Comparison of Gastrostomy Techniques in Stroke Patients With Dysphagia: An Entropy-Balanced Analysis

Abstract

Introduction

Enteral nutrition is commonly placed via percutaneous endoscopic gastrostomy (PEG) or radiologically inserted gastrostomy (RIG) techniques. This study compared perioperative outcomes of PEG and RIG in adults with dysphagia caused by cerebral infarction.

Methods

Adult stroke patients who underwent either PEG or RIG between 2018 and 2020 at a tertiary care center were reviewed retrospectively. Differences in baseline characteristics between PEG and RIG patients were adjusted using entropy-balanced weights. Multivariable weighted logistic and linear regressions were subsequently developed to evaluate the independent association between RIG and outcomes of interest.

Results

217 stroke patients met inclusion criteria, of whom 37 (17.0%) received PEG and 180 (83.0%) received RIG. Compared to PEG, patients with RIG were more commonly Medicare beneficiaries and had a higher burden of comorbidities as measured by the Charlson comorbidity index. Time to achieve goal feeds was comparable between PEG and RIG (3 d [interquartile range 2-5] vs 4 d [interquartile range 3-5], respectively, P = 0.059). After multivariate adjustment, RIG was associated with significantly lower odds of reoperation (adjusted odds ratio [AOR] 0.10, 95% CI 0.02-0.50, P = 0.005), cerebrovascular accident (AOR 0.24, 95% CI 0.00-0.74, P = 0.030), and intensive care unit admission (AOR 0.14, 95% CI 0.03-0.70, P = 0.017). Risk factors for in-hospital mortality among RIG included arrhythmia (AOR 6.54, 95% CI 1.67-15.48, P = 0.009), myocardial infarction (AOR 4.78, 95% CI 2.25-10.23, P = 0.009), and obesity (AOR 4.48, 95% CI 1.03-9.61, P = 0.047).

Conclusions

While both techniques are effective methods of enteral feeding in stroke patients, RIG may confer lower perioperative morbidity. Local referral patterns and individual patient comorbidities could influence outcomes following PEG or RIG, necessitating careful patient selection.