of asthma was based on parent- or self-reported physician-diagnosed asthma assessed at the 6-year and 20-year follow-ups. Using a questionnaire rather than medical records as a source of information on the diagnosis may have introduced some measurement error. We partially evaluated this type of error at baseline by calling all of the parents who had reported that their child had physician-diagnosed asthma. All of the parents verified the information provided, which indicates that they were well informed about the presence of asthma. We think it is reasonable to assume that young adults at the 20-year follow-up were also well aware of their diagnosed asthma. The National Social Insurance Institution partially reimburses patients whose asthma diagnosis fulfills the diagnostic criteria for the cost of asthma medication. The criteria are presented by the personal physician and evaluated independently by a specialist at the National Social Insurance Institution before approval of the diagnosis. This reduces heterogeneity in diagnostic practice between physicians, but naturally does not eliminate changes in diagnostic practice over time. The Finnish health-care system has broad coverage for the entire population, and the majority of persons with asthma are likely to receive medical attention. Special reimbursement for the cost of asthma medication offered people a strong economic incentive for getting their asthma clinically diagnosed.
In addition, our analysis included information on the age of asthma onset, provided first by the parents and then by the subjects themselves in the 20-year follow-up survey. Pattaro et al. showed that using the age of asthma onset increased the reliability of incidence estimates. Assessment of the age of onset for asthma is problematic both conceptually and empirically. Conceptually, a subject has asthma when the diagnostic criteria are fulfilled, but given the nature of the disease, the diagnosis is always made after some delay. The delay depends, among other things, on the health behavior and activity of the individual and his or her access to health care. After the actual diagnosis, the patient often recalls symptoms and signs that may have started before the physician-made diagnosis. Given these limitations, the results should be interpreted cautiously.
Information on the frequency and type of respiratory infections occurring during the 12-month period in early childhood was based on parents' reporting in the baseline questionnaire. This was a limitation of the study, but it is not likely to have introduced systematic error, because the parents did not know whether their child would develop asthma when reporting about infections at baseline. In addition, we were able to strengthen our findings by using information on hospitalized respiratory infections from the National Hospital Discharge Register database, which has been consistently evaluated to maintain highly complete and reliable data. Both of our information sources on infections included strengths and limitations. Reporting in the questionnaire may have included some misclassification. For example, information on recurrent otitis media is susceptible to error, but the error is not likely to be systematic in relation to the development of asthma. Further, the association for recurrent otitis media was greater than for acute otitis media and thus consistent with the hypothesis that the greater the number of ear infections the higher the risk of developing asthma. The hospital diagnoses did not include any information on primary health-care visits, because that information started to be registered in Finland only after 2011. However, we were able to include all hospital diagnoses from birth to the age of 6 years and to estimate the associated risk of developing asthma, whereas in the questionnaire analyses we probably missed infections from the time period before data collection. Since both parts of the analyses were restricted to persons who did not have previous or current asthma at the beginning of follow-up, recall bias was not an issue when the occurrence of infections was assessed.
We adjusted our analyses for a number of potential confounders that represented the most common determinants of asthma.
The original hygiene hypothesis—based on the results presented in 1989—suggested that early-life infections would protect against atopic disease. However, the results of these studies have been consistent with a protective effect, mainly based on indirect evidence using the number of younger siblings or day-care attendance as proxy measures for the occurrence of respiratory infections. In addition, in a prospective birth cohort study, Illi et al. reported that infectious diseases other than LRTIs in the first 3 years of life would reduce the risk of asthma development by the age of 7 years. However, they also presented figures suggesting that LRTIs were actually predictors of asthma risk.
There is also contradictory evidence that childhood respiratory infections are associated with development of asthma and atopy later in childhood, but none of those longitudinal studies covered adulthood. In 2 prospective birth cohort studies of children selected on the basis of a high risk for asthma (i.e., history of asthma or allergies in at least 1 parent), LRTIs in early life were a significant risk factor for asthma development at age 6–7 years. Additionally, Kusel et al. found in a birth cohort that children with acute respiratory illness and atopy by the age of 2 years constituted the highest-risk group for asthma at the ages of 5 and 10 years. In the Oslo Birth Cohort Study—a prospective longitudinal study of 2,540 children—LRTIs (including bronchiolitis, bronchitis, and pneumonia) during the first year of life were determinants of asthma at the ages of 4 and 10 years. URTIs (such as otitis media) in early childhood have also been linked to asthma at the ages of 4–6 years. In addition, 2 similar reports on data from the European Community Respiratory Health Survey (20- to 44-year-old subjects), with retrospective collection of information on early respiratory infections during the first 5 years of life, suggested that childhood respiratory infections could influence the risk of asthma in adulthood. In our previous population-based incident case-control study among working-age adults (the Finnish Environment and Asthma Study), we found that the occurrence of respiratory infections during the previous 12 months was a strong determinant of adult-onset asthma. According to our systematic literature search, ours is the first prospective cohort study to have assessed the relationship between childhood respiratory infections and the development of asthma up to adulthood.
Discussion
Main Findings
The results of our population-based, 20-year prospective cohort study show that the occurrence of respiratory tract infections during early childhood predicts the development of asthma up to young adulthood. We were able to utilize both questionnaire data that included information on the occurrence of respiratory infections during the 12 months preceding baseline data collection (part 1) and registry-based data that included information on respiratory infections leading to hospitalization (part 2). In part 1, persons with at least 1 early LRTI had a 2.8-fold incidence of asthma by the age of 12 years and a 2.1-fold incidence by the age of 27 years in comparison with persons without LRTIs. The risk of developing asthma up to young adulthood was also increased in relation to having at least 1 LRTI that required hospitalization by the age of 6 years (part 2), suggesting a 2-fold risk of asthma by the ages of 18 and 27 years in comparison with persons without LRTIs.
Although our results are consistent with the causal hypothesis, there are alternative explanations, such as the presence of one or several common causes for both the studied infections and asthma (genetic, environmental, or their joint effect), which could partially explain the observed association between early infections and the risk of asthma. For example, children with a genetic propensity to asthma could be more prone to frequent and/or severe respiratory infections. Even in this case, respiratory infections predict the development of asthma and are likely to be harmful in relation to asthma development.
The risk of asthma had the strongest association with infections occurring before the age of 3 years. The immunological and physiological development of the respiratory system has been found to be fastest during early childhood, which could make the respiratory tract more susceptible to the effects of infections at those ages. Respiratory infections at that time may cause permanent changes, which could have an influence on asthma development all the way to adulthood.
Validity of Results
The strengths of our study include the prospective follow-up of a large population-based cohort from early childhood to the ages of 20–27 years. The response rate at baseline was high (80.3%). The present study population—including those who participated in either the 6-year follow-up or the 20-year follow-up or both—included 86.8% of the baseline population. There were no substantial differences between the baseline population and the present study population (Table 2), suggesting that any major selection bias was unlikely.
The definition of asthma was based on parent- or self-reported physician-diagnosed asthma assessed at the 6-year and 20-year follow-ups. Using a questionnaire rather than medical records as a source of information on the diagnosis may have introduced some measurement error. We partially evaluated this type of error at baseline by calling all of the parents who had reported that their child had physician-diagnosed asthma. All of the parents verified the information provided, which indicates that they were well informed about the presence of asthma. We think it is reasonable to assume that young adults at the 20-year follow-up were also well aware of their diagnosed asthma. The National Social Insurance Institution partially reimburses patients whose asthma diagnosis fulfills the diagnostic criteria for the cost of asthma medication. The criteria are presented by the personal physician and evaluated independently by a specialist at the National Social Insurance Institution before approval of the diagnosis. This reduces heterogeneity in diagnostic practice between physicians, but naturally does not eliminate changes in diagnostic practice over time. The Finnish health-care system has broad coverage for the entire population, and the majority of persons with asthma are likely to receive medical attention. Special reimbursement for the cost of asthma medication offered people a strong economic incentive for getting their asthma clinically diagnosed.
In addition, our analysis included information on the age of asthma onset, provided first by the parents and then by the subjects themselves in the 20-year follow-up survey. Pattaro et al. showed that using the age of asthma onset increased the reliability of incidence estimates. Assessment of the age of onset for asthma is problematic both conceptually and empirically. Conceptually, a subject has asthma when the diagnostic criteria are fulfilled, but given the nature of the disease, the diagnosis is always made after some delay. The delay depends, among other things, on the health behavior and activity of the individual and his or her access to health care. After the actual diagnosis, the patient often recalls symptoms and signs that may have started before the physician-made diagnosis. Given these limitations, the results should be interpreted cautiously.
Information on the frequency and type of respiratory infections occurring during the 12-month period in early childhood was based on parents' reporting in the baseline questionnaire. This was a limitation of the study, but it is not likely to have introduced systematic error, because the parents did not know whether their child would develop asthma when reporting about infections at baseline. In addition, we were able to strengthen our findings by using information on hospitalized respiratory infections from the National Hospital Discharge Register database, which has been consistently evaluated to maintain highly complete and reliable data. Both of our information sources on infections included strengths and limitations. Reporting in the questionnaire may have included some misclassification. For example, information on recurrent otitis media is susceptible to error, but the error is not likely to be systematic in relation to the development of asthma. Further, the association for recurrent otitis media was greater than for acute otitis media and thus consistent with the hypothesis that the greater the number of ear infections the higher the risk of developing asthma. The hospital diagnoses did not include any information on primary health-care visits, because that information started to be registered in Finland only after 2011. However, we were able to include all hospital diagnoses from birth to the age of 6 years and to estimate the associated risk of developing asthma, whereas in the questionnaire analyses we probably missed infections from the time period before data collection. Since both parts of the analyses were restricted to persons who did not have previous or current asthma at the beginning of follow-up, recall bias was not an issue when the occurrence of infections was assessed.
We adjusted our analyses for a number of potential confounders that represented the most common determinants of asthma.
Synthesis With Previous Knowledge
The original hygiene hypothesis—based on the results presented in 1989—suggested that early-life infections would protect against atopic disease. However, the results of these studies have been consistent with a protective effect, mainly based on indirect evidence using the number of younger siblings or day-care attendance as proxy measures for the occurrence of respiratory infections. In addition, in a prospective birth cohort study, Illi et al. reported that infectious diseases other than LRTIs in the first 3 years of life would reduce the risk of asthma development by the age of 7 years. However, they also presented figures suggesting that LRTIs were actually predictors of asthma risk.
There is also contradictory evidence that childhood respiratory infections are associated with development of asthma and atopy later in childhood, but none of those longitudinal studies covered adulthood. In 2 prospective birth cohort studies of children selected on the basis of a high risk for asthma (i.e., history of asthma or allergies in at least 1 parent), LRTIs in early life were a significant risk factor for asthma development at age 6–7 years. Additionally, Kusel et al. found in a birth cohort that children with acute respiratory illness and atopy by the age of 2 years constituted the highest-risk group for asthma at the ages of 5 and 10 years. In the Oslo Birth Cohort Study—a prospective longitudinal study of 2,540 children—LRTIs (including bronchiolitis, bronchitis, and pneumonia) during the first year of life were determinants of asthma at the ages of 4 and 10 years. URTIs (such as otitis media) in early childhood have also been linked to asthma at the ages of 4–6 years. In addition, 2 similar reports on data from the European Community Respiratory Health Survey (20- to 44-year-old subjects), with retrospective collection of information on early respiratory infections during the first 5 years of life, suggested that childhood respiratory infections could influence the risk of asthma in adulthood. In our previous population-based incident case-control study among working-age adults (the Finnish Environment and Asthma Study), we found that the occurrence of respiratory infections during the previous 12 months was a strong determinant of adult-onset asthma. According to our systematic literature search, ours is the first prospective cohort study to have assessed the relationship between childhood respiratory infections and the development of asthma up to adulthood.
SHARE
