Diabetes-specific tube feeds improved glycaemia, with modest effect sizes

Design

Systematic review and meta-analysis of randomised trials

Population

Tube-fed patients with hyperglycaemia: 483 in critical care, 137 in long-term care

Primary outcome

Glycaemic control on diabetes-specific versus standard formula

Effect

Critical care mean glucose -0.54 mmol/L (95% CI -0.78 to -0.31); long-term care HbA1c -0.92%

A systematic review examined diabetes-specific nutritional formulas against standard formulas for tube-fed patients with hyperglycaemia, handling critical care and long-term care separately because the populations and goals differ.

Five randomised trials allocating 483 critically ill patients and three trials allocating 137 long-term care patients met inclusion. In critical care, the specific formulas lowered mean blood glucose by 0.54 mmol/L (95% CI -0.78 to -0.31, p<0.001, I-squared 9%, moderate certainty), and also reduced the coefficient of variation of glucose by 6.56% and the standard deviation by 0.77 mmol/L, though both of those carried I-squared of 98% and low certainty. In long-term care, HbA1c fell by 0.92% (95% CI -1.72 to -0.13, p=0.023, low certainty).

The authors are candid that outcomes beyond glycaemia were uncertain — either unreported, inconsistent, or underpowered. This is evidence that the formulas do what they are designed to do, not yet evidence that it changes what happens to the patient.

• Eight randomised trials in total, split across critical care (483 patients) and long-term care (137).
• Mean glucose in critical care fell 0.54 mmol/L, with low heterogeneity and moderate certainty.
• Outcomes other than glycaemic control were too sparsely reported to judge.

Why it matters

Tube feeding is where hyperglycaemia in inpatients is most often generated and least often planned for.

Don't overread it

Only the critical-care mean glucose result had moderate certainty and low heterogeneity. The variability and HbA1c results sit at I-squared 98% and 85% and are rated low certainty.

The statistics, in plain English

A 0.54 mmol/L drop is roughly 10 mg/dL — real, consistent across those trials, and small. The HbA1c figure of 0.92% looks larger, but it rests on three small long-term care trials that disagreed with each other considerably.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.