Hypoalbuminemia is a poor predictor of survival after percutaneous endoscopic gastrostomy in elderly patients with dementia

OBJECTIVE: Percutaneous endoscopic gastrostomy (PEG) is widely used for long-term enteral nutrition in patients with dementia and inadequate oral intake, although its benefit in prolonging the survival of the patient is not clear. Patients are often referred for PEG placement after a significant weight loss or fall in serum albumin. It is not known whether this delay in referral adversely affects the survival. Our aim was to determine the survival after PEG placement in patients with inadequate oral intake secondary to cognitive impairment and to determine whether the nutritional parameters at the time of placement of PEG predict survival. METHODS: After excluding patients with cancer, cerebrovascular accident (CVA), and anoxic encephalopathy, 56 consecutive patients with dementia and inadequate oral intake were included in the study. The following parameters were analyzed to predict survival benefit: demographics, reason for admission to the hospital, nutritional indices, Karnofsky performance status before and after PEG, comorbid medical illnesses, and complications after PEG. A control group was selected from patients admitted to the geriatric division who did not receive PEG. Patients with CVA, cancer, and anoxic encephalopathy were also excluded in the control group. Both groups were followed-up for 6 months. RESULTS: One patient was lost to follow-up and complete data were available for 55 patients. The control group of 33 patients was comparable to the patients in age, gender, and comorbid illnesses. During a 6-month follow-up period, the mortality was 44% among patients and 26% among controls (p = 0.03). The only factor that predicted poor survival in patients at 6 months was a serum albumin of <2.8 g/dl. The predictive value of low serum albumin was higher in patients with sepsis on hospital admission and in patients younger than 85 yr. The mean serum albumin of the patients was significantly lower than the control group (p = 0.0001). CONCLUSIONS: Only 50% of demented patients with inadequate oral intake are likely to survive beyond 6 months after PEG placement. No improvement in performance status is likely to occur in any patient. These limited benefits should be realized before initiating PEG feeding. Since the survival in the age-matched control group with a higher mean serum albumin was better than in the patients who had PEG, there may be some benefit in performing PEG before the onset of severe hypoalbuminemia.