Adverse perioperative outcomes among patients undergoing gastrointestinal cancer surgery: Quantifying attributable risk from malnutrition.

Background: Preoperative malnutrition adversely impacts perioperative outcomes among patients with gastrointestinal (GI) cancer. The attributable risk (AR) that nutrition status contributes towards negative outcomes is poorly understood.

Methods: Adults undergoing GI cancer surgeries were identified within the American College of Surgeons National Surgical Quality Improvement Program database (2005-2017). Emergency surgeries, outpatients, and cases with an American Society of Anesthesiologists status above III were excluded. Adjusted multivariable models were constructed to determine the associations between markers of nutrition status (body mass index, >10% weight loss in last 6 months, functional status, and serum albumin level) and adverse perioperative outcomes (presence and number of complications, death, 30-day readmission, and length of stay). Predictive accuracy statistics and population AR (PAR) were determined.

Results: The final sample included 78,662 cases. Patients with >10% weight loss 6 months preceding surgery (compared with those who did not), had a significantly increased risk of complications (Relative Risk = 1.28; 95% CI, 1.20-1.37) and odds of death (odds ratio [OR] = 1.37; 95% CI, 1.18-1.59). A totally dependent functional status (compared with independent status) was associated with a 3.3-times higher odds of death (OR = 3.30; 95% CI, 1.53-7.15). Multivariable models were not predictive of adverse outcomes; PAR from the markers ranged 1%-2%.

Conclusion: Ten percent weight loss in preceding 6 months was associated with increased risk of adverse perioperative outcomes among adults undergoing GI cancer surgery. The contribution of nutrition status markers to surgical outcomes as assessed by PAR was small (1%-2%), a finding not previously reported. Future intervention studies should include validated nutrition risk markers, control for effects of perioperative variables, and evaluate PAR within the immediate/long-term postoperative periods.