The relationship between body mass index and mortality is not linear in patients requiring venovenous extracorporeal support
Elwin Tham,Stuart Campbell,Heather Hayanga,Jeffrey Ammons,Wei Fang,Penny Sappington,Paul McCarthy,Alper Toker,Vinay Badhwar,J W Awori Hayanga
DOI: https://doi.org/10.1016/j.jtcvs.2023.11.041
Abstract:Objective: Morbid obesity may influence candidacy for venovenous extracorporeal membrane oxygenation (VVECMO) support. Indeed, body mass index (BMI) >40 is considered to be a relative contraindication due to increased mortality observed in patients with BMI above this value. There is scant evidence to characterize this relationship beyond speculating about the technical challenges of cannulation and difficulty in optimizing flows. We examined a national cohort to evaluate the influence of BMI on mortality in patients requiring VVECMO for severe acute respiratory syndrome coronavirus 2 infection. Methods: We performed a retrospective cohort analysis on National COVID Cohort Collaborative data evaluating 1,033,229 patients with BMI ≤60 from 31 US hospital systems diagnosed with severe acute respiratory syndrome virus coronavirus 2 infection from September 2019 to August 2022. We performed univariate and multivariable mixed-effects logistic regression analysis on data pertaining to those who required VVECMO support during their hospitalization. A subgroup risk-adjusted analysis comparing ECMO mortality in patients with BMI 40 to 60 with the 25th, 50th, and 75th BMI percentile was performed. Outcomes of interest included BMI, age, comorbidity score, body surface area, and ventilation days. Results: A total of 774 adult patients required VVECMO. Of these, 542 were men, median age was 47 years, mean adjusted Charlson Comorbidity Index was 1, and median BMI was 33. Overall mortality was 47.8%. There was a nonsignificant overall difference in mortality across hospitals (SD, 0.31; 95% CI, 0-0.57). After mixed multivariable logistic regression analysis, advanced age (P < .0001) and Charlson Comorbidity Index (P = .009) were each associated with increased mortality. Neither gender (P = .14) nor duration on mechanical ventilation (P = .39) was associated with increased mortality. An increase in BMI from 25th to 75th percentile was not associated with a difference in mortality (P = .28). In our multivariable mixed-effects logistic regression analysis, there exists a nonlinear relationship between BMI and mortality. Between BMI of 25 and 32, patients experienced an increase in mortality. However, between BMI of 32 and 37, the adjusted mortality in these patients subsequently decreased. Our subgroup analysis comparing BMIs 40 to 60 with the 25th, 50th, and 75th percentile of BMI found no significant difference in ECMO mortality between BMI values of 40 and 60 with the 25th, 50th, 75th percentile. Conclusions: Advancing age and higher CCI are each associated with increased risk for mortality in patients requiring VVECMO. A nonlinear relationship exists between mortality and BMI and those between 32 and 37 have lower odds of mortality than those between BMI 25 and 32. This nonlinear pattern suggests a need for further adjudication of the contraindications associated with VVECMO, particularly those based solely on BMI.