Since our last systematic review [10
] and the ATS statement that followed it [11
], valuable new data relevant to the population burden of occupation in adult asthma have appeared. In particular, multiple new analyses based on large general population-based asthma incidence studies allow for more reliable estimates of PAR. These studies collectively yield a median value of 16.3%, quite close to the findings in previous reviews [10
]. Expanding the pool of studies to include a heterogeneous array of case-control and cross-sectional studies still results in an overall PAR estimate only modest different (17.6%), while limiting this to adult-onset asthma only (which includes all of the longitudinal analyses above, but also substantially increases the study pool) yields and estimate of 16.9%.
We excluded from this review occupational asthma incidence studies derived from surveillance data, given that sources markedly underestimate the proportion of cases attributable to work-related factors. Recent data from Finland indicate that, even after excluding officially recognized occupational asthma cases, excess risk of disease is still evident on epidemiologic grounds [52
]. The remaining risk is consistent with under-detection of one half to two-thirds of cases proportionally, even for well recognized risk groups such as bakers, fur workers, and painters. In contrast to under-reporting through surveillance, there is likely to be a bias toward over-attribution in clinical case series in which the PAR is derived from the ratio "probable" occupational cases with a denominator of all asthma cases identified in a registry, clinic, or hospital data base. In particular, case series that assign a case definition of occupational asthma solely because disease has occurred in a high risk job may overestimate the proportion of all cases that are work-related. We addressed this problem by excluding such case series in the integrated estimates shown in Table , although we acknowledge that case series with rigorous diagnostic criteria do indeed provide useful insights [53
We have not weighted our estimates taking into account the size of the study populations reported. We have, however, provided the number of asthma cases in each study type (Table ); these data indicate that the longitudinal general population studies have accounted for a major part (more than nine in ten) of the asthma cases upon which the PAR estimates have been based.
There are remarkably few studies, only five, presenting separate estimates for males and females. The median estimates with regard to gender were quite similar, but there is clearly a need for more studies that stratify by gender as well as other potential covariates that may be of interest, such as smoking and atopy. We have taken a simplistic approach regarding exposure classification consistent with a PAR approach. Thus we have not analysed the impact of more specific exposures such as flour dust or diisocyanates, nor have synthesized data from industry or occupation-specific studies that may be relevant to exposure-specific risk estimates.
There have been other recent reports that were not included in our analysis that may be relevant to broader questions of occupational factors in asthma across working groups. For example, a population-based case-control study from Finland provides insights on various occupational groups associated with increased asthma risk, but does not allow for combined risk estimates from which a PAR estimate can be derived [54
]. There have also been a number of studies addressing the relative frequency work-aggravated asthma, a subject beyond the scope of this systematic review [55
The analysis presented here yields an estimate of the PAR for asthma associated with work-related exposures that is quite consistent with past estimates. The range of the single estimates from each study is quite wide, but we consider a value of at least 15% and potentially as high as 20% to be the most accurate range of the likely population burden of asthma attributable to occupational exposures.
One key lesson clinicians should take from these data is that, when assessing patients of working age who have asthma, the occupational history should be carefully considered, in particular job duties held when the asthma first became manifest. In the same vein, the public health perspective should take into account the preventive implications of such findings. These data underscore the need for further actions to reduce the occupational exposure likely to lead to work-related asthma, on both the individual and population level.