Evaluation of a device for insertion of a small intestinal feeding tube in the critically ill: A prospective multicentre international cohort study

Background Critically ill patients sometimes require insertion of a feeding tube into the small intestine. This study evaluated the ease of placement of a proprietary small intestinal feeding tube incorporating a distal tip camera. Methods investigator-initiated, industry-funded, prospective cohort study was conducted at four intensive care units across two countries. Patients were eligible if the treating clinician requested small intestinal feeding tube insertion. Operators were physicians or senior dietitians, and attempts were categorised as "inexperienced" (≤4 attempts) or "experienced" (≥5 attempts). The primary outcome was the overall probability of successful small intestinal placement. Secondary outcomes included procedural time and the posterior probabilities of success according to operator classification and experience, estimated using Bayesian hierarchical regression models incorporating minimally informative priors and random intercepts for clustering within operators. Results The device was used by 11 operators in 68 patients. Prior 24-h cumulative gastric residual volumes were 454 (standard deviation: 453) millilitres. The overall probability of success was 75% (95% credible interval: 61–86%). Experienced users and inexperienced users were successful in 30 of 38 (79%) and 21 of 30 (70%) attempts, respectively. Overall median (interquartile range) procedural times for successful versus unsuccessful placement were 23 (10–35) and 30 (20–42) minutes, respectively. In a multivariable analysis, the posterior probabilities of improved success (odds ratio >1) were 83% (gastrokinetic drug prescription), 82% (operative versus nonoperative admission), and 65% (device experience). Conclusions The probability of successful small intestinal placement of feeding tubes by motivated clinicians in a critical care environment approximated the prespecified threshold of 80%, without the need for extensive specific device experience.