Validation of image interpretation for direct vision-guided feeding tube placement

Unguided (blind) tube placement commonly results in lung (1.6%) and oesophageal (5%) misplacement, which can lead to pneumothorax, aspiration pneumonia, death, feeding delays, and increased cost. Use of real-time direct vision may reduce risk. We validated the accuracy of a guide to train new operators in the use of direct vision-guided tube placement.

Using direct vision, operators matched anatomy viewed to anatomical markers in a preliminary operator guide. We examined how accurately the guide predicted tube position, specifically whether respiratory and gastrointestinal placement could be differentiated.

A total of 100 patients each had one tube placement. Placement was aborted in 6% because of inability to enter or move beyond the oesophagus. In 15 of 20 placements in which the glottic opening was identified, the tube was maneuvered to avoid entry into the respiratory tract. Of 96 tubes that reached the oesophagus, 17 had entered the trachea; all were withdrawn pre-carina. One or more specific characteristics identified each organ, differentiating the trachea-oesophagus (P < 0.0001), oesophagus-stomach, and stomach-intestine in 100%. End-of-procedure tube position was ascertained by pH ≤4.0 (gastric) of aspirated fluid and/or x-ray (gastric or intestinal). In patients with a trauma risk (13%), it was avoided by identification that the tube remained within the nasal, oesophageal, or gastric lumen.

Operators successfully matched anatomy seen by direct vision to images and descriptions of anatomy in the "operator guide." This validated that the operator guide accurately facilitates interpretation of tube position and enabled avoidance of lung trauma and oesophageal misplacement.