Development of an Evidence-Based Surgical Decision Making Algorithm for Bladder Exstrophy Related Continence Procedures

Bladder exstrophy (BE), is a rare congenital malformation that occurs 1:40,000 live births, (Nelson, 2005). Occurring during the tenth week of pregnancy, BE manifests as failure of the abdominal wall to close and results in the protrusion of the anterior bladder through the lower abdominal wall. 

In the modern staged repair, the abdominal wall portion of the defect is ideally closed in the first 72 hours of life.  The staged approach then requires a continence procedure, typically undertaken in the middle elementary years.  There are two main continence procedures, the bladder neck reconstruction (BNR), and the continent urinary diversion, (CUD). The former is designed to achieve continence through urethral voiding while the latter provides continence with urinary emptying occurring via a continent catheterizable channel created through the abdominal wall into the bladder.  While CUD is performed by a large number of surgeons throughout the world for a variety of diagnosis, BNR is performed by only a few select individuals due to it’s high level of technical difficulty.

The use of functional reconstruction in bladder exstrophy has resulted in dramatic improvement in the success of reconstruction, (Gearhart, Mathews, 2012).  However, even in the most experienced and successful hands BNR failure rates are approximately 30%.  The procedure requires a high level of technical surgical acumen and it is important to remember that urinary retention is the most common symptom after BNR surgery, (Surer, et al., 2001).  Children plagued by post BNR urinary retention typically experience pain and suffering due to bladder spasms, readmissions, increased anesthesia inductions, increased medication requirements, prolonged length of stay, and recalcitrant incontinence. We know that incontinence may have a negative impact on social function and self esteem, (Gearhart, 2001).

It is well documented that two of the most reliable predictors of eventual urinary continence are the size of the bladder template at birth and a successful primary closure, (Gearhart, Mathews, 2012). However, through the appraisal of additional measures that began to be formally evaluated at the author's institution in 2005, there are a number of additional variables significantly affecting continence outcomes post BNR.   These include but are not limited to pelvic floor strength, pelvic floor relaxation ability, and post voiding urinary residual.  Currently, an evidence-based tool to incorporate all variables in the surgical decision making process does not exist.  Lack of such a tool increases the risk of incorrect surgical candidate choice, thus increasing the risk of surgical failure and continued incontinence. 

An exhaustive literature search focused on decision-making and algorithm development resulted in a modest but valuable yield of 36 total articles.  Further critique of quality and appropriateness further narrowed the findings to 15 readily translatable articles. The final cut included no Level I evidence, and only two Level II articles both describing quasi-experimental studies.  The majority of the evidence was Level III, non-experimental , represented by 10 articles, a mix of A and B quality. No Level IV evidence was identified but there were 3 Level V articles that provided evidence on algorithm development.  

The evidence discovered through the search process established that there is a means by which to extract and evaluate the data that exists in the patient charts at the author's institution in order to create decision making points to guide and aid providers in an exceedingly difficult decision.

The evidence repeatedly pointed to data mining (DM) as a means to extract patient data in order to ascertain which data points end in failed BNR and which result in a successful BNR outcome. The protection of patients, their privacy and their rights remain the utmost importance.  The evidence overwhelmingly favors retrospective chart review, extracting data and utilizing appropriate statistical analysis to evaluate data in order to draw meaningful conclusions.  Most commonly cited in the literature was multivariate logistic regression analysis.  Also utilized were Fisher’s exact test and Wilcoxon rank sum.  Some univariant analysis such as Mann-Whitney U test was utilized to compare attributes of survivors vs. non-survivors in one cancer related study.  Also in that study, tumor grade and survival was evaluated through Pearson X2.  However, such studies dealt with concrete outcome measures such as tumor markers.  Decision making points in bladder exstrophy patients is less exacting and will include variability from patient to patient.  Recursive partitioning was evaluated in several studies, with the most complete description from Shaikh, N., et al., 2012 , in which they developed a decision making tree through CART analysis, (one specific tool that utilizes recursive partitioning), of a previous cohort study and then validated the tree by applying it to a different data set.

Evaluation of numerous studies, their statistical analysis, and their application setting clearly indicate that CART analysis is likely the most applicable and effective tool to address the challenge of surgical decision making faced by Johns Hopkins department of pediatric urology.   Evaluation of various multivariate regression models illustrates that the goal will be to evaluate multiple variables and analyze each separately but also in simpatico, as they relate to one another.  Such analysis will clarify and define the importance and impact of each variable as a stand alone, as well as in conjunction with one another.  Traditional statistical methods are poorly suited for this sort of multiple comparison, (Lewis, R., 2000).  According to Lewis, CART analysis is superior to many traditional multivariate techniques in that it does not require that predictor variables be evenly distributed and still performs well when different groups of patients have significantly different degrees of variance.   In addition, traditional multivariate regression models in the literature concentrate on probabilities, which is not consistently applicable in clinical practice.  In addition CART analysis is able to accurately manage a large number of variables in building a model, whereas traditional linear regression has difficulty with large numbers of variables.  

The evidence clearly points in the direction of algorithm creation derived out of DM from the only existing set of inclusive data points available, those at the author's institution department of pediatric urology.  With data mining complete and statistical analysis firmly underway, the completion of the evidence-based surgical decision making algorithm is on time and on target to be complete and ready for presentation, dissemination and implementation by March 15, 2014.

 References

Gearhart, J. (2001). The bladder exstrophy-epispadius-cloacal exstrophy complex. In J. Gearhart, R. Rink & P. Mouriquand (Eds.), Pediatric urology(pp. 511). Philadelphia, Pennsylvania: WB Saunders Company.

Gearhart, J., & Matthews, R. (2012). Exstrophy-epispadius complex. Campbell's urology(pp. 3325). Philadelphia, Pennsylvania: Elsevier Saunders.

Lewis, R.J. (2000).  An Intorduction to Classification and Regression Tree (CART) Analysis. Annual Meeting of the Society for Academic Emergency Medicine, San Francisco, CA, 2000.

Nelson, C., Dunn, R., & Wei, J. (2005). Contemporary epidemiology of bladder exstrophy in the united states. The Journal of Urology, 173(5), 1728-1731.

Novak, T., Costello, J., Orosco, R., Sponseller, P., Mack, E., & Gearhart, J. (2009). Failed exstrophy closure: Management and outcome. Journal of Pediatric Urology, 6, 381. doi: 10.1016/j.jporol.2009.10.009

Shaikh, N. M.,MPH., Hoberman, A. M., Rockette, H. P., & Kurs-Lasky, M. (2012). Development of an algorithm for the diagnosis of otitis media. Academic Pediatrics, 12(3), 214.

Surer, I., Baker, L., Jeffs, R., & Gearhart, J. (2001). Modified young-dees-leadbetter bladder neck reconstruction in patients with successful primary bladder closure elsewhere: A single institution experience. The Journal of Urology, 165(6), 2438-2440. doi: 10.1016/s0022-5347(05)66224-6

Yerkes, E., Adams, M., Rink, R., Pope, J., & Brock, J. (2000). How well do patients with exstrophy actually void? The Journal of Urology, 164(3), 1044-1047. doi: 10.1016/s0022-5347(05)67426-1