- Email: [email protected]
Effects of body condition score and score change on the reproductive performance of dairy cows: a meta-analysis
Theriogenology 59 (2003) 801±812
Effects of body condition score and score change on the reproductive performance of dairy cows: a meta-analysis F. LoÂpez-Gatiusa,*, J. YaÂnizb, D. Madriles-Helma a
Department of Animal Production, University of Lleida, Escuela TeÂcnica Superior de IngenierõÂa Agraria, Avda. Alcalde Rovira Roure 177, 25198 Lleida, Spain b Department of Animal Production, University of Zaragoza, Zaragoza, Spain Received 28 January 2002; accepted 15 April 2002
Abstract A meta-analysis was conducted to evaluate the effects of body condition score (BCS) at parturition and at ®rst AI, and of body condition change during the early lactation period on the variables pregnancy rate at ®rst AI and number of days open in dairy cattle. Inclusion criteria for the publications were: comparison of at least two groups of animals of different categories of BCS or change (independent variables) and consideration of pregnancy rate at ®rst AI or number of days open (dependent variables). Fifteen papers were selected and 23 excluded. Data corresponding to 7733 cows from 11 studies described in 10 papers were included in the analysis of pregnancy rate at ®rst AI, and those derived from 4529 cows from 11 studies described in 10 papers were used to analyze effects on the number of days open. Analyses were strati®ed according to study design and milk production characteristics. We de®ned low, intermediate and high categories of BCS for values lower than 2.5, from 2.5 to 3.5, or higher than 3.5, respectively. The categories of body condition change were increase (gain in score), slight loss (0±0.5 point loss), moderate loss (0.6±1 point loss), and severe loss (over 1 unit loss). Intermediate body condition at parturition and at ®rst AI, and a slight loss during the early lactation period were used as reference categories. The effects of body condition on the variable pregnancy rate at ®rst AI were highly heterogeneous, while all the studies considering the number of days open presented homogeneous results. A clear association between body condition category and pregnancy rate at ®rst AI was detected only when the effect of a low score at parturition was analyzed: pregnancy rate at ®rst AI signi®cantly decreased by about 10% in cows delivering in poor condition. Animals with a high BCS at parturition showed a signi®cant reduction in the number of days open of 5.8 or 11.7 when compared with animals with an intermediate or low body condition, respectively. In animals in the high body condition category at ®rst AI, the variable days open showed a signi®cant drop of 11.9 or 24.1, compared to animals in the intermediate or low categories, respectively. A severe loss in score during early lactation was related to a signi®cant (10.6) increase in * Corresponding author. Tel.: 34-973-702-500; fax: 34-973-238-264. E-mail address: [email protected] (F. LoÂpez-Gatius).
0093-691X/02/$ ± see front matter # 2002 Elsevier Science Inc. All rights reserved. PII: S 0 0 9 3 - 6 9 1 X ( 0 2 ) 0 1 1 5 6 - 1
802
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
the number of days open, while a slight or moderate body condition change, either loss or gain in score, was not signi®cantly related to the variable days open. We found that the number of days open was a good indicator of the effects of BCS or change in score on reproductive performance in dairy cattle. The variable pregnancy rate at ®rst AI, however, yielded heterogeneous results among studies. # 2002 Elsevier Science Inc. All rights reserved. Keywords: Dairy cows; Body condition score; Fertility; Meta-analysis
1. Introduction Body condition score (BCS) has proved useful as a management tool for assessing the nutritional status of dairy cows [1]. Cows are usually scored on a 5-point scale: from 1, indicating thin, to 5, indicating fat [2]. It is well known that cows enter a state of negative energy balance during early lactation [3]. As lactation progresses, the cow can regain its energy balance, but the consequences of the previous imbalance on reproductive functions are dif®cult to evaluate, since several variables besides nutritional factors may be involved. For instance, excessive loss of body condition during early lactation has been related to metabolic and infectious diseases [4,5]. Discrepancies regarding the effects of BCS and body condition change on the reproductive performance of dairy cows are, nevertheless, common in the literature. Body condition at parturition has been described as a risk factor [6,7], or as having no effect [5,8] on reproductive performance. There is also disagreement concerning the effect of body condition change on reproductive ef®ciency. Several authors have noted a linear relationship between postpartum body condition loss and fertility [9,10], but con¯icting results have also been reported [5,6]. Meta-analytical methods use statistical procedures to quantify research evidence and to investigate the heterogeneity of results among different studies testing the same hypothesis [11]. The objective of this study was to evaluate, through meta-analysis, the effects of BCS at parturition and at ®rst AI, and of body condition change during the early lactation period, on pregnancy rate at ®rst AI, and on the number of days open (the time interval between parturition and conception). 2. Materials and methods 2.1. Literature search and selection of papers All English-language papers published in peer-reviewed journals during the period 1982 to June 2001 that analyzed the effects of BCS or body condition change on conception rate at ®rst AI and number of days open in dairy cows, were included in the study. Papers were identi®ed through the CAB computer database (CAB Abstracts, Cab International, Oxon, UK), and the systematic review of citations in all the papers retrieved. Search terms were combined descriptors of BCS (body condition, nutrition, nutritional interactions, energy status, energy balance) and reproduction terms (parturition, postpartum, fertility, conception, pregnancy, days open). The analysis focused on lactating dairy or dual-purpose cows and excluded heifers before their ®rst parturition. Papers including the use of treatments
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
803
that could interfere with the validity and reliability of the results (e.g. BST or experimental nutritional additives) were also discarded. A paper was excluded from the meta-analysis if: the relationship between body condition and reproductive performance was not quanti®ed [1±3,6,12,13]; results could not be used to calculate effect estimates, i.e. studies failing to give related statistics as standard errors or con®dence intervals [14±16], or those in which results were only expressed as medians [17]; reproductive performance was measured by reproductive variables other than days open or conception rate at ®rst service [8,10,18]; categories of BCS or change were unknown [19]; or were referred to as dependent variables with reproductive variables considered as independent variables [20]; review papers provided results referring to the same original data only [21±25]; or animal groups showing similar BCS or change were compared [26±28]. On these grounds, 23 reports were excluded. Of the remaining reports, only papers that compared at least two groups of animals of different BCS category or change (independent variables) and quoted the variables conception rate at ®rst AI or the number of days open (dependent variables) were included in the meta-analysis. Fifteen papers were ®nally selected [4,5,7,9,29±39]. 2.2. Data organization and abstracting Features regarding study design, effect estimates, con®dence intervals and standard errors of the estimates, and signi®cance levels were recorded by the same abstractor. Study characteristics were analyzed to identify factors that could in¯uence the value of the effect estimate provided by the individual authors, as possible modi®ers of the summary effect calculated by the meta-analysis. Sources of heterogeneity among articles were also analyzed. Papers were classi®ed according to the BCS scale used (Table 1), the study design (experimental or observational, and number of animals, Table 2), and production characteristics (milk production and lactation number, Table 3). The different BCS scales (1±4, 1±5, and 0±5 points) were combined according to proportions to give one reference scale (1±5 points) [2]. We de®ned the categories low, intermediate, or high BCS for values lower than 2.5, from 2.5 to 3.5, or higher than 3.5, respectively. Categories of BCS change were increase (a gain in points), slight loss (loss of 0±0.5 points), moderate loss (loss of 0.6±1 points), and severe loss (greater than 1 unit loss). Medium BCS at parturition and at ®rst AI, and slight body condition loss during the early lactation period were used as reference categories. In several studies in which these categories differed slightly from the established de®nition [4,35,37], animal traits were proportionally classi®ed into the corresponding categories. Animals showing traits of equal
Table 1 Classification of studies according to the body condition scale used Body condition score
Studies (reference numbers)
1±4 (3 units) 1±5 (4 units) 0±5 (5 units)
[29,30] [4,5,7,9,31±37] [38,39]
804
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
Table 2 Classification of studies according to the study design Study design
Studies (reference numbers)
Type of study Experimental Observational
[29,30,34,35,39] [4,5,7,9,31±33,36±38]
Number of animals <100 100±1000 >1000
[29,30,35,37,39] [4,5,7aa,7bb,9,32,33,36,38] [7cc,7dd,31,34]
(a, b, c, d) Trials 1, 3, 2 and 4, respectively [7].
Table 3 Classification of studies according to production characteristics Production characteristics
Studies (reference numbers)
Milk production (305 days) <7000 kg 7000±10000 kg >10000 kg
[33,39] [4,5,29±32,34,35,38] [7,9,36,37]
Lactation number 1st lactation 2nd lactation All lactations
[7aa,7bb,39] [7cc,7dd,9,29,30,37] [4,5,31±36,38]
(a, b, c, d) Trials 1, 3, 2 and 4, respectively [7].
body condition category were matched and treated as a single group, and their pooled reproductive performance was calculated [5,30,38]. 2.3. Expression of estimates for the meta-analysis For the categorical dependent variable pregnancy rate at ®rst AI, the relative risk of conception between two groups of animals of different category of BCS or change was either retrieved from the paper as an odds ratio [4,7,9,31±33], or was calculated by dividing least square means [36,39] or crude rates [35] of pregnancy (from a trait of a BCS or change category) and its reference group. In one study [34], the relative risk was derived from the linear regression coef®cient. For all odds ratios, the variance of the relative risk was calculated either by exponentiating the standard error of the logistic regression coef®cient [7,32±34], or from the upper or lower limit of the 95% con®dence interval [9,31]. If the latter were not consistent, the clearest variance of the relative risk was chosen among the possible alternative calculations. In four studies [4,35,36,39], the variance of the relative risk was obtained by 2 2 contingency tables as described by Petitti [11]. For the continuous dependent variable number of days open, the effect size, de®ned as the difference in days open between two animal groups of different category of BCS or change, was either obtained from the paper as a linear regression coef®cient [7,33,34], or
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
805
was calculated as the difference among least square means of groups belonging to a given BCS or change category [5,29,30,35±38]. The variance of the effect number of days open was calculated as described by Petitti [11], or from the standard error of the linear regression coef®cient. 3. Statistical methods 3.1. Summary estimates We adapted the general variance-based methods proposed by Petitti [11] to obtain the summary estimates and their 95% con®dence intervals (see formulae). The effect estimates, de®ned as the difference in reproductive performance between two groups of different BCS or change category, were weighted through the inverse term of their variance. The meta-analytic summary estimate (effect number of days open or relative risk of conception at ®rst AI) was inversely compounded to the size of its deviation. The variance and 95% con®dence interval of the summary estimate was then calculated. The null estimate or non-signi®cance of effect was 0 for the summary effect size of the number of days open, and 1 in the case of the summary relative risk of conception at ®rst AI. A lower value meant reproductive disadvantage and a higher value a relative improvement, for animals with different BCS or change with respect to the reference categories. 3.2. Testing the homogeneity of results and evaluating sources of heterogeneity The homogeneity of the results of studies combined in a partial meta-analysis was tested by calculating the Q-statistic equivalent to a chi-square distribution (see formulae) in order to examine the quality of the summary estimates of effects. The null-hypothesis of homogeneity of results was rejected when the value of Q exceeded the bounds of the chi-distribution with the number of studies minus one degree of freedom. 3.3. Formulae used in the meta-analysis: general variance-based methods For the categorical dependent variable pregnancy rate at ®rst AI, the formulae applied were: s P 1 1 i ln RRi i w P ln RRS ln CI ln RRS 1:96 P ; ; wi var i i wi i wi X fwi ln RRS ln RRi g Q i
where, RRS is the summary estimate of the relative risk of conception at ®rst AI between cows of a BCS or BC-loss category and the reference or control group, RRi is the relative risk of conception in the ith study, wi is the weight assigned to the ith study, var i is the variance of the relative risk in the ith study, CI is the 95% con®dence interval of the summary estimate, Q: refers to a chi-square distribution with the number of estimates
806
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
minus 1 degrees of freedom to test for homogeneity between the results of different studies. For the continuous dependent variable number of days open, the formulae applied were: s P X wi Di 1 1 i ; CI DS 1:96 P ; Q DS P ; wi fwi DS Di g var i i wi I wi i where, DS is the summary estimate of the difference in days open between cows of a BCS or change category and the reference or control group, Di is the effect size of the difference in the ith study, wi is the weight assigned to the ith study, var i is the variance of the effect size in the ith study, CI is the 95% con®dence interval of the summary estimate, Q: refers to a chi-square distribution with the number of estimates minus 1 degrees of freedom to test for homogeneity between the results of different studies. 4. Results 4.1. Effects of BCS and change on pregnancy rate at ®rst AI Table 4 shows the effects of BCS at parturition and at ®rst AI, and of body condition change during the early lactating period on pregnancy rate at ®rst AI. Ten studies performed on a total of 7733 cows were included in the analysis: three experimental [34,35,39], and seven observational [4,7,9,31±33,36]. Two trials from study 7 were independently analyzed: 7a (652 primiparous cows), and 7c (1040 multiparous cows). BCS at parturition was associated with the relative risk of conception only in cows showing a low BCS at parturition. Animals in poor condition showed a signi®cant (9%) reduction in pregnancy rate (relative risk: 0.91; CI: 0.84, 0.99) at ®rst AI compared to Table 4 Effect of BCS and score change on pregnancy rate at first AI Variable
Relative risk of conception
95% CI
Pooled number of cows
Reference numbers
Inter-study homogeneity
BCS at parturition (reference category: 2.5±3.5) <2.5 0.91 0.84, 0.99 >3.5 1.04 0.95, 1.13
3644 3723
[4,7aa,7cb,34] [4,7aa,7cb,34,35]
Yesc Nod
BCS at first AI (reference category: 2.5±3.5) <2.5 0.91 0.85, 0.98 >3.5 1.05 0.97, 1.13
2122 1786
[32,34,36] [32,34]
Noe Yesf
Body condition change during early lactation (reference category: <0.5 point loss) 0.5±1 point loss 0.96 0.91, 1.01 3975 [9,31,33,34,35,39] >1 point loss 0.9 0.82, 0.99 3837 [9,31,33±35] Increase 1.03 0.98, 1.09 3474 [9,31,34,39]
Yesg Noh Noi
(a, b) Trials 1 and 2, respectively [7]. (c) Q 0:8, 3 d.f., w2 7:8. (d) Q 30:8, 4 d.f., w2 9:5. (e) Q 10:9, 2 d.f., w2 6. (f) Q 2:8, 1 d.f., w2 3:8. (g) Q 6:4, 5 d.f., w2 11:1. (h) Q 17:1, 4 d.f., w2 9:5. (i) Q 9, 3 d.f., w2 7:8.
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
807
animals in intermediate condition. The summary estimate was based on one experimental [34] and three observational [4,7a,7c] studies, involving 3644 medium or high producing cows in different lactation periods. Studies 7c and 34 were performed on over 1000 animals. All four papers reported homogeneous effects. A high BCS at parturition was not associated with pregnancy rate at ®rst AI. Results from two experimental [34,35] and three observational [4,7a,7c] studies performed on a variable number of cows of different lactation number and milk production were heterogeneous. BCS at ®rst AI was not associated with the relative risk of conception. The ®ndings of one experimental [34] and two observational [32,36] studies evaluating the effect of a low BCS at AI on pregnancy rate were heterogeneous. Based on results from one experimental [34] and one observational [32] trial, the effect of a high BCS at AI on pregnancy rate was not signi®cant (relative risk: 1.05; CI: 0.97, 1.13). Body condition change during the early lactation period was not associated with the relative risk of conception. We analyzed the results of three experimental [34,35,39] and three observational [9,31,33] studies derived from large study populations (3474±3975 animals). The studies included a variable number of cases, lactations, and milk production ®gures. The summary estimate of the effect of body condition change on pregnancy rate at ®rst AI was either non-signi®cant (for a moderate body condition loss) or heterogeneous (for a gain or a severe loss in body condition). 4.2. Effects of BCS and change on days open Table 5 shows the effects of BCS at parturition and at ®rst AI, and of body condition change during the early lactation period on days open. Ten studies performed on a total number of 4529 cows were included in the analysis: four experimental [29,30,34,35] and six observational [5,7,33,36±38]. Two trials from study 7 were independently analyzed: 7b (683 primiparous cows), and 7d (1168 multiparous cows). Table 5 Effect of BCS and change on number of days open Variable
Summary effect size (days)
95% CI
Pooled number of cows
Reference numbers
BCS at parturition (reference category: 2.5±3.5) <2.5 5.9 1.4, 10.4 >3.5 5.8 9.8, 1.9
3078 3755
[7ba,7db,29,34] Yesc [5,7ba,7db,29,30,34,35,37,38] Yesd
BCS at first AI (reference category: 2.5±3.5) <2.5 12.2 4.8, 19.7 >3.5 11.9 19.5, 4.3
1418 1211
[34,36] [34]
Yese (Single study)
[5,30,33,34,37] [5,30,33,34,37] [5,33,34]
Yesf Yesg Yesh
Body condition change (reference category: <0.5 point loss) 0.5±1 point loss 3.5 2.5, 9.6 2012 >1 point loss 10.6 3.9, 17.3 1863 Increase 3.7 10.7, 3.3 1823
Inter-study homogeneity
(a, b) Trials 3 and 4, respectively [7]. (c) Q 4:1, 3 d.f., w2 7:8. (d) Q 12, 8 d.f., w2 15:5. (e) Q 0:1, 1 d.f., w2 3:8. (f) Q 1:5, 4 d.f., w2 9:5. (g) Q 4:4, 4 d.f., w2 9:5. (h) Q 1:7, 2 d.f., w2 6.
808
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
Cows with a low BCS at parturition remained open for approximately 6 further days, compared to cows in the intermediate body condition category. The effect size was signi®cant (CI: 1.4, 10.4) and included two experimental studies on medium milk producing cows [29,34], and two observational studies involving high milk producers [7b,7d]. The study population for studies 7d and 29 was greater than 1000 animals. The results of the four reports were homogeneous. Cows showing a high BCS at parturition were open approximately 6 days less than cows with a medium body condition. The size of the effect size was signi®cant (CI: 9.8, 1.9) and was based on the outcomes of experimental [29,30,34,35] and observational [5,7c,7d,37,38] studies. Studies were homogeneous results and included a variable number of medium or high milk producing cows. Only two studies [34,36] evaluated the effect of a low BCS at the time of ®rst AI on days open. The summary effect size indicated that cows in poor condition at ®rst AI needed 12.2 further days to become pregnant (CI: 4.8, 19.7), compared to cows in the medium body condition category. This analysis of one experimental [34] and one observational [36] study was homogeneous and included 1418 medium or high milk producers. A single experimental study [34] performed on 1211 animals related a high BCS at ®rst AI with a 12-day decrease in the number of days open (CI: 19.5, 4.3). A moderate loss of body condition (0.5±1 points) during the early lactation period was not signi®cantly associated with an increased number of days open (summary effect size: 3.5 days; CI: 2.5, 9.6). The papers included were homogeneous and involved experimental [30,34] and observational [5,33,37] studies performed on different population sizes and on animals showing different lactation numbers and milk production. Results from these same studies showed a signi®cant increase of 10.6 days open (CI: 3.6, 17.3) in cows suffering a severe body condition loss (greater than 1 point) during the early lactation period, relative to animals undergoing a slight loss in body condition. Effects were also homogeneous. No signi®cant effect could be established for cows experiencing a body condition increase (gain in points) during early lactation. The homogeneous summary estimate (effect size: 3.7 days; CI: 10.7, 3.3) included one experimental [34] and two observational [5,33] studies performed on cows showing low or medium milk yields. 5. Discussion 5.1. Analysis of heterogeneity We selected the variables pregnancy rate at ®rst AI and number of days open for our analysis since they are the factors most commonly used to evaluate possible biological effects of BCS, or score change, on reproductive performance. The results of the effects of body condition category on pregnancy rate at ®rst AI were highly heterogeneous. Heterogeneity was detected in at least one subset of all the sections considered: BCS at parturition, at ®rst AI, and body condition change during the early lactation period. In contrast, when the variable number of days open was considered, all the studies presented homogeneous results.
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
809
Statistical heterogeneity among studies occurs when the original results combined in a meta-analysis are suf®ciently different to reject the hypothesis of homogeneity [11]. Investigating possible sources of heterogeneity identi®es factors that can lead to heterogeneity. The experimental studies included in the meta-analysis all involved low numbers of animals (lower than 100, except for study 34) of low or medium milk production, whereas the observational trials were usually performed on a large number of medium and high milk producers. Thus, the main source of heterogeneity seems to be the type of study. Nevertheless, the effects of body condition status or change on the variable number of days open were highly homogeneous among studies. In this case, the type of study did not contribute to heterogeneity. It is likely that the variable pregnancy rate at ®rst AI is itself the persistent source of heterogeneity in the meta-analysis. In observational studies, this variable can be strongly affected by farm policy regarding the voluntary waiting period or by failure to detect estrus. BCS is dynamic and is strictly related to the physiological cycle of cows; it decreases in early lactation, is restored during mid-lactation, and reaches a near steady state during late lactation [40]. Thus, a longer parturition to AI interval should improve postpartum metabolic recovery. For instance, better results would be expected when a herd is allowed 90 days to ®rst AI than 60 days. However, in both cases cows could have similar body conditions. Additional factors that may in¯uence the pregnancy rate should be taken into account in the design of future studies aimed at evaluating the effects of body condition on pregnancy rate at ®rst AI. In contrast, the homogeneity of results from studies based on the variable number of days open is consistent with the idea that the average number of days open, better described as the parturition to conception interval, is the best predictor of overall reproductive performance in a herd. The voluntary waiting period, estrus detection accuracy and pregnancy rate at ®rst service are only components of this indicator [41]. 5.2. Effects of BCS and change on pregnancy rate at ®rst AI A clear association between BCS and pregnancy rate at ®rst AI was found only for the effect of a low BCS at parturition. The results of all the studies were homogeneous; pregnancy rate at ®rst AI showed a signi®cant drop of about 10% in cows delivering in poor condition. This reduced fertility could be a consequence of prolonged anovulatory intervals, which is frequent in thin cows and has a negative impact on ®rst service conception rates [42]. 5.3. Effects of BCS and change on number of days open BCS at parturition and at ®rst AI were good indicators of the relationship between the nutritional status of the cow and the number of days open. In animals showing a good body condition (score above 3.5) at parturition the number of days open was signi®cantly reduced (5.8 or 11.7, respectively) compared to animals in the intermediate (score 2.5±3.5) or low (score under 2.5) body condition category. At ®rst AI, animals in good condition were related to a signi®cant decrease in days open when compared with animals in the intermediate or low body condition category (11.9 or 24.1, respectively). As expected, a closer relationship (an effect double the size) between BCS and days open was noted at ®rst
810
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
AI than at parturition. At ®rst AI, most cows are recovering from metabolic postpartum stress and their energy balance will not become positive until next parturition. On the contrary, BCS at parturition is a worse indicator. Depending on the effects of the period of negative energy balance during the early lactation period, BCS may undergo a further drop. Mobilization of body fat stores following parturition may be limited in thin cows and considerable in fat animals. Some thin cows regain their BCS before the time of ®rst AI, while over-conditioned cows may reach the time of AI in poor condition. This rationale was supported by the analysis of effects of body condition change on the number of days open. A severe loss of body condition (a drop in score of over 1 unit) during the early lactation stage was related to a signi®cant 10.6 increase in the number of days open. Ovarian follicular dynamics in the postpartum dairy cow appear to be unaffected by a negative energy balance [42]. Our ®ndings were consistent with this; a moderate body condition change, either loss or gain, was not signi®cantly related to the number of days open. However, as noted above, a severe postpartum body condition loss resulted in an increased number of days open. In this case, it is possible that postpartum ovarian activity is affected. Our results show that the variable number of days open is a good predictor of the effects BCS or change will have on reproductive performance. Better results were obtained in cows in good body condition (score higher than 3.5) at parturition and ®rst AI, and in those showing a slight or moderate loss in postpartum body condition (less than 1 unit). Given that nutritional status is associated with management strategy, special attention should be paid to the cows' nutritional needs during the periparturition period, between late pregnancy and early lactation (also called the transition period [43]). Cow management during this period is acquiring special relevance in the prevention of metabolic disorders [43,44]. Overfeeding during the dry period has been related to a predisposition to accumulate fat in adipose tissue, increased lipolysis postpartum and a lower ability of the adipose tissue to reesterify mobilized fatty acids [45]. Excessive lipid mobilization from adipose tissue has been linked to a greater incidence of periparturient health disorders [43]. We could add that this is a major factor affecting the reproductive performance of cows showing good body condition at parturition. Studies aimed at evaluating the effects of BCS or score change during the dry period on the subsequent number of days open would be of great interest. Acknowledgements The authors thank Ana Burton for assistance with the English translation. References [1] Hady PJ, Domecq JJ, Kaneene JB. Frequency and precision of body condition scoring in dairy cattle. J Dairy Sci 1994;77:1543±7. [2] Edmonson AJ, Lean IJ, Weaver LD, Farver T, Webster G. A body condition scoring chart for Holstein dairy cows. J Dairy Sci 1989;72:68±78. [3] Butler WR. Nutritional interactions with reproductive performance in dairy cattle. Anim Reprod Sci 2000;60/61:449±57.
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
811
[4] Heuer C, Schukken YH, Dobbelaar P. Postpartum body condition score and results from the ®rst test day milk as predictors of disease, fertility, yield, and culling in commercial dairy herds. J Dairy Sci 1999;82:295±304. [5] Ruegg PL, Milton RL. Body condition scores of Holstein cows on Prince Edward Island, Canada: relationships with yield, reproductive performance, and disease. J Dairy Sci 1995;78:552±64. [6] Gearhart MA, Curtis CR, Erb HN, Smith RD, Sniffen CJ, Chase LE, et al. Relationship of changes in condition score to cow health in Holsteins. J Dairy Sci 1990;73:3132±40. [7] Markusfeld O, Galon N, Ezra E. Body condition score, health, yield and fertility in dairy cows. Vet Rec 1997;141:67±72. [8] Waltner SS, McNamara JP, Hillers JK. Relationships of body condition score to production variables in high producing Holstein dairy cattle. J Dairy Sci 1993;76:3410±9. [9] Domecq JJ, Skidmore AL, Lloyd JW, Kaneene JB. Relationship between body condition scores and conception at ®rst arti®cial insemination in a large dairy herd of high yielding Holstein cows. J Dairy Sci 1997;80:113±20. [10] Suriyasathaporn W, Nielen M, Dieleman SJ, Brand A, Noordhuizen-Strasse EN, Schukken YH. A cox proportional-hazards model with time-dependent covariates to evaluate the relationship between bodycondition score and the risks of ®rst insemination and pregnancy in a high producing dairy herd. Prev Vet Med 1998;37:159±72. [11] Petitti DB. Meta-analysis, decision analysis, and cost-effectiveness analysis. Methods for quantitative synthesis in medicine. Oxford: Oxford University Press; 2000. [12] Domecq JJ, Skidmore AL, Lloyd JW, Kaneene JB. Relationship between body condition scores and milk yield in a large dairy herd of high yielding Holstein cows. J Dairy Sci 1997;80:101±12. [13] Prandi A, Messina M, Tondolo A, Motta M. Correlation between reproductive ef®ciency, as determined by new mathematical indexes, and the body condition score in dairy cows. Theriogenology 1999;52: 1251±65. [14] Garnsworthy PC, Jones GP. The in¯uence of body condition at calving and dietary protein supply on voluntary food intake and performance in dairy cows. Anim Prod 1987;44:347±53. [15] Gearhart MA, Curtis CR, Erb HN, Smith RD, Sniffen CJ, Chase LE, et al. Relationship of body condition score and changes in condition score with health, reproductive performance and productivity in New York Holstein dairy herds. Acta Vet Scand 1988;84(Suppl):122±5. [16] Holter JB, Stolnick MJ, Hayes HH, Bozak CK, Urban WE, McGilliard ML. Effect of prepartum dietary energy on condition score, postpartum energy, nitrogen partitions, and lactation production responses. J Dairy Sci 1990;73:3502±11. [17] Treacher RJ, Reid IM, Roberts CJ. Effect of body condition at calving on the health and performance of dairy cows. Anim Prod 1986;43:1±6. [18] Senatore EM, Butler WR, Oltenacu PA. Relationships between energy balance and postpartum ovarian activity and fertility in ®rst lactation dairy cows. Anim Sci 1996;62:17±23. [19] Ducker MJ, Morant SV. Observations on the relationships between the nutrition, milk yield, live weight and reproductive performance of dairy cows. Anim Prod 1984;38:9±14. [20] Wildman EE, Jones GM, Wagner PE, Boman RL, Troutt Jr HF, Lesch TN. A dairy cow body condition scoring system and its relationship to selected production characteristics. J Dairy Sci 1982;65:495±501. [21] Broster WH, Broster VJ. Body score of dairy cows. J Dairy Res 1998;65:155±73. [22] Burkholder WJ. Use of body condition scores in clinical assessment of the provision of optimal nutrition. J Am Vet Med Assoc 2000;217:650±4. [23] Ducker MJ, Morant SV, Fischer WJ, Haggett RA. Nutrition and reproductive performance of dairy cattle. Part 2. Prediction of reproductive performance in ®rst lactation dairy heifers subjected to controlled nutritional regimes. Anim Prod 1985;41:13±22. [24] Nebel RL, McGilliard ML. Interactions of high milk yield and reproductive performance in dairy cows. J Dairy Sci 1993;76:3257±68. [25] Ruegg PL. Body condition scoring in dairy cows: relationships with production, reproduction, nutrition, and health. Comp Cont Educ Pract Vet 1991;13:1309±13. [26] Aeberhard K, Bruckmaier RM, Kuepfer U, Blum JW. Milk yield and composition, nutrition, body conformation traits, body condition scores, fertility and diseases in high yielding dairy cows: Part 1. J Vet Med Ser A 2001;48:99±110.
812
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
[27] Reist M, Koller A, Busato A, Kuepfer U, Blum JW. First ovulation and ketone body status in the early postpartum period of dairy cows. Theriogenology 2000;54:685±701. [28] Snijders SEM, Dillon PG, O'Farrel KJ, Diskin M, Wylie ARG, O'Callaghan D, et al. Genetic merit for milk production and reproductive success in dairy cows. Anim Reprod Sci 2001;65:17±31. [29] Garnsworthy PC, Topps JH. The effect of body condition of dairy cows at calving on their food intake and performance when given complete diets. Anim Prod 1982;35:113±9. [30] Jones GP, Garnsworthy PC. The effects of body condition at calving and dietary protein content on drymatter intake and performance in lactating dairy cows given diets of low energy content. Anim Prod 1988;47:321±33. [31] Loef¯er SH, Vries MJ, Schukken YH. Effects of time of disease occurrence, milk yield, and body condition on fertility of dairy cows. J Dairy Sci 1999;82:2589±604. [32] Loef¯er SH, Vries MJ, Schukken YH, de Zeeuw AC, Dijkhuizen AA, de Graaf FM, et al. Use of AI technician scores for body condition, uterine tone and uterine discharge in a model with disease and milk production parameters to predict pregnancy risk at ®rst AI in Holstein dairy cows. Theriogenology 1999;51:1267±84. [33] Gillund P, Reksen O, GroÈhn YT, Karlberg K. Body condition related to ketosis and reproductive performance in Norwegian dairy cows. J Dairy Sci 2001;84:1390±6. [34] Pryce JE, Coffey MP, Simm G. The relationship between body condition score and reproductive performance. J Dairy Sci 2001;84:1508±15. [35] Butler WR, Smith RD. Interrelationships between energy balance and postpartum reproductive function in dairy cattle. J Dairy Sci 1989;72:767±83. [36] Moreira F, Risco C, Pires MFA, Ambrose JD, Drost M, DeLorenzo M, et al. Effect of body condition on reproductive ef®ciency of lactating dairy cows receiving a timed insemination. Theriogenology 2000;53:1305±19. [37] Ruegg PL, Goodger WJ, Holmberg CA, Weaver LD, Huffman EM. Relation among body condition score, serum urea nitrogen and cholesterol concentrations, and reproductive performance in high producing Holstein dairy cows in early lactation. Am J Vet Res 1992;53:10±4. [38] Pedron O, Cheli F, Senatore E, Baroli D, Rizzi R. Effect of body condition score at calving on performance, some blood parameters, and milk fatty acid composition in dairy cows. J Dairy Sci 1993;76:2528±35. [39] Ducker MJ, Hagget RA, Fisher WJ, Morant SV, Bloom®eld GA. Nutrition and reproductive performance of dairy cattle. Part 1. The effect of level of feeding in late pregnancy and around the time of insemination on the reproductive performance of ®rst lactation dairy heifers. Anim Prod 1985;41:1±12. [40] Gallo L, Carnier P, Cassandro M, Mantovani R, Bailoni L, Contiero B, et al. Change in body condition score of Holstein cows as affected by parity and mature equivalent milk. J Dairy Sci 1996;79:1009±15. [41] Fetrow J, Stewart S, Eicker S. Reproductive health programs for dairy herds: analysis of records for assessment of reproductive performance. In: Youngquist RS, editor. Current therapy in large animal theriogenology. Philadelphia: Saunders; 1997, p. 441±51. [42] Beam SW, Butler WR. Effects of energy balance on follicular development and ®rst ovulation in postpartum dairy cows. J Reprod Fertil 1999;54(Suppl):411±24. [43] Drackley JK. Biology of dairy cows during the transition period: the ®nal frontier? J Dairy Sci 1999;82:2259±73. [44] Gerloff BJ. Dry cow management for the prevention of ketosis and fatty liver in dairy cows. Vet Clin North Am Food Anim Pract 2000;16:283±92. [45] Rukkwamsuk, Wensing T, Geelen MJH. Effect of overfeeding during the dry period on the rate of esteri®cation in adipose tissue of dairy cows during the periparturient period. J Dairy Sci 1999;82:1164±9.
Effects of body condition score and score change on the reproductive performance of dairy cows: a meta-analysis F. LoÂpez-Gatiusa,*, J. YaÂnizb, D. Madriles-Helma a
Department of Animal Production, University of Lleida, Escuela TeÂcnica Superior de IngenierõÂa Agraria, Avda. Alcalde Rovira Roure 177, 25198 Lleida, Spain b Department of Animal Production, University of Zaragoza, Zaragoza, Spain Received 28 January 2002; accepted 15 April 2002
Abstract A meta-analysis was conducted to evaluate the effects of body condition score (BCS) at parturition and at ®rst AI, and of body condition change during the early lactation period on the variables pregnancy rate at ®rst AI and number of days open in dairy cattle. Inclusion criteria for the publications were: comparison of at least two groups of animals of different categories of BCS or change (independent variables) and consideration of pregnancy rate at ®rst AI or number of days open (dependent variables). Fifteen papers were selected and 23 excluded. Data corresponding to 7733 cows from 11 studies described in 10 papers were included in the analysis of pregnancy rate at ®rst AI, and those derived from 4529 cows from 11 studies described in 10 papers were used to analyze effects on the number of days open. Analyses were strati®ed according to study design and milk production characteristics. We de®ned low, intermediate and high categories of BCS for values lower than 2.5, from 2.5 to 3.5, or higher than 3.5, respectively. The categories of body condition change were increase (gain in score), slight loss (0±0.5 point loss), moderate loss (0.6±1 point loss), and severe loss (over 1 unit loss). Intermediate body condition at parturition and at ®rst AI, and a slight loss during the early lactation period were used as reference categories. The effects of body condition on the variable pregnancy rate at ®rst AI were highly heterogeneous, while all the studies considering the number of days open presented homogeneous results. A clear association between body condition category and pregnancy rate at ®rst AI was detected only when the effect of a low score at parturition was analyzed: pregnancy rate at ®rst AI signi®cantly decreased by about 10% in cows delivering in poor condition. Animals with a high BCS at parturition showed a signi®cant reduction in the number of days open of 5.8 or 11.7 when compared with animals with an intermediate or low body condition, respectively. In animals in the high body condition category at ®rst AI, the variable days open showed a signi®cant drop of 11.9 or 24.1, compared to animals in the intermediate or low categories, respectively. A severe loss in score during early lactation was related to a signi®cant (10.6) increase in * Corresponding author. Tel.: 34-973-702-500; fax: 34-973-238-264. E-mail address: [email protected] (F. LoÂpez-Gatius).
0093-691X/02/$ ± see front matter # 2002 Elsevier Science Inc. All rights reserved. PII: S 0 0 9 3 - 6 9 1 X ( 0 2 ) 0 1 1 5 6 - 1
802
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
the number of days open, while a slight or moderate body condition change, either loss or gain in score, was not signi®cantly related to the variable days open. We found that the number of days open was a good indicator of the effects of BCS or change in score on reproductive performance in dairy cattle. The variable pregnancy rate at ®rst AI, however, yielded heterogeneous results among studies. # 2002 Elsevier Science Inc. All rights reserved. Keywords: Dairy cows; Body condition score; Fertility; Meta-analysis
1. Introduction Body condition score (BCS) has proved useful as a management tool for assessing the nutritional status of dairy cows [1]. Cows are usually scored on a 5-point scale: from 1, indicating thin, to 5, indicating fat [2]. It is well known that cows enter a state of negative energy balance during early lactation [3]. As lactation progresses, the cow can regain its energy balance, but the consequences of the previous imbalance on reproductive functions are dif®cult to evaluate, since several variables besides nutritional factors may be involved. For instance, excessive loss of body condition during early lactation has been related to metabolic and infectious diseases [4,5]. Discrepancies regarding the effects of BCS and body condition change on the reproductive performance of dairy cows are, nevertheless, common in the literature. Body condition at parturition has been described as a risk factor [6,7], or as having no effect [5,8] on reproductive performance. There is also disagreement concerning the effect of body condition change on reproductive ef®ciency. Several authors have noted a linear relationship between postpartum body condition loss and fertility [9,10], but con¯icting results have also been reported [5,6]. Meta-analytical methods use statistical procedures to quantify research evidence and to investigate the heterogeneity of results among different studies testing the same hypothesis [11]. The objective of this study was to evaluate, through meta-analysis, the effects of BCS at parturition and at ®rst AI, and of body condition change during the early lactation period, on pregnancy rate at ®rst AI, and on the number of days open (the time interval between parturition and conception). 2. Materials and methods 2.1. Literature search and selection of papers All English-language papers published in peer-reviewed journals during the period 1982 to June 2001 that analyzed the effects of BCS or body condition change on conception rate at ®rst AI and number of days open in dairy cows, were included in the study. Papers were identi®ed through the CAB computer database (CAB Abstracts, Cab International, Oxon, UK), and the systematic review of citations in all the papers retrieved. Search terms were combined descriptors of BCS (body condition, nutrition, nutritional interactions, energy status, energy balance) and reproduction terms (parturition, postpartum, fertility, conception, pregnancy, days open). The analysis focused on lactating dairy or dual-purpose cows and excluded heifers before their ®rst parturition. Papers including the use of treatments
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
803
that could interfere with the validity and reliability of the results (e.g. BST or experimental nutritional additives) were also discarded. A paper was excluded from the meta-analysis if: the relationship between body condition and reproductive performance was not quanti®ed [1±3,6,12,13]; results could not be used to calculate effect estimates, i.e. studies failing to give related statistics as standard errors or con®dence intervals [14±16], or those in which results were only expressed as medians [17]; reproductive performance was measured by reproductive variables other than days open or conception rate at ®rst service [8,10,18]; categories of BCS or change were unknown [19]; or were referred to as dependent variables with reproductive variables considered as independent variables [20]; review papers provided results referring to the same original data only [21±25]; or animal groups showing similar BCS or change were compared [26±28]. On these grounds, 23 reports were excluded. Of the remaining reports, only papers that compared at least two groups of animals of different BCS category or change (independent variables) and quoted the variables conception rate at ®rst AI or the number of days open (dependent variables) were included in the meta-analysis. Fifteen papers were ®nally selected [4,5,7,9,29±39]. 2.2. Data organization and abstracting Features regarding study design, effect estimates, con®dence intervals and standard errors of the estimates, and signi®cance levels were recorded by the same abstractor. Study characteristics were analyzed to identify factors that could in¯uence the value of the effect estimate provided by the individual authors, as possible modi®ers of the summary effect calculated by the meta-analysis. Sources of heterogeneity among articles were also analyzed. Papers were classi®ed according to the BCS scale used (Table 1), the study design (experimental or observational, and number of animals, Table 2), and production characteristics (milk production and lactation number, Table 3). The different BCS scales (1±4, 1±5, and 0±5 points) were combined according to proportions to give one reference scale (1±5 points) [2]. We de®ned the categories low, intermediate, or high BCS for values lower than 2.5, from 2.5 to 3.5, or higher than 3.5, respectively. Categories of BCS change were increase (a gain in points), slight loss (loss of 0±0.5 points), moderate loss (loss of 0.6±1 points), and severe loss (greater than 1 unit loss). Medium BCS at parturition and at ®rst AI, and slight body condition loss during the early lactation period were used as reference categories. In several studies in which these categories differed slightly from the established de®nition [4,35,37], animal traits were proportionally classi®ed into the corresponding categories. Animals showing traits of equal
Table 1 Classification of studies according to the body condition scale used Body condition score
Studies (reference numbers)
1±4 (3 units) 1±5 (4 units) 0±5 (5 units)
[29,30] [4,5,7,9,31±37] [38,39]
804
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
Table 2 Classification of studies according to the study design Study design
Studies (reference numbers)
Type of study Experimental Observational
[29,30,34,35,39] [4,5,7,9,31±33,36±38]
Number of animals <100 100±1000 >1000
[29,30,35,37,39] [4,5,7aa,7bb,9,32,33,36,38] [7cc,7dd,31,34]
(a, b, c, d) Trials 1, 3, 2 and 4, respectively [7].
Table 3 Classification of studies according to production characteristics Production characteristics
Studies (reference numbers)
Milk production (305 days) <7000 kg 7000±10000 kg >10000 kg
[33,39] [4,5,29±32,34,35,38] [7,9,36,37]
Lactation number 1st lactation 2nd lactation All lactations
[7aa,7bb,39] [7cc,7dd,9,29,30,37] [4,5,31±36,38]
(a, b, c, d) Trials 1, 3, 2 and 4, respectively [7].
body condition category were matched and treated as a single group, and their pooled reproductive performance was calculated [5,30,38]. 2.3. Expression of estimates for the meta-analysis For the categorical dependent variable pregnancy rate at ®rst AI, the relative risk of conception between two groups of animals of different category of BCS or change was either retrieved from the paper as an odds ratio [4,7,9,31±33], or was calculated by dividing least square means [36,39] or crude rates [35] of pregnancy (from a trait of a BCS or change category) and its reference group. In one study [34], the relative risk was derived from the linear regression coef®cient. For all odds ratios, the variance of the relative risk was calculated either by exponentiating the standard error of the logistic regression coef®cient [7,32±34], or from the upper or lower limit of the 95% con®dence interval [9,31]. If the latter were not consistent, the clearest variance of the relative risk was chosen among the possible alternative calculations. In four studies [4,35,36,39], the variance of the relative risk was obtained by 2 2 contingency tables as described by Petitti [11]. For the continuous dependent variable number of days open, the effect size, de®ned as the difference in days open between two animal groups of different category of BCS or change, was either obtained from the paper as a linear regression coef®cient [7,33,34], or
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
805
was calculated as the difference among least square means of groups belonging to a given BCS or change category [5,29,30,35±38]. The variance of the effect number of days open was calculated as described by Petitti [11], or from the standard error of the linear regression coef®cient. 3. Statistical methods 3.1. Summary estimates We adapted the general variance-based methods proposed by Petitti [11] to obtain the summary estimates and their 95% con®dence intervals (see formulae). The effect estimates, de®ned as the difference in reproductive performance between two groups of different BCS or change category, were weighted through the inverse term of their variance. The meta-analytic summary estimate (effect number of days open or relative risk of conception at ®rst AI) was inversely compounded to the size of its deviation. The variance and 95% con®dence interval of the summary estimate was then calculated. The null estimate or non-signi®cance of effect was 0 for the summary effect size of the number of days open, and 1 in the case of the summary relative risk of conception at ®rst AI. A lower value meant reproductive disadvantage and a higher value a relative improvement, for animals with different BCS or change with respect to the reference categories. 3.2. Testing the homogeneity of results and evaluating sources of heterogeneity The homogeneity of the results of studies combined in a partial meta-analysis was tested by calculating the Q-statistic equivalent to a chi-square distribution (see formulae) in order to examine the quality of the summary estimates of effects. The null-hypothesis of homogeneity of results was rejected when the value of Q exceeded the bounds of the chi-distribution with the number of studies minus one degree of freedom. 3.3. Formulae used in the meta-analysis: general variance-based methods For the categorical dependent variable pregnancy rate at ®rst AI, the formulae applied were: s P 1 1 i ln RRi i w P ln RRS ln CI ln RRS 1:96 P ; ; wi var i i wi i wi X fwi ln RRS ln RRi g Q i
where, RRS is the summary estimate of the relative risk of conception at ®rst AI between cows of a BCS or BC-loss category and the reference or control group, RRi is the relative risk of conception in the ith study, wi is the weight assigned to the ith study, var i is the variance of the relative risk in the ith study, CI is the 95% con®dence interval of the summary estimate, Q: refers to a chi-square distribution with the number of estimates
806
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
minus 1 degrees of freedom to test for homogeneity between the results of different studies. For the continuous dependent variable number of days open, the formulae applied were: s P X wi Di 1 1 i ; CI DS 1:96 P ; Q DS P ; wi fwi DS Di g var i i wi I wi i where, DS is the summary estimate of the difference in days open between cows of a BCS or change category and the reference or control group, Di is the effect size of the difference in the ith study, wi is the weight assigned to the ith study, var i is the variance of the effect size in the ith study, CI is the 95% con®dence interval of the summary estimate, Q: refers to a chi-square distribution with the number of estimates minus 1 degrees of freedom to test for homogeneity between the results of different studies. 4. Results 4.1. Effects of BCS and change on pregnancy rate at ®rst AI Table 4 shows the effects of BCS at parturition and at ®rst AI, and of body condition change during the early lactating period on pregnancy rate at ®rst AI. Ten studies performed on a total of 7733 cows were included in the analysis: three experimental [34,35,39], and seven observational [4,7,9,31±33,36]. Two trials from study 7 were independently analyzed: 7a (652 primiparous cows), and 7c (1040 multiparous cows). BCS at parturition was associated with the relative risk of conception only in cows showing a low BCS at parturition. Animals in poor condition showed a signi®cant (9%) reduction in pregnancy rate (relative risk: 0.91; CI: 0.84, 0.99) at ®rst AI compared to Table 4 Effect of BCS and score change on pregnancy rate at first AI Variable
Relative risk of conception
95% CI
Pooled number of cows
Reference numbers
Inter-study homogeneity
BCS at parturition (reference category: 2.5±3.5) <2.5 0.91 0.84, 0.99 >3.5 1.04 0.95, 1.13
3644 3723
[4,7aa,7cb,34] [4,7aa,7cb,34,35]
Yesc Nod
BCS at first AI (reference category: 2.5±3.5) <2.5 0.91 0.85, 0.98 >3.5 1.05 0.97, 1.13
2122 1786
[32,34,36] [32,34]
Noe Yesf
Body condition change during early lactation (reference category: <0.5 point loss) 0.5±1 point loss 0.96 0.91, 1.01 3975 [9,31,33,34,35,39] >1 point loss 0.9 0.82, 0.99 3837 [9,31,33±35] Increase 1.03 0.98, 1.09 3474 [9,31,34,39]
Yesg Noh Noi
(a, b) Trials 1 and 2, respectively [7]. (c) Q 0:8, 3 d.f., w2 7:8. (d) Q 30:8, 4 d.f., w2 9:5. (e) Q 10:9, 2 d.f., w2 6. (f) Q 2:8, 1 d.f., w2 3:8. (g) Q 6:4, 5 d.f., w2 11:1. (h) Q 17:1, 4 d.f., w2 9:5. (i) Q 9, 3 d.f., w2 7:8.
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
807
animals in intermediate condition. The summary estimate was based on one experimental [34] and three observational [4,7a,7c] studies, involving 3644 medium or high producing cows in different lactation periods. Studies 7c and 34 were performed on over 1000 animals. All four papers reported homogeneous effects. A high BCS at parturition was not associated with pregnancy rate at ®rst AI. Results from two experimental [34,35] and three observational [4,7a,7c] studies performed on a variable number of cows of different lactation number and milk production were heterogeneous. BCS at ®rst AI was not associated with the relative risk of conception. The ®ndings of one experimental [34] and two observational [32,36] studies evaluating the effect of a low BCS at AI on pregnancy rate were heterogeneous. Based on results from one experimental [34] and one observational [32] trial, the effect of a high BCS at AI on pregnancy rate was not signi®cant (relative risk: 1.05; CI: 0.97, 1.13). Body condition change during the early lactation period was not associated with the relative risk of conception. We analyzed the results of three experimental [34,35,39] and three observational [9,31,33] studies derived from large study populations (3474±3975 animals). The studies included a variable number of cases, lactations, and milk production ®gures. The summary estimate of the effect of body condition change on pregnancy rate at ®rst AI was either non-signi®cant (for a moderate body condition loss) or heterogeneous (for a gain or a severe loss in body condition). 4.2. Effects of BCS and change on days open Table 5 shows the effects of BCS at parturition and at ®rst AI, and of body condition change during the early lactation period on days open. Ten studies performed on a total number of 4529 cows were included in the analysis: four experimental [29,30,34,35] and six observational [5,7,33,36±38]. Two trials from study 7 were independently analyzed: 7b (683 primiparous cows), and 7d (1168 multiparous cows). Table 5 Effect of BCS and change on number of days open Variable
Summary effect size (days)
95% CI
Pooled number of cows
Reference numbers
BCS at parturition (reference category: 2.5±3.5) <2.5 5.9 1.4, 10.4 >3.5 5.8 9.8, 1.9
3078 3755
[7ba,7db,29,34] Yesc [5,7ba,7db,29,30,34,35,37,38] Yesd
BCS at first AI (reference category: 2.5±3.5) <2.5 12.2 4.8, 19.7 >3.5 11.9 19.5, 4.3
1418 1211
[34,36] [34]
Yese (Single study)
[5,30,33,34,37] [5,30,33,34,37] [5,33,34]
Yesf Yesg Yesh
Body condition change (reference category: <0.5 point loss) 0.5±1 point loss 3.5 2.5, 9.6 2012 >1 point loss 10.6 3.9, 17.3 1863 Increase 3.7 10.7, 3.3 1823
Inter-study homogeneity
(a, b) Trials 3 and 4, respectively [7]. (c) Q 4:1, 3 d.f., w2 7:8. (d) Q 12, 8 d.f., w2 15:5. (e) Q 0:1, 1 d.f., w2 3:8. (f) Q 1:5, 4 d.f., w2 9:5. (g) Q 4:4, 4 d.f., w2 9:5. (h) Q 1:7, 2 d.f., w2 6.
808
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
Cows with a low BCS at parturition remained open for approximately 6 further days, compared to cows in the intermediate body condition category. The effect size was signi®cant (CI: 1.4, 10.4) and included two experimental studies on medium milk producing cows [29,34], and two observational studies involving high milk producers [7b,7d]. The study population for studies 7d and 29 was greater than 1000 animals. The results of the four reports were homogeneous. Cows showing a high BCS at parturition were open approximately 6 days less than cows with a medium body condition. The size of the effect size was signi®cant (CI: 9.8, 1.9) and was based on the outcomes of experimental [29,30,34,35] and observational [5,7c,7d,37,38] studies. Studies were homogeneous results and included a variable number of medium or high milk producing cows. Only two studies [34,36] evaluated the effect of a low BCS at the time of ®rst AI on days open. The summary effect size indicated that cows in poor condition at ®rst AI needed 12.2 further days to become pregnant (CI: 4.8, 19.7), compared to cows in the medium body condition category. This analysis of one experimental [34] and one observational [36] study was homogeneous and included 1418 medium or high milk producers. A single experimental study [34] performed on 1211 animals related a high BCS at ®rst AI with a 12-day decrease in the number of days open (CI: 19.5, 4.3). A moderate loss of body condition (0.5±1 points) during the early lactation period was not signi®cantly associated with an increased number of days open (summary effect size: 3.5 days; CI: 2.5, 9.6). The papers included were homogeneous and involved experimental [30,34] and observational [5,33,37] studies performed on different population sizes and on animals showing different lactation numbers and milk production. Results from these same studies showed a signi®cant increase of 10.6 days open (CI: 3.6, 17.3) in cows suffering a severe body condition loss (greater than 1 point) during the early lactation period, relative to animals undergoing a slight loss in body condition. Effects were also homogeneous. No signi®cant effect could be established for cows experiencing a body condition increase (gain in points) during early lactation. The homogeneous summary estimate (effect size: 3.7 days; CI: 10.7, 3.3) included one experimental [34] and two observational [5,33] studies performed on cows showing low or medium milk yields. 5. Discussion 5.1. Analysis of heterogeneity We selected the variables pregnancy rate at ®rst AI and number of days open for our analysis since they are the factors most commonly used to evaluate possible biological effects of BCS, or score change, on reproductive performance. The results of the effects of body condition category on pregnancy rate at ®rst AI were highly heterogeneous. Heterogeneity was detected in at least one subset of all the sections considered: BCS at parturition, at ®rst AI, and body condition change during the early lactation period. In contrast, when the variable number of days open was considered, all the studies presented homogeneous results.
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
809
Statistical heterogeneity among studies occurs when the original results combined in a meta-analysis are suf®ciently different to reject the hypothesis of homogeneity [11]. Investigating possible sources of heterogeneity identi®es factors that can lead to heterogeneity. The experimental studies included in the meta-analysis all involved low numbers of animals (lower than 100, except for study 34) of low or medium milk production, whereas the observational trials were usually performed on a large number of medium and high milk producers. Thus, the main source of heterogeneity seems to be the type of study. Nevertheless, the effects of body condition status or change on the variable number of days open were highly homogeneous among studies. In this case, the type of study did not contribute to heterogeneity. It is likely that the variable pregnancy rate at ®rst AI is itself the persistent source of heterogeneity in the meta-analysis. In observational studies, this variable can be strongly affected by farm policy regarding the voluntary waiting period or by failure to detect estrus. BCS is dynamic and is strictly related to the physiological cycle of cows; it decreases in early lactation, is restored during mid-lactation, and reaches a near steady state during late lactation [40]. Thus, a longer parturition to AI interval should improve postpartum metabolic recovery. For instance, better results would be expected when a herd is allowed 90 days to ®rst AI than 60 days. However, in both cases cows could have similar body conditions. Additional factors that may in¯uence the pregnancy rate should be taken into account in the design of future studies aimed at evaluating the effects of body condition on pregnancy rate at ®rst AI. In contrast, the homogeneity of results from studies based on the variable number of days open is consistent with the idea that the average number of days open, better described as the parturition to conception interval, is the best predictor of overall reproductive performance in a herd. The voluntary waiting period, estrus detection accuracy and pregnancy rate at ®rst service are only components of this indicator [41]. 5.2. Effects of BCS and change on pregnancy rate at ®rst AI A clear association between BCS and pregnancy rate at ®rst AI was found only for the effect of a low BCS at parturition. The results of all the studies were homogeneous; pregnancy rate at ®rst AI showed a signi®cant drop of about 10% in cows delivering in poor condition. This reduced fertility could be a consequence of prolonged anovulatory intervals, which is frequent in thin cows and has a negative impact on ®rst service conception rates [42]. 5.3. Effects of BCS and change on number of days open BCS at parturition and at ®rst AI were good indicators of the relationship between the nutritional status of the cow and the number of days open. In animals showing a good body condition (score above 3.5) at parturition the number of days open was signi®cantly reduced (5.8 or 11.7, respectively) compared to animals in the intermediate (score 2.5±3.5) or low (score under 2.5) body condition category. At ®rst AI, animals in good condition were related to a signi®cant decrease in days open when compared with animals in the intermediate or low body condition category (11.9 or 24.1, respectively). As expected, a closer relationship (an effect double the size) between BCS and days open was noted at ®rst
810
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
AI than at parturition. At ®rst AI, most cows are recovering from metabolic postpartum stress and their energy balance will not become positive until next parturition. On the contrary, BCS at parturition is a worse indicator. Depending on the effects of the period of negative energy balance during the early lactation period, BCS may undergo a further drop. Mobilization of body fat stores following parturition may be limited in thin cows and considerable in fat animals. Some thin cows regain their BCS before the time of ®rst AI, while over-conditioned cows may reach the time of AI in poor condition. This rationale was supported by the analysis of effects of body condition change on the number of days open. A severe loss of body condition (a drop in score of over 1 unit) during the early lactation stage was related to a signi®cant 10.6 increase in the number of days open. Ovarian follicular dynamics in the postpartum dairy cow appear to be unaffected by a negative energy balance [42]. Our ®ndings were consistent with this; a moderate body condition change, either loss or gain, was not signi®cantly related to the number of days open. However, as noted above, a severe postpartum body condition loss resulted in an increased number of days open. In this case, it is possible that postpartum ovarian activity is affected. Our results show that the variable number of days open is a good predictor of the effects BCS or change will have on reproductive performance. Better results were obtained in cows in good body condition (score higher than 3.5) at parturition and ®rst AI, and in those showing a slight or moderate loss in postpartum body condition (less than 1 unit). Given that nutritional status is associated with management strategy, special attention should be paid to the cows' nutritional needs during the periparturition period, between late pregnancy and early lactation (also called the transition period [43]). Cow management during this period is acquiring special relevance in the prevention of metabolic disorders [43,44]. Overfeeding during the dry period has been related to a predisposition to accumulate fat in adipose tissue, increased lipolysis postpartum and a lower ability of the adipose tissue to reesterify mobilized fatty acids [45]. Excessive lipid mobilization from adipose tissue has been linked to a greater incidence of periparturient health disorders [43]. We could add that this is a major factor affecting the reproductive performance of cows showing good body condition at parturition. Studies aimed at evaluating the effects of BCS or score change during the dry period on the subsequent number of days open would be of great interest. Acknowledgements The authors thank Ana Burton for assistance with the English translation. References [1] Hady PJ, Domecq JJ, Kaneene JB. Frequency and precision of body condition scoring in dairy cattle. J Dairy Sci 1994;77:1543±7. [2] Edmonson AJ, Lean IJ, Weaver LD, Farver T, Webster G. A body condition scoring chart for Holstein dairy cows. J Dairy Sci 1989;72:68±78. [3] Butler WR. Nutritional interactions with reproductive performance in dairy cattle. Anim Reprod Sci 2000;60/61:449±57.
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
811
[4] Heuer C, Schukken YH, Dobbelaar P. Postpartum body condition score and results from the ®rst test day milk as predictors of disease, fertility, yield, and culling in commercial dairy herds. J Dairy Sci 1999;82:295±304. [5] Ruegg PL, Milton RL. Body condition scores of Holstein cows on Prince Edward Island, Canada: relationships with yield, reproductive performance, and disease. J Dairy Sci 1995;78:552±64. [6] Gearhart MA, Curtis CR, Erb HN, Smith RD, Sniffen CJ, Chase LE, et al. Relationship of changes in condition score to cow health in Holsteins. J Dairy Sci 1990;73:3132±40. [7] Markusfeld O, Galon N, Ezra E. Body condition score, health, yield and fertility in dairy cows. Vet Rec 1997;141:67±72. [8] Waltner SS, McNamara JP, Hillers JK. Relationships of body condition score to production variables in high producing Holstein dairy cattle. J Dairy Sci 1993;76:3410±9. [9] Domecq JJ, Skidmore AL, Lloyd JW, Kaneene JB. Relationship between body condition scores and conception at ®rst arti®cial insemination in a large dairy herd of high yielding Holstein cows. J Dairy Sci 1997;80:113±20. [10] Suriyasathaporn W, Nielen M, Dieleman SJ, Brand A, Noordhuizen-Strasse EN, Schukken YH. A cox proportional-hazards model with time-dependent covariates to evaluate the relationship between bodycondition score and the risks of ®rst insemination and pregnancy in a high producing dairy herd. Prev Vet Med 1998;37:159±72. [11] Petitti DB. Meta-analysis, decision analysis, and cost-effectiveness analysis. Methods for quantitative synthesis in medicine. Oxford: Oxford University Press; 2000. [12] Domecq JJ, Skidmore AL, Lloyd JW, Kaneene JB. Relationship between body condition scores and milk yield in a large dairy herd of high yielding Holstein cows. J Dairy Sci 1997;80:101±12. [13] Prandi A, Messina M, Tondolo A, Motta M. Correlation between reproductive ef®ciency, as determined by new mathematical indexes, and the body condition score in dairy cows. Theriogenology 1999;52: 1251±65. [14] Garnsworthy PC, Jones GP. The in¯uence of body condition at calving and dietary protein supply on voluntary food intake and performance in dairy cows. Anim Prod 1987;44:347±53. [15] Gearhart MA, Curtis CR, Erb HN, Smith RD, Sniffen CJ, Chase LE, et al. Relationship of body condition score and changes in condition score with health, reproductive performance and productivity in New York Holstein dairy herds. Acta Vet Scand 1988;84(Suppl):122±5. [16] Holter JB, Stolnick MJ, Hayes HH, Bozak CK, Urban WE, McGilliard ML. Effect of prepartum dietary energy on condition score, postpartum energy, nitrogen partitions, and lactation production responses. J Dairy Sci 1990;73:3502±11. [17] Treacher RJ, Reid IM, Roberts CJ. Effect of body condition at calving on the health and performance of dairy cows. Anim Prod 1986;43:1±6. [18] Senatore EM, Butler WR, Oltenacu PA. Relationships between energy balance and postpartum ovarian activity and fertility in ®rst lactation dairy cows. Anim Sci 1996;62:17±23. [19] Ducker MJ, Morant SV. Observations on the relationships between the nutrition, milk yield, live weight and reproductive performance of dairy cows. Anim Prod 1984;38:9±14. [20] Wildman EE, Jones GM, Wagner PE, Boman RL, Troutt Jr HF, Lesch TN. A dairy cow body condition scoring system and its relationship to selected production characteristics. J Dairy Sci 1982;65:495±501. [21] Broster WH, Broster VJ. Body score of dairy cows. J Dairy Res 1998;65:155±73. [22] Burkholder WJ. Use of body condition scores in clinical assessment of the provision of optimal nutrition. J Am Vet Med Assoc 2000;217:650±4. [23] Ducker MJ, Morant SV, Fischer WJ, Haggett RA. Nutrition and reproductive performance of dairy cattle. Part 2. Prediction of reproductive performance in ®rst lactation dairy heifers subjected to controlled nutritional regimes. Anim Prod 1985;41:13±22. [24] Nebel RL, McGilliard ML. Interactions of high milk yield and reproductive performance in dairy cows. J Dairy Sci 1993;76:3257±68. [25] Ruegg PL. Body condition scoring in dairy cows: relationships with production, reproduction, nutrition, and health. Comp Cont Educ Pract Vet 1991;13:1309±13. [26] Aeberhard K, Bruckmaier RM, Kuepfer U, Blum JW. Milk yield and composition, nutrition, body conformation traits, body condition scores, fertility and diseases in high yielding dairy cows: Part 1. J Vet Med Ser A 2001;48:99±110.
812
F. LoÂpez-Gatius et al. / Theriogenology 59 (2003) 801±812
[27] Reist M, Koller A, Busato A, Kuepfer U, Blum JW. First ovulation and ketone body status in the early postpartum period of dairy cows. Theriogenology 2000;54:685±701. [28] Snijders SEM, Dillon PG, O'Farrel KJ, Diskin M, Wylie ARG, O'Callaghan D, et al. Genetic merit for milk production and reproductive success in dairy cows. Anim Reprod Sci 2001;65:17±31. [29] Garnsworthy PC, Topps JH. The effect of body condition of dairy cows at calving on their food intake and performance when given complete diets. Anim Prod 1982;35:113±9. [30] Jones GP, Garnsworthy PC. The effects of body condition at calving and dietary protein content on drymatter intake and performance in lactating dairy cows given diets of low energy content. Anim Prod 1988;47:321±33. [31] Loef¯er SH, Vries MJ, Schukken YH. Effects of time of disease occurrence, milk yield, and body condition on fertility of dairy cows. J Dairy Sci 1999;82:2589±604. [32] Loef¯er SH, Vries MJ, Schukken YH, de Zeeuw AC, Dijkhuizen AA, de Graaf FM, et al. Use of AI technician scores for body condition, uterine tone and uterine discharge in a model with disease and milk production parameters to predict pregnancy risk at ®rst AI in Holstein dairy cows. Theriogenology 1999;51:1267±84. [33] Gillund P, Reksen O, GroÈhn YT, Karlberg K. Body condition related to ketosis and reproductive performance in Norwegian dairy cows. J Dairy Sci 2001;84:1390±6. [34] Pryce JE, Coffey MP, Simm G. The relationship between body condition score and reproductive performance. J Dairy Sci 2001;84:1508±15. [35] Butler WR, Smith RD. Interrelationships between energy balance and postpartum reproductive function in dairy cattle. J Dairy Sci 1989;72:767±83. [36] Moreira F, Risco C, Pires MFA, Ambrose JD, Drost M, DeLorenzo M, et al. Effect of body condition on reproductive ef®ciency of lactating dairy cows receiving a timed insemination. Theriogenology 2000;53:1305±19. [37] Ruegg PL, Goodger WJ, Holmberg CA, Weaver LD, Huffman EM. Relation among body condition score, serum urea nitrogen and cholesterol concentrations, and reproductive performance in high producing Holstein dairy cows in early lactation. Am J Vet Res 1992;53:10±4. [38] Pedron O, Cheli F, Senatore E, Baroli D, Rizzi R. Effect of body condition score at calving on performance, some blood parameters, and milk fatty acid composition in dairy cows. J Dairy Sci 1993;76:2528±35. [39] Ducker MJ, Hagget RA, Fisher WJ, Morant SV, Bloom®eld GA. Nutrition and reproductive performance of dairy cattle. Part 1. The effect of level of feeding in late pregnancy and around the time of insemination on the reproductive performance of ®rst lactation dairy heifers. Anim Prod 1985;41:1±12. [40] Gallo L, Carnier P, Cassandro M, Mantovani R, Bailoni L, Contiero B, et al. Change in body condition score of Holstein cows as affected by parity and mature equivalent milk. J Dairy Sci 1996;79:1009±15. [41] Fetrow J, Stewart S, Eicker S. Reproductive health programs for dairy herds: analysis of records for assessment of reproductive performance. In: Youngquist RS, editor. Current therapy in large animal theriogenology. Philadelphia: Saunders; 1997, p. 441±51. [42] Beam SW, Butler WR. Effects of energy balance on follicular development and ®rst ovulation in postpartum dairy cows. J Reprod Fertil 1999;54(Suppl):411±24. [43] Drackley JK. Biology of dairy cows during the transition period: the ®nal frontier? J Dairy Sci 1999;82:2259±73. [44] Gerloff BJ. Dry cow management for the prevention of ketosis and fatty liver in dairy cows. Vet Clin North Am Food Anim Pract 2000;16:283±92. [45] Rukkwamsuk, Wensing T, Geelen MJH. Effect of overfeeding during the dry period on the rate of esteri®cation in adipose tissue of dairy cows during the periparturient period. J Dairy Sci 1999;82:1164±9.