Abstract
Introduction: Percutaneous Endoscopic Gastrostomy (PEG) is accepted as an efficient method to provide long-term enteral nutrition. PEG accidental dislodgement (device exteriorization confirmed by expert evaluation) rate is high and can lead to major morbidity.
Objective: To identify independent risk factors for PEG accidental dislodgement.
Methods: Retrospective, single-center study, including consecutive patients submitted to PEG procedure, for 38 consecutive months. Every patient had 12 months minimum follow-up after PEG placement. Univariate analysis selected variables with at least marginal association (p < .15) with the outcome variable, PEG dislodgement, which were included in a logistic regression multivariate model. Discriminative power was assessed using area under curve (AUC) of the receiver operating curve (ROC).
Results: We included 164 patients, 67.7% (111) were female, mean age was 81 years. We report 59 (36%) PEG dislodgements, of which 13 (7.9%) corresponded to early dislodgements. The variables with marginal association were hypoalbuminemia (p = .095); living at home (p = .049); living in a nursing home (p = .074); cerebrovascular disease (CVD) (p = .028); weight change of more than 5 kg, either increase or decrease (p = .001); psychomotor agitation (p < .001); distance inner bumper-abdominal wall (p = .034) and irregular appointment follow-up (p = .149). At logistic multivariate regression, the significant variables after model adjustment were CVD OR 4.8 (CI 95% 2.0–11.8), weight change OR 4.7 (CI 95%1.6–13.9) and psychomotor agitation OR 18.5 (CI 95% 5.2–65.6), with excellent discriminative power (AUC ROC 0.797 [CI95% 0.719–0.875]).
Conclusion: PEG is a common procedure and accidental dislodgement is a frequent complication. CVD, psychomotor agitation and weight change >5 kg increase the risk of this complication and should be seriously considered when establishing patients’ individual care requirements.
Authors’ contributions
Rui de Sousa Magalhães was involved in all stages of the manuscript construction: from the conduct, reporting, conception and design, to the acquisition of data or analysis, literature review and writing the final paper; Tiago Cúrdia Gonçalves was involved in the conception, design and discussion of the final paper to be published; Bernardo Sousa Pinto was involved in the statistical analysis and in the discussion of the final paper to be published; Bruno Rosa was involved in the conception, design and discussion of the final paper to be published; Carla Marinho was involved in the discussion of the final paper to be published; José Cotter was involved in the conception, design and discussion of the final paper to be published.
Disclosure statement
The authors report no conflict of interest.