Killing Underweighted Low Viable New-Born Piglets – A First Step to a Reliable and Comprehensive Decision

Lukas Geiping Tierärztliche Hochschule Hannover: Tierarztliche Hochschule Hannover https://orcid.org/0000-0001-7648-402X Maria Hartmann Stiftung Tierärztliche Hochschule Hannover: Stiftung Tierarztliche Hochschule Hannover Lothar Kreienbrock Stiftung Tierärztliche Hochschule Hannover: Stiftung Tierarztliche Hochschule Hannover Elisabeth grosse Beilage (  elisabeth.grosse.beilage@tiho-hannover.de ) Tierärztliche Hochschule Hannover: Tierarztliche Hochschule Hannover

The study was conducted on three commercial pig producing farms and one research farm in Lower Saxony and North Rhine Westphalia from May to October 2020. The sow herds belong to three genetic lines (BundesHybridZuchtProgramm, Topigs, Danish Genetics) frequently used in Germany. In all herds, sows were inseminated with semen from Pietrain boars. In each herd, litters from three consecutive batches were included in the study. The study comprised a total of 529 piglets out of 99 litters (Table 1), ten piglets were excluded from further analysis due to missing data. Approximately one week before the expected farrowing date the sows were moved to conventional farrowing pens equipped with a crate, fully slatted oors and a heated creep area. Herd 1 also used four pens for free farrowing equipped with a partly slatted concrete oor. Farrowing supervision took place during the normal working hours and farrowing induction was not routinely used in any herd. All farms used nurse sows and additional feeders for suckling piglets to ensure adequate milk and feed intake. All sow herds were vaccinated quarterly against porcine reproductive and respiratory syndrome virus, Erysipelothrix rhusiopathiae and parvovirus.
-Piglet management Litters were included 12 to 24 hours after birth (day 1) before cross fostering and before litter equalization was done. Tail docking and teeth grinding were performed on day 2. On day 3 all piglets were treated with an iron preparation and male piglets were surgically castrated according to the German Animal Protection Act (16).

Data collection
For individual identi cation, all piglets were marked with numbered ear tags on day 1.
-Clinical variables evaluated at day 1 Body weight (BW) at birth was the inclusion criterion and piglets with a body weight ≤ 1.0 kg (low weight group, LW) were compared to piglets with a body weight > 1.0 kg (normal weight group, NW)). While all piglets from a selected litter weighing ≤ 1.0 kg were included in the LW, the number of piglets included in the NW was restricted to two piglets per litter. In total the LW comprises 328 piglets and the NW 191 piglets. For further analysis the study and control group were subdivided based on whether they were dead or alive at day 5 (low weight group alive (LWA) and low weight group dead (LWD), normal weight group alive (NWA) and normal weight group dead (NWD)).
The data recorded on day 1 also comprised the sex, the rectal temperature and the crown-rump length (CRL), de ned as the length from the occiput to the base of the tail. The vitality was assessed based on a score from 0 to 3 (Table 2). A score of 0 was given if the piglets showed no signs of reduced vitality. A score of 1 was given for piglets showing one sign of moderately reduced vitality, score 2 was recorded when a piglet showed two signs of moderately reduced vitality and score 3 when one or more signs of severely reduced vitality were diagnosed. Piglets with a score of 3 were categorized as non-viable and were killed immediately by the farmer by blunt force trauma and subsequent bleeding or by the investigator by intravenous injection of a lethal dose of Pentobarbital (Release ® 500 mg/ml WDT, dosage: 450 mg/5 kg BW).
The piglets were also scored for IUGR based on the scheme evaluated by (17,18). A score of 1 (mild IUGR) was given if the piglet showed at least one sign of IUGR and a score of 2 (severe IUGR) when more than one sign was proven ( Table 2). Blood (0.5 mL) was collected from all piglets on day 1 by punctuation of the V. cava cranialis with a 23 G needle and a 1 mL syringe and used for analysis of laboratory measurements to compare these with the survivability of neonatal piglets (Table 3). A part of the blood was immediately analysed on-farm using three different hand-held devices. With the Accutrend® Plus by Roche glucose and lactate concentrations were measured in whole blood as previously described (19,20). Both devices were checked before each batch with control solutions provided by the manufacturer (Accutrend® Control G2; BM-Control G2). The range for glucose concentrations measurable with this device ranges from 20 to 600 mg/dl. Values below this range were classi ed as 19 mg/dl. Lactate concentration was measured from 0.8 to 22 mmol/L; higher values were classi ed as 23 mmol/L. The Hemocue Hb 201 Analyzer was used to measure haemoglobin concentration (21). The measuring scale runs from zero to 15.9 mmol/L.
To assess the amount of colostrum the piglets have ingested, the immunocrit was measured, as previously described (22). This method is a cost e cient and easy way to determine the IgG-content in serum samples. All piglets included in the study that died or had to be killed until day 5 were weighed and collected for a standardized post-mortem examination ( Table 4). Reasons for killing were a vitality of score 3 (Table 2), anomalies or lacerations. The cause of death was recorded (crushing, starvation, infection, anomalies, other) as well as whether the piglet had been killed or died spontaneously. The diagnosis "crushing" was assigned when the piglet had broken bones or when typical internal or external lesions or bleedings were detected. "Starvation" was diagnosed when the piglet was emaciated, and ribs or other prominent bones were easily visible. The diagnosis "infection" included all piglets showing signs of enteritis, pneumonia or arthritis. Under "anomaly" splay legs, blind anus and other congenital malformations incompatible with survival have been summarized. The diagnosis "other" was used when the cause of death could not be ascertained.    Calculation of the minimum sample size required for the study was performed using NCSS-PASS-software for a Wilcoxon rank-sum test comparing two groups (type I error 5%, power 80%). The sample size calculated was 564 piglets.
After general description data was analysed using a linear logistic regression analysis following a backward selection for multivariable modelling. Herds and batches per herd were included as random effects in this model to incorporate the hierarchical structure of the data.

Results
Birth weight, vitality score and IUGR-score were signi cantly associated with a higher chance of mortality (univariate model, Table 6). The other clinical variables crown-rump length (CRL) and rectal temperature (RT) also showed lower values in LW than NW piglets and within both groups, piglets dying until day 5 had lower values compared to survivors (Fig. 1a-c).
The same effects were found in the laboratory measurements glucose and immunocrit but not lactate and haemoglobin (Fig. 2a-d).
Higher scores in vitality (VS) and intrauterine growth retardation (IUGR) were associated with higher mortality until day 5 ( Table 5).
The variables appropriate to predict the death a new-born piglet until day 5 of age were identi ed in a logistic regression model. The nal multivariate model included the clinical parameters: body weight, rectal temperature and the IUGR-and vitality score. Laboratory measurements were not included as they are not accessible to the farmer. For the discussion and completeness of the data another model was used including immunocrit and glucose (see additional le 1). Crown-rump length was highly correlated with body weight at day 1 ( Fig. 3) and, therefore, not considered in the nal model. BW 4, RT 2, VS 0 and IUGR 0 were used as reference values. In the multivariate logistic regression model, a bodyweight < 0.82 kg, a vitality score 1 or 2 and a rectal temperature ≤ 37.5°C were identi ed to have an effect on the prediction of a death of a new-born piglet until day 5 (Table 6).
Based on the adjusted Odds ratios of the results of the logistic model, prognostic values for the probability of a new-born piglet to die until day 5 of age were derived, considering the different combinations of body weight, rectal temperature and vitality score (Table 7).

Discussion
Care of new-born piglets is one of the main animal welfare issues in piglet-producing farms (6, 14,23). While most studies are focused on improvement of piglet survivability (24-26) only a very few publications refer to the requirements for compromised new-born piglets with low chance for survival (2,23,27,28). As the options for intensive care are limited under the standard conditions of piglet production, some farmers chose a pragmatic approach by killing new-born piglets they expect having less economical value (11). However, this practice is neither in accordance with the different animal welfare acts nor accepted by the majority of people in Europe (29). The opposite, also practiced on some farms, ignoring compromised new-born piglets and let them die without or only a little intervention is also not acceptable, as this may result in avoidable pain and/or suffer (30,31). A rst step into a way out of this dilemma might be found in a validated assessment scheme facilitating an estimation of the probability of death of a new-born piglet until ve days of age. This scheme is aimed at giving farmers a clear guideline leading to a comprehensive decision about care or killing of a compromised new-born piglet.
Based on literature (10, 20, 32) a set of clinical and laboratory measurements associated with increased mortality in new-born piglets was selected. The selection was focussed on clinical variables that can be assessed easily by farmers and resulted in body weight, rectal temperature, crown-rump-length, a vitality score and an IUGR score. As clinical variables are often estimated being "subjective" compared to "objective" laboratory parameter, additionally a set of laboratory measurements was selected, expected to be appropriate in predicting the death of a new-born piglet (33)(34)(35).
The farmers were participating voluntarily in the study. The study herds were managed according to the rules of good farming practice, ensuring standard housing conditions and extensive care of the pigs. By the selection of the study herds we tried to exclude an in uence of poor housing or management on the results.
In a logistic regression analysis the clinical variables body weight, rectal temperature and vitality, measured and scored at the day of birth, have been identi ed being associated with death of piglets until day 5. Piglets are born without immunoglobulins and are highly dependent on the early intake of colostrum (3). The subjective measurement of milk intake (abdominal palpation) as a part of the vitality score and the objective measurement of the immunocrit in this study clearly underlined the importance on neonatal survival (for the in uence of the immunocrit in the logistic regression model: see additional le 1).
Since piglets are born with no brown adipose tissue they are only able to maintain their body temperature through liver and muscle glycogen. These reserves only last for 16 to 24 hours (36). Blood glucose levels are described in various studies as highly signi cant to survival (33,34). But the increase of glucose in agony or through catecholamine release in stressful situations aggravates the interpretation (4,37). The univariate model showed a signi cant association between mortality and glucose, but the multivariate model showed no signi cant association (see additional le 1). Rectal temperature proved to be a better indicator of hypoglycaemia and hypothermia and is further easier to assess.
Hypoxia during prolonged births or birth in the last third of a litter leads to an increase of lactate (38, 39). Higher haemoglobin levels at birth make piglets less prone to hypoxia due to the higher oxygen binding capacity (40). No association to survival was seen for both variables in this study.
As a nal outcome the probability of death was estimated for various combinations of scores by means of the estimated regression model parameters. A probability of 94 %, 89 % and 86 % to die until day 5 was determined for the combination rectal temperature ≤ 37.5°C and a vitality score 2 in the body weight groups < 0.82 kg, 0.82-0.95 kg and 0.96-1.39 kg (Table 7).
Body weight and rectal temperature can be easily and quickly measured by the farmer. The vitality score (41) is based on ndings assessing the colour of the skin, the movement and the lling of the abdomen ( Table 2). The vitality score with three items and ndings that need to be assigned to one of only three categories (unaffected, moderate, and severe) is estimated to be appropriate for the application by farmers. Otherwise, a short introduction by the herd attending veterinarian will ensure a reliable application.
The results of this study clearly show that the death of a new-born piglet can be predicted by a simple clinical score applicable for veterinarians and farmers. Nonetheless, it needs to be emphasized the proposed method is not ready to be used in practice. Based on these results veterinarians, farmers and particularly ethicists need to clarify in an extensive process which probability of death will justify the killing of a new-born piglet.
Declarations Figure 1 a-c: Descriptive statistics of metric clinical variables in the low weight group and normal weight group dying or surviving until day 5 of age. Figure 2 a-d: Descriptive statistics of metric laboratory variables in the low weight group and normal weight group dying or surviving until day 5 of age.