M. Rojnica1, C. S. Lyttle2, G. Z. Mak1, M. B. Slidell1 1University Of Chicago,Pediatric Surgery,Chicago, IL, USA 2University Of Chicago,Center For Health And The Social Sciences,Chicago, IL, USA
Introduction:
Thoracoscopic repair (TR) of esophageal atresia (EA) has gained acceptance as an alternative to the gold-standard of open repair (OR) via thoracotomy. To date, studies comparing these approaches have been retrospective in nature and typically conclude with a call for prospective randomized trials. Currently, no consensus exists on selection criteria for TR versus OR. Using published results from EA repair as well as national outcomes data, we propose selection criteria for considering TR, but also hypothesized that the sample size required for a prospective randomized trial comparing the two approaches is prohibitively large, making such a trial infeasible.
Methods:
We identified 1109 neonates with EA in the 2009 Kids' Inpatient Database (KID). Potential selection criteria which may preclude a thoracoscopic approach were developed based on patient weight and the absence of major cardiac anomalies. Two objective endpoints for power calculations we chosen: anastomotic leak and mortality. To compare TR versus OR we estimated a mean leak rate of 12.2% vs. 10.4% and mortality rates of 0.9% vs. 1.1% respectively based upon contemporary published data.
Results:
The mean birth weight of our cohort was 2476 grams. Proposed inclusion criteria were compared. Cohort number one consisted of weight > 2kg and no major cardiac anomalies, and cohort number two was patients with a weight > 3kg and no major cardiac anomalies. This yielded cohorts where 72% of the infants were >2kg and 27% were >3kg. When additional cardiac exclusion criteria were applied 58% of >2kg infants and 21% of >3kg infants met our criteria for TR of esophageal atresia. Even with the liberal inclusion criteria of patients >2kg and no major cardiac anomalies, our power calculations show that a prospective randomized trial would require 8,148 patients per study arm for a total of 16,296 patients in order to detect a difference in leak rates (p >0.05) and 130,636 patients to detect a difference in mortality between the two techniques.
Conclusion:
With approximately 1100 EA births annually in the USA, a prospective randomized trial using liberal inclusion criteria requires a prohibitively large sample size. Use of conservative selection criteria of weight >3kg and no major cardiac anomalies would make this study even less feasible.