The cross-sectional methodology

In Crete, there are no published reports on weed flora in vineyards. Garcia-Ortega et al, in a survey on occupational sensitization to Diplotaxis erucoides, found that workers in vineyards with rhinoconjunc-tivitis were sensitized to several pollens. Moreover, grape workers were found to be more likely to have skin disorders than were citrus or tomato work-ers, and these disorders may be causally associated with crop-specific exposures and lack of protective equipment. 

Therefore, it is possible that the high prevalence of atopy in grape farmers could be associated with their exposure to different irritants found in their workplace (ie, open-field cultivation and exposure to a variety of inhaled agents such as pollens, molds, mites, bacteria, and pesticides), rather than grape growing itself. We are well aware of the inherent limitations of the cross-sectional design of the present study. First, the study refers to a relatively small area of research in northern Crete. 

Furthermore, the cross-sectional methodology used in this study is not optimal for the assessment of causal relationships, but it suggests only the possibility of the association between the occupational exposure and the disease. Since we had statistically significant differences between the two groups in terms of age and smoking status, we used multivariate regression models to correct these imbalances.

 Selection bias may have affected the results observed in this study. First, a possible bias might exist in the reporting of symptoms, so that those with respiratory symptoms would be more likely to respond to the questionnaire and participate in the study. For that reason, the nonresponders from both groups were contacted by telephone, and they were asked about self-reported allergic rhinitis and asthma. 

These prevalence rates were lower than the prevalence rates of allergic rhinitis and asthma found in responders, and they could indicate the presence of a possible selection bias in the study. However, these differences could also be explained by the difference in the definition of the cases (self-reported disease vs clinical diagnosis based on questionnaire and medical examinations). We cannot rule out that the individual, social, and educational status of each subject had an impact on the study findings. However, we have no indication from our data of differential selection bias in the compared groups.