Sow mortality includes sows found dead. However, sows euthanised on farm due to trauma or disease are generally also included in studies on mortality. Both of these two kinds of unplanned removal lead to urgent need for replacement gilts, loss of income from slaughter, and an extra cost for destruction of the carcasses. Besides the loss in production, there is a risk that the sows are suffering from pain during their last days alive. Annual mortality rates reported previously, mainly including sows found dead, varied from 3.4% to 6.9% of sows in production 123. Risk factors for sow mortality have been identified. Higher mortality was reported during summer months 245, both during the days before expected farrowing and the days just after farrowing 467.

In two recent studies, from Denmark and USA, the proportion euthanised and found dead were of equal magnitude 89 whereas a Swedish study reported more than two times as many euthanised sows as found dead 10. The primary findings at post mortem examination of sows that die in the herd vary between studies. Heart failure was the most common finding in two Canadian studies 411 whereas locomotory related findings was reported as most common in others 912.

This study is a part of a larger investigation of sow removal in Swedish commercial herds. Removal pattern 10 and factors affecting length of productive life 13 have been investigated. The aim of this study was to get information on post mortem diagnosis of sows euthanised or found dead, i.e. to obtain the proper diagnoses. Moreover, the study was to evaluate the association between the clinical symptoms observed on farm and the post mortem findings.

This study was based on material collected from a sow pool with 2200 crossbred Landrace × Yorkshire sows in the south central part of Sweden. The sow pool had a central unit supplying 13 satellite units with pregnant sows within a leasing system. In the central unit the newly weaned sows were housed in groups of 50 on deep straw bedding for one week. During this period oestrus was checked twice daily starting three days after weaning and artificial insemination was used. During the first eight weeks of pregnancy the sows were kept in smaller groups (9–15 sows per pen) on concrete/partially slatted floor with access to straw. Pregnancy check was performed twice with ultrasound scan, at 4 and 8 weeks. After the second check, pregnant sows were moved to large pens (50 sows per pen) with deep straw or peat bedding.

Three weeks before expected farrowing, sows were transported to satellite units where they on arrival were housed in groups on deep straw bedding. A couple of days before expected farrowing, the sows were moved into the individual farrowing pens where they stayed with their litter during the lactation period (approximately 5 weeks). After weaning, sows were transported back to the central unit either to be mated for the next reproductive cycle or to be sent for slaughter. All satellite units were located within a one-hour drive from the central unit and sows were transported in a two-storey ventilated lorry.

At the site of the central unit there was also a multiplying unit with 250 purebred Yorkshire and Landrace sows producing crossbred gilts for the sow pool. The routines in this multiplying unit were the same as in the sow pool itself as described above, except that these sows stayed at the same location throughout the reproductive cycle. Growing gilts were housed in pens with concrete/partially slatted floors. The feed in the sow pool was based on home produced grain (mainly barley) supplemented with commercial concentrate. In the central unit, the multiplying unit and in ten of the satellite units the feed was given in liquid form, while in the remaining four satellite units dry feed was given.

During the 32-week period, 816 sows (n = 709) and gilts (n = 107) were removed from the sow pool including the multiplying unit. Average parity number at removal was 4.1. Of the removed sows/gilts, 3.9% were found dead, 12.0% were euthanised and the rest were sent to slaughter. This corresponds to an annual removal rate of sows at 46.7% of sows in production.

Of 32 sows/gilts found dead 17 (53%) were post mortem examined, and of 98 sows euthanised 79 (81%) were examined (Table 2). The distribution of the 96 sows/gilts examined was evaluated against all 130 sows/gilts found dead or euthanised during the period studied regarding parity number, season, reproductive stage and observed symptoms (data not shown). The 96 sows/gilts post mortem examined were considered to be a representative sample of the 130 (results not shown).

The proportion of sows/gilts being medically treated a short time before death was higher among animals being euthanised (58%) than for animals found dead (29%), see Table 2.

Of the 96 post mortem examined carcasses, 70 were from sows and 26 from gilts. The average parity number at removal for the sows/gilts was 2.8 for those found dead and 2.1 for those euthanised (Table 2). The highest number of euthanised and found dead were found in parity 0 (gilts), see Figure 1. The distribution of sows found dead did not deviate from the parity number distribution in the herd (p < 0.7), whereas the distribution of sows/gilts euthanised tended to deviate from the parity number distribution in the herd (p < 0.08). Examined sows had a higher proportion euthanized in parity 1 and 2 but a lower proportion in parity 7 than expected. Most sows/gilts were found dead in May whereas the number of sows/gilts euthanised was high in February, March, July and August (Figure 2). The distribution of sows found dead did not deviate from the expected even distribution when grouped into 3-month periods (p < 1.0), whereas the distribution of sows/gilts euthanised deviated from the expected even distribution during the period (p < 0.03).

Table 3 shows the distribution of sows/gilts euthanised or found dead according to reproductive stage. The main proportion of the examined sows/gilts were in the period = 28 d of gestation at death (37.5%), followed by sows in the weaning to next service period (24.0%), Table 3. In all the reproductive stages shown in Table 3, a higher proportion of sows/gilts was euthanised than found dead. The number of animals found dead was highest among sows/gilts in the period = 28 d of gestation (9 animals). Of these, 4 died after 100 d of gestation. The incidence rate ratio (IR) for sows euthanised or found dead during the two gestation periods and during the lactating period was approximately 1. However, the IR during weaning to service period was about 9 times higher than in the two gestation periods and the lactation period.

The PAD categories are shown in Table 4. Most sows/gilts had disease related diagnoses. Arthritis was the single most common finding (36.4%), found in euthanised animals with average parity number 1.9 (Table 4). Almost all arthritis cases (34 out of 35) were chronic, purulent type and commonly found in elbow, stifle and/or shoulder. In 16 cases the arthritis was polyarthritic and in 14 cases polyarthritic and periarthritis. In 5 cases of purulent arthritis, samples were taken for bacteriological culture and 3 were found negative and 2 positive (Streptococcus equisimilis and Arcanobacterium pyogenes).

The incidence of osteochondrosis was 13.5%, found in animals with average parity number 2.1 (Table 4). All animals with PAD osteochondrosis were euthanised. Among sows/gilts found dead, circulatory/cardiac failure and trauma related injuries were common main findings. Circulatory/cardiac failure was characterized pathoanatomically by an acute pulmonary oedema and passive congestion of lungs and liver with absence of gross and microscopic findings suggestive of other diseases. Three of the four animals that died due to circulatory/cardiac failure were gilts. The highest average parity number was found among animals with the PAD abscess in spinal cord (3.7), see Table 4.

The herd staff recorded the clinical symptoms before death of the sows/gilts that were euthanised or found dead. Table 5 shows the association between the clinical observations and the PAD categories. More than one clinical symptom could be reported per individual, but among sows found dead, 71% were found dead without any previous symptom. Lameness was the most common observed symptom for the sows/gilts with the primary findings fracture, arthritis and osteochondrosis. Notably, according to the clinical observations, fracture was expected to be found in 43% of the sows with the primary post mortem finding arthritis.

Sows/gilts that got the primary finding trauma miscellaneous, circulatory/cardiac failure or unknown were in most cases found dead without any previous observed symptoms. Sitting (i.e. unable to stand up with the hind legs) was one commonly observed symptom for sows with PAD category abscess in spinal cord or heart/lung inflammatory reaction.

The pathological-anatomical findings, including most of the incidental findings, from the 96 examinations are listed in Table 6. The most common finding was arthritis (44.8%), followed by one or more abscesses (38.5%) and teeth injuries (31.0%). Teeth injuries were in approximately 50% of the cases found in the incicives.

This study was based on material from one herd, but the proportion of sows/gilts found dead and euthanised agrees with the findings in a larger study based on 21 Swedish piglet producing herds 10. The high proportion of euthanised sows/gilts in both studies (12% and 10.5%) is partly due to the animal welfare legislation in Sweden, which states that only sows and gilts in normal body condition and without lameness are allowed to be transported to slaughter.

The proportion euthanised sows of all removed sows/gilts was in the present study about the same as that in a Danish study (10%) 8. However the proportion found dead was markedly lower (4%) in the present study than in the Danish study (11%). Moreover the proportion of sows/gilts found dead of sows in production (2.1%) was lower than in previous studies from other countries 135. The lower mortality found in the present study could be due to differences in housing, management system and climate. The high number of animals euthanised or found dead in low parity numbers corresponds to the age structure with many young animals in the herd. Other studies have shown higher mortality risk in higher parity numbers 513.

Most of the examined sows/gilts (79 of 96) in the present study were euthanised. Other studies of post mortem examined sows have had more equal proportion found dead and euthanised 89. The variation in proportion found dead versus euthanised also influences the most common PAD at post mortem examination. The most common reason for euthanasia was lameness which agrees with a previous report 14. It has earlier been reported that concrete/slatted floor increases the incidence of lameness 15 and in the present study sows/gilts were during a part of their gestation housed on concrete floor in groups.

In the present study the most common PAD was arthritis causing lameness resulting in the high proportion of euthanised sows/gilts. The high proportion of arthritis agrees with other studies on sow mortality 8916. In the present study, all arthritis cases except one were purulent and of chronic character. In nearly half of the cases more than one joint was affected, i.e. polyarthritic. The cause of these infectious arthritis cases was likely an initial trauma, e.g. due to group-housing (mounting during oestrus or rank order fighting) and secondary infection with bacterial infection with bacterial spread to multiple joints. The exact aetiology of these arthritis cases is questionable because of the lesions' chronic character. In the present study, the 35 arthritis cases were from relatively young animals (average parity number 1.9) and 10 were from gilts, which is remarkable. The background for this needs to be further studied.

Another common cause of lameness or leg weakness is osteochondrosis 17. In present study osteochondrosis was the second most common primary diagnosis (all euthanised) and was not only found among young sows but also in older sows. Heredity plays a significant role for the development of osteochondrosis 1718. Since 1988, osteochondrosis has been included in the Swedish breeding evaluation. The finding that osteochondrosis was common among the post mortem examined sows/gilts, i.e. is still a significant problem, indicates the importance of increasing the selection pressure on constitution and osteochondrois. Besides genetic selection, feeding intensity 19, growth rate 18 and mechanical stress influence the incidence of osteochondrosis. In the present study, fracture was a common primary diagnosis in sows euthanised in the period weaning to service. Since the herd after weaning kept the sows loose-housed in large pens on deep straw bedding in the breeding area, the fractures probably occurred when sows in oestrus mounted each other or fought to establish rank order.

The proportion of animals with circulatory/cardiac failures in the present study was lower than in two Canadian studies 24 reporting circulatory failure as the most common primary finding for sow mortality. In the present study, circulatory/cardiac failure accounted for death in approximately 25% of the found dead post mortem examined sows, and among those, 3 out of 4 were non-mated gilts (parity 0). The probable cause of circulatory/cardiac failure in the present study may have been stress due to grouping. In the Canadian studies the average age for sows diagnosed with circulatory failure was higher, 2.3 years 2 and 4.1 parities 4, respectively. In one of these studies 2 circulatory failure was associated with heat stress. The lack of circulatory/cardiac failures due to heat stress in older sows in the present study may be due to a cold Swedish summer in 2006. The incidence of sows with gastric ulcers found in the present study was low compared with other studies 8920. In the present study, sows had access to straw and straw has been reported to decrease the risk of mortality due to gastrointestinal disorders 12.

Abscess in the spinal cord accounted for 7% of the PAD. In other studies this diagnosis has not been reported. An association between tail biting and carcass abscesses have been reported 21. Although tail status of the carcasses was not recorded, clinical observations do not indicate problem with tail biting in the herd. In the present study no sow had a primary finding related to urinary or reproductive organs, which differs from other studies 91220. This may be due to the advantages of the loose housing system since urinary tract disease has been reported to be more common in sows tethered or kept in individual stalls 22 due to ascending infections. The lower incidence of reproductive problems in the present study compared with other studies may be due to the same reason 22. The factors behind might be that loose-housed sows get better physical condition than individual stalled/tethered sows, which may result in fewer problems at farrowing, e.g. secondary uterine inertia (laber).

In the present study, the herd staff recorded the clinical symptoms before death of the sows/gilts that were euthanised or found dead. For sows/gilts with PAD fracture the herd staff observation was correct in 60% of the cases. However, in 43% of the sows/gilts with the PAD arthritis, the herd staff predicted a fracture. This is interesting since it shows that a significant proportion of the sows with arthritis may incorrectly, in the recording of removal reasons, be reported as having fractures. In an earlier study of removal reasons 10, 3.3% of all removed sows (including animals sent to slaughter) had the removal reason "leg fracture". The results from the present study suggest that the incidence of fracture in our earlier study may in fact be lower than reported. This also shows the importance of post mortem examinations to get correct diagnosis. Moreover, knowledge about the aetiology is important for prophylactic measures i.e. for improvements in the herd.

According to the pathological-anatomical findings, including most of the incidental findings, teeth injuries were recorded for 31% of the examined carcasses. This is lower than in two other studies, where 85% had lesions 23 and 42.5% had severe teeth wear 24. In the present study teeth injuries were often found in the incicives and may have arisen when the sows chewed at steel tubes of the feeding stalls while waiting for the feed. This biting is a common behaviour that often starts when the pigs can hear the feed coming. These results on teeth indicate that teeth injuries may be a problem for sows, influencing their well being and should be further investigated.

The results from present study show the importance of post mortem examination to obtain the proper diagnoses for sows found dead or being euthanised. The finding that most of the post mortem examined sows had arthritis as main finding needs to be confirmed on a larger number of animals and herds. If this is a general problem, further investigation is needed to find out more about the aetiology so prophylactic measures can be implemented.

The authors declare that they have no competing interests.

LE participated in the design of the study, evaluated the post mortem findings, performed the descriptive statistics and drafted the manuscript. LES participated in the design of the study, evaluated the post mortem findings and helped to draft the manuscript. NL applied for funding of the project, participated in the design of the study and helped to draft the manuscript. KB participated in the design of the study, evaluated the post mortem findings and helped to draft the manuscript. KA participated in the design of the study and helped to draft the manuscript. AMD participated in the design of the study, evaluated the post mortem findings and contributed significantly to the manuscript. All authors read and approved the final manuscript.