The authors pooled nine comparative, non-randomised studies comprising 813 patients to compare uniportal robotic-assisted thoracic surgery with multiportal RATS for anatomical lung resection. Postoperative complications were reported in 16.0% of uniportal patients and 30.2% of multiportal patients. When combined in a random-effects meta-analysis the pooled risk ratio was 0.49 (95% CI 0.22–1.06; p = 0.069), with substantial heterogeneity (I² = 66%). The study team observed that this pooled result did not reach statistical significance. Several subgroup and sensitivity checks were performed; the authors noted that findings were sensitive to individual studies and that overall evidence certainty was limited.
The meta-analysis did not find statistically significant differences in rates of major complications, conversion to open procedures, or perioperative mortality. Data indicated neither approach showed a clear safety advantage for major adverse outcomes, but the cohort of included studies carried a high risk of bias that weakens confidence in those null findings.
Investigators reported a mean reduction in console time favoring uniportal approaches: mean difference −23.5 minutes (95% CI −35.6 to −11.4). The meta-analysis did not show a significant pooled difference in overall operative time when all studies were combined. Blood loss differences were likewise non-significant across pooled comparisons.
The authors suggested that apparent time savings for console activity might reflect case selection, team familiarity, or differences in operative era rather than an intrinsic advantage of a single-port configuration. Exploratory subgroup analyses hinted at interactions by comparator port count; those exploratory signals could reflect confounding by surgeon experience or institutional adoption timelines rather than causal effects.
Day‑1 postoperative pain scores were lower after uniportal procedures in pooled estimates (mean difference −0.6 points, 95% CI −0.9 to −0.3). The review team emphasized that the 0.5‑point magnitude falls below commonly used thresholds for clinical importance. They also reported that this pain reduction lost statistical significance when digitised values were excluded in sensitivity analysis (p = 0.092). In short, a small early pain benefit was observed in pooled observational data, but the clinical meaningfulness and robustness of that finding are uncertain.
No significant pooled differences emerged for lymph-node yield or R0 resection rates between uniportal and multiportal robotic techniques. The cohort showed comparable surgical-quality metrics across approaches in available reports. The study team cautioned that these oncologic and nodal endpoints were reported in observational series and that follow-up duration or staging mix were not uniformly detailed in the publication abstract; the source did not report additional trial-level staging or long-term oncologic outcome data within the abstract.
The authors assessed risk of bias with ROBINS-I V2 and rated seven of nine included studies at critical risk of bias. Certainty of evidence was evaluated using GRADE and judged low to very low for key outcomes reported. Heterogeneity was moderate-to-high for several pooled analyses (for example, I² = 66% for overall complications). The review team explicitly warned that subgroup signals—such as differences in length of stay or operative time by comparator port count—could reflect confounding by operative era, surgeon experience, or selection bias in the observational literature.
The protocol registration (PROSPERO CRD420261432924) was reported in the abstract, and the authors declared no competing interests relevant to the article.
All included studies were non-randomised comparative series; seven were rated at critical risk of bias, which substantially limits causal inference. The evidence base is therefore observational and of very-low certainty for many outcomes according to the authors. Some pooled differences (console time and early pain) were sensitive to the inclusion or exclusion of individual studies and lost statistical strength in sensitivity analyses.
The source did not report randomized trial data, detailed long-term oncologic outcomes, or granular methodology on case selection and perioperative pathways in the abstract. The dataset leaves unanswered questions about whether any observed perioperative differences persist after robust adjustment for surgeon experience and institutional learning curves, and whether long-term oncologic equivalence holds across broader populations. Future randomized or carefully risk‑adjusted prospective studies were implied as necessary to resolve these uncertainties.