OBJECTIVE: Using a large national database, we sought to better define the relationship between obesity measures and early clinical outcomes following mitral valve surgery for degenerative disease. METHODS: For the outcomes of in-hospital mortality, postoperative cerebrovascular event (CVA), and deep sternal wound infection (DSWI), a retrospective cohort study was performed using data acquired from the United Kingdom National Adult Cardiac Surgery Audit. Multivariable Cox proportional hazard regression modeling was used to investigate associations with individual measures of obesity. Progressively adjusted body mass index (BMI)-specific hazard ratios (HRs) were plotted against mean BMI values in each World Health Organization category using floated variances to investigate specific shapes of association. RESULTS: Multivariable Cox proportional hazard modeling failed to demonstrate an association between mortality and an increase in BMI of 5 points (HR, 0.93, 95% confidence interval [CI], 0.81-1.07), a BMI quintile increase (HR, 0.98; 95% CI, 0.90-1.07), or being classed "obese" by World Health Organization standards (HR, 1.03; 95% CI, 0.74-1.42). A 5-point BMI increase was associated with an increased hazard of DSWI (HR, 1.38; 95% CI, 1.08-1.77) but was not associated with perioperative CVA (HR, 1.05; 95% CI, 0.91-1.21). The shape of association between BMI and mortality appeared approximately U-shaped. DSWI appeared linear, whereas CVA demonstrated an inverted U, or a possible hourglass. CONCLUSIONS: Although individual measures of obesity were not associated with an increased mortality risk on regression modeling, the U-shaped relationship between mortality and increasing BMI demonstrates lower mortality risks in lower obesity classes. Increasing BMI was associated with an increased hazard for DSWI.
OBJECTIVE: Using a large national database, we sought to better define the relationship between obesity measures and early clinical outcomes following mitral valve surgery for degenerative disease. METHODS: For the outcomes of in-hospital mortality, postoperative cerebrovascular event (CVA), and deep sternal wound infection (DSWI), a retrospective cohort study was performed using data acquired from the United Kingdom National Adult Cardiac Surgery Audit. Multivariable Cox proportional hazard regression modeling was used to investigate associations with individual measures of obesity. Progressively adjusted body mass index (BMI)-specific hazard ratios (HRs) were plotted against mean BMI values in each World Health Organization category using floated variances to investigate specific shapes of association. RESULTS: Multivariable Cox proportional hazard modeling failed to demonstrate an association between mortality and an increase in BMI of 5 points (HR, 0.93, 95% confidence interval [CI], 0.81-1.07), a BMI quintile increase (HR, 0.98; 95% CI, 0.90-1.07), or being classed "obese" by World Health Organization standards (HR, 1.03; 95% CI, 0.74-1.42). A 5-point BMI increase was associated with an increased hazard of DSWI (HR, 1.38; 95% CI, 1.08-1.77) but was not associated with perioperative CVA (HR, 1.05; 95% CI, 0.91-1.21). The shape of association between BMI and mortality appeared approximately U-shaped. DSWI appeared linear, whereas CVA demonstrated an inverted U, or a possible hourglass. CONCLUSIONS: Although individual measures of obesity were not associated with an increased mortality risk on regression modeling, the U-shaped relationship between mortality and increasing BMI demonstrates lower mortality risks in lower obesity classes. Increasing BMI was associated with an increased hazard for DSWI.