Understanding why patients die after gastrostomy tube insertion: a retrospective analysis of mortality.

Objectives: To understand the causes of mortality of inpatients receiving a percutaneous endoscopic gastrostomy (PEG) tube compared with a survival curve predicted from a model proposed by Levine et al (2007).

Design: A retrospective study of patients receiving a PEG over an 18-month period.

Setting: Royal United Hospital Bath, a district general hospital in the southwest of England.

Patients: Fifty-five cases, with 44 found eligible for inclusion.

Interventions: A Levine score was calculated for this cohort. A survival curve after PEG was produced and compared with the Kaplan-Meier curve predicted by the Levine model.

Main Outcome Measures: Mortality over a period of 1 year.

Results: The mortality at 1, 3, 6, and 12 months was 16%, 20%, 25%, and 28%, respectively. This matched the predicted death rate from the Levine model closely (Pearson's rank correlation coefficient = 0.96).

Conclusions: The authors found that the mortality of patients receiving a PEG followed that predicted for a similar cohort of patients without PEGs in the Levine model. This suggests that the deaths observed were due to underlying comorbidities, can provide a baseline for mortality targets for PEG services, and is useful patient information regarding the risks and benefits of the procedure. The findings demonstrate that PEG does no harm and supports the accepted opinion that nutrition support is associated with a better outcome. Furthermore, they show that most deaths occur within the first month of placement and would support arguments for delaying placement until outcome from the underlying condition is more predictable.