Construct Validity of the Chicken Crop Model in the Simulation of Laparoscopic Pyeloplasty

Introduction

D ismembered pyeloplasty has been considered the gold standard for management of ureteropelvic junction obstruction (UPJO), because this procedure is almost universally applicable to different clinical scenarios. Since the first cases of laparoscopic pyeloplasty (LP) were reported by Schuessler and associates in 1993, ¹ this technique has developed worldwide as a minimally invasive alternative to match the success rates of open surgery with decreased morbidity, shorter hospital stay, shorter convalescence, and improved cosmetic outcome in experienced hands. ^2,3 Because of the inherent technical complexity and long learning curve, however, LP has only been performed at a handful of medical centers by proficient urologists.

Advances in laparoscopic surgery have brought a change in training methodology. Skill laboratory has become an essential and indispensable component of laparoscopic training programs. Training surgeons outside the operating room translates into better performance and fewer complications. Use of specific training models can be an effective way for surgeons to enhance their laparoscopic skills before operating on patients.

Several training models for LP have been developed in recent years. ^{4 –11} Among them is a training model using chicken crop reported by Ramachandran and colleagues. ¹¹ We have used this model in our workshop and assessed its construct validity—the ability to discriminate between inexperienced and experienced subjects.

Materials and Methods

Model

A dead, plucked chicken was purchased from a local market. A middle longitudinal incision was made through the neck and thorax to expose the chicken crop and esophagus. The lower segment of the esophagus was ligated, and the crop was thoroughly cleaned and filled with water to simulate the dilated human renal pelvis. The upper segment of esophagus was used to simulate the human ureter. The chicken was then placed in a left lateral position within a pelvic trainer to simulate a right-sided ureteropelvic junction (UPJ) (Fig. 1).

OPEN IN VIEWER

FIG. 1.

The chicken crop was used to simulate the dilated human renal pelvis, and the upper segment of esophagus was used to simulate the human ureter.

A laparoscopic dismembered Anderson-Hynes pyeloplasty was performed using 4-0 polyglactin sutures (Fig. 2 A–C). At the end of the procedure, 20 mL of water was infused via the upper segment of the esophagus to detect leakage from the anastomosis. The quality of the anastomosis was assessed by an independent observer, who was blinded to the subjects. The criteria used to analyze the quality included exact tissue sutured, equal bite on each side, equal stitch intervals, no tissue tear, and a water-tight anastomosis. The total quality score ranged from 0 to 10 (0 to 2 for each item); the higher the score, the higher the quality.

OPEN IN VIEWER

FIG. 2.

(A) The redundant portion of the “pelvis” was excised, and the “ureter” was spatulated. (B) Posterior wall of the anastomosis was completed. (C) Completed anastomosis.

Results

Those in the most experienced group completed the procedure in a mean time of 33.80 minutes. Those in the less experienced group averaged 55.20 minutes, and those in the group with no experience averaged 92.60 minutes. There was a significant difference between the groups using one-way ANOVA (P<0.001). The Tukey multiple-comparisons test showed a significant difference between the most experienced group and the less experienced group (P<0.01), between the less experienced group and the group with no experience (P<0.001), and between the most experienced group and the group with no experience (P<0.001).

With regard to quality score, one-way ANOVA also showed a significant difference between the groups (P<0.001). The Tukey multiple-comparisons test showed a significant difference between the less experienced group and the group with no experience (7.2 vs 4.0, P<0.01), and also between the most experienced group and the group with no experience (9.0 vs 4.0, P<0.001). There was, however, no significant difference between the most experienced group and the less experienced group (9.0 vs 7.2, P>0.05) (Fig. 3).

OPEN IN VIEWER

FIG. 3.

Operative time and quality scores for groups after completion of the procedure. Values are expressed as mean±standard error of the mean.

Discussion

Although it is believed that LP is rapidly moving toward replacing open surgery as the gold standard in the management of UPJO, ^3,12 the great technical complexity and steep learning curve impede its extensive application. Operators have to complete several complex steps in a limited working space, including the delicate tailoring of the ureter and renal pelvis, intracorporeal suturing and knotting, prevention of kinking of the ureter, and antegrade stent placement.

Urologic residents need proper training to perfect their technique to the level of a more experienced laparoscopic surgeon before they practice LP in the operating room. Although pelvic trainers can provide the necessary basic-skills training for laparoscopic surgeons, specific training models are needed to help them grasp those difficult techniques in a shorter time. A valid simulation should provide an environment as close as possible to reality, must mimic the visual–spatial and real-time characteristics of the procedure, and must provide realistic haptic feedback. ¹³ Convenience and cost should also be taken into consideration. Several training models for LP had been developed, including animal models and in vitro training models.

Live animal models could provide sufficiently realistic simulation of laparoscopic conditions such as pneumoperitoneum, blood circulation, bleeding, and anatomic environment. ¹⁴ McDougall and collaborators ⁴ developed a porcine model in 1997. In this model, unilateral UPJO was created by ligating the UPJ over a 5F catheter for 6 weeks. The time-consuming preparation made this model both inconvenient and costly. Fu and colleagues ⁵ used a short segment of porcine small intestine to simulate the enlarged renal pelvis. The intestinal wall is much thicker than the renal pelvis wall, however, and intestinal mucosa is much different from urothelium. Moreover, live animal models need a complete set of equipment and instrumentation, anesthesia, and care. Ethical consideration and costs also limit the use of live animals in many countries and regions.

In vitro training models present obvious advantages with regard to availability, versatility, low cost, and absence of the need for extreme care and anesthesia. The LapED 4-in-1 model developed by Abraham and colleagues ⁶ provided content and face validity for both LP and vesicourethral anastomosis. The model is inexpensive and reusable, but not available in most countries and regions currently. Raza and coworkers ⁷ also introduced an LP model using a surgical glove. Nevertheless, synthetic models do not handle like live tissue at present.

Models composed of ex vivo animal organs share similar qualities and appearance with humans. In 2006, Ooi and associates ⁸ developed a simple model using chicken skin that had the advantage of low cost and ready availability. The tough chicken skin makes needle insertion very difficult, however, and tissue tear never happens even with brutal manipulation. Trainees could not comprehend the fragility of the renal pelvis and ureter, and therefore learn to manipulate tissue meticulously with this model.

Yang and coworkers ⁹ used a carp swim bladder and porcine ureter to construct a training model that demonstrated satisfactory fidelity and practicality; however, the LP procedure cannot be simulated completely because the ureter is separate from the “renal pelvis” at the beginning, and the swim bladder is too thin so that it is easily torn during suturing. Moreover, obtaining the fresh swim bladder and porcine ureter is not easy. A porcine bladder and urethra were used for the UPJ simulation in the training model developed by Teber and colleagues. ¹⁰ Trainees could practice the whole LP procedure with this model. The porcine bladder wall, however, is much thicker than the human renal pelvis wall, and the caliber of porcine urethra is much larger than that of the human ureter. All of these make this model less of fidelity.

In the model developed by Ramachandran and colleagues, ¹¹ the chicken crop and esophagus share similar size, appearance, flexibility, thickness, and texture with human dilated renal pelvis and ureter. This model can be used to simulate both left and right UPJ. It allows trainees to practice the complete LP procedure, and provides sufficient surrounding tissue for preliminary dissection and mobilization, adding to the realistic nature of the simulation. ¹¹ Practicing with this model, trainees had to manipulate tissue gently and exactly to get satisfying tailoring, perform sutures from many different angles, and handle the sutures without getting them tangled. After practicing, trainees could get acquainted with the steps of LP, enhance their manual dexterity and coordination, and improve their skill of intracorporeal suturing and knotting. It may help the trainees shorten the learning curve and improve their operative performance.

Operative time is the most common objective measure to evaluate proficiency and to differentiate between experienced and inexperienced subjects. ^15,16 Some studies used volume of leakage as a single objective measure to evaluate anastomotic quality. ¹⁵ The quality of the UPJ anastomosis, however, is closely related to factors such as the distance between sutures, placement of the first stay suture without tension, and the positioning of a Double-J stent. ⁵ As a single factor, leakage volume is not enough to determine the anastomotic quality, so a multiple-criteria assessment was used to evaluate the anastomotic quality. ^7,11

An ideal training model should have subjective validities such as face validity and content validity, and objective validities such as concurrent validity, predictive validity, and constructive validity. Construct validity is a mandatory, and one of the most valuable, assessments of a simulator before its acceptance as a competency-evaluating device. In this regard, the simulator must be able to distinguish the experienced from the inexperienced surgeon. ¹⁷

We tested this model for construct validity by examining whether operative time and quality scores were consistent with the degree of experience of the surgeons. In this study, there is a strong negative correlation between time to complete the procedure and subjects' experience. The differences in time to complete the procedure were statistically significant between the three groups.

A positive correlation between quality scores and subjects' experience was also found. A statistical difference in quality score was shown between the most experienced group and the group with no experience and between the less experienced group and the group with no experience. Although the most experienced group had a higher average quality score than that of the less experienced group (9.0 vs 7.2), we did not find a significant difference. This may be because of the small sample size of this study. The difference in operative time and quality score proved that this model was able to distinguish between subjects with different levels of experience and consequently has good construct validity.

This model is inexpensive and convenient, because a dead chicken provides all materials we need. This model, however, may be unavailable in some countries or regions, because of the restriction of chicken viscera.

The predictive validity (the degree of concordance between the outcomes on the training model and real-life performance) of this model should be assessed in future studies.

Conclusion

The chicken crop model exhibits good construct validity. It is relatively cost-effective and readily available, and can be used to reproduce the technical complexity of LP.