Total parenteral nutrition vs enteral nutrition

There are many small trials in various sub-groups of patients (i.e.: lymphoma, radiotherapy) comparing enteral and total parenteral nutrition. This review will discuss only meta-analyses of various subgroups for the sake of brevity as the findings are remarkably homogenous. A large meta-analysis of 7 studies comparing total parenteral nutrition (TPN) with standard care and 20 studies comparing tube feeding to TPN was done by Braunschweig et al [57]. A total of 1,828 patients were included in these randomized prospective studies. Standard care patients had a relative risk of infection of 0.77 when compared to TPN and enterally fed patients had a relative infection risk of 0.66. If catheter infections were excluded from the meta-analysis, the increased risk of infection associated with TPN persisted.

A much smaller meta-analysis using mostly unpublished studies was done on 230 surgical and trauma patients [58]. The overall risk of infection complications was 2.5 times higher (p = 0.0001) in those randomized to TPN as opposed to enteral feedings early in the course of hospitalization. Rather than increasing the risk of aspiration and pneumonia, enteral feedings reduced that overall risk of pneumonia to 40% of that seen in TPN patients.

Since pancreatitis causes severe inflammation which may lead to transluminal spread of bacteria, it would be reasonable to suppose that TPN would decrease the infectious risks in patients with pancreatitis by decreasing pancreatic activity. The combined risk for infection for pancreatitis patients in two trials (n = 70) of enteral feeding was 0.61 when compared to the TPN route [59].

Chemotherapy patients may have long periods of anorexia and poor nutrition. Would TPN mitigate this malnutrition and allow a more vigorous defense against infection? Many small studies examining this question were published in the 1980's and 1990's. None showed an infection benefit with TPN. A meta-analysis of 15 studies comparing TPN with routine care was published in 1990 [60]. Three month survival and chemotherapy response were not improved by the use of TPN. The risk of infection was 3.2 times higher in those receiving TPN (p < 0.005) even if catheter-related infections were excluded from the analysis. If catheter infections were included, the risk of infection was 4 times greater in those receiving TPN. A recent Brazilian study [61] found that a central venous catheter through which TPN was given had a 3.3 times greater risk of becoming infected than a similar catheter not used for TPN.

Is the flood of intravenous nutrients responsible for the increased rate of infections seen with TPN or is it the lack of enteral feeding and the presence of a central venous catheter? Some insight into the answer is gained by the result of studies comparing TPN with intravenous feedings of other types.

A study of 248 Swedish post-operative patients were randomized to TPN or glucose and electrolyte solutions until they could eat and drink "freely" without the use of any intravenous support [62]. They found that those supported with glucose and electrolytes had a 5% mortality as opposed to 2% with TPN, but this difference was not significant. Both groups had similar rates of wound infections and pneumonia. Similar results were seen in 117 post-pancreatic resection patients in New York [63].

Reducing the amount of glucose and protein in the TPN admixture did not affect the infection rate in 40 patients with a variety of underlying conditions. Six of 21 hypo-caloric TPN patients developed infections as did 10 of 19 standard TPN patients. This difference is not statistically significant due to the small sample size [64]. Thus, there is no data to support the idea that increased delivery of nutrients intravenously alone increases the rate of non-catheter infection, but the size of the samples and the duration of the TPN may be an important variable in these studies. Enteral feedings seem to protect against infection.

There are several potential explanations for this apparent protection provided by enteral feedings. There is evidence reviewed by Kudsk [65] that levels of IgA and numbers of circulating lymphocytes from gut-associated lymphoid tissue are strongly affected by enteral feeding. These lymphocytes migrate to non-gut tissues such as the lung and can alter the immune response in non-gut locales. The neuroendocrine system of the gut also affects the regulation of inflammation outside of the gut. This neuroendocrine system is modulated by the presence of dietary constituents [66,67]. The value of enteral feeding may be thus dependent on cellular and neuroendocrine factors.