Colonoscopy has become a routine modality for the evaluation and treatment of colonic disorders in children. Although generally considered a safe procedure, colonoscopy has the potential for complications. Most informed consent forms describe adverse effects of sedative medicines, perforation, infection, and bleeding as possible risks of colonoscopy. However, the frequency of these complications remains unclear. There are limited pediatric data showing complication rates, and most prior studies involved small numbers of procedures. Furthermore, there might be additional complications that need to be added to consent protocols, because most current complication data are extrapolated from the much larger experience in adult populations.A review of pediatric literature demonstrated a wide array of potential complications and no large multicenter studies during the past 25 years. The largest study was conducted in [1978][1979] and included a survey of 1400 proctosigmoidoscopies performed during an 18-month period at 25 medical centers in the U.S. 1 This study was only published in abstract form and reported complications in 0.2% of all colonoscopy procedures. These complications included one hemorrhage requiring transfusion and one perforation. In the second largest study, Stringer et al 2 prospectively described a series of 250 flexible colonoscopies done in the United Kingdom during a 3.5-year period with no adverse events. Several other studies reported complications (Table 1) in pediatric colonoscopy, but the majority of these studies examined a small number of procedures.Determination of potential colonoscopy-related complications and their expected frequency in children can lead to enhanced risk-benefit analysis by physicians and patients and improved informed consent. There is a lack of systematic, multicenter studies in the pediatric population. Therefore, we conducted the current study to examine the frequency, type, and clinical determinants of immediate complications encountered during pediatric Address requests for reprints to: Kalpesh Thakkar, MD, 6621 Fannin St CCC 1010, Houston, TX 77030. kthakkar@bcm.tmc.edu; fax: 832-825-3633. Simple logistic regression was used to estimate the crude risk of cardiopulmonary events and gastrointestinal bleeding for variables that were identified as significantly associated with the outcomes during categorical analysis (P < .05). Multiple logistic regression models included all variables determined significant on univariate analysis. In constructing the models, we followed the commonly used heuristic of 10 outcome events for each explanatory variable. The adjusted risks estimate for each independent predictor variable was expressed as odds ratios. All analyses were performed by using SAS software (SAS Institute, Inc, Cary, NC).
HHS Public Access
ResultsDuring the study period, 8841 colonoscopies were recorded at 12 PEDS-CORI sites, and 1049 (11.9%) were excluded for incomplete procedure notes, duplicate records, or patient age older than 18 years. Thus, a total of 77...