Objective To develop a job-exposure matrix (JEM) that estimates exposure to eight variables representing different aspects of shiftwork among female workers.

Methods Occupational history and shiftwork exposure data were obtained from a population-based breast cancer case–control study. Exposure to light at night, phase shift, sleep disturbances, poor diet, lack of physical activity, lack of vitamin D, and graveyard and early morning shifts, was calculated by occupational code. Three threshold values based on the frequency of exposure were considered (10%, 30% and 50%) for use as cut-offs in determining exposure for each occupational code. JEM-based exposure classification was compared with that from the OccIDEAS application (job-specific questionnaires and assessment by rules) by assessing the effect on the OR for phase shift and breast cancer. Using data from the Australian Workplace Exposure Study, the specificity and sensitivity of the threshold values were calculated for each exposure variable.

Results 127 of 413 occupational codes involved exposure to one or more shiftwork variables. Occupations with the highest probability of exposure shiftwork included nurses and midwives. Using the 30% threshold, the OR for the association between phase shift exposure and breast cancer was decreased and no longer statistically significant (OR=1.14, 95% CI 0.92 to 1.42). The 30% cut-off point demonstrated best specificity and sensitivity, although results varied between exposure variables.

Conclusions This JEM provides a set of indicators reflecting biologically plausible mechanisms for the potential impact of shiftwork on health and may provide an alternative method of exposure assessment in the absence of detailed job history and exposure data.