Factors related to the level of occurrence of bovine abortion in Chilean dairy herds

Preventive Veterinary Medicine 110 (2013) 183–189

Contents lists available at SciVerse ScienceDirect

Preventive Veterinary Medicine journal homepage: www.elsevier.com/locate/prevetmed

Factors related to the level of occurrence of bovine abortion in Chilean dairy herds Paula Gädicke a,b , Gustavo Monti c,∗ a b c

Universidad Austral de Chile, Chile Department of Pathology and Preventive Veterinary Medicine, Faculty of Veterinary Sciences, Universidad, de Concepción, Chile Department of Preventive Veterinary Medicine, Faculty of Veterinary Sciences, Universidad Austral de Chile, Chile

a r t i c l e

i n f o

Article history: Received 31 October 2011 Received in revised form 13 November 2012 Accepted 20 November 2012 Keywords: Bovine Abortion Incidence

a b s t r a c t The objectives of this study were (1) to estimate the frequency and dynamics of bovine abortion syndrome; (2) to identify groups of cows affected by abortion; and (3) to assess the characteristics of herd management and lactation associated with abortion rates. The study was performed using farmers’ historical records for 77 dairy herds in the south of Chile (Bio-Bio, Los Lagos and Los Ríos Regions) collected between 2001 and 2005. These records included 44,959 lactations from 20,977 cows. In addition, farm management practices were assessed through a questionnaire involving 127 herds. The herds were selected according to the farmers’ willingness to participate and the existence of high-quality electronic records assessed by the practitioners advising the farms. The frequency distribution of observed, inferred and general abortions was estimated by the incidence rate (IR). A hierarchical logistic regression analysis with random intercept was performed to assess the association between herd management and lactation characteristics and the occurrence of abortion. An IR of 1.74 per 100 cow-months at risk was estimated. General abortions were highest in first-parity cows (IR: 1.85 per 100 cow-months at risk). Abortion cases inferred from individual records were most frequent in the first trimester of gestation and decreased over time, whereas observed abortions increased in accordance with gestation time. The period of highest risk for abortion was around 82 days of gestation. Management practices such as a tap drinking system for cows, a closed herd, vaccination against leptospirosis, exclusive use of pasture for cows, animal density, the time that a calf stays with its dam and breed type were associated with the risk of abortion. The results of this study demonstrate that there is a large underestimation of abortion rates when only farmers’ abortion records are analysed, and there are several factors associated with the risk of abortion. © 2012 Elsevier B.V. All rights reserved.

1. Introduction The primary reason for preventing bovine abortion or its sequelae is to reduce the impact on a herd’s reproduc-

∗ Corresponding author at: Institute of Preventive Medicine, Faculty of Veterinary Sciences (UACH), PO Box 567, Valdivia, Chile. Tel.: +56 63221064; fax: +56 63 293233. E-mail address: [email protected] (G. Monti). 0167-5877/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.prevetmed.2012.11.022

tive performance by taking appropriate control measures. Bovine abortion is a limiting factor for the dairy business because it decreases milk production and the potential number of replacements for the herd, increases feeding and medical treatment costs, increases the number of artificial inseminations required to obtain a calf and increases culling rates (Thurmond and Picanso, 1990). For example, in Chilean dairy herds, it is estimated that losses are as great as US $143.32 for lactations in which an abortion occurs (Gädicke et al., 2010).

184

P. Gädicke, G. Monti / Preventive Veterinary Medicine 110 (2013) 183–189

Several determinants are associated with abortion in cattle; some are specific infectious agents (Brucella Abortus, Leptospira spp., BVDV, etc.), whereas others are related to the characteristics of the cow, such as age, breed and milk production level. Other determinants include the presence of an additional corpus luteum, prior abortion, retained placenta or metritis in the early postpartum period (Gröhn et al., 1990; Markusfeld-Nir, 1997; Duchens et al., 2008). Moreover, nutritional supplementation and other aspects of feeding, such as energy or protein levels in the diet (Forar et al., 1996; López-Gatius et al., 2008; Szenci, 2008), are noted as putative causes. Finally, environmental determinants, such as management practices and climatic factors (Labèrnia et al., 1996), are also associated with bovine abortion syndrome (BAS) (Gädicke and Monti, 2008). However, management practices have not received much attention in the literature (Forar et al., 1995; Szenci, 2008). Although a multiple-cause origin is recognised, few studies on abortion approach the problem independently of these causes (Thurmond et al., 1990, 1994, 2005; Rafati et al., 2010). Most studies are related to infectious causes or specific agents and are based on intensive indoor dairy production systems. Fewer studies (Atkinson et al., 2000; Campero et al., 2003; Moore et al., 2009) focus on extensive or free-grazing production systems, which represent a large proportion of dairy farming around the world (FAO, 2010). Accurate estimation of the frequency of occurrence is crucial for monitoring herd performance. Therefore, the frequency distribution of BAS incidence and an estimation of the risk factors under Chilean production conditions may contribute to current knowledge and aid in the design of more realistic and efficient control programmes in extensive production systems. The objectives of this study were as follows: first, to estimate the frequency and dynamics of BAS; second, to identify which cows are likely to abort; and, finally, to assess the characteristics of herd management and lactation associated with abortion rates. 2. Material and methods 2.1. Herds and management The study used a retrospective design. The area under study is located between 36◦ 00 and 44◦ 04 South and 71◦ 00 West and the Pacific Ocean (Instituto Geográfico Militar, 2008). This area supports almost 60% of the total cattle population of the country (Velis et al., 2006). In terms of herd size, the herds selected represent typical medium to large dairies in the centre of southern Chile. In southern Chile, as in many developing countries, milk production is based on perennial pastures because costs are lower compared with indoor concentrate-based feeding. Pasture is a major component of the diet for dairy cows, and supplementation with cereal grains is often used to increase milk production. Dairy cows are usually fed concentrate twice daily during milking, and they receive a new pasture lot after each milking (Pulido et al., 2009). In this study, dairy herds consisted of Holstein (57.1%) and Black-Pied X Holstein cattle (23.8%). In 93.7% of the farms, all replacements were bred from their own herd. The feeding system was grazing pasture year round and

was supplemented with silage (89.7% of the farms), concentrates (88.9% of the farms) and hay (62.2% of the farms). In addition, most farmers offered year-round mineral supplements to heifers and cows (85.8% of the farms), and cows were divided into production groups (59.8% of the farms). Reproductive management was AI and breeding, year round (all herds), with visual heat detection without the use of aiding devices (95% of the farms). A bull was used on oestrus cows (35.4% of the farms) (after 3 AI, on average). Prostaglandin treatment was applied by farmers (83.5%) to control the reproductive cycle but was not used for all animals; 27.5% of farmers used GnRH treatment. Ninety-four percent of the farms applied vaccines against brucellosis; 84.2% applied vaccines against leptospirosis (Leptospira interrogans serovar Hardjo-bovis and L. interrogans serovar Pomona); 48.8% applied vaccines against infectious bovine rhinotracheitis (IBR); and 40.9% applied vaccines against bovine viral diarrhoea virus (BVDV). Most of the farms were certified by the official animal health service (SAG) as free of infection for brucellosis (96.8%) and tuberculosis (79.5%), but the majority of farms only applied control measures for leptospirosis (68.5%). 2.2. Data collection Herds were selected by convenience based on farmers’ willingness to cooperate with the project and by the presence of high-quality electronic records (assessed by the farm advisor). Initially, 127 herds participated and included the requested information. Reproductive information was collected by veterinarians during routine herd health visits (at least once monthly). All cows that calved from 1 January 2000 to 6 months before the date of administration of the questionnaire on each farm were included. The cows had been observed for at least 6 months. Pregnancy diagnosis was performed by transrectal palpation between 35 and 60 days after AI. The reproductive records of the farms included the AI date. Cows that were served with cleanup bulls were removed from the dataset. The herds that were included in the study could be characterised as having both high management standards and high yielding cows (305-day production yield of 7979 lt). At all farms, a personal interview was conducted and a questionnaire was administered by four interviewers who were trained by the senior author. The questionnaire (5 pages with 18 topics, mostly open questions) was previously validated with 10 farmers and practitioners, and information was obtained about herd management practices relating to replacements, feeding systems, breeding and reproductive management and biosafety measures on farms (A copy of the questionnaire is available upon request from the corresponding author). 2.3. Dataset descriptions Records obtained from the 127 herds were assessed for consistency and accuracy, validity and recording year. Analysis was performed based on two reduced but consistent datasets that aimed to pursue different goals because not all farms have the same level of information. The first dataset (dataset A) consisted of farmers’ historical records

P. Gädicke, G. Monti / Preventive Veterinary Medicine 110 (2013) 183–189

from 77 dairy herds located in the south of Chile (Bio-Bio, Los Lagos and Los Ríos Regions) that had complete and reliable reproductive and productive records from 2001 to 2005. This dataset included 44,959 lactations from 20,977 cows. This dataset provided the most extensive historical data on individual cows for the analysis of trends over time. Herd sizes ranged from 10 to 438 cows. Second, a reduced dataset (dataset B) was created that included herds from dataset A that had complete reproductive and productive records at the herd and lactation levels. Dataset B included 42 herds, with 4426 cows and 5167 lactations. Information was coupled with the production and management practices described above. This dataset provided the most extensive information on herd and individual characteristics. Herd sizes ranged from 11 to 271 cows. 2.4. Case definition An observed abortion case was defined as an event corresponding with the interruption of gestation 42 days after conception, visualised and recorded by the farmer (the software used by farmers includes a code that synthesises several signs, such as uterine discharge or a foetus found or verified by the veterinarian). An inferred abortion case was defined as an abortion event inferred by the analysis of the differences between the dates of two consecutive AIs, which should be between 90 and 260 days. The 90 days limit resulted from considering the shortest gestation time for abortion (42 days) in addition to time to recover (27 days) and become cyclic again (21 days). The definitions of observed and inferred abortion were considered mutually exclusive. Finally, a general case of abortion included the occurrence of either of the two types. Animals at risk included those cows that were present in the herd at 42 days after their last AI until one of the following events: (1) Abortion on a registered date for cows that had an observed abortion. We verified that these records did not corresponded to stillbirths or a calf carried to term or to a gestational length longer than 260 days and that the animal was born dead. (2) Leaving the herd due to culling or death (for both types of abortion). (3) The end of the foetal period (260 gestation days) for all types of abortion. (4) Re-insemination 90 days or more since the last AI (for inferred abortions). Trimesters of pregnancy were as follows: first, conception to the 90th day of gestation; second, the 91st to 180th day of gestation; third, 181 or more days of gestation. 2.5. Statistical analysis Frequency distribution of observed, inferred and general abortion cases were estimated by the incidence rate (IR) using dataset A. Although several indicators were available, we preferred IR because it is most frequently used in the literature, facilitates comparison with our results,

185

and captures the time at risk of animals that are lost to follow-up. Additionally, given that abortions are not uniformly distributed during pregnancy, IR better represents the changes in abortion per unit of time. The total number of abortions was determined using dataset A and stratified by parity and the year of the most recent calving. In addition, the IR was estimated for the first abortion case only, using lactation as the unit of analysis. The IR was calculated as follows: number of cases of abortion/100 cow-months at risk. The difference in frequencies stratified by parity and by year of calving were tested by a chi-square test (p < 0.05) for all types of abortions. The differences in IR by lactation, year and gestation trimester were assessed by the Log Rank test for trend using p < 0.05. Frequencies that were greater than the expected values were identified by the standardised Pearson residual (>2). To identify the period of increased risk of abortion, the hazard curve over the foetal period (days 42–260 of gestation) was obtained using a nonparametric survival analysis as the derivate of the cumulative hazard (Cleves et al., 2004). The associations between herd management characteristics and lactation characteristics in first abortion cases were assessed using dataset B. Parity, calving season and accumulated milk production at 120 days (1% percentile) for days of milk in lactations with abortion were included in the analysis. All contemporary lactations that took place during the 12 months following the date the survey data were obtained were considered at-risk lactations. A hierarchical logistic regression model with random intercept (Rabe-Hesketh and Skondral, 2005) was performed to analyse the association between herd management and lactation characteristics with the occurrence of an abortion case. A random-intercept logistic regression was used to relax the assumption of conditional independence among the responses for the same herd. In this model, the effects of explanatory variables are assumed to be the same for each level 2 (herd). Potential confounders were considered as first-order interactions (Dohoo et al., 2010). To evaluate the model’s goodness of fit, AIC and BIC indicators were used, and the model was constructed using variables selected by a backward approach. The model was conducted using the GLLAMM programme (Rabe-Hesketh and Skondral, 2005) with Stata V.10SE (StataCorp, 2007; College Station, TX, USA). 3. Results 3.1. Frequency of general, observed and inferred abortion cases In total, 5745 general abortions were registered, of which 5218 were the only cases within the lactation, and 4289 (82.2%) of these were inferred. In addition, 487 lactations had 2 abortion events, 366 (75%) of which were inferred cases. Finally, 40 lactations had 3 abortion events, of which 29 (72.5%) were inferred. The proportion of lactations with general abortion cases ranged from 2.5% to 30.8% between herds, with a median of 10.4%.

1.70;1.79 1.743 299,453 5218 1.39;1.47 1.432 **

Different letters states statistical differences (p < 0.05) of values of a variable within the column. Different number states statistical differences (p < 0.05) of values within the row.

300.131 4289 0.27;0.31 0.291 319,376 929 Total

*

1.36;1.44 0.33;0.39 0.73;0.89 1.39a3 0.36b3 0.81c3 299,453 171,211 50,217 4189 622 407 1.33;1.41 0.08;0.11 0.00;0.01 1.37a2 0.09b2 1.97* 10-5c2 300,131 172,053 50,753 0.02;0.03 0.22;0.27 0.68;0.83 319,376 185,120 53,914 69 454 406 1 2 3 Gestation Trimester

0.02a1 0.24b1 0.75c1

4120 168 1

1.76;1.93 1.65;1.84 1.59;1.81 1.54;1.73 1.85a3 1.74a3 1.70b3 1.63b3

IR Cow-months

96,439 74,560 54,018 74,436 1783 1301 917 1217

# of Abortion 95% CI

1.43;1.58 1.34;1.52 1.29;1.49 1.28;1.44 1.50a2 1.43a2 1.38ab2 1.36b2

IR Cow-months

96,718 74,742 54,088 74,583 1456 1067 750 1016 0.28;0.35 0.26;0.33 0.25;0.34 0.22;0.29

# of Abortion 95% CI

0.31a * 1 ** 0.29a1 0.29a1 0.26b1

Cow-months

103,682 79,649 57,402 78,643 327 234 167 201

IR # of Abortion

1 2 3 ≥4 Lactation

General Inferred

Using dataset B and after evaluating interactions and potential confounders, the final model contained 7 variables that characterised information at the herd level (Table 2). These variables were associated with the risk of general abortion cases while controlling for the follow-up time of the cow in the herd. The variables included in the final model could be grouped into management factors and host characteristics relating to intrinsic and extrinsic determinants associated with abortion in cattle. Variables related to herd management were as follows: accessible tap water for drinking for the cows (OR = 0.57); pasture grazing only (OR = 0.11), which implies that cows do not receive any concentrate supplements; and animal density (OR = 0.74), expressed in number of cows/ha. Another group of variables was related to biosafety measures applied in the herd during the previous year (closed herd (OR = 0.82)) and the use of Leptospira vaccine the year before the administration of the questionnaire (OR = 0.61). The time (in days) that a calf stayed with its dam (OR = 0.83) was associated with a reduced risk of abortion in the subsequent lactation. Finally, host characteristics, such as the breed type used in the herd, were associated with a decreased risk of abortion. Using Holstein cattle as a reference, the following

Observed

3.2. Associations between herd management characteristics and characteristics of lactation with general abortion

Variable

The IR per 100 cow-months at risk was 1.74 for general abortion cases (95% CI 1.70; 1.79), 0.29 for observed cases (95% CI 0.27; 0.31), and 1.43 for inferred cases (95% CI 1.39; 1.47). The IRs for general, observed and inferred abortion were not uniform across lactation (Log Rank test for trend p < 0.05) (Table 1), and there was a clear tendency in all types of cases for the incidence rate to decrease with an increase in age. First-lactation cows had the highest rate for general abortion cases, observed cases and inferred cases. The group containing ≥4 lactations had the lowest rate for general cases, observed cases and inferred cases. IR registered for general, observed and inferred abortions was not uniform along trimesters of gestation (Log Rank test for trend p < 0.05) (Table 1). The IR for general abortions was higher than expected in the first trimester but lower than expected in the second trimester. The abortion rate during the third trimester did not differ from expected values. The IR for observed abortions increased during each trimester of gestation; abortion during the first and second trimesters occurred at lower rates than expected. However, the third trimester had a higher rate than expected. In contrast, the overall tendency of IR for inferred abortions decreased over the course of gestation. Abortion rates in the second and third trimesters of gestation were smaller than expected if the incidence was independent of the trimester. However, abortion in the first trimester register had a higher rate than expected. We observed statistically significant differences among the years of calving in which the abortions occurred for general and inferred cases (p < 0.05), but there was no clear trend over time.

95% CI

P. Gädicke, G. Monti / Preventive Veterinary Medicine 110 (2013) 183–189

Table 1 Incidence rate (IR) (expressed as number of new cases per 100 cow months at-risk), for observed, inferred and general abortion cases, by lactation number and gestation trimester of occurrence.

186

P. Gädicke, G. Monti / Preventive Veterinary Medicine 110 (2013) 183–189

187

Table 2 Logistic hierarchical regression model for herd management characteristics and characteristics of lactation with first general abortion case occurrence. Variable

Class

Herd (n)

Lact (n)

Abortion %

OR

Tap drinking water

No Yes No Yes Holstein Black Frison Black-Pied Holstein Red-Pied Holstein Other No Yes No Yes

28 14 39 3 24 10 3 4 1 8 34 11 31

3395 1172 4866 301 3355 1122 351 298 41 930 4237 1001 4166

14.96 10.67 13.83 7.97 14.96 10.96 13.68 3.69 3.71 13.76 13.43 15.38 13.03

Ref. 0.57 Ref. 0.11 Ref. 0.54 0.86 0.12 0.30 Ref. 0.82a Ref. 0.61a

Pasture grazing for cows Breed type

Closed herd Use Leptospira vaccine

Continuos variables Animal density Time that Calf stays with its dam (at herd level) Follow-up time of the cows in the study a b

Lact. with abortion No Yes No Yes

¯ X 1.99 1.93 1.02 0.79

(SD) (1.03) (0.77) (1.17) (1.07) (days)

Unit (cows/ha) (cows/ha) (Days) (days)

ORb Ref. 0.74 Ref. 0.83 1.002

95% CI 0.40

0.80

0.06

0.23

0.37 0.47 0.06 0.10

0.79 1.58 0.28 0.89

0.53

1.25

0.41

0.92

95% CI 0.62

0.88

0.72 1.002

0.94 1.003

OR adjusted with interaction effect. OR per unit of change; Lact = lactations; CI = confidence interval.

breed types had a decreased risk of abortion: Black Friesian (OR = 0.54), Black-Pied Holstein (OR = 0.85), Red-Pied Holstein (OR = 0.12) and others (OR = 0.30). We adjusted the differences in follow-up periods in each herd by including time in the model. Herd-level management variables included in the survey and other variables related to the current lactation, such as parity, calving season and 120-day cumulative milk production, did not have statistically significant associations in the multivariable model. The intraclass correlation was estimated at 0.03. The inspection of the distribution of standardised Pearson residuals of the logistic regression model for general abortion cases and herd management indicates that the model has an acceptable fit to the data and model assumptions. 4. Discussion Overall, the study estimated an IR of 1.74 cases of general abortion per 100 cow-months at risk that was greater (IR = 1.17) than what was reported by MarkusfeldNir (1997), who used similar case definitions and a similar method of estimation. The definition of inferred abortion used in this study was conservative and was assessed to minimise the risk of confounding a case with undetected heats. Although we cannot definitively conclude this for each individual case, under the production conditions of Chilean farming, it is very unlikely that herdsmen missed an average of 4 heats for so many animals. Other potential risks are linked to the lack of recoding of heats by the herdsmen, but this was ruled out because good record keeping was an inclusion criterion for participating in the study. This study uses an extensive dataset to examine abortion morbidity in extensive production systems based on grazing. We consider this study of substantial value for other

less developed countries with similar production conditions given that most studies have focused on infectious diseases and one specific agent (Patitucci et al., 1999; Thobokwe and Heuer, 2004; Santos et al., 2005; Moura et al., 2012) or on evaluating diagnostic submissions to laboratories (Paredes et al., 2002; Campero et al., 2003; Paredes and Moroni, 2005). Given the scarcity of studies that take a more general approach, estimations of morbidity could provide evidence to practitioners and farmers for the assessment of the overall impact of the problem on their production efficiency. We found that first-lactation cows had a higher IR (1.85 abortions per 100 cow-months at risk) for general abortion than older cows, in accordance with Markusfeld-Nir (1997). This finding suggests that heifers with low or no immunity are exposed to abortion-inducing agents when they are introduced into the milking herd. Most heifers in our study were reared on the same farm but were not kept in the lactating group. In addition, heifers’ susceptibility increases due to the increased stress of beginning lactation and the nutritional demand that this implies, and these heifers are still growing (Svensson and Hiltgren, 2007). Therefore, heifers were more likely to be affected by the cumulative influence of metabolic, endocrine, energy imbalance and health components (Szenci, 2008). Under the conditions of this study, we found that inferred abortions occurred more often during the first trimester (1.37 cases for 100 cow-months at risk) than during other periods, which is consistent with other studies (Forar et al., 1995; Markusfeld-Nir, 1997) that reported a greater frequency of observed and inferred abortions in the period between 35 and 90 days after conception. However, other studies have reported a peak incidence during the second trimester (Thurmond and Picanso, 1990; Carpenter et al., 2006, 2007; Lee and Kim, 2007). Abortions, especially those caused by pathogens, may be “trimester specific”; for example, BVDV-induced abortions occur in the first

188

P. Gädicke, G. Monti / Preventive Veterinary Medicine 110 (2013) 183–189

trimester (Moennig and Liess, 1995), and abortions due to Neospora caninum occur in the second trimester (Wouda et al., 1998; Moura et al., 2012). An additional reason for a high level of inferred early abortions may be that in herds with better than average oestrus detection, a cow with inferred abortion would likely be detected sooner than in herds with poor oestrus detection; thus, this level reflects better management practices (Forar et al., 1996). However, we do not have exact records on oestrus detection for the herds included in the study. We introduced the term inferred abortion because we observed a relatively large time gap between two consecutive AIs. Using a tap drinking water system was associated with a decrease in the risk of abortion in the multivariable model (OR: 0.57; 95% CI 0.40; 0.80) compared to herds that do not have a drinking water system. This association may be explained by the way in which some pathogens, such as leptospirosis, are transmitted. Leptospira is endemic to the area (serovars of L. interrogans and Leptospira borgpeterseii; Zamora and Riedemann, 1999). In addition, the interaction term that includes Leptospira vaccination was statistically significant, providing evidence that leptospirosis is an important determinant of BAS in the study area (Alacid, 2001; Paredes and Moroni, 2005). Repeated abortions in cattle are rare. Our estimation of repeated abortions was higher than that reported by another study (Thurmond and Picanso, 1990), which reported the proportion of abortions to be only slightly higher for cows with previous abortions than for those with no such history (14.5% and 12.1%, respectively). In addition, aborting cows have a shorter productive life in the herd than non-aborting cows (Gädicke et al., 2010) because farmers preferentially cull cows that have aborted (Bell et al., 2010). Some associations between the risk of abortion and factors such as production intensity and other management practices were assessed. No statistically significant association between abortion and cumulative milk production up to 120 days postpartum was found. For single pregnancies, previous studies found associations between a high proportion of abortions and increased milk yield (Melendez and Pinedo, 2007; Silva-del-Río et al., 2009) near the AI time. One possible explanation for our results is that the above-mentioned studies included embryonic and foetal mortality (López-Gatius et al., 2002; Silva-del-Río et al., 2009), whereas we included only the foetal period. Moreover, the increase in milk production has been described as an important reason for the increase in the twinning rate, and cows carrying twins represent an important risk factor for pregnancy failure (López-Gatius et al., 2002; Silva-delRío et al., 2009). Another possible explanation is that in our dataset most farms were likely to be high yielding; therefore, there was not a sufficient number of herds with low production to observe such an effect. Other variables associated with a decreased risk of abortion are the exclusive use of pasture for cows (OR: 0.11) and animal density (OR = 0.74), expressed in mass cows/ha, which may be related to less-intensive management systems (Pulido et al., 2009). However, these results should be considered with care due to possible

non-causal associations. The number of farms that used pasture exclusively for cows was very low (n = 5). Although the association was statistically significant, it may not constitute a causal relationship. However, the non-exclusive use of pasture does not necessarily imply an increased use of concentrate for cows because they may use silage supplementation as well. The potential relationship between concentrate use and abortion could be assessed directly from the feeding records of cows. Unfortunately, these data were not available in the records of the herds analysed. One additional variable associated with a decreased risk of abortion was the time that a calf stays with its dam (OR: 0.83). The time that a farmer allows a calf to stay in the rearing unit is usually connected to the intensification of the production system, and it has been reported that this time decreases in high-producing dairy herds (Svensson and Hiltgren, 2007). Early foetal loss is the most common complication of gestation in high-producing dairy herds (Melendez and Pinedo, 2007; López-Gatius et al., 2008). However, the shorter time for which a calf stays with its dam cannot be directly associated with abortion; instead, it may indicate a more intensively managed herd or may simply be a spurious association. Finally, an important intrinsic variable associated with the risk of abortion is the breed type. We found that the Holstein-type herd has a higher risk of abortion than the other breed types. The genetic association with reproductive traits is complex and difficult to explain, but one study (Arikan and Rodway, 2001) suggests that the Holstein breed appears to have lower B-carotene levels in comparison to other breeds (Guernsey and Jersey breeds). However, to measure the potential effect of breed type, an experimental design, such as a cohort study, would be required to assess this type of relationship. 5. Conclusions The results of this study demonstrate that there is a large underestimation of abortion rates when only farmers’ abortion records are analysed. It would be a useful task for farmers and consultants to analyse reproductive records in more detail given that a considerable number of possible cases of abortion may be identified. Management practices such as a tap drinking system for cows, a closed herd, vaccinating against leptospirosis, the exclusive use of pasture for cows, animal density, the time that a calf stays with its dam and breed type were associated with greater risk of abortion. Conflict of interest statement None of the authors of this paper has a financial or personal relationship with other people or organisations that could inappropriately influence or bias the content of the paper. Acknowledgments This study is part of the first author’s doctoral thesis, founded by a CONICYT (Chilean National Council of

P. Gädicke, G. Monti / Preventive Veterinary Medicine 110 (2013) 183–189

Science and Technology) doctoral fellowship D-21050139. Additionally, financial support for the work was provided by Austral University Projects DID S2006-54 and MECESUP AUS 0005. References Alacid, M., 2001. Descripción epidemiológica del Síndrome Abortivo ˜ Bovino en rebanos sometidos a monitoreo por el Servicio Agrícola y Ganadero en ocho comunas de la Provincia de Valdivia, periodo 1999–2001. Escuela de Medicina Veterinaria, Universidad Austral de Chile. Arikan, S., Rodway, R., 2001. Seasonal variation in bovine luteal concentrations of B Carotene. Turk. J. Vet. Anim. Sci. 25, 165–168. Atkinson, R.A., Cook, R.W., Reddacliff, L.A., Rothwell, J., Broady, K.W., Happer, P.A.W., Ellis, J.T., 2000. Seroprevalence of Neospora caninum infection following an abortion outbreak in a dairy cattle herd. Aust. Vet. J. 78, 262–266. Bell, M.J., Wall, E., Russell, G., Roberts, D.J., Simm, G., 2010. Risk factors for culling in Holstein-Friesian dairy cows. Vet. Rec. 167, 238–240. Campero, C., Moore, D., Odeon, A., Cipolla, A., Odrizola, E., 2003. An etiology of bovine abortion in Argentina. Vet. Res. Commun. 27, 359–369. Carpenter, T.E., Chriel, M., Andersen, M.M., Wulfson, L., Jensen, A.M., Houe, H., Greiner, M., 2006. An epidemiologic study of late-term abortions in dairy cattle in Denmark, July 2000–August 2003. Prev. Vet. Med. 77, 215–229. Carpenter, T.E., Chriel, M., Greiner, M., 2007. An analysis of an earlywarning system to reduce abortions in dairy cattle in Denmark incorporating both financial and epidemiologic aspects. Prev. Vet. Med. 78, 1–11. Cleves, M., Gould, W., Gutierrez, R., 2004. An Introduction to Survival Analysis Using Stata. Stata Press, TX, USA. Dohoo, I., Martin, W., Stryhn, H., 2010. Veterinary Epidemiologic Research, second ed. VER Inc., Charlottetown, Canada. Duchens, A., Apiolaza, P., Melendez, R., 2008. Factors affecting pregnancy loss (30–60 ds) in Holstein cows in central Chile. In: Buiatrics, W.A.f. (Ed.), XXV Jubilee World Buiatrics Congress, Budapest, Hungary, p. 215. Food and Agriculture Organization for the United Nations (FAO). FAOSTAT. 2010. http://faostat3.fao.org/home/index.html#DOWNLOAD STANDARD Forar, A., Gay, J., Hancock, D., 1995. The frequency of endemic fetal loss in dairy cattle: a review. Theriogenology 43, 989–1000. Forar, A., Gay, J., Hancock, D., Gay, C., 1996. Fetal loss frequency in ten Holstein dairy herds. Theriogenology 45, 1505–1513. Gädicke, P., Monti, G., 2008. Aspectos epidemiológicos y de análisis del síndrome de aborto bovino. Arch. Med. Vet. 40, 223–234. Gädicke, P., Vidal, R., Monti, G., 2010. Economic effect of bovine abortion syndrome in commercial dairy herds in Southern Chile. Prev. Vet. Med. 97, 9–19. Gröhn, Y., Erb, H., McCulloch, C., Saloniemi, H., 1990. Epidemiology of reproductive disorders in dairy cattle: association among host characteristics, diseases and production. Prev. Vet. Med. 8, 25–39. Instituto Geográfico Militar, 2008. Mapas de Chile. Online. http://www.igm.cl (accessed 15.06.08). Labèrnia, J., López-Gatius, F., Santolaria, P., López-Béjar, M., Rutllant, J., 1996. Influence of management factors on pregnancy attrition in dairy cattle. Theriogenology 45, 1247–1253. Lee, J.-I., Kim, I.-H., 2007. Pregnancy loss in dairy cows: the contributing factors, the effects on reproductive performance and the economic impact. J. Vet. Sci. 8, 283–288. López-Gatius, F., Santolaria, P., Yániz, J., Rutllant, J., López-Béjar, M., 2002. Factors affecting pregnancy loss from gestation day 38 to 90 in lactating dairy cows from a single herd. Theriogenology 57, 1251–1261. López-Gatius, F., Szenci, O., Bech-Säbat, G., García-Ispierto, I., Serranod, B., Santolaria, P., Yániz, J., 2008. Factors of non-infectious nature affecting early foetal loss in high producing dairy herds in north-eastern Spain. In: Buiatrics, W.A.f. (Ed.), XXV Jubilee World Buiatrics Congress, Budapest, Hungary.

189

Markusfeld-Nir, O., 1997. Epidemiology of bovine abortions in Israeli dairy herds. Prev. Vet. Med. 31, 245–255. Melendez, P., Pinedo, P., 2007. The association between reproductive performance and milk yield in Chilean Holstein cattle. J. Dairy Sci. 90, 184–192. Moennig, V., Liess, B., 1995. Pathogenesis of intrauterine infections with bovine viral diarrhea virus. Vet. Clin. North Am. 11, 477–487. Moore, D.P., Pérez, A., Agliano, S., Brace, M., Cantón, G., Cano, D., Leunda, M.R., Odeón, A.C., Odriozola, E., Campero, C.M., 2009. Risk factors associated with Neospora caninum infections in cattle in Argentina. Vet. Parasitol. 161. Moura, A.B., Souza, A.P., Sartor, A.A., Bellato, V., Texeira, E.B., 2012. Neospora caninum antibodies in dairy cattle of Lages Municipality, Santa Catarina State, Brazil. Arch. Vet. Med. 44, 117–122. Paredes, E., Moroni, M., 2005. Principales causas de abortos diagnosticadas en el período 2003–2005 en fetos bovinos examinados en el Instituto de Patología Animal, Universidad austral de Chile. In: XII Congreso Latinoamericano de Buiatria, Valdivia, Chile. Paredes, E., Pradenas, M., Arancibia, A., 2002. Principales causas de aborto bovino diagnosticadas en el Instituto de Patología Animal de la Universidad Austral de Chile; período 1990–1999. In: XII Congreso Chileno de Medicina Veterinaria, Chillán, Chile. Patitucci, A., Charleston, W., Alley, M., O’Connor, R., Pomoroy, W., 1999. Serological study of a dairy herd with a recent history of Neospora abortion. N Z Vet. J. 47, 28–30. ˜ R., Lemaire, P., Wittwer, F., Orellana, P., Waghorn, G., Pulido, R., Munoz, 2009. Impact of increasing grain feeding frequency on production of dairy cows grazing pasture. Livest. Sci. 125, 109–114. Rabe-Hesketh, S., Skondral, A., 2005. Multilevel and Longitudinal Modeling Using Stata. Stata Press, TX, USA. Rafati, N., Mehrabani-Yeganeh, H., Hanson, T.E., 2010. Risk factors for abortion in dairy cows from commercial Holstein dairy herds in the Tehran region. Prev. Vet. Med. 96, 170–178. Santos, A., Navarro, I., Bracarense, A., Freire, R., Maranda, E., Ogawa, L., Alfieri, A., Freitas, J., Vodotto, O., 2005. Arq. Bras. Med. Vet. Zootec. 57, 545–547. Silva-del-Río, N., Colloton, J., Fricke, P., 2009. Factors affecting pregnancy loss for single and twin pregnancies in a high-producing dairy herd. Theriogenology 71, 1462–1471. StataCorp, 2007. Stata Statistical Software/SE 10.0 for Windows. College Station, TX: StataCorp. LP. Svensson, C., Hiltgren, J., 2007. Associations between housing, management, and morbidity during rearing and subsequent first-lactation milk production of dairy cows in southwest Sweden. J. Dairy Sci. 91, 1510–1518. Szenci, O., 2008. Factors, which may affect reproductive performance in dairy cattle. In: Buiatrics, W.A.f. (Ed.), XXV Jubilee World Buiatrics Congress. Hungarian Veterinary Journal, Budapest, Hungary, pp. 107–111. Thobokwe, G., Heuer, C., 2004. Incidence of abortion and association with putative causes in dairy herds in New Zealand. N Z Vet. J. 52, 90–94. Thurmond, M., Blanchard, P., Anderson, M., 1994. An example of selection bias in submissions of aborted bovine fetuses to a diagnostic laboratory. J. Vet. Diagn. Invest. 61, 269–271. Thurmond, M., Branscum, A., Johnson, W.O., Bedrick, E.J., Hanson, T.E., 2005. Predicting the probability of abortion in dairy cows: a hierarchical Bayesian logistic-survival model using sequential pregnancy data. Prev. Vet. Med. 68, 223–239. Thurmond, M., Picanso, J., 1990. A surveillance system for bovine abortion. Prev. Vet. Med. 8, 41–53. Thurmond, M., Picanso, J., Jameson, C., 1990. Considerations for use descriptive epidemiology to investigate fetal loss in dairy cows. JAVMA 197, 1305–1312. Velis, H., Olivares, R., Neumann, E., 2006. Estudio de la ganadería bovina ˜ regiones del Maule, del Bíobio, de la Araucanía y de Los lagos. Ano 2005 (Ed.) Instituto Nacional de Estadísticas. Santiago, Chile. Wouda, W., Moen, A., Schukken, H., 1998. Abortion risk in progeny of cows after a Neospora caninum epidemic. Theriogenology 49, 1311–1316. Zamora, J., Riedemann, S., 1999. Wild animals as reservoirs of leptospirosis in Chile. Revision of studies in the country. Arch. Med. Vet. 31, 151–156.