The frequency of twinning has increased from 1.4% in 1983 to 2.4% in 19931 and from 3.4% in 1996 to 4.8% in 2004.2 Increased milk production has been reported as an important risk factor for the increased twinning rate.1,3 Lopez et al3 reported that the occurrence of multiple ovulations increased steadily from 0.0% (0/30) to 51.6% (31/60) for cows producing from < 30 to ≥ 50 kg (66 to 110 lb) of milk/d during the 14 days preceding estrus. It is well-known that milk production per cow is steadily increasing in the United States; thus, it is also expected that the occurrence of twin parturitions will continue to rise as milk production increases.4
Twinning is an undesirable condition in dairy cattle because of the detrimental effect on the dams' survival, milk production, and reproduction.5 Furthermore, there is no increase in the number of replacement heifers despite the increase in the total number of calves born because of increased stillbirth for twins, compared with singleton parturition.1,5 Gregory et al6 evaluated natural twinning in beef cattle and reported that birthing twins was associated with higher frequency of retained placenta, dystocia, and stillbirth and increased days from calving to conception. An economic analysis of naturally occurring twinning in dairy cattle estimated that a twin birth costs $108.00 US dollars more to the farmer than a single birth.7
The occurrence of stillbirth parturitions among cows that have singleton parturition is associated with decreased survival and increased calving-to-conception interval, independent of assistance at calving.8 Nielen et al5 reported that stillbirth incidence was 16% higher for twins than for singleton parturitions. Therefore, the negative affect of twin births on the dams' performance may be exacerbated by increased stillbirths.
The purpose of the study reported here was to evaluate the hypothesis that twin parturitions will decrease dams' survival and milk production and increase calving-to-conception interval and that at least 1 stillbirth twin calf will affect the dams' hazard of culling or death, milk production, and calving-to-conception interval.
Materials and Methods
Farms and management—Calvings from 20 farms (19 from upstate New York and 1 from Indiana) were used. The farms were enrolled by convenience; databases were provided by farm owners or farm veterinarians. The criteria for selection were use of a specific dairy farm record database softwarea; use of the 5-point default CES system provided by the software; good calf data records, defined as farms with < 10% missing calf information (dead or alive, male or female, single or twin calving); and good milk production data, either by automatically recording daily milk weights in the parlor or by entering Dairy Herd Improvement Association milk test information into the database. Thirty dairy farms originally offered their records for the research; however, 8 farms were excluded because of poor records on calving ease (> 25% missing data), and 2 dairy farms were excluded because of poor milk production data records (incomplete data for lactations). Most farms were from upstate New York, where 19 farms were enrolled that had from 478 to 4,846 lactating Holstein cows housed in 6-row free stall barns. There was 1 farm from Indiana that had 3,947 lactating Holstein cows housed in 4-row free stall barns. The rolling herd average was 11,800 kg (25,960 lb) of milk. All farms fed a total mixed ration that was formulated to meet or exceed the National Research Council requirements for lactating Holstein cows weighing 650 kg (1,430 lb) and producing 45 kg (99 lb) of 3.5% fat corrected milk. Part of the data used for this study was used in a previous study,8 which excluded twin births and analyzed the effects of stillbirth parturition on the dams' survival and reproductive performance.
Study design and data collection—A retrospective observational cohort study design was used. Data were collected from the dairy records database from May 2005 until September 2006. For all analyses, cows were allocated into 3 cohorts according to the status of the newborn calf: cows that delivered a singleton (either live or dead), cows that delivered live twins, and cows that delivered twins of which at least one of the calves was dead at birth or immediately after. To exclude abortions, all cows that delivered dead calves before 270 days of gestation were excluded from the data analysis. Furthermore, gestation duration was added in the model as a fixed effect.
Additional information collected included season of calving and CES. A 5-point scale CES system was used by all farms enrolled in the study. Calving ease score 1 was defined as calvings that occurred easily without assistance; score 2 was defined as unassisted calvings, but with more difficulty for the cow, compared with CES 1; score 3 was defined as calvings requiring easy assistance from a person; score 4 was defined as vaginally delivered calvings requiring the calf position to be corrected or hard traction to be applied to deliver the calf; and score 5 was defined as calvings requiring fetotomy or caesarean section. There were 70 cows that had a CES of 5, and these cows were included in the CES 4 group.
To facilitate analyses and interpretation of results, the variables CES, parity, and season of calving were dichotomized before they were used in all multivariable analyses. Calving ease scores were divided into unassisted calvings, which were CESs 1 and 2, and assisted calvings, which were CESs 3 and 4. Parity group was divided into primiparous and multiparous cows. Seasons of calving were defined as warm months, from April through September, and cold months, from October through March.
For primiparous cows, the age at first calving was also used as an explanatory variable in the model. As reported in other studies,9 we categorized age at first calving into 3 categories: the low quartile (< 695 days), the 2 middle quartiles (≥ 695 and < 779 days), and the high quartile (≥ 779 days). This categoric variable was then used in the model as a fixed effect.
Statistical analysis—For milk production analysis, the outcome variable used was the average daily milk weight, which is provided on a month of lactation basis. Milk weights were calculated and extracted from the database. Data were structured so that each cow had 10 lines of information and each line contained the average daily milk weight for a specific month of lactation, a number from 1 to 10 that consisted of the respective milk test number, and all of the other information used in the statistical model, which did not change from line to line. A single table for the entire data set was created in a software program.b A separate table was generated that contained only 1 line of data/cow; this spreadsheet was used to perform descriptive statistics and survival and reproductive performance analyses. To assess whether dichotomization of the variable CES affected its capability to explain variability, we performed a sequential ANOVA with milk production as the outcome variable and dichotomized CES and 5-point scale CES as the independent variables. The amount of leftover sum of squares explained by having a 5-point scale CES, after accounting in the model the dichotomized CES, was low; therefore, the amount of variance explained by having extra levels in the variable CES was little, and we used the dichotomized CES for a more parsimonious model.
A mixed modelc was used to analyze the milk production data. The outcome variable was test day milk weight (kilograms), which was modeled as a Gaussian variable. The normality of the residuals assumption was assessed by plotting the residuals against the predicted values by use of a quantile-quantile plot. The following independent variables were entered in the model as fixed effects: twin (1 = singletons, 2 = live twins, and 3 = at least 1 dead twin), calving ease group (assisted or unassisted), calving season (calving at warm or cold season), reproductive status (pregnant or nonpregnant at the end of data collection), gestation duration, age at first calving (used for primiparous cows), previous lactation ME305 for multiparous cows, and test day number (from 1 to 10). Primiparous cows were modeled separately from multiparous cows because some relevant variables were available for only 1 group or the other, such as ME305, which was used in the model for multiparous cows but not for primiparous cows, and age at first calving, which was only relevant for primiparous cows. Farm and cow were included in the model as random effects; the random effect of cow was nested within farm. Interactions between twin and all other fixed effect variables were tested; the same was done for the variable milk test number. Variables were manually excluded from the model when their interaction terms and respective main effects were not significant (P ≥ 0.05); the nonsignificant variables were removed in descending order of probability value, until only significant variables (P < 0.05) were left in the model. Significant interaction terms between categoric variables were further analyzed via stratified analysis.
The data were longitudinally collected and therefore had a series of repeated measures of milk weights throughout the lactation, which implied that data points were correlated within each research subject. To account for this, a unique identification number was assigned for each cow, and this variable was controlled in the model as a random effect. A first-order autoregressive covariance structure was used to model the covariance within cow.
The effect of twins on cow survival was analyzed by use of the Cox proportional hazard.d The time variable used in the model was days from calving until culled, death, or censored. A censoring variable was used to identify animals that had been sold or died (not censored) and animals that were alive at the end of data collection (censored).10 The variables used in the model were calf status (singleton, live twin, and at least 1 dead twin), parity (primiparous and multiparous), CES (assisted and unassisted), and season of calving (warm and cold). For the level 3 variable calf status, the baseline category used in the computation of hazard ratios was the group of cows that delivered singleton calves. Farm was used in the model as a strata variable to adjust for the effect of location. Backward variable elimination considering main effects and 2-way interactions of twin with all of the other independent variables on cow survival was performed.10 Variables were retained in the model when either their main effect or their interaction with the variable twin was significant (α ≤ 0.05). To evaluate the difference in the hazard of culling or death between cows that had live twins and cows that had at least 1 dead twin, a contrast comparing the 2 groups was made.e
Effect of twins on calving to conception interval was evaluated by use of Cox proportional hazards analysis.d The time series variable for this model was the calving-to-conception interval or days from calving to censure. A censoring variable was used to identify cows that were pregnant (not censored) from cows that were not pregnant by the end of data collection (censored).The variables used in this model were calf status (singletons, live twins, and at least 1 dead twin), parity (primiparous and multiparous), CES (assisted and unassisted), and season of calving (warm and cold). Farm was also used in the model as a strata variable to adjust for the effect of location. The variable selection criteria and the interaction tests used for this model were the same as those for the dam's survival model. To evaluate the difference in the hazard of conceiving between cows that had live twins and cows that had at least 1 dead twin, a contrast comparing the 2 groups was made.e
Kaplan-Meier survival analysisf was performed to determine the differences in survival curves (reproduction and survival) among the 3 cohorts (singletons, live twins, and at least 1 dead twin). Data are reported as mean ± SE.
Results
Descriptive statistics—A total of 33,868 cows were used. The prevalence of twin calvings was 4.6%, with a range among farms of 2.7% of 524 to 7.0% of 1,219. Of all twin parturitions, 74.1% of 1,569 cows had 2 live calves, 13.4% had 1 live calf and 1 dead calf, and 12.4% had 2 dead calves. Prevalence of twin parturitions for primiparous cows was 1.3% of 12,050, of which 63.5% were live-calf parturitions and 36.5% were parturitions with at least 1 dead twin). Prevalence of twins for multiparous cows was 6.5% of 21,818, of which 75.3% were live-calf parturitions, and 24.7% were parturitions with at least 1 dead twin. A total of 10.4% of all calvings were classified as CES of ≥ 3, with a range among farms of 2.4% of 1,219 to 36.0% of 591).
Daily milk weight (kg) or difference in milk weight in primiparous cows with categoric fixed effects used in a mixed model.
Least square mean ± SE values for monthly milk production of primiparous cows that had singleton calves (black line), live twins (dark gray line), or at least 1 dead twin calf (light gray line).
Citation: Journal of the American Veterinary Medical Association 231, 9; 10.2460/javma.231.9.1390
Least square mean ± SE values for monthly milk production of primiparous cows that had singleton calves (black line), live twins (dark gray line), or at least 1 dead twin calf (light gray line).
Citation: Journal of the American Veterinary Medical Association 231, 9; 10.2460/javma.231.9.1390
Least square mean ± SE values for monthly milk production of primiparous cows that had singleton calves (black line), live twins (dark gray line), or at least 1 dead twin calf (light gray line).
Citation: Journal of the American Veterinary Medical Association 231, 9; 10.2460/javma.231.9.1390
Effect of twins on dams' milk production—Milk production was modeled separately for primiparous and multiparous cows because some relevant variables were only available for 1 group or another, such as ME305, which was used in the model for multiparous cows but not for primiparous cows, and age at first calving, which was only relevant for primiparous cows and therefore not used for multiparous cows. For primiparous cows, twinning significantly decreased milk yield, compared with singleton parturitions. Cows that gave birth to singletons produced 1.21 kg/d (2.66 lb/d) more (P = 0.005) milk than cows that had live twins and 1.93 kg/d (4.25 lb/d) more (P = 0.002) milk than cows that had at least 1 dead twin (Table 1). Although there was a numeric difference (mean ± SE, 0.71 ± 0.6 kg/d [1.56 ± 1.32 lb]) in milk yields between cows with live twins and those with dead twins, a significant (P = 0.5) difference was not observed. The mean ± SE LSMs were 33.10 ± 0.04 kg/d (72.82 ± 0.09 lb/d), 31.88 ± 0.39 kg/d (70.14 ± 0.86 lb/d), and 31.17 ± 0.48 kg/d (68.57 ± 1.06 lb/d) for cows that gave birth to singletons, live twins or at least 1 dead twin, respectively (Figure 1).
Reproductive status at the end of data collection was also highly significant in the model (Table 2). Nonpregnant cows produced 2.57 kg/d (5.65 lb/d) less than did cows that were pregnant by the end of the data collection. The mean ± SE LSMs for nonpregnant cows and pregnant cows were 30.77 ± 0.32 kg/d (67.69 ± 0.70 lb) and 32.84 ± 0.25 kg/d (72.25 ± 0.55 lb/d), respectively. By the end of data collection, an interaction of twin group and reproductive status was also detected. Daily milk yields did not differ between singletons, live twins, and dead twin groups for cows that were pregnant by the end of data collection; however, daily milk yield was lower for cows with at least 1 dead twin, compared with cows with live twins, and was lower for cows with singletons, compared with the nonpregant cows.
Milk yield (LSM ± SE; kg/d) in primiparous cows of various reproductive statuses and age groups that had singleton calves, live twins, or at least 1 dead twin calf.
Age at first calving was highly significant in the model. The mean ± SE LSMs were 31.66 ± 0.40 kg/d (69.65 ± 0.88 lb/d), 31.0 ± 0.30 kg/d (68.20 ± 0.66 lb/d), and 33.49 ± 0.40 kg/d (73.68 ± 0.88 lb/d) for the low-, middle-, and high-age groups, respectively. There was no difference in milk yields between cows in the lowand middle-age groups. The high-age group of primiparous cows had higher daily milk yields, compared with lowand middle-age groups (P < 0.001). A highly significant interaction of twin group and age group on milk production was detected (Table 2). Primiparous cows in the high-age group did not have decreased milk production because of having live twins or at least 1 dead twin, compared with cows that had singletons. However, daily milk yield was decreased in cows that had live twins and at least 1 dead twin, compared with cows that had singletons, for the middleand low-age groups, which explained the interaction of twin groups and age groups on daily milk yields.
The variable gestation duration was significant and retained in the model; however, the interaction of twin group and gestation duration was not significant and therefore was excluded from the model. The variables season at calving and CES were also included in the model, but were not significant and were not retained. The detrimental effect of twins on daily milk yields did not decrease as days in milk increased (Figure 1); the interaction of twin group and milk test number was not significant.
Daily milk weight (kg) or difference in milk weight in multiparous cows with categoric fixed effects used in a mixed model.
Least square mean ± SE values for monthly milk production of primiparous cows. See Figure 1 for key.
Citation: Journal of the American Veterinary Medical Association 231, 9; 10.2460/javma.231.9.1390
Least square mean ± SE values for monthly milk production of primiparous cows. See Figure 1 for key.
Citation: Journal of the American Veterinary Medical Association 231, 9; 10.2460/javma.231.9.1390
Least square mean ± SE values for monthly milk production of primiparous cows. See Figure 1 for key.
Citation: Journal of the American Veterinary Medical Association 231, 9; 10.2460/javma.231.9.1390
Milk yield (LSM ± SE; kg/d) in multiparous cows of various reproductive statuses and age groups that had singleton calves, live twins, or at least 1 dead twin calf.
For multiparous cows, daily milk yields were also significantly decreased by twin parturitions. The mean ± SE LSMs for daily milk yields were 36.50 ± 0.1 kg/ d (80.30 ± 0.22 lb/d), 35.73 ± 0.1 kg/d (78.61 ± 0.22 lb), and 35.0 ± 0.2 kg/d (77.0 ± 0.44 lb/d) for cows with singletons, live twins, and at least 1 dead twin, respectively. Both twin groups produced significantly less milk, compared with the singleton group. Furthermore, for multiparous cows, the decrease in daily milk yield for cows with live twins, compared with cows with at least 1 dead twin (0.78 ± 0.3 kg/d [1.72 ± 0.66 lb/d]), was significant (P = 0.027; Table 3).
As for primiparous cows, the variable reproductive status at the end of the data collection was significant. Nonpregnant cows produced 0.49 kg/d (1.1 lb/d) less milk than pregnant cows (P < 0.001). The interaction of twin group and reproductive status on daily milk yields was also significant. Season at calving was also significant (P < 0.001); cows that calved during the warm season produced 1.09 kg/d (2.40 lb/d) less than cows that calved in the cold season. The interaction of twin group and season at calving on daily milk yield was not significant and was excluded from the model. In contrast to the primiparous cows' model, the variable calving ease group was significant. Cows with a nonassisted calving produced 0.83 kg/d (1.83 lb/d) more milk than did cows with an assisted calving. Nevertheless, the interaction of twin group and CES on daily milk yield was not significant. The interaction of twin group and milk test number was significant; the difference in milk production between cows with singletons, live twins, and at least 1 dead twin decreased with time; by the tenth milk test number, there was no difference in milk production among the groups (Figure 2). The variables gestation duration, previous lactation ME305, and milk test number and the interactions of twin group and milk test number and of gestation duration and milk test number on milk production were also significant and controlled for in the model.
Kaplan-Meier survival analysis of survival in cows that had singletons (solid line), live twins (middle interrupted line), or at least 1 dead twin (inner interrupted line).
Citation: Journal of the American Veterinary Medical Association 231, 9; 10.2460/javma.231.9.1390
Kaplan-Meier survival analysis of survival in cows that had singletons (solid line), live twins (middle interrupted line), or at least 1 dead twin (inner interrupted line).
Citation: Journal of the American Veterinary Medical Association 231, 9; 10.2460/javma.231.9.1390
Kaplan-Meier survival analysis of survival in cows that had singletons (solid line), live twins (middle interrupted line), or at least 1 dead twin (inner interrupted line).
Citation: Journal of the American Veterinary Medical Association 231, 9; 10.2460/javma.231.9.1390
Effect of twins on dams' calving-to-conception interval—The occurrence of twin parturition significantly increased the calving-to-conception interval, compared with singleton parturition. Cows that had twins (live or dead) had a 22% decrease in the hazard of conceiving, compared with cows that had singletons (hazard ratio, 0.78; P < 0.001). Cows that had live twins were at a 23% increased hazard of conceiving, compared with cows that had at least 1 dead twin (hazard ratio, 1.23; P = 0.041). Multiparous cows had a 10% decreased hazard of conceiving, compared with primiparous cows (hazard ratio, 0.90; P < 0.001). The variables season at calving and calving ease group were also highly significant and retained in the model (Table 4). Assisted calving also affected reproductive performance; cows that had assisted parturitions were at a 28% decreased hazard of being detected pregnant, compared with cows that had an unassisted parturition (hazard ratio, 0.72; P < 0.001).
Cox proportional hazards analysis of dam survival and calving to conception interval.
A Kaplan-Meier survival analysis was performed to determine the differences in days from calving-to-conception among the 3 groups (cows with singletons, live twins, or at least one dead twin). There was a significant (P = 0.025) difference in the median interval from calving to conception between cows that had live twins and dead twins (220 days vs 250 days; Figure 3). The median interval from calving to conception for cows that had singletons was 175 days, which was significantly less than that for both twin groups.
Effect of twins on dams' survival—Twin parturition significantly increased the hazard of culling or death, compared with singleton parturitions (Table 4). Different contrasts were used to make comparisons among the 3 groups of twins. Cows that had twins (live or dead) were at a 42% increased hazard of culling or death, compared with cows that had singletons (hazard ratio, 1.42; P < 0.001). The occurrence of at least 1 dead twin did not significantly affect survival time, compared with cows that had live twins (hazard ratio, 0.89; P = 0.309). Multiparous cows were at a 122% increased hazard of culling or death, compared with primiparous cows (P < 0.001). Season at calving and the interaction of twin group and season at calving significantly influenced hazard of culling or death and were retained in the model. Cows that had twins in the warm season were at a 19% increased hazard for culling or death, compared with cows that had twins in the cold season (P = 0.041). Cows that had assisted calving were at higher hazard of culling or death, compared with cows that had unassisted parturitions (hazard ratio, 1.13; P < 0.001; Table 5).
Kaplan-Meier survival analysis of survival in cows that had singletons (solid line), live twins (middle interrupted line), or at least 1 dead twin (inner interrupted line).
Citation: Journal of the American Veterinary Medical Association 231, 9; 10.2460/javma.231.9.1390
Kaplan-Meier survival analysis of survival in cows that had singletons (solid line), live twins (middle interrupted line), or at least 1 dead twin (inner interrupted line).
Citation: Journal of the American Veterinary Medical Association 231, 9; 10.2460/javma.231.9.1390
Kaplan-Meier survival analysis of survival in cows that had singletons (solid line), live twins (middle interrupted line), or at least 1 dead twin (inner interrupted line).
Citation: Journal of the American Veterinary Medical Association 231, 9; 10.2460/javma.231.9.1390
A Kaplan-Meier survival analysis was also performed to determine the differences in survival time among the 3 calf status groups (Figure 4). Medium survival time was not available because > 60% of the cows were still alive by the end of the data collection period.
Discussion
The twinning rate observed in this study was 4.6%, which was similar to the 4.2% twinning rate observed in > 2 million calving events over 8 years in Minnesota dairies.2 The incidence of twinning has been increasing over time from 1.4% in 1983 to 2.4% in 1993 and from 3.4% in 1996 to 4.8% in 2004.1,2 Increased milk production was reported to be the most important risk factor for increased twinning rate.1 Other risk factors include cystic ovarian disease and the use of hormones and antimicrobials.1,5,11 In the present study, primiparous cows had lower twinning and stillbirth rates (within twin parturitions), compared with multiparous cows, which is in agreement with previous reports1,2,5 and may be explained by the lower double ovulation rates, compared with multiparous cows. Twin births have been associated with increased risk of stillbirths and abortions, and stillbirths have been associated with decreased performance in the subsequent lactation.5,8,12 Bicalho et al8 found a significant increase in the calving-to-conception interval as well as the culling risk for cows that had stillbirths; in that study, only singleton parturitions were included. Mangurkar et al12 reported that cows that had stillbirths produced 100 to 400 kg (220 to 880 lb) less milk, compared with cows that had live calves. The effects of stillbirths on lactation performance are similar to the effects of twinning. Therefore, cows that had twins were allocated into those that had at least 1 stillborn calf and those that had both live twins to determine whether the detrimental effects were not only attributable to a higher proportion of stillbirths and whether there was an additive effect of stillbirth on twinning.
Twinning has been reported as a cause for decreased milk production in the subsequent lactation.7 The effects of twinning on milk production were analyzed separately for primiparous and multiparous cows in the present study to allow important variables to be included in the model. Overall, primiparous cows that gave birth to singletons produced more milk than cows that had live twins and cows that had at least 1 stillborn twin. Although a numeric difference in milk yields between the cows with live twins and at least 1 dead twin was present, a significant difference was not observed. In contrast, multiparous cows that had at least 1 stillborn twin produced significantly less milk than cows that had both live twins. The difference in milk production persisted throughout the lactation. Multiparous cows that produced singletons produced significantly more milk than did cows that had live twins, and cows that had live twins produced significantly more milk than did cows that had at least one stillborn twin. The difference in milk production decreased with the number of milk tests and was similar by the fourth milk test. This was in contrast to primiparous cows, in which the difference in milk production persisted. Nielen et al5 found a 144-kg (316.8-lb) reduction in milk production by 100 days in milk in cows that had twins, compared with cows that had singletons; for cows that had at least one twin stillborn, the difference was an additional 437 kg (961.4 lb). Moreover, the difference in milk production in that study persisted past 270 days in milk. A 207-kg (455.4-lb) difference was found between cows that had singletons versus twins, and an additional 520-kg (1,144-lb) difference was found if at least one of the twins were stillborn; however, the difference in milk production was diminished by the end of lactation, similar to the pattern observed in multiparous cows in the present study.5
It is important to emphasize that milk production differences between cows with twins versus cows with singletons could be even bigger considering that cows with twins were at higher risk of culling or death and therefore contributed less milk production data than cows with singletons.
For primiparous cows, age at first calving was used to group cows and the high-age group cows produced more milk than did cows in the middleand low-age group, as previously reported.9 In the present study, live and stillbirth twinning did not significantly affect milk production in the high-age group, but calving twins (live or stillborn) affected cows in the middle-age group; in addition, calving stillborn twins significantly affected milk production in the low-age group. Lowage group cows have been reported to have a higher incidence of stillbirths (19.8%), compared with medium(16.1%) and high(13.5%) age group cows.9 When accounting for all variables in the model, the variable calving ease group (assisted vs nonassisted calving) was not significant and was excluded from the primiparous cow model. In a review by Fourichon et al,13 it was reported that in 8 of 13 studies, there was no effect of dystocia on milk yield, whereas only 5 found losses. In the present study, it is possible that the effect of calving ease group was accounted for by the other variables included in the model, such as age group of the heifers or even the occurrence of twin parturitions, which might explain its exclusion from the model. However, calving ease group was significant for the multiparous cow model; multiparous cows that had assisted calvings produced 0.8 kg (1.76 lb) less milk than cows without assisted calving.
Reproductive status at the end of the study period (pregnant vs nonpregnant) was significant for milk production in primiparous and multiparous cows. Furthermore, interactions of the variables twin parturition and reproductive status on milk production were significant for both models (primiparous and multiparous); the occurrence of twin parturition decreased milk production with more intensity among the cows that were not pregnant by the end of the study period, compared with pregnant cows. The authors believe that there is no causal relationship between subsequent pregnancy and milk production, although it is biologically plausible to infer that cows that were pregnant by the end of the study were not as affected by postpartum disorders that have been associated with twining and stillbirths, such as retained fetal membranes, hypocalcemia, metritis, and displaced abomasums, which affect milk production and reproduction simultaneously. As reported by Nielen et al,5 intrauterine antimicrobial treatment was higher among twinning cows, and this was explained by a higher frequency of retained placenta and endometritis after twin calving.
The effect of twinning on fertility in the present study was a 22% decreased hazard of pregnancy. The median calving-to-conception interval for cows that had live twins was 220 days and was 250 days if at least 1 twin was stillborn, compared with 175 days for cows that had singletons. Previous studies5,14,15 also found that the mean calving-to-conception interval as well as inseminations per conception were increased for cows that had twins. Twin parturitions increase the risk of several diseases that are risk factors for impaired reproductive performance, such as retained fetal membranes, metritis, and abortion.1,5,11,16 In the present study, abortions were excluded by removing all cows that had stillborn calves before 270 days of gestation, and gestation duration was added to all multivariable models. Thus, the effects attributed to twin parturitions in this study were not confounded by occurrence of abortions or premature calvings. The authors have reported that the occurrence of stillbirth parturition among cows with singleton calves increased the median days that cows were nonpregnant by 88.8 In the present study, cows with at least 1 stillborn twin calf had an increase in the calvingto-conception interval of 30 days, compared with cows with 2 live twins, and an increase of 75 days, compared with cows with singletons. It is possible that intrauterine bacterial growth is accelerated by the presence of a dead calf in the uterus; this would increase the risk of metritis, leading to impaired reproductive performance.
In the present study, twinning increased the risk of culling by 42%. Nielen et al5 reported that the relative risk for failure to become pregnant was 1.4 for cows that had twins, compared with cows that had singletons. Moreover, cows that had twins were culled more often for infertility than for low milk production, and cows that had singletons were culled more often for low milk production than for infertility.5 A recent study by Thomsen et al17 used a so-called loser cow score that was based on clinical examination of the individual cow. The loser cow score included the clinical signs lameness, tarsal lesions, other cutaneous lesions, and condition of coat, and the clinical sign scores were then converted to an index. Among several factors associated with increased loser cow score, cows that gave birth to twins were 238% as likely to be classified as a loser cow.17 In that study, the hazard ratio for death or euthanasia was 5.69 for loser cows, compared with nonloser cows. In the present study, cows that had twins in the warm season were at a 19% increased hazard for culling or death, compared with cows that had twins in the cold season. Heat stress decreases milk production and reproductive efficiency.18,19 Therefore, twin parturitions and heat stress may have a synergetic relationship that amplifies the hazard of culling or death.
ABBREVIATIONS
| CES | Calving ease score |
| ME305 | Mature equivalent milk for a 305-day lactation |
| LSM | Least square means |
Dairy Comp 305, Valley Agricultural Software, Tulare, Calif.
JMP, SAS Institute Inc, Cary, NC.
MIXED, SAS Institute Inc, Cary, NC.
TPHREG, SAS Institute Inc, Cary, NC.
CONTRAST statement of TPHREG, SAS Institute Inc, Cary, NC.
MedCalc for Windows, version 9.2.0.0, MedCalc software, Mariakerke, Belgium.
References
- 1.↑
Kinsel ML, Marsh WE & Ruegg PL, et al. Risk factors for twinning in dairy cows. J Dairy Sci 1998;81:989–993.
- 2.↑
Silva del Rio N, Stewart S & Rapnicki P, et al. An observational analysis of twin births, calf sex ratio, and calf mortality in Holstein dairy cattle. J Dairy Sci 2007;90:1255–1264.
- 3.↑
Lopez H, Caraviello DZ & Setter LD, et al. Relationship between level of milk production and multiple ovulations in lactating dairy cows. J Dairy Sci 2005;88:2783–2793.
- 4.↑
Lucy MC. Reproductive loss in high-producing dairy cattle: where will it end? J Dairy Sci 2001;84:1277–1293.
- 5.↑
Nielen M, Schukken YH & Scholl DT, et al. Twinning in dairy cattle: a study of risk factors and effects. Theriogenology 1998;32:845–862.
- 6.↑
Echternkamp SE, Gregory KE & Dickerson GE, et al. Twinning in cattle: III. Effect of twinning on dystocia, reproductive traits, calf survival, calf growth and cow productivity. J Anim Sci 1990;68:3133–3144.
- 7.↑
Beerepoot GM, Dykhuizen AA & Nielen M, et al. The economics of naturally occurring twinning in dairy cattle. J Dairy Sci 1992;75:1044–1057.
- 8.↑
Bicalho RC, Galvão KN & Cheong SH, et al. Effect of stillbirth on dam's survival and reproduction performance in Holstein dairy cows. J Dairy Sci 2007;90:2797–2803.
- 9.↑
Ettema JF, Santos JEP. Impact of age at calving on lactation, reproduction, health, and income in first-parity Holstein on commercial farms. J Dairy Sci 2004;87:2730–2742.
- 10.↑
Cantor AB. Extending SAS survival analysis techniques for medical research. Cary, NC: SAS Institute Inc, 1997;112–113.
- 11.
Bendixen PH, Oltenacu PA, Andersson L. Case-referent study of cystic ovaries as a risk indicator for twin calvings in dairy cows. Theriogenology 1989;31:1059–1066.
- 12.↑
Mangurkar BR, Hayes JF, Moxley JE. Effects of calving ease-calf survival on production and reproduction in Holsteins. J Dairy Sci 1984;67:1496–1509.
- 13.↑
Fourichon C, Seegers H & Bareille N, et al. Effects of disease on milk production in the dairy cow: a review. Prev Vet Med 1999;41:1–35.
- 14.
Chapin CA, Van Vleck LD. Effects of twinning on lactation and days open in Holsteins. J Dairy Sci 1980;63:1881–1886.
- 15.
Erb RA, Morrison RA. Effects of twinning on reproductive efficiency in a Holstein-Friesian herd. J Dairy Sci 1959;42:512–519.
- 16.
Eddy RG, Davies O, David C. An economic assessment of twin births in British dairy herds. Vet Rec 1991;129:526–529.
- 17.↑
Thomsen PT, Østergaard S & Houe H, et al. Loser cows in Danish dairy herds: risk factors. Prev Vet Med 2007;79:136–154.
- 18.
Srikandakumar A, Johnson EH. Effect of heat stress on milk production, rectal temperature, respiratory rates and blood chemistry in Holstein, Jersey and Australian milking zebu cows. Trop Anim Health Prod 2004;36:685–692.
- 19.
Hansen PJ, Arechiga CF. Strategies for managing reproduction in the heat-stressed dairy cow. J Anim Sci 1999;77(suppl 2):36–50.
