Abstract

Purpose: Radiological gastrostomy catheter insertion is a well-established procedure to provide enteral nutrition to patients unable to do so themselves. It carries risks and an associated 30 day mortality rate. After referring clinicians perceived increase in early mortality after gastrostomy insertion, we aimed both to determine whether this was a true difference or random variation and how best to communicate this.

Material and Methods: Data for 696 consecutive gastrostomy insertions in 661 patients performed between 2014 and 2020 were collected retrospectively. A literature search was performed for gastrostomy series with 30 day mortality data. Mortality variation was compared to random variation and visualised with a funnel plot. Mortality rate was compared to existing literature. Survival analyses were performed to compare survival among indications.

Results: We found 30 day mortality variability to be indistinguishable from random variation and visualised it to demonstrate no significant outlier and also showed that it compared favourably to published literature. Patients with motor neurone disease had a significantly worse survival than those with ear, nose and throat cancers at 30 days and one year (p<0.05).

Conclusion: Mortality after gastrostomy catheter insertion varies over time but statistical methods can be used to help assess whether the variability is random. A funnel plot is an ideal visualisation for such data and should not be limited to comparing institutions or operators. Some patient groups have significantly worse outcomes after gastrostomy than others.

Keywords: Gastrostomy; Mortality rate; Statistics; Radiological