The Relationship Among Obesity, Nutritional Status, and Mortality in the Critically Ill* Malcolm K. Robinson, MD1; Kris M. Mogensen, MS, RD, LDN, CNSC2; Jonathan D. Casey, MD3; Caitlin K. McKane, BS, RN4; Takuhiro Moromizato, MD5; James D. Rawn, MD1; Kenneth B. Christopher, MD6

Introduction: The association between obesity and mortality in critically ill patients is unclear based on the current literature. To clarify this relationship, we analyzed the association between obesity and mortality in a large population of critically ill patients and hypothesized that mortality would be impacted by nutritional status. Methods: We performed a single-center observational study of 6,518 adult patients treated in medical and surgical ICUs between 2004 and 2011. All patients received a formal, in-person, and standardized evaluation by a registered dietitian. Body mass index was determined at the time of dietitian consultation from the estimated dry weight or hospital admission weight and categorized a priori as less than 18.5 kg/m2 (underweight), 18.5–24.9 kg/m2 (normal/referent), 25–29.9 kg/m2 (overweight), 30–39.9  kg/m2 *See also p. 240. 1 Department of Surgery, Brigham and Women’s Hospital, Boston, MA. 2 Department of Nutrition, Brigham and Women’s Hospital, Boston, MA. 3 Department of Medicine, Brigham and Women’s Hospital, Boston, MA. 4 Department of Nursing, Brigham and Women’s Hospital, Boston, MA. 5 Department of Medicine, Hokubu Prefectural Hospital, Kunigami District, Okinawa Prefecture, Japan. 6 The Nathan E. Hellman Memorial Laboratory, Renal Division, Brigham and Women’s Hospital, Boston, MA. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website (http://journals.lww.com/ ccmjournal). Dr. Robinson provided expert testimony for various law firms. Ms. Mogensen served as board member for ThriveRx (member of the Nutrition Advisory Board for ThriveRx, which is a home infusion company); lectured for Massachusetts Dietetic Association (honorarium for lecture on essential fatty acid deficiency, presented at the Association Annual Meeting on April 11, 2014) and for the American Society for Parenteral and Enteral Nutrition (honoraria for speaking at Clinical Nutrition Week 2014 and presentation of a Webinar on critical care nutrition on April 8, 2014); and received royalties from Wolf Rinke Associates (royalties for the publication/continuing education self-study course “Nutritional Support of the Critically Ill Adult”). The remaining authors have disclosed that they do not have any potential conflicts of interest. For information regarding this article, E-mail: [email protected] Copyright © 2014 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/CCM.0000000000000602

Critical Care Medicine

(obesity class I and II), and more than or equal to 40.0 kg/m2 (obesity class III). Malnutrition diagnoses were categorized as nonspecific malnutrition, protein-energy malnutrition, or well nourished. The primary outcome was all-cause 30-day mortality determined by the Social Security Death Master File. Associations between body mass index groups and mortality were estimated by bivariable and multivariable logistic regression models. Adjusted odds ratios were estimated with inclusion of covariate terms thought to plausibly interact with both body mass index and mortality. We utilized propensity score matching on baseline characteristics and nutrition status to reduce residual confounding of the body mass index category assignment. Results: In the cohort, 5% were underweight, 36% were normal weight, 31% were overweight, 23% had class I/II obesity, and 5% had class III obesity. Nonspecific malnutrition was present in 56%, protein-energy malnutrition was present in 12%, and 32% were well nourished. The 30-day and 90-day mortality rate for the cohort was 19.1 and 26.6%, respectively. Obesity is a significant predictor of improved 30-day mortality following adjustment for age, gender, race, medical versus surgical patient type, DeyoCharlson index, acute organ failure, vasopressor use, and sepsis: underweight odds ratio 30-day mortality is 1.09 (95% CI, 0.80– 1.48), overweight 30-day mortality odds ratio is 0.93 (95% CI, 0.80–1.09), class I/II obesity 30-day mortality odds ratio is 0.80 (95% CI, 0.67–0.96), and class III obesity 30-day mortality odds ratio is 0.69 (95% CI, 0.49–0.97), all relative to patients with body mass index 18.5–24.9 kg/m2. Importantly, there is confounding of the obesity-mortality association on the basis of malnutrition. Adjustment for only nutrition status attenuates the obesity–30-day mortality association: underweight odds ratio is 0.74 (95% CI, 0.54–1.00), overweight odds ratio is 1.05 (95% CI, 0.90–1.23), class I/II obesity odds ratio is 0.96 (95% CI, 0.81–1.15), and class III obesity odds ratio is 0.81 (95% CI, 0.59–1.12), all relative to patients with body mass index 18.5–24.9 kg/m2. In a subset of patients with body mass index more than or equal to 30.0 kg/m2 (n = 1,799), those with either nonspecific or protein-energy malnutrition have increased mortality relative to well-nourished patients with body mass index more than or equal to 30.0 kg/m2: odds ratio of 90-day mortality is 1.67 (95% CI, 1.29–2.15; p < 0.0001), fully adjusted. In a cohort of propensity score matched patients www.ccmjournal.org

87

Robinson et al (n = 3,554), the body mass index–mortality association was not statistically significant, likely from matching on nutrition status. Conclusions: In a large population of critically ill adults, the association between improved mortality and obesity is confounded by malnutrition status. Critically ill obese patients with malnutrition have worse outcomes than obese patients without malnutrition. (Crit Care Med 2015; 43:87–100) Key Words: body mass index; critical illness; malnutrition; mortality; nutrition; obesity

O

besity is a well-recognized problem in the United States with an estimated 34% of adults having a body mass index (BMI) of 30.0 kg/m2 or higher (1). This is of great concern not only because obesity-related treatments account for an estimated 21% of medical spending but, more importantly, because of the presumed impact of obesity on mortality risk (2). Although it is well documented that obese individuals have shorter life expectancies over the long term (3, 4), it is unclear whether obesity confers an increased risk of death for hospitalized patients. One might naturally expect that obese critically ill patients are at increased risk of death compared with those of normal weight. Obesity is associated with a substantially increased risk of hypertension, dyslipidemia, type 2 diabetes, coronary heart disease, and respiratory problems, all of which could theoretically and adversely impact survival in the ICU (3, 4). It may be difficult to perform diagnostic and therapeutic procedures in obese individuals making their care more challenging. Furthermore, the hospitalized obese patient is at increased risk for life-threatening complications, such as thromboembolic events, pulmonary complications, catheter-related sepsis, and sepsis in general and acute kidney injury (5–11). However, the expectation that obesity increases mortality risk in the critically ill is far from established in the literature. Studies in the critically ill show that obesity is associated with increased ICU and/or in-hospital mortality (12–15), no difference in mortality (16–20) and improved mortality (21–25) relative to nonobese patients. It is unclear why there are such disparate results regarding the obesity-mortality relationship in the ICU. One possibility is that nutritional status is not known or properly controlled for in these studies, which leads to the observed widely divergent results. Of note, there appears to be a correlation between malnutrition and outcomes of hospitalized patients (26–34), and malnutrition is present in a substantial minority of the hospitalized (35) and in critically ill medical and surgical patients (36, 37). Furthermore, patients with morbid obesity under evaluation for bariatric surgery are found to have a high prevalence of nutritional deficiencies (38–40). As malnutrition in obese patients appears prevalent (38– 40) and nutrition status may play a role in patient outcomes (26–34), we hypothesized that nutrition status is a driver of mortality in the obese critically ill. To test this hypothesis, we performed a single-center observational study of critically ill 88

www.ccmjournal.org

patients among whom a formal structured nutrition evaluation was performed by a registered dietitian to define nutritional status. The objective of this study was to determine the impact of nutrition status on the relationship between BMI and mortality in critically ill patients.

MATERIALS AND METHODS Source Population Administrative and laboratory data from individuals admitted to medical and surgical ICUs at the Brigham and Women’s Hospital (BWH) in Boston, MA, served as the source population. BWH is a 793-bed teaching hospital with 100 ICU beds. BWH provides primary and tertiary care to an ethnically and socioeconomically diverse population within eastern Massachusetts and the surrounding region. Data Sources Nutrition data on critically ill patients were collected prospectively at the time of initial nutrition consultation by a registered dietitian between 2004 and 2011 and stored in an electronic database built for research and quality improvement purposes. Mortality data and International Classification of Diseases, 9th Edition, Clinical Modification (ICD-9-CM), coding data were obtained from a central computerized administrative data registry called the Research Patient Data Registry (RPDR) (41) that serves as a central clinical data warehouse for all inpatient and outpatient records at Partners HealthCare sites including BWH. The RPDR has been used for other clinical research studies, and mortality and coding data from the RPDR have been validated in at least one prior study (42). Approval for the study was granted by the Partners Human Research Committee Institutional Review Board. At the BWH during 2004–2011, there were a total of 25,686 hospitalized patients, who are 18 years old or older, who were assigned the Current Procedural Terminology (CPT) code 99291 (critical care, first 30–74 min). We have previously validated the accuracy of CPT code 99291 assignment for ICU admission in our administrative database (42). Of the 25,686 patients, 6,518 patients were evaluated by a registered dietitian. The study cohort includes 6,518 ICU patients who were evaluated by a registered dietitian. Exposure of Interest and Comorbidities The exposure of interest was BMI determined at the time of dietitian consultation from the height and the hospital admission weight or estimated dry weight. BMI was categorized a priori as less than 18.5 kg/m2 (underweight), 18.5–24.9 kg/m2 (normal/referent), 25–29.9 kg/m2 (overweight), 30–39.9 kg/m2 (obesity class I and II), and more than or equal to 40.0 kg/m2 (obesity class III). Class I and class II obesity groups were combined for improved model fit to the observed data. Height was determined first by records from the Partners HealthCare electronic medical record and then in order by Dana-Farber Cancer Institute records, surgery records, outside hospital transfer records, or patient report. Estimated dry weight is determined January 2015 • Volume 43 • Number 1

Clinical Investigations

by the weight prior to fluid resuscitation (earliest hospitalization weight and most recent outpatient clinic weight) or the lowest postdiuresis weight. Per Joint Commission mandate and hospital policy, all medical and surgical patients are screened for nutritional risk by the registered nurse within 24 hours of ICU admission (43). Additionally, registered dietitians screened all ICU patients, and those patients deemed at risk for malnutrition are formally evaluated by a registered dietitian in person using a structured objective assessment. Malnutrition diagnoses were determined by a registered dietitian based on prior studies (44, 45) using clinical judgment and a combination of available factors (Supplemental Table 1, Supplemental Digital Content 1, http://links.lww.com/CCM/B56). To meet criteria for nonspecific malnutrition, the patient has known malnutrition risk factors (inadequate nutrient intake of kcal, protein, and micronutrients) with metabolic stress and/or overt signs of malnutrition (wasting of muscle and fat stores) without supporting anthropometric or biochemical data present. To meet criteria for protein-energy malnutrition (PEM), patients must have a combination of disease-related weight loss, decrease in ideal body weight (46), overt muscle wasting, peripheral edema, inadequate kcal or protein intake, decrease in albumin, transferring, or total lymphocyte count. A patient need not be below an albumin threshold to be diagnosed if other PEM criteria are present. Malnutrition was considered to be present if the patient was diagnosed by a registered dietitian (RD) with nonspecific malnutrition, mild PEM, moderate PEM, severe PEM, marasmus, or kwashiorkor. Malnutrition was considered to be absent if patients were diagnosed as well nourished or at risk for malnutrition. Nutrition diagnoses were categorized a priori into nonspecific malnutrition, any PEM, or malnutrition absent. Patient energy needs are determined based on BMI by literature precedent (47–50) as outlined in supplemental data (Supplemental Digital Content 2, http://links.lww.com/CCM/B57). The Deyo-Charlson index was used to assess the burden of chronic illness (51) using International Classification of Diseases, 9th Edition (ICD-9), coding algorithms which are well studied and validated (52, 53). Patient type was defined as medical or surgical and incorporated the Diagnostic-Related Grouping methodology (54). Sepsis was defined by the presence of any of the following ICD-9-CM codes: 038.0–038.9, 020.0, 790.7, 117.9, 112.5, or 112.81, 3 days prior to critical care initiation to 7 days after critical care initiation (55). Acute kidney injury was defined as Risk, Injury, Failure, Loss, and End-stage Kidney (RIFLE) class Injury or Failure (56) occurring between 3 days prior to critical care initiation and 7 days after critical care initiation. We applied the serum creatinine criteria only to determine the maximum RIFLE class (57). We classified patients according to the maximum RIFLE class (class Risk, class Injury, or class Failure) (58) defined as a fold change in serum creatinine from preadmission serum creatinine (57). RIFLE class was determined in patients with preadmission baseline creatinine available from 7 to 365 days prior to hospital admission with the creatinine closet to hospital Critical Care Medicine

admission recorded. Mechanical ventilation was defined by the presence of ICD-9-CM code 96.7× during hospitalization (59). Noncardiogenic acute respiratory failure was identified by the presence of ICD-9 codes for respiratory failure or pulmonary edema (518.4, 518.5, 518.81, and 518.82) and mechanical ventilation (96.7×), excluding congestive heart failure (428.0– 428.9) (60) following hospital admission. Malignant neoplasm history is defined by the presence of any of the following ICD9-CM codes prior to hospital discharge date: 140–209 (61). Inotropes or vasopressors were considered to be present if prescribed 3 days prior to critical care initiation to 7 days after critical care initiation (62). Using electronic pharmacy records, exposure to inotropes and vasopressors was determined for dopamine, dobutamine, epinephrine, norepinephrine, phenylephrine, milrinone, and vasopressin. In a random subset, Acute Physiologic and Chronic Health Evaluation (APACHE II) scores were determined at critical care initiation (63). Records of the administration of parenteral nutrition (PN) prior to critical care initiation was determined by CPT code 99.15 and confirmed by pharmacy records. Acute organ failure was adapted from Martin et al (55) and defined by a combination of ICD-9-CM and CPT codes relating to acute organ dysfunction assigned from 3 days prior to critical care initiation to 30 days after critical care initiation (42, 55). Acute failure is the summation of the number of acute organ failure categories (respiratory failure, cardiovascular failure, renal, hepatic, hematologic, metabolic, and/or neurologic) (55) present by ICD-9-CM code assignment. We have shown that the acute organ failure variable is strongly associated with mortality following critical care (42, 62). Assessment of Mortality Information on vital status for the study cohort was obtained from the Social Security Administration Death Master File, which has high sensitivity and specificity for mortality (64–67). We have validated the accuracy of the Social Security Administration Death Master File for in-hospital and out-of-hospital mortality in our administrative database (42). The censoring date was March 15, 2012, and 100% of the cohort had at least 90-day follow-up. Endpoints The primary endpoint was 30-day all-cause mortality following critical care initiation. The prespecified secondary endpoint was 90-day all-cause mortality. Statistical Analysis Categorical covariates were described by frequency distribution and compared across malnutrition groups using contingency tables and chi-square testing. Continuous covariates were examined graphically (e.g., histogram and box plot) and in terms of summary statistics (mean, sd, median, and interquartile range) and compared across exposure groups using one-way analysis of variance. The outcome considered was mortality. www.ccmjournal.org

89

Robinson et al

Unadjusted associations between BMI groups and outcomes were estimated by bivariable logistic regression models. Adjusted odds ratios (ORs) were estimated by multivariable logistic regression models with inclusion of covariate terms thought to plausibly associate with both BMI and mortality. For the primary model (30-day mortality), specification of each continuous covariate (as a linear vs categorical term) was adjudicated by the empiric association with the primary outcome using Akaike Information Criterion; overall, model fit was assessed using the Hosmer-Lemeshow test. To reduce potential bias from the nonrandomized assignment of the diagnosis of obesity, we constructed propensity scores for the allocation of obesity and used these in the primary and secondary analyses (68–70). Utilizing logistic regression, propensity scores were calculated for each cohort subject to estimate the probability for the presence or absence of a diagnosis of obesity (BMI ≥ 30.0 kg/m2). Covariates selected for the propensity score development included age, sex, race, patient type, Deyo-Charlson index, and nutrition status. Two smaller cohorts were obtained where a subject (with obesity) was matched to a control subject (without obesity) on the basis of the propensity score. We utilized Mahalanobis metric matching within calipers defined by the propensity score to match the smaller cohorts (69) by using the STATA matching algorithm “psmatch2” (StataCorp, College Station, TX) (71). Utilizing the publically available “sgmediation” STATA program, we used nonparametric bootstrapping analyses (72, 73) to test the meditational model of hs-C-reactive protein (hs-CRP) (as a linear term) at critical care initiation as a mediator of the relationship between BMI and 30-day mortality. For the time to mortality, we estimated the survival curves according to group with the use of the Kaplan-Meier method and compared the results by means of the log-rank test. All p values presented are two-tailed; values below 0.05 were considered significant. All analyses are performed using STATA 12.0MP statistical software.

Patient Characteristics of the Study Population

Table 1.

6,518

n Gender, n (%)  Female

2,837 (43.5)

 Male

3,681 (56.5)

Race, n (%)  Nonwhite

1,350 (20.7)

 White

5,168 (79.3)

Age, mean (sd)

63.8 (16.5)

Patient type, n (%)  Medical

3,138 (48.1)

 Surgical

3,380 (51.9)

 Trauma

608 (9.3)

 Burn

71 (1.1)

 Coronary artery bypass

299 (4.6)

Body mass index, n (%)  < 18.5  kg/m2

314 (4.8)

 18.5–24.9  kg/m

2

2,369 (36.4)

 25.0–29.9  kg/m

2

2,036 (31.2)

 30.0–39.9  kg/m  ≥ 40.0 kg/m

1,471 (22.6)

2

328 (5.0)

2

Body mass index, mean (sd)

27.4 (7.0)

Malnutrition absent, n (%)  Not at risk for developing malnutrition  At risk for developing malnutrition

47 (0.7) 2,076 (31.9)

Nonspecific malnutrition, n (%)  Nonspecific protein-calorie malnutrition

3,641 (55.9)

Specific malnutrition, n (%)

RESULTS Table 1 shows demographic characteristics of the study population. The majority of patients were men (56.5%), white (79.3%), and 48.1% had medically related diagnosis-related groups. The mean age at hospital admission was 63.8 years (sd 16.5). In the cohort, 5% were underweight, 36% were normal weight, 31% were overweight, 23% had class I/II obesity, and 5% had class III obesity. Nonspecific malnutrition was present in 56%, PEM was present in 12%, and 33% were well nourished. The 30- and 90-day mortality rates were 19.1% and 26.6%, respectively, and 20.5% patients had sepsis per ICD9-CM coding. The mean hospital length of stay for the cohort was 18.6 days (15.5 d). The mean time between hospital admission date and ICU admission was 2.6 days (sd 5.4). For the 76% of the cohort, the hospital admission date and ICU admission date was within 72 hours, and 11.2% of the cohort had hospital stays of more than 7 days prior to ICU admission. Patient characteristics of the study cohort were stratified according to BMI (Table 2). Factors that significantly differed 90

www.ccmjournal.org

 Mild protein-calorie malnutrition

324 (5.0)

 Moderate protein-calorie malnutrition

152 (2.3)

 Severe protein-calorie malnutrition  Marasmus  Kwashiorkor

33 (0.5) 244 (3.7) 1 (0.0)

Mortality rates, %  30 d

19.1

 90 d

26.6

 365 d

37.5

Sepsis, n (%)

1,333 (20.5)

between stratified groups included age, gender, patient type, inotropes or vasopressor exposure, malignant neoplasm, nutrition status, acute organ failure, acute kidney injury, mechanical ventilation, sepsis, albumin, prealbumin, hs-CRP, and length of stay (Table 2). In the multivariable adjusted analysis, age, January 2015 • Volume 43 • Number 1

Clinical Investigations

patient type, Deyo-Charlson index, acute organ failure, inotropes or vasopressors, and sepsis were all significantly associated with 30-day mortality (Table 3). Primary Outcome Patients with class III obesity had a decreased odds of 30-day crude mortality: underweight OR 30-day mortality is 1.06 (95% CI, 0.79–1.42), overweight 30-day mortality OR is 0.96 (95% CI, 0.83–1.11), class I/II obesity 30-day mortality OR is 0.85 (95% CI, 0.72–1.01), and class III obesity 30-day mortality OR is 0.72 (95% CI, 0.52–0.99), all relative to patients with BMI 18.5–24.9 kg/m2 (Fig. 1 and Table 4). Obesity in the cohort is a significant predictor of the odds of 30-day mortality following adjustment for age, gender, race, medical versus surgical patient type, Deyo-Charlson index, acute organ failure, vasopressor use, and sepsis: underweight OR 30-day mortality is 1.09 (95% CI, 0.80–1.48), overweight 30-day mortality OR is 0.93 (95% CI, 0.80–1.09), class I/II obesity 30-day mortality OR is 0.80 (95% CI, 0.67–0.96), and class III obesity 30-day mortality OR is 0.69 (95% CI, 0.49–0.97), all relative to patients with BMI 18.5–24.9 kg/m2 (Table 4, model 1). Obesity in the cohort is similarly predictive of odds of 90-day mortality prior to and following multivariable adjustment (Table 4). Furthermore, individually running the adjusted model (model 1) with and without terms for metastatic disease or acute kidney injury, the 30-day mortality estimates in each case are similar indicating that the BMI-mortality relationship is not materially confounded by metastatic disease or acute kidney injury (data not shown). Confounding and Effect Modification Patients with any malnutrition had a significantly a higher Deyo-Charlson index relative to patients without malnutrition (χ2, p < 0.001). In addition, patients with any malnutrition had significantly more episodes of sepsis (χ2, p < 0.001), more use of inotropes or vasopressors (χ2, p < 0.001), and more acute kidney injury (χ2, p < 0.001) all relative to patients without malnutrition. Importantly, there is confounding of the obesity-mortality association on the basis of malnutrition. Adjustment for only nutrition status attenuates the BMI-mortality association: underweight OR 30-day mortality is 0.74 (95% CI, 0.54–1.00); overweight 30-day mortality OR is 1.05 (95% CI, 0.90–1.23); class I/II obesity 30-day mortality OR is 0.96 (95% CI, 0.81–1.15); and class III obesity 30-day mortality OR is 0.81 (95% CI, 0.59–1.12), all relative to patients with normal BMI (data not shown). Furthermore, when nutrition status is added to the fully adjusted model, the BMI-30-day mortality association is rendered nonsignificant (Table 4, model 3). Addition of terms for mechanical ventilation, trauma, and time between hospital and ICU admission did not materially alter the significance of the observed BMI-mortality relationship with or without additional adjustment for nutrition status (Supplemental Table 1, Supplemental Digital Content 1, http:// links.lww.com/CCM/B56). In addition to confounding, there is significant effect modification of the obesity-mortality relationship according Critical Care Medicine

to malnutrition status (p-interaction: 30-day mortality, p < 0.0001). The result of the Hosmer-Lemeshow test for goodness-of-fit of each model was not significant (p = 0.90 for model 1, 0.28 for model 2, and 0.09 for model 3), indicating good fit of the models in the cohort concerning the similarity of expected and observed cases of 30-day mortality. Subanalyses We next utilized propensity score matching on baseline characteristics and nutrition status to reduce residual confounding of the BMI category assignment. We assessed the odds of death in a smaller cohort of propensity score matched patients (n = 3,554). Again, the propensity score matched on the baseline characteristics and nutrition status (BMI ≥ 30.0 kg/ m2, n = 1,777; BMI < 30.0 kg/m2, n = 1,777). In the matched cohort, crude all-cause mortality rates were 17.3% (95% CI, 15.5–19.0; 717 deaths) in patients with obesity and 17.6% (95% CI, 15.8–19.3; 776 deaths) in patients without obesity. In the matched cohort, the odds of 30- and 90-day mortality lost statistical significance, likely from matching on nutrition status (Table 4). The Kaplan-Meier plot demonstrating survival grouped according to obesity in the propensity score matched cohort had no significant difference between the two curves (p = 0.063, log-rank test, plot not shown). To evaluate the contribution from energy delivery, we studied data from a subset of the cohort who had at least 1 electronically documented RD follow-up after initial RD consultation (n = 2,572). In this subset, 5.9% of the patients received PN in the first 48 hours of an ICU admission, whereas 11.6% of patients received PN during the 8 days following an ICU admission and 4.2% began PN 8 days following an ICU admission. We next addressed the hypothesis that underfeeding is driving the nonspecific malnutrition mortality association. The nonspecific malnutrition group does not appear to be underfed relative to the malnutrition absent group (Supplemental Table 2, Supplemental Digital Content 1, http://links.lww.com/CCM/ B56). Comparing the nonspecific malnutrition group with the malnutrition absent group, the proportion of total kcal delivered by the enteral route is significantly less than the malnutrition absent group (81.5–72.9%), and there is little clinically relevant difference in the percent of total kcal needs met (30.1–36.0%). Exposure to any PN by day 2 involves a small proportion of the nonspecific malnutrition group patients that increases by day 8. Further, the percent kcal needs met were similar across the BMI categories (Supplemental Table 3, Supplemental Digital Content 1, http://links.lww.com/CCM/B56). PN exposure was highest in the BMI less than 18.5 kg/m2 group but not statistically different across BMI groups. From the data, it does not appear that the kcal delivery is different in patients with BMI more than 29.99 kg/m2 (Supplemental Table 3, Supplemental Digital Content 1, http://links.lww.com/CCM/B56). To reduce potential bias from the nonrandomized use of early PN, we constructed propensity scores for the allocation of early PN (68–70). Utilizing logistic regression, propensity scores were calculated for a cohort subject subset to estimate the probability for the presence or absence of www.ccmjournal.org

91

Robinson et al

Table 2.

Stratified Patient Characteristics of the Cohort Body Mass Index, kg/m2 ≤ 18.5

Variable

314

n Age, mean (sd)

18.5–24.9

2,369

63.3 (19.3)

64.5 (15.6)

 Female

191 (60.8)

1,068 (45.1)

 Male

123 (39.2)

1,301 (54.9)

245 (78.0)

1,877 (79.2)

69 (22.0)

492 (20.8)

 Medical

181 (57.64)

1,165 (49.18)

 Surgical

133 (42.36)

1,204 (50.82)

Gender, n (%)

Race, n (%)  White  Nonwhite Patient type, n (%)

Deyo-Charlson index, n (%)  0–3

247 (78.7)

1,941(81.93)

 4–6

43 (13.7)

321 (13.55)

 ≥ 7

24 (7.6)

107 (4.5)

Inotropes/ vasopressors, n (%)

137 (43.6)

1,126 (47.5)

Malignant neoplasm, n (%)

169 (53.8)

1,049 (44.3)

Nutrition status, n (%)  Malnutrition absent

26 (8.3)

 Nonspecific malnutrition

678 (28.6)

87 (27.7)

1,316 (55.6)

201 (64.0)

375 (15.8)

Body mass index, mean (sd)

17.0 (1.2)

22.3 (1.4)

Acute Physiologic and Chronic Health Evaluation II, mean (sd)c

23.0 (8.3)

22.7 (8.3)

 Protein-energy malnutrition

Acute organ failure, n (%)  0

66 (21.0)

 1

88 (28.0)

733 (30.9)

 2

81 (25.8)

682 (28.8)

 3

53 (16.9)

377 (15.9)

 ≥ 4

26 (8.28)

258 (10.9)

Sepsis, n (%)

61 (19.4)

445 (18.8)

10 (6.0)

119 (9.5)

64 (20.4)

561 (23.7)

106 (33.8)

957 (40.4)

Acute kidney injury, n (%)

d

Noncardiogenic acute respiratory failure, n (%) Mechanical ventilation, n (%)

319 (13.5)

8 (2.6)

51 (2.2)

Albumin, g/dL, mean (sd)

3.3 (0.7)

3.5 (0.7)

Prealbumin, mg/dL, mean (sd)

12.6 (6.2)

13.0 (6.2)

hs-C-reactive protein, mg/L, mean (sd)

94.5 (83.4)

101.6 (90.4)

Length of stay, mean (sd)

14.5 (10.8)

18.4 (16.3)

Parenteral nutrition prior to ICU, n (%)

e

p values determined by chi-square test. b p values determined by Kruskal-Wallis test. c Acute Physiologic and Chronic Health Evaluation II available in 146 cohort patients. d Acute kidney injury determined in 3,627 cohort patients. e Parenteral nutrition given prior to admission to the ICU. Albumin available in 6,199 cohort patients, prealbumin available in 3,341 cohort patients, and hs-C-reactive protein available in 3,689 cohort patients. a

92

www.ccmjournal.org

January 2015 • Volume 43 • Number 1

Clinical Investigations

Body Mass Index, kg/m2 25.0–29.9

30.0–39.9

p

≥ 40.0

2,036

1,471

328

65.0 (15.6)

62.6 (14.7)

57.5 (13.9)

< 0.0001a < 0.0001

735 (36.1)

650 (44.2)

193 (58.8)

1,301 (63.9)

821 (55.8)

135 (41.2)

1,629 (80.0)

1,172 (79.7)

245 (74.7)

0.26 407 (20.0)

299 (20.33)

83 (25.3) 0.002

954 (46.86)

675 (45.89)

163 (49.7)

1,082 (53.14)

796 (54.11)

165 (50.3) 0.33

1,645 (80.8)

1,166 (79.3)

266 (81.1)

289 (14.2)

220 (15.0)

45 (13.7)

102 (5.0)

85 (5.8)

17 (5.2)

1,009 (49.6)

771 (52.4)

180 (54.9)

0.002a

916 (45.0)

607 (41.3)

102 (31.1)

< 0.0001a < 0.0001

713 (35.0)

575 (39.1)

131 (39.9)

1,204 (59.1)

849 (57.7)

185 (56.4)

119 (5.8)

47 (3.2)

12 (3.7)

27.3 (1.4)

33.5 (2.7)

46.9 (7.8)

25.4 (8.3)

25.3 (6.3)

260 (12.8)

163 (11.1)

27 (8.2)

600 (29.5)

416 (28.3)

101 (30.8)

579 (28.4)

392 (26.7)

76 (23.2)

317 (15.6)

260 (17.7)

66 (20.1)

280 (13.8)

240 (16.3)

58 (17.7)

382 (18.8)

341 (23.2)

104 (31.7)

< 0.0001a

134 (11.8)

112 (13.1)

32 (15.3)

0.004a

480 (23.6)

355 (24.1)

90 (27.4)

0.33a

888 (43.6)

741 (50.4)

187 (57.0)

< 0.0001a

42 (2.1)

34 (2.3)

4 (1.2)

0.76a

3.5 (0.7)

3.4 (0.7)

3.3 (0.7)

< 0.0001b

13.3 (6.5)

13.5 (6.9)

12.9 (7.4)

< 0.0001b

110.6 (107.1)

111.2 (96.8)

< 0.0001b

23.0 (4.1)

0.37b < 0.0001

18.8 (15.3)

Critical Care Medicine

19.0 (14.6)

131.3 (110.0)

< 0.0001b

21.4 (17.6)

< 0.0001b

www.ccmjournal.org

93

Robinson et al

Multivariable Adjusted Associations Between Covariates and 30-Day Mortality Table 3.

Variable

OR

95% CI

p

0.69

Body mass index 1.06

0.79–1.42

2

1.00

Referent

2

0.96

0.83–1.11

0.59

 30.0–39.9  kg/m

0.85

0.72–1.01

0.061

 ≥ 40.0 kg/m

0.72

0.52–0.99

0.040

1.03

1.02–1.03

< 0.001

 Male

0.95

0.83–1.08

0.41

 Female

1.00

Referent

 White

1.00

Referent

 Nonwhite

0.95

0.80–1.11

 Medical

1.00

Referent

 Surgical

0.51

0.44–0.58

 0–3

1.00

Referent

 4–6

1.33

1.12–1.59

0.001

 ≥ 7

1.77

1.37–2.30

< 0.001

 0

1.00

Referent

 1

1.88

1.42–2.50

< 0.001

 2

2.71

2.05–3.59

< 0.001

 3

2.76

2.05–3.72

< 0.001

 ≥ 4

5.90

4.37–7.98

< 0.001

Inotropes/ vasopressors

1.21

1.05–1.39

0.009

Sepsis

1.44

1.24–1.68

< 0.001

 < 18.5  kg/m2  18.5–24.9  kg/m  25.0–29.9  kg/m

2

2

Age (per 1 yr) Gender

Race 0.51

Patient type < 0.001

Deyo-Charlson index

Acute organ failure

OR = odds ratio. Estimates for each variable are adjusted for all other variables in the table.

early PN (PN exposure during the 8 days after ICU admission). Covariates selected for the propensity score development included age, sex, race, patient type, Deyo-Charlson index, sepsis, acute organ failure, inotrope/pressor use, and nutrition status. In this subset of propensity score matched patients (n = 588), the early PN-mortality association was not statistically significant (OR for 30-day mortality, 1.16; 95% CI, 0.75–1.79; p = 0.51). Further model building using BMI in place of nutrition status in the propensity 94

www.ccmjournal.org

score development did not alter the absence of an early PN-mortality association. As albumin is a negative phase reactant (62), we explored whether increased inflammation (elevated hs-CRP) may associate with lower albumin, malnutrition diagnosis assignment, and the outcomes observed. A total of 6,199 cohort patients had albumin measured at or very near the time of hospital admission. In cohort patients with albumin measured, we note that there are small but statistically different albumin levels between the BMI groups (Table 2). Mean (sd) serum albumin declined with worsening nutrition status (well nourished albumin = 3.69 g/dL [0.60 g/dL], nonspecific malnutrition albumin = 3.37 g/dL (0.71 g/dL), and PEM albumin = 3.13 g/dL (0.63 g/ dL); p < 0.001). In contrast to adjustment for nutrition status, additional adjustment for albumin does not materially alter the point estimates of the BMI-mortality association (Table 4, model 2 vs model 3). In the cohort, 3,639 patients had albumin and hs-CRP measured. In these 3,639 patients, the correlation between albumin and hs-CRP was significant but small (r = –0.13, p < 0.05). hs-CRP is statistically different between the BMI groups with higher hs-CRP in those with higher BMI (Table 2). When hsCRP is analyzed as the exposure of interest, we find that there is a 1.3-fold increase in the odds of 30-day mortality when hs-CRP is more than or equal to 100 mg/L compared with hs-CRP less than 100 mg/L (OR, 1.34; 95% CI, 1.10–1.64; p =0.003 adjusted for age, gender, race, Deyo-Charlson index, and patient type). To test the proportion of the association between BMI and 30-day mortality that was mediated by inflammation (hs-CRP), we used a nonparametric bootstrapping analyses approach (72, 73). Results based on 5,000 bootstrapped samples indicated that although the total effect of BMI and 30-day mortality was significant (total effect = 0.98, se = 0.2, p < 0.001), the direct effect was not (direct effect = –0.0008, se = 0.0009, p = 0.4). hs-CRP mediates the relationship between BMI and 30-day mortality (indirect effect 95% CI, 0.00007– 0.0004). Mediation of BMI and 30-day mortality by hs-CRP is significant (p < 0.05) in this analysis, as the 95% CIs for the indirect effect do not include zero (72, 73). In a subset of patients with BMI more than or equal to 30.0 kg/m2 (n = 1,799), mortality rates differ for those with and without malnutrition (Fig. 2). In patients with BMI more than or equal to 30.0 kg/m2, those without malnutrition have a 30- and 90-day mortality of 13.2% and 18.9%, respectively, whereas in those with either nonspecific malnutrition or PEM, the 30- and 90-day mortality is 19.8% and 30.4%, respectively. Cohort patients with BMI more than or equal to 30.0 kg/m2 and either nonspecific malnutrition or PEM have increased mortality relative to obese patients without malnutrition: OR of 30-day mortality is 1.58 (95% CI, 1.21–2.07; p = 0.001) that of patients without malnutrition adjusted for age, gender, race, Deyo-Charlson index, and patient type (Table 5). Additional adjustment by hs-CRP does not alter the nutrition-mortality association, and there is no significant effect modification of the nutrition-mortality relationship according to hs-CRP levels January 2015 • Volume 43 • Number 1

Clinical Investigations

has been previously shown to increase mortality in a number of chronic illnesses, including chronic obstructive pulmonary disease (79), congestive heart failure (80), chronic kidney disease (81), end-stage renal disease (82), and cirrhosis (83). Malnutrition has been reported to be associated with longer ICU stays and higher rates of complication (36). Our study suggests that one must account for the nutritional status of the obese patient before drawing any conclusions about obesity on ICU outcome. There are little data on malnourished critically ill obese individuals or the relationFigure 1. Time-to-event curves for mortality. Unadjusted event rates were calculated with the use of the Kaplan-Meier methods and compared with the use of the log-rank test. Categorization of body mass index ship between malnutrition and (BMI) is per the primary analyses. The global comparison log-rank p value is < 0.0001. medium- to long-term mortality outcomes in critical care. Assessment of nutritional status in the ICU can be challenging, (p-interaction: 30-day mortality, p = 0.31). In obese patients which may explain the paucity of data on malnourished obese with hs-CRP and albumin measured (n = 1,011), the correpatients in the ICU. Previous studies have shown that routine lation between hs-CRP and albumin is small but significant (r = –0.17; p < 0.001). Obese patients with hs-CRP more than nutrition assessment variables, such as skin-fold thickness and measures of immune function, may be unreliable in the setting of or equal to 100 mg/L do not have significantly higher odds of critical illness (84) and in obese individuals (85). For that reason, mortality (Table 5). Finally, obese patients with albumin less than 3.4 g/dL have higher mortality relative to those with albu- BMI is often and improperly used as a surrogate for malnutrition in the ICU (86). However, BMI is an extremely poor surrogate for min more than or equal to 3.4 g/dL (Table 5). determining nutritional status and many with a BMI more than or equal to 30.0 kg/m2 can in fact be malnourished (87). DISCUSSION A second major finding of this study is that obese critically ill In this study, we determined whether the association between patients who are malnourished have a marked increase in morBMI and all-cause mortality following critical care initiation tality compared with those without malnourishment. This indiwas modified by nutrition status. There are several important cates that nutrition status may be a unique risk factor and may be novel findings from this study. First, our study suggests that independent from BMI. In addition, our results suggest that obese malnutrition affects the association between obesity and morindividuals are not protected from the adverse effects of malnourtality. When malnutrition is not included in the statistical analishment. Our data contradict the presumption that obese indiyses, obesity appears to be protective for critically ill patients in viduals are in less need of nutrition evaluation and intervention. the medical and surgical ICUs. A protective effect of obesity, The present study may have limitations. The definition of previously termed “the obesity paradox,” has been reported malnutrition utilized in this study was put into practice in with multiple conditions, including HIV/AIDS (75), end- 2000 and is based on older literature (44, 45). These protocols stage renal disease (76), and acute heart failure (77). A cohort have been utilized on a daily basis in over 30,000 inpatients study of 16,812 ICU patients demonstrated that obese patients and are currently the basis for nutrition treatment recommenhad 26% lower mortality at 30 days and 43% lower mortal- dations in the hospital under study. Per the American Society ity at 1 year following adjustment for potential confounding for Parenteral and Enteral Nutrition and the Academy of by age, gender, race, diabetes, coronary artery disease, stroke, Nutrition and Dietetics, in adults there is no approach to diaghypertension, obesity-related cancers, and kidney disease (78). nosis of malnutrition that is universally accepted (88). Others However, the analyses in this latter study were not adjusted for have used nutrition screening that involves variables such as nutrition status (78). appetite, unintentional weight loss, anthropometric variables, In contrast, our study suggests that there is neither a and laboratory values (89–95). Malnutrition in the obese may decreased nor increased risk of mortality in critically ill obese be better detected with pre-ICU weight loss, sarcopenia (96), patients compared with nonobese patients when one controls or small musculature (97) seen on ICU admission CT scan all for nutrition status. This is not surprising as malnutrition of which we are unable to asses in the cohort. Critical Care Medicine

www.ccmjournal.org

95

Robinson et al

Table 4.

Unadjusted and Adjusted Associations Between Malnutrition and Mortality Mortality OR (95% CI)a BMI < 18.5 kg/m2

Variable

OR (95% CI)

25–29.9 kg/m2 p

OR (95% CI)

≥ 40.0 kg/m2

30–39.9 kg/m2 p

OR (95% CI)

p

OR (95% CI)

p

30-day mortality Crude

1.06 (0.79–1.42)

0.69

0.96 (0.83–1.11)

0.59

0.85 (0.72–1.01)

0.061

0.72 (0.52–0.99)

0.040

Adjustedb

1.09 (0.80–1.48)

0.60

0.93 (0.80–1.09)

0.38

0.80 (0.67–0.96)

0.016

0.69 (0.49–0.97)

0.032

Adjusted + albuminc

1.01 (0.74–1.39)

0.94

0.95 (0.81–1.11)

0.51

0.80 (0.67–0.96)

0.016

0.67 (0.47–0.94)

0.021

Adjusted + nutritiond

0.78 (0.56–1.08)

0.13

1.01 (0.86–1.19)

0.89

0.90 (0.75–1.07)

0.24

0.77 (0.55–1.08)

0.14

PS matched cohorte

0.85 (0.45–1.60)

0.61f

1.07 (0.83–1.37)

0.60g

1.04 (0.83–1.29)

0.74h

0.86 (0.61–1.22)

0.40i

Crude

1.17 (0.91–1.51)

0.22

0.94 (0.83–1.08)

0.38

0.80 (0.69–0.93)

0.004

0.63 (0.47–0.84)

0.001

Adjustedb

1.22 (0.93–1.60)

0.16

0.91 (0.79–1.05)

0.19

0.75 < 0.001 (0.64–0.88)

0.61 (0.45–0.83)

0.002

Adjusted + albuminc

1.10 (0.83–1.45)

0.52

0.92 (0.79–1.06)

0.24

0.73 < 0.001 (0.62–0.87)

0.57 < 0.001 (0.41–0.78)

Adjusted + nutritiond

0.81 (0.60–1.08)

0.15

1.01 (0.87–1.16)

0.92

0.86 (0.73–1.02)

0.079

0.70 (0.51–0.96)

0.025

PS matched cohorte

0.93 (0.54–1.61)

0.80f

1.10 (0.88–1.38)

0.39g

1.03 (0.85–1.25)

0.79h

0.79 (0.58–1.09)

0.15i

90-day mortality

OR = odds ratios, BMI = body mass index, PS = propensity score. a Referent in each case is BMI 18.5–24.9 kg/m2. b Model 1: estimates adjusted for age, gender, race (white and nonwhite), Deyo-Charlson index, type (surgical vs medical), acute organ failure, vasopressor/ inotrope use, and sepsis. c Model 2: estimates adjusted for age, gender, race (white and nonwhite), Deyo-Charlson index, type (surgical vs medical), acute organ failure, vasopressor/ inotrope use, sepsis, and serum albumin. d Model 3: estimates adjusted for all covariates in model 1 and nutrition status (malnutrition absent, nonspecific malnutrition, or protein-energy malnutrition). e n = 3,554, with 1,777 BMI ≥ 30 kg/m2 and 1,777 with BMI < 30 kg/m2. f n =80. g n =791. h n = 1,454. i n =323.

We observed a higher Deyo-Charlson index in the malnourished patients in our cohort which may have contributed to the attenuation of the obesity-mortality association observed in our outcome data. Despite adjustment for Deyo-Charlson index, unmeasured confounders related to a higher chronic disease burden in malnourished patients may have contributed to our observed associations. APACHE II scores were not uniformly available for our study due to lack of physiological data in the administrative dataset. Hence, severity of illness was estimated using an acute organ failure score that strongly correlates with mortality (Table 3). Insufficient adjustment for severity of illness may have contributed to our observed results despite adjustment for age and sepsis, utilizing acute organ score as an adjustor for severity of illness. 96

www.ccmjournal.org

The study was conducted at a single academic medical center, which may not be reflective of practices at all hospitals. Other variables that were not measured may have influenced mortality independently of BMI, which may have led to bias in our estimates. Despite multivariable adjustment for potential confounders, there may be residual confounding that contributes to the differences in outcomes in our study. BMI may simply be a reflection of the condition of the patient, for which we are unable to fully adjust. Further, in the various ICUs under study while malnutrition diagnosis protocols are uniform, nutrition delivery is not standardized. In our study, sepsis, comorbidity, and illness severity are defined based on administrative ICD-9-CM coding. Algorithms January 2015 • Volume 43 • Number 1

Clinical Investigations

may be associated with increased complications (86). It has also been observed that increased nutrition, with greater intakes of energy and protein, is associated with better clinical outcomes of critically ill, underweight patients (BMI < 25 kg/m2) and a subset of middle-range obese patients (98). Late initiation of PN in the critically ill is shown to be associated with a less complicated ICU course (99). Our study is an observational study, and thus, causality regarding malnutrition, obesity, and mortality is limited. Furthermore, our study does not 2 address the utility of nutritional Figure 2. Time-to-event curves for mortality in patients with body mass index ≥ 30.0 kg/m (n =1,799). Unadjusted event rates were calculated with the use of the Kaplan-Meier methods and compared with the use supplementation in improving of the log-rank test. Categorization of nutrition status is per the secondary analyses. The global comparison log- patient outcomes. With regard to rank p value is < 0.0001. the definition of malnutrition in general, it is not clear if outcomes are related to an epiphenomenon of disease or reflect alterations developed to recode ICD-9-CM-coded data into a Deyoin nutrient delivery (100, 101). Although our analysis indicates Charlson index are well studied, validated, and well suited for that a diagnosis of malnutrition attenuates the survival advantage use in administrative datasets (51, 52). Although we based our of obesity, our diagnosis of malnutrition may be confounded by determinants of sepsis on ICD-9-CM codes utilized and valiseverity-of-illness markers that may not be related to nourishdated by others (53), adjusting for covariates determined by ment and factors that may be more closely linked to nourishment ICD-9-CM coding is a limitation. ICD-10-CM coding informabut may not be sensitive to response to nourishment. tion is not available in the administrative dataset under study. The present study has several strengths. A nutritional Nutritional supplementation in the ICU has been shown to professional (i.e., a registered dietitian) made an in person affect outcomes. A negative energy balance while in the ICU assessment of nutritional status of all cohort patients. Plasma albumin was utilized as part of the determination of nutrition Table 5. Adjusted Associations Among status. Although plasma albumin levels may be affected by Albumin, hs-CRP, Malnutrition, and 30-Day nonnutritional conditions, albumin has been used as a surroMortality in Cohort Patients With Body gate of nutritional status and been shown to predict long-term Mass Index ≥ 30.0 kg/m2 outcomes in the critically ill patients with chronic obstructive pulmonary disease (102), older patients requiring mechanical Variable OR (95% CI) p ventilation (103), and in a diverse population of critically ill Nutrition statusa patients (25). Judicious evaluation of albumin along with the other assessments made by the registered dietitian in our study  Malnutrition 1.58 (1.21–2.07) 0.001 was critical for the standardization of the assessment of nutri Well nourished 1.00 (referent) tion status. The database utilized to ascertain nutrition status hs-CRPb in this study was designed for clinical documentation of nutrition status in addition to research and quality purposes. 1.25 (0.90–1.75) 0.10  ≥ 100 mg/L Another strength of the study was that the determination of all < 100  mg/L 1.00 (referent) cause mortality was based on the Social Security Administration Albuminc Death Master File. This allows for very accurate long-term follow-up of patients following discharge and has been validated in  < 3.4  g/dL 2.67 (2.06–3.44) < 0.001 our administrative database (42). Finally, our study has a suffi1.00 (referent)  ≥ 3.4 g/dL cient cohort and effect size that increases the reliability of our hs-CRP = hs-C-reactive protein. mortality estimates (n = 6,518, 30-day mortality rate = 19.1%). a

n = 1,799, malnutrition is defined as nonspecific malnutrition or proteinenergy malnutrition. b n = 1,011 patients with hs-CRP drawn. c n = 1,710 patients with albumin drawn. All estimates are adjusted for age, gender, race (white and nonwhite), DeyoCharlson index, and type (surgical vs medical).

Critical Care Medicine

CONCLUSIONS In the critically ill, the association between improved survival and high BMI observed in this study appears to be confounded www.ccmjournal.org

97

Robinson et al

by the presence of malnutrition. Once adjustment is made for nutrition status, BMI more than or equal to 30.0 kg/m2 does not appear to either increase or decrease mortality in the critically ill. The disparate conclusions of other studies regarding the obesity-mortality relationship in the critically ill may due to lack of control for nutritional status. Finally, malnutrition near the time of ICU admission appears to be an important driving force for outcome in the obese critically ill patients where obese patients with malnutrition may have a markedly higher risk of death compared with obese patients without malnutrition.

ACKNOWLEDGMENT This article is dedicated to the memory of our dear friend and colleague Nathan Edward Hellman, MD, PhD.

REFERENCES

1. Flegal KM, Carroll MD, Ogden CL, et al: Prevalence and trends in obesity among US adults, 1999-2008. JAMA 2010; 303:235–241 2. Brown CV, Neville AL, Rhee P, et al: The impact of obesity on the outcomes of 1,153 critically injured blunt trauma patients. J Trauma 2005; 59:1048–1051 3. NHLBI Obesity Education Initiative Expert Panel on the Identification, Evaluation, and Treatment of Obesity in Adults (US): Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults. WMJ 1998; 97:20–21, 24–25, 27–37 4. Adams KF, Schatzkin A, Harris TB, et al: Overweight, obesity, and mortality in a large prospective cohort of persons 50 to 71 years old. N Engl J Med 2006; 355:763–778 5. Varon J, Marik P. Management of the obese critically ill patient. Crit Care Clin 2001; 17:187–200 6. Marik P, Varon J: The obese patient in the ICU. Chest 1998; 113:492–498 7. Boulanger BR, Milzman DP, Rodriguez A: Obesity. Crit Care Clin 1994; 10:613–622. 8. Wilson AT, Reilly CS: Anaesthesia and the obese patient. Int J Obes Relat Metab Disord 1993; 17:427–435 9. Wang HE, Griffin R, Judd S, et al: Obesity and risk of sepsis: A population-based cohort study. Obesity 2013; 21:E762–E769 10. Druml W, Metnitz B, Schaden E, et al: Impact of body mass on incidence and prognosis of acute kidney injury requiring renal replacement therapy. Intensive Care Med 2010; 36:1221–1228 11. Kumar G, Majumdar T, Jacobs ER, et al, Milwaukee Initiative in Critical Care Outcomes Research (MICCOR) Group of Investigators: Outcomes of morbidly obese patients receiving invasive mechanical ventilation: A nationwide analysis. Chest 2013; 144:48–54 12. Brown CV, Neville AL, Rhee P, et al: The impact of obesity on the outcomes of 1,153 critically injured blunt trauma patients. J Trauma 2005; 59:1048–1051; discussion 1051 13. Byrnes MC, McDaniel MD, Moore MB, et al: The effect of obesity on outcomes among injured patients. J Trauma 2005; 58:232–237 14. Bochicchio GV, Joshi M, Bochicchio K, et al: Impact of obesity in the critically ill trauma patient: A prospective study. J Am Coll Surg 2006; 203:533–538 15. Bercault N, Boulain T, Kuteifan K, et al: Obesity-related excess mortality rate in an adult intensive care unit: A risk-adjusted matched cohort study. Crit Care Med 2004; 32:998–1003 16. Tremblay A, Bandi V: Impact of body mass index on outcomes following critical care. Chest 2003; 123:1202–1207 17. Ciesla DJ, Moore EE, Johnson JL, et al: Obesity increases risk of organ failure after severe trauma. J Am Coll Surg 2006; 203:539–545 18. Alban RF, Lyass S, Margulies DR, et al: Obesity does not affect mortality after trauma. Am Surg 2006; 72:966–969 19. Newell MA, Bard MR, Goettler CE, et al: Body mass index and outcomes in critically injured blunt trauma patients: Weighing the impact. J Am Coll Surg 2007; 204:1056–1061; discussion 1062-1054

98

www.ccmjournal.org

20. Brown CV, Rhee P, Neville AL, et al: Obesity and traumatic brain injury. J Trauma 2006; 61(3):572–576. 21. Aldawood A, Arabi Y, Dabbagh O: Association of obesity with increased mortality in the critically ill patient. Anaesth Intensive Care 2006; 34:629–633 22. Hutagalung R, Marques J, Kobylka K, et al: The obesity paradox in surgical intensive care unit patients. Intensive Care Med 2011; 37:1793–1799 23. Marik P, Doyle H, Varon J. Is obesity protective during critical illness? An analysis of a national ICU database. Crit Care Shock 2003; 6:156–162 24. Nasraway SA Jr, Albert M, Donnelly AM, et al: Morbid obesity is an independent determinant of death among surgical critically ill patients. Crit Care Med 2006; 34:964–970; quiz 971 25. Peake SL, Moran JL, Ghelani DR, et al: The effect of obesity on 12-month survival following admission to intensive care: A prospective study. Crit Care Med 2006; 34:2929–2939 26. Incalzi RA, Gemma A, Capparella O, et al: Energy intake and in-hospital starvation. A clinically relevant relationship. Arch Intern Med 1996; 156:425–429 27. Gallagher-Allred CR, Voss AC, Finn SC, et al: Malnutrition and clinical outcomes: The case for medical nutrition therapy. J Am Diet Assoc 1996; 96:361–366, 369; quiz 367–368 28. Dempsey DT, Mullen JL, Buzby GP: The link between nutritional status and clinical outcome: can nutritional intervention modify it? Am J Clin Nutr 1988; 47(2 Suppl):352–356 29. Herrmann FR, Safran C, Levkoff SE, et al: Serum albumin level on admission as a predictor of death, length of stay, and readmission. Arch Intern Med 1992; 152:125–130 30. Fiaccadori E, Lombardi M, Leonardi S, et al: Prevalence and clinical outcome associated with preexisting malnutrition in acute renal failure: A prospective cohort study. J Am Soc Nephrol 1999; 10:581–593 31. Landi F, Onder G, Gambassi G, et al: Body mass index and mortality among hospitalized patients. Arch Intern Med 2000; 160:2641–2644 32. Middleton MH, Nazarenko G, Nivison-Smith I, et al: Prevalence of malnutrition and 12-month incidence of mortality in two Sydney teaching hospitals. Intern Med J 2001; 31455–461 33. Reilly JJ Jr, Hull SF, Albert N, et al: Economic impact of malnutrition: A model system for hospitalized patients. JPEN J Parenter Enteral Nutr 1988; 12:371–376 34. Potter JF, Schafer DF, Bohi RL: In-hospital mortality as a function of body mass index: An age-dependent variable. J Gerontol 1988; 43:M59–M63 35. Butterworth CE Jr: Editorial: Malnutrition in the hospital. JAMA 1974; 230:879 36. Giner M, Laviano A, Meguid MM, et al: In 1995 a correlation between malnutrition and poor outcome in critically ill patients still exists. Nutrition 1996; 12:23–29 37. Popkin BM, Gordon-Larsen P: The nutrition transition: Worldwide obesity dynamics and their determinants. Int J Obes Relat Metab Disord 2004; 28(Suppl 3):S2–S9 38. Kaidar-Person O, Person B, Szomstein S, et al: Nutritional deficiencies in morbidly obese patients: a new form of malnutrition? Part A: Vitamins. Obes Surg 2008; 18:870–876 39. Kaidar-Person O, Person B, Szomstein S, et al: Nutritional deficiencies in morbidly obese patients: A new form of malnutrition? Part B: Minerals. Obes Surg 2008; 18:1028–1034 40. Schweiger C, Weiss R, Berry E, et al: Nutritional deficiencies in bariatric surgery candidates. Obes Surg 2010; 20:193–197 41. Murphy SN, Chueh HC: A security architecture for query tools used to access large biomedical databases. Proc AMIA Symp 2002; 552–556 42. Zager S, Mendu ML, Chang D, et al: Neighborhood poverty rate and mortality in patients receiving critical care in the academic medical center setting. Chest 2011; 139:1368–1379 4 3. Acute Inpatient Prospective Payment System, DRG resources. 2013. Available at: http://www.cms.gov/Medicare/Medicare-Feefor-Service-Payment/AcuteInpatientPPS/index.html?redirect=/ AcuteInpatientPPS/. Accessed December 29, 2013 January 2015 • Volume 43 • Number 1

Clinical Investigations 44. Swails WS, Samour PQ, Babineau TJ, et al: A proposed revision of current ICD-9-CM malnutrition code definitions. J Am Diet Assoc 1996; 96:370–373 45. Blackburn GL, Bistrian BR, Maini BS, et al: Nutritional and metabolic assessment of the hospitalized patient. JPEN J Parenter Enteral Nutr 1977; 1:11–22 46. Simopoulos AP: Obesity and body weight standards. Annu Rev Public Health 1986; 7:481–492 47. Fleisch A: [Basal metabolism standard and its determination with the “metabocalculator”]. Helvetica Medica Acta 1951; 18:23–44 48. Karkeck J: Adjusted body weight for obesity. Am Diet Assoc RenalDiet Practice Group Newsl 1984; 3:6 49. Jacobs DG, Jacobs DO, Kudsk KA, et al, EAST Practice Management Guidelines Work Group: Practice management guidelines for nutritional support of the trauma patient. J Trauma 2004; 57:660–678; discussion 679 50. McClave SA, Martindale RG, Vanek VW, et al, A. S. P. E. N. Board of Directors, American College of Critical Care Medicine, Society of Critical Care Medicine: Guidelines for the Provision and Assessment of Nutrition Support Therapy in the Adult Critically Ill Patient: Society of Critical Care Medicine (SCCM) and American Society for Parenteral and Enteral Nutrition (A.S.P.E.N.). JPEN J Parenter Enteral Nutr 2009; 33:277–316 51. Charlson ME, Pompei P, Ales KL, et al: A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40:373–383 52. Quan H, Sundararajan V, Halfon P, et al: Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care 2005; 43:1130–1139 53. Quan H, Li B, Couris CM, et al: Updating and validating the Charlson comorbidity index and score for risk adjustment in hospital discharge abstracts using data from 6 countries. Am J Epidemiol 2011; 173:676–682 54. Rapoport J, Gehlbach S, Lemeshow S, et al: Resource utilization among intensive care patients. Managed care vs traditional insurance. Arch Intern Med 1992; 152:2207–2212 55. Martin GS, Mannino DM, Eaton S, et al: The epidemiology of sepsis in the United States from 1979 through 2000. N Engl J Med 2003; 348:1546–1554 56. Bellomo R, Ronco C, Kellum JA, et al: Acute renal failure - definition, outcome measures, animal models, fluid therapy and information technology needs: The Second International Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group. Crit Care 2004; 8:R204-R212 57. Bagshaw SM, Uchino S, Cruz D, et al: A comparison of observed versus estimated baseline creatinine for determination of RIFLE class in patients with acute kidney injury. Nephrol Dial Transplant 2009; 24:2739–2744 58. Hoste EA, Clermont G, Kersten A, et al: RIFLE criteria for acute kidney injury are associated with hospital mortality in critically ill patients: A cohort analysis. Crit Care 2006; 10:R73 59. Al-Alwan A, Ehlenbach WJ, Menon PR, et al: Cardiopulmonary resuscitation among mechanically ventilated patients. Intensive Care Med 2014; 40:556–563 60. Cooke CR, Erickson SE, Eisner MD, et al: Trends in the incidence of noncardiogenic acute respiratory failure: The role of race. Crit Care Med 2012; 40:1532–1538 61. Dreyfus B, Kawabata H, Gomez A: Selected adverse events in cancer patients treated with vascular endothelial growth factor inhibitors. Cancer Epidemiol 2013; 37:191–196 62. Purtle SW, Moromizato T, McKane CK, et al: The association of red cell distribution width at hospital discharge and out-of-hospital mortality following critical illness. Crit Care Med 2014; 42:918–929 63. Knaus WA, Draper EA, Wagner DP, et al: APACHE II: A severity of disease classification system. Crit Care Med 1985; 13:818–829 64. Cowper DC, Kubal JD, Maynard C, et al: A primer and comparative review of major US mortality databases. Ann Epidemiol 2002; 12:462–468

Critical Care Medicine

65. Sohn MW, Arnold N, Maynard C, et al: Accuracy and completeness of mortality data in the Department of Veterans Affairs. Popul Health Metr 2006; 4:2 66. Schisterman EF, Whitcomb BW: Use of the Social Security Administration Death Master File for ascertainment of mortality status. Popul Health Metr 2004; 2:2 67. Newman TB, Brown AN: Use of commercial record linkage software and vital statistics to identify patient deaths. J Am Med Inform Assoc 1997; 4:233–237 68. Rubin DB: Estimating causal effects from large data sets using propensity scores. Ann Intern Med 1997; 127(8 Pt 2):757–763 69. D’Agostino RB Jr: Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998; 17:2265–2281 70. Rosenbaum P, Rubin D: Constructing a control group using multivariate matched sampling methods that incorporate the propensity score. Am Stat 1985; 39:33–38 71. Leuven E, Sianesi B: PSMATCH2: Stata module to perform full Mahalanobis and propensity score matching, common support graphing, and covariate imbalance testing. 2003 Available at: http://ideas. repec.org/c/boc/bocode/s432001.html. Accessed July 18, 2013 72. Preacher KJ, Hayes AF: SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behav Res Methods Instrum Comput 2004; 36:717–731 73. Preacher K, Rucker D, Hayes A: Addressing moderated mediation hypotheses: Theory, methods, and prescriptions. Multivariate Behav Res 2007; 42:185–227 74. Doweiko JP, Nompleggi DJ: The role of albumin in human physiology and pathophysiology, Part III: Albumin and disease states. JPEN Journal of Parenteral and Enteral Nutrition 1991; 15:476–483 75. Chlebowski RT, Grosvenor M, Lillington L, et al: Dietary intake and counseling, weight maintenance, and the course of HIV infection. J Am Diet Assoc 1995; 95:428–432; quiz 433-425 76. Kalantar-Zadeh K, Abbott KC, Salahudeen AK, et al: Survival advantages of obesity in dialysis patients. Am J Clin Nutr 2005; 81:543–554 77. Fonarow GC, Srikanthan P, Costanzo MR, et al: An obesity paradox in acute heart failure: analysis of body mass index and inhospital mortality for 108,927 patients in the Acute Decompensated Heart Failure National Registry. Am Heart J 2007; 153:74–81 78. Abhyankar S, Leishear K, Callaghan FM, et al: Lower short- and long-term mortality associated with overweight and obesity in a large cohort study of adult intensive care unit patients. Crit Care 2012; 16:R235 79. Schols AM, Slangen J, Volovics L, et al: Weight loss is a reversible factor in the prognosis of chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1998; 157(6 Pt 1):1791–1797 80. Anker SD, Ponikowski P, Varney S, et al: Wasting as independent risk factor for mortality in chronic heart failure. Lancet 1997; 349:1050–1053 81. Lawson JA, Lazarus R, Kelly JJ: Prevalence and prognostic significance of malnutrition in chronic renal insufficiency. J Ren Nutr 2001; 11:16–22 82. De Lima JJ, da Fonseca JA, Godoy AD: Baseline variables associated with early death and extended survival on dialysis. Ren Fail 1998; 20:581–587 83. Caregaro L, Alberino F, Amodio P, et al: Malnutrition in alcoholic and virus-related cirrhosis. Am J Clin Nutr 1996; 63:602–609 84. Phang PT, Aeberhardt LE: Effect of nutritional support on routine nutrition assessment parameters and body composition in intensive care unit patients. Can J Surg 1996; 39:212–219 85. Gray DS, Bray GA, Bauer M, et al: Skinfold thickness measurements in obese subjects. Am J Clin Nutr 1990; 51:571–577 86. Dvir D, Cohen J, Singer P: Computerized energy balance and complications in critically ill patients: an observational study. Clin Nutr 2006; 25:37–44 87. Jensen GL, Hsiao PY: Obesity in older adults: Relationship to functional limitation. Curr Opin Clin Nutr Metab Care 2010; 13:46–51 88. White JV, Guenter P, Jensen G, et al, Academy of Nutrition and Dietetics Malnutrition Work Group, A.S.P.E.N. Malnutrition Task Force, A.S.P.E.N. www.ccmjournal.org

99

Robinson et al Board of Directors: Consensus statement of the Academy of Nutrition and Dietetics/American Society for Parenteral and Enteral Nutrition: characteristics recommended for the identification and documentation of adult malnutrition (undernutrition). J Acad Nutr Diet 2012; 112:730–738 89. Ferguson M, Capra S, Bauer J, et al: Development of a valid and reliable malnutrition screening tool for adult acute hospital patients. Nutrition 1999; 15:458–464 90. Perioperative total parenteral nutrition in surgical patients. The Veterans Affairs Total Parenteral Nutrition Cooperative Study Group. N Engl J Med 1991; 325:525–532 91. Ingenbleek Y, Carpentier YA: A prognostic inflammatory and nutritional index scoring critically ill patients. Int J Vitam Nutr Res 1985; 55:91–101 92. Buzby GP, Mullen JL, Matthews DC, et al: Prognostic nutritional index in gastrointestinal surgery. Am J Surg 1980; 139:160–167 93. Laporte M, Villalon L, Thibodeau J,et al: Validity and reliability of simple nutrition screening tools adapted to the elderly population in healthcare facilities. J Nutr Health Aging 2001; 5:292–294 94. Kaiser MJ, Bauer JM, Ramsch C, et al, Mini Nutritional Assessment International Group: Frequency of malnutrition in older adults: A multinational perspective using the mini nutritional assessment. J Am Geriatr Soc 2010; 58:1734–1738 95. van Venrooij LM, van Leeuwen PA, Hopmans W, et al: Accuracy of quick and easy undernutrition screening tools--Short Nutritional Assessment Questionnaire, Malnutrition Universal Screening Tool, and modified Malnutrition Universal Screening Tool--in patients undergoing cardiac surgery. J Am Diet Assoc 2011; 111:1924–1930 96. Stenholm S, Harris TB, Rantanen T, et al: Sarcopenic obesity: Definition, cause and consequences. Curr Opin Clin Nutr Metab Care 2008; 11:693–700

100

www.ccmjournal.org

97. Jordao AA, Bellucci AD, Dutra de Oliveira JE, et al: Midarm computerized tomography fat, muscle and total areas correlation with nutritional assessment data. Int J Obes Relat Metab Disord 2004; 28:1451–1455 98. Alberda C, Gramlich L, Jones N, et al: The relationship between nutritional intake and clinical outcomes in critically ill patients: Results of an international multicenter observational study. Intensive Care Med 2009; 35:1728–1737 99. Casaer MP, Mesotten D, Hermans G, et al: Early versus late parenteral nutrition in critically ill adults. N Engl J Med 2011; 365:506–517 100. Jensen GL, Mirtallo J, Compher C, et al, International Consensus Guideline Committee: Adult starvation and disease-related malnutrition: A proposal for etiology-based diagnosis in the clinical practice setting from the International Consensus Guideline Committee. JPEN J Parenter Enteral Nutr 2010; 34:156–159 101. White JV, Guenter P, Jensen G, et al, Academy of Nutrition and Dietetics Malnutrition Work Group, A.S.P.E.N. Malnutrition Task Force, A.S.P.E.N. Board of Directors: Consensus statement: Academy of Nutrition and Dietetics and American Society for Parenteral and Enteral Nutrition: Characteristics recommended for the identification and documentation of adult malnutrition (undernutrition). JPEN J Parenter Enteral Nutr 2012; 36:275–283 102. Moran JL, Green JV, Homan SD, et al: Acute exacerbations of chronic obstructive pulmonary disease and mechanical ventilation: A reevaluation. Crit Care Med 1998; 26:71–78 103. Dardaine V, Dequin PF, Ripault H, et al: Outcome of older patients requiring ventilatory support in intensive care: Impact of nutritional status. J Am Geriatr Soc 2001; 49:564–570

January 2015 • Volume 43 • Number 1