A Meta – analysis of the Impact of Parity on Dystocia and Stillbirth in Holstein Cattle

Received: Revised: Accepted: 2014–04–06 2014–08–05 2014–08–07 Dystocia and stillbirth are major factors reducing the productivity of dairy cattle. The objective of this study was to determine the effect of parity on the rates of dystocia and stillbirth. A meta–analysis was conducted to investigate the impact of first parity (primiparous) and later parities (multiparous) on dystocia and stillbirth in Holstein cattle. A total of 30 and 19 papers were analyzed for evaluation of two traits. Results revealed that primiparous cattle are more susceptible to dystocia [Odds Ratio (OR) = 2.68, 95% Confidence interval (CI) 2.51 to 2.85], stillbirth (OR = 2.18, 95% CI 1.84 to 2.58) as compared with multiparous. These results supported the opinion about the importance of considering primiparous and multiparous as different traits in genetic evaluation and shed light on the importance of improving genetics and environment of heifers to minimize the effect of dystocia and stillbirth in Holstein cattle. All copyrights reserved to Nexus® academic publishers


INTRODUCTION
Meta-analyses can be defined as systematic reviews with pooled data (Ton et al., 2007).It was considered as a valuable method with unique properties: establishing whether scientific findings are consistent (Cook et al., 1998) and can be generalized across populations (Burrin and Britton, 1986), limit bias, improve reliability and accuracy of conclusions (Collett, 1994) and increase the power and precision of treatment effects (Bell and Bauman, 1997).
In recent years, breeders have shown increasing interest in selection of functional traits in dairy cattle (Mark, 2004), therefore they have focused to shift selection from traits that increasing-profit to reducing-costs traits (De Maturana, 2007).Health management has been emphasis in order to minimize losses due health disorders (Beaudea et al., 2000).
Dystocia and stillbirth are related terms; as dystocia associated with approximately 50% of calf mortality cases at birth (Mee, 2008).These two traits may result in direct losses due to calf and, dam mortality and premature culling, as well as indirect costs due to additional veterinary services, labor and treatment (Szucs et al., 2009).
Dystocia and stillbirth are generally scored on categorical or binary scales which make them sensitive to subjectivity (Dekkers, 1994).
Dystocia may be defined as delayed or difficult parturition.It's an important problem in Holstein cattle since one birth of every 5 to first parity dams need assistance (Philipsson, 1996).Stillbirths are defined as a calf that dies just before, during, or within the first 24 to 48 h after birth with at least 260 days of gestation (Meyer et al., 2001;Chassagne et al., 1999).
Several studies revealed that primiparous and multiparous cows clearly differ in the rate of dystocia and stillbirths.Meyer et al., (2001) confirmed that statistical analysis of the two traits could be best when considering primiparous and multiparous cows as separated traits.
The aim of this study is to view an extract of estimations (Odds ratio) for the effect of primiparous and multiparous on dystocia and stillbirth in Holstein cattle.

MATERIALS AND METHODS
Extensive literature search of scientific electronic search engines (PubMed, Google Scholar, CAB, ISI Web of Knowledge, Science Direct, SciQuest, and Scirus) was conducted to identify primary studies carried out between 1980 and 2013.Following rigorous screening for appropriate subject matter, high quality of studies, and adequate statistical reporting, were extracted for meta-analysis.Several keyword combinations (dystocia, stillbirth, odds ratio, calving problems, Holstein cattle, meta-analysis) were used.Criteria examined included randomization of study, recording, statistical analysis.Analytic techniques described by Dohoo et al., (2003).
Articles were selected to meet the following criteria: (1) published in English (2) published as peer reviewed original articles (3) must had information about dystocia and stillbirth in first and later parity (4) articles of Holstein cows only were included in the analysis (5) The non-peer-reviewed studies were assessed and included in the meta-analysis if they met the selection criteria.

A Meta-analysis of the Impact of Parity on Dystocia and Stillbirth in Holstein Cattle
ISSN: 2307-8316 (Online); ISSN: 2309-3331 (Print) The articles selected were generally American or European as shown in Table 2 and 3. Thirty papers were used to evaluate the impact of primiparous and multiparous cows on dystocia and nineteen papers for stillbirth.The scoring of dystocia was not constant in all papers, whereas stillbirth was recorded as dichotomous.The definition of dystocia was not standardized across studies (Table 1).Most studies classified dystocia within 5 categories including unassisted, easy, moderate, difficult, and very difficult.Some studies were recorded dystocia with four categories: = easy (nonassisted), 2 = moderate assistance (veterinarian called as precaution), 3 = difficult, 4 = very difficult with veterinary assistance.Some else recorded dystocia with three categories: no assist, easy and hard or two categories: unassisted and assisted.

Study name
No

Data Analysis Analysis of Potential Publication Bias
A funnel plot: two modes were available, one which plots a study's effect size against its standard error and another which plots effect size against precision.
In the absence of bias the plot would be symmetric about the summary effect (Duval andTweedie2000): Test the rank correlation (Kendall's tau) between the standardized effect size and the variances (or standard errors) of these effects (Begg and Mazumdar, 1994).
Test the standardized effect (the regression of effect size divided by standard error on precision (inverse of standard error).

Analysis of Heterogeneity
Heterogeneity of the estimated OR was assessed using the Cochran's Q statistic chi square test (Egger et al., 2001).If there was evidence of heterogeneity, then a random model (inverse variance) is preferred.The degree of heterogeneity was assessed by the I 2 (I squared) statistic.This describes the percentage of the variability in effect estimates that is due to heterogeneity rather than sampling error (chance) (Higgins et al., 2003).

Effect of Primiparous and Multiparous on Dystocia
Results shows that I 2 = 99.72, and P = 0.000.An I 2 value >50 may be considered indicative of substantial heterogeneity.In such case a random model is considered more suitable than fixed model (Rabiee et al., 2012).Estimated OR with a random model for dystocia is OR = 2.68, 95% CI 2.51 to 2.85 and the corresponding estimates in a fixed model is OR = 2.61, 95% CI 2.60 to 2.61 (Table 4).
It's obvious from   10615.92 whereas the adjusted OR was 2.42, 95% CI (2.26, 2.60), 13133.952.Egger's linear regression method, quantifies the bias captured by the funnel plot.Egger's method uses the actual values of the effect sizes and their precision.The rank correlation test of Begg and Mazumdar (1994) showed that there was no significant correlation between effect and study size (P = 0.39).This was also confirmed by the regression test of Egger et al. (1997), which showed no significant association between study size and effect (intercept = 1.45,P = 0.36).
Heterogeneity in studies could be belonging to many reasons such as: studies conducted by different people, in different areas, with different definitions and at different times, which create a heterogeneous population of studies.Differences between studies in terms of the definition or measurement of outcomes, may lead to differences in observed effects (Lean et al., 2009).As I 2 (I square) was significant (99.72), hence the sources of heterogeneity of response were investigated by meta-regression.
In our research, heterogeneity could be attributed to differences in definition of dystocia (categories).To investigate the validity of using this factor as predictor factor, data were analyzed using ANOVA.T-test was confirmed the significant (P < 0.05) differences between OR estimates.Hence, data were subjected to meta-regression.Two types of regression were used: fixed effect regression which shows that the slope is 0.13, 95%CI (0.12, 0.13), P = 0.000 with intercept 0.32, 95%CI (0.30, 0.34), P = 0.000 (Figure 2) and mixed effect regression which shows that the slope is 0.11, 95%CI (0.02, 0.20), P = 0.000 with intercept 0.52, 95%CI (0.12, 0.91), P = 0.000 (Figure 3).The significant effects of two regressions confirmed the effect of dystocia categories on the value of log OR.It was shown from the two Figures (2, 3) that the OR increased as category increasing.
Figure ( 4) illustrate each study represented by a circle proportional to its weight in the analysis.This view identifies which studies have the greatest impact on the slope of the regression line.Studies with five categories have more impact on slope.Forest plots were used to provide illustration of the calculated OR per study as well as the overall pooled effect of all studies in the plot.The forest plot is a graphical presentation of the results that displays the point estimate and confidence interval of the effect observed in each study, along with the summary estimate and its confidence interval (Dohoo et al., 2003).A forest plot of the studies of dystocia was shown in Figure ( 5).

Effect of Primiparous and Multiparous on Stillbirth
Results shows that I 2 = 99.92, and P = 0.000.Estimated OR with random model for stillbirth is OR = 2.18, 95% CI 1.84 to 2.58 and the corresponding estimates in fixed model is OR = 2.06, 95% CI 2.05 to 2.07 (Table 5).Duval and Tweedie (2000) reported that: when there was no missing study in the funnel plot, the observed and adjusted OR is identical.Figure (6) shows that there was no bias as the studies were distributed symmetrically.The estimate of observed and adjusted OR are identical.(OR = 2.18, 95% CI 1.84 to 2.58).
Egger's linear regression method was applied and results shows that the intercept (β0) is 1.76, 95% CI (21.70, 25.24), with t = 0.158, df = 17.The one-tailed p-value is 0.43, and the two-tailed p-value is 0.87.Begg and Masumdar rank correlation was used also and the Tau value was -0.29, P (one tailed) = 0.04 and P (two tailed) = 0.08.It's obvious that the heterogeneity is detected in stillbirth but when there is heterogeneity that cannot readily be explained, one analytical approach is to incorporate it into a random effects model.In such case, we were unable to define the causing factors and then unable to apply metaregression.A Forest plot for stillbirth was shown in Figure (7).
Results revealed that the test of heterogeneity confirmed the existance of a substantial heterogeneity in dystocia and stillbirth.So the estimation of OR by random model is more accurate in two mentioned traits.The OR of dystocia (2.68) is higher than stillbirth (2.18) which means that heifers is more likely to have dystocia as compared with stillbirth.Although the studies in dystocia were more as compared with stillbirth, results show that bias associated with estimation of OR was present in dystocia only.The current study confirmed that primiparous cows were most likely to have dystocia and stillbirth as compared with multiparous cows.Results also indicate that differentiation can be made among primiparous and multiparous cows in the risk of having dystocia and stillbirth.These differences among cows could be useful to aid the better management to minimize their harmful effects in the dairy herds; particularly both traits have a low heritability (Lin et al., 1989).
Figure 1: A funnel plot for detection bias of OR in dystocia Figure 2: Illustrate the heterogeneity as a Figure4: Studies represented by a circle

Figure 5 :
Figure 5: Forest plots of dystocia studies

Figure 7 :
Figure 6: A funnel plot of studies effect size against their standard error for stillbirth

Table 4 :
Odds ratio of dystocia for different studies Weight

Table 5 :
Odds ratio of stillbirth for different studies