Objective:

To assess the association between perioperative nutritional support and early postoperative albumin recovery indicators after gastrectomy for gastric cancer, and to develop a prediction model for postoperative albumin improvement.

Approach:

• Study Design: Retrospective cohort study of 1,529 adults who underwent gastrectomy for gastric cancer, with preoperative and early postoperative albumin measurements.
• Nutritional Support Definition: Perioperative nutritional support was classified as binary based on available dataset.
• Outcome Measures: Primary outcome was postoperative albumin improvement; secondary outcomes included total protein improvement and change in albumin.
• Statistical Analysis: Multivariable linear and logistic regression analyses, and a gradient boosting model using 46 preoperative and perioperative variables.

Key Findings:

• Albumin improved in 233 of 1,529 patients (15.2%), with higher rates in the nutritional support group (19.2%) compared to the no/standard support group (12.8%).
• Nutritional support was associated with a higher Δ albumin (β = 0.572 g/L, p = 0.015).
• The gradient boosting model achieved an AUROC of 0.711 and an AUPRC of 0.422.

Interpretation:

The prediction model demonstrated moderate discrimination for early albumin improvement.

Limitations:

• The study is retrospective and may have inherent biases.
• External validation and prospective evaluation of the prediction model are needed.

Conclusion:

Further investigation is warranted.