BACKGROUND: Recent studies in the USA have shown a lower postoperative mortality rate in mildly obese patients, described as the 'obesity paradox'. The results from the relatively obese population in Western countries may not be generalizable to Asian countries, prompting the present study to investigate the relationship between body mass index (BMI) and outcomes after gastrointestinal surgery. METHODS: Patients who underwent gastrectomy or colorectal resection for stage I-III cancer between July and December 2010 were identified from a nationwide inpatient database in Japan. Multivariable logistic regression models for in-hospital mortality and postoperative complications, and a linear regression model for total costs were established, with adjustment for age, sex, co-morbidities, cancer stage and BMI. Restricted cubic spline functions were used to consider potential non-linear associations between BMI and the outcomes. RESULTS: Among 30 765 eligible patients, associations between BMI and the outcomes were U-shaped, with the lowest mortality, morbidity and total costs in patients with a BMI of around 23·0 kg/m(2) . A BMI of 18·5 kg/m(2) was associated with significantly greater mortality (odds ratio (OR) 2·04, 95 per cent confidence interval 1·64 to 2·55), postoperative complications (OR 1·10, 1·03 to 1·18) and total costs (difference €1389, 1139 to 1640) compared with a BMI of 23·0 kg/m(2) . Patients with a BMI exceeding 30·0 kg/m(2) had significantly higher rates of postoperative complications and total costs than those with a BMI of 23·0 kg/m(2) , but no significant association was evident between a BMI of more than 23·0 kg/m(2) and in-hospital death. CONCLUSION: Unlike previous studies in the USA, in the present national Japanese cohort of patients undergoing surgery for gastrointestinal cancer, those who were either underweight or overweight had more postoperative complications and greater perioperative costs than those of normal weight.
BACKGROUND: Recent studies in the USA have shown a lower postoperative mortality rate in mildly obesepatients, described as the 'obesity paradox'. The results from the relatively obese population in Western countries may not be generalizable to Asian countries, prompting the present study to investigate the relationship between body mass index (BMI) and outcomes after gastrointestinal surgery. METHODS:Patients who underwent gastrectomy or colorectal resection for stage I-III cancer between July and December 2010 were identified from a nationwide inpatient database in Japan. Multivariable logistic regression models for in-hospital mortality and postoperative complications, and a linear regression model for total costs were established, with adjustment for age, sex, co-morbidities, cancer stage and BMI. Restricted cubic spline functions were used to consider potential non-linear associations between BMI and the outcomes. RESULTS: Among 30 765 eligible patients, associations between BMI and the outcomes were U-shaped, with the lowest mortality, morbidity and total costs in patients with a BMI of around 23·0 kg/m(2) . A BMI of 18·5 kg/m(2) was associated with significantly greater mortality (odds ratio (OR) 2·04, 95 per cent confidence interval 1·64 to 2·55), postoperative complications (OR 1·10, 1·03 to 1·18) and total costs (difference €1389, 1139 to 1640) compared with a BMI of 23·0 kg/m(2) . Patients with a BMI exceeding 30·0 kg/m(2) had significantly higher rates of postoperative complications and total costs than those with a BMI of 23·0 kg/m(2) , but no significant association was evident between a BMI of more than 23·0 kg/m(2) and in-hospital death. CONCLUSION: Unlike previous studies in the USA, in the present national Japanese cohort of patients undergoing surgery for gastrointestinal cancer, those who were either underweight or overweight had more postoperative complications and greater perioperative costs than those of normal weight.