Materials and Methods

Sampling frame and selection of participants—Colorado and Kansas were selected as the region to collect survey data. To develop a survey mailing list, producer lists were generated for beef cattle, dairy cattle, swine, and small ruminants through communication with livestock producer groups in Colorado and Kansas. These included the Kansas Livestock Association, Colorado Livestock Association, Colorado Cattlemen's Association, Kansas Pork Association, Kansas Sheep Association, Colorado Wool Growers Association, Kansas Meat Goat Association, and Kansas Farm Bureau. From January 24, 2011, to January 31, 2011, all members of the Kansas Livestock Association, Colorado Cattlemen's Association, Kansas Sheep Association, Colorado Wool Growers Association, and Kansas Meat Goat Association were sent letters of invitation via postal mail on their respective association letterhead to participate in the survey; the Kansas Pork Association and Colorado Livestock Association elected to use Kansas State University and Colorado State University letterhead, respectively, for their members' survey invitations. The Kansas Farm Bureau included a column about the survey in their electronic newsletter and provided information on their website for electronic survey sign-up.

Additionally, an article was published in the Western Dairy Newsletter,²⁶ focusing on FMD and introducing the study Further publicity was generated through website announcements and Colorado and Kansas extension emails. A Livestock Contact Survey website was developed and publicized, providing information for producers about the survey as well as an electronic sign-up form. To reach as many producers as possible, flyers were distributed in Colorado at 2 large auction markets (High Plains Livestock Exchange in Brush, Colo, and Centennial Livestock in Fort Collins, Colo; January 2011), the National Western Stock Show (Denver; January 2011), and Dairy Day (Greeley, Colo; January 26, 2011) and through 4-H extension agents throughout the state. Personal visits to sheep, goat, and cattle producers in northern Colorado were made by one of the investigators (MS) accompanied by a private veterinarian (selected on the basis of interest in the project and the type of producers they had as clients). In Kansas, invitation letters to participate in the survey were mailed to operations in the state that had a Confined Animal Feeding Permit, operations that received a commodity milk payment, and members of the Local Harvest organization, an organization of farmers selling their products directly to the public. The Kansas Livestock Association sent out 3,650 letters, the Kansas Pork Association 460, and the Colorado Cattlemen's Association and Colorado Wool Growers Association 1,615; an additional 2,000 letters were sent out from the remaining lists. Duplication of invitation letters was highly likely because multiple organizations and mailings were involved. Producers were asked to volunteer to participate in the survey by returning a postcard with their name and preferred communication information. To ensure that information collected during the study remained anonymous, communication information and survey responses were kept separate throughout the study.

Survey questionnaire—Approval for the survey was obtained from the Kansas State and Colorado State University Review Boards for Research Involving Human Subjects. A 6-page questionnaire^a was developed to collect contact data from producers. The questionnaire was mailed to 6 producers in Colorado and Kansas to pretest it for clarity. After the suggested revisions were made, the questionnaire (which consisted of 17 multiple-choice and open-ended questions) was sent electronically or by postal mail, according to respondent preference, to all enrolled participants. The mailings included a cover letter, the 6-page questionnaire, and a postage-paid return envelope. For participants who requested to receive the questionnaire via email, a cover letter was attached to the email along with the questionnaire. The same survey was sent on a quarterly basis to all participants (in March, June, September, and December, 2011) to capture variation in movement throughout the year. Every participant was sent the survey each quarter, regardless of whether they responded to the previous surveys. A reminder email or postcard was sent to nonresponders approximately 4 weeks after each survey was sent.

Classification of type of operation—Participants in the survey were asked the current number and type of livestock at the operation as well as the primary type of operation. Responses from participants who returned a survey but did not own any cattle, pigs, or small ruminants were not included in the survey analysis. Because some participants selected > 1 type of operation, a classification system was used to assign an operation type to all participants. Participants' operations were categorized into 9 types on the basis of the operation description chosen on the survey and type and number of livestock in the operation. Operation types were classified according to the types of contacts within the livestock production system. For example, large and small feedlots were separated to represent backgrounding (< 3,000 livestock) and finishing feedlots (> 3,000 livestock), which fit into different locations in the production chain and thus would be expected to have different types of contacts. Similarly, large swine operations were likely part of an integrated swine business, so interactions would be different from those of small swine operations.

Cow-calf operations included all operations that described their operation as commercial cow-calf or beef seed stock. Small cow-calf operations had < 100 cattle, and large cow-calf operations had > 100 cattle. Small feedlots included all operations characterized as stocker grazer, beef backgrounder, or cattle finishing feedlots that had < 3,000 cattle. Large feedlots included all cattle finish feeder operations that had > 3,000 cattle. If a participant selected > 1 type of operation, the type with the greater number of livestock was selected as the assigned operation type. Operations with mostly beef cattle that selected commercial cow-calf operation along with another type were assigned a cow-calf operation type. Dairy operations included seed stock producers, commercial dairy operations, and calf raisers. The limited number of dairy participants precluded segregation into small and large herd categories. Sheep, dairy goat, and meat goat operations were combined into a small ruminant operation type. Swine operations with < 1,000 pigs and > 1,000 pigs were categorized as small and large swine herds, respectively. Because 34 of 532 (6.4%) participants reported having swine and beef cattle, we added a mixed beef and swine operation type to the analysis. Operations that included both species and for which > 40% of the livestock were either beef or swine and > 11% were of the other type were classified as mixed beef and swine operations.

Estimation of direct contact rates and distance of contacts—Participants were asked to record all incoming and outgoing shipments of livestock (direct contacts) for the 7 days after the survey was received. The survey included 2 pages for recording direct contacts, including the date of movement, species moved, number of animals, source of livestock, destination of livestock, and distance traveled. For the source and destination, a brief description of the type of location was requested (eg, auction or feedlot). An additional page was provided for cow-calf and sheep producers to record livestock movements for the previous 3 months. This was done to allow more accurate assessment of contacts in those operations because they were expected to have fewer movements overall than were other types. The record included month of movement, species of livestock, number of shipments, approximate distance the livestock were moved, and destination and source of animals.

For all operation types, movements to pasture or grazing, pens, headquarters, stock fields, crop fields, and calving pens that were < 11 km (< 6.9 miles) were classified as movement within the operation. Those that were > 11 km were included in outgoing direct contact movements because of the distance traveled and the increased concerns of contact with livestock from other operations. Other reported movements could not be clearly classified for contact with a particular operation type: movements of livestock to a veterinary clinic, for semen testing, or for embryo transfer were classified as veterinary visits, and movement to rodeos, petting zoos, fairs, and shows were classified as visits to a show. Contacts were not considered in our analysis if the destination or source listed only a city or state because the type of contact was unknown. The number and proportion of excluded contacts were recorded. The contact rate was calculated by counting the number of contacts with each participant's operation for each destination-source combination. The total number of contacts was divided by the interval of time covered by the survey to obtain the daily direct contact rate. Daily rates generated for movements over 7-day and 3-month intervals were combined, and a mean number of contacts per quarter (ie, mean quarterly contact rate) was calculated by multiplying the daily rate by 91 for each livestock type category.

Estimation of indirect contact and distance of contacts—Indirect contacts were defined as contacts between operations other than livestock movements, including movement of persons and equipment or materials (eg, vehicles or feed). Survey participants were asked to record the number of visits by indirect contacts expected to regularly interact with livestock operations and the approximate distance that was traveled (from the visitor's base of operations) to reach their facility. The survey listed 28 types of visitors or vehicles; a free-form entry was also provided so that respondents could record any potential indirect contact that was not included in the list. All participants were asked to complete a record for the 7 days following receipt of the survey, and cow-calf and sheep producers were also asked to record the number of visits by each type of visitor and approximate distances that each traveled to reach their operation during the previous 3 months. The 7-day and 3-month records were divided by the time interval to calculate daily indirect contact rates, and the combined rates were used to calculate the mean quarterly indirect contact rate for each contact source and livestock type category combination in the same manner described for direct contacts.

Statistical analysis—Descriptive statistics (counts, percentages, and mean and median data) were calculated for survey responses and 10th and 90th percentiles were calculated for herd sizes and contact rates with commercially available statistical software.^b The data were evaluated for normal distribution with a frequency histogram and a Shapiro-Wilk test and were nonparametric.

Results

Response to survey—A total of 2,400 surveys were sent out to producers in Colorado and Kansas who volunteered to participate in the study, with 1,136 (47.3%) surveys returned from 532 unique operations. Six of 1,136 returned surveys were excluded from analysis because no livestock were on the premises.

Most (65.6%) of the 1,130 surveys included in the analysis were from Kansas, with 742 responses received from 391 unique participants. The remaining 34.3% were received from Colorado, with 388 responses received from 141 unique participants. Surveys were returned from 93 of 105 counties in Kansas and 42 of 64 counties in Colorado. The first quarterly survey (for December 2010 through February 2011) had the highest overall response rate, with 354 of 628 (56.4%) surveys returned. The highest response rates for each state were also found for the first survey (Table 1). Numbers of unique participants for each operation type are summarized (Table 2). Cow-calf producers comprised 305 of 532 (57.3%) of the operations that participated in the survey. The survey was completed by the owner of the operation for 1,011 of 1,136 (89%) surveys and by the manager for 79 (7%); 40 (3.5%) surveys were completed by individuals who had another role in the operation, and this information was not provided for the remaining 6 (0.5%) surveys. Herd size distribution for various livestock operation types is provided (Table 3). The mean beef cow herd size was 369 cattle.

Direct contacts—Although equine movements were not requested, 182 outgoing equine movements were reported: 160 (88%) from cow-calf operations and 22 (12%) from small feedlot and mixed beef and swine operations; these were excluded from the contact analysis. In addition, 38 of 4,742 outgoing movements and 483 of 3,219 incoming movements were excluded from analysis because the type of contact was left blank or no specific operation type was recorded for the source or destination. These movements accounted for 6.5% of 7,961 total direct movements reported for the year. Of these, 320 movements for the small swine operation type were excluded because the source for incoming livestock was left blank, and 103 movements were excluded for 1 small ruminant operation that had livestock brought in from out of state with no further details provided. These reported contacts could not be categorized according to the source and therefore could not be accurately included in contact calculations.

The mean number of outgoing direct contacts during each quarter varied within and among operation types (Table 4). Mean values were reported because these are used as input parameters for disease modeling, and the median value was sometimes 0. Large swine operations had the highest outgoing direct contact rates for all quarters, with other large swine operations as destinations (range of mean values, 5.9 to 24.53/quarter). For dairy operations moving cattle to auction and to other dairy operations, mean outgoing direct contact rates ranged from 2.6 to 10.34/quarter. For large cow-calf operations, the greatest outgoing direct contact rate in a quarter was reported for auction destinations (range of mean values, 1.28 to 2.88/quarter). Mixed beef and swine, small swine, small feedlot, small cow-calf, and small ruminant operations reported similar outgoing direct contact rates overall, with mean values for all quarters ranging from 0 to 4.73.

Data were summarized for reported outgoing movements in which the types of other livestock operations (if any) in contact with the shipped animals was not clear (Table 5). These included visits to a veterinarian, a show, or to another site within an operation. Large cow-calf, small cow-calf, dairy, mixed beef and swine, small ruminant, and small feedlot operations were the only operations to report movements of < 11 km to home grazing. No operation type had a mean of > 2 visits to shows in a quarter.

The mean number of incoming direct contacts also fluctuated by quarter. In most quarters, incoming direct contact rates were highest for large feedlots and dairies (Table 6). Incoming movements from auctions to dairies were only reported during June through August. Large feedlots reported the highest number of incoming direct contacts from auctions overall, peaking with a mean of 11.6 contacts in March through May.

Indirect contacts—Mean total indirect contacts for livestock operations were summarized for the year (Table 7). The number of indirect contacts varied substantially by operation type and contact source. For large feedlots, the mean number of indirect contacts through feed trucks, livestock haulers, and manure haulers each exceeded 725 for the year (181/quarter). For dairy operations, the highest mean numbers of indirect contacts were reported for milk trucks (434.25 [109/quarter]), manure haulers (282.25 [71/quarter]), and feed trucks (146.65 [37/quarter]). Dairy operations in our study had a mean of 280 indirect contacts/quarter.

The estimated distance traveled to each livestock operation type was similar across the quarters of the year (range of mean values, 66 to 77 km/quarter [41.25 to 48.13 miles/quarter]) for all reported indirect contact sources except livestock haulers. Livestock haulers traveled a median of 113 km (70.63 miles) for each contact in the months of September to November, compared with approximately 64 km (40 miles) for each contact for the rest of the year (data not shown). Distances traveled by all indirect contacts to each type of operation are reported (Table 8). The longest median distance traveled by indirect contacts (109 km [68.13 miles]) was to large feedlot operations.

Discussion

The present study was performed to estimate direct and indirect contact rates and contact distance distributions among livestock operations in Colorado and Kansas. The region selected represented the west central region of the United States, which includes a wide variety of livestock operations types. Previous studies have been conducted to evaluate contact rates in the state of California,^16,23 The Netherlands,²⁷ and New Zealand,²⁸ but because of differences in operation types and sizes as well as regionally specific management practices, the present study provides information specific to direct and indirect contacts in the west central United States. Because of the practical limitations of enrolling participants in this study, true random sampling of livestock operations was not possible. Self-selected volunteers participated in the study. As such, the response to our survey was subject to volunteer bias. Reasons for declining to participate may have included perceived time requirement and concerns regarding privacy and confidentiality of information. Previous studies^16,27,28 of livestock operation contact rates encountered similar limitations, but investigators concluded that the results were still broadly representative of the respective surveyed populations.

To reach as many livestock producers from as broad a range as possible, invitation letters and flyers were distributed through a wide variety of organizations and methods. One result of this was a high probability of duplicate invitations and an inability to enumerate the number of unique invitations made or calculate the initial survey invitation response rate. During the year, the survey response rates of the self-selected volunteers decreased (from 354/628 [56.4%] for the first quarter to 244/591 [41.3%] for the fourth and final quarter). The apparent decrease in response rates throughout the study period may have resulted from participants becoming uninterested or perceiving little direct benefit from taking the time to complete the survey. Most participants were located in Kansas, with 1,609 surveys sent and 744 returned, whereas 791 surveys were sent to Colorado producers and 392 were returned. This was expected because of the larger number of livestock operations located in Kansas.²⁴ The survey responses did include all livestock production types and a range of herd sizes within each type. Dairy and large feedlot operation types were the least commonly reported. Given that 34 of 532 (6.4%) participants reported owning both swine and beef cattle, we considered this a potentially important conduit for the spread of disease from a swine operation to a beef cattle operation, and a mixed beef and swine production type was included in the analysis. Mixed livestock production systems could increase contact rates among species, leading to larger outbreaks of infectious diseases (such as FMD) that infect multiple species. An additional consideration is the removal of all movements with an undetermined source or destination; however, these represented a small proportion (< 7%) of the reported movements. Most movements removed were incoming contacts for small swine and small ruminant operations, so contacts for those categories may have been underestimated. The reported equine movements were not included in the analysis because the survey did not ask for equine movements and therefore the reported numbers were not considered to be reliable estimates of overall equine movement.

The 2007 USDA National Agricultural Statistics Service census reported the mean number of cattle on feed per feedlot as 1,665 in Colorado and 2,022 in Kansas,²⁹ indicating a relatively large number of small feedlots, compared with large feedlots, as was also reflected in the responses to our survey. Both large and small feedlots in this survey had substantial numbers of direct and indirect contacts. Previously reported data from California included only a few small feedlots.^16,23

The typical beef cow herd size in Colorado and Kansas has been reported as approximately 60 cattle,²⁹ whereas the mean beef cow herd size was 369 in the present study. Some of this apparent herd size difference was attributable to the fact that participants in our quarterly surveys were asked to record the current number of livestock, including calves and bulls, whereas the 2007 USDA census results reflect only the number of cows in a herd. Because the total number of livestock was reported each quarter, we did find an apparent decrease in the number of cattle reported in later quarters of the year as calves were (most likely) weaned and removed from the operation. Still, this survey population appeared to represent larger cow-calf herds within Colorado and Kansas than previously described.

In previously reported data from California, approximately 30% of beef cattle herds were kept at multiple locations.²³ The number of movements to other locations that were part of the same operation in the present study suggested that Kansas and Colorado beef herds are similarly managed on multiple sites. Multiple locations that were part of the same operation could be located some distance apart from one another. For purposes of this report, we included direct contacts between 2 locations that were part of the same operation when these locations were > 11 km apart. Owing to the likelihood of fence-line contact with other operations, transmission to other locations is still a concern even for movement within an operation. This may have overrepresented contact rates if the relocated herd remains isolated with no contact with other livestock. The movements of < 11 km were included in the outgoing direct contact rate within an operation; this distance was chosen on the basis of a natural break in the data from the survey (not shown).

The mean herd size of dairy herds has been reported as 283 cows in Colorado and 150 cows in Kansas.²⁹ The median herd size in this survey for dairy operations was 138 cattle, with a mean of 1,274 cattle for the 19 dairy herds for which survey responses were received. Dairy operations in our study had 280 indirect contacts/quarter, compared with an indirect contact rate of 234/mo found in small dairies (defined as those comprising < 1,000 cattle) in California.¹⁶ The apparently lower indirect contact rate among dairies in the present study was most likely attributable to a generally smaller size of herds in Colorado and Kansas, compared with those in California. However, the number of survey participants that reported data for dairy operations was low, and caution should be exercised to avoid overinterpretation of the reported contact rates for this type of livestock.

The typical herd size of swine has been approximated at 750 pigs in Colorado and 1,300 pigs in Kansas.²⁹ Large swine operations in our study had a high number of direct incoming and outgoing contacts overall, which included any movements reported to another location even if part of the same operation. Most large swine production is part of an integrated industry, with animals shipped within operations and then directly to slaughter,³⁰ making it unique, compared with the other livestock operation types in our study population. The high number of direct contacts for large swine facility to large swine facility most likely represented this vertical integration in the swine industry and may suggest a risk for direct transmission of disease predominantly within the integrated system and not to other types of livestock operations.

Multiple production types reported high numbers of outgoing and incoming direct contacts with auction markets, suggesting markets are an important potential distribution and surveillance point for infectious disease spread. Bates et al¹⁶ also found that large numbers of livestock were purchased from livestock auctions in herds in a 3-county region in California. All cattle operations in our study reported livestock incoming from auctions. Multiple herds and livestock types mix at auction markets before dispersal to individual herds, providing substantial opportunity for disease transmission. Continued education to producers on the risks of bringing outside livestock onto premises and the importance of quarantine and open versus closed herds remains important to prevent the possible spread of disease agents such as infectious bovine rhinotracheitis virus, BVDV, Tritrichomonas foetus, porcine epidemic diarrhea virus, and PRRSV. Separation of sick animals from healthy animals to prevent direct exposure is the basis for controlling the spread of endemic pathogens.^31,32 Additionally, cleaning and disinfecting livestock trailers may also help control disease spread. The results of the present study provide data that may help to quantify the magnitude of the risk from contact among herds through auction markets.

Indirect contacts are a potential risk for disease spread, particularly for a highly contagious disease such as FMD.^12,33 Indirect contact could allow disease spread after livestock movement controls are in place, resulting in a longer outbreak.¹¹ Some indirect contacts must be maintained for animal welfare reasons and for continuity of business and long-term continuation of livestock production even in the presence of an outbreak. Delivery of feed, supplies, and laborers will be necessary and will require increased efforts in biosecurity and disinfection to control risk. In our study, large feedlots had the highest mean number of indirect contacts in a year as well as high numbers of incoming direct contacts from auctions and small feedlots, putting them at substantial risk for introduction and spread of disease. The results of this study did differ substantially from the previous research¹⁶ on indirect contacts for large beef operations in California; for example, approximately 8 mean monthly contacts/y with livestock haulers was previously reported,¹⁶ compared with approximately 726 livestock hauler interactions for large feedlots and 38 for small feedlots during the year of our study. These contacts may occur over long distances as well, increasing the risk of long-distance transmission of some diseases during an outbreak in the Midwest. Data from the present study provide estimates of the number and types of indirect contacts among livestock operations in the central United States, which can be used for planning the resources that would be required to institute biosecurity and disinfection procedures needed to control indirect transmission of infectious disease during an outbreak in this region. For example, the large number of feed truck visits and the need to deliver feed daily have substantial welfare impact and magnify the need for well-planned biosecurity and disinfection. The high number of indirect contacts found in this study support the premise that cleaning and disinfection are imperative to prevent the transmission and persistence of disease agents such as PRRSV, Salmonella spp, BVDV, and Escherichia coli O157. Both Salmonella enterica serovar Dublin and E coli O157 have been shown to persist in cattle manure,^34,35 and indirect transmission of BVDV and PRRSV has also been reported.^36–38

The direct and indirect contact rates reported here can also be used to inform efforts to model transmission and control of infectious diseases such as FMD that infect multiple species. The results can lead to biosecurity improvements for emergency planning during an infectious disease outbreak among livestock as well as provide data to parameterize simulation models to evaluate control methods during a possible outbreak. The contact data from this study are specific to the central United States, allowing modeling with region-specific parameters for increasing the validity of results.⁷ To our knowledge, no prior estimates of the number of direct and indirect contacts were available for this region, and simulation models of livestock disease outbreaks were therefore lacking an essential element to provide valid model results and evaluate alternate control methods in this important agriculture region. Valid model results are essential for planning and decision making, including the relative importance of different control strategies such as biosecurity and control of livestock movement.