Sample EPID 168 commentary (Fall 1997)

Mercedes Carnethon

A confluence of factors including the social environment, allergic sensitization, environmental tobacco smoke (ETS), viral infections, and heredity influence the onset and course of asthma in children (1,2). Even with the multifactorial etiology of asthma, researchers commonly study isolated determinates of disease, so this review will focus on select studies to form an exposure profile of young asthmatics.

Cunningham et al. collected information on respiratory symptoms and history of ETS exposure via parental questionnaire in 11,534 children in 24 communities in the US and Canada (3). Exposure information included categorized data on the cigarettes smoked per day, as well as past smoking with an emphasis on the prenatal period. As compared to children who were never exposed to ETS at home, those currently exposed had significantly higher odds of wheeze with a cold (OR = 1.65; 95% CI, 1.45-1.88) and emergency room visits for wheeze (OR = 1.63; 95% CI, 1.20-2.21). For physician diagnosed asthma, only smoking during pregnancy showed significantly increased odds (OR= 2.70; 95% CI, 1.13-6.45) (3).

Early exposure to allergens and respiratory infections influences the age of onset of asthma (3-6). A prospective study analyzing the relationship between atopy, exposure to the household dust mite, and development of asthma demonstrated a relative risk (RR) of active asthma 4.8 times greater in children exposed to high levels of dust-mite antigen (> 10 µg of DER p I per gram of dust) than non-exposed children (4). Researchers collected data on the incidence of viral respiratory infections and conducted skin prick tests for evidence of sensitization annually during the first 5 years and again at 11 years of age. However, strict use of adult standards for sensitization (wheal > 4mm in diameter) may have decreased the sensitivity of the test in children. The cohort of 67 children were selected on the basis of one parent's having atopy (a background risk present in at least 33% of the UK's population) to aid in the collection of events and improve cohort retention (4). Children were divided into four groups: 1) those who had never wheezed; 2) those who had wheezed at least once in their lives; 3) those with active asthma; and 4) those with asthma who were receiving medication (4).

Data collected on viral infections allowed researchers to un-confound the relationship between respiratory infections before the age of 5 and the development of asthma. Results indicate that active asthma is strongly associated with both atopy and sensitivity to the house-dust mite at age 11 (RR = 14.6, p <.005) when compared with non-atopic children. In a child sensitized specifically to the house dust mite, the RR increased to 19.7 (p<.001). Final conclusions by the authors suggest that exposing young children to high levels of foreign proteins in the home is an important health risk (4). Strengths of this prospective design included the ability to establish causality between infantile exposure and the presence of disease at age 11. However, analysis of the time of sensitization onset did not reveal whether dust mite sensitization preceded or followed wheezing at the final examination. Some children had clear sensitivity before wheezing, while others were wheezing before evidence of sensitization (4).

School attendance is a substantial part of a child's social environment. Johnston et al. used a cohort of 108 British schoolchildren reporting asthma symptoms on a health questionnaire to collect rates of self-reported upper respiratory infections and asthma symptoms and correlated them with the rates of hospital admissions for asthma (7). Self-reported respiratory symptoms or a drop in Peak Expiratory Flow Rates (PEFR) were supplemented by a visit to the home by investigators to obtain a nasal aspirate and serum specimens for virologic testing and confirmation. Comparisons were made with the school holiday calendar (7).

Results demonstrated a strong correlation between respiratory infections and hospital admissions for asthma (r = .72, p < .0001). Infections were more frequent during periods of school attendance (87%), and peaks were seen at the beginning of each new school term for asthma, with lows during the holiday season (7). These trends-- falling admissions during holidays (p < .005) followed by peaks each term (p< .001), were corroborated by Storr and Lenney over an 11-year period in a retrospective study (8).

Classifying the outcome, asthma or persistent wheeze, is a methodologic challenge that can alter the results of studies, as demonstrated by Cunningham et al. In the Cunningham study, using wheeze as the endpoint demonstrated a much stronger association with the exposure than did physician confirmed asthma (3). If the distinction, which is dependent on severity and frequency of episodes, is of clinical significance, then close attention must be paid to the endpoints chosen. However, for epidemiologic study, combining endpoints allows for the collection of more events, improving the power to detect significant differences.

The potential for confounding due to social class was not addressed in this brief review, but is an important factor to contend with. When Weitzman et al. entered factors highly associated with poverty (ETS, respiratory viruses during infancy, psychological stressors, poor dietary intake, access to medical care, and outdoor air pollution) into a regression analysis, no independent effect of poverty was demonstrated in his cross-sectional data set (2). These results provide evidence that characteristics of a poverty-related environment, rather than income itself, contributes to much of the increased rates of asthma in children living in poverty (2).

Retrospective and cross-sectional studies of behavioral exposures like ETS, lacks measurement reliability. Cunningham et al. speculates that "the lack of association between physician diagnosed asthma and current smoking are due to biases in the reporting of smoking because of it's lack of social acceptability" (3). Additionally, upon diagnosis of asthma, parents may quit smoking or be more likely to deny it. If these biases were present, then estimates of the relationship between ETS and asthma may be biased towards the null. Wheeze may be unaffected by this bias if it were associated primarily with respiratory infection, in which case parents may have continued smoking (3).

Prospective analyses provide the only evidence of temporality, but Sporik's study demonstrates that this ability is limited by the interrelatedness of the exposures (as discussed earlier) (4). Correlational studies are restricted in their ability to establish causal relationships in the event of erroneously attributing group characteristics to individuals. The inability to control for confounding from temperature, humidity, and allergen exposure (all established risk factors for the transmission of some respiratory viral infections and airborne allergens) present further limitations for the use of correlational designs. Even with these limitations, correlational analyses are informative and can identify areas for further study.

The study of childhood asthma requires flexibility in study designs and reliance on the consistency of findings and on the biologic plausibility of relationships to establish causality. Large cross-sectional studies like that by Cunningham demonstrate robust statistically significant findings that are primarily weakened by the selection of two different endpoints for etiologically similar phenomenon with equally severe consequences (3). Future study designs (particularly in smaller studies) could rely on more objective measures of exposure like the collection of urine cotinine levels to study exposures to ETS or review of medical records during pregnancy for a record of maternal smoking in a retrospective study. While evidence of the heritability of allergic diseases, including asthma, cannot be discounted, ample epidemiologic research suggests that the environment potentiates the onset of asthma in children and shapes the course of disease.

Word Count: 1261