Diabetes-specific tube feeds move the glucose, and not much else

Design

Systematic review and meta-analysis of randomised controlled trials, critical care and long-term care analysed separately; PROSPERO-registered

Population

483 critically ill patients across five trials and 137 long-term care patients across three trials, all with hyperglycaemia on tube feeding

Primary outcome

Glycaemic control and clinical outcomes with diabetes-specific versus standard formulas

Effect

Critical care: mean glucose -0.54 mmol/L (95% CI -0.78 to -0.31, I squared 9%, moderate certainty); coefficient of variation -6.56% (-12.55 to -0.57, I squared 98%, low certainty); glucose SD -0.77 mmol/L (-1.49 to -0.06, low certainty). Long-term care: HbA1c -0.92% (-1.72 to -0.13, low certainty)

Hyperglycaemia during enteral feeding in critical illness is common, and formulas designed for it — lower in rapidly absorbed carbohydrate, higher in monounsaturated fat and fibre — cost more than standard ones. This review pooled five randomised trials in 483 critically ill patients and three in 137 long-term care patients to see what that money buys.

In critical care, diabetes-specific formulas lowered mean blood glucose by 0.54 mmol/L (95% CI -0.78 to -0.31), with low heterogeneity and moderate certainty — the one solid finding here. They also reduced the coefficient of variation of glucose by 6.56% and its standard deviation by 0.77 mmol/L, but both with I² of 98% and low certainty, meaning the trials disagreed almost completely. In long-term care, HbA1c fell by 0.92% with low certainty.

Everything else — infections, length of stay, mortality, ventilator days — was either not reported, inconsistent, or underpowered. The authors are candid: small effect sizes, imprecision and substantial heterogeneity reduce certainty for most outcomes.

A reduction of 0.54 mmol/L in mean glucose is real and modest. Whether it justifies the cost difference depends on what your unit pays and how tightly glucose is otherwise controlled — and it cannot be justified on outcomes, because the outcomes have not been shown.

• The proven effect is on glucose, not on any clinical outcome
• 0.54 mmol/L is the reliable figure; the variability results carry almost complete heterogeneity
• Weigh the cost difference locally; the case rests on glycaemic control alone
• Insulin protocols remain the primary tool for hyperglycaemia during feeding
• Long-term care evidence rests on three trials and 137 patients

Why it matters

It gives a precise, honest figure to set against the price difference, which is the actual decision a unit is making.

Don't overread it

Only glycaemic measures improved; no infection, ventilation, length-of-stay or mortality benefit was demonstrated.

The statistics, in plain English

An I squared of 98% means almost all the variation between trials is real disagreement rather than chance, which makes a pooled estimate close to meaningless for those outcomes — the authors label them low certainty for exactly that reason. The mean glucose result is different: I squared of 9% means the trials agreed, and moderate certainty is the highest rating anything here received. Small effects on a surrogate with no outcome data is the profile of a product decision, not a clinical one.

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.