Sepsis is a frequent and lethal syndrome. Multiple phase 2 and phase 3 clinical trials have failed to discover novel sepsis therapies that improve outcome (1). This failure has been attributed to the heterogeneity of study populations (2). Closer scrutiny of phenotypes and sub-phenotypes of patients that display strong survival signals in sepsis may enable us to understand novel mechanisms to improve treatment.
Body mass index (BMI) is one such phenotype (3–13). More than 25% of adults admitted to ICUs in the United States are overweight or obese (14–16). Observational studies suggest an association between obesity (BMI ≥ 30 kg/m2) and survival in sepsis, with an absolute mortality reduction ranging between 5% and 15% in obese compared with normal weight patients (4–9). The biologic basis for this survival advantage is unclear, but theories include greater metabolic reserve (17), renin-angiotensin system activation (18), and secretion of immunomodulatory mediators such as leptin and soluble-tumor necrosis factor receptor-2 by adipose tissue (19).
This obesity survival “paradox” has not been consistently reproduced (10–13) and studies of sepsis and obesity have been limited by small numbers and failure to adjust for potential confounders such as illness severity (11), comorbidities (10 , 11), site of infection (6 , 7 , 10–13), recent weight loss (4–13), or geographic location (4–6 , 10–12). Also, studies measured BMI at different time points relative to initiation of sepsis resuscitation (20–22). Obtaining weight measurements after fluid resuscitation can falsely elevate estimates. Likewise, measuring weights later in hospitalization may yield falsely low values if the patient has suffered weight loss due to prolonged hospitalization or sepsis catabolism. Both effects would confound meaningful interpretation of the impact of obesity on sepsis outcomes (21). Further, several studies used administrative codes to define sepsis (4–6 , 12), which lack granularity around the timing of sepsis and are less accurate than sepsis identification using clinical data (23).
Therefore, the true association between obesity and sepsis survival is unknown and compels further investigation with larger and more rigorous studies. Understanding whether obesity actually improves survival is important. If obesity is truly protective, the underlying mechanisms warrant additional research, which would yield insights into sepsis pathogenesis and novel sepsis therapies. If harmful, investigators may need to treat obese septic adults with different treatments. We, therefore, evaluated a large clinical data repository to rigorously measure associations between survival and BMI using detailed clinical data to identify sepsis and to adjust for potential confounders.
Study Design, Data Source, and Population
We performed patient care from a cohort of U.S. hospitals for which the electronic heath record systems are provided by Cerner (North Kansas City, MO); the hospitals are well distributed by region, volume, teaching status, and urban/rural status (24). This dataset has previously been used to evaluate frequency and trends in sepsis in the United States (23). We included adults greater than or equal to 20 years old admitted as inpatients or under observation status or who died in the emergency department in calendar years 2009–2015. The National Institutes of Health Office of Human Subject Research Protections deemed that institutional review board approval was not required as the study used de-identified data alone. The Strengthening the Reporting of Observational Studies in Epidemiology guidelines for reporting cohort studies were followed (25).
We defined sepsis as concurrent infection and organ dysfunction consistent with Sepsis-3 criteria (26). We defined presumed serious infections as a blood culture draw and administration of greater than or equal to 4 days of new antibiotics, including at least one IV antibiotic day, as previously described (23). We defined organ dysfunction using the Sequential Organ Failure Assessment (SOFA) score. Sepsis was deemed present in a patient with suspected infection and a concurrent increase in SOFA by greater than or equal to 2 points from baseline. Septic shock was defined as sepsis with a septic shock diagnosis code (International Classification of Diseases, 9th Edition, Clinical Modification 785.52) or use of greater than or equal to 1 vasopressor(s) within 2 days of the culture draw. Further details about the case detection method are provided in eAppendices 1–3 (Supplemental Digital Content 1, http://links.lww.com/CCM/E420).
We identified the subset of sepsis cases with BMI, weight, or height data on or preceding the day of sepsis diagnosis (defined as the index blood culture draw). For those without direct BMI measurements, we calculated BMI as weight in kilograms divided by height in meters squared (kg/m2). We categorized BMI as follows: underweight (< 18.5 kg/m2); normal weight (18.5–24.9 kg/m2); overweight (25.0–29.9 kg/m2); obese class I (30.0–34.9 kg/m2); obese class II (35.0–39.9 kg/m2); and obese class III (≥ 40.0 kg/m2, morbidly obese) (3). We excluded patients if their BMI values were less than 10 kg/m2 or greater than or equal to 100 kg/m2 or if they had discordant BMI data (i.e., two or more recorded BMI values that fell into ≥ 2 BMI categories on the latest date on or before sepsis diagnosis). Sepsis cases without BMI data were excluded.
Implementation and Statistical Analyses
Using normal BMI as the reference BMI category, we reported the proportion of patients with short-term mortality (defined as in-hospital death or discharge to hospice) and the adjusted odds ratios (aOR) of short-term mortality (aOR [95% CI]). Variables evaluated in previous studies (4–13 , 17–22) and others considered clinically relevant were included in our multivariate logistic regression models (eAppendices 4–6, Supplemental Digital Content 1, http://links.lww.com/CCM/E420). We used generalized estimating equations to account for clustering of patients from the same hospital (27). Because the normal weight BMI category was compared with the five other BMI categories, we reported a p value corrected for multiple testing (i.e., Bonferroni correction). We reported secondary outcomes of length of ICU and hospital stay across BMI categories. SAS version 9.4 (SAS Institute, Cary, NC) was used for all statistical analyses. Power calculations were conducted using R software version 3.4.0 (R Foundation for Statistical Computing, Auckland, New Zealand) using the powerMediation package. Source code for analyses related to statistical computing (e.g., R or SAS scripts) are provided (eAppendices 7 and 8, Supplemental Digital Content 1, http://links.lww.com/CCM/E420). The underlying assumptions of logistic regression were tested (eAppendix 9, Supplemental Digital Content 1, http://links.lww.com/CCM/E420).
We conducted prespecified sensitivity analyses in sepsis patients (eAppendix 7, Supplemental Digital Content 1, http://links.lww.com/CCM/E420). Because malnutrition confounds the association between obesity and mortality in critical illness (28), we used serum albumin data as a surrogate for malnutrition and performed a separate sensitivity analysis to determine whether outcomes across BMI categories varied by albumin level. To account for early versus late antibiotic administration, we performed a separate sensitivity analysis to assess whether outcomes varied across BMI categories by the timing of antibiotic initiation relative to the day of index blood culture draw (day –2, –1, 0, +1, and +2). Interactions were tested in certain variables (age, gender, ethnicity, albumin quartile, SOFA quartile, Elixhauser quartile, site of infection, and blood culture positivity) because mortality varied across BMI status and these variables. Last, we analyzed a separate cohort of patients with billing codes for gastrointestinal bleeding and at least one blood transfusion but without clinical markers of presumed infection to ascertain whether the association of BMI with short-term mortality was consistent in a select group of nonseptic hospitalized patients, in whom no effect had previously been described. This cohort with gastrointestinal bleeds was chosen as they were likely to have no infection, sepsis, or antibiotic administration, compared with cohorts with exacerbations of congestive heart failure or chronic obstructive pulmonary disease.
Among 2,529,158 adult encounters admitted to 139 hospitals between 2009 and 2015, we identified 138,511 sepsis encounters. Of these, 61,609 had no assessable BMI data, 6,859 had incomplete data, and 7,554 had BMI data only after sepsis diagnosis (Fig. 1). Complete data for 62,489 sepsis encounters and 55,038 adult inpatients with sepsis were available for analysis.
Baseline Characteristics of Sepsis Cases
Tables E1 and E2 (Supplemental Digital Content 1, http://links.lww.com/CCM/E420) summarizes baseline characteristics for 55,038 adults with sepsis according to BMI categories. The BMI proportions were underweight (6%; BMI < 18.5 kg/m2), normal weight (33%; BMI = 18.5–24.9 kg/m2), overweight (28%; BMI = 25.0–29.9 kg/m2), obese class I (16%; BMI = 30.0–34.9 kg/m2), obese class II (8%; BMI = 35.0–39.9 kg/m2), and obese class III (9%; BMI ≥ 40 kg/m2, morbidly obese). Proportions of patients initiating antibiotics on day –2, –1, 0, +1, and +2 relative to the index blood culture day were similar across BMI categories. Procedures performed following sepsis onset were also similar across BMI categories, except central venous catheter insertion which was more frequent in the morbidly obese than those of normal weight BMI (29% [1,415/4,881] vs 22% [3,930/18,164]). Adults with BMI data had similar baseline characteristics to those without BMI data, except for admission year and urban-rural setting (Table E4, Supplemental Digital Content 1, http://links.lww.com/CCM/E420).
Unadjusted mortality (death in hospital or discharge to hospice) was inversely proportional to BMI category: underweight (31%; 1,080/3,520), normal weight (24%; 4,300/18,164), overweight (19%; 2,897/15,193), obese class I (16%; 1,458/8,916), obese class II (16%; 678/4,364), and obese class III (14%; 704/4,881) (Table E3, Supplemental Digital Content 1, http://links.lww.com/CCM/E420). Median ICU length of stay (9 d [interquartile range (IQR), 6–15 d]) was similar across BMI categories for the 14,511 adults admitted to the ICU. Median hospital length of stay (8 d [IQR, 6–13 d]) was similar across BMI categories for the 55,038 adults.
On multivariate analysis, the aOR (95% CI) of short-term mortality (death or hospice) was 1.62 (1.50–1.74; p < 0.0001) for underweight BMI; 0.73 (0.70–0.77; p < 0.0001) for overweight BMI; 0.61 (0.57–0.66; p < 0.0001) for obese class I BMI; 0.61 (0.55–0.67; p < 0.0001) for obese class II BMI; and 0.65 (0.59–0.71; p < 0.0001) for obese class III BMI relative to normal weight BMI (Table E3, Supplemental Digital Content 1, http://links.lww.com/CCM/E420). These results were similar when patients with BMI data were divided into BMI quintiles and quintile 3 (including median BMI = 26.67) was used as the reference group in logistic regression.
Results were consistent in sensitivity analyses (Figs. 2–4; Table E5, Supplemental Digital Content 1, http://links.lww.com/CCM/E420; and Figs. E1–E15 [Supplemental Digital Content 2, http://links.lww.com/CCM/E421; Supplemental Digital Content 3, http://links.lww.com/CCM/E422; Supplemental Digital Content 4, http://links.lww.com/CCM/E423; and Supplemental Digital Content 5, http://links.lww.com/CCM/E424—legend, Supplemental Digital Content 6, http://links.lww.com/CCM/E425], respectively). Statistically significant interactions were demonstrated between BMI and gender (p = 0.035), and BMI and albumin quartile (p = 0.001); but not with age, ethnicity, SOFA quartile, Elixhauser quartile, site of infection, or blood culture positivity. In adults with albumin levels less than 2 g/dL, those with overweight and obese class I BMIs had improved survival but not those with obese class II or class III BMIs. Analysis of a separate cohort of patients with gastrointestinal bleeds and without infection also showed a similar trend of lower mortality with increasing BMI (Table E6, Supplemental Digital Content 1, http://links.lww.com/CCM/E420; and Fig. E16, Supplemental Digital Content 5, http://links.lww.com/CCM/E424—legend, Supplemental Digital Content 6, http://links.lww.com/CCM/E425).
In a cohort of 55,038 adults hospitalized with sepsis, short-term mortality (death or hospice) was lower in overweight and obese patients compared with those with normal weight BMIs. Underweight patients had increased mortality compared with normal weight BMIs. This relationship persisted after adjusting for multiple potential confounders including demographic factors, admission year, hospital-level factors, infection factors, and severity of illness. These results were consistent across several sensitivity analyses including comorbidities, site of infection, and timing of antibiotics.
At the turn of the 21st century, Fleischmann et al (29) first described improved survival with increasing BMI in a cohort of patients with end-stage renal disease. This association has since been reported in acute and chronic illnesses such as congestive heart failure (30), acute coronary syndrome (31), chronic obstructive pulmonary disease (32), and critical illness (33–35). Prior systematic reviews (21 , 22) have suggested improved sepsis survival with increasing BMI and worse survival with decreasing BMI, but their interpretations were limited by weaknesses of their source studies including low power due to small sample sizes, inconsistent timing of BMI measurements, and incomplete adjustment for potential confounders (21). Our study used detailed clinical, laboratory, and physiologic data to strengthen the finding that obesity is independently associated with improved short-term survival in adults with sepsis.
We found better short-term survival in those with higher BMI in sepsis. Several large well-conducted studies have reported poorer long-term survival with higher BMI in healthy adults and those with diabetes (36–38). We consider these investigations to be complementary since our study evaluated short-term outcomes (39) and does not preclude the possibility of long-term adverse consequences of obesity in sepsis. Notably, two studies with both short- and long-term outcomes in surgical sepsis and pneumonia reported improved 30-d survival in the obese but no difference in 5-year survival (35 , 40). We observed improved survival in obese (vs normal weight) patients in a separate cohort of nonseptic adults with gastrointestinal bleeds, a population in which, to the best of our knowledge, this phenomenon has not been previously described. This observation along with evidence from previous studies on other nonseptic populations (30–32) suggests that the survival advantage associated with higher BMI may not be sepsis-specific but a general pattern in acute illness.
We observed lower mortality rates in patients with higher BMIs and caution against labeling this as “the obesity paradox.” BMI is a poor proxy for adiposity and may not accurately represent fat content, the proportions of muscle and fat, or overall body composition (41). Future prospective studies should quantify adipose tissue at or before sepsis diagnosis by either performing anthropometric indices of increased adiposity (e.g., waist circumference, waist-to-hip ratio, percentage visceral versus total body fat ); or assessing body composition with CT or dual-energy x-ray absorptiometry. Abdominal CT to measure visceral adipose tissue to subcutaneous adipose tissue ratios may be informative, but its utility is diminished if performed after fluid resuscitation in sepsis (43). Prior studies reported that the absolute quantity of muscle tissue is increased in those with obese BMIs compared with those with normal weight BMIs (44), and that muscle wasting occurs early and rapidly during the first week of critical illness and is more severe among those with multiple organ failure (45). Finally, low skeletal muscle area as assessed by CT scan during the early stage of critical illness is an independent risk factor for mortality (46). Future studies should therefore also explore the contribution of muscle mass on sepsis survival.
We detected a significant interaction between BMI and albumin levels, a surrogate marker for malnutrition. This finding echoes that of a prior study where obese critically ill patients had similar survival to nonobese ones after adjusting for objective measures of malnutrition (28). However, sample sizes in these groups were relatively small. Also, albumin is a negative acute-phase protein, so it is unclear whether these low albumin levels represent malnutrition or immune response differences. Investigators should determine whether better biomarkers of malnutrition can be used than albumin in large multicenter studies (47 , 48)
Reports from animal models of obesity and sepsis are inconsistent with our findings. Obese mice and rats are at higher risk of death than nonobese ones (49). These models use young animals in contrast to elderly human patients who develop sepsis; survival is not risk-adjusted for illness severity or baseline imbalances at the time of sepsis onset; and animals do not receive antimicrobial therapy or other supportive measures. These differences in baseline factors and interventions make it difficult to analyze and extrapolate the effect of obesity in animal models to human disease.
We highlight several strengths of our study. Although retrospective in nature and investigating a known phenomenon among patients with sepsis, this is the largest analysis conducted to date assessing the association between BMI and sepsis mortality. Our large cohort included sepsis patients admitted to over 130 U.S. hospitals that were well distributed by region. Thus, our findings are unlikely to be biased by regional hospital practices. Furthermore, our results are representative of more recent patterns of sepsis care compared with another large database study from 1996 to 2008 that predates our study period (4). We used detailed electronic clinical data to apply SOFA/Sepsis 3 definitions rather than using administrative codes for sepsis (23). We used BMI data on or before the day of sepsis onset (i.e., dry body weight) to minimize potential weight gain due to fluid resuscitation (9) or potential weight loss following the catabolic effects of sepsis (45). We adjusted for several known confounders such as illness severity, comorbidities, site of infection, recent weight loss, or geographic location. We performed several sensitivity analyses, which showed a consistent association between BMI and sepsis survival. We specifically did not adjust for other interventions, such as mechanical ventilation or dialysis following sepsis onset, as these could mediate the effect of BMI on sepsis survival. Our findings are unlikely to have a “selective-survivor” effect as our cohort included all adults 20 years old or older. For example, if we had limited our cohort to only those adults 65 years old or older, it is possible that healthier obese patients would survive to age 65, but less healthy adults may die before age 65. Despite several strengths of study design, our analysis is not without limitations.
Our findings are hypothesis-generating and do not imply causality. Residual confounding remains possible because quantitative data on smoking status (50) and weight changes were not included in our multivariate regression model. However, we adjusted our findings using the Elixhauser comorbidity index, which includes illnesses caused by smoking such as cardiac disease, chronic pulmonary disease, and cancer. Also, our multivariate model adjusted for qualitative weight loss using the Elixhauser comorbidity index. We did not determine the appropriateness of antibiotics because pathogens were not always identified and susceptibility data were not always available. Our findings may not be generalizable to other countries or those younger than 20 years old. Despite a comprehensive search algorithm, BMI was available or could be indirectly calculated in only 49% of sepsis encounters, which exceeded the BMI availability (32%) of the largest prior retrospective study that explored this question (4). Finally, the sizable proportion of missing BMI data was unlikely to have introduced systematic bias as the populations with and without BMI availability were similar.
Future prospective studies that aim to assess the effect of BMI on sepsis outcome will need to: 1) uniformly record weight and height measurements; 2) quantify body composition, smoking consumption, and weight loss; 3) adjust for potential confounders; 4) report the effects of interventions used in sepsis; and 5) assess long-term mortality. Based on data from our study, we estimate that such a cohort will need to enroll over 8,500 sepsis cases (Table E7, Supplemental Digital Content 1, http://links.lww.com/CCM/E420) to have sufficient statistical power to detect true mortality differences across all BMI groups using normal weight BMI as the reference group.
In conclusion, our analysis of a large data repository shows that higher BMIs are associated with higher sepsis short-term survival after controlling for potential confounders. These findings emphasize the need for well-conducted prospective clinical trials that assess obesity’s impact on survival in sepsis and that aim to understand the exact underlying mechanisms. These studies could enhance our understanding of sepsis pathogenesis and inform therapeutic strategies.
1. Marshall JC. Why have clinical trials in sepsis
failed? Trends Mol Med 2014; 20:195–203
2. Scicluna BP, van Vught LA, Zwinderman AH, et al; MARS consortium: Classification of patients with sepsis
according to blood genomic endotype: A prospective cohort study. Lancet Respir Med 2017; 5:816–826
4. Arabi YM, Dara SI, Tamim HM, et al; Cooperative Antimicrobial Therapy of Septic Shock (CATSS) Database Research Group: Clinical characteristics, sepsis
interventions and outcomes in the obese patients with septic shock: An international multicenter cohort study. Crit Care 2013; 17:R72
5. Gaulton TG, Marshall MacNabb C, Mikkelsen ME, et al. A retrospective cohort study examining the association between body mass index
in severe sepsis
. Intern Emerg Med 2015; 10:471–479
6. Prescott HC, Chang VW, O’Brien JM Jr, et al. Obesity
and 1-year outcomes in older Americans with severe sepsis
. Crit Care Med 2014; 42:1766–1774
7. Sakr Y, Madl C, Filipescu D, et al. Obesity
is associated with increased morbidity but not mortality
in critically ill patients. Intensive Care Med 2008; 34:1999–2009
8. Wacharasint P, Boyd JH, Russell JA, et al. One size does not fit all in severe infection: Obesity
alters outcome, susceptibility, treatment, and inflammatory response. Crit Care 2013; 17:R122
9. Wurzinger B, Dünser MW, Wohlmuth C, et al. The association between body-mass index and patient outcome in septic shock: A retrospective cohort study. Wien Klin Wochenschr 2010; 122:31–36
10. Adamzik M, Frey UH, Möhlenkamp S, et al. Aquaporin 5 gene promoter–1364A/C polymorphism associated with 30-day survival
in severe sepsis
. Anesthesiology 2011; 114:912–917
11. Chalkias A, Nitsotolis T, Papalexandrou A, et al. Sagittal abdominal diameter may effectively predict future complications and increased mortality
in intensive care unit patients with severe sepsis
. J Crit Care 2013; 28:964–969
12. Kuperman EF, Showalter JW, Lehman EB, et al. The impact of obesity
on sepsis mortality
: A retrospective review. BMC Infect Dis 2013; 13:377
13. Sakr Y, Alhussami I, Nanchal R, et al; Intensive Care Over Nations Investigators: Being overweight is associated with greater survival
in ICU patients: Results from the intensive care over nations audit. Crit Care Med 2015; 43:2623–2632
14. Akinnusi ME, Pineda LA, El Solh AA. Effect of obesity
on intensive care morbidity and mortality
: A meta-analysis. Crit Care Med 2008; 36:151–158
15. Oliveros H, Villamor E. Obesity
in critically ill adults: A systematic review and meta-analysis. Obesity
(Silver Spring) 2008; 16:515–521
16. Hogue CW Jr, Stearns JD, Colantuoni E, et al. The impact of obesity
on outcomes after critical illness
: A meta-analysis. Intensive Care Med 2009; 35:1152–1170
17. Ng PY, Eikermann M. The obesity
conundrum in sepsis
. BMC Anesthesiol 2017; 17:147
18. Kershaw EE, Flier JS. Adipose tissue as an endocrine organ. J Clin Endocrinol Metab 2004; 89:2548–2556
19. Winkler G, Kiss S, Keszthelyi L, et al. Expression of tumor necrosis factor (TNF)-alpha protein in the subcutaneous and visceral adipose tissue in correlation with adipocyte cell volume, serum TNF-alpha, soluble serum TNF-receptor-2 concentrations and C-peptide level. Eur J Endocrinol 2003; 149:129–135
20. Trivedi V, Bavishi C, Jean R. Impact of obesity
on sepsis mortality
: A systematic review. J Crit Care 2015; 30:518–524
21. Pepper DJ, Sun J, Welsh J, et al. Increased body mass index
and adjusted mortality
in ICU patients with sepsis
or septic shock: A systematic review and meta-analysis. Crit Care 2016; 20:181
22. Wang S, Liu X, Chen Q, et al. The role of increased body mass index
in outcomes of sepsis
: A systematic review and meta-analysis. BMC Anesthesiol 2017; 17:118
23. Rhee C, Dantes R, Epstein L, et al; CDC Prevention Epicenter Program: Incidence and trends of sepsis
in US hospitals using clinical vs claims data, 2009-2014. JAMA 2017; 318:1241–1249
24. DeShazo JP, Hoffman MA. A comparison of a multistate inpatient EHR database to the HCUP Nationwide Inpatient Sample. BMC Health Serv Res 2015; 15:384
25. von Elm E, Altman DG, Egger M, et al; STROBE Initiative: The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: Guidelines for reporting observational studies. Lancet 2007; 370:1453–1457
26. Singer M, Deutschman CS, Seymour CW, et al. The third international consensus definitions for sepsis
and septic shock (sepsis
-3). JAMA 2016; 315:801–810
27. Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika 1986; 73:13–22
28. Robinson MK, Mogensen KM, Casey JD, et al. The relationship among obesity
, nutritional status, and mortality
in the critically ill. Crit Care Med 2015; 43:87–100
29. Fleischmann E, Teal N, Dudley J, et al. Influence of excess weight on mortality
and hospital stay in 1346 hemodialysis patients. Kidney Int 1999; 55:1560–1567
30. Oreopoulos A, Padwal R, Kalantar-Zadeh K, et al. Body mass index
in heart failure: A meta-analysis. Am Heart J 2008; 156:13–22
31. Niedziela J, Hudzik B, Niedziela N, et al. The obesity
paradox in acute coronary syndrome: A meta-analysis. Eur J Epidemiol 2014; 29:801–812
32. Singanayagam A, Schembri S, Chalmers JD. Predictors of mortality
in hospitalized adults with acute exacerbation of chronic obstructive pulmonary disease. Ann Am Thorac Soc 2013; 10:81–89
33. Hutagalung R, Marques J, Kobylka K, et al. The obesity
paradox in surgical intensive care unit patients. Intensive Care Med 2011; 37:1793–1799
34. Pickkers P, de Keizer N, Dusseljee J, et al. Body mass index
is associated with hospital mortality
in critically ill patients: An observational cohort study. Crit Care Med 2013; 41:1878–1883
35. Utzolino S, Ditzel CM, Baier PK, et al. The obesity
paradox in surgical intensive care patients with peritonitis. J Crit Care 2014; 29:887.e1–887.e5
36. Di Angelantonio E, Bhupathiraju ShN, Wormser D, et al. Body-mass index and all-cause mortality
: Individual-participant-data meta-analysis of 239 prospective studies in four continents. Lancet 2016; 388:776–786
37. Tobias DK, Pan A, Jackson CL, et al. Body-mass index and mortality
among adults with incident type 2 diabetes. N Engl J Med 2014; 370:233–244
38. Khan SS, Ning H, Wilkins JT, et al. Association of body mass index
with lifetime risk of cardiovascular disease and compression of morbidity. JAMA Cardiol 2018; 3:280–287
39. Pepper DJ, Sun J, Suffredini AF, et al. Body-mass index and all-cause mortality
. Lancet 2017; 389:2284
40. Prescott HC, Chang VW. Overweight or obese BMI is associated with earlier, but not later survival
after common acute illnesses. BMC Geriatr 2018; 18:42
41. Christopher KB. The body mass index
paradox. Crit Care Med 2015; 43:2693–2694
42. Pichard C, Kyle UG, Morabia A, et al. Nutritional assessment: Lean body mass depletion at hospital admission is associated with an increased length of stay. Am J Clin Nutr 2004; 79:613–618
43. Pepper DJ. Adipose tissue measurements in patients with sepsis
. Crit Care Med 2017; 45:e331
44. Woo J, Leung J, Kwok T. BMI, body composition, and physical functioning in older adults. Obesity
(Silver Spring) 2007; 15:1886–1894
45. Puthucheary ZA, Rawal J, McPhail M, et al. Acute skeletal muscle wasting in critical illness
. JAMA 2013; 310:1591–1600
46. Weijs PJ, Looijaard WG, Dekker IM, et al. Low skeletal muscle area is a risk factor for mortality
in mechanically ventilated critically ill patients. Crit Care 2014; 18:R12
47. Elia M, Lunn PG. Biological markers of protein-energy malnutrition. Clin Nutr 1997; 16(Suppl 1):11–17
48. Combs GF Jr, Trumbo PR, McKinley MC, et al. Biomarkers in nutrition: New frontiers in research and application. Ann N Y Acad Sci 2013; 1278:1–10
49. Mittwede PN, Clemmer JS, Bergin PF, et al. Obesity
and critical illness
: Insights from animal models. Shock 2016; 45:349–358
50. Huttunen R, Laine J, Lumio J, et al. Obesity
and smoking are factors associated with poor prognosis in patients with bacteraemia. BMC Infect Dis 2007; 7:13