Introduction: Decisions to pursue enteral feeding with percutaneous endoscopic gastrostomy tube (PEG) may lead to prolong hospitalization in stroke patients. Prediction models in intracerebral hemorrhage (ICH) have shown conflicting data, lacked objective speech pathology findings, and have included race predictor without biological basis. We hypothesized that stroke severity and findings from instrumental swallowing assessments would predict PEG placement and race would not be a significant predictor after controlling for other variables.
Methods: Using our prospectively obtained stroke registry, we analyzed consecutive ICH cases at our comprehensive stroke center. Patients transferred to hospice or died within 3 days were excluded. Backward stepwise elimination using logistic regression evaluated potential predictors of PEG placement. Fiberoptic Endoscopic Evaluation of Swallowing (FEES) and Modified Barium Swallow (MBSS) studies utilized the validated Penetration-Aspiration (PEN-ASP) scale; an 8-item scale: 1-2, normal; 3-5, penetration; 6-8 aspiration.
Results: Between June-14 and April-15, 290 ICH patients were treated by our stroke service (65.5% hospital transfers). Overall, 64 (21.4%) patients received PEG tubes with higher rates in non-whites (table). Ten variables with p<0.1 were identified (see table). In the final model, race was not significantly associated with PEG (p=0.092), but total NIHSS, sub-item 1a of the NIHSS, presentation to our ED and GCS remained independent predictors of PEG placement (p<.05). PEG patients had significantly worse PEN-ASP scores for both thin and nectar-thick liquids.
Conclusions: Our work represents the largest cohort of ICH patients that underwent surgical feeding tube placement. We did not find an independent association with race as previously reported. Future research regarding dysphagia and stroke should include objective assessments of swallow function such as the PEN-ASP score.