Nutrition therapy is a core component of care for critically ill patients, yet large randomized trials have reported inconsistent effects of protocolised feeding on clinical outcomes. The NEED trial previously demonstrated that an evidence-based feeding protocol improved delivery of enteral nutrition (EN) but did not reduce overall mortality. Physiological differences by age may alter response to nutritional interventions; this secondary, post hoc analysis evaluated whether age modifies the association between implementation of an evidence-based feeding protocol and clinical outcomes in critically ill adults.
This analysis used patient-level data from the multicentre cluster-randomised NEED trial conducted across intensive care units in China. Adult patients newly admitted to participating ICUs were included. Clusters (ICUs) were randomised to active implementation of a feeding protocol or to usual care. The analysed cohort comprised 2,672 patients with a median age of 62 years (IQR 49–74); 1,798 (67.3%) were male. The original trial registration is ISRCTN 12233792.
Age was modelled both as a continuous variable and as prespecified categorical groups: ≤65 years, 66–75 years, and ≥76 years. The primary effect-modification analysis tested the interaction between treatment allocation (protocol versus usual care) and age on 28-day all-cause mortality using mixed-effects Cox regression models. Secondary outcomes—reported as exploratory—included organ failure, persistent multiple organ failure, ICU length of stay, ICU-free days, infection within 7 days, and feeding intolerance. Mixed-effects logistic regression was used for categorical secondary outcomes where applicable. Per-protocol and sensitivity analyses were conducted to assess robustness.
Across the 2,672 patients, the effect of the evidence-based feeding protocol on 28-day mortality varied by age. When age was modelled continuously, each 10-year increase in age was associated with a higher adjusted hazard ratio (HR) for mortality in the protocol versus usual care comparison (adjusted HR per 10-year increase 1.28; 95% CI 1.13–1.46), with a continuous interaction P < 0.001. In prespecified age-stratified analyses, the feeding protocol was associated with a statistically significant reduction in 28-day mortality among patients aged ≤65 years (adjusted HR 0.64; 95% CI 0.44–0.92). No mortality benefit was observed in the 66–75 years group (adjusted HR 1.01; 95% CI 0.64–1.60) or in patients aged ≥76 years (adjusted HR 1.29; 95% CI 0.89–1.88). The categorical interaction for age groups and mortality had P = 0.010.
An exploratory analysis of persistent multiple organ failure showed a similar age-dependent pattern (interaction P = 0.017), suggesting the protocol’s association with organ-failure outcomes may also vary by age. Other secondary endpoints listed in the original methods—organ failure, ICU length of stay, ICU-free days, early infection, and feeding intolerance—were examined as exploratory outcomes, though the abstract reports the age-dependent signal most clearly for persistent multiple organ failure. Detailed effect estimates for all secondary endpoints were not reported in the abstract and therefore are not available from this source.
The authors report that results were consistent in per-protocol and sensitivity analyses, indicating the observed age-treatment interaction for mortality and persistent multiple organ failure was robust to those additional analytical approaches. Specifics of the per-protocol definitions and sensitivity models are reported in the trial publication; detailed parameters beyond consistency statements were not provided in the abstract.
These secondary, hypothesis-generating findings indicate age-related heterogeneity in the association between an evidence-based feeding protocol and clinical outcomes in critically ill adults. In this cohort, younger patients (≤65 years) appeared to derive mortality benefit from protocol implementation, whereas older patients did not show the same benefit and point estimates suggested no advantage or potential attenuation with advancing age. If confirmed, this could imply that age should be considered when applying standardised feeding protocols and when designing nutritional trials in heterogeneous ICU populations.
This analysis is a post hoc secondary study of a cluster-randomised trial and therefore is exploratory. Causal inferences are limited by the secondary nature of the analysis and potential unmeasured confounding. The abstract does not present full details of secondary outcome effect sizes, nutrient delivery metrics by age subgroup, or granular per-protocol definitions; these details require review of the full publication. The authors conclude that prospective studies are needed to confirm age-related treatment heterogeneity and to clarify whether feeding strategies should be tailored by age in critically ill patients.
In this secondary analysis of the NEED trial (n = 2,672), the association between implementation of an evidence-based enteral nutrition feeding protocol and 28-day mortality varied by age: benefit in patients ≤65 years but not in older age groups. Findings were consistent across sensitivity analyses but are exploratory and warrant prospective confirmation. The original trial is registered as ISRCTN 12233792.