The dangers of dairy farming The injury experience of 600 workers followed for two years.код для вставкиСкачать
American Journal of Industrial Medicine 21:637450 (1992) The Dangers of Dairy Farming: The Injury Experience of 600 Workers Followed for Two Years David S. Pratt, MD, Laura H. Marvel, BSN, Diane Darrow, BS, Lorann Stallones, PhD, John J. May, MD, and Paul Jenkins, MS In order to better understand the work-related injuries sustained on central New York dairy farms, we undertook a two-year population-based study of 600 farmers and farm workers on 201 dairy farms. During the observation period, 1984-1986, 151 persons had 200 injuries, giving an injury rate of 16.6%/year (166 injuries/1,000 workers/year). Men were injured more often than women (p 5 0.01). Injured workers were older (p 5 0.01), worked more hours (p 5 0.001), and had heavier workloads than noninjured workers (p 5 0.001). The growing and harvest seasons had the most injuries; winter the fewest. More than 2/3 of the injuries occurred in the afternoon. Owner/operators, often the most experienced, knowledgeable people on the farms, were most often hurt. Those working more than 60 hours/week, with greater than 30 acres under tillage/worker, had a relative risk of 2.76 compared with all other workers. The attributable risk for this group was 51%. There were two fatalities, both involved owner/operators. Our findings suggest that previous studies may have underestimated the risks faced by farmers. Dairy farming in central New York is very dangerous work. Those who own and operate these dairy farms are most often hurt and killed. Analysis of events on individual farms will be reported separately. 0 1992 Wiley-Liss, Inc. Key words: farm injuries, occupational injuries, agricultural exposure, dairy farming, injury epidemiology INTRODUCTION Rural America’s landscape is dotted with the familiar silhouette of family farms. Over the past three decades, economic changes and improved agricultural practices have resulted in a decreasing number of farms producing more goods and services [Census of Agriculture, 19871. Because agriculture is often associated with fresh air, safe, hard work, and robust good health, reports to the contrary are often surprising to those not familiar with the farm as a work place. Beginning in the 1950s, agricultural engineers and safety specialists reported information that began to paint a disturbing picture of real The New York Center for Agricultural Medicine and Health, The Mary Imogene Bassett Hospital, Cooperstown, NY. Lorann Stallones’ current address is the Department of Environmental Health, Colorado State University, Fort Collins, CO . Address reprint requests to David S. Pratt, M.D., Director, New York Center for Agricultural Medicine and Health, One Atwell Road, Cooperstown, NY 13326. Accepted for publication September 24, 1991. 0 1992 Wiley-Liss, Inc. 638 Pratt et al. danger on American farms. [Hoff, 1969; Jensen, 1980; Knapp and Piercy, 1966; Schnieder, 19801. These reports had some methodological limitations, but drew attention to potentially serious hazards for farmers. Numerous clinical reports from emergency rooms and other sources described farm injuries and their severity. [Cogbill and Busch, 1985; Doyle, 1984; Maxim et al., 1954; McKinnon et al., 1967; Powers, 19501. These reports served to support the agricultural engineers in their concerns. Well-designed surveillance of agricultural injuries is scant. Both the National Safety Council and the U.S. Labor Department’s Bureau of Labor Statistics report work-related agricultural injury rates. Each has limitations that are substantial. The Bureau of Labor Statistics samples farms with 11 or more employees (about 5 % of all U.S. farms) and reviews only the injuries that regulations require be kept on file (OSHA Form 200). The National Safety Council uses volunteers to report on a sample selected by a cooperating group or agency in each of 30 states. Recent criticism and recommendations from the National Research Council should help improve the data collection problems. [National Research Council, 19871. We were concerned that clinical reports and emergency room reviews omitted less serious, but important work-related illnesses, while the above noted surveys might not sample or collect data in a representative fashion. In order to deal with these apparent problems, we chose to study in detail the work experience on nearby dairy farms in Otsego County, New York. Our hope was to be able to accurately report the injury experience of a defined population of dairy workers, followed for two years. METHODS The Study Group The farms in the study were selected from the U.S. Department of Agriculture’s (USDA) list of all farms in Otsego County, New York. This list is based on the Agricultural Census done by the United States Department of Agriculture, and its annual subsample update. At the time of the enumeration in 1984, there were 540 farms with annual gross receipts of at least $10,000 operating in the county. More than 95% of these were dairy farms. All farms in the final study group were dairy operations. Every farm operator was contacted by mail and asked to participate in a twoyear study of agricultural injuries. We received agreements (verbal and written) from those responsible for 220 farms. We tested our sample of farms for selection bias by comparing their herd size with the herd size for all farms in the county. No difference was noted. Zip codes of participating farms were also plotted to evaluate geographic distribution. Study farms were distributed in each township of the county. Over a two-year observation period, 19 farms left the study. Eighteen farms left either because of financial hardship or government buyout programs. One farm no longer wanted to participate. We analyzed the size and worker characteristics of the farms that stopped providing information and found no bias introduced. At the completion of the study in 1986, information was available on 600 workers from 201 farms. A worker was defined as a person who worked at least 10 hours/week on the farm in any capacity and was older than 10 years of age. We used 10 years of age as a cut-off for farmers since no one under 10 years worked more than one hour per day and children under 10 years were not felt to fairly represent the Dangers of Dairying 639 worker pool. We used 10 hours per week as the lower limit of “workers” since this would only amount to about one average farmer’s day. Those working less than 10 hours a week could not be thought of as farmers. Data Collection Study farms were observed for 24 months between 1984 and 1986. Initial demographic information was obtained by mail questionnaire. The same information was reconfirmed during the second year by telephone. All workers on the farms were enumerated, as were their work hours. The characteristics of each farm were recorded, including size (in terms of animals and acres), line of business, receipts, and equipment in use. After the enrollment questionnaire, each farm was called every other month for two years. Two phone interviewers made all the calls. Calls were structured using a written statement. Phone respondents (usually the farm wife) defined all injuries on their own farms. No attempt was made, beyond the written statement, to prompt or discourage reporting of any job-related injury. No reports were excluded, except, for the purposes of this study, injuries in those who did not meet our definition of a worker. All work-related incidents were recorded. Injuries were often written on the family calendar, and farms were anticipating our calls. Information collected on each injury included person injured, age, sex, time, place, extent of injury, and consequences in terms of medical care and time off work. During the study period, over 3,200 phone calls were made. Information was entered into a data base on an IBM-XT PC. All data were edited after entry and checked for accuracy. Analysis was done using SAS software on a Digital Equipment Corporation Microvax machine. For categorical variables, all comparisons reported were made using chi-square. Analysis of continuous dependent variables, such as injury rates, acres under tillage, and number of co-workers, was performed using t-tests for dichotomous independent variables and analysis of variance for independent variables with more than two levels. A p value of less than or equal to 0.05 was taken as significant. Those variables found to be significant at the univariate level were entered into a multiple logistic regression equation with the absence or presence of an injury as the outcome. Stepwise procedures were then employed to obtain a reduced model containing only independent predictors of an injury. As a check on the accuracy of the reporting from the study farms, we reviewed the emergency room records of 70 injuries reported to have been seen at a local emergency room. We uncovered one additional injury in this way. In no instances were cases missing from the emergency room records if they reported having received care there. Injuries were scored for severity from one to four. Simple injuries not requiring medical attention or lost time from work were given a score of one, more serious injuries requiring medical attention and/or time off work were given higher scores, as explained in detail below. RESULTS Crude Injury Rates Two hundred injuries occurred among the 600 workers on 201 farms during the 1984-1986 observation period. The crude rate of injury was 166 injuries/l,000 640 F’ratt et al. 6 0.4 6 0.3 4 3 Injuries per 2 1,m hours 1 worked Injurles per loo workers 0.1 0 0 11-20 21-30 31-40 41-50 51-60 61-70 71, Age Inlury Rate RaW1000 hrs. worked Fig. 1. Injury rates adjusted for farmers’ age and mean hours worked. workerdyear, or 16.6%. Men had an overall injury rate nearly three times higher than women. Persons There were 151 individuals injured during the study. The majority of workers (79%) were injured once; twenty-one workers (14%) were injured twice. Ten additional workers had three or more injuries, including one man with six events. Women workers numbered 137; they sustained 19 injuries for an annual rate of 69/1,000 or 6.9%. The 463 men in the study had 181 injuries for an annual rate of 195/1,000 of 19.5%. The women were injured at a rate about one-third that of men (rate ratio of 1:2.8) when workers at risk was used as the denominator. This difference was highly significant (p 5 0.001). In order to explore the male/female disparity further, we calculated injury rates by mean numbers of hours worked. Women had 1.59 injuries/l ,OOO workerdhour and men 2.98 injuries/l ,OOO workerdhour. This difference is significant (p 5 0.05) and confirms the unadjusted trends. Age-specific injury rates were examined to look for special risk groups. Figure 1 shows that injuries are highest in the 31-40 and 51-60 age groups. Age-specific rates, though helpful, need further correction for exposure to give a full picture. Figure 1 also shows injury rates for worker groups corrected for reported hours of exposure. This figure demonstrates the persistence of peaks in 31-40 and 51-60 age-groups. The pattern of injury by decade remains similar after the adjustment for hours of exposure. In all but five injuries, we were able to identify the type of worker involved. Examination of the injury pattern in each worker group produced some striking results. The owner/operators made up about one-third of all workers, yet had 62% of the injuries followed by relatives (24%) and hired men (14%). As shown in Table I, the increased number of injuries is explained in part by the size of this group and its Dangers of Dairying 641 TABLE I. Injuries by Worker, Rate on the Farm, and Population at Risk (Otsego County, NY) Owners Relatives Hired men Injuries" Population at riska Mean weekly hours at risk Injury rate/ population Injury/hours 122 46 27 203 292 100 79.0 51.7 53.7 0.60 0.16 0.27 1.54 0.89 0.50 "Five of the 200 injuries were incurred by people in which the type of worker was unknown. We were unable to contact these farms for verification of worker type. high number of hours on the job. When we corrected for these factors, we still found that owner/operators had the highest injury rate (p I0.05). The gender, worker type, and age variables were highly related, in that 99% of the ownedoperators were men and 73% of these were in the fourth, fifth, and sixth decades of life. Multiple logistic regression procedures revealed that sex and age do not contribute significantly to the probability of an injury in an equation which contained the type of worker. Thus, the relationship between age, gender, and injury rates may be affected by the preponderance of injuries among the owner/operators. Place The dairy farm consists of a wide variety of work environments. In order to explore the dangers of these different locales, we examined injuries by their place of occurrence. Figure 2 shows the location of the worker when the injury occurred. The barn was obviously the most common location for injuries (p 5 0.002). This frequency of barn injuries is not surprising, given the nature of dairy farming and the limited field work that goes on during the five winter months (November through March). Another factor in the high number of barn injuries is the type of work done in this environment. It is in the barn where the dairy worker interacts with both machinery and animals-an average of 50 milking cows twice daily, 365 days per year. As shown in Table 11, it was the interactions with machinery, followed by those with animals, which most commonly caused injury. Time Although very few farms reported the exact hour of an injury, 93% were able to report that an injury occurred in the morning or afternoon. Two-thirds of the injuries took place in the afternoon. Injuries occurred every day of the week. No day had a significantly increased number when compared with other days of the week. Only 60% of respondents could recall the day of the reported injury. Injuries were much more common during the growing seasons. This difference was significant (p 50.01) (Figure 3). Injury Detail Body part injured is shown in the human form diagram (Figure 4). It can be readily seen that the extremities were the most frequently injured parts of the body. In general, illnesses like colds and urinary tract infections were not reported as work-related and hence do not appear in the data base. In our chest/lung category (13% of all injuries), 52% were trauma related (rib fractures, usually), while 48% 642 Pratt et al. 60% SOX 40% 30% 20% 10% 0% Barn Field Sllo Barnyard Other Shop Location Fig. 2. Percentage of injuries by location on the farm. TABLE 11. Farm Injuries by Causative Factor (Otsego County, NY Dairy Farms, 1984-1986) Machinery Animals Other Falls Respiratory Chemical 35% 32% 16% 8% 7% 2% were ailments like silo filler’s disease and organic dust toxic syndrome (0DTS)both of which could be considered work-related injuries. Table I11 outlines the severity of the self-reported injuries based on medical visits and lost time from work. Seventy percent sought medical care, demonstrating that few trivial injuries were reported. The injuries sustained were often disabling; 31% of those injured lost time from work. Eighteen percent were serious enough to put people off work for four or more days. The average lost time was 10.7 days; the median was eight days. Having seen the high rate of injuries for those in the 3 1-40 age-group, we were interested to see if they had more severe injuries. No greater severity in this age-group was found. Likewise, we noted no increase in disability or severity analyzed by age or sex. Farm Characteristics Associated With Injuries Injured persons tended to work on larger farms with more acres under tillage (p 0.001), but with fewer co-workers (p 5 0.001). They also worked more hours than non-injured persons (p 5 0.0001). The long hours and smaller crews suggested that 5 Dangers of Dairying 643 20% 15% 10% 5% 0% Wlnter - (DJF) Month Sprlng (MAMI Summer' (JJA) Season (p<O.Ol) Fig. 3. Percentage of farm injuries by season. individual workloads were important. Two ratios were set up to examine this workload hypothesis. These ratios were: workers to milking cows, and workers to acres under tillage. On most dairy farms that grow their own forage, these ratios are, by necessity, proxies for one another. Both ratios were highly and directly associated with injuries (p 5 0.0001). Owner/Operators After age and exposure adjustments, owner/operators emerge from our data as being at particular risk for injury. These men and women worked longer hours, were older, and, even when injured, seldom missed days of work. Only 27% of injured owners took a day off, whereas 52% of hired help took days off with injuries. Detailed analysis showed that owners working more than 60 hours per week, and with farm workloads greater than 30 acres per worker, had a relative risk of 2.78 compared with all other workers. They had an attributable risk of 51%, based on the above characteristics. Seasonal Patterns Earlier in the Results section, we described the injuries by location on the farm. Figure 5 shows the increase in field injuries as the work becomes more intense in the spring and early summer. The field injuries rise again (apparently) with haying and the taking in of corn in the fall. Barn injuries, by contrast, appear more evenly distributed throughout the year. Another aspect of the patterns illustrated in Figure 5 confirms our clinical experience. Here we can see a gradual increase in silo injuries as the year draws closer to the corn harvest. In November, a full 25% of all silo-related injuries occur. 644 Pratt et al. Fig. 4. Body location of farm injuries. Deaths During the two years’ study, two men lost their lives. The first man, age 39, died when a tractor, with a front end loading device, turned over and crushed him. His tractor was not equipped with a rollover protective structure. He was found several hours after the incident and pronounced dead at the scene. The second death occurred when a 52-year-old farmer suffered a fatal heart attack while hammering on a corn picker in his barnyard. He had had a prior history of coronary artery disease. He died at a local hospital within hours of his infarction. These two events are described, but not further analyzed. Two events do not allow fair estimates of fatality rates on our farms. Dangers of Dairying 645 TABLE 111. Injuries by Severity Score for 1984-1986 Observation (Otsego County, NY) Score number Number of iniuries Definition 59 No medical care, no lost time Sought medical care, no lost time Sought medical care, 5 3 days lost Sought medical care, 2 4 days lost 1 2 3 4 % 30 39 13 I1 21 31 18 30% i.. ... * ... ... ...... .... .,.(.. ..... .._.. .. .. , . . .. ... ... . . ; j. ’ . . . 20% .. 16% 10% i 6% 0% J F M A M J J A S O N D Month +Barn *Fleld ..*,. Silo Fig. 5 . Percentage of injuries by farm location and month. DISCUSSION Our results show that a surprisingly high rate of 166 injuries per 1,000 workers occurred each year (16.6%). Since some workers sustained more than one injury, the rate for individual workers injured is somewhat lower, but still quite high at 12.5% of all workers injured per year. Thus a worker has a 1 in 8 chance of being injured each year on the job. Occupational injuries are reported differently by various sources as shown in Table IV [Jansson, 1987; Michigan Department of Public Health, 1989; National Safety Council, 1989; U.S. Department of Labor Bureau of Labor Statistics, 19831. The National Safety Council’s 1989 Accident Facts reported the rate of injuries on dairy farms in 35 states based on voluntary study data. They estimate a rate of 22.3 farm work injuries per 1,000,000 hours on the job. After correcting our rate to meet the NSC definition of an injury: “. . . an injury . . . requiring professional care or causing one half day’s lost work . . . ,” we had 42.7 injuries/l,000,000 hours worked, or about 2.0 times their estimate. This difference may be the result of the method of active surveillance we used and a maximum 60-day recall, as opposed to their 90-day recall and volunteer surveyors providing reports. We chose a more liberal definition of an injury based on our clinical experience of farmer stoicism and doctor-avoidance. Had we used the National Safety Council’s 646 Pratt et al. TABLE IV. Comparison of Other Data Sources With Otsego County Dairy Farms Injuries/] O6 hours workedlyear Sweden 1983 Michigan 1987 National Safety Council 1988 (Dairv) Otsego County, New York 1984-1986 24.6 28.5 22.3 42.7 TABLE V. Variations in Iniurv Definitions Pratt et al. Injury definition Any selfdefined acute illness episode National Safety Council Michigan Jansson BLS Episode requiring professional care or causing f i day’s lost work Same as National Safety Council Episodes requiring care in hospitals or emergency rooms Self-defined reported according to OSHA standards definition, many important injuries would have been omitted. A good example is a farmer who was sprayed with caustic hay preservative, burning both his hands and face. He neither sought care nor lost time from work. In other settings he might well have done both. We feel his was an important injury to report. A recent comprehensive one-year study from Sweden reported an injury rate of 24.6 injuries/1,000,000 hours worked [Jansson, 19871. This figure is very close to the NSC estimate, but well below the rate we report. The Swedish estimate is based on injuries requiring medical care-which is available at no charge to injured workers. Their reporting relied generally on emergency room records. The Swedish farms involved included diverse agricultural activities, including dairying, beef production, pork, and feed grains. This diversity may help explain the lower rate. The U. S. Department of Labor reports injuries in American business as well. In the last published Bulletin, they reported injuries in all productive agriculture (SIC codes 01-02) for 1981 to be 12.8 injuries per 100 full-time workers [U.S. Department of Labor Bureau of Labor Statistics, 19831. This rate is calculated on the basis of a 40-hour week. The injury case definition used by the Bureau of Labor Statistics (BLS) is similar to ours. When we adjust our data for full-time workers (77% of the work force), we find a rate of 8.24 injuries/lOO workerdyear. This rate is smaller than the BLS all farm estimate. The small farm (1 1-19 workers) BLS figure, however, is quite a bit lower at 6.3 injuries/lOO workerdyear. Unfortunately, the BLS does not collect any information on farms smaller than 11 workers (roughly 95% of all American farms). Recently, the Michigan Department of Public Health published a one-year study of injuries in two counties [Michigan Department of Public Health, 19891. They used written questionnaires and required farms to report injuries that met the NSC definition. This survey covered over 2,000 workers on 785 farms. The estimated rate of injuries for this farm group was 50/1,000 workerdyear. Information was obtained on only two injuries on any given farm during the study period. Adjusting our rate to their methods, we find 108 injuries/l ,000 workers/ year. This is nearly twice Michigan’s and may be affected by the fact they used a Dangers of Dairying 647 one-year recall period, limited case reporting, and surveyed several kinds of agriculture, including grains, dairy, and others. The Michigan group also reported their incidence in the injuries/l ,000,000 format. This is shown in Table V along with other estimates cited. These varied methods of data collection, injury definition, recall period, and rate reporting argue strongly for a unified system of defining and following agricultural injuries. Improving the quantity and the quality of farming injury data would help considerably in planning intervention strategies and evaluating them. We were surprised to see the marked difference in injuries to men compared with women (2.8:l). While, in general, women are less often injured than men, we did not anticipate the high ratio of men to women. In data presented by the National Health Interview Survey (NHIS), they found injury rates for white women to be comparable to white men in agriculture. Black women, however, had an injury experience similar to that seen in our figures. The NHIS black men to women ratio was 3.9:l [Kaminski and Spirtas, 19801. Injuries seen in emergency rooms in Sweden, reported by Jansson, showed a sex ratio of 6.76:l male to female. [Jansson, 19871. One possibility is that women do fewer risky tasks on the farm. We did not find that women differed appreciably from men when we examined the type of injuries they sustained, where they were injured, or the severity of their injuries. Another, more interesting possibility, is that women are safer workers. Injury rates on the job by sex are very hard to locate beyond the NHIS figures described above. One of the most interesting published studies on causative or explanatory factors in work place injury came from Kriebel [ 19821. He identified a number of factors which helped explain differences in injuries among various kinds of work. One of his postulates is that injury frequency is explained in part by rate of new hires. This factor would reflect the relative number of inexperienced workers on the job at any one time. To test this hypothesis in the study group, injury rates were examined by age decile. In contrast to Kriebel’s hypothesis, we found that injury rates were highest in the 3rd and 5th decile. These findings are similar to Jansson’s experience in Sweden [Jansson, 19871. Another factor of importance in Kriebel’s work is the correlation of energy consumed per worker with injury severity. We found that injuries associated with machinery were more severe. While only 50% of animal injuries required a visit to a health provider, 80% of those associated with machinery necessitated medical attention. Thus, in this instance, we have evidence to support Kriebel’s observation. We were very interested to note that the people who own and operate the farms are most often hurt. This certainly has potential economic significance. Our observation was shared by Sell and others in a description of power-take-off (PTO) injuries. [Sell et al., 19851. The relative risk of hard-working owner/operators was 2.76. Owners with high-risk characteristics had nearly three times the likelihood of injury compared with other workers. This finding has important implications. The owner/ operators are the most experienced individuals in farming today. Their risk suggests that the best of the work force are at the highest risk-a very serious situation. An even greater concern arises from the recent trend in American farming toward larger farms, more cows, more mechanization, and fewer workers. These trends for the Northeast are shown in Figure 6. The economic trends are moving dairy farming in New York toward increasing risk for workers [New York Department of Agriculture and Markets, 19901. The occurrence of disabling injuries on small farms is a serious problem. Given 648 Pratt et al. INCREASE IN LARGE HERDS ON DAIRY FARMS IN NEW YORK 1984 - 1988 0.370 0.350 0.330 0.310 0.290 0.270 0.250 4 I 1984 1985 1986 1987 1988 Years CHANGE IN NUMBERS OF ALL NORTHEAST AGRICULTURAL WORKERS 1984 170r 160.- 0 \ ' 0 150-140130 - 1988 \ -- 120 -1104 i 1984 1985 1986 1987 1988 Years Fig. 6. Increase in large herds on New York dairy farms contrasted with the decrease in numbers of all Northeast agricultural workers. the fact that owner/operators were most often hurt and that 20% of all injuries entailed four or more days of disability, it seems that the injuries we reviewed were likely to have serious economic consequences. Just over 30% of the injuries had attendant lost time. This is considerably smaller than the Swedish experience, where they had 46%. This may be explicable, in part, based upon the insurance scheme used by the Swedes to compensate injured farm workers [Jansson, 19871. The finding that most of our injuries occurred in the afternoon is consistent with many other reports of both fatal and non-fatal farm injuries [Hoff, 1969; Jansson, Dangers of Dairying 649 1987; Pugh et al., 1974; Field and Toemehlen, 1982; Goodman et al., 19851. Most authors have postulated fatigue as explanatory of afternoon injury rates. CONCLUSIONS Our investigation strongly supports other information suggesting that agricultural work injuries are frequent, severe, and often disabling. It does appear that women in our dairy group have fewer injuries than men. This experience has been reported for much of injury research [Baker et al., 19841. New injuries on our farms follow several trends and tend to occur more frequently when the workloads are greatest. Other studies support this finding [Michigan Department of Public Health, 19891. The serious plight of the owner/operator deserves special mention. Our information suggests that these men and women work almost 80 hours per week and have the highest risk on the farm. Both of our deaths were owner/operators. The owner/ operator is the linchpin of America’s family farm. Richard Rhodes’ recent fictionalized account of a year on a Missouri family farm repeatedly shows the skill, knowledge, and dedication of these people [Rhodes, 19891. It seems to us that the social situation described in our study deserves public review. A recent study done by economists at Cornell reported that returns to labor and management were negative on 31 % of New York dairy farms. The mean income per operator was $1 1,042 per year [Smith et al., 19871. If this were representative of our owner/operator’s return, they would be earning $2.70 per hour. One could conclude that the dairy farmers are keeping the price of food low by keeping labor costs down (his or her time). Our data suggest that these long hours result in injuries and illness. Are family farmers providing a de facto subsidy of our food prices at a personal cost in injuries, illnesses, and deaths? Clearly, further study is needed, but a dark picture of family farming is emerging, which should cause all of us to pause and reflect on the “inputs” of our food system and what their value represents. The analysis of the experiences of individual farms will be reported in a separate manuscript. We have purposely avoided addressing farms as a unit of analysis for this paper to the extent possible. ACKNOWLEDGMENTS This work was supported, in part, by funds from the Stephen C. Clark Foundation and by the people of New York State. REFERENCES Baker SP, O’Neill B, Karpf R (1984): “The Injury Fact Book.” Lexington, Massachusetts, Lexington Books, p 313. Census of Agriculture (1989): Bureau of Census. U.S. Department of Commerce, Washington, D.C. Cogbill TH, Busch HM (1985): The spectrum of agricultural trauma. J Emerg Medicine 3:205-210. Doyle Y (1984): Farming and industrial accidents in the Midlands during 1981: A pilot study. Ir Med J 11~211-283. Field W, Toemoehlen RL (1982): Analysis of fatal and non-fatal accidents involving children. Paper 82-5501, St. Joseph, Missouri: American Society of Agricultural Engineers. 650 Pratt et al. Goodman R, Smith JD, Sikes RK, Rogers DL, Mickey JL (1985): Fatalities associated with farm tractor injuries: An epidemiologic study. Public Health Rep 100:329-333. Hoff DR (1970): Accidents in agriculture: A survey of their causes and prevention. Cooperative Extension, Cornell University, Ithaca, New York, Information Bulletin 1. Jansson B (1987): The yield of systems for continuous and periodic injury surveillance in emergency care with emphasis of farm work-related accidents. Scand J Soc Med 15:247-252. Jensen DV (1980): A summary of total farm accidents in Wisconsin from 1944-1978. In “Engineering a Safer Food Machine.” St. Joseph, Missouri: American Society of Agricultural Engineers. Kaminski R, Spirtas R (1980): Industrial characteristicsof persons reporting morbidity during the health interview surveys conducted in 1969-1974 An explanatory review. Cincinnati, Ohio DHHS (NIOSH) Publ. NO. 80-123. Knapp LW, Piercy L (1966): An epidemiological study of power take-off accidents. Bulletin 10, Institute of Agricultural Medicine and Environmental Health. Iowa City, Iowa: The University of Iowa. Kriebel D (1982): Occupational injuries: Factors associated with frequency and severity. Int Arch Occup Environ Health 50:209-218. Maxim EW, Webster FS, Willander DA (1954): The cornpicker hand. J Bone Joint Surg 36A:21-29. McKinnon DA, Robinson DW, Masters EW (1967): Haybaler injuries. J Trauma 7:265-274. Michigan Department of Public Health (1989): Farm injury in Michigan. Lansing. National Research Council (1987): “Counting Injuries and Illnesses in the Workplace: Proposals for a Better System.” Washington, D.C.: National Academy of Sciences. National Safety Council (1989): “Accident Facts.” Chicago. New York Department of Agriculture and Markets (1990): “Agricultural Statistics 1988-1989.” Albany. Powers JH (1950): Farm injuries. N Engl J Med 243:979-983. Pugh AR, Stockey WE, Phillips GH (1974): Accidents to farm and rural non-farm people-1972. Research Bulletin 1069, Wooster, Ohio: Ohio Cooperative Extension Service. Rhodes R (1989): “Farm.” New York: Simon and Schuster. Schnieder RD (1980): The Nebraska farm accident example 1970-1979. In “Engineering a Safer Food Machine.” St. Joseph, Missouri: American Society of Agricultural Engineers. Sell WE, Campbell WP, Field WE (1985): Summary of on-farm F’TO related accidents. Paper #85-5513. St. Joseph, Missouri: American Society of Agricultural Engineers. Smith SF, Knoblauch WA, Putnam LD (1987): Dairy Farm Management Business Summary, New York, 1987. AE Ris 88-8 Cornell University, Ithaca, New York Department of Agricultural Economics. U.S. Department of Labor Bureau of Labor Statistics (1983): “Occupational Injuries and Illnesses in the United States by Industry, 1981.” Washington, D.C.: Government Printing Office.