Erschienen in:
01.12.2013 | Original Article
Predictive Factors for Early Mortality After Percutaneous Endoscopic and Radiologically-Inserted Gastrostomy
verfasst von:
Faidon-Marios Laskaratos, Martin Walker, Mary Walker, Janitha Gowribalan, Despoina Gkotsi, Victoria Wojciechowska, Ajay Arora, Anthony Jenkins
Erschienen in:
Digestive Diseases and Sciences
|
Ausgabe 12/2013
Einloggen, um Zugang zu erhalten
Abstract
Background
Gastrostomy insertion is of benefit to selected patients, but 30-day mortality is as high as 54 % in some patient groups. The current study examines risk factors associated with 30-day mortality in a cohort of patients who underwent percutaneous endoscopic gastrostomy (PEG) or radiologically-inserted gastrostomy (RIG) in a district general hospital over a 2.5 year period.
Methods
A retrospective review of case notes was performed for all patients who underwent a PEG (n = 53) or RIG (n = 40) insertion in the period January 2009–July 2011. PEG/RIG re-insertions were excluded. Demographic, clinical, and biochemical data were analysed. Multivariate regression analysis was used to identify risk factors for early mortality after gastrostomy insertion.
Results
The indications for gastrostomy insertion were similar in the PEG and RIG groups and included mainly dysphagic stroke, chronic neuromuscular disease and head and neck cancer. The patients in the RIG group were older and had a higher incidence of cardiovascular co-morbidities. The overall 30-day mortality was 11 % in the PEG and 40 % in the RIG group. The multivariate regression analysis suggested that cardiovascular co-morbidities and RIG insertion were independent risk factors for early mortality. The main cause of death 30 days after gastrostomy insertion was pneumonia, which was significantly more common in the RIG group.
Conclusions
Our data suggest that PEG should be the procedure of choice when considering gastrostomy insertion and RIG should be reserved as a second-line approach for cases in which PEG is technically difficult or contraindicated.