Thirty years of the Ross procedure, and two meta-analyses that undercut their own subgroups

All three genuinely new, all published yesterday.

1. Late results of the Ross procedure — thirty years

Journal · Published: Annals of Thoracic Surgery, 3 September 2026 — genuinely new, 1 day old David TE, David CM, Samman, Runeckles, Ho, Ouzounian (Toronto)

Most Ross series report the first two decades. This reports thirty years. 212 consecutive patients, median age 34, operated 1990–2004 — freestanding aortic root replacement in 108, aortic root inclusion in 104 — followed prospectively for a median of 26.6 years, with serial echocardiography.

• 30-year survival 79.4% (95% CI 71.1–85.6) — 9.1% lower than the Canadian general population matched for age and sex
• 47 patients (22%) required Ross-related reintervention: 64 surgical and 12 percutaneous procedures
• Cumulative incidence at 30 years — any Ross-related reintervention 28.4%; autograft reoperation 19.1%; homograft reoperation 17.1%
• Autograft dysfunction 38.0%, pulmonary homograft dysfunction 62.6%
• Reoperation on the autograft was associated with dilated aortic annulus, older age, and freestanding root replacement

The authors’ conclusion is unusually plain: the Ross procedure “is a good aortic valve substitute for young adults, but it does not restore normal lifespan, and the function of both valves deteriorates over time, including the development of autograft valve stenosis.”

Interpretation. This is the series that matters on this question — Tirone David’s group, prospective follow-up to a median of 26.6 years, which almost nothing in cardiac surgery can match. And it complicates the case for the Ross that has been building over the last decade, which rested largely on the observation that Ross patients appear to have survival approaching the general population.

At thirty years they do not: 9.1 percentage points below matched controls, with more than a fifth reoperated and both valves degenerating. The specific finding that freestanding root replacement predicts autograft reoperation is technically actionable, and the emergence of autograft stenosis — not just insufficiency — is a mode of failure that shorter series would not have caught.

None of this makes the Ross a bad operation for a 34-year-old facing mechanical valve anticoagulation for life. It does mean the consent conversation should include a roughly one-in-five chance of autograft reoperation and a two-in-three chance of homograft dysfunction across a normal remaining lifespan. That is a different conversation from “restores normal survival.”

Read the paper · PMID 42692133

2. Remote ischaemic preconditioning in noncardiac surgery

Journal · Published: Anesthesia & Analgesia, 3 September 2026 — genuinely new, 1 day old Labanca, Oliva, Fresilli, Beretta, Zangrillo et al.

Systematic review and meta-analysis of RCTs of RIPC in adults undergoing noncardiac surgery, searched to June 2025, Cochrane RoB 2, random-effects GLMM for rare binary outcomes. Primary outcome all-cause mortality at longest follow-up.

• 79 RCTs, 9,340 patients — predominantly single-centre; abdominal surgery the commonest setting (25 studies); RIPC mostly by upper-limb blood-pressure cuff
• Mortality overall: OR 0.79 (95% CI 0.51–1.24), P=0.30 — null
• Subgroup receiving RIPC before anaesthesia induction: OR 0.37 (0.17–0.81), P=0.013
• Exploratory: stroke OR 0.41 (0.22–0.78, P=0.006); length of stay −0.84 days (P<0.001); peak postoperative NGAL −10.27 (P=0.008)

Interpretation. Look at the arithmetic before the subgroup: 79 trials across 9,340 patients is an average of 118 patients per trial, predominantly single-centre. That is the exact profile of a literature prone to small-study effects — and it is worth noting that the same journal published a Bayesian simulation study on precisely this problem in goal-directed haemodynamic therapy two days earlier (still on my outstanding list, paywalled).

Given that, a null primary result with a positive timing subgroup should be read cautiously. “Before induction” is a plausible mechanistic hypothesis, not a post-randomisation accident, which puts it ahead of most subgroups. But it remains a subgroup within a null meta-analysis of small trials, and RIPC’s history is a cautionary one: promising in preclinical work and small trials, then negative in ERICCA and RIPHeart, the two large cardiac-surgery RCTs that finally tested it properly.

The honest summary is that RIPC still has not had its adequately powered multicentre trial in noncardiac surgery, and this meta-analysis — despite 79 studies — does not substitute for one.

Read the paper · PMID 42691252

3. Intravenous lidocaine in spine surgery — procedure-specific effects

Journal · Published: Regional Anesthesia & Pain Medicine, 3 September 2026 — genuinely new, 1 day old Felippe, Bersot, Gonzalez, Wegner et al.

The premise: previous meta-analyses pooled heterogeneous spine procedures and may have obscured procedure-specific effects. Searched to June 2026. 10 RCTs.

• Pain at 24 h: 9 trials, 655 patients — MD −0.83 (95% CI −1.36 to −0.30), p=0.002, moderate certainty — but I²=89% and a prediction interval of −2.69 to 1.02, which crosses zero
• Opioid consumption: MD −11.64 mg IV morphine equivalents (−16.14 to −7.14, p<0.001)
• Post hoc subgroup by procedure (test for subgroup differences p=0.005): instrumented fusion or complex spine MD −1.23 (−1.81 to −0.64), exceeding the 1.0-point minimal clinically important difference; decompression MD −0.20 (−0.61 to 0.21) — essentially nothing
• Baseline pain severity explained ~53% of between-study heterogeneity; infusion rate did not modify effect (p=0.97)
• No effect on PONV or length of stay

The authors label the subgroup finding post hoc, exploratory and hypothesis-generating.

Interpretation. The prediction interval is the number to carry: −2.69 to 1.02 means that in a new trial, lidocaine could plausibly reduce pain by nearly three points or increase it by one. A significant pooled mean with I²=89% and that prediction interval is a description of heterogeneity, not of a reliable effect.

Which is exactly why the procedure-specific analysis is interesting. The split — meaningful benefit in instrumented fusion, nothing in decompression — is mechanistically coherent, since the two operations differ enormously in tissue injury and baseline pain, and baseline pain severity explained over half the heterogeneity. The finding that infusion rate did not matter points the same way: the variance is in the patient and the operation, not the dose.

Treat it as the authors ask — hypothesis-generating. But if you are choosing where to spend a lidocaine infusion, this says instrumented fusion rather than decompression, and that is a more useful message than the pooled estimate.

Read the paper · PMID 42692530