A Case-control Study of Farmwork-related Injuries in Olmsted County, Minnesota

AUTHOR ABSTRACT

There were many variables included in the analyses of this study and interpretation of the statistical significance assigned to them needs to be considered in light of the potential for making a Type I error, or concluding that the difference is real when in fact the difference occurred by chance. Since the significance level, which is the margin of error associated with a Type I error, was pre-set at 0.05, for every 20 statistical tests performed, the probability of at least one being statistically significant has been calculated to be 0.64 (Rothman 1990). As indicated below, three of the six significant findings relevant to the secondary hypotheses, and three of the five significant findings relevant to the secondary hypotheses would have remained significant if a significance level of less than or equal to 0.02 had been selected. The cost of increasing the P value, however, increases the chance for making incorrect conclusions of no differences (Rothman 1990). In addition, since all analyses were based on a prior hypotheses and the resulting significant findings were consistent with previous reported studies o findings that fit with nature, less concern can be given to these findings reflecting Type I errors instead of true differences.

The following relevant findings were identified through matched analyses of the data from the current study.

Primary hypotheses:

• A higher proportion of cases than controls reported the pre- existence of arthritis (p=0.02).
• A borderline significant finding was that a higher proportion of cases, than controls reported the pre-existence of stress, depression or other psychological problems (p=0.051).
• A higher proportion of cases than controls reported regular use of prescription pain medicine (p=0.046).
• A significant trend was indicated such than an increase in the number of farm hours worked was associated with an increase risk of injury (p=0.002).
• The association between lifetime involvement in farming on the basis of the four levels of exposure, "full-time year round", "full- time seasonal", "part-time year round", and "part-time seasonal", was significant (p=0.008), indicating the greater the amount of exposure, the greater the risk of injury.

Secondary hypotheses:

• Cases, compared with controls, had a higher index of injury history for farmwork related injuries that ever restricted their usual activities for any length of time (p<0.0001).
• Cases, compared with controls, had a higher index of injury history for farmwork related injuries for which they ever sought medical care (p<0.0001).
• A nearly significant finding indicated that cases reported more serious injuries in their lifetime than controls (p=0.053).
• Unmatched analysis indicated that there was a significant trend for an increasing number of acres in active production being associated with an increased risk of injury (p=0.018). Matched analysis, however, resulted in a non-significant finding except when examined by type of enterprise. The number of acres was significantly associated with the risk of injury for crop farmers (p=0.017) but not for livestock farmers (p=0.29).

Other findings

• Female cases were significantly more likely than males cases to incur a farmwork related injury to the back or spine (p=0.001).
• Female cases were significantly more likely than female controls to report one pre-existing medical condition compared to those respondents reporting zero pre-existing medical conditions (p=0.03).
• When farm variables were considered simultaneously in multivariate modeling, the only factor which remained significantly associated with the risk of injury was the number of hours spent farming per week.
• When select pre-existing medical conditions and medications were considered simultaneously with the number of hours spent farming per week, arthritis and the number of farm hours contributed separately and significantly to the risk of injury.

SOURCE: Minneapolis, Minnesota: University of Minnesota; 1990. n.p.

NLOM ID#: No ID#.