With an annual fatality rate of 27 per 100,000 workers, farming is the sixth deadliest occupation in the U.S., second only to sales/delivery transportation in the absolute number of occupational deaths [1]. Rates of non-fatal agricultural injuries are much higher than severe injuries, but all injuries combined can generally be expected to impact about 10% of farm workers each year [2,3]. Precise injury estimates are complicated by how injury data is collected, as farmers are known to delay medical care, save for the most acute health events [4,5]. Health hazards are unequally distributed across farming types/operations. For example, Pratt et al. [6] found that, after controlling for age and numbers of work hours, dairy farm owners appeared to be at particularly high risk with about 17 injuries per 100 dairy farm owners per year. Other studies have indicated farmers in dairy production are at somewhat lower risk compared to those in other agricultural areas, such as beef production [7].

How dairy farmers get injured has been examined in several prior studies. Use of heavy equipment remains the most common cause of fatal or disabling injury (e.g., tractor rollover, rotating shaft entanglement) [8]. Less severe dairy farm injuries are usually the result of falls and interactions with animals [9,10]. Other causes of injuries include chemical exposures, confined space (e.g., manure lagoons), and use of motorized equipment (e.g., all-terrain vehicles [ATV]) and power tools without personal protective equipment. Ways in which distal factors, including characteristics of the farm environment and farmers’ sociodemographic profile, might also influence dairy injuries are not well known. A recent meta-analysis of 32 studies, focused on all agricultural operations from around the world, found that males, full-time farmers, farm owners, those with a history of prior injury, hearing loss, regular medication use, and stress/depression were associated with ~1.5–2.0 higher odds of agricultural injury (usually self-reported) [11].

While nearly all agricultural injuries are theoretically preventable, identifying the modifiable factors that can reduce injury risk in dairy operations is a challenge. Developing a better understanding of injury determinants may be particularly important for dairy farmers due to the unique, technologically dynamic environment in which they work. Minimal research has been conducted on dairy injury antecedents beyond the most proximal causes (e.g., tractor rollover, animal assault). Methods that combine primary and passive data collection, via linking survey data on general dairy farm exposures to injury documentation in electronic health records (EHR), may offer advantages in terms of minimizing respondent burden and negating injury recall biases. There are burgeoning models of the secondary use of EHR data to capture various disease events and certain types of injuries [12,13], but such methods have only recently been tested to augment farm injury surveillance [14]. The purpose of this study was to examine associations between medically-attended agricultural injuries and: (1) personal sociodemographic characteristics, and (2) farm environment features and general safety practices. This was an exploratory study designed to inform future agricultural injury prevention research and safety initiatives, thus analyses were conducted without a priori hypotheses.

Surveys were mailed to all 1293 study-eligible dairy farmers in the target region, with 935 (72%) of them responding. There were 315 survey respondents excluded from the main analysis because they did not consent to having their survey responses linked to their EHR data, yielding a final analytical sample of 620 dairy farmers for the injury analyses. As outlined in Table 1, the sample was predominantly middle-age (4%were younger than age 30; 25% were age 60 or older) male dairy owners who live and work on their farm. Dairy farms in this target region tend to be mid-size with about 2–3 total workers per farm, and few large industrial operations as are commonplace in other parts of the U.S. [21]. In addition, dairy farms were generally not specialized, as there were considerable land and equipment resources dedicated to agricultural production (see Table 2). Comparisons between survey respondents and non-respondents on available personal characteristics in the EHR found that both groups had similar gender proportions, average age, and racial/ethnic background (data not shown).

Fifty medically-attended agricultural injuries were observed, which translates into an estimated 5.7 injuries per 1000 dairy farmers per year, a rate somewhat higher than that observed among all Minnesota farmers (dairy and non-dairy) [18]. About three-fourths of all injury cases were coded as farm accidents (E849.1). Univariate models found that household income, health insurance, farm residence, number of tractors, and worker safety training were significantly associated with agricultural injuries, and were thus tested for potential inclusion in the multivariable model (see Tables 1 and 2). In the final reduced multivariable model, the odds of agricultural injury were significantly greater among farmers who have private individually-purchased health insurance (odds ratio [OR]=4.25; 95%confidence interval [CI]: 1.31, 13.84), do not live at their dairy operation (OR=2.91; CI: 1.27, 6.67), and do not provide safety training to their workers (OR=4.27; CI: 1.00, 18.21) (see Table 3).

Dairy farming in north-central Wisconsin remains a risky occupation, with 8% of respondents having experienced a medically-attended agricultural injury over the previous 14 years. Prior research finds that such injuries are typically the result of machinery contact, animal assaults, or falls [8–10], but our findings also indicate that agricultural injuries are more common in dairy farmers who do not live at their dairy operation, do not provide safety training to their workers, and have a private individually-purchased health insurance policy.

These socio-environmental determinants of agricultural injury may have implications for farm safety programs and initiatives. As previously studied and advocated for [22,23]. safety is an important core component of training/education for farm owners and workers. Though we did not assess if respondents received safety training directly, nearly three of every four farmers in our study failed to provide safety training to all of their workers, thus highlighting the potential population-level value of targeting this modifiable risk factor. Why farm residence and health insurance type predicts agricultural injuries is less clear, but these factors perhaps suggest that some subgroups of dairy farmers are more apt to benefit from farm-level safety training or other injury prevention programs. Living on the farm was protective against injury risk. This may reflect more compact operations where there is less need for travel or use of various (dangerous) production implements. Having private/commercial health insurance coverage, presumably via a spouse’s employer or union plan, was also a protective factor as compared to an individually-purchased health insurance policy. Reasons for this are obviously quite speculative, particularly in light of the rapidly changing healthcare coverage environment in the U.S., but private insurance could perhaps be a general marker of economic vitality within farm families where there is dual income; one householder typically managing the farm and another householder working at a job where commercial health insurance is available for the family. This may free up resources to use, or otherwise more readily prioritize, injury prevention activities such as equipment protections.

It was somewhat surprising that other farm/farmer factors did not predict injuries. For example, full-time farmers and farm ownership were strong risk factors in a previous meta-analysis of studies not specific to dairy [11], but these factors did not predict injury risk in this sample of Wisconsin dairy farmers. Given their prominence in current farm injury prevention initiatives [24,25], use of tractor rollover protections was also not significantly associated with injuries in our study. The rather weak correlation between these factors and injury risk may have been partially a function of a major methodological limitation in our study; the temporality of exposure and outcome assessments. Exposures were essentially measured at study end and injuries were assessed looking back 14 years from that time point. This general approach, though common in agricultural injury research [11], assumes that exposures were reasonably representative throughout the whole study timeframe. But reverse causation is a possibility for some observations. For example, those who experienced a tractor rollover injury in the past may have been more apt to retrofit rollover protections on their current tractors, thereby obscuring the association between tractor rollover protection use (as measured today) and prior injuries.

The main study strengths were the objective capture of agricultural injuries and the defined source population. But ascertainment of agricultural injuries using E-codes alone, while practical and specific, may lack sensitivity. A recent surveillance model developed by Scott et al. estimated that E-codes alone only captured about 63% of all medically-attended agricultural injuries that occurred in Maine [26], a state where external cause of injury coding is not mandated. E-codes are mandated in Wisconsin and would presumably capture a higher fraction of all medically-attended injuries, but it is possible that some agricultural injuries can be missed without also using complementary methods such as reviews of ambulance records. Other methodological limitations were the homogeneity of the source population and the reliance on self-reports to assess most exposure variables, which may be subject to recall or self-presentation biases. Also, farm workers (who are not owners) and field fatalities that did not receive medical attention were generally not captured. Future research should attempt to confirm the associations observed in this study using prospective methods that include more non-owner farm workers and field fatalities, as well as direct observation of exposures wherever possible.

Though study findings were correlational, they reinforce prior observations that injuries to dairy farmers remain common and suggest that said farmers who do not live at their dairy operation, do not provide safety training to their workers, or have individually-purchased health insurance are more likely to be injured compared to those who live at their dairy operation, provide safety training, or have private/commercial health insurance, respectively. More research is needed to confirm these findings in prospective studies and to promote the general health and safety of dairy farmers.