عوامل مدیریت عملیات در ارتباط با مرگ زودرس پس از تولد کره اسبهای ایالات متحده

Of 7320 equine foals reported born alive during 1997 on 1043 operations that had equids on 1 January 1997, and that participated in the United States National Animal Health Monitoring System (NAHMS) Equine 1998 Study, 120 foals were reported to have died (by either euthanasia or natural causes) within the first 2 days of a live birth. The weighted estimate was 1.7% mortality (standard error=0.5) within the first 2 days of live birth for all foals born on operations in the 28 states included in the study. A multivariable logistic-regression model revealed that foals born in the southern region were more likely to have been reported to have died within the first 2 days of live birth than in the western region. In addition, the following operation-level factors were associated with increased odds of a foal dying within the first 2 days of live birth: not routinely testing newborn foals for adequate absorption of colostral immunoglobulins during the first 2 days of life; adding new resident equids to the operation during 1997; having non-resident equids stay on the operation for 1–30 days during 1997; never requiring an official health certificate (for operations that had non-resident equids stay on the operation for 1–30 days); using something other than straw or hay as the predominant bedding type; and feeding equids a vitamin-mineral supplement/premix with forage and/or grain.

The objective of equine breeders is to produce healthy foals that will grow to fulfill their aimed purpose. The neonatal period is a critical phase in an equid’s life. Morley and Townsend (1997) reported that 5% of Thoroughbred foals died within 2 weeks of birth in four western-Canadian provinces. In a prospective study of foal health and management in Texas, Cohen (1994) stated that the risk of death in foals decreased with age, and was highest during the first 7 days of life. Haas et al. (1996) found that 74% of foal deaths occurred within 48 h of parturition in a large mare herd in Manitoba. The results of the National Animal Health Monitoring System (NAHMS) Equine 1998 Study of the United States Department of Agriculture (USDA) estimated that 1.7% of foals born alive either died of natural causes or were euthanized within 2 days of birth in the United States during 1997 (USDA, 1998a). Conditions or agents that have been described as causes of foal mortality include congenital anomalies (von Matthiessen, 1993, Campbell-Beggs et al., 1995, Doige, 1996 and Theoret et al., 1997), equine herpesvirus infection (Dixon et al., 1978), exposure of the pregnant mare to fescue toxicosis (Putnam et al., 1991; Cross et al., 1995), failure of colostral immunoglobulin transfer to foals (McGuire et al., 1977), septicemia (Carter and Martens, 1986), haemorrhage from the umbilicus (Haas et al., 1996), diarrhea (Cohen, 1994), pneumonia (Prescott et al., 1989), trauma (Cohen, 1994), and starvation/exposure (Haas et al., 1996). Only a few studies have examined risk factors involved in early post-natal foal mortality. Cohen (1994) did not identify factors that were significantly associated with foal mortality-possibly because the low number of deaths observed in his study limited the power. Haas et al. (1996) found that failure of passive transfer, poor mothering ability, dystocia, low birth weight, lack of rainfall and low environmental temperatures were associated with foal death on a pregnant-mare operation in Manitoba. However, their study population was atypical of United States equine populations in that assistance to foals was minimal and foal mortality was quite high (22% before reaching 10 days of age) (Haas et al., 1996). The purpose of this investigation was to identify management and other operation-level factors that were associated with foal death within 2 days of live birth in the United States during 1997, based on data from the NAHMS Equine 1998 Study.

Data used in this study were from the first NAHMS national study of equine health. The NAHMS Equine 1998 Study was designed to permit inferences to be drawn to the total population of equids in 28 states that accounted for about 78% of the farms with horses and ponies (and 78% of the horses and ponies on farms) based on the 1992 Census of Agriculture (USDA, 1998a). Because the 1992 Census of Agriculture did not include operations with <US$ 1000 of sales of agricultural products during the year and less than five horses, the total population of equids in these states was not known at the time that the study was designed. However, the 1992 Census of Agriculture represented the most-reliable state-level estimates of equine numbers at the time of the design of the NAHMS Equine 1998 Study. Participants in the National Swine Survey (USDA, 1992) and the National Dairy Heifer Evaluation Project (Heinrichs et al., 1994) (which were the first two NAHMS national studies) were selected from both list frames and area frames. The area frames served to adjust estimates for incompleteness of the lists (including recent producer transition in and out of business) (Heinrichs et al., 1994). With reduced budgets, subsequent NAHMS national studies relied solely upon sampling from NASS lists of agricultural producers, and did not make use of area frames (Losinger et al., 1997). For the NAHMS Equine 1998 Study, it was possible to produce lists of operations with large numbers of equids (such as boarding stables, riding and training facilities, race tracks and other service providers). However, it would have been very difficult and costly to produce lists of all premises with ≥1 equids in the states included in the study. Therefore, area-frame sampling (in addition to sampling from the lists) was absolutely essential for the Equine 1998 Study to be successful. The Equine 1998 Study had multiple objectives. Participants who had three or more horses present on their operations on 1 January 1998 were eligible to continue in subsequent phases of data collection with federal and state veterinary medical officers and animal health technicians (USDA, 1999). On three separate seasonal visits, veterinary medical officers and animal health technicians gathered data on lamenesses in horses (USDA, 2000b). The veterinary medical officers and animal health technicians collected fecal and grain samples from operations to test for the presence of Salmonella (Traub-Dargatz et al., 2000). Blood-serum specimens were taken from horses on 837 operations to test for antibodies to equine arteritis virus (USDA, 2000a). Considerations for determining the study’s overall sample size included not only the available budget and the desired levels of precision of estimates from various stages of data collection, but also workloads that the NASS field enumerators, the federal and state veterinary medical officers and animal health technicians, and laboratories could handle. Moreover, because participation was voluntary on the part of operators, a certain level of refusal had to be anticipated. Since this was the first NAHMS national study to address equids, participation rates were quite unknown at the time that the study was designed. The precision of population estimates from sample surveys depends on not only the size of the sample, but also the heterogeneity of the population (Sukhatme and Sukhatme, 1970). Once the number of sample units has been established, the objective is then to select a sample in such a way that the variance of population estimates is minimized. Stratified sampling, which consists of dividing the population into strata and drawing one random sample from each stratum, is intended to yield a better cross-section of the population and more precise population estimates than unstratified sampling (Sukhatme and Sukhatme, 1970). In general, the more homogeneous the units within strata, the greater the precision of the resulting population estimates; the more units within a stratum, the larger the sample that should be drawn therefrom; and the greater the variability of units within a stratum, the larger the sample that should be drawn therefrom (Sukhatme and Sukhatme, 1970). If participation rates are expected to vary by stratum, then this should also be taken into account in designing the survey. Large variation in weights can reduce the precision of survey estimates (Cox and Cohen, 1985). If a few respondents have extremely large weights compared to the majority of the respondents, then the population estimates will be heavily dependent upon responses given by the respondents with the large weights. The weight-smoothing process mitigated the impact of some very large weights. Generally, operations selected from the area frames had larger weights than operations selected from the list frames, and contributed more to the error of estimates. The analyses of foal mortality were based on data from a subset of Equine 1998 Study participants (i.e. those that had live-foal births during 1997). Although the Equine 1998 Study was not designed specifically to yield estimates of early-postnatal mortality of foals as related to risk factors, the sample size was large enough to yield estimates with an acceptable level of precision. One limitation of this study is that the data were collected during March and April of 1998, and respondents were asked to recall the number of foals that died within 2 days of live birth during the entire previous year. Thus, some of the respondents might not have been able to recall exactly the age at which their newborn foals died. Respondents with a large number of foals might not have known the exact number of foals that were born or that died. For many operations, births and especially deaths (the maximum number of deaths per operation was four) were rare and sufficiently important events that the recollection of respondents may have been reasonably accurate. The exact age at the death of early post-natal foals may have been more difficult to recall. In any sample survey, a certain amount of non-sampling error is inevitable (Sukhatme and Sukhatme, 1970). Using a questionnaire from a prior NAHMS survey of the dairy industry, Erb et al. (1996) found that respondents gave discrepant answers on about 8.5% of the questions when the questionnaire administration was repeated at a different date. Recollection may be less accurate as the time elapsed from the event and the number of observations increase. Although every effort was made to verify that the data were collected consistently, we cannot know for certain how accurately respondents provided data on age of foals at death and the number of births and deaths per operation. We did not attempt to quantify recall bias by repeated administration of the questionnaire. Misclassification bias might have occurred if respondents who managed their equids less intensively were more likely to misclassify neonatal deaths as intra-partum (thus rendering them “Acts of God” rather than management errors). Insurance considerations (particularly for more valuable equids) could also have led to misclassification bias. Another limitation of this study is that no information was collected on how individual mares or foals were managed. Data were gathered at the operation-level, not at the individual animal-level, regarding routine procedures performed on neonatal foals, on where foals were born, and on equine management practices for operations. Therefore, we cannot conclude from this study that a specific management practice caused foal mortality — but rather, that foals born on operations which used a particular practice had a different odds of being reported as dying within the first 2 days of live birth than foals born on operations which did not employ the particular practice. We obviously have a better idea of how an individual foal was handled for the 329 sample respondents who reported the birth of only one live foal during 1997 versus the 714 respondents who reported the birth of more than one foal. However, the operations with fewer equids were generally selected from within the randomly selected geographic areas of the December 1997 NASS Area Survey. These operations were sampled at a much lower proportion, and represented a much larger number of operations in the population than the larger operations that had been selected from lists. The final model indicated that operation size (as measured by the number of resident equids on 1 August 1997) was not a significant predictor. In general, an analysis which would have ignored intra-operation correlation (and assumed independent responses) would have underestimated the true variance of parameter estimates and led to test statistics with inflated Type-I errors. The use of SUDAAN permitted the computation of estimates and standard errors that took into account both the sample design parameters and the clustering of births on the same operations (Bieler and Williams, 1995). We made numerous statistical comparisons, and might well have erroneously rejected at least one true null hypothesis (Neter and Wasserman, 1974). Moreover, the ability of the model to predict outcomes for new data may be less than that observed from the data upon which the model was based (Neter and Wasserman, 1974). However, the purpose of this investigation was to find management factors potentially associated with neonatal foal mortality (rather than to construct a predictive model). No significant differences were reported in the percent of operations by primary use of equids nor by primary function of operation between the southern and western regions (USDA, 1998a). Larger percentages of operations in both the southern and western regions used equids primarily for farm or ranch work than in other regions of the country (USDA, 1998a). However, the US equine industry is certainly very heterogeneous within the regions defined for this analysis. The southern region included Florida (where over one-half of the horses were used in showing) and Kentucky (where the majority of the people involved in the horse industry participated in racing) (American Horse Council Foundation, 1996). The NAHMS Equine 1998 Study was not designed to provide state-level estimates. The higher mortality risk shown for the southern region (compared to the western region) may have been due in part to the recall bias described above. If a foal is very important to the economic well-being of an owner/operator, the owner/operator may have a better capacity to remember (or better records — such as computerized records — to indicate) the age at which foals died. Such operators may also have a higher level of observation at foaling, and different criteria for culling foals. Climatic differences could have had an impact on differences in foal mortality: the southern region is generally warmer and more humid than the western states. Cohen (1994) reported that the practice of assessing passive immunity was significantly associated with decreased incidence of septicemia and pneumonia in foals. Routinely testing newborn foals for adequate absorption of colostral immunoglobulins may indicate a high level of care of early post-natal foals. Although no other variables related specifically to the care of foals within the first 2 days of life passed the variable screening, most respondents who indicated the routine testing of foals for adequate absorption of colostral immunoglobulins during the first 2 days of life also typically had the foals examined by a veterinarian, dipped the navels with antiseptic, and gave the foals an enema during the same time frame. Associations between these variables and neonatal foal mortality were weak (P≥0.25). The perceived value of a newborn foal varies considerably depending on the intended use, the breed, and bloodline within breed. Usual costs to produce a new foal include the stud fee, housing and transportation to the breeding facility (unless the semen is shipped), monitoring the mare for foaling, veterinary and other care of the mare and foal, etc. The results of this analysis suggest that even if one only considers the cost of the test for adequate absorption of colostral immunoglobulins and the hypothetical value of a foal, one has a lower expected cost when one tests the foal for adequate absorption of colostral immunoglobulins except where the hypothetical value of the foal is <∼US$ 1000. If a veterinarian is present within 2 days of a live birth and evaluates the foal for adequate absorption of colostral immunoglobulins, it is likely that the veterinarian is able to examine the foal and intervene for other health problems as well. Early diagnosis of inadequate absorption of colostral immunoglobulins can lead to interventions that may prevent mortality in foals during the early post-natal period (McGuire et al., 1977). In the NAHMS Equine 1998 Study, no data were collected on the number of foals with inadequate absorption of colostral immunoglobulins, nor on interventions for foals with failure of passive transfer on study farms. Biosecurity practices are important to reduce the risk of cause-specific morbidity and mortality due to infectious diseases (Doherr et al., 1998). Contact with other animals can result in the spread of infectious agents which cause disease (Doherr et al., 1998). While introducing equids from outside the operation (both as permanent acquisitions and as temporary visitors) was associated with increased foal mortality within 2 days of a live birth, no information was collected on whether the foals born on an operation were born to resident mares or visiting mares, nor on whether visiting non-resident equids were broodmares. The health risk to foals born to visiting mares and to resident mares is potentially different. Traub-Dargatz et al. (1988) reported that foals born to visiting mares were more likely to develop diarrhea than foals born to resident mares. The practice of requiring a health certificate for non-resident equids that visited for <30 days may either reduce the risk of disease exposure to foals, or reflect a higher level of equine health care on the operation. In the present study, information was collected on the types of bedding used for equids during 1997, and on the predominant bedding used. However, no questions were asked specifically on the bedding used for newborn foals. Most respondents who did not use straw or hay as the predominant type of bedding for equids on the operation indicated that wood shaving, chips or sawdust was the predominant bedding used. Traub-Dargatz et al. (1988) found that the use of wood shavings in foaling stalls was associated with a higher occurrence of foal diarrhea. Further research is needed to determine whether pathogens that cause diarrhea in foals are less prevalent in straw or hay bedding than in other types of bedding, or whether the type of bedding is a marker for some other associated management risk factor. No information was collected on the type of vitamin-mineral supplements fed to equids nor on the reasons for feeding it. It is possible that this variable entered the model purely by chance, because the result was contrary to what one might have anticipated. Some respondents may have fed the vitamin-mineral supplement premix as a remedy to a perceived equine-health problem. Further research is required to determine the possibility that some of the respondents who fed a vitamin-mineral supplement premix to equids may have fed too much, resulting in altered viability of newborn foals. We do not know whether the vitamin-mineral supplement premix was fed to broodmares or just to other equids; respondents reported feeding the supplement if it was fed to one or more equids on the operation during 1997 (USDA, 1998b). Equine owners should pay particular attention to the nutritional requirements of pregnant mares in determining the need for supplementation of pregnant mares with vitamin and mineral supplements.