www.elsevier.comrlocateranireprosci
Epidemiology of reproductive performance in dairy
cows
Y.T. Grohn
¨
a,), P.J. Rajala-Schultz
a,ba
Section of Epidemiology, Department of Population Medicine and Diagnostic Sciences, College of Veterinary Medicine, Cornell UniÕersity, Ithaca, NY 14853 ,USA
b
Department of Veterinary PreÕentiÕe Medicine, College of Veterinary Medicine, The Ohio State UniÕersity,
Columbus, OH 43210, USA
Abstract
The objectives of this presentation are to review results of our previous and on-going research with respect to the risk factors and consequences of poor reproductive performance in dairy cows, and to develop an economic framework to optimize decisions related to dairy cow reproductive performance. To make profitable breeding and replacement decisions, the farmer must account for factors including age, production level, lactation stage, pregnancy status, and disease history of the cows in the herd. Establishing the interrelationships among disease, milk yield, reproduction, and herd management is necessary for developing a decision model for disease treatment, insemina-tion, and replacement.
The data for the studies reviewed in this presentation incorporate health, production, and management components from Holsteins in the Northeast USA and Ayrshires from Finland. Data were analyzed using the Cornell Theory Center Supercomputer. The effect of risk factors on reproductive disorders was modeled with logistic regression, and on conception, insemination, and culling with survival analysis. The effect of reproductive disorders on milk yield was analyzed with mixed models. Economic optimization of reproductive performance was done with dynamic
Ž .
programming DP .
High milk yield, high parity, and calving in winter were risk factors for several reproductive disorders. These disorders, in turn, delayed insemination and conception in dairy cows, and some of them increased the risk of culling. Dystocia, retained placenta, and early metritis led to a short-term drop in milk production. High milk yield was not a major factor in delaying conception, except in first parity cows. However, higher yielders were more likely to be inseminated, and less likely to be culled. Non-pregnant cows had a higher risk of being culled.
)Corresponding author. Tel.:q1-607-253-3571; fax:q1-607-253-3083.
Ž .
E-mail address: ytg1@cornell.edu Y.T. Grohn .¨
0378-4320r00r$ - see front matterq2000 Elsevier Science B.V. All rights reserved. Ž .
Reproductive performance of dairy cows influenced a herd’s profitability, and good heat detection and conception rates provided opportunities for management control. It was not always economically advantageous to get cows pregnant as soon as possible, and there was no one optimal value for the calving interval length for all cows in a herd.q2000 Elsevier Science B.V.
All rights reserved.
Keywords: Dairy cows; Reproduction; Epidemiology
1. Introduction
Profitable milk production relies upon a careful, efficient and cost-effective manage-ment of dairy herds. Poor reproductive performance is an important production-limiting factor. Management decisions on dairy farms should be economically justified. Our ultimate goal is to help farmers make the best decisions regarding management of their dairy herds. To do this, we must understand the biological parameters involved, namely milk yield, disease, and reproductive performance.
Although many studies on reproductive performance in dairy cows have been conducted, few have integrated together epidemiology and economics. Many of them have been constrained by the lack of three factors: a large health database, computing power, and sophisticated statistical tools. Fortunately, we have had access to high quality
Ž Ž .
data on Finnish Ayrshires beginning with Grohn et al. 1986 , and continuing until the
¨
.present ; such data have not been previously available in the United States, where Holsteins predominate. Therefore, we established a health and production database in
Ž Ž .
the Northeast USA beginning with Grohn et al.
¨
1995 , and continuing until the .present to study the epidemiology and economics of production diseases under North-eastern management.
In this presentation, our focus is reproductive performance in dairy cows. We will
Ž . Ž .
concentrate on three aspects of it: 1 risk factors for poor reproductive performance; 2 Ž .
consequences of poor reproductive performance; and 3 economic optimization of reproductive performance. The material for this review was primarily taken from our on-going epidemiological research on New York Holsteins and Finnish Ayrshires.
2. Risk factors for poor reproductive performance
2.1. Increased risk of reproductiÕe disorders
The dairy industry’s major goal is to provide milk for the consumer market. Over the past several decades, milk yield of cows has increased markedly. However, one negative
Ž
impact of this improvement is reduced fertility e.g., cows open for longer, more .
services per conception and higher incidence of reproductive problems, which, in turn,
Ž Ž ..
energy balance in some cows, especially those still growing, and disease may occur. Selective culling muddies the issue further. High yielders are more likely to remain in the herd, and receive more veterinary treatment, even when they become ill, than low yielders.
To address this question, we conducted several studies in both Finnish Ayrshires and
Ž .
New York Holsteins e.g., Grohn et al., 1990, 1994, 1995; Rajala and Grohn, 1998a .
¨
¨
Ž Lactational incidence risks for reproductive disorders in our latest Finnish study Rajala.
and Grohn, 1998a were: dystocia, 2.1%; retained placenta, 3.1%; metritis, 3.2%; silent
¨
heat, 8.1%; ovarian cysts, 7.3%; and other infertility, 1.9%. In our New York study ŽGrohn et al., 1995 , incidences were higher: retained placenta, 7.4%; metritis, 7.6%;¨
. and ovarian cysts, 9.1%.Using logistic regression, we analyzed the effect of several risk factors on the
Ž .
occurrence of reproductive disorders. In one study Grohn et al., 1990 , first parity cows
¨
Ž .
were most likely to have dystocia Table 1 . Older cows were more likely to have retained placenta and ovarian cysts. Winter calvers had higher risks of early metritis, silent heat, ovarian cysts, and other infertility than summer calvers did. Most reproduc-tive disorders were interrelated. For example, dystocia was a risk factor for retained
Table 1
Ž .
Risk factors parity, calving season, diseases for seven diseases, in 61,124 Finnish Ayrshires. All seven models included community as a proxy of managementa
Risk factors Diseases
DystociaRetained placentaOvarian CystsEarly metritisLate metritisSilent heatOther infertility
Parity
Values are odds ratios, which measure how much more or less likely the outcome is among observations
Ž . Ž
with a given risk factor or level of risk factor if)2 categories , compared with those without it or reference
.
category, which has an odds ratio of 1.0 . Milk yield was also included in these models.
U
placenta and metritis. Results were similar in our study on New York Holsteins, for
Ž .
parity and calving season, but there were fewer disease associations Grohn et al., 1995 .
¨
Cows with higher milk yield in the previous lactation were more likely to have retained placenta, early metritis, silent heat, ovarian cysts, and other infertility than were cows with lower milk yield. Cows with higher milk yield in the current lactation were more likely to have ovarian cysts than were cows with lower milk yield. In a similarŽ .
study on New York Holsteins Grohn et al., 1995 , higher milk yield was associated
¨
only with ovarian cysts.In addition to breed difference, other factors may account for the greater number of disorders associated with milk yield in Ayrshires than Holsteins. Finnish farms are much smaller than New York farms so management differs. The Finnish studies contained over five times as many cows as our New York study, so smaller differences become significant. Also, in New York, a farmer may treat animals for disease, but in Finland only veterinarians treat animals.
2.2. Risk factors for delayed conception
Many factors influence reproductive performance in dairy cows, only some of which are under the farmer’s control. Milk yield may play an important role in reproduction. However, the relationship between milk yield and conception is difficult to study because of the confounding effect of culling. Conception does not necessarily occur at the first breeding; cows may have to be inseminated several times. A cow that remains
Ž .
open i.e., does not conceive for a long time or has low milk yield is more likely to be culled. Some researchers have argued that high yielders are less fertile than low yielders are; this association may, however, be complicated by selective culling.
The objective of our New York study was to measure the effect of 60-day cumulative
Ž .
milk yield on rates of conception and first breeding in lactation Eicker et al., 1996 . Data were from 15,320 Holsteins calving between June 1990 and November 1993, in 26 herds. We used survival analysis, which allows inclusion of all cows, whether or not they had conceived or been bred by the end of the study. Thus, the loss of information was minimized. Parity, calving season, and herd as a proxy of management were included as confounders.
Ž Retained placenta, metritis, and ovarian cysts were risk factors for conception Table .
2 . Cows with retained placenta had a 14%, those with metritis a 15%, and those with ovarian cysts a 21% lower conception rate than cows free of these disorders. Current cumulative 60-day milk yield had no effect on conception rates, except among the
Ž .
highest yielders. They had a slightly lower but not significant conception rate than did the lowest yielders.
Because these results may be a consequence of farmers intentionally delaying insemination of high yielding cows, we also studied the association between milk yield and time to first insemination. As 60-day milk yield increased, so did insemination rates. The highest yielders were nearly 30% more likely to be inseminated than were the lowest yielders.
Table 2
Effect of milk yield, parity, calving season, and disease on conception in 13,307 New York Holsteins
a
Risk factor Hazard ratio
( ) First 60-day cumulatiÕe milk yield kg
F1582 1.0
Hazard ratios for factors in proportional hazards model. The hazard ratio is the ratio between two relative
Ž . Ž .
risks of an event e.g., conception . For example, if a cow has a hazard ratio of 0.92 parityG3 , then she has an 8% reduced likelihood of conceiving than a first parity cow.
U
p-0.05.
UU
p-0.01.
conception rate than their herdmates. Older cows and sick cows were less likely to conceive. In contrast, the rate of being bred increased with 60-day milk. These findings demonstrate that farmers are making rational decisions by breeding young, healthy, high yielding cows.
Among 30,036 multiparous Finnish Ayrshires, the lowest yielders were less likely to
Ž .
conceive than were average yielders Harman et al.,1996a . Among 11,761 heifers, the highest yielders were less likely to conceive than were average yielders. A number of disorders decreased conception probability in both multiparous and primiparous cows: anestrus, ovulatory dysfunction, other infertility, late metritis, and clinical ketosis ŽHarman et al., 1996b ..
3. Consequences of poor reproductive performance
3.1. Lowered milk yield due to reproductiÕe disorders
Ž
in the analysis is important; a single summary measure of milk yield e.g., 305-day
. Ž
yield may give a completely different answer than monthly milk yields Detilleux et al., .
1994 .
Ž .
In one of our studies Rajala and Grohn, 1998b , we estimated the effects of dystocia,
¨
retained placenta, and metritis on milk yield using repeated, monthly test day milk yields taken on 37,776 Finnish Ayrshire cows. All three disorders had a significant effect on milk yield. Their impact varied across parities. Milk losses were greatest immediately following diagnosis. For instance, in parity 2, daily losses from dystocia, retained placenta, and early metritis during the 2 weeks following diagnosis were 2.2, 1.4, and 1.3 kgrday, respectively.There will probably always be some loss of milk yield after a disease. It is important to use proper methodology when quantifying the loss, i.e., it is more accurate to use monthly measurements of milk yield rather than a single, summary lactational measure. Only in this way can true losses be observed. The measurement of loss of milk yield following disease is important. Knowing whether the milk loss due to disease is small and temporary or larger and more sustained helps a farmer decide whether it is worth keeping the cow in the milking herd.
3.2. Shortened productiÕe life
Culling is a complex issue: many factors are involved. Dairy cows are culled for
Ž . Ž .
involuntary i.e., death, acute disease and infertility and voluntary i.e., low yield reasons. Both biology and management affect the culling decision. Reproductive failure
Ž .
is a major reason for involuntary culling Fetrow, 1988 . However, when making a decision, the dairy farmer considers at least five major reasons: parity, milk yield, stage of lactation, diseases, and conception status.
Using survival analysis with time-dependent covariates, we studied the effect of seven diseases on culling in New York Holsteins. The effects of milk yield and conception status on culling were also studied, and the interactions between these
Ž .
covariates and stage of lactation were accounted for Grohn et al., 1998 . Cows that had
¨
not yet conceived, as well as older cows and cows with certain diseases, had a higherŽ .
risk of culling. For instance, open cows i.e., before conception were seven-and-a-half times more likely to be culled than pregnant cows.
We have also conducted a series of studies on Finnish Ayrshires, of increasing complexity, starting with only disease effects on culling in the model, then adding
Ž
pregnancy status effects, and finally adding milk yield effects Rajala-Schultz and .
Table 3
Effects of pregnancy status and number of inseminations on culling in 39,727 Finnish Ayrshire cows calving in 1993. The model also included parity, season, and herd
Pregnancy status
a
Ž . Ž .
Lactational stage of culling day Lactational stage of conception day Risk ratio
0–150 0–150 1.0
Stage of lactation when the farmer was assumed to know when a cow was pregnant; 0 refers to open cows
Ži.e., cows that did not conceive at all ..
the 15 diseases studied, retained placenta was the only disorder without an effect on culling.
Effects of pregnancy status and number of inseminations on culling of Finnish Ayrshires are in Table 3. The data are expressed as risk ratios, which measure the risk of a cow being culled depending on the time when she was known to be pregnant
Ž .
compared with a reference level i.e., pregnant by day 150 after calving, RRs1.0 . The knowledge about a cow’s pregnancy status had a different effect on culling at different stages of lactation. The later the cow conceived, the higher was the risk of her being culled. Cows not inseminated at all were at 10 times higher risk of being culled than cows inseminated only once. Cows inseminated more than once were slightly less likely to be culled than cows inseminated only once.
4. Economic optimization of reproductive performance
Reproductive performance of a dairy herd affects profitability of the farm. Reproduc-Ž
.
produced , and may increase the cost of veterinary services and, as shown above, it also affects the culling rate. Insemination, treatment, and culling decisions represent an area of dairy herd management affecting profitability. To make rational decisions, the farmer needs a valid estimate of the future profitability of each cow, accounting for factors including age, production level, lactation stage, pregnancy status, and disease history.
Ž .
Dynamic programming DP can be used for optimizing a sequence of interrelated decisions. The objective of DP models developed for optimizing breeding and replace-ment decisions in dairy herds is to maximize the expected net present values from
Ž
current and replacement cows over a given decision horizon see, e.g., DeLorenzo et al. Ž1992 . Based on net present values, decisions are made monthly to either 1 keep and.. Ž .
Ž . Ž .
inseminate, 2 keep and not inseminate, or 3 replace a cow. If the net present value of a cow at a certain stage is lower than that of a replacement heifer, a decision to replace her should be made. Similarly, if the net present value from a cow leaving her open is lower than breeding her immediately, a decision should be made to breed her.
Ž .
We used a DP model, described by McCullough 1992 , to determine the structure of
Ž .
an ‘‘optimal herd’’ under Finnish conditions Rajala-Schultz et al., 1999a . The calving interval resulting from the optimal replacement and breeding decisions over a 5-year decision horizon was 363 days; however, there was considerable seasonal variation within parities, depending on the month of calving. It was optimal to have most calvings in the fall, and to keep the calving interval of these cows approximately 1 year. We also estimated the effects of reproductive performance on optimal breeding decisions and
Ž .
profitability of Finnish dairy herds Rajala-Schultz et al., 1999b . It was not optimal to start breeding cows calving in spring and early summer immediately after the voluntary waiting period; instead, it was preferable to allow the calving interval to lengthen for these cows so that their next calving was in the fall. However, cows calving in the fall should be bred immediately after the voluntary waiting period. These findings were mainly due to seasonal milk pricing and higher milk production among fall calvers in Finland. Reproductive performance had a considerable impact on a herd’s profitability; good heat detection and conception rates provided opportunities for management control.
5. Conclusions
Having quantified these biological parameters, they can be incorporated into an economic model to see their impact on a farm’s profitability. The result of this DP approach is the net present value of a cow; this model also provides an insemination value for each cow in a herd, indicating how much could be spent in trying to get a cow pregnant. This information helps farmers make rational decisions to maximize the profitability of their dairy herds. Thus, seeing the farm as an enterprise where all
Ž .
management decisions including breeding and treatment decisions must be economi-cally justified is a strategy for optimizing herd health and efficiency, and, ultimately, its livelihood.
Acknowledgements
The initial work of this project was the AVMA Foundation Research Proposal ‘‘The effect of high milk production on disease and reproduction in dairy cows: logistic and survival analysis,’’ 1991–1992. Major funding for the project was provided by the
Ž
USDA National Research Initiative Competitive Grants Program award number
92-. Ž .
03884 , 1992–1995 and award number 94-02316 , 1994–1997. The current research on optimizing and delivering dairy herd health and management decisions is funded by the
Ž .
USDA Rural Fund of America award number 97-05048 , 1998–2000, and the USDA
Ž .
National Research Initiative Competitive Grants Program award number 98-02405 , 1998–2001. This research was conducted using the resources of the Cornell Theory
Ž .
Center Cornell University, Ithaca, NY . The most recent research conducted on Finnish Ayrshires was supported by the Finnish Academy.
References
DeLorenzo, M.A., Spreen, T.H., Bryan, G.R., Beede, D.K., van Arendonk, J.A.M., 1992. Optimization model: insemination, replacement, seasonal production and cash flow. J. Dairy Sci. 75, 885–896.
Detilleux, J.C., Grohn, Y.T., Quaas, R.L., 1994. Effects of clinical ketosis on test day milk yields in Finnish¨ Ayrshire cattle. J. Dairy Sci. 77, 3316–3323.
Dhaliwal, G.S., Murray, R.D., Dobson, H., 1996. Effects of milk yield, and calving to first service interval, in
Ž .
determining herd fertility in dairy cows. Anim. Reprod. Sci. 41 2 , 109–117.
Eicker, S.W., Grohn, Y.T., Hertl, J.A., 1996. The association between cumulative milk yield and days open¨ and days to first breeding in New York State Holstein cows. J. Dairy Sci. 79, 235–241.
Fetrow, J., 1988. Culling of dairy cows. In: 20th annual meeting of AABP, Phoenix, Arizona, AABP. Grohn, Y.T., Eicker, S.W., Ducrocq, V., Hertl, J.A., 1998. The effect of disease on culling Holstein dairy¨
cows in New York State. J. Dairy Sci. 81, 966–978.
Grohn, Y.T., Eicker, S.W., Hertl, J.A., 1995. The association between previous 305-day milk yield and disease¨ in New York State dairy cows. J. Dairy Sci. 78, 1693–1702.
Grohn, Y.T., Erb, H.N., McCulloch, C.E., Saloniemi, H.S., 1990. Epidemiology of reproductive disorders in¨ dairy cattle: associations among host characteristics, disease and production. Prev. Vet. Med. 8, 25–39. Grohn, Y.T., Hertl, J., Harman, J.L., 1994. The effect of early lactation milk yield on reproductive disorders in¨
dairy cows. Am. J. Vet. Res. 55, 1521–1528.
Harman, J.L., Grohn, Y.T., Erb, H.N., Casella, G., 1996a. Event-time analysis of the effect of 60-day milk¨ production on dairy cow interval-to-conception. Am. J. Vet. Res. 57, 634–639.
Harman, J.L., Grohn, Y.T., Erb, H.N., Casella, G., 1996b. Event-time analysis of the effect of calving season,¨ parity, and concurrent disease on interval-to-conception. Am. J. Vet. Res. 57, 640–645.
McCullough, D.A., 1992. Effects of model specifications and exogenous variables on a stochastic dynamic insemination and replacement model for dairy cattle. MS thesis, Univ. Florida, Dept. of Dairy Science, Gainesville, pp. 168.
Rajala, P.J., Grohn, Y.T., 1998a. Disease occurrence and risk factor analysis in Finnish Ayrshire cows. Acta¨ Vet. Scand. 39, 1–13.
Rajala, P.J., Grohn, Y.T., 1998b. Effects of dystocia, retained placenta, and metritis on milk yield in Finnish¨ Ayrshire cows. J. Dairy Sci. 81, 3172–3181.
Rajala-Schultz, P.J., Grohn, Y.T., 1999a. Culling of dairy cows: Part I. Effects of diseases on culling in¨ Finnish Ayrshire cows. Prev. Vet. Med. 41, 195–208.
Rajala-Schultz, P.J., Grohn, Y.T., 1999b. Culling of dairy cows: Part II. Effects of diseases and reproductive¨ performance on culling in Finnish Ayrshire cows. Prev. Vet. Med. 41, 279–294.
Rajala-Schultz, P.J., Grohn, Y.T., 1999c. Culling of dairy cows: Part III. Effects of disease, pregnancy, and¨ milk yield on culling in Finnish Ayrshire cows. Prev. Vet. Med. 41, 295–309.
Rajala-Schultz, P.J., Grohn, Y.T., Allore, H.G., 1999a. Optimizing replacement decisions on Finnish dairy¨
Ž .
herds: price effects of calving pattern, herdlife and calving interval. Acta Vet. Scand., May 20 in press. Rajala-Schultz, P.J., Grohn, Y.T., Allore, H.G., 1999b. Optimizing breeding decisions for Finnish dairy herds:¨
Ž .
