Design: systematic review and random-effects meta-analysis of four comparative studies (one randomised trial, three propensity score-matched), PROSPERO registered
Population: 787 patients undergoing purely off-clamp robot-assisted partial nephrectomy
Primary outcome: Trifecta achievement, with renal function, complications and margins as secondary outcomes
Effect: pooled RR 1.17 (95% CI 0.97–1.41, I² 86.5%); matched studies 1.26 (1.05–1.52); randomised trial 0.97 (0.92–1.04), non-inferior; perioperative eGFR difference −3.89 (−6.16 to −1.62) with no difference at three months

Whether the renal parenchyma needs suturing after off-clamp robot-assisted partial nephrectomy is a genuine open question: renorrhaphy costs ischaemic time and nephron mass, and omitting it risks bleeding and urinary leak. A meta-analysis pooled four comparative studies — one randomised trial and three propensity score-matched cohorts, 787 patients in total — with Trifecta achievement as the primary outcome.

The overall pooled estimate showed no significant difference (risk ratio 1.17, 95% confidence interval 0.97 to 1.41) with heterogeneity of 86.5%, and the heterogeneity has an obvious source: the study designs disagreed. The propensity-matched studies favoured the sutureless approach (1.26, 1.05 to 1.52), while the randomised trial returned 0.97 (0.92 to 1.04), meeting its prespecified non-inferiority criterion without showing superiority. Sutureless surgery was associated with a smaller perioperative fall in estimated glomerular filtration rate (mean difference −3.89 mL/min/1.73m², −6.16 to −1.62), though there was no difference at three months. Major complications, transfusion and positive margins did not differ.

The authors name the likely explanation for the divergence, and it is the useful part: in an observational study the surgeon decides intraoperatively whether to suture, and that decision is made on how the resection bed looks. Easy beds go unsutured and do well; difficult beds get sutured and do less well. Propensity matching on preoperative variables cannot touch a decision made after the tumour is out. The randomised evidence is the trustworthy part, and what it supports is omitting routine renorrhaphy while repairing selectively where it is needed — not abandoning the stitch.

Read the randomised result, not the pooled one: non-inferior, not superior.
Omit routine renorrhaphy only where selective repair remains available and is used when needed.
Do not quote the smaller perioperative eGFR fall as a lasting renal benefit; it was gone by three months.
Note that urinary and vascular complications were sparsely reported — the outcomes most relevant to omitting the stitch are the least measured.
Treat the propensity-matched advantage as intraoperative selection bias until a second trial says otherwise.

Why it matters

It is a clean demonstration of how an intraoperative decision defeats propensity matching, in a question where the observational answer is the more flattering one.

The statistics, in plain English

An I² of 86.5% in a four-study meta-analysis means the studies were not estimating the same thing, and pooling them gives a number (1.17, 0.97 to 1.41) that describes neither group. The split is structural rather than random: the randomised trial's tight interval (0.92 to 1.04) sits almost entirely inside the matched studies' interval but centred on no effect, while the matched studies (1.05 to 1.52) exclude no effect. When designs disagree that cleanly, the design that removes selection bias is the one to believe. The eGFR difference of −3.89 units is a perioperative measurement, and its disappearance by three months is the more informative finding.