Prioritization of early nutritional support, especially enteral nutrition, in critically ill patients is gaining widespread acceptance in veterinary medicine.1 When treating patients with certain diseases, such as pancreatitis, or after major gastric or duodenal surgery, enteric feeding tubes such as jejunostomy tubes are indicated1–3 to provide nutrients via the enteral route while bypassing the pancreas or surgical site. Although a more recent report4 described the use of esophagostomy tubes in patients with acute pancreatitis, feeding aborad to the pancreas is still the most commonly accepted technique, especially in patients undergoing abdominal surgery.
The surgical placement of enteral feeding tubes originates from techniques used in human medicine, with a large variety of techniques decribed.5,6 Several techniques have been described in veterinary medicine for surgical jejunostomy tube placement, such as a pursestring,3,7,8 an interlocking box placement technique,9 tunneling with a needle,3 or an inverting longitudinal tunnel (Witzeling or Witzel jejunostomy).3 In this technique, the full-thickness intestinal wall is inverted around the jejunostomy tube and sutured.3 A modification used in our institution is to make a seromuscular incision over the preplanned length of the tunnel and complete the tunnel by suturing the incised seromuscular layers over the jejunostomy tube (seromuscular incision modification). A laparoscopically assisted jejunostomy placement technique has been described that uses a pursestring suture for securing the jejunostomy tube in the jejunum.10
A commonly used surgical variation of a jejunostomy tube placement technique is placing a jejunostomy tube through a percutaneous endoscopically placed gastrostomy tube3 as well as via surgically placed gastrostomy tubes. Another technique that has been described is use of a nasojejunal tube.3 Recently developed variations in placement include endoscopically assisted nasojejunal feeding tube placement2 and a fluoroscopic technique for wire-guided nasojejunal tube placement.10,11
Two potential adverse effects of surgical placement of a jejunostomy tube include leakage around the tube leading to peritonitis as well as decrease in intestinal luminal area, which may become clinically relevant in smaller size patients. Although decrease in luminal area with the Witzel tunnel technique has been described in human patients,5 to our knowledge, no study to date has objectively compared 3 commonly used placement techniques for decrease in luminal area, nor has any study leak tested different jejunostomy placement techniques with the tubes in place in the intestinal segments.
The aim of the study reported here was to evaluate and compare pursestring, Witzel (seromuscular inversion), and seromuscular incision jejunostomy tube placement techniques for leakage pressure and decrease of intestinal luminal area in vitro. We hypothesized that the jejunostomy site would fail, resulting in leakage, at the lowest pressure for the pursestring technique, followed by the seromuscular incision technique and then the Witzel tunnel technique. We also hypothesized that the pursestring technique would result in the smallest decrease in intestinal luminal area, followed by the seromuscular incision technique and then the Witzel tunnel technique.
Materials and Methods
This study was approved by the Institutional Animal Care and Use Committee of the University of Florida. Euthanasia was performed in accordance with the AVMA Guidelines on Euthanasia (June 2007).
Specimen preparation
Sections of jejunum (50 cm in length) were harvested immediately prior to euthanasia from 10 mixed-breed dogs and stored in saline (0.9% NaCl) solution–soaked towels during instrumentation and in a saline solution bath during testing. The orad and aborad ends of the resected jejunum were clamped with Rochester-Carmalt forceps. Three jejunostomy tubesa (8F) were inserted through an antimesenteric stab incision, at equally spaced intervals, with at least 10 cm between 2 sites. Three insertion techniques were evaluated: pursestring only, with 4 stitches, equally spaced and approximately 1 mm from the stab incision; a 3-cm-long inversion technique with all layers (Witzel tunnel technique) in which the jejunostomy tube was inserted through a stab incision after which a full-thickness intestinal wall tunnel was created by means of suturing serosa to serosa in a simple continuous pattern starting aborad to the jejunostomy tube insertion site and continuing for 3 cm in an orad direction; and a modification of the inversion technique where the seromuscular layers were incised over a distance of 3 cm to create a tunnel (seromuscular incision), with the stab incision for luminal entry at the most aborad edge of the incision. The tunnel was created by means of suturing the incised seromuscular layers over the jejunostomy tube in a simple continuous suture pattern. A stainless steel ruler was used to premeasure all tunnels.
All tubes were fed into the lumen for a distance of 10 cm in the same direction. A new No. 15 scalpel blade was used for all bowel segments. A 7F double lumen central venous catheterb was placed at the orad and aborad end of the bowel segment to allow luminal filling and simultaneous pressure recording. The double lumen central venous catheters were secured with a pursestring technique. For all procedures, a new package of 4–0 polydioxanonec suture was used. The jejunostomy tubesa were marked with colored tape for identification purposes. During ultrasonography and leak testing, the jejunostomy tubes were only identified by their color code and not by technique. Placement of the 3 jejunostomy tubes in each segment was in random order. One surgeon (MR) placed all jejunostomy tubes and central line catheters.
Initial filling of the bowel lumen was accomplished with 50-mL syringes filled with dyed saline solution (1 mL of 1% methylene blued in 1,000 mL of saline solution) through a dedicated port of one double lumen central catheter while simultaneously venting the solution from the intestinal segment via the second central venous catheter. After the initial filling, perfusion was performed at 4 mL/min with saline solutione dyed with methylene blue, and the intraluminal pressure was measuredf through the second lumen of the double lumen central catheter. The bowel lumen was filled with saline solution to a pressure of 6 mm Hg prior to performing the ultrasonographic measurements and was maintained at 6 mm Hg during the ultrasonographic evaluation.
Ultrasonography
Ultrasonography was performed with a 5- to 17-MHz transducerg with spatial compounding (Figure 1). The intestinal segment was suspended in a warm saline solution bath during transport to and from the ultrasonography room and during the ultrasonographic evaluation. Both cine loops and stills were made of each of the jejunostomy tube sites and of 1 area of normal jejunum and were saved on a picture archiving and communication systemh in DICOM (Digital Imaging and Communications in Medicine) format for measurements. Acquisition of images started at the insertion site and was continued for 3 cm in the direction of the intraluminal portion of the jejunostomy tube. One radiologist (RFG) performed all ultrasonographic measurements, with 1 person (AS) dedicated to luminal filling during the examination. Cross-sectional images representative of the area orad to the insertion site, at the site of the jejunostomy tubes entering the lumen, and approximately 1.5 cm aborad of the insertion site with the tube traveling in the tunnel or intraluminally were selected for subsequent measurements.
Transverse ultrasonographic images of a 50-cm-long canine jejunal specimen with jejunostomy tubes placed in vitro via 3 techniques (pursestring, Witzel [seromuscular inversion], and seromuscular incision). Images were obtained at a control site (A) and at the pursestring technique site (B) with the jejunostomy tube free within the intestinal lumen, seromuscular inversion site (C), and seromuscular incision site (D). Two double lumen central venous catheters were placed at each intestinal end for luminal filling and leak testing. Specimens were harvested immediately prior to euthanasia from mixed-breed dogs with the orad and aborad ends clamped and stored in saline (0.9% NaCl) solution–soaked towels during instrumentation. Intestinal luminal areas were measured ultrasonographically with the intestinal segments suspended in a warm undyed saline solution bath with the intestinal lumen filled with dyed saline solution (intraluminal pressure, 6 mm Hg). Leak testing was performed by means of infusion of dyed saline solution (4 mL/min) until each jejunostomy site had failed. Intestinal luminal area and leakage pressure for the 3 tube placement techniques were compared.
Citation: American Journal of Veterinary Research 76, 5; 10.2460/ajvr.76.5.467
All measurements were made during a single session by an observer (MDW) blinded to the tube insertion technique, on representative images, chosen by the ultrasonographer who performed the imaging. The mucosal-luminal surface on a transverse image was measured by tracing the mucosal lining to allow for a calculation of intestinal luminal area in cm2. Intestinal luminal area was measured in a normal, noninstrumented area of the intestinal segment (baseline), and for each jejunostomy tube at the insertion site and at a different site with the tube traveling intraluminally. To facilitate comparison, the ratio of the intestinal luminal area at the tube insertion site, compared with the baseline, was calculated for each technique.
Leak testing
After completing the ultrasonographic evaluation, the intestinal loop was filled with dyed saline solution with an infusion pump (at a continuous rate of 4 mL/min) until leakage was noted (failure of the insertion site or the suture line) by 1 observer (GWE) blinded to the insertion technique. Pressures were planned to be capable of exceeding normal resting (2 to 6 mm Hg)12,13 and peak intraluminal pressures (15 to 25 mm Hg).13 The intraluminal pressures at which the different jejunostomy tube placement sites leaked were recorded. The failures of the insertion sites were also assigned a ranking score (1 for the first to leak and 3 for the last to leak) within 1 specimen to allow comparison between specimens. When 1 site leaked, that insertion site was clamped off with Doyen intestinal forceps across the jejunostomy tube entry site in a side-clamping manner with the concave side of the forceps positioned toward the antimesenteric border. When the first 2 sites had failed, the appropriate section of bowel was clamped off and the infusion continued through the double lumen central catheter in the section with the third jejunostomy tube to avoid spurious decrease of intraluminal pressure because of potential leakage at the 2 failed sites. All pressure measurements were performed by a dedicated study member (MR), with another member dedicated to luminal filling (AS).
Statistical analysis
For all statistical analyses, a linear model was used with dog as a random block effect and group as the fixed treatment effect. If the treatment effects were significant, the least squares means were calculated to determine effect sizes and directions. The least squares means were compared with Tukey multiple comparison tests, and adjusted P values were reported. All models were checked for fit, specifically for homogeneity of variances. If the variances were heterogeneous, transformations were performed to improve fit. Commercial softwarei was used to conduct all analyses. The baseline, insertion point, and tube measurements were each analyzed separately with a general linear models procedure.j
After fitting models including the random block effect for dogs, residuals were studied for evidence of non-constant variance. The intestinal luminal data (baseline, insertion point, and tube) did not show evidence of non-constant variance, the residuals were symmetrically distributed around zero, and a Q-Q plot of the residuals did not show evidence of nonnormal distribution. Evidence of heteroskedasticity was found in the residuals for leak pressure, so to address this, the data were transformed by taking the natural logarithm. The log-transformed data did not appear to violate the model's assumptions.
Raw data and details of model fit were presented (Appendix 1 and 2).
Results
Dogs
The median body weight of the dogs was 22.5 kg (range, 18.0 to 31.6 kg). All dogs had an intact, 50-cm-long section of midjejunum available.
Surgery and instrumentation
All tube placement surgeries were completed on each jejunal specimen as planned. During preparation of one of the seromuscular incision jejunostomy tube sites, the submucosa was incised over a small area (< 2 mm). The jejunostomy tube placement was continued as planned, and the musculoserosal layers were routinely apposed over the tube in a simple continuous suture pattern.
Ultrasonography
Baseline jejunal luminal circumference and normal intestinal luminal area measurements were similar between groups (P ≥ 0.73; Table 1). A significantly (P < 0.05) larger decrease in mean ± SD intestinal luminal area was observed at the jejunostomy insertion site in specimens treated with the Witzel technique (1.07 ± 0.52 cm2), compared with the pursestring technique (1.70 ± 0.47 cm2).
—Comparison of intestinal luminal circumference, luminal area, and leakage pressure in intestinal specimens from dogs with jejunostomy tubes placed in vitro via 3 techniques (pursestring, Witzel, and seromuscular incision).
| Variable | Witzel (n = 10) | Seromuscular incision (n = 9) | Pursestring (n = 10) | P value |
|---|---|---|---|---|
| Baseline comparisons | ||||
| Luminal circumference (mm) | 5.26 ± 0.85 | 5.15 ± 0.68 | 5.21 ± 0.49 | 0.94 |
| Intestinal luminal area (cm2) | 1.60 ± 0.59 | 1.54 ± 0.43 | 1.72 ± 0.44 | 0.73 |
| Outcome comparisons | ||||
| Intestinal luminal area (cm2) | ||||
| Insertion site | 1.07 ± 0.52a | 1.16 ± 0.36a,b | 1.70 ± 0.47b | 0.01 |
| Luminal jejunostomy tube | 1.30 ± 0.46a,b | 1.09 ± 0.47a | 1.74 ± 0.49b | 0.02 |
| Leak testing | ||||
| Leakage pressure (mm Hg) | 119.50 ± 77.83 | 90.55 ± 57.73 | 139.80 ± 73.61 | 0.25 |
| Leak test rank | 2 (1–3) | 2 (1–2) | 3 (1–3) | 0.22 |
Specimens were harvested immediately prior to euthanasia from 10 mixed-breed dogs with the orad and aborad ends clamped, and stored in saline (0.9% NaCl) solution–soaked towels during instrumentation. Intestinal luminal areas were measured ultrasonographically with the intestinal segments suspended in a warm undyed saline solution bath with the intestinal lumen filled with dyed saline solution (intraluminal pressure, 6 mm Hg). Leak testing was performed by means of infusion of dyed saline solution (4 mL/min) until each jejunostomy site had failed. Intestinal luminal area and leakage pressure were compared between the 3 tube placement techniques.
Data are reported as mean ± SD or median (range). For leak rank test, the median was also the mode.
Within a row, groups with different superscript letters are significantly (P < 0.05) different.
A larger decrease in intestinal luminal area was also observed at the jejunostomy insertion site in specimens treated with the seromuscular incision technique (1.16 ± 0.36 cm2), compared with the pursestring technique (1.70 ± 0.47 cm2); however, this difference was not significant (P = 0.06).
A larger decrease in intestinal luminal area was observed at the site with the jejunostomy tube traveling intraluminally in the seromuscular incision group (1.09 ± 0.47 cm2), compared with the pursestring group (1.74 ± 0.49 cm2), but no significant (P = 0.083) difference was found between the treatment groups.
Leakage pressure
All jejunostomy tube sites were tested to failure. The lowest pressure (23 mm Hg) at which any of the jejunostomy tube sites leaked was for one placed via the seromuscular incision technique where the submucosa was inadvertently incised, and the second lowest was for a jejunostomy tube placed via the pursestring technique (25 mm Hg). The seromuscular incision with full-thickness incision was censored from statistical analysis of leak testing. Mean pressure at which the jejunostomy sites started leaking was 117.5 ± 71 mm Hg (range, 25 to 252 mm Hg; median, 109 mm Hg [1 seromuscular incision specimen censored]).
The maximum leakage pressure for any jejunostomy tube was 252 mm Hg (recorded for the Witzel technique). Mean ± SD leakage pressure was lowest for the seromuscular incision group (90.55 ± 57.73 mm Hg), and highest for the pursestring group (139.80 ± 73.61 mm Hg). The difference of mean leakage pressures between techniques was not significant either for dog effect (P = 0.338) or treatment effect (P = 0.251).
Median and mode (most frequent score within each group) scores for failures of the insertion sites were lower (rank, 2) for the Witzel and seromuscular incision techniques, compared with the pursestring technique (rank, 3), but these differences were not significant (P = 0.220).
Discussion
In the present in vitro study evaluating 3 jejunostomy techniques in jejunal segments obtained postmortem from dogs, the Witzel technique created a significant decrease in intestinal luminal area at the insertion site. The seromuscular incision technique decreased the intestinal luminal area, but this difference was not significant in the present study. The pursestring technique had the least and nonsignificant decrease in intestinal luminal area. The seromuscular incision technique had a greater decrease in intestinal luminal area at the intraluminal site of measurement versus at the tube insertion point, which was unexpected. We suggest that this may have been a result of a full-thickness folding of the intestinal wall around the aborad portion of the incision at this site.
In this study, all measured leakage pressures were higher than previously reported physiologic intraluminal resting pressures of 2 to 6 mm Hg,12,13 for dogs with one specimen failing within the range of peristaltic pressures (seromuscular incision specimen, 23 mm Hg) and a second specimen failing at the upper limit of reported intestinal peristaltic pressures of 15 to 25 mm Hg13 for dogs (pursestring specimen, 25 mm Hg).
The seromuscular incision specimen that failed at 23 mm Hg had an area of full-thickness incision into the intestinal tract. We chose to report the result for the pressure testing for this specimen, considering that this is something that could happen clinically as well, without the option to redo the jejunostomy placement. However, the result was censored from the statistical analysis of the pressure test results to allow for comparison of techniques.
In the present study, we measured the effect on intestinal luminal area and leakage test results of 3 jejunostomy techniques in dog cadaver specimens. The pursestring technique was chosen for comparison with the Witzel tunnel technique (and the modification of the Witzel technique) because we felt that this technique would provide the least inversion of the jejunum at the site and therefore the least decrease in intestinal luminal area.
Reports of previous studies that used a tunneling technique describe a tunnel of 2.5 cm (veterinary patients)9 or 6.4 cm (human patients).6 We chose to use a 3-cm-long tunnel, measured from entrance into the lumen to the exit site, which we felt would be long enough to get representative measurements and a complete seal as well as short enough to make it clinically applicable. No clear rules or guidelines exist regarding the appropriate intraluminal length of jejunostomy tubes. In previous studies,8,9,14,15 the length has varied from 10 cm to 20 to 25 cm, 1 long loop, or as far as possible. Even though the intraluminal length should not have interfered with the outcome of the present study, we chose to standardize the intraluminal length at 10 cm.
Use of ultrasonography to measure the intestinal lumen with feeding tube placement has not been previously described. As the most representative measurement for decrease in luminal size, we chose to measure the area within the lumen on a cross-sectional ultrasound image. Furthermore, to account for different specimen sizes, ratios of the instrumented area with a normal, noninstrumented area were made to allow interspecimen comparison.
Previous studies9,16–24 have used different filling media, ranging from air to dye at a broad range of filling rates (1 to 150 mL/min). We preferred dyed saline solution, which enabled us to perform ultrasonography without having to switch between setups, and adding the dye allowed for easier identification of jejunostomy site failure during subsequent leakage testing. The filling rate chosen was between the minimum (60 mL/h [in rat colon])19 and maximum (1 L/min [in equine jejunum])18 rates reported.
Resting intraluminal jejunal pressure in unanesthetized dogs has been reported as mean ± SD of 6.8 ± 0.77 mm Hg in the proximal jejunum.12 Another study13 found a normal intraluminal pressure of 2 to 4 mm Hg and a normal intraluminal pressure of 15 to 25 mm Hg during peristalsis. In a different study,25 peak intraluminal pressure in canine small intestine was reported to be at least 85 cm H2O (62.5 mm Hg), although another experimental study13 reported that following small intestinal obstruction in dogs, lymphatic and capillary stasis occurred at a mean intraluminal pressure of 30 mm Hg, atrioventricular shunting at 44 mm Hg, and total venous occlusion at 50 mm Hg. It is therefore estimated that with naturally occurring mechanical obstruction, physiologic pressure will most likely not exceed 50 mm Hg,13 suggesting that it is unlikely that jejunostomy tubes will have to withstand similar or still higher pressures; rather, they likely would only be subjected to peristaltic pressure, not obstructive or peak bursting pressures.
All jejunostomy sites tested in the present study leaked (failed) at pressures that would be considered higher than resting intraluminal pressures. A prior in vivo study25 of dogs used intraluminal pressures from 15 to 25 mm Hg for leak testing, considering this is the normal physiologic intraluminal peristaltic pressure range. In this study, when evaluating the individual jejunostomy sites, for one specimen where the submucosa was inadvertently incised, leakage occurred at 23 mm Hg, for one pursestring specimen at 25 mm Hg, and for all others at ≥ 30 mm Hg. These values are lower than the peak intraluminal pressure (62.5 mm Hg) previously reported as normal for dogs,26 but at the high end of physiologic intraluminal peristaltic pressure (25 mm Hg) as previously reported.13
This study had several limitations. One limitation was that only the jejunostomy site itself was tested. No enteropexies were performed as they are in patients. Arguably, fixation of the loop of jejunum and jejunostomy site to the abdominal wall musculature would most likely decrease the amount of leakage. Our reason for not including enteropexies in our experimental technique was to enable evaluation of each jejunostomy technique on its own. Adding an enteropexy would have added another potential variable to the study. Another limitation was that a small sample size prevented detection of significant reductions in intestinal luminal areas at the insertion site (seromuscular incision technique) or at the site with the luminal jejunostomy tube (Witzel technique).
A previous study9 pressure tested 3 jejunostomy techniques after removing the jejunostomy tube; the interlocking box technique significantly outperformed both the pursestring and Witzel techniques. In that experimental design, cadavers were used within 2 hours after euthanasia, as opposed to intestinal loop harvesting at the time of euthanasia,9 as was performed in our study. No attempt was made to seal off leaking segments. We suggest that their model9 would more accurately test the effect of early jejunostomy removal, whereas our model more closely represents the clinical situation with a jejunostomy tube in place. Additionally, the pursestring and Witzel technique in that study were fixed to the abdominal muscular layer (ie, abdominal wall fixation) with 2 sutures, whereas this study did not include abdominal wall fixation.
The Witzel and seromuscular incision techniques both created a decrease in intestinal luminal area (diameter), the clinical importance of which warrants further investigation especially when considering jejunostomy tube placement in small dogs and cats. On the basis of leakage pressures, we suggest that all 3 tested jejunostomy techniques may be safe; however, in vivo assessment is needed prior to making any clinical recommendations regarding these findings.
Acknowledgments
The authors thank Laura Cuddy for technical assistance.
Footnotes
Argyle polyvinylchloride feeding tube, 8F, 16 inch, Covidien, Mansfield, Mass.
Arrow, 7F, 6-inch polyurethane central venous cathether, Teleflex Medical, Research Triangle Park, NC.
4–0 PDS II, Ethicon Inc, Johnson & Johnson, Cincinnati, Ohio.
Methylene Blue 1%, Taylor Pharmaceuticals, Decatur, Ill.
0.9% NaCl, 1 L, Baxter Healthcare, Deerfield, Ill.
IntelliVue MP40, Philips Healthcare, Andover, Mass.
L17–5 MHz, iU22, Philips Healthcare, Andover, Mass.
AMICAS PACS, Merge Healthcare, Chicago, Ill.
SAS, version 9.3, SAS Institute Inc, Cary, NC.
PROC GLM, SAS, version 9.3, SAS Institute Inc, Cary, NC.
References
1. Chan DL, Freeman LM. Nutrition in critical illness. Vet Clin North Am Small Anim Pract 2006; 36: 1225–1241.
2. Campbell SA, Daley CA. Endoscopically assisted nasojejunal feeding tube placement: technique and results in five dogs. J Am Anim Hosp Assoc 2011; 47:e50–E55.
3. Heuter K. Placement of jejunal feeding tubes for post-gastric feeding. Clin Tech Small Anim Pract 2004; 19: 32–42.
4. Mansfield CS, James FE, Steiner JM, et al. A pilot study to assess tolerability of early enteral nutrition via esophagostomy tube feeding in dogs with severe acute pancreatitis. J Vet Intern Med 2011; 25: 419–425.
5. Tapia J, Murguia R, Garcia G, et al. Jejunostomy: techniques, indications, and complications. World J Surg 1999; 23: 596–602.
6. Minard G. The history of surgically placed feeding tubes. Nutr Clin Pract 2006; 21: 626–633.
7. Crowe DT, Devey JJ. Clinical experience with jejunostomy feeding tubes in 47 small animal patients. J Vet Emerg Crit Care 1997; 7: 7–19.
8. Hewitt SA, Brisson BA, Sinclair MD, et al. Evaluation of laparoscopic-assisted placement of jejunostomy feeding tubes in dogs. J Am Vet Med Assoc 2004; 225: 65–71.
9. Daye RM, Huber ML, Henderson RA. Interlocking box jejunostomy: a new technique for enteral feeding. J Am Anim Hosp Assoc 1999; 35: 129–134.
10. Wohl JS. Nasojejunal feeding tube placement using fluoroscopic guidance: technique and clinical experience in dogs. J Vet Emerg Crit Care 2006; 16 (suppl 1): S27–S33.
11. Beal MW, Brown AJ. Clinical experience utilizing a novel fluoroscopic technique for wire-guided nasojejunal tube placement in the dog: 26 cases (2006–2010). J Vet Emerg Crit Care (San Antonio) 2011; 21: 151–157.
12. Tasaka K, Farrar JT. Intraluminal pressure of the small intestine of the unanesthetized dog. Pflugers Arch 1976; 364: 35–44.
13. Shikata J, Shida T, Amino K, et al. Experimental studies on the hemodynamics of the small intestine following increased intraluminal pressure. Surg Gynecol Obstet 1983; 156: 155–160.
14. Song EK, Mann FA, Wagner-Mann CC. Comparison of different tube materials and used of Chinese finger trap or four friction suture technique for securing gastrostomy, jejunostomy, and thoracostomy tubes in dogs. Vet Surg 2008; 37: 212–221.
15. Jergens AE, Morrison JA, Miles KG, et al. Percutaneous endoscopic gastrojejunostomy tube placement in healthy dogs and cats. J Vet Intern Med 2007; 21: 18–24.
16. Nieto JE, Dechant JE, Snyder JR. Comparison of one-layer (continuous Lembert) versus two-layer (simple continuous/Cushing) hand-sewn end-to-end anastomosis in equine jejunum. Vet Surg 2006; 35: 669–673.
17. Gandini M, Bertuglia A. In vitro evaluation of an inverted end-to-end equine jejunojejunal anastomosis using skin staples. Vet Surg 2006; 35: 678–682.
18. Gandini M. In vitro evaluation of a closed-bowel technique for one-layer hand-sewn inverting end-to-end jejunojejunostomy in the horse. Vet Surg 2006; 35: 683–688.
19. García-Olmo D, Payá J, Lucas FJ, et al. The effects of the pharmacological manipulation of postoperative intestinal motility on colonic anastomoses. An experimental study in a rat model. Int J Colorectal Dis 1997; 12: 73–77.
20. Wheeless CR Jr, Zanagnolo V, Bowers D, et al. The effect of growth hormone on the bursting strength of ileal anastomotic segments in radiation-injured rat bowel. Gynecol Oncol 1998; 70: 121–122.
21. Ikeuchi D, Onodera H, Aung T, et al. Correlation of tensile strength with bursting pressure in the evaluation of intestinal anastomosis. Dig Surg 1999; 16: 478–485.
22. Downey DM, Harre JG, Dolan JP. Increased burst pressure in gastrointestinal staple-lines using reinforcement with a bioprosthetic material. Obes Surg 2005; 15: 1379–1383.
23. Semevolos SA, Reed SK, Gamble K. In vitro bursting pressures of jejuna enterotomy closures in llamas. Vet Surg 2007; 36: 64–67.
24. Goto T, Kawasaki K, Fujino Y, et al. Evaluation of the mechanical strength and patency of functional end-to-end anastomoses. Surg Endosc 2007; 21: 1508–1511.
25. Saile K, Boothe HW, Boothe DM. Saline volume necessary to achieve predetermined intraluminal pressure during leak testing of small intestinal biopsy sites in the dog. Vet Surg 2010; 39: 900–903.
26. Otterson MF, Sarr MG. Normal physiology of small intestinal motility. Surg Clin North Am 1993; 73: 1173–1192.
Appendix 1—Raw data
Raw data used for statistical analyses in a study comparing intestinal luminal circumference, luminal area, and leakage pressure in intestinal specimens from dogs (n = 10) with jejunostomy tubes placed in vitro via 3 techniques (pursestring, Witzel, and seromuscular incision).
| Intestinal luminal area (cm2)* | Ratio of measurement at tube site and normal area | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Jejunostomy placement technique (group) | Dog | Body weight (kg) | Luminal circumference (mm) | Normal intestine | Insertion point | With the tube traveling in the lumen | Insertion site | Intraluminal site | Leakage pressure (mm Hg) | Leakage test rank |
| Witzel (seromuscular inversion) | 1 | 31.6 | 4.331 | 1.043 | 0.755 | 1.825 | 0.72 | 1.749 | 37 | 1 |
| 2 | 19 | 6.804 | 2.319 | 1.159 | 1.469 | 0.499 | 0.633 | 73 | 2 | |
| 3 | 22 | 5.688 | 1.323 | 1.077 | 0.77 | 0.814 | 0.582 | 252 | 3 | |
| 4 | 18 | 5.685 | 2.259 | 1.449 | 1.45 | 0.641 | 0.641 | 233 | 3 | |
| 5 | 21.4 | 4.011 | 1.146 | 0.885 | 0.648 | 0.772 | 0.565 | 192 | 2 | |
| 6 | 26.7 | 5.62 | 1.831 | 1.869 | 2.134 | 1.021 | 1.165 | 102 | 1 | |
| 7 | 20.4 | 4.785 | 1.748 | 1.036 | 0.823 | 0.592 | 0.47 | 54 | 2 | |
| 8 | 23 | 4.434 | 0.487 | 0.543 | 1.264 | 1.114 | 2.595 | 55 | 1 | |
| 9 | 28 | 5.814 | 1.925 | 1.804 | 1.3 | 0.937 | 0.675 | 88 | 3 | |
| 10 | 26 | 5.432 | 2.017 | 0.197 | 1.383 | 0.097 | 0.685 | 109 | 2 | |
| Seromuscular incision | 1 | 31.6 | 5.655 | 1.26 | 1.822 | 1.205 | 1.446 | 0.956 | 44 | 2 |
| 2 | 19 | 5.705 | 2.167 | 1.802 | 1.221 | 0.831 | 0.563 | 23 | 1 | |
| 3 | 22 | 5.859 | 2.12 | 1.365 | 1.729 | 0.643 | 0.816 | 95 | 1 | |
| 4 | 18 | 5.245 | 1.751 | 1.262 | 1.088 | 0.72 | 0.621 | 215 | 2 | |
| 5 | 21.4 | 3.71 | 0.83 | 0.67 | 0.846 | 0.807 | 1.019 | 115 | 2 | |
| 6 | 26.7 | 4.539 | 1.432 | 0.669 | 0.753 | 0.467 | 0.525 | 120 | 2 | |
| 7 | 20.4 | 5.496 | 2.191 | 1.347 | 1.954 | 0.614 | 0.891 | 49 | 1 | |
| 8 | 23 | 4.834 | 1.161 | 1.034 | 1.029 | 0.89 | 0.886 | 102 | 2 | |
| 9 | 28 | 5.722 | 1.533 | 1.259 | 0.501 | 0.821 | 0.327 | 33 | 2 | |
| 10 | 26 | 4.772 | 1.592 | 1.083 | 0.748 | 0.68 | 0.469 | 42 | 1 | |
| Pursestring | 1 | 31.6 | — | — | — | — | — | — | 240 | 3 |
| 2 | 19 | 5.262 | 1.433 | 1.511 | 2.433 | 1.054 | 1.697 | 193 | 3 | |
| 3 | 22 | 5.849 | 2.298 | 2.385 | 2.233 | 1.0378 | 0.971 | 130 | 2 | |
| 4 | 18 | 4.861 | 1.37 | 1.848 | 1.365 | 1.348 | 0.996 | 149 | 1 | |
| 5 | 21.4 | 4.929 | 1.456 | 1.143 | 1.452 | 0.785 | 0.997 | 25 | 1 | |
| 6 | 26.7 | 6.082 | 2.514 | 1.735 | 2.212 | 0.69 | 0.879 | 228 | 3 | |
| 7 | 20.4 | 5.243 | 2.001 | 1.962 | 1.858 | 0.98 | 0.928 | 116 | 3 | |
| 8 | 23 | 4.657 | 1.484 | 1.361 | 1.409 | 0.919 | 0.949 | 176 | 3 | |
| 9 | 28 | 4.726 | 1.288 | 2.3 | 1.867 | 1.785 | 1.449 | 30 | 1 | |
| 10 | 26 | 5.346 | 1.668 | 1.079 | 0.911 | 0.646 | 0.546 | 111 | 3 | |
For intestinal luminal area, 3 measurements were obtained: one with no tube inserted (baseline), one at the insertion site, and one where the tube was tunneled (or traveling intraluminally 1 cm aborad to the insertion site for the pursestring technique). These measurements were obtained for all 3 sites in an attempt to counteract variability within the specimens.
— = Data not available.
Appendix 2—Explanation of model fit
The intestinal luminal areas and leak pressures were analyzed with a commercial software package.i The baseline, insertion point, and tube measurements were each analyzed separately with a general linear models procedure.j The study design was a randomized complete block design with dog-to-dog differences included as a random block effect. A key assumption of this type of linear model is the assumption of constant variance. This assumption was checked by visually inspecting both residual plots and Q-Q plots after the models were fit. If the model fit well, least squares means were calculated for significant effects and compared by means of Tukey adjustment for post hoc multiple comparisons.
The overall model was not significant (P = 0.138) for intestinal luminal area at baseline, and neither the jejunostomy tube placement technique (P = 0.930) nor the random dog effect (P = 0.086) explained a significant amount of the variability in the data. The overall model was significant (P = 0.026) for the intestinal luminal area at the insertion point. This model fit very well, as indicated by the residual plot and the Q-Q plot. Whereas the dog effect was not significant (P = 0.093), the group differences were highly significant (P = 0.009). Examination of the least squares mean comparisons indicated that the pursestring technique (mean, 1.72 cm2) had a significantly higher mean than seromuscular inversion (mean, 1.08 cm2; P = 0.008) and seromuscular incision (mean, 1.23 cm2; P = 0.045).
Although the overall model was not significant (P = 0.158) for the intestinal luminal area with the tube traveling in the lumen, the lack of overall significance appeared to be driven by the lack of a significant (P = 0.443) dog effect. In fact, if the effect of having multiple dogs in the study was removed from the model, the overall model achieved significance (P = 0.021). However, because of the structure of the experiment, the dog effect remained in the model, although we did examine the group differences by least squares mean comparisons. The residual plots and Q-Q plots showed that the model fit the data well, and the homogeneous variance assumption appeared to be met. Pursestring technique once again had the highest mean (mean, 1.78 cm2), compared with seromuscular inversion (mean, 1.31 cm2) and seromuscular incision (mean, 1.11 cm2), although the only significant (P = 0.008) pairwise comparison was between pursestring technique and seromuscular inversion.
Linear models fit to intestinal luminal area and leak pressure data.
| Variable | df | F | P value |
|---|---|---|---|
| Intestinal luminal area at baseline (n = 17) | 11 | 1.78 | 0.138 |
| Dog | 9 | 2.13 | 0.086 |
| Placement technique | 2 | 0.07 | 0.930 |
| Intestinal luminal area at insertion point (n = 17) | 11 | 2.83 | 0.026 |
| Dog | 9 | 2.07 | 0.093 |
| Placement technique | 2 | 6.36 | 0.009 |
| Intestinal luminal area with tube (n = 17) | 11 | 1.70 | 0.158 |
| Dog | 9 | 1.05 | 0.443 |
| Placement technique | 2 | 5.01 | 0.020 |
| ln(leak pressure) (n = 18) | 11 | 1.28 | 0.312 |
| Dog | 9 | 1.23 | 0.338 |
| Placement technique | 2 | 1.50 | 0.251 |
df = Degrees of freedom.
When the model was fit to the raw leak pressure data, the residuals indicated the presence of heteroskedasticity. As the values of leak pressure increased, so did the variance. To adjust for this effect, the data were transformed by taking the natural logarthim. The residual plots and Q-Q plots indicated that the model fit the transformed data well, although it was not significant (P = 0.312) overall.
