Assessment of indications for percutaneous endoscopic gastrostomy--development of a predictive model.

Objective: Percutaneous endoscopic gastrostomy (PEG) is used for long-term enteral nutrition in neurological patients with dysphagia (NEUR), in head and neck cancer patients prior to chemoradiation therapy (head and neck malignancy group [HNM]), or in cases of oropharyngeal or esophageal tumor obstruction or stricture (OBSTR). Considerable morbidity and overall mortality is reported. Aim was to analyze the complication rates and mortality with PEG and to identify subgroups with poor outcomes.

Material And Methods: Patients underwent PEG (n = 401) in a single tertiary care center. Indications, characteristics, and causes of death were recorded.

Results: Number of patients in groups: HNM 135 (34%), OBSTR 74 (18%), and NEUR 192 (48%); follow-up time median (interquartile range): 17 (39) months; the time PEG used for feeding: 4 (7) months. A total of 91 patients (23%) had 110 complications, 31 patients (8%) had early (≤30 days) complications, and 49 patients (12%) major complications. Two deaths (0.5%, 2 peritonitis) were related to PEG. The 30-day mortality was 11% (n = 47). According to multivariate analysis, an increased 30-day mortality was associated with ≥75 years of age, American Society of Anesthesiologists (ASA) class IV, a Charlson comorbidity index (CCI) ≥4, body mass index (BMI) <18.5, and ongoing antibiotic therapy. With this model, 95% specificity was obtained in the 30-day survival figures.

Conclusion: The presented predictive model derived from our analysis may recognize patients with poor outcome when referred for PEG. The parameters in the present model (age, ASA class, CCI score, BMI, and data of ongoing antibiotic treatment) are easily measurable, and it is possible to integrate them into everyday work.