Comparison of primary closure of incisional hernias in horses with and without the use of prosthetic mesh support


Present address: Edisto Equine Clinic 7796 White Point Road, Yonges Island, South Carolina 29449, USA. Email:


Reasons for performing study: Repair of incisional hernias in horses has been described previously; however, this report describes the outcome of primary closure of incisional hernias in a large number of horses and compares these results with those of mesh implantation.

Objective: To report the perioperative care, complications and long-term outcome of primary closure of incisional hernias in horses and to compare these results with a second population of horses in which prosthetic mesh was used.

Methods: Medical records of horses undergoing an incisional herniorrhaphy between 1998 and 2009 were reviewed. Information obtained included case details, factors from the initial surgery that contributed to the hernia formation, method of hernia repair and outcome. Comparisons between horses with and without mesh were made using logistic regression.

Results: Thirty-eight horses with primary closure and 9 horses with mesh implantation met inclusion criteria. Long-term follow-up for cases in which a mesh was not used was available for 25 cases; of these, 21 horses (84%) had a normal cosmetic appearance and 4 (16%) had a visible defect. There was no significant difference between the 2 repair methods in terms of age, sex, breed, weight, size of the hernia, number of defects, timing of the repair or cosmetic outcome. Horses in which a mesh was used had significantly longer duration of surgery and hospitalisation, and were significantly more likely to develop post operative complications while having a longer duration of convalescence prior to return to use.

Conclusions: Primary apposition of incisional hernias in horses without the use of mesh support appears to result in a good cosmetic outcome while avoiding the complications associated with mesh implantation in this population of horses.

Potential relevance: Surgical time, duration of hospitalisation, and post operative complications may be reduced by using this technique of primary repair and avoiding mesh implantation.


Incisional hernias can develop following abdominal surgery in all species with an incidence reported in the horse of 8.1% (Mair and Smith 2005) to 16% (Gibson et al. 1989) within 4 months of surgery. Risk factors for development of incisional hernias in the horse have been previously reported and include incisional drainage, use of chromic gut, repeat laparotomy, excessive incisional oedema, castrated male sex, post operative leucopenia and post operative pain (Gibson et al. 1989). Other studies have confirmed incisional drainage or infection to be the single most important predictive risk factor for development of an incisional hernia with 19.3% of horses with incisional drainage or infection reported to develop hernias vs. 2.6% of horses with no known incisional drainage or sepsis (Mair and Smith 2005). Another study estimated that the odds of an incisional hernia were 62.5 (95% confidence interval: 24.2–202.2) times greater in horses with an incisional infection (Ingle-Fehr et al. 1997).

Conservative therapy involving application of a long-term pressure bandage may resolve the problem in certain cases, eliminating the need for surgical correction (White 1996; Freeman et al. 2002; Freeman 2005). If the ring fails to decrease in size or bridging soft tissue is not found after 2–6 months surgical correction should be considered (Freeman 2005; Stick 2006).

There are a number of different methods reported for surgical repair of ventral midline abdominal hernias in horses, including reconstruction and primary closure (Cook et al. 1996), body wall apposition and addition of a prosthetic mesh placed subcutaneously to support the repair (Kelmer and Schumacher 2008), bridging the body wall defect with mesh placed subcutaneously (van der Velden and Klein 1994), prosthetic mesh placed subfascially but retroperitoneally (Elce et al. 2005; Vilar et al. 2009) or laparoscopic intraperitoneal placement (Caron and Mehler 2009). Use of a synthetic mesh has been advocated to reduce the risk of re-herniation in ‘large’ hernias or any hernia in which primary closure results in excess tension (Stick 2006). The benefits of using synthetic mesh must be weighed against potential complications, which include intestinal adhesions to the mesh and infection of the implant that may necessitate removal (Freeman et al. 2002). In a report evaluating retroperitoneal, subfascially positioned prosthetic mesh for repair of incisional hernias, there was a 62% incidence of surgical site infection, although only one horse developed an infection of the mesh. All of the horses in the study experienced episodes of recurrent colic following the procedure which may have been due to intra-abdominal adhesions (Elce et al. 2005). In another report evaluating a prosthetic mesh repair placed in a similar fashion, there was an overall 13% fatality rate associated with infection or recurrent abdominal pain (Tulleners and Fretz 1983). In another report, a much more favourable prognosis was obtained with only one of 36 horses repaired with a mesh experiencing post operative drainage and all but one horse surviving to discharge, but a long-term follow-up was not reported (Edwards 2002). In an attempt to reduce the incidence of gastrointestinal tract adhesions, mesh has been placed subcutaneously following apposition of the body wall (van der Velden and Klein 1994; Kelmer and Schumacher 2008) which significantly reduces the risk of adhesion formation as long as the abdomen is not entered; however, this superficial location with little soft tissue protection may increase the risk of infection (Freeman 2005). In addition, the contents of the abdomen are not able to be explored, and the location of the mesh external to the external sheath may be biomechanically inferior to the more traditional retroperitoneal or intraperitoneal overlay techniques (Cobb et al. 2005).

Primary apposition of the body wall defect without the use of a mesh has the benefit of avoiding adhesions to the mesh and reduced risk of infection as there is minimal foreign material in the wound. The primary focus of the studies evaluating treatment of abdominal hernias in horses has been directed at the evaluation of mesh repairs. In one previously published abstract, using cases from the same institution as the present paper, mesh repair compared favourably to primary repair. Unfortunately, records and data for those cases were lost, preventing further analysis (Davis and Rakestraw 2002). To the authors' knowledge, there are no detailed reports evaluating a large number of horses with ventral midline incisional hernias closed primarily without the use of mesh implantation. The objectives of this paper are to report the technique, perioperative care, complications and outcome of primary closure of incisional hernias in horses and to compare the results of primary closure to prosthetic mesh implantation.

Materials and methods

The medical records of horses of all ages admitted to Texas A&M University Veterinary Medical Center from 1998–2009 were examined. Records for cases in which an incisional herniorrhaphy was performed were included in the study. These included hernias of all sizes, including those that had areas of drainage and regardless of the hospital that performed the initial surgery that led to hernia formation. Horses were excluded from this study if the herniorrhaphy was performed due to acute body wall dehiscence or if the hernia resulted from something other than a previous ventral midline celiotomy.

Information obtained from the medical record included case details, intended use, initial surgery that resulted in hernia formation, time until the hernia was first identified, size of the hernia, time from when the hernia was noted until repair, method of repair (primary repair or mesh implant), surgery time, and pre- and post operative care. Specifically, information obtained regarding pre- and post operative care consisted of days feed was withheld, pre- and post operative diet recommendations, duration of an abdominal bandage, duration of antimicrobial and nonsteroidal anti-inflammatory drug (NSAID) administration and post operative exercise recommendations. Follow-up information was obtained by re-evaluation at our clinic or telephone communication with the owner or referring veterinarian. Specific follow-up information obtained included whether any complications had developed associated with the repair of the hernia, cosmetic appearance of the previous hernia site, current use of horse and if the horse returned to intended use and duration of convalescence prior to returning to work. Owners were asked to characterise the hernia site as normal appearance, normal appearance but areas that felt abnormal upon palpation of the area of repair or presence of visible abnormalities.

Surgical technique for primary closure

In the 7–14 days prior to surgery, owners and referring veterinarians were encouraged to feed a diet of no hay and a complete pelleted feed at 50–75% of the label recommendations to decrease bulk within the abdomen and decrease overall bodyweight. Preoperative examination of the hernia usually consisted of thorough palpation of the area; the hernia was not routinely examined ultrasonographically. Preoperatively, horses received systemic antimicrobials (procaine penicillin G 22,000 iu/kg bwt i.m. q. 12 h and gentamicin sulphate 6.6 mg/kg bwt i.v. q. 24 h) NSAIDs (flunixin meglumine 1.1 mg/kg bwt i.v. q. 12 h) and tetanus prophylaxis (tetanus toxoid 1 ml i.m. once). Antimicrobial and NSAID administration was continued for variable amounts of time based on clinical assessment and clinician preference. After induction of general anaesthesia, horses were positioned in dorsal recumbency and the ventral abdomen was clipped, prepared and draped in a routine fashion. A linear incision was made over the ventral midline defect through the skin and subcutaneous tissues extending 2–5 cm beyond the palpable body wall defect. The skin and subcutaneous tissues were then elevated and undermined as one layer over the defect to a point extending 2–4 cm beyond the fibrous hernia ring. Depending on clinician preference, need for abdominal exploration or health of the tissues, the hernia sac was either resected along with all abnormally thin tissue to the level of the fibrous hernia ring to allow for clear identification of healthy tissue and apposition of the body wall, or left intact and inverted into the abdomen. If the hernia sac was resected, the abdomen was explored and, if needed, adhesiolysis or resection of any adhesed bowel was performed. If multiple small hernias were present and defects located in close proximity to one another, the thin bands between the defects were resected to allow for easier, more cosmetic closure. Suture material and patterns used to close the defect varied, but consisted of No. 2 or 3 polydioxanone or polyglactin 910 on a reverse cutting needle and combinations of interrupted suture patterns and simple continuous patterns. Interrupted patterns were used in areas of the repair that were deemed to be under the most tension and any areas that were relatively tension free were closed with a simple continuous pattern. Any excess skin was resected to allow for cosmetic apposition prior to routine subcutaneous and skin closure. A stent bandage consisting of a rolled sterile towel and held in place with 0 polypropylene in a cruciate pattern was placed over the incision. An abdominal support bandage was placed on the horse as it hung from the hoist just before placement in the recovery stall or was placed immediately following recovery.

Surgical technique mesh implantation

Horses were prepared for surgery and positioned as described above. The incision, initial dissection and decision to resect or invert the hernia sac were also as described above. The edges of the fibrous ring were approximated, if possible, as described above. A double layer of knitted polypropylene mesh (Bard Mesh)1 was cut to a size to span the closed body wall by approximately a 5 cm margin on all sides. The mesh was tacked in place with simple interrupted sutures of 0 polydioxanone placed every 2 cm and the margin was secured with a simple continuous line of 2-0 polydioxanone. The subcutaneous tissue was closed with 2-0 polydioxanone in a walking pattern ensuring that bites incorporated body wall, mesh and subcutaneous tissue so that all dead space was obliterated. A second layer of subcutaneous tissue was closed in a simple continuous pattern at the cut edges of the skin. The skin edges were trimmed as needed to allow for apposition and closed with skin staples. A stent bandage and abdominal support bandage was placed as above.

Post operative care

Horses were maintained on antibiotics and NSAIDs for a variable amount of time following surgery. The abdominal support bandage consisted of a sterile cotton pad placed directly over the incision and held in placed with adhesive tape (Elasticon)2 wrapped around the entirety of the horses' abdomen from just cranial to the incision to just caudal to the incision. An abdominal support bandage was maintained for a minimum of 3 days post operatively. After 2004, if abdominal support was recommended for >7 days, a hernia belt3 was used. The duration of the abdominal support bandage varied based on clinician preference and size of the preoperative defect. Horses were confined to a stall for a minimum of 4 weeks following surgery. Horses were then allowed turnout in a restricted enclosure for a minimum of 2 weeks prior to unrestricted turn out at 6 weeks following surgery at the earliest. In addition to the use and duration of an abdominal support bandage, duration of stall confinement and paddock confinement, other modifications of post operative care consisted of diet changes. In cases where the hernias were large (>15 cm diameter) or there was excessive tension, the complete pelleted feed diet was maintained for 2–3 weeks post operatively in order to keep the animal at a lower bodyweight and reduce bulk within the abdominal cavity.

Data analysis

Data were analysed using descriptive and inferential methods. For descriptive purposes, categorical data were summarised using contingency tables; continuous data that appeared to follow a Gaussian distribution were summarised with means ± s.d., whereas variables that appeared non-Gaussian were summarised using medians and ranges. When necessary, data were transformed to meet distributional assumptions of methods used for inferential analyses. Associations between the dichotomous dependent variable of hernia repair method (mesh-treated vs. not) and independent variables were assessed using logistic regression analysis. The association between a given independent variable with mesh repair was summarised as the odds ratios (ORs) derived by exponentiation of coefficients of logistic regression modelling, and the 95% confidence interval for each OR was estimated by maximum likelihood methods. For purposes of analysis, weight was categorised as ≥488.6 kg (the median for all 47 horses) or <488.6 kg, and Quarter Horses and American Paint Horses were combined and considered relative to other breeds. When there were 0-values in any cell of a contingency table for an association of repair method with a given variable, Fisher's exact test was used to assess the statistical significance of the association between variables. The association between outcome (apparently normal, palpable but nonvisible defect and visible defect) and continuous variables was assessed using ANOVA. Inferential analysis of the associations between the dichotomous dependent variable of primary hernia repair method (inversion of the hernia sac or resection of the hernia sac) and the outcome was assessed using Chi-squared tests. A significance level of P<0.05 was used for all analyses, which were performed using S-PLUS version 8.04.


The study population included 38 horses that had a primary hernia repair without mesh support and 9 horses that had a primary apposition with mesh support. The primary apposition group without the use of mesh consisted of 6 stallions, 20 mares and 12 geldings. There were 20 Quarter Horses, 4 Saddlebreds, 3 Arabians, 2 Thoroughbreds, one American Miniature, one Draught cross, 3 Warmblood, one donkey and 3 horses did not have their breed listed. The mean ± s.d. bodyweight of the primary repair group was 466 ± 116.8 kg. The 9 horses repaired with mesh implantation included one stallion, 6 mares and 2 geldings. Breeds represented included 4 Quarter Horses, 2 Thoroughbreds, one Saddlebred and one Warmblood; the breed was not recorded for one horse. The mean bodyweight of the mesh repair group was 481 ± 110 kg. There was no significant difference between the 2 treatment groups in distributions of age, weight, sex or breed (Table 1).

Table 1. Comparison of variables between 38 horses undergoing primary hernia repair vs. mesh repair treated at the Texas Veterinary Medical Center
VariableClosure without meshMesh supportOdds ratio (95% CI)P value
  • *

    Indicates statistically significant difference between the 2 groups examined.

Mean ± s.d. age (years)7.6 ± 4.57.8 ± 4.31.0 (0.8–1.2)0.9233
 Male18 (47%)3 (33%)1 (NA)NA
 Female20 (53%)6 (67%)1.3 (0.6–2.9)0.4537
 Other14 (37%)5 (56%)1 (NA)NA
 Quarter Horse/Paint24 (63%)4 (44%)0.7 (0.3–5.6)0.8360
 <488.6 kg20 (54%)4 (50%)1 (NA)NA
 ≥488.6 kg17 (46%)4 (50%)1.2 (0.3–5.6)0.8360
Cosmetic outcome    
 Normal18 (72%)2 (33%)1 (NA)NA
 Palpable only3 (12%)1 (17%)3.0 (0.3–43.4)0.4394
 Visible4 (16%)3 (50%)6.7 (0.8–54.1)0.0830
Short-term complications*    
 No34 (89%)5 (56%)1 (NA)NA
 Yes4 (11%)4 (44%)6.8 (1.3–36.3)0.0297
Long-term complications*    
 No25 (96%)3 (50%)1 (NA)NA
 Yes1 (4%)3 (50%)25.0 (1.9–321.8)0.0195
Mean ± s.d. length (cm)12.1 ± 7.117.0 ± 7.91.1 (<1.0–1.3)0.0950
Mean ± s.d. width (cm)8.5 ± 4.46.8 ± 1.40.9 (0.7–1.1)0.2976
<70 cm219 (50%)3 (33%)1 (NA)NA
≥70 cm219 (50%)6 (67%)2.0 (0.4–9.2)0.3773
Mean ± s.d. duration of surgery (min)*102 ± 51.6188 ± 83.61.3 (1.1–1.6)
(OR for every 15 min of surgery)
Mean ± s.d. duration of antimicrobials (days)*3.1 ± 2.512 ± 7.21.5 (1.1–2.0)0.0008
Mean ± s.d. duration of hospitalisation (days)*5.1 ± 2.610.1 ± 7.01.3 (>1.0–1.6)0.0296
Mean ± s.d. time to use (months)*3.9 ± 2.06.8 ± 2.21.8 (1.1–3.0)0.0407

The initial abdominal surgery that resulted in a midline hernia consisted of exploratory laparotomy for colic in 36 cases, caesarean section in 4 cases, ovariohysterectomy in one case, cystotomy in 2 cases, exploration following penetrating abdominal wound in one case and the cause was not recorded in 3 cases.

The mean time from surgery until the hernia was first identified was 30 days with a range of 7–90 days. The time from when the hernia was first noted until repair was 3–36 months with a mean of 7.5 months and a median of 5.5 months.

Primary repair without mesh support in 38 horses

The mean length of the hernia repaired by primary closure was 12.1 cm with the longest measuring 28 cm and the shortest 2 cm. The mean width was 8.5 cm with the widest measuring 18.0 cm and the narrowest 2 cm. The mean and median total areas for the primary closure group were 107 and 69 cm2, respectively. Twelve horses in this group had >1 palpable hernia. Ten horses had 2 distinct, separate hernias and 2 horses were described as having multiple small hernias. The largest hernia repaired in the manner described was 28 × 15 cm.

The hernia sac was resected in 29 of 38 cases and inverted into the abdomen in 8; this information was not available for one horse. The average surgery time for horses undergoing primary apposition without mesh support was 102 ± 51.6 min. In 3 cases (8%) adhesions were noted between the previous incision and the caecum (n = 1), the jejunum (n = 1) and both the caecum and the large colon (n = 1). All 3 horses required resection of the involved portion of the gastrointestinal tract.

Following surgery the mean duration of antibiotics for horses in which the hernia was repaired by primary closure was 3.1 ± 2.5 days and the mean duration of NSAIDs was 2.7 ± 1.9 days. Mean duration of hospitalisation was 5.1 ± 2.6 days. An abdominal support bandage was used following discharge in 20 cases (53%), the mean duration of abdominal support following discharge was 22.6 days (median 10 days, range 2–65 days). Horses were confined to a stall for a mean of 43 ± 18.5 days.

Outcome of primary repair without mesh

Short-term complications occurred in 4 of 38 cases (11%), including one horse that re-herniated the caudal 3 cm of the repair, 2 horses that developed diarrhoea and fever in the immediate post operative period and one horse that developed a 3 cm pressure sore over the cranial aspect of the incision caused by slippage of the abdominal support bandage. The case that re-herniated was managed with prolonged abdominal support bandage and the small hernia was no longer visible but remained palpable after 45 days of support. Both cases of fever and diarrhoea resolved without complication. The bandage sore healed without complication following an additional 30 days of hospitalisation at the owners' request. One horse died of colic 3 weeks following repair. Necropsy revealed a small intestinal volvulus with no adhesions to the repair site and with good healing of the hernia repair.

Long-term follow-up data were available for 25 of 38 cases (66%). Failure of the hernia repair was reported for one case (4%), a 10-year-old donkey that had a 10 × 10 cm hernia following a routine exploratory celiotomy. The donkey was diagnosed with a cystolith 14 months after the colic surgery and it was decided to perform a cystotomy via a ventral midline approach and concurrently repair the incisional hernia. The surgical approach for this patient required a caudal incision necessitating reflection of the penis in order to gain access to the bladder. The repair initially appeared to heal well but the hernia was noted to have recurred 6 weeks following discharge from the hospital.

The repair site was characterised as normal in 18 cases (72%), normal appearance but abnormal palpation in 3 (12%) and abnormal appearance and palpation in 4 cases (16%). There was no significant difference in the cosmetic outcome when comparing resection vs. inversion of the hernia sac (P = 0.7343). Twenty-three horses (92%) returned to their previous level of use including 4 brood mares, 12 western performance horses, 6 English performance horses and one racehorse. The mean time to return to use was 4 months (median 3 months, range 2–8 months).

Mesh repair group

Short-term complications were recorded in 4 of 9 cases (44%). These consisted of 3 cases of implant infection prior to discharge and one horse that developed signs of colic 6 days post operatively. An exploratory surgery was performed via a flank laparotomy 6 days post operatively and small intestine was found to be adhesed to the incision but not directly to the mesh. The adhered bowel was freed but the horse was subjected to euthanasia due to continued signs of colic 2 days later. The 3 cases of implant infection were first observed at 3, 7 and 12 days post operatively. All 3 cases were initially managed with long-term antimicrobial therapy, which resulted in temporary resolution of the clinical signs associated with implant infection. However, 2 of these 3 horses ultimately required implant removal at 7 and 8 months post operatively due to a return of incisional drainage. These 2 horses were included in the group of 3 that developed long-term complications. Also included in the group of long-term complications was one horse that showed signs of colic 12 weeks post operatively. Long-term follow-up was available for 6 of 9 horses following mesh implantation and the results are presented in Table 1.

Comparison of primary repair and mesh-repaired groups

There was no significant difference between the 2 groups in regards to age, breed, sex or size of the hernia. Horses repaired with mesh support had significantly longer surgery times, longer duration of antimicrobial administration, longer duration of hospitalisation, more short- and long-term complications and longer period of convalescence prior to return to use. There was no significant difference between the 2 repair methods in regards to cosmetic outcome (Table 1).

There was no significant difference between groups in length, width or area of the defects. The area data for the primary repair group did not follow a Gaussian distribution: the mean and median areas for the primary group were 107 and 69 cm2, respectively. The mean and median areas for the mesh group were 113 and 120 cm2, respectively. Consequently, a categorical variable for area was created using 70 cm2 (approximately the median value for the primary group). There was no significant difference in the proportion of horses with lesions having an area >70 cm2 between the 2 repair methods (Table 1).

There was no significant difference between the 2 groups in regards to duration of NSAID use (P = 0.1107) or duration of stall rest (P = 0.1038). Distribution of duration of bandaging for the primary repair group was skewed: the median value was 10 days as the mean was 22.6 days. Consequently, data were log-transformed for analysis. The logarithms of duration of days of bandaging did not differ significantly between the 2 groups (P = 0.7010). Defects were categorised as single or multiple (>1 defect) for purposes of analysis and there was no significant difference in proportion of horses with multiple defects between the 2 repair methods (P = 0.0864).

Factors affecting cosmetic outcome regardless of repair method

Cosmetic outcome was reported for 31 of 47 horses (66%). There was no statistically significant difference between the cosmetic outcome and variables assessed including length of the defect, width of the defect, total area of the defect, time until the hernia was identified, time from identification of the defect to repair, duration of post operative abdominal support and number of defects present (Table 2).

Table 2. Comparisons of mean ± s.d. between normal, palpable and visible regardless of the treatment method
VariableNormal (n = 19)Palpable only (n = 4)Visible (n = 7)P value
  1. There was no statistically significant difference between outcome regardless of method and of the variables examined.

Length (cm)14.0 ± 6.47.0 ± 1.416.1 ± 10.00.1273
Width (cm)9.2 ± 4.46.0 ± 0.88.5 ± 5.20.4329
Area (cm2)121.9 ± 93.741.5 ±7.5163.0 ± 128.20.4329
ln (time from identification to repair [months])1.7 ± 0.52.0 ± 0.72.0 ± 0.70.4067
Days to identification36.2 ± 26.323.0 ± 6.228.7 ± 7.10.6427
Duration of bandaging days27 ± 19.828 ± 3213 ± 6.90.5214
>One defect26% (5/19)25% (1/4)29% (2/7)0.9901


Primary closure of incisional hernias without the use of mesh appears to have a good outcome, with 23 of 25 horses (92%) returning to previous use and 21 cases (84%) being normal in appearance. Of these 21 horses with normal appearance, 18 were normal in both appearance and palpation and 3 were normal in appearance but upon palpation had defects noted by the owners. This compares favourably with the mesh implantation group where only 3 of 6 (50%) were considered to be normal in appearance and of those, 2 had palpable defects noted by the owner. Previous reports indicate that repair of incisional hernias in large horses (>450 kg) and repair of large hernias increase the risk of complications and are more likely to require mesh augmentation of the repair (Tulleners and Fretz 1983; Elce et al. 2005). In our study we were able to obtain a similar cosmetic outcome without the use of a mesh with no significant difference between the 2 groups of repair techniques in terms of case details or size of the defect. In addition, horses repaired by primary closure without mesh support had significantly shorter surgery times, shorter duration of hospitalisation, fewer complications and faster return to work compared with horses in which mesh was used.

Although it is difficult to compare results of this study to results of other studies, the 84% of visually normal horses in our report compares favourably with the 79% nonrecurrence rate reported previously in the only other study of primary closure of incisional hernias without the use of mesh (Cook et al. 1996). The results of our study also compare favourably to the 78.3% ‘excellent’ outcome reported by Tulleners and Fretz (1983) where a subfascial mesh was used to augment the repair.

When performing an incisional herniorrhaphy, the hernia sac can either be inverted and the abdomen not entered (closed technique) or it can be incised and resected (open technique) (Gibson et al. 1989). Opening and resecting the hernia sac has previously been associated with a better cosmetic appearance (Baxter 1992) and has the advantage of allowing exploration of the abdomen. In our study, 3 of 30 horses in which the abdomen was entered were found to have intra-abdominal adhesions, which were thought might potentially lead to future episodes of abdominal pain and consequently required further intervention; this has been reported in other case series (Cook et al. 1996). Although there was no significant difference in cosmetic outcome between resecting the hernia sac and inverting the hernia sac, horses in which the hernia sac was inverted tended to have a higher incidence of visible defects. A potential advantage to resecting the sac and all abnormally thin tissue to the level of the fibrous hernia ring is that it may allow for clear identification of healthy tissue and apposition of the body wall resulting in a more cosmetic appearance.

As expected, primary apposition of incisional hernias without the use of a mesh had a significantly shorter duration of surgery as compared with the use of a mesh. The 86 min increase in mean surgery time is due to the additional procedures involved in mesh implantation including measuring and cutting the mesh, placing it in the desired position and suturing it in place. The added time may also be due to lack of surgeon experience with mesh implantation as that procedure is performed much less frequently at our clinic. Length of surgery and anaesthesia has been associated with increased complications following surgery, most commonly increased surgical site infections, in both small and large animals (Beal et al. 2000; Ahern et al. 2010). This reduction in surgery time may help decrease the rate of surgery site infections.

Horses treated without mesh implantation received antimicrobials post operatively for a significantly shorter time compared with horses that had mesh implantation. The increased duration of antibiotics was likely due to clinician concern for development of an infection because this is a reported complication following mesh implantation (White 1996). Duration of hospitalisation was also significantly longer for horses treated by mesh implantation. This may have been due to clinicians' desire to treat these horses longer with injectable antibiotics. These differences in hospitalisation and treatment are likely to result in less cost to the owner

The method of mesh implantation used in this report has been described previously, where the outcome appeared to be better than we obtained, with only 25% reported to have visible abnormalities and only one surgical site becoming infected (van der Velden and Klein 1994; Kelmer and Schumacher 2008). However, in those reports, the sizes of the defects were not recorded and the outcome was considered good unless there was evidence of re-herniation making comparison with our results difficult. The disadvantages of placing the mesh externally to the rectus sheath includes minimal soft tissue protection of the mesh (Freeman et al. 2002) and less than ideal positioning to counter the biomechanical forces placed on a ventral midline incision (Cobb et al. 2005). In addition, we used a double layer of mesh instead of a single layer; this double layer has been used previously (Kelmer and Schumacher 2008; Vilar et al. 2009) but remains controversial. Benefits of double layer mesh include added mechanical strength and ability to hold suture but disadvantages include increased foreign material in the wound (Vilar et al. 2009). It is possible that the trend seen in this report of a poorer cosmetic outcome in the mesh support group was due to a foreign body reaction to the double layer mesh. Further evaluation comparing methods of mesh placement is warranted.

Incisional hernias have been reported to occur and be noted by owners within 4 months of the initial surgery (Wilson et al. 1995). This is substantiated by our report where all hernias were noted within 90 days of the initial surgery. Many authors recommended delayed repair of an incisional hernia beyond resolution of any infection or acute inflammation (Kelmer and Schumacher 2008). This delay allows for increased strength and definition of the fibrous hernia ring (McIlwraith and Robertson 1998). The amount of time that elapsed between the initial surgery and the hernia repair did not affect the cosmetic outcome regardless of technique in our report. This may be due to the fact that in our population of horses a mean of 7.5 months elapsed between when the hernia was noted and when it was repaired. This is more than the 4 months reported by Kelmer and Schumacher (2008) and is probably sufficient time for inflammation and infection to resolve and for a fibrous ring to develop; it is possible that for this reason no difference in outcome was noted.

In addition to delaying surgical repair until conditions were appropriate, the recommendation was made in all cases to change diet to decrease bodyweight and bulk within the abdomen both before and after surgery. The exact duration of diet modification could not be ascertained from the medical records as this was often accomplished via telephone conversations with owners and referring veterinarians prior to arrival at the clinic and following discharge.

Use of an abdominal support bandage has been advocated previously as either part of the post operative treatment (White 1996) or as a stand-alone treatment (Stick 2006). In our population of horses, an abdominal bandage was used for a mean of 22.6 days in horses in which a mesh was not used. The duration of bandaging did not have a significant effect on cosmetic outcome regardless of repair technique, although horses with a visible defect tended to have a shorter duration of post operative bandaging. In previous studies examining celiotomy incisions, the use of an abdominal support bandage following surgery appeared to decrease the incidence of incisional complications (Smith et al. 2007). One complication noted with the use of abdominal bandages was a sore associated with slipping of the bandage. This can be avoided by including a breast strap to prevent the caudal movement of the bandage.

Incisional complications have been reported to occur more commonly in horses weighing >300 kg (Wilson et al. 1995). Presumably, complications would occur more commonly following repair of incisional hernias on horses weighing >300 kg. However, in our report, there was no significant difference between the cosmetic outcome groups and bodyweight regardless of repair method. Bodyweight may contribute to hernia development but was not found to be a risk factor in this study because of the similar distribution of bodyweights of the horses in the 3 outcome groups and the small number of horses in each group.

Rupture of the obliquus internus abdominus (OIA) has been reported following repair of incisional hernias in large horses (>590 kg); in addition to the size of the horse, it was speculated that tension on the repair site may contribute to this complication (Elce et al. 2005). In our study, subcutaneous placement of the mesh was used and rupture of the OIA was not observed. Two of the 3 horses that ruptured the OIA in the previous study (Elce et al. 2005) were mid-gestation mares, whilst our study did not include any pregnant mares. In addition, abdominal support bandages were used to aid in support of the repair site. Only 3 horses in our population were >600 kg, all of which healed without complication.

There are several limitations of this study. First, the number of horses studied was relatively modest, particularly for the mesh group. The size of the study population probably limited our power to detect differences among groups. Moreover, not all horses were available for follow-up, which further limited our power to detect differences among groups with respect to cosmetic outcome. The wide confidence interval for some of the odds ratios reflects the large degree of imprecision in the estimates resulting from the small numbers of horses in the study. Performing a similar study on a larger number of horses may be of benefit to more accurately detect significant differences in outcome.

Surgical treatment type was not assigned randomly in this study. Thus, differences between groups could have resulted from a selection bias: horses might have received mesh treatment rather than no mesh for specific yet undetermined reasons and these differences could have caused confounded results (i.e. the observed differences between groups were attributable to the confounding factors and not the treatment itself). Factors that might have affected which surgical technique was selected included age, sex, breed, weight of horse, size of hernia and timing of the herniorrhaphy in relation to the initial surgery and in relation to identification of the hernia. No significant difference between the 2 groups was observed for any of these factors. Nevertheless, it is possible that there were other unmeasured factors that influenced the decision for surgical technique that may have influenced the observed results.

This study was conducted at a single centre. Thus, different results might be observed at centres where the types of horses (e.g. greater diversity in bodyweight distribution than that observed for this study) and preferences of surgeons for technique and case selection might differ. In addition multiple surgeons performed the described procedures and, as a result, surgeon variation may have affected the outcome. We also relied on owners' assessment of cosmetic outcome in the majority of cases and although we attempted to simplify the questionnaire it is possible that their lack of medical training could have resulted in inaccurate assessment of the surgical repair. As with any retrospective study, the quality of data may have been less than that for a study in which data are collected prospectively.

At our clinic, primary apposition of incisional hernias in horses without the use of mesh support appeared to result in a good cosmetic outcome while avoiding some of the complications associated with mesh implantation. In addition, when a mesh was not used, study horses had a shorter duration of hospitalisation and a quicker return to work; in fact the odds of returning to use were nearly 2-fold greater per month for the primary treated group relative to the mesh treated group (Table 1). There was no difference in hernia size between the 2 treatment groups. The authors believe that with appropriate pre- and post operative care and surgical technique, incisional hernias repaired in this manner have an excellent prognosis and cosmetic outcome.

Conflicts of interest

No conflicts of interest have been declared.

Manufacturers' addresses

1 Davol inc., Cranston, Rhode Island, USA.

2 Johnson & Johnson, Skillman, New Jersey, USA.

3 CM, Norco, California, USA.

4 TIBCO, Seattle, Washington, USA.