Objectives: To estimate the incidence of specific IgE sensitization and allergic respiratory symptoms among UK bakery and flour mill workers; and to examine the roles of flour aeroallergen and total dust exposures in determining these outcomes.

Methods: A cohort of 300 new employees, without previous occupational exposure to flour, were followed prospectively for a median (range) of 40 (1-91) months. Cases-defined as those developing work-related symptoms or a positive skin prick test to flour or alpha-amylase during follow up--were compared with controls, matched for duration of employment. Exposures to flour aeroallergen and total inhalable dust were estimated using a questionnaire and personal sampling techniques.

Results: Incidence rates for work-related eye/nose and chest symptoms were 11.8 and 4.1 cases per 100 person years (py), respectively. Fewer employees developed positive skin prick tests to flour (2.2 cases per 100 py) or alpha-amylase (2.5 cases per 100 py). Positive skin tests to occupational allergens were more common among those with new work-related symptoms. There were clear relationships between the risks of developing work-related symptoms or a positive skin prick test and three categories of estimated exposure to total dust or flour aeroallergen. Atopic employees were more likely to develop a positive skin prick test-but not work-related symptoms. These findings were unaffected by age, sex or cigarette smoking.

Conclusions: In this population, many work-related symptoms which develop after first employment in modern UK bakeries or flour mills were not accompanied by evidence of IgE sensitization to flour or alpha-amylase. Although average dust exposures were within current occupational standards, the risks of development of upper and lower respiratory symptoms and of specific sensitization were clearly related to total dust and/or flour aeroallergen exposure. The incidence of work-related chest symptoms in the presence of a positive skin test to flour or alpha-amylase in this setting was approximately 1 case per 100 py.