Influence of dystocia on lamb mortality and postpartal ewe health compared to uneventful vaginal delivery and plannend caesarean section

Strong breed influences have been shown, with reported dystocia rates varying from 4.0% in Merino to 34.0% in Dorset Horn ewes in two studies by the same author (George 1975, 1976). Other authors showed that Texel lambs were three times more likely to require veterinary assistance at birth than lambs of other studied breeds (Scottish Blackface, Mule), with only 44.0% of all Texel lambs born without assistance (Dwyer and Bünger 2012).

Many other studies on ovine dystocia have been conducted in veterinary hospitals or refer to case cohorts presented to veterinary practices. They are thus based on a pre-selected case load and do not allow conclusions regarding the incidence of dystocia in the wider sheep population. In such instances, the proportion of maternal causes of dystocia is commonly over-represented (Scott 1989, Sobiraj 1994, Brounts et al. 2004, Ennen et al. 2013, Voigt et al. 2021).

Dystocia has been attributed to a significant percentage of peripartal ewe deaths and is also a major contributor to perinatal lamb mortality (Bruce et al. 2021). An extended duration of labour prior to presentation for veterinary attention has been shown to significantly increase the odds of stillbirth (Voigt et al. 2021). Stillbirth was also the predominant cause of lamb loss in a field study conducted in 17 large flocks in Germany, and accounted for an average of 45% of lamb losses from birth to sale or retention as breeding animals (Voigt et al. 2019). In a study by Regueiro et al. (2021), 75% of lamb deaths up to the age of 72 hours occurred in primiparous ewes following a prolonged parturition. In Australia, a recent meta-analysis of 11 studies covering a total of 14,413 lamb post mortem examinations revealed that the overall proportion of lamb mortality attributable to dystocia was 47% in these cases (Bruce et al. 2021). Dystocia was also reported as the most frequent cause of periparturient ewe deaths following 595 post mortem examinations from 40 farms in Australia carried out in 2019 and 2020, identifying 41% (2019) and 29% (2020, respectively) of these deaths as attributable to dystocia (McQuillan et al. 2021).

Whether dystocia results in the death of the ewe and/or the lambs is strongly influenced by the level of husbandry, as well as the adequacy, skill and timescale of potential interventions. Inappropriate treatments and neglect (Scott 2003, 2005) or limited observation in extensive husbandry systems (Bruce et al. 2021) contribute to high mortality. Planned lambing management and increased farmer education have been shown to be associated with lower stillbirth rates in a German study (Voigt et al. 2019).

Current data on dystocia-related ewe and lamb mortality are frequently derived from field situations and thus reflect varying levels of observation and care (Hawkins et al. 2021). Such studies also almost invariably lack a control group. Stillbirth rates can additionally be confounded by infectious causes of abortion as shown by the frequent presence of these agents in stillborn lambs from field submissions (Kirkbride 1993, Clune et al. 2021). The true influence of dystocia on ewe and lamb mortality thus needs to be assessed in well-managed and standardized circumstances providing an optimum level of observation and care, in the absence of other flock health issues and infectious abortion agents. We therefore conducted a retrospective whole-flock study in a closely-monitored university sheep flock in good health and free from infectious abortions to assess the frequency and type of dystocia, and any potential influence of birth difficulties on stillbirth and perinatal lamb mortality as well as postpartal ewe health and survival.

Caesarean sections were carried out following a standard protocol using a left flank approach under local anaesthesia with procaine hydrochloride (various suppliers) in right lateral recumbency (Winter 1999, Thorne and Jackson 2000, Khan and Erdogan 2019, Vermunt et al. 2019). Following a left flank incision, the uterus was exteriorized. In case of multiple litters, both horns were incised. Following removal of and contact with the lambs, the ewes were sedated using 0.1 mg/kg body weight (bw) xylazine intramuscularly (various suppliers) because of longer than usual suture times caused by less experienced surgeons (veterinary students). To counter-act uterine contractions induced by the side-effects of xylazine, 0.15 mg clenbuterol (Planipart, Boehringer Ingelheim, Germany) were applied intravenously per animal as approved by the above experimental licenses. The uterine incisions were sutured using a double-layer continuous inverting suture pattern (Cushing suture, Surgicryl monofilament, USP 1 – EP 4, SMI AG, Belgium). A simple continuous suture was applied to jointly close the peritoneum and transverse muscle (Surgicryl monofilament, USP 1 – EP 4, SMI AG, Belgium), followed by a joint simple continuous suture of the two oblique abdominal muscles (Surgicryl PGA, USP 1 – EP 4, SMI AG, Belgium), subcutaneous adaptation (Surgicryl PGA, USP 1 – EP 4, SMI AG, Belgium) and skin closure using metal staples (Manipler AZ – 35W, B Braun, Germany). All ewes received a peri-operative course of at least five days of amoxicillin (15mg/kg body bw, various suppliers) and three days of non-steroidal anti-inflammatory drugs, using either meloxicam (various suppliers, 0.5mg/kg bw subcutaneously) or flunixin-meglumine (various suppliers, 2.2mg/kg bw subcutaneously). From 2018 onwards, additional application of metamizole (20 mg/kg bw intravenously, various suppliers) was required by the experimental license. Immediately after completion of surgery, 70 µg carbetocin (Depotocin, Veyx, Germany) were applied subcutaneously per animal.

All late pregnant ewes were closely monitored continuously between 6.00 and 24.00 h by veterinary students and staff during lambing time. Between 0.00 and 6.00 h, regular observation intervals were extended to 2–3 hours, but whenever a ewe showed signs of imminent parturition at these times, the students on duty kept monitoring the animal until lambing was completed. If a parturition was failing to progress, the ewe was given assistance at lambing by either the student or veterinary surgeon on duty. All cases requiring a higher level of intra-uterine manipulation were given a minimum three day course of amoxicillin at the discretion of the veterinary surgeon. Additional treatments were administered as required on a case by case basis. Any weak lambs were given intensive care including ewe or cow colostrum and assistance with suckling if required, plus additional treatments if necessary depending on their clinical condition(s). Surplus or rejected lambs were bottle-fed or fostered. The ewes and their lambs were placed in individual pens for the first days of life, before being housed together as a group until weaning at around three to four months of age. Detailed clinical and treatment records were kept for all individual animals and were available for retrospective evaluation.

Data were managed using Microsoft Excel for Mac­kintosh version 16.74 and statistically analysed using R (R version 4.3.1 [2023-06-16]). The data set was checked for completeness, and a total of 51 missing values for lamb birth weight (5.2%) and 40 missing values for lamb sex (4.1%) out of the 980 individual lamb records were imputed via a non-parametric multivariable imputation approach by chained random forest algorithm with 1000 trees (missRanger R package; Mayer 2019). This method combines random forest imputation (Stekhoven and Bühlmann 2012, Wright and Ziegler 2017) with predictive mean matching (van Buuren and Groothuis-Oudshoorn 2011) and thus iterates multiple times until the average out-of-bag (OOB) prediction error of predictive models stops to improve. The same method was used to impute two missing values of ewe age (0.2%) and one missing value of placental shedding (0.3%) out of the 576 studied parturitions. The OOB prediction error for all four parameters stopped improving at 5.5 to 0%, indicating very accurate predictions. Imputation was therefore accepted.

Parturitions were categorized as follows: parturition at term by either uneventful vaginal delivery (group I; n=283), delayed progress at parturition / assisted vaginal delivery (dystocia, group II; n=75) or near-term planned caesarean section (n=218) on day 147 to 149 of gestation (group III). In case of multiple lambs per litter, this classification was applied to all lambs born on that occasion, irrespective of the number of lambs requiring assistance during one particular parturition. Mummified foetuses were counted for the assessment of litter size and the presence of dead foetuses, but were excluded from analyses regarding stillbirth and perinatal mortality, since their death occurred during the course of pregnancy and was not birth-related. Uni- and multivariable Bayesian logistic regressions (arm package in R; Gelman and Su 2022) were applied to asses a potential association of the mode of delivery (groups I to III) with stillbirth and lamb perinatal mortality, taking into account the potential confounders ewe age and parity (primiparous, secondiparous, parity ≥ 3), litter size (single, twins, multiples ≥ 3), lamb sex, lamb birth weight and lambing season. In addition, a potential association of the mode of delivery with retained foetal membranes and increased body temperature (fever) in the postpartal period was also assessed. Both these analyses included the predictors age, parity, litter size and the presence of dead lambs. For the evaluation of fever, the occurrence of retained foetal membranes (yes/no) was additionally included in the analysis. Not every ewe lambed every year, so parity and age contained different information.

Univariable Bayesian logistic regressions were initially performed for the four response variables and their corresponding predictors. The predictors were subsequently ranked for their relative importance by means of p value and Chi Square Statistics as an effect size. Final multivariable models were selected by the automated brute-force model selection method with multimodel inference for variable importance using the glmulti package in R. The brute-force approach provided the list of models ranked by the Akaike information criterion (AIC), where the model with the lowest AIC contained the most informative combination of predictors. Results showing a p value of p<0.05 were considered significant, while p<0.1 was considered a tendency.

Near-term elective caesarean section was performed in 218 cases on days 147 (n=129), 148 (n=87) or 149 of gestation (n=2). Uneventful lambing was recorded in 283 cases, while 75 parturitions were classified as dystocia (20.9% of the studied 358 vaginal deliveries). The average gestational period in these cases was 150.2 days (median: 150; range: 145–158 days) based on 252 pregnancies with a known mating date following oestrus synchronization.

Four ewes died or were euthanized within 14 days of giving birth (4/576; 0.7%). These included two cases (2/218; 0.9%) following elective caesarean section, one case following assisted delivery (1/75; 1.3%) and one case after uneventful parturition (1/283; 0.4%). The causes of death included gangraenous mastitis (n=1), perforating abomasal ulcers (n=1), euthanasia due to traumatic leg injury (n=1) and one case of metritis and peritonitis following elective caesarean section and retained foetal membranes. The uterine suture was intact in this animal. Only the latter case was considered directly related to parturition.

The overall stillbirth rate was 3.4% of all lambs born, while overall perinatal mortality was 7.1%; An additional 2.1% of the lambs were born at term in a mummified state, predominantly as part of multiple litters. Retained placenta was diagnosed in 74 of the 576 cases (12.8%), and metritis was present on 7 occasions (1.2%). Fever was observed in the postpartal period in 66 cases (11.5%). These included the seven cases diagnosed with clinical metritis. Table 1 shows the descriptive results and further characteristics of the cases by mode of delivery for the three parturition categories.

The vast majority (84%) of the 75 vaginal deliveries classified as dystocia were of foetal origin. Only 12% of the cases were attributed to maternal dystocia, with delays in the birth process/uterine inertia the most frequently observed maternal cause. Table 2 shows details of the causes of dystocia diagnosed in the 75 assisted vaginal deliveries and their frequency in relation to dystocia cases and all examined vaginal deliveries.

Uneventful wound healing was recorded in 131 of the performed 218 elective caesarean sections (60.1%), while minute findings such as short-lived swelling or minimal, short-lived serosanguinous discharge were noted in 69 cases (31.7%). These were still considered part of a normal healing process, since no abnormal findings were present at the time of removal of the skin staples. Normal wound healing therefore occurred in 200 of the 218 cases (91.7%). In the 18 animals diagnosed with wound healing disorders, twelve (5.5%) were categorized as slight, five (2.3%) as moderate and one (0.5%) as severe (full-length dehiscence of the skin suture). All animals with wound healing disorders made a full recovery.

Similarly, only birth weight (p<0.001) and mode of delivery (p=0.006) were initially significant for perinatal mortality in univariable analyses, with no additional tendencies identified for any of the other studied predictors. Brute force selection retained birth weight, litter size, mode of delivery, lamb sex and ewe age in the final multivariable model for perinatal mortality, and ranked their relative importance in that order, while parity and season were ranked as least important and thus dismissed from the final model (Fig. 2). In the multivariable model, birth weight (p<0.001) was the most significant factor, with decreased odds of perinatal mortality associated with an increased birth weight (OR=0.30, 95% CI 0.23–0.40; p<0.001). All other predictors also showed varying levels of significance in the following order: Litter size (p=0.005), mode of delivery (p=0.015), lamb sex (p=0.020) and ewe age (p=0.022). Upon pairwise comparison of the delivery categories, a significant difference was seen between dystocia and planned caesarean section, with increased odds of perinatal mortality for dystocia (OR=2.81; 95% CI 1.19–6.63; p=0.014), and between uneventful deliveries and dystocia, with decreased odds following uneventful parturitions (OR=0.42; 95% CI 0.18–0.98; p=0.042), but no significant difference was observed in this context between planned caesarean section and uneventful vaginal delivery (p=0.863). Lower odds of perinatal mortality were seen for lambs from higher litter sizes, lambs from younger ewes and female lambs (Table 3). Figure 2 illustrates the model-averaged importance of terms and the predicted probabilities of perinatal mortality following the final multivariable model.

The percentage of retained placenta following near-term elective caesarean section was considerably higher than following the two vaginal delivery categories, and the mode of delivery was significant in the univariable analysis (p<0.001). Ewe age showed a tendency (p=0.077), while parity, litter size and the presence of dead lambs were not significant in univariable analyses. Brute force selection retained the mode of delivery and the presence of dead lambs in the final model and ranked their importance in this order, while age, parity and litter size were ranked as less important and thus not included in the final model (Fig 3). In the multivariable analysis, the mode of delivery remained highly significant (p<0.001), with decreased odds of retained placenta following uneventful vaginal delivery (OR=0.06, 95% CI 0.02–0.16; p<0.001) or dystocia (OR=0.09, 95% CI 0.02–0.35; p<0.001) in comparison to near-term elective caesarean section, and no significant difference between the two vaginal delivery categories (p=0.848). The presence of a dead (mummified or stillborn) foetus showed a tendency (p=0.054), with increased odds of retained foetal membranes in cases carrying a dead foetus (OR=2.77; 95% CI 0.96–7.95; p=0.059). Figure 3 illustrates the model-averaged importance of terms and the predicted probabilities of retained placenta following the final multivariable model.

The mode of delivery showed a tendency (p=0.060) for the development of postpartal fever episodes >40°C in univariable analyses, while the presence of retained foetal membranes was significant (p<0.001) in this context. In the majority of cases (59/66, 89.4%), these fever episodes were not associated with any compromised clinical condition of the affected animals, but clinical metritis was diagnosed in the remaining seven cases. Age, parity, litter size and the presence of dead lambs were not significant in this context in univariable analyses. Brute force selection retained the presence of retained foetal membranes and the mode of delivery in the final multivariable model, with retained placenta remaining significant (p<0.001) and the mode of delivery remaining a tendency (p=0.094) (Fig 4). There were increased odds of developing fever episodes following retained placenta (OR=3.92, 95% CI 1.98–7.76; p<0.001). Pairwise comparison of the modes of delivery only showed a tendency between uneventful vaginal delivery and dystocia, with decreased odds for uneventful parturitions (OR=0.46, 95% CI 0.19–1.10; p=0.094). There was no significant difference between planned caesarean section and uneventful vaginal delivery (p=0.998), or between caesarean section and dystocia (p=0.131). Figure 4 illustrates the model-averaged importance of terms and the predicted probabilities of postpartal fever episodes following the final multivariable model.

A previous case control study compared elective caesarean sections with uneventful vaginal deliveries, but did not assess any potential impact of dystocia on ewe and lamb health and survival (Waage and Wangensteen 2013). The brute-force approach for automated model selection and multimodel inference represents a significant advancement over this approach or the commonly employed backward or forward variable selection methods. Unlike these sequential methods, the brute-force approach systematically assesses all possible combinations of predictors to identify the optimal model that provides the most accurate representation of the data. Given the frequent disagreement regarding the ‘final optimal’ models obtained through forward and backward variable selections, the brute-force approach ranks the models and suggests not only a single “best model” but also a set of qualitatively similar models. These models typically fall within 2 units of AIC and feature different combinations of predictors, irrespective of the classical significance of those predictors. Moreover, similar to the Random Forest algorithm, the brute-force approach generates variable importance plots, but based on distinct decision criteria. While neither backward nor forward selections can produce variable importance plots, and the Random Forest algorithm cannot suggest an “optimal” model, the brute-force approach enhances both the quality and quantity of inferences drawn from the data.

The causes of dystocia observed in the studied flock largely confirm the results of previous whole-flock studies, with foetal causes by far outnumbering maternal causes of dystocia in field situations, and foetal maldisposition the most frequently observed condition (Grommers et al. 1985, Dwyer and Bünger 2012). Only one case of caesarean section due to dystocia was observed in the 14.5-year study period, accounting for 0.3% of deliveries not intended for the elective procedure, making clinical obstetrical indications for caesarean section a very rare event in the studied flock. Comparison of dystocia rates between published studies is difficult, as case definitions and breed characteristics vary. The breed used in the present study typically produces lambs with a lean body conformation. The observed dystocia rate of nearly 21% of all litters therefore seems relatively high. Close and frequent observation and early assistance at lambing following the definition of dystocia as no progress being made within 30 minutes or more (Anderson 2014) were aimed at maximizing lamb survival, but may have led to some potential over-diagnosis in contrast to other publications, which used, for instance, a definition of one to two hours (Grommers et al. 1985) or simply “when assistance was required” (Holst et al. 1997). In some cases classified as dystocia, the ewe may thus still have lambed without assistance, even though this may have further compromised lamb viability. This potential over-diagnosis most likely applies to some of the cases classified as uterine inertia or a delay in the birth process with no other obvious problems, since frequent observation itself and thus possible interruption of the birth process may have been the cause of the delay on some occasions.

Despite early intervention in cases showing delays in the birth process, dystocia was highly significant for the occurrence of stillbirth, with significantly increased odds of stillbirth for dystocia in comparison to elective caesarean section as well as uneventful vaginal delivery. The difference in stillbirth rates between planned caesarean section (0%) and uneventful vaginal delivery (3.9%) was also statistically significant. Even if classified as uneventful, the vaginal birth process itself thus increases the stillbirth risk to the lamb. Birth weight was an equally significant predictor in this context, but contrary to expectations the risk did not increase with increased birth weight. While heavier lambs are more likely to require assistance at birth (Dwyer and Bünger 2012) and are thus more likely to be exposed to the effects of dystocia, it was particularly low birth weights that were associated with higher odds of stillbirth in our study population. This phenomenon is therefore most likely unrelated to parturition difficulties but associated with factors affecting foetal development such as, for instance, placental insufficiency leading to underweight lambs, which are then less viable than lambs with normal or high birth weights. For perinatal mortality, i.e. cumulative deaths to include stillbirth and any postnatal mortality up to the second day of life (Wong et al. 2021), low birth weights outweighed the other predictors in relative importance, but all studied parameters also showed different levels of significance. For litter size and ewe age, the results interestingly indicated better perinatal survival for lambs from triplet or multiple litters in comparison to singles and twins, and poorer perinatal survival for lambs born to older ewes. This observation contradicts previous studies such as, for instance, Holst et al. (1997), which indicate inferior survival rates for triplets, and the common assumption that older, more experienced ewes are more successful at raising their lambs than younger sheep. These observations most likely reflect the high standards of care given to each and every live-born lamb, and successful hand-rearing and fostering practices for surplus and rejected lambs, and show that field data are most likely frequently confounded by a ewe’s over-stretched mothering capability, or husbandry and management factors. Lambs born following dystocia still showed decreased odds of perinatal survival in comparison to the other two delivery groups, but, in contrast to stillbirth alone, there was no longer a significant difference between uneventful delivery and planned caesarean section. This can be explained by the fact that an even slightly pre-term caesarean section carries a risk of prematurity in the delivered lambs (Zoller et al. 2015). While all non-mummified lambs were delivered alive by planned caesarean surgery, sequels of prematurity were the predominant cause of death or reason for euthanasia in lambs delivered this way despite the application of dexamethasone prior to the procedure to aid foetal lung maturation. The perinatal mortality rate of 5.2% following planned caesarean section was, however, still numerically lower but very similar to the 6.7% seen following uneventful vaginal delivery.

In field situations, the effect of dystocia goes far beyond the direct effect of a delayed vaginal delivery causing lamb asphyxia, hypoxic damage or birth injuries. Recent studies have shown that postpartal maternal grooming behaviour and the ewe-lamb bond are negatively affected in dystocic ewes (Regueiro et al. 2021, Redfearn et al. 2023), most likely triggered by exhaustion or pain following a difficult or traumatic parturition. Shortening the birth process by early, pre-planned assistance immediately after the start of the expulsion phase significantly increased lamb viability and had a positive effect on maternal behaviour (Regueiro et al. 2023), and is thus advisable to limit these negative behavioural effects. In the current study, the negative behavioural impact of dystocia was minimized by early intervention, close supervision of neonatal lambs and by the high level of care given to all cases where the ewe-lamb bond or ewe and lamb health appeared compromised. The effects of dystocia on stillbirth and perinatal mortality seen in the current study are therefore largely restricted to direct effects of the duration and difficulty of the vaginal birth process on the lamb and are not confounded by reduced maternal care, as all lambs rejected by their dams or otherwise requiring assistance were given adequate colostrum, shelter and care and were subsequently hand-reared or fostered. It is therefore highly likely that cumulative effects of dystocia on lamb vigour and maternal behaviour will lead to more pronounced differences in perinatal mortality between eutocic and dystocic ewes in field situations, with mismothering and subsequent starvation and exposure further adding to perinatal mortality following dystocia (Regueiro et al. 2021).

Of the four ewe fatalities occurring within 14 days post partum, only one case was birth-related, with a diagnosis of retained foetal membranes, metritis and subsequent peritonitis in a ewe 6 days after elective caesarean section. This death accounted for 0.2% of the studied 576 parturitions. Parturition-related ewe mortality can thus be considered very low in healthy animals under close observation and good management conditions, and this is a standard that should be strived for. In field situations, high periparturient ewe mortality is frequently associated with underlying health issues (Sobiraj 1994, Voigt et al. 2021), inappropriate treatments or poor hygiene (Scott 2005), or limited observation in extensive farming systems (Bruce et al. 2021). Labour shortage or the nature of extensive farming are quite often limiting factors for optimal flock surveillance at lambing time. Efforts have therefore recently been made to use animal-fitted sensor technology and artificial intelligence algorithms to identify and ultimately predict the time of parturition in ewes (Smith et al. 2020, Fogarty et al 2021). While there are currently still limitations to this technique, precision livestock farming approaches hold great promise for improving individual sheep welfare at lambing time.

Even under the given conditions of the operations being performed by inexperienced veterinary students leading to prolonged surgery times, caesarean section was associated with excellent clinical recovery, survival rates and wound healing. The surgical protocol described here is also suitable for operations under field conditions, with the exception that sedation with xylazine is not usually necessary, and the use of clenbuterol in sheep is prohibited in field cases under German law. Its use in this study was thus only possible due to the experimental status of the animals.

The results following caesarean delivery are in accordance with the findings previously reported for ewes undergoing elective caesarean surgery (El-Guindy and El-Ghannam 1973, Thurley 1973, Waage and Wangensteen 2013, Paping et al. 2023). Higher post-surgical mortality in field cases carried out following dystocia (Scott 1989, Sobiraj 1994, Brounts et al. 2004, Voigt et al. 2021) or a high incidence of wound healing disorders seen by some authors (Ennen et al. 2013) are thus attributable to the effects of prolonged labour, contamination of uterine contents by prior delivery attempts, underlying health issues or, possibly, differences in suture material or suture techniques.

Under the circumstances of elective procedures and timely intervention in dystocic ewes, retained placenta can be seen as the most important risk factor for immediate postpartal ewe health as determined by fever episodes, although the majority of these cases were not associated with a compromised general condition. Only seven animals were diagnosed with clinical metritis, precluding reliable statistical analyses, but five of these cases were preceded by retained foetal membranes. One of these animals died after developing subsequent peritonitis. The two remaining metritis cases had undergone a very difficult assisted delivery following simultaneous presentation of multiple lambs but had subsequently cleansed normally.

The high frequency of retained foetal membranes in the cases undergoing near-term elective surgical delivery (29.4%; n=218) by far exceeded the rate seen in dystocic field cases following emergency caesarean section (12.6%; n=206) by the same group of authors (Voigt et al. 2021). Recognized factors predisposing to the condition in cattle include insufficient placental maturation associated with premature delivery or shortened gestations (Sheldon 2019). The premature delivery after an average gestational length of 147.4 days as opposed to 150.2 days for natural deliveries in the same flock is thus an obvious explanation for this high incidence in our study. It is interesting to note that the antepartal application of dexamethasone lead to enhanced placental maturation in cattle, but did not reduce the incidence of retained foetal membranes following induced parturitions in this species (Hartmann et al. 2013). A similar phenomenon is likely in the ovine cases described here although no histological examinations were carried out to assess the process of placental maturation. In contrast to our results, a much lower percentage of placental retention was observed following elective caesarean section by other authors: Waage and Wangensteen (2013) only report a single case of retained placenta in 162 ewes undergoing elective caesarean surgery. The authors however do not provide a case definition, so the time allowed for placental shedding prior to classification as retention may have been longer. Also, elective surgery was carried out much closer to term, with caesarean sections performed on day 146.7 (+ 1.7), which was very similar to the average gestational length of 146.4 days (+ 1.7) in the same flock (Waage and Wangensteen 2013).

A study involving dairy ewes from Greece suggested a four-fold increased risk for retained placenta following assisted deliveries, and a linear increase of risk with litter size (Leontides et al. 2000). We could, however, not find a statistically significant difference in the odds of developing retained placenta between parturitions classified as dystocic or eutocic, or associated with litter size in our data set.