No significant association between perioperative serum sodium and intestinal anastomotic leaks: a systematic review and meta-analysis

Background Anastomotic leakage (AL) rates have been reported as high as 40% with associated mortality approaching 27%. Non-osmotic activation of vasopressin may result in hyponatremia. It has been suggested that perioperative sodium changes may function as an inexpensive and widely available biomarker for early identification of AL. Methods A systematic review and meta-analysis was performed, registered with PROSPERO (CRD42024522436). English language observational studies were eligible regardless of publication date. Methodological quality was evaluated using the QUADAS 2 tool. Quantitative synthesis was conducted using Meta Mar version 3.5.1. Results Among 121 screened records, five studies fulfilled the inclusion criteria. One study demonstrated a high risk of bias related to patient selection, while the remaining studies were assessed as low risk across all evaluated domains. The combined cohort included 2,034 patients, with AL occurring in 188 individuals (9%). Although all studies measured preoperative serum sodium, only three reported these values, and three studies assessed postoperative sodium levels. Pooled analysis showed no statistically significant differences in preoperative (SMD −0.45, p=0.33) or postoperative sodium (SMD −3.61, p=0.15) between patients with and without AL. In contrast, individual studies consistently demonstrated lower postoperative sodium levels among patients who developed AL that were statistically significant. Two studies identified postoperative hyponatremia or a postoperative decline in sodium as an independent predictor of AL. Reported sodium cutoffs ranging from 130 to 139.5 mEq/L yielded sensitivities between 23% and 70% and specificities between 72% and 93%. Conclusions Current evidence does not demonstrate a statistically significant association between perioperative serum sodium levels and anastomotic leakage. While individual studies have suggested a potential relationship, the substantial heterogeneity and methodological limitations limit the reliability of these findings. Further well-designed prospective studies are required before serum sodium can be considered a clinically useful biomarker.