Piglet mortality – A parallel comparison between loose-housed and temporarily confined farrowing sows in the same herd

ABSTRACT

In literature, piglet mortality is described as a multifactorial complex influenced by factors as litter size, age and health of the sow, farrowing system, management etc. In this study, a parallel comparison was made between two farrowing systems; a temporarily confined (TC) (farrowing – 3 days after) versus a loose sow (L). On average, 0.4 more pigs per litter survived until weaning if the sow was temporarily confined (TC) compared with being loose (L). Cause of death was recorded according to a strict template. Underweight and crushing was the most common causes. For crushing, during 1–3 days after birth, an interaction between sow age and farrowing system was observed, with differences between farrowing systems only for intermediate-aged and older sows (>parity 2). There were no significant differences between systems regarding farrowing duration or number of stillborn pigs, but a significant increase in farrowing problems was recorded for TC-sows.

KEYWORDS:

• Piglet mortality
• farrowing system
• temporarily confined
• loose sows
• underweight
• crushing

Introduction

Over the years, a considerable number of studies have been performed with the aim of reducing the excessive mortality rate within piglet production. Despite extensive effort in research on reducing piglet mortality by means of breeding, housing, close environment, hygiene, improved management routines etc., high pre-weaning mortality remains a problem in commercial piglet production (Gård & Djurhälsan, 2017a). Piglet mortality is a multifactorial complex and can be explained by a variety of factors, such as the sow and her maternal characteristics (Andersen et al., 2005), the new-born piglet and its vitality (Amdi et al., 2017; Hole et al., 2018), litter size (Pedersen et al., 2006), parity (Marchant et al., 2000), factors in the local environment (housing, temperature, cleanliness) and management routines (Andersen et al., 2007; Muns Vila & Tummaruk, 2016). In the end the proportion of piglets surviving is a result of these factors interacting (Edwards, 2002).

A complete elimination of piglet mortality is probably unattainable. According to Fraser et al. (1995), the pig cannot be compared with other production animal species because the sow produces a large number of offspring, so the mother’s investment in each individual offspring is lower than for other production animals. The competition between piglets at the udder in the suckling stage is strong and only the most vital piglets that are sufficiently powerful and strong in this competition survive (Fraser et al., 1995; Edwards, 2002; Illmann et al., 2018). Fraser et al. (1995) claim that pigs overproduce progeny as a way to increase their vitality in order to explain why farmers are continually wrestling with high pig mortality in practical pig production. Breeding goals for larger litter sizes within pig breeding companies additionally reinforce this struggle. However, for welfare and economic reasons it is important to continue work on reducing piglet mortality. In general, the number of weaned pigs per litter has been increasing to date, but this is mainly because of an increasing number of live-born piglets rather than lower piglet mortality (Gård & Djurhälsan, 2017a).

Since the causes of the early piglet mortality are mainly non-infectious, such as crushing by the sow and/or starvation (Bille et al., 1974; Svendsen et al., 1986; Svendsen, 1992; Edwards, 2002; Pedersen et al., 2006; Westin et al., 2015a) the effect of confining the sow around farrowing or during the whole suckling period has been investigated in several earlier studies. Results from these studies are however contradictory. In studies by Condous et al. (2016), Hales et al. (2014) and Moustsen et al. (2013) , the preweaning mortality was reduced when the sow was confined at farrowing compared to being loose-housed. On the other hand, findings by KilBride et al. (2012), Pedersen et al. (2011) and Weber et al. (2007) showed no differences in preweaning piglet mortality between crated or loose housed sows. KilBride et al. (2012) found a tendency for more crushed piglets when the sow was kept loose at farrowing, but no significant effect on total piglet mortality.

In Sweden, free movement of the sows is stipulated by law (Swedish National Board of Agriculture, 2017) and access to nest-building material prior to farrowing is strongly recommended (Westin and Algers, 2006), as temporarily confinement prevents the sow from performing nest-building behaviour in a satisfactory way. Swedish welfare regulations are strict and in general sows have to be loose-housed during farrowing and lactation. Confinement is only permitted in exceptional cases if the sow e.g. shows aggressive behaviour towards her piglets or the stockperson, or if she crushes many piglets (confinement only allowed during the first seven days) (Swedish National Board of Agriculture, 2017). Furthermore, litter material (in Sweden most often straw) for nest-building behaviour by sows has to be provided (Swedish National Board of Agriculture, 2017). Westin et al. (2015b) found decreased farrowing duration with increased nest-building activity, and Fraser et al. (1997) concluded that the possibility for a sow to move around is an important factor in achieving easy and normal farrowing without farrowing problems. Negative effects of confinement on farrowing duration, stress and plasma oxytocin concentrations have been reported by Thodberg et al. (2002) and Oliviero et al. (2008, 2010). According to Lawrence et al. (1997), opioids inhibit oxytocin in the pig during farrowing, while Algers and Uvnäs-Moberg (2007) concluded that a low plasma oxytocin concentration is associated with prolonged farrowing in sows. Prolonged farrowing may result in reduced vitality and survival of the piglets. Several studies have shown an increased number of stillborn piglets with extended farrowing times (Olsson & Svendsen, 1989; Zaleski & Hacker, 1993; Van Dijk et al., 2005; Feyera et al., 2018). However, a number of studies comparing confined and loose sows before and during farrowing have not been able to find any difference in the number of stillborn pigs (Cronin et al., 2000; Pedersen et al., 2011; Moustsen et al., 2013; Hansen et al., 2017).

A high piglet mortality within Swedish pig production has led to discussions about more powerful mortality-reducing methods. One proposal is to allow temporary confinement of the sow during farrowing and 3 days thereafter. The effect of this on piglet mortality, sows and consumer perceptions of Swedish pig production has been discussed widely among the Swedish public.

The present study was performed in this context, with the aim to compare production results, piglet mortality, farrowing duration and impact on sow and piglet medical treatments and morbidity during farrowing and suckling period in two different farrowing systems: one system designed for a completely loose sow and one system for a temporarily confined sow during, and three days after farrowing. Well-described and implemented management measures were also included in the system concepts. The overall objective was to provide information about the effect on piglet mortality, when having the sow temporary confined compared to loose, in typical Swedish farrowing pens.

Material and methods

Experimental facility and design

The effect of housing at farrowing on piglet mortality, sow morbidity/treatments and piglet morbidity/treatments during the farrowing/suckling period was studied on a piglet production farm in southern Sweden during a period of 16 months (October 2015-January 2017). The farm comprised approximately 130 sows (Norwegian Landrace × Swedish Yorkshire) in a batch farrowing system with seven groups of 18 sows (farrowing interval 22 weeks and 3, 3, 3, 3, 3, 3 and 4 weeks between farrowings). The sows and their piglets were monitored from introduction of the sows into the farrowing pens (approximately 3–4 days before the first sow in the batch was expected to give birth) until weaning of the piglets (at about 32–33 days of age).

In total, 321 farrowings by 176 sows were randomly distributed on the basis of parity number, to one of two parallel housing systems at farrowing: TC- Sow temporarily confined at farrowing and until 3 days after and L- Sow loose at farrowing.

Housing

During mating/insemination and gestation, gilts and sows were group-housed in deep-litter straw systems with individual feeding stalls. Farrowing took place in one of two farrowing sections, each of which comprised one row (9 pens) with farrowing pens for temporarily confined sows (TC) and one row (9 pens) with farrowing pens for loose-housed sows (L).

The total area of the TC and L pens was the same: 6.5 m² (3.35 m × 1.95 m). Approximately 53% of the pen area (1.75 m × 1.95 m) consisted of solid concrete with floor heating and 47% (1.60 m × 1.95 m) of slatted floor (cast iron).

After weaning, the sows were moved to the insemination/gestation section, while the weaned piglets were moved to one of three weaning sections. Wet feeding was used in all sections.

General management and experimental design

Working hours in the pig herd were 06.00–15.00 Monday-Friday and on weekends with farrowings and about 07.00–13.00 during other weekends. Most of the gilts/sows were mated/inseminated with HK Scan Hampshire boar/semen, except a few sows per batch used as recruitment mothers, which were inseminated with Landrace or Yorkshire semen.

The farrowing pens were cleaned twice a day and given 200 g chopped straw on the solid area in the pen (including the creep area of the piglets) after each cleaning (total 400 g per day). In order to permit nest-building, the sows in the TC farrowing system were kept loose as long as possible prior to farrowing. However, if a stockperson judged that a sow would give birth during the coming night, and was able to get a powerful milk jet from a teat of the particular sow, then the sow was confined at the end of the working day. In such cases, when a sow was confined before farrowing, the time of confinement and the time of farrowing (estimate for night-time farrowing) was recorded. If the stockperson had misjudged the farrowing time and the sow did not give birth during the night, no changes were made in the morning (i.e. the confinement of the sow was continued).

Upon confinement of the TC sows, the heating lamp from the piglet creep area was moved to the area behind the sow and extra straw (1 kg) was added (on the cast iron floor under the heating lamp). Similarly, extra straw (1 kg) was given to the L sows before expected farrowing (when there was milk in the udder). However, the extra straw given to the L sows was placed on the solid concrete area in the pen and it was not possible to move the heating lamp from the piglet creep area, where the lamp was protected, since the sow was loose.

Farrowing was not pharmaceutically induced. The time when farrowing began and ended was recorded when personnel were present in the sow unit i.e. ‘actual time’. For sows that farrowed when no personnel were present, the farrowing duration was estimated i.e. ‘estimated time’, based on observations made by a stock-person when leaving in the evening and returning in the morning.

After farrowing, the rectal temperature of the sow was measured and numbers of stillborn and live-born piglets were recorded. Furthermore, the heating lamps in the TC pens were moved back to the piglet creep area. All live-born piglets in the study were tattooed with individual numbers with an ear tattoo equipment in their ears and weighed after birth. If a sow gave birth to more piglets than the number of functional teats of that sow, the personnel were allowed to cross-foster (equalize) piglets within the first two days between litters within the same farrowing system. In both systems (TC and L), the piglets were locked into their creep area during the first three feeding occasions of the sow (0.5–1 h per occasion) after farrowing.

During the first week after farrowing, sows and litters in both systems were given 600 g straw per pen and day. The piglets were given commercial piglet feed + peat without antibiotics from day four in the creep area. Tail docking was not performed. Iron injections were given and male piglets were surgically castrated within 4 days of age. The sows in the TC pens were only temporarily confined during farrowing and 3 days thereafter, and thus all sows in the two systems (TC and L) were kept loose in their farrowing pens from days 4 to 33 after farrowing. Due to the design of the confinement gate, sows in the TC system were somewhat more restricted than sows in the L system when being loose (Figure 1).

Figure 1. Over-view of the two farrowing systems compared in the study. Picture A- TC-pen in which sows were temporarily confound at farrowing + movable heating lamp behind the sow. Picture B- TC-pen in which sows were loose 3 days after farrowing. Picture C- L-pen with loose sows at farrowing and during the whole lactation period.

The total number of stillborn piglets was recorded by the personnel and divided into three categories: mummified, dead ante partum (before farrowing) and dead intra partum (during farrowing). If there was doubt about whether a dead piglet died intra partum or was live-born, a simplified post mortem examination was performed by testing ‘lung in water’ (to determine if the piglet had breathed or not). All live-born piglets that died during the suckling period were recorded with date and cause of death. No post mortem examinations of these live-born piglets were performed, but cause of death was recorded according to the standard template presented in Table 1. In order to pay special attention to the difficult competition situation for underweight piglets in large litters, underweight was defined as a cause of death for all dead piglets with a birth weight ≤ 0.9 kg. To use ‘underweight’ as an under-lying cause of death rather than the actual death cause, means that the results of this study have to be compared with caution in relation to other studies.

Table 1. Template used to classify causes of death of live-born piglets.

Cause of death (recordings in the herd)	Description	Cause of death (after merging death causes in the statistical analyzes)
Underweight at birth	Piglets, ≤900 g at birth, dead due to starvation, crushing or euthanasia since they were unable to cope due to underweight.	Underweight
Starvation	Piglets, dead due to starvation but without signs of underweight, weakness, splay-leg problems or malformation.	Starvation
Crushing	Piglets, which died or had to be euthanized due to crushing of the sow without earlier, recorded disabilities.	Crushed by the sow
Weak or splay-leg at birth	Piglets, dead due to starvation, crushing or euthanasia since they were unable to cope due to weakness or splay-leg problems (>900 grams).	Others
Malformation at birth	Piglets, dead due to starvation, crushing or euthanasia since they were unable to cope due to malformation.	Others
Diarrhea	Piglets, dead due to clear signs of diarrhea.	Others
Joint/claw inflammation	Piglets, which had to be euthanized due to not recovering after treatment of joint/claw inflammation with antibiotics.	Others
Bitten to death	Piglets, which were bitten to death by the sow	Others
Others	Used when not possible to identify any of the other death causes above.	Others

Morbidity and treatments of sows and/or piglets were recorded in the same way as mortality, with date and cause. A coding system for these recordings, developed previously by the veterinarian treating the study herd, was used. A total of 10 different codes were used for morbidity and treatments in sows (Table 2) and six for morbidity/treatments in piglets (Table 3).

Table 2. Template used to classify morbidity/treatments in the sows.

Treatments/morbidity in the sows (recordings in the herd)	Treatments/morbidity in the sows (after merging recordings in the statistical analyzes)
Farrowing problems (assistance needed)	Farrowing problems
MMA (mastitis, metritis, agalactia)	MMA
Failure of milk let down	Others
Udder inflammation (not associated with farrowing)	Others
Joint inflammation	Others
Claw abscess	Others
Lameness	Others
Reduced appetite	Others
Shoulder lesion	Others
Others	Others

Table 3. Template used to classify treatments/morbidity in the piglets.

Treatments/morbidity in the piglets (recordings in the herd)	Treatments/morbidity in the piglets (after merging recordings in the statistical analyzes)
Piglet diarrhea (0–7 days)	Piglet diarrhea
Claw abscess/(joint inflammation)	Claw abscess
Middle ear inflammation/meningitis	Others
Diarrhea during the latter part of the suckling period (>7 days)	Others
Lesions	Others
Others	Others

Statistical analyses

Data on the sow cards for the herd were transferred manually into Microsoft Excel 2007 and then manually doubled-checked against the data recorded in the WinPig production monitoring programme (Gård & Djurhälsan, 2017b) used in the herd.

Piglet mortality and causes of death were classified into three piglet age groups (1–3 days, 4–7 days, >7 days; day 1 = day of birth) depending on when, in relation to birth, the piglets died. In the final calculations only mortality figures for the first 3 days and for the whole preweaning period were compared, since the results showed that no more detailed breakdown was relevant. Due to a low number of recordings for some morbidity/treatment codes used for sows and piglets, these codes were merged into the category ‘others’, as described in Tables 2 and 3. The different morbidity/treatment codes in sows and piglets were treated as binary variables (occurrence or not in the sow or litter).

To adjust for different numbers of live-born piglets produced by the sows, an equalization was performed (if possible) within the first two days. The number of piglets in the litter after this equalization was taken as the number of ‘at risk’ piglets (=live-born + received–transferred).

Statistical analyses were performed using SAS (SAS, 2017), with pen as experimental unit. Number of total born, live-born, ‘at risk’ and weaned piglets and farrowing duration were tested for normality and analyzed with mixed linear model (PROC MIXED). The model had the fixed factors farrowing system (TC/L), sow age category (young (parity 1 + 2), intermediate (parity 3 + 4) and old (parity ≥ 5)) and the interaction between these two factors and the random factor sow.

To adjust for the small (non-significant) numerical difference in number of ‘at risk’ piglets between farrowing systems, this variable was introduced in the model as a covariate when analyzing the number of weaned piglets per pen (=litter). When analyzing farrowing duration, the introduction of live born as a covariate in the model was not significant and did not influence the result, and was therefore not included in the analysis.

Stillborn piglets and piglet mortality were not normally distributed and were analyzed with PROC GLIMMIX. A logistic regression model with response equal to events/‘at risk’ was used. The explanatory variables in this model were farrowing system (TC/L), sow age category (young (parity 1 + 2), intermediate (parity 3 + 4) and old (parity ≥ 5)) and the interaction between these two factors. Sow was a repeated factor in this model with correlation structure compound symmetry but, since the response was also influenced by the number of piglets in the litter, ‘at risk’ was included as a covariate to give satisfactory goodness of fit. To compare differences between levels, Tukey’s post-hoc test was used with p < 0.05 considered significantly different.

To analyse the morbidity and treatments in sows and piglets, Fisher’s exact test was used.

Results

During the 16-month study period, data were gathered for a total of 318 litters. From these litters, a total of 4564 ‘at risk’ piglets and 3609 weaned piglets were recorded. Overall calculations of production and mortality (whole preweaning period), with pen/litter as the experimental unit, are presented in Table 4 and Figure 2. In Figure 2, cause of mortality is presented in relation to number of piglets ‘at risk’ in the litter. Average overall preweaning mortality (of live-born piglets) was 20.9%, ranging from 16.3% for young sows (gilts and parity 2 sows) to 20.6% for intermediate sows (parity 3 + 4) and 28.5% for old sows (parity ≥ 5). The most common reasons why piglets died before weaning were underweight (1.1 piglet per litter) and crushing by the sow (1.1 piglet per litter) (Table 4). Distribution of parity number was similar in the two treatments.

Figure 2. Number of dead piglets per litter within different litter sizes and with specified causes of death.

Table 4. Overview of litters, production results and mortality in the study.

Age of sow, parity	Total	1 + 2	3 + 4	≥5
No. of litters in the study, total	321
No. of litters deleted from the study	3^a
No. of litters left in the study	318	127	120	71
No. of piglets ‘at risk’	4564	1797	1695	1072
No. per litter
Total born	15.1	14.4	15.0	16.8
Stillborn	0.8	0.6	0.9	1.1
Live-born	14.3	13.9	14.1	15.6
‘At risk’	14.3	14.1	14.1	15.1
Dead before weaning	3.0	2.3	2.9	4.3
Weaned	11.3	11.8	11.2	10.8
Causes of death, no. of piglets per litter
Underweight	1.1	0.8	0.9	1.9
Crushed by the sow	1.1	0.8	1.2	1.4
Starvation	0.2	0.2	0.3	0.3
Other	0.6	0.6	0.5	0.7

^aOne sow died during farrowing and two sows farrowed before confinement in system TC.

More piglets died in larger than in smaller litters (Figure 2) (Spearman correlation p < 0.001). The number of piglets that died due to underweight or crushing by the sow was 1.3, 2.5 and 3.6 for litters with 13, 15 and 17 piglets ‘at risk’, respectively (Figure 2). A similar relation was seen for starvation and other causes of death (Figure 2).

Comparison of the two farrowing systems revealed no significant difference in numbers of total born, live-born or ‘at risk’ piglets between the systems (Table 5). However, the number of weaned piglets was significantly higher in the system with temporarily confined sows than in the system with loose sows. Age category (in general older sows performed worse) was a significant factor for most of the variables listed in Table 5. In the model for weaned piglets, ‘at risk’ was used as a covariate (p < 0.001).

Table 5. Number of total born, live born, ‘at risk’, dead before weaning and weaned (mean ± std).

	Farrowing system		p-value
	Temporarily confined (TC)	Loose (L)	Farrowing system	Age category
No. of litters	157	161
No. per litter
- total born	15.3 ± 3.9	15.0 ± 3.8	0.23	0.008**
- stillborn	0.8 ± 1.2	0.8 ± 1.3	See Table 6
- live-born	14.5 ± 3.6	14.2 ± 3.4	0.20	0.03*
- ‘at risk’^a	14.4 ± 2.7	14.3 ± 2.5	0.51	0.07
- dead before weaning	2.8 ± 2.4	3.8 ± 2.8	See Table 6
- weaned	11.6 ± 1.9	11.1 ± 2.1	0.03*	<0.001***

^aTransfer of piglets was always made within system but sometimes to/from sows not in the study.

Effect of farrowing system and age category; in all tests in the table, the interaction was not significant.

The number of stillborn piglets was 0.8 piglets per pen, with no difference between the TC and L farrowing systems. Most of the preweaning mortality (54%) occurred 1–3 days after birth (520 out of totally 955 dead piglets) (Table 6). The probability of mortality depended on the number of piglets at risk, and therefore this variable was included as a covariate in the model. In total, ‘underweight’ and ‘crushed by the sow’ were the most common causes of death. The percentage of piglets that died due to underweight day 1–3 was greater in the L system (Table 6). Furthermore, a significant interaction between farrowing system and age category was found for the cause of death ‘crushed by the sow’, at day 1–3 and for the whole preweaning period (Table 6).

Table 6. Preweaning mortality. Cause of death, age and percent of piglets per litter, which died before weaning (lsmeans ± SE).

	Farrowing system		p-value
	Temporarily confined (TC)	Loose (L)	Farrowing system	Age category
No. of litters	157	161
Stillborn, % of total born piglets per litter	4.0 ± 0.6	4.7 ± 0.6	0.26	0.13
Preweaning mortality, % of ‘at risk' piglets per litter
Day 1–3
- underweight	4.1 ± 0.5	5.3 ± 0.6	0.04*	0.04*
- crushed by the sow	2.6 ± 0.4	5.2 ± 0.6	Interaction
parity 1 + 2	2.8 ± 0.6	3.3 ± 0.7	0.57
parity 3 + 4	1.9 ± 0.6	6.6 ± 1.0	<0.001***
parity ≥5	3.3 ± 0.8	6.5 ± 1.2	0.03*
- starvation^a	-	-
- others	0.4 ± 0.2	0.8 ± 0.3	0.12	0.30
Whole preweaning period
- underweight	5.9 ± 0.7	6.5 ± 0.7	0.40	0.002**
- crushed by the sow	5.8 ± 0.6	7.6 ± 0.7	Interaction
parity 1 + 2	5.6 ± 0.9	5.2 ± 0.9	0.77
parity 3 + 4	5.0 ± 0.9	10.1 ± 1.2	0.001***
parity ≥5	7.0 ± 1.2	8.6 ± 1.4	0.39
- starvation	1.6 ± 0.4	1.5 ± 0.4	0.77	0.12
- others	3.7 ± 0.6	4.0 ± 0.6	0.75	0.84

^aThe number of piglets starved to death day 1–3 is not enough to find a suitable model.

Effect of farrowing system and age category; if only p-values for main effects are given, the interaction was not significant.

Age category was significant for the causes of death ‘underweight, day 1–3’, and ‘underweight, whole preweaning period’ (Table 6). Mortality due to underweight was significantly higher in older sow litters.

The interactions between farrowing system and age category for the causes of death ‘crushed by the sow, day 1–3’ and ‘crushed by the sow, whole preweaning period’ are described in detail in Table 6. For ‘crushed by the sow, day 1–3’ a statistically significant difference between farrowing systems was found for intermediate-aged and old sows, but not for young sows. For ‘crushed by the sow, whole preweaning period’ a statistically significant difference between farrowing systems was found for intermediate-aged sows, but not for young and old sows.

Since personnel were not present at all farrowings, the actual farrowing duration (notes of both start and end of farrowing) was only recorded for 95 out of 318 litters (30%) (Table 7). If possible, farrowing duration was estimated when personnel were not in place during the whole farrowing. Such recordings were made for 59% of all farrowings. For the remaining farrowings (11%), no farrowing time was recorded at all and therefore was not included in any analysis. As can be seen in Table 7, there was no significant difference between farrowing systems or age categories for either actual or estimated farrowing duration.

Table 7. Actual and estimated farrowing duration (h) (mean ± SE).

	Farrowing system		p-value
	Temporary confined (TC)	Loose (L)	Farrowing system	Age category
Actual time registration
No. litters	49	46
Farrowing duration, h	4.5 ± 0.2	4.2 ± 0.2	0.32	0.92
Estimated time registration
No. litters	91	98
Farrowing duration, h	4.8 ± 0.1	4.8 ± 0.1	0.85	0.45

Effect of farrowing system and age category.

As time was recorded, when TC sows were confined, it was possible to calculate the time each sow was confined before farrowing. Sixteen percentage (22 sows), of the TC sows were confined less than one hour before farrowing and another 50% (70 sows) farrowed during evening/night-time when confined at 15 o’clock when stockperson left the stable. The remaining sows (34% = 48 sows) were confined more than 15 h before farrowing.

When comparing confinement time with actual farrowing time it was found that only 49 of the 140 TC-sows had recordings of actual farrowing time. For 16 sows, confined less than 15 h before farrowing, the average farrowing time was 4.9 h compared to 4.3 h per farrowing for 33 sows confined more than 15 h before farrowing.

Treatments and morbidity in sows and piglets are presented in Table 8. Significantly more farrowing problems, were recorded in the TC system (4.5%) compared with the L system (0.6%) (p = 0.03).

Table 8. Morbidity/treatments in sows and piglets. Percent of pens with morbidity/treatments.

	Farrowing system		p-value
	Temporarily confined (TC)	Loose	Farrowing system	Age category
No. of litters	157	161
No. of special notes	2^a	0
Morbidity/treatments in sows, % of pens
- farrowing problems	4.5	0.6	0.03*	0.48
- MMA	4.5	9.3	0.12	0.01*
- others	1.9	5.0	0.22	0.18
Morbidity/treatments in pigs, % of pens
- piglet diarrhoea	8.9	10.6	0.45	0.41
- claw abcess	48.4	55.9	0.31	0.03*
- others	9.6	16.8	0.07	0.36

^aTwo TC sows had special notes on their sow journals of ‘restless at confinement’.

Effect of farrowing system and age category. Due to the small number of events, this was analyzed with Fisher’s exact test.

Discussion

The choice between having the sow temporarily confined or loose at farrowing may influence production variables such as piglet mortality, farrowing duration and farrowing problems. In the present study, significantly more piglets were weaned in pens with temporarily confined sows than in pens with loose-housed sows. This result was partly explained by how many piglets were crushed by the sow. However, for piglet crushing as a cause of death, there was an interaction between farrowing system and age category of the sow, with no difference between farrowing system for gilts and young sows (parity ≤ 2) but with a difference for intermediate-aged and older sows for crushing day 1–3. There are two possible explanations: (i) maternal responsiveness can be expected to decrease with increased parity number, due to older sows being heavier and clumsier and having more health problems, e.g. claw/leg problems and teat damage, and/or (ii) older sows have larger litters and more underweight piglets. For piglet crushing during the whole preweaning period there was a significant difference between farrowing system only for intermediate-aged sows since the piglet mortality increased slightly after the confinement period for older sows. In the present study, the mortality rate ranged from 16.3% for gilts and young sows (parity ≤2) to 20.6% for intermediate-aged sows (parity 3 + 4) and was as high as 28.5% for older sows (parity ≥5).

Hales et al. (2014) suggest that contradictory results of the effect of sow confinement on preweaning mortality in different studies might be explained by the level of live-born piglets per litter. The number of live-born per litter in the present study was 14.3, which is somewhat lower than in studies by Hales et al. (2014) and Moustsen et al. (2013) but higher than in the studies by KilBride et al. (2012) and Weber et al. (2007). The number of live-born piglets in the present study are in line with the Swedish overall average of 14.0 live-born pigs per litter reported by Gård and Djurhälsan (2017c). However, the preweaning mortality in the present study (19.4%) was higher than the Swedish overall average of 17.1% for 2016 (Gård & Djurhälsan, 2017c). This discrepancy might be explained by factors such as herd size, working hours, management restrictions due to the study in the experimental herd as well as the proportion of gilt litters in the study. As mentioned in the introduction, piglet mortality is known to be a multifactorial complex rather than being solely influenced by farrowing system. Management is another important factor (Andersen et al., 2007, 2009; Rosvold et al., 2017). Management procedures, such as drying and warming immediately after birth, litter equalization, use of nurse sows and/or supplemental milkers are important factors in reducing preweaning mortality. Furthermore, having well-trained personnel with a well-developed eye for the animals is important to prevent piglet mortality. In the present study, use of management procedures to reduce piglet mortality was limited, partly due to restrictions when performing the study. Litter equalization was only permitted within each farrowing system and nurse sows were not used. Since the herd was rather small, only nine sows farrowed per batch in each system and farrowing was not induced pharmaceutically. This strongly limited the possibilities to even out litter size. In addition, only one permanent stockperson, not living on the farm and only working part-time, was responsible for the management. Therefore, 24-h surveillance of farrowings was not possible and trainees or replacement personnel had to be used during weekends and holidays. These are all factors that may have influenced the piglet mortality in the herd. However, some management procedures for better piglet survival were performed. The piglets were locked into their creep area during the first three feeding occasions of the sow (0.5–1 h per occasion) after farrowing and the TC system had a movable heating lamp behind the sow during farrowing. This management procedure is possible to use together with a confined sow and was therefore included in the TC concept. A movable heating lamp behind the confined sow was first used in Swedish studies on piglet mortality in the late 1980s (Svendsen et al., 1986). Those studies identified a positive effect of having a heat lamp placed behind the confined sow during farrowing, to enhance piglet survival. This expected positive effect might have a confounding effect on farrowing system.

The proportion of gilt litters in the present study was lower (14%) than the Swedish average (23.2%) (Gård & Djurhälsan, 2017b). In general, gilt litters have lower piglet mortality than sow litters (Hales et al., 2014 ). The lower proportion of gilt litters in the present study is explained by the fact that the study herd produces its own recruitment gilts and some of these gilts farrowed outside the batches and therefore were not included in the study.

More than half (54%) of the preweaning mortality recorded, occurred within the first three days after farrowing and 81% occurred within the first week after farrowing. This is a typical pattern of piglet mortality (Bille et al., 1974; Svendsen and Bille, 1981; Marchant et al., 2000; Pedersen et al., 2006; Moustsen et al., 2013). Crushing by the sow and/or starvation are often concluded to be major causes of preweaning mortality (Nielsen et al., 1974; Svendsen et al., 1986; Marchant et al., 2000; Strange et al., 2013; Westin et al., 2015a; Pandolfi et al., 2017). However, the occurrence of death causes depends on how death causes are defined. In the present study, underweight, defined as mortality of piglets with birth weight ≤0.9 kg, proved to be a common cause in addition to crushing. This was due to that we devoted particular attention to underweight piglets and defined underweight piglets as a certain mortality group, even though the actual cause of death probably was starvation and hunger, sometimes also resulting in crushing by the sow. It is well-known that the competition between siblings in large litters is tough and that smaller, weaker and later-born pigs are clearly handicapped in this competition (Milligan et al., 2002; Andersen et al., 2011). The number of underweight piglets increases with increasing litter size (Quesnel et al., 2008; Rutherford et al., 2013), due to limits in the number of foetuses which can develop optimally in the sow uterus (limited uterine capacity). Therefore, it is not generally desirable to have too many piglets born in a litter. However, the genetic selection for increasingly large litters in modern pig breeding generates an increase not only in the total number of piglets born, but also in the within-litter variation in birth weight and in the number of underweight piglets (Quiniou et al., 2002; Wolf et al., 2008). Underweight piglets have higher mortality (Milligan et al., 2002; Baxter et al., 2008) and require more work by personnel in litter equalization and use of nurse sows in order to survive (Muns Vila & Tummaruk, 2016). In the present study, the mortality due to underweight was 1.6% in litters with <12 live-born piglets, compared with 15% in litters with ≥18 piglets. There has also been speculation that crushing of piglets in large litters is not always accidental, but is a deliberate way for the sow to protect herself and reduce her maternal investment (Andersen et al., 2005). Thus, the continuing quest for ever-larger litters can be questioned.

In the present study, no significant difference in farrowing duration was seen between temporarily confined and loose sows. However, the actual farrowing duration was only recorded for a limited number of the farrowings in the study (95 out of 318 farrowings). Prolonged farrowing duration has negative effects, since it has been shown to correlate with increased number of stillborn piglets (Olsson & Svendsen 1989; Zaleski & Hacker. 1993; Van Dijk et al., 2005; Oliviero et al., 2010). Stillborn pigs generally constitute about 4–8% of the total number born (Nielsen et al., 1974; English & Edwards, 1996), but figures above 10% of total number of piglets born have also been reported (Moustsen et al., 2013; Strange et al., 2013; Hales et al., 2014). In the present study, there were 0.82 stillborn piglets per litter (0.12 mummified foetuses, 0.68 stillborn ante partum (excluding mummified) and 0.02 stillborn intra partum). This is in average 5.3% stillborn and no significant difference was found between the loose and temporarily confined farrowing systems. Nevertheless, there were significantly more records of farrowing problems for the temporarily confined (TC) system. Similar results have been reported by Cronin et al. (2000). Therefore, it can be speculated that farrowing problems is likely to increase when sows are temporarily confined even though no prolonged farrowing duration was found in this study.

Conclusion

In literature, piglet mortality is described as a multifactorial complex influenced by a number of factors, such as litter size, age and health of the sow, farrowing system, management regime etc. In this study, a parallel comparison was made between two farrowing pen systems in the same herd. On average, 0.4 more pigs per litter survived until weaning if the sow was temporarily confined (TC) at farrowing and for three days after, compared with being left loose (L). However, for piglet crushing by the sow during days 1–3, an interaction between sow age and farrowing system was observed, with differences in piglet mortality between farrowing systems (less crushing in TC-pens) only for intermediate-aged and older sows (>parity 2). There were no significant differences between farrowing systems regarding farrowing duration or number of stillborn pigs, but an increase in farrowing problems was recorded for sows temporarily confined at farrowing.

Acknowledgements

This work was supported by Stiftelsen Lantbruksforskning (V1450008) and Partnership Alnarp (Project 740). The competent assistance of Ida Berglund and Lisa Persson at the research facility is much appreciated.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Anne-Charlotte Olsson http://orcid.org/0000-0002-4713-479X

Jos Botermans http://orcid.org/0000-0003-1155-8328

Additional information

Funding

This work was supported by Stiftelsen Lantbruksforskning (V1450008) and Partnership Alnarp (Project 740).