The epidemiological evidence comes from almost every type of epidemiological design: case–control studies, ecological studies, large cross-sectional studies,[8,9] and large prospective cohort studies, in children and adults. To put some order and appropriately summarize these studies, several systematic reviews and meta-analyses have been published. In 2009, Etminan et al. reported a comprehensive systematic review and meta-analysis of the published literature on acetaminophen use and the risk of wheezing disorders in children and adults up to the end of October 2008 and after a thorough review and selection process, included 19 studies comprising 425,140 subjects, including children and adults (13 cross-sectional studies, four cohort studies and two case–control studies). Throughout the paper, it can be seen that point estimates of the odds ratios (ORs) always fell, in each and every study, against acetaminophen exposure and only in a small minority of studies does the confidence interval cross the line of no statistical significance. The pooled OR for all studies, for adults, for children and for almost any other subgroup of studies is consistently near 1.5, with a narrow 95% CI.
In 2011, García-Marcos et al. published an even more comprehensive systematic review and meta-analysis on early exposure to acetaminophen and allergic disorders, which included the studies of the meta-analysis by Etminan et al. (although appropriately dissected into substudies when appropriate) and some others published since 2008, totaling an outstanding number of 41 substudies. After several meta-analyses with different criteria, the authors concluded that, "no matter what type of study, age or time of exposure, there is always a positive association between [acetaminophen] exposure and respiratory symptoms." The pooled ORs obtained in this paper are also around 1.5. The consistency of the association found in almost every epidemiological study urged the authors to investigate the possibility of publication bias. Even after correcting for the possible unpublished negative results using the Duval and Tweddie's 'trim and fill' method, the pooled OR remained statistically and scientifically significant. In 2011 and 2012, some other epidemiological studies have shown the same results.[13,14] It is highly improbable that studies with negative results have remained unpublished because, from the journals editors' point of view, adding new studies to the large list of positive results is far less attractive than showing off a negative study.
Another feature of this epidemiological association which deserves special attention is dose-response relationship, which was already found in the first epidemiological case–control study and also in the two large ISAAC studies in adolescents and children aged 6–7 years[8,9] and a recent longitudinal birth cohort study in Ethiopia. For instance, in this study a dose-dependent increased risk of incident wheeze was found with acetaminophen exposure, with an adjusted OR for one to three acetaminophen tablets per month of 1.88 (1.03–3.44) and 7.25(2.02–25.95) for ≥ four tablets.
Expert Rev Resp Med. 2013;7(2):113-122. © 2013 Expert Reviews Ltd.