Two-way Interaction Effects of Perioperative Complications on 30-Day Mortality in General Surgery.

backgroundMultiple perioperative complications increase mortality risk, and certain complications synergistically increase this risk to a greater degree than might be expected if the complications were independent, but these effects are not well established.

methodsThis is a retrospective cohort study of 422,827 intraabdominal general surgery patients (American College of Surgeons National Surgical Quality Improvement Program 2005-2011). Eight complications were evaluated: acute respiratory failure (ARF), acute kidney injury (AKI), sepsis/septic shock, stroke, cardiac arrest (CA), myocardial infarction (MI), deep vein thrombosis/pulmonary embolus, and transfusion. Each combination of two complications (28 total) was modeled using a Cox model for 30-day mortality, with adjustment for preoperative comorbidities and risk factors. Additive interaction was determined with the relative excess risk due to interaction (RERI). A positive RERI indicates that the mortality risk with both complications is greater than the sum of the individual mortality risks. Bonferroni correction was applied (α = 0.05/28 = 0.0018).

resultsSeven combinations demonstrated positive interaction: sepsis-CA (RERI 88.1; p < 0.0001), ARF-AKI (RERI 50.5; p < 0.0001), AKI-sepsis (RERI 33.9; p < 0.0001), sepsis-stroke (RERI 33.9; p < 0.0001), ARF-stroke (RERI 32.3; p < 0.0001), AKI-MI (RERI 24.5; p = 0.0013), and ARF-sepsis (RERI 19.2; p < 0.0001). Two combinations demonstrated negative interaction: ARF-CA (RERI -65.1; p = 0.0017) and CA-transfusion (RERI -52.0, p < 0.0001).

conclusionsInteraction effects exist between certain complications to increase the risk of short-term mortality. ARF, AKI, sepsis, and stroke were most likely to be involved in positive interactions. Further research into the mechanisms for these effects will be necessary to develop strategies to minimize the compounding effects of multiple complications in the perioperative period.