The effects of in-utero smoke exposure on future asthma sufferers
Kristel Van Steen got involved in the study when she spent two years at the Harvard School of Public Health (Boston, USA) as a post-doctoral fellow, from 2003 till 2005. The direction she took on her return to Belgium was the uncommon one of using and developing novel statistical tools to disentangle the genetic underpinnings of complex diseases, including the fields of asthma and allergies. She continued to work on the project she had gotten involved in while in Boston and still maintains collaborations with the asthma team at the Channing Laboratories. As she says, ‘in a former life I was a biomedical statistician, so here my role was examining the longitudinal data. This is complex data which involve accounting for correlated measurements over times. Because we observe each individual multiple times, we may find that the individuals differ in some systematic way. For instance, treatment response may be higher for all observation periods for individual i than for other individuals with the same covariate values because of characteristics of that individual that do not depend on the considered explanatory variables. Such a pattern can be represented by including in the models an “effect” that is specific to each individual. It is this approach we took while developing the models for this study.’ Working in Belgium on a project being run in the USA sometimes proved problematic, with different time differences an issue, but also involving data protection issues which require many administrative steps to be overcome. But the advantages of using the CAMP data as a given offset the disadvantages: ‘the beautiful thing about these data is that it is longitudinal. This is quite unusual. It takes time, it costs money, and it’s an effort for the study investigators and the kids, with their families. In other studies you usually only get a snapshot view of what is going on.’ And the statistical analysis was vital in countering some of the criticisms made of the study, a study which the researcher says needs replication and should be extended by also taking into account the genetic blueprint of the children. But that is the next step.
Disentangling the results and forming conclusions
Kristel Van Steen picks up the story of interpreting the results of the study both medically and statistically: ‘The CAMP study was primarily designed to assess the primary effects of inhaled anti-inflammatory medication on lung function in asthmatic children, not to specifically evaluate the effects of in utero exposures on treatment response. Thus, it is important to consider whether our findings could be explained by design-related biases. Three major issues of concern are statistical power, exposure misclassification, and residual confounding by both current environmental tobacco smoke exposure and differences in adherence between IUS exposed and unexposed children. Of the 150 children with reported IUS, only 39 were randomized to budesonide. Therefore, we first assessed whether our inability to detect improvement in airway responsiveness among children exposed to IUS was due to low statistical power. This is unlikely because although low power may limit our ability to detect a statistical difference between treatment arms, the observed mean difference in response in the exposed groupwas a fraction of that observed in the unexposed group. Furthermore, there was a protocol in place for open-label rescue use of inhaled beclomethasone. By the end of the clinical trial, both the nedocromil and placebo groups had significantly more rescue use of beclomethasone than the budesonide groups. This disproportionate use of supplemental inhaled corticosteroids by the non-budesonide groups would be expected to reduce the likelihood that we would detect a difference between treatment arms. With regard to exposure misclassification, we recognize that IUS designation was based on self-report, but if anything this might lead to an underestimation of the prevalence of smoking during pregnancy.’
The conclusions to be drawn from the study are the following. In utero smoke exposure reduces age-related improvements in airway responsiveness among children with asthma. Moreover, prenatal smoke exposure appears to severely blunt the beneficial effects of inhaled corticosteroid use on airways responsiveness, this medication being the primary recommended treatment for persistent asthma. There are important considerations for if children who are exposed to smoke in the womb respond less well to the various inhalers on the market then they might need other types of medication to be developed. Whilst more work needs to be carried out in analysing the action mechanisms at work in these processes, these results also obviously emphasise the importance of preventing this exposure through increasing the advice and support given to pregnant women, particularly by doctors and other health care workers, and ensuring that they do not smoke whilst pregnant.