Friday, 31 October 2014

CALORIES - Early administration of enteral or parenteral nutrition to critically ill adults

Paper Link

Harvey S et al for the CALORIES Trial investigators
New England Journal of Medicine (2014); 371:1673-84

At a glance

  • Intervention : Patients randomised to EN or PN in first 36 hours of admission, continued up to 5 days
  • Randomised - Yes
  • Controlled - Yes : Head to head
  • Placebo used - No
  • Blinded - No
  • Population / Sample size calculation - 2388 patients in 33 English ICUs
  • Groups same at randomisation - Yes
  • Groups treated the same after randomisation - Mostly, though some differences exist, such as use of prokinetics
  • Intention to treat analysis - Yes
  • Follow-up - Almost 100% at 90 days
  • Significant primary endpoint - There was no statistical difference in 30 day mortality
  • Precision of results - Powered to detect 20% relative risk reduction which may have underestimated sample size
  • Harms analysed - Yes.  Serious complications were similar in both groups.


The background

The role of total parenteral nutrition in critically ill patients has been a source of controversy for decades.

Multiple studies in the 1970s and 80s had demonstrated that using TPN in these patients when compared with EN resulted in worse outcomes, particularly infectious complications.  TPN rapidly fell from favour, particularly as protocols for delivery of nutrition via the enteral route improved.

For years, the prevailing wisdom was that TPN was significantly inferior to EN in critically ill patients, and should be reserved for those patients in whom enteral nutrition was impossible for at least a week, assuming normal nutritional status prior to admission.

However, there has recently been a softening of this industry-wide stance.  Proponents of TPN have argued that the early trials were performed in a different context, a time when blood sugar control was relatively poor, prior to improved central line technology and before widespread use of bundles of care designed to reduce CVC infection rates.  Additionally, EN is not always smooth sailing.  Underfeeding is common and possibly associated with aspiration and ventilator associated pneumonia though there is little evidence to support this is.

Conflicting guidelines have led to the confusion - for example, the current European guidelines recommend adding TPN if enteral nutrition cannot be established within 2 days of admission, and the American guidelines suggest waiting until at least a week.

Harvey et al have tried to clarify the use of early TPN when compared with early EN in this recently released trial in critically ill adults.

The trial

It is important to understand the clinical question being asked here.  The study examines whether using early EN or PN changes outcome in generic adult critically ill patients who are within 36 hours of admission.  It doesn't look specifically at the subgroup in whom enteral nutrition can't be established.  The study hypothesis was that TPN would achieve better caloric intake initially than EN.  Additionally the potential for a reduction in infectious complications related to EN (due to aspiration etc) along with the improved safety profile of modern day PN delivery would tip the balance in favour of TPN.

The trial randomised 2388 patients in 33 adult ICUs in England with an unplanned admission to the ICU to either TPN or EN within 36 hours of admission.  Therapy was continued for up to 5 days, until death or discharge, or when full oral nutrition was introduced.

This was a pragmatic trial - designed to avoid being too restrictive in a trial sense, so that the results might be more applicable to the real world.  However, this does result in the loss of some discrimination when analysing the results, as fewer variables are controlled.  This concept is explored more in the accompanying editorial by Deborah Cook.

The study sample calculation estimated the baseline mortality in the enteral nutrition group would be 32%, and was powered to detect an absolute risk reduction of 6.4% based on a recent meta-analysis.  Perhaps this was an optimistic figure that has led to under powering of the trial.

The primary endpoint was 30 day all cause mortality.

The nutritional goals was 25kcal/kg actually body weight in both groups, including non-nutritional sources of calories such as propofol.  Parenteral nutrition required the insertion of a central venous line, while enteral nutrition was administered by nasogastric or nasojejunal tubes, as guided by a national standard.  Glycaemic control was maintained less than 10mmol/L (180mg/dL).

The caloric intake was similar between the two groups, with nutritional goals unable to be met in the majority of patients in both groups.  The reasons for this are not clear, and are particularly confusing for the PN group.  For the most part the groups were treated equally, though more patients crossed over from the PN group to EN than the other way around.  Patients in the enteral group received prokinetic agents more often than the PN group.  The two groups received insulin infusions and vasopressors in similar frequencies, though the doses are not clear.

Aside from the powering of the trial, the biggest methodological limitation of the study was that it was unblinded.  This is reflected in the higher likelihood of patients crossing over from the TPN group to the EN group than the other way around.

The results

There was no difference between the groups with respect to the primary endpoint, even after adjustment for variations in baseline risk factors.

Contrary to expectations, there was no difference in infectious complications between the groups, and hypoglycaemia was actually less in the TPN group.    Fairly predictably, vomiting was more likely in the EN group.  Other that that, there was no difference recorded in the other 15 secondary endpoints.

What does it mean?

That's an interesting question, and your perspective probably depends on the baggage you carry into the trial.

Proponents will argue that this is good evidence that TPN is well tolerated, does not lead to infectious complications if used for a short period early on.  This study does seem to support the assertion that the delivery of TPN is now safer than in the early studies, and this may reflect better standards of central line care and glycaemic management.

Naysayers will retort that it is an expensive, unnecessary intervention that does little to improve calorie delivery, so why do it?  Additionally, there appears to be a dose-related effect on infectious complications with TPN, and it remains possible that the inability to reach target feeding rates in the TPN "protected" this group from harm.

Interestingly, neither method was particularly effective in achieving the pre-defined goal for calorie delivery, something reported in other trials in this area.  This represents a significant limitation in interpreting the results, as one of the reasons early TPN is advocated is that it was thought to be a more reliable source of calories than EN.  The inability to successfully meet these targets in the context of a trial highlights the significant hurdles to achieving adequate nutrition with either EN or PN, including product availability, interruption for procedures and competing clinical priorities.

Does a middle ground exist?  Feeding a patient enterally is cheap, efficient, has similar risks to TPN and likely has a beneficial effect on gut epithelial function.  It might be argued that this trial supports the notion that when enteral nutrition is not possible, starting TPN is not unreasonable.  But it could also be argued that patients who can't be fed enterally are a different group of patients to those included in this trial.  Trials such as the Caesar TOPUP trial have examined this concept and found no benefit, or even harm, with this approach.

Additional Resources