The validity of central venous oxygen saturation (ScvO₂) as a surrogate for mixed venous oxygen saturation (SvO₂) in surgical patients remains uncertain.

We systematically searched PubMed, Embase, the Cochrane Library, China National Knowledge Infrastructure (CNKI), Wanfang, VIP, and the Chinese BioMedical Literature & Retrieval System (SinoMed) up to March 31 ^st 2026 for observational studies reporting paired perioperative ScvO₂ and SvO₂ measurements in surgical patients. Two reviewers independently screened studies, extracted data, and assessed quality using the Newcastle–Ottawa Scale (NOS). The primary outcome was clinical interchangeability, pre-defined as both the pooled mean difference (MD) 95% confidence interval (CI) and 95% limits of agreement (LOA) within ±5%. Correlation was a secondary outcome via Fisher's z transformation. For studies reporting multiple measurements within a single perioperative phase, arithmetic means were used to avoid unit-of-analysis errors. Random-effects models were prespecified for all pooled analyses, supplemented by pre-planned subgroup and sensitivity analyses to explore heterogeneity.

Twenty-eight studies including 1,369 patients were analyzed. In studies reporting aggregated perioperative data (10 studies, n = 524), ScvO₂ and SvO₂ were not clinically interchangeable: pooled MD was −0.49% (95% CI, −1.34 to 0.35; I ² = 70.3%), and LOA (−12.15% to 11.17%) exceeded the predefined ±5% threshold for clinical acceptability, consistent with observations across perioperative phases. The pooled correlation was 0.72 (5 studies, n = 197; 95% CI 0.57–0.82; I ² = 64.3%). Subgroup analyses showed narrower LOA and higher correlation in non-cardiovascular and off-pump procedures.

ScvO₂ is positively correlated with SvO₂ in the perioperative period. However, wide individual-level LOA indicate that the two measurements are not interchangeable in surgical patients. LOA are narrower in non-cardiovascular and off-pump procedures but remain outside clinically acceptable limits.