The rate of complications caused by undetected lung misplacement of feeding tubes is relatively small (estimated at 0.01%) (Jones, 2020) when compared with the relatively common major complications caused by lung misplacement that is detected (0.5%) (Krenitsky, 2011). Based on the number of tubes used, an estimated 4000 complications due to identified lung misplacement can be expected annually in the UK (Krenitsky, 2011). Guided tube placement has the potential to almost eliminate these complications by pre-empting deep lung placement of the tube. One method, Cortrak, facilitates this by tracing the path of an electromagnet at the tip of the tube guide-wire, warning in real-time of potential deviation into the lung.
Unfortunately, up to the end of February 2016 there were 54 adverse events reported on the Manufacturer and User Facility Device Experience (MAUDE) database in the USA; 98% were lung misplacements, undetected by Cortrak operators, commonly resulting in pneumothorax (77%), pneumonitis (21%) and death (17%) (Bourgault et al, 2017). In a parallel report of MAUDE events, 17 of 25 events resulted in pneumothorax, despite only six being fed (Metheny and Meert, 2017). These data show that, unless misplacement is detected before the tube has advanced deeply into the lung, pneumothorax, pneumonitis and death are possible. Failure to detect lung misplacement has been reported to be as high as 2.1% with Cortrak (Bryant et al, 2015), similar to the 1.4% seen in blind placement (Krenitsky, 2011), and was thought to be due to inadequate training.
Training needs to be based on accurate interpretation of the electromagnetic (EM) trace. Manufacturer's guidance for confirming gastric placement is summarised in two criteria: ‘[a] For a correct nasogastric (NG) tube placement (in typical anatomy) the EM trace will progress straight down the vertical with no deviations above the horizontal midline,’ and [b] ‘the green dot and yellow trace should be in the bottom left quadrant’ (Avanos Medical Inc, 2018).
However, in a previous audit of UK-wide data, four of six undetected Cortrak lung misplacements had EM traces that largely overlapped with the screen area expected from a gastrointestinal (GI) placement and either met the manufacturer's guidance for confirming gastric placement or alternatively could be confused with misplacement of the receiver or unusual anatomy (Taylor et al, 2019). In retrospective reviews, when Cortrak GI traces were compared with the manufacturer's guidance, 38% were found to deviate by >5 cm from the sagittal midline, potentially indicating lung placement, and only 70% were accurately predicted to be intra-gastric by the tube's placement reaching the lower left quadrant (Taylor et al, 2017a; 2017b). In contrast, interpretation using the ‘GI flexure system’, adopted by the authors, accurately predicted all lung placements in both series (Taylor et al, 2017a; 2017b; 2020). This system involves identifying the pre-gastro-oesophageal junction (GOJ), gastric body, pyloric and superior duodenal flexures from their characteristic EM-trace shape.
For this article, the authors prospectively compared the GI flexure system with the manufacturer's guidance to determine the accuracy of trace interpretation.
Methods
The authors prospectively compared the accuracy of the ‘GI flexure system’ for interpreting the EM trace against the recommendations in the manufacturer's guidance. The cohort consisted of consecutive patients requiring intestinal tube placement admitted to the intensive care unit (ICU) at Southmead Hospital, Bristol. GI flexures were identified from trace characteristics as follows:
In addition, ST determined whether the chest-level trace was within 5 cm of the vertical midline and that the gastric body flexure was in the lower left anterior screen quadrant to meet the manufacturer's guidance for confirming gastric placement.
Possible artefacts due to receiver misplacement were identified from the trace being >5 cm rotated or displaced from the sagittal midline or the pre-GOJ or pyloric flexures being >5 cm below or above Cortrak's horizontal screen line, respectively. Subsequent analysis determined whether artefact affected trace interpretation. Trace GI flexures, midline deviation and screen quadrant were identified (ST) and compared using sensitivity and specificity analysis. Traces that are potentially difficult to interpret were highlighted. These traces were interpreted against the screen area in which historical GI placements had appeared (Taylor et al, 2017a), the GI flexure system and manufacturer guidance. The authors provide here a narrative comparison of which method could differentiate safe from unsafe tube placement.
These studies were audits of standard practice. All records and interventions were clinically necessary and anonymised.
Results
The authors observed 185 consecutive patients admitted to the ICU between 4 April 2016 and 1 November 2019 requiring primary nasointestinal (NI) tube placement and for which an X-ray was available to confirm the tube's position. Patient demography and disease characteristics are presented in Table 1.
| Parameter | Median or n | IQR or % | |
|---|---|---|---|
| Group | n | 185 | 74.3 |
| Age (years) | 52.4 | 36.6−67.7 | |
| Sex | Male | 132 | 71.4 |
| APACHE 2 | Score | 16 | 10−20 |
| Disease | Medical | 42 | 22.7 |
| Neurosurgical (non-trauma) | 30 | 16.2 | |
| Surgery (general) | 53 | 28.6 | |
| Trauma | 60 | 32.4 | |
| Consciousness | Awake | 35 | 18.9 |
| Sedated | 136 | 73.5 | |
| Unconscious | 14 | 7.6 | |
| Airway | Normal | 24 | 13 |
| Endotracheal | 144 | 77.8 | |
| Tracheostomy | 17 | 9.2 | |
| Tube placement | Day | 4 | 3−5.7 |
| Initially respiratory tract | 21 | 11.4 | |
| Stomach | 5 | 2.7 | |
| Duodenum part 1 | 9 | 4.9 | |
| Duodenum part 2 | 2 | 1.1 | |
| Duodenum part 3 | 9 | 4.9 | |
| Duodenum part 4 | 50 | 27.7 | |
| Jejunum | 110 | 58.7 |
IQR=interquartile range
Lung and receiver misplacement
Of 185 placements, 21 (11.4%) were suspected of initially entering the respiratory tract because of clinical symptoms (1.6%) or the EM trace was seen to be entering the left bronchus (3.2%), both bronchi (1.6%), or right bronchus (4.9%). No tube was left in the lung prior to feeding. Receiver unit misplacement occurred in 23.2% of traces (Table 2). In most cases traces appeared to be displaced by less than 5 cm. The cause of receiver misplacement, although not recorded, was commonly due to patient obesity, position or movement, making it difficult to maintain optimal position.
| Artefact | Type | n | % |
|---|---|---|---|
| None | 141 | 76.2 | |
| Rotation | Anti-clockwise | 13 | 7 |
| Clockwise | 5 | 2.7 | |
| High | 8 | 4.3 | |
| & left shift | 1 | 0.5 | |
| Displaced | Left | 13 | 7 |
| Right | 3 | 1.6 | |
| Reversed* | Patient prone | 2 | 1.1 |
Trace interpretation using manufacturer guidance
GI flexures were not always in their expected quadrants: pre-GOJ (top left), gastric body and pyloric (bottom left: manufacturer criterion [b] and superior duodenal (bottom right) (Table 3). Consequently, differentiating oesophageal from gastric body, pyloric and superior duodenal flexures by expected quadrant was only accurate in 74.3%, 81.1% and 94.9% of placements, respectively or, if restricted to traces without artefact, 74.6%, 79.4% and 96%, respectively. Specifically, 26.5% of traces were oesophageal (pre-GOJ flexure) on entry into the bottom left quadrant and 24.9% had reached the gastric body flexure in the top left quadrant.
| GI flexure | n | Quadrant | Sensitivity | Specificity | Accuracy | |||
|---|---|---|---|---|---|---|---|---|
| % | % | |||||||
| Top | Bottom | |||||||
| Right | Left | Right | Left | |||||
| Pre-GOJ | 185 | 7.6 | 63.2 | 2.7 | 26.5 | |||
| Gastric body | 185 | 0.0 | 24.9 | 0.0 | 75.1 | 75.1 | 73.5 | 74.3 |
| Pyloric | 180 | 0.0 | 3.3 | 7.8 | 88.9 | 88.9 | 73.5 | 81.1 |
| Superior duodenal | 168 | 6.0 | 0.6 | 92.3 | 1.2 | 92.3 | 97.3 | 94.9 |
NB Not all 185 tube traces reached the pyloric or superior duodenal flexures
Lung vs GI traces: manufacturer vs GI flexure guidance
Figure 1 shows EM trace ID numbers, including L/R indicating left/right lung, from the patient's perspective. EM traces are shown on ‘Anterior’ and ‘Lateral’ screens relative to the chest, xiphisternum and abdomen.
Traces of lung placements were almost entirely within anterior and lateral screen areas covered by traces from placements confirmed, from historical data, to be safely GI (Taylor et al, 2017a) (Figure 1a). In addition, although some lung trace deviations from the midline, on the Anterior screen, were acute or large (826L & R), many had a shallower angle, closer to the midline (Figure 1a. 645, 705) than GI trace deviations (Figure 1b, 703, 727). None of the lung placements reached the lower left quadrant. However, maximal chest-level GI deviation from the sagittal midline was >5 cm in 42% and even after eliminating possible artefacts 33% deviated >5 cm (median 3.5 cm, IQR: 2-7 cm). 3.2% of GI traces deviated >5 cm at a similar level to the bronchi. Finally, traces from patients with anatomical GI defects, although safe, deviated from the sagittal midline and sometimes never entered the lower left quadrant (Figure 1c, 796). Overall, differentiation of lung from GI traces can be difficult when based on screen position alone.
In contrast, GI traces, without anatomical defect, showed all the expected GI flexures regardless of the presence of screen artefacts (Table 4). Where anatomical defects dilated the stomach into the chest cavity (Figure 1c, 796 paralysed hemi-diaphragm, 819 hiatus hernia), the lateral trace did not always steadily shallow through the stomach, nor did the anterior trace always become relatively close to the sagittal midline where the pre-GOJ flexure was expected. However, all anterior GI traces eventually moved to the right through the pylorus and deepened laterally on entering the duodenum, then turned through the superior duodenal flexure. The GI flexure system confirmed GI placement and differentiated pre-GOJ from gastric body flexures in all except one trace; that trace was confirmed as GI at the superior duodenal flexure. All lung placements were identified because they failed to meet all GI flexure criteria.
| GI flexure | GI | Lung | Non-GI | Sensitivity | Specificity | Accuracy | |
|---|---|---|---|---|---|---|---|
| True | False | True | False | ||||
| Pre-GOJ | 163 | 0 | 21 | 1* | 99.4 | 100 | 99.5 |
| Gastric body | 163 | 0 | 21 | 1* | 99.4 | 100 | 99.5 |
| Pyloric | 159 | 0 | 20 | 1* | 99.4 | 100 | 99.4 |
| Superior duodenal | 151 | 0 | 17 | 0 | 100 | 100 | 100 |
NB Not all 185 tube traces reached the pyloric or superior duodenal flexures. All tubes were ultimately GI, but the GI traces from patients whose tube initially entered the lung are not shown to avoid double counting in the same patient
The trace had not met GI flexure criteria until it reached the superior duodenal flexure
Discussion
Primary outcome
In our opinion, manufacturer guidance fails to accurately predict gastric position; it appears that criterion a falsely predicts potential lung placement in 33-42%, including 3.2% at the level of bronchi, and criterion [b] fails to predict gastric placement in about 25%. In contrast, the GI flexure system confirmed 99.5% of GI placement on reaching the gastric body flexure and 100% by the superior duodenal flexure; lung placement was always detected by an absence of GI flexure characteristics.
Cortrak trace interpretation
This prospective study appears to validate previous findings (Taylor et al, 2017a; 2017b) that the manufacturer's guidance is too simplistic to accurately confirm gastric position of a feeding tube. Accuracy in confirming post-oesophageal GI placement by quadrant only improved when accepting that a tube must reach the pyloric or superior duodenal flexures. However, if intending gastric placement, this approach would risk unintended intestinal placement. In addition, the high proportion of safe GI placements failing to meet the manufacturer's guidance risks unnecessary procedure failure. More importantly, traces reaching the lower left quadrant (criterion b) were oesophageal in 26.5% and, in a previous study, comprised four of the six lung misplacement ‘never events’ in the UK (Taylor et al, 2019). Such overlap makes operator trace interpretation difficult.
Conversely, the authors validated the accuracy of GI flexures in predicting gastric placement in ≥99.4% of cases. In the one exception, GI placement was confirmed on reaching the superior duodenal flexure. All lung placements were confirmed by the absence of GI flexures. Receiver misplacement made little difference to the results and may be impossible to avoid as specific care had already been taken to avoid this.
Expert centres have suggested that up to 75 supervised EM-guided tube placements are required to achieve competence (McCutcheon et al, 2018). In contrast, the manufacturer's training for operators consists of a 2-hour training session and a minimum of three supervised placements, of which two may be re-traces of previous placements (Avanos Medical Inc, 2018). Healthcare centres are expected to extend the training but there is no manufacturer or other published guidance on which to base this.
Implications for practice
Evidence-based guidelines are essential to identify feeding tube misplacement and prevent the potentially life-threatening complications associated with the misplacement per se, or due to delivery of feed or drugs through a misplaced tube. Operators, predominantly nurses, must be expert at interpreting guided tube placement and knowing when to seek alternative confirmation because they are responsible for confirming whether the tube is safe to use. This and previous studies were prompted by an investigation into a fatal, undetected Cortrak tube misplacement and the need for a systematic guide for Cortrak trace interpretation.
Conclusion
Manufacturer guidance did not accurately differentiate tube placement in the lung from gastric placement in at least 25% of traces in this study. It is particularly prone to failure when anatomy is non-typical or abnormal. Conversely, the ‘GI flexure system’ used by the authors identified all but one GI placement as GI when the trace reached the gastric body flexure; the one exception was confirmed as GI at the superior duodenal flexure. The GI flexure system identified all lung misplacements by the absence of GI flexures.