Recruitment and retention of farm owners and workers for a six-month prospective injury study in New Zealand: a feasibility study

A feasibility study simulating a six month prospective cohort study was designed and undertaken. The levels of farm and worker participation and retention were analysed to determine the feasibility of the methods for wider deployment.Recruitment levels were comparable with other studies, with 24% of farms and 36% of non-owner workers participating. Once recruited, retention was high at 85% and 86% respectively.The main challenges identified were in the recruitment process. Once recruited, farms and workers tended to complete the study, indicating that prospective studies in this the agricultural workforce may be feasible. Issues encountered and potential solutions for future studies are discussed.Agriculture is widely recognised as one of the most hazardous industries in both industrialised and developing countries [1]. In New Zealand, agriculture is among the top three industries for fatal and non-fatal occupational injury [2,3].Information available from descriptive epidemiological studies has highlighted potential avenues for reducing rates of injury in the agricultural sector [4,5]. However, in the early 1990s researchers noted a dearth of risk factor and detailed exposure information, and that this has hindered the development of properly informed injury control interventions [6-8].One facet of this has been the collection of time-exposed information on occupational exposures. Much of the literature has used persons-exposed exposure estimates to calculate injury rates. While these can help with identifying exposures with high injury burden, they can be substantially incorrect when attempting to ascribe risk. This was demonstrated by Nordstrom et al. [9] when they compared the injury rate ratios for males versus females when calculated using persons-exposed and time-exposed denominators. Using persons-exposed denominators, they found a rate ratio of 2.4. This dropped to 0.9 when using time-exposed denominators. If the persons-exposed exposure estimate had been u