Ketamine for analgosedation, parenteral nutrition's old trade-off, and renal perfusion phenotypes

1. Ketamine for analgosedation in mechanically ventilated adults

Journal · Published: Critical Care Medicine, 7 September 2026 — genuinely new, 1 day old Casamento, Ghosh, Said, Eastwood, Peck, Jones et al. (Melbourne)

Prospective, double-blind, randomised, placebo-controlled trial at two university-affiliated Melbourne ICUs, September 2022 – December 2024. Ventilated adults already receiving an opioid infusion for analgosedation — cardiac surgical patients excluded. Low-dose ketamine infusion 0.15 mg/kg/hr or placebo. Primary outcome: hourly opioid dose.

• 538 screened; 413 ineligible; 120 randomised (59 ketamine, 61 placebo)
• Median hourly opioid dose in fentanyl equivalents: 64 µg/hr (IQR 36–89) with ketamine vs 77 µg/hr (47–100) with placebo
• Median difference −13.0 µg/hr, 95% credible interval −26.6 to 2.4; probability of benefit 95.1%
• No differences in most secondary outcomes, including delirium and significant adverse events

Interpretation. Read the statistics carefully, because the framing is Bayesian and the same numbers would read differently in frequentist dress. A 95% credible interval of −26.6 to 2.4 crosses zero; “probability of benefit 95.1%” is an honest way of presenting that, but it is not a demonstration of effect. The point estimate — 13 µg of fentanyl per hour, roughly a 17% reduction — is real but modest.

Two design details matter more than the result. 413 of 538 screened patients were ineligible, which is a very tight funnel and limits who this applies to. And cardiac surgical patients were explicitly excluded — so for your practice this does not speak to the post-cardiotomy population at all.

The delirium finding is the one worth carrying: no difference. Ketamine’s reputation as either a delirium-sparing or delirium-provoking sedative gets no support here in either direction.

It also adds a third data point to the ketamine thread in this archive. The RSI trial found more cardiovascular collapse with ketamine at induction; the induction-agent network meta-analysis agreed (RR 1.31 for haemodynamic instability). This trial, using it as a low-dose infusion adjunct rather than an induction bolus, found it safe. Those are compatible: the dose and the moment are what differ, and conflating them is how ketamine’s reputation got muddled in the first place.

Read the paper · PMID 42704275

2. Early parenteral nutrition versus enteral nutrition alone

Journal · Published: JPEN, 6 September 2026 — genuinely new, 2 days old Cai, Bao, Yin, Yang, Wang, Liu. PROSPERO CRD42023462386.

RCTs from January 2000 to January 2025 comparing PN (total or supplemental) with EN alone in critically ill adults. Cochrane RoB 2, GRADE. 20 RCTs, 12,865 patients.

• 30-day mortality: RR 0.96 (95% CI 0.90–1.03), P=0.29 — null
• 90-day mortality: RR 0.95 (0.86–1.04), P=0.26 — null
• ICU mortality: RR 0.92 (0.86–0.99), P=0.03 — lower with PN
• Shorter mechanical ventilation duration; less gastrointestinal intolerance
• Overall infections: RR 1.12, P=0.002. Bloodstream infections: RR 1.27, P=0.005

Interpretation. The pattern is internally odd and worth naming: ICU mortality lower, but 30- and 90-day mortality flatly null. A real survival benefit confined to the ICU and vanishing by day 30 usually means either competing risks — patients surviving ICU to die on the ward — or a marginal p=0.03 that will not replicate. With 30- and 90-day estimates both sitting close to 1.0 with tight intervals, the second is more likely, and the ICU-mortality result should not be quoted on its own.

The infection signal is the durable finding, and it is the same one EPaNIC and CALORIES produced: parenteral nutrition buys nutritional adequacy and costs bloodstream infections, RR 1.27 here. That trade-off has been stable across two decades of trials.

Sits directly alongside the nutrition guidelines this archive covered over the past fortnight — ESPEN in the ICU, ESPEN in surgery, and the AAST/ACS trauma protocol, all of which reserve PN for when enteral is contraindicated or insufficient. This meta-analysis is consistent with that positioning rather than a challenge to it.

Read the paper · PMID 42701873

3. Early renal perfusion phenotypes and renal recovery

Journal · Published: Shock, 7 September 2026 — genuinely new, 1 day old Fu, Ke, Lv, Wang, Zhang et al. Six Chinese ICUs. NCT05866250.

Prospective multicentre cohort. Critically ill adults with or at risk of AKI, enrolled within 24 h of ICU admission. Doppler-derived semi-quantitative renal perfusion (SQP) and renal resistive index (RRI) measured at admission, 24 h and 48 h, with group-based trajectory modelling.

• 456 patients; 47.4% had AKI at admission, 55.5% met AKI criteria within 72 h
• Two SQP phenotypes: sustained-high perfusion (49.1%) and sustained-low perfusion (50.9%)
• Sustained-high SQP: lower odds of persistent AKI, adjusted OR 0.56 (95% CI 0.35–0.89), P=0.015 — though attenuated in expanded sensitivity models
• Also associated with less renal non-recovery at 7 days, faster recovery, and less RRT on days 3 and 7
• RRI trajectories did not discriminate renal outcomes

Interpretation. Two things here, and the negative one may be more useful.

RRI did not work. The renal resistive index has been the standard bedside Doppler measure for a decade, and in serial trajectory form it failed to separate outcomes that a cruder perfusion score did. That is directly relevant to the renal-resistive-index-guided MAP titration study sitting on this archive’s outstanding list — if RRI trajectories do not track recovery, using RRI to titrate pressure looks less promising than it did.

The positive finding is a flow finding. Sustained renal perfusion predicts renal recovery, and it does so independently of the KDIGO creatinine-and-urine-output classification the authors note “may not capture early physiologic differences.” That is the fourth paper in three weeks pointing the same way: the off-pump tissue oxygenation study, the pulse pressure and AKI cohort, the Danish MAP-target trial where the highest-pressure arm had the most AKI, and now this. Perfusion, not pressure, and not creatinine.

Restraint is warranted: single-country cohort, the primary association attenuated under sensitivity analysis, and the authors themselves ask for external validation. But the convergence across four very different study designs is becoming hard to dismiss.

Read the paper · PMID 42704056