PEG Insertion in Patients with Dementia Does Not Improve Nutritional Status and Has Worse Outcomes as Compared with PEG Insertion for Other Indications

Background: Percutaneous endoscopic gastrostomy (PEG) tubes are commonly utilized as a method of enteral feeding in patients unable to obtain adequate oral nutrition. Although some studies have shown improved mortality in select populations, the safety and effectiveness of PEG insertion in patients with dementia compared with those with other neurological diseases or head and neck malignancy remains less well defined. Objective: To evaluate the nutritional effectiveness, rate of rehospitalization, and risk of mortality among patients with dementia compared with patients with other neurological diseases or head and neck cancers who undergo PEG placement. Materials and Methods: We conducted a retrospective analysis from a prospective database of patients who underwent PEG placement at an academic tertiary center between 2008 and 2013. The following data were collected: indication for PEG, patient demographics, biochemical markers of nutritional status rehospitalization, and survival rates. Results: During the study period, 392 patients underwent PEG tube placement. Indications for PEG were dementia (N=165, group A), cerebrovascular accident (N=124, group B), and other indications such as oropharyngeal cancers and motor neuron disease (N=103, group C). The mean follow-up time after PEG was 18 months (range, 3 to 36 mo). No differences in baseline demographics were noted. PEG insertion in the dementia (group A) neither reduced the rehospitalization rate 6 months' postprocedure compared with groups B and C (2.45 vs. 1.86 and 1.65, respectively; P=0.05), nor reduced the mortality rate within the first year post-PEG placement (75% vs. 58% and 38% for groups A, B, and C, respectively, P=0.001), as well, it did not improve survival at 1 month after the procedure (15% vs. 3.26% and 7.76%, for groups A, B, C, respectively, P<0.01). The presence of dementia was also associated with shorter mean time to death (7.2 vs. 8.85 and 8 mo for groups A, B, C, respectively, P<0.05). The rate of improvement of the nutritional biomarker albumin was lower in the dementia group [3.1. to 2.9 vs. 3.2 to 3.3 and 3 to 3.3 g/dL for groups A, B, and C, respectively (P<0.02)]. Multivariate regression analysis showed that the presence of dementia was an independent predictor for mortality rate within the first year and 1-month mortality rate in patients undergoing PEG insertion with odds ratio 3.22 (95% confidence interval, 1.52-4.32) and odds ratio 2.52 (95% confidence interval, 1.22-3.67). Conclusions: PEG insertion in patients with dementia neither improve both short-term and long-term mortality nor rehospitalization rate as compared with patients who underwent PEG placement for alternate indications such as other neurological diseases or head and neck malignancy and even was associated with shorter time to death. Furthermore, PEG insertion in patients with dementia did not improve albumin. Therefore, careful selection of patients with dementia is warranted before PEG placement weighing the risks and benefits on a personalized basis.