a b s t r a c t

Purpose: This study aimed to investigate the probability of 30-day mortality based on body mass index (BMI) as- sessment combined with pneumonia severity index in patients with community-acquired pneumonia and to determine whether being underweight is an independent risk factor contributing to 30-day mortality.

Basic procedures: A prospectively collected database was analyzed retrospectively. Multivariable logistic regression analysis was performed to determine whether BMI is an independent predictor of mortality in patients with CAP by adjusting for PSI and other factors found significant in univariable analysis. Mortality predictability of BMI and PSI was evaluated using area under the receiver operating characteristic curve analyses.

Main findings: A total of 1403 patients were assessed in this study. In multivariable regression analysis, severe thin- ness (BMI b 16 kg/m²), hypoalbuminemia (albumin b 3.3 mg/dL), and PSI IV and V were predictive factors for 30-day mortality in patients with CAP. In terms of Mortality prediction, the accuracy of PSI was 0.67 (95% confi- dence interval [CI], 0.63-0.71) as measured by the area under the receiver operating characteristic curve. When hypoalbuminemia was combined with PSI, the predictive accuracy significantly increased to 0.71 (95% CI, 0.66- 0.75; P = .02). The addition of severe thinness to PSI and hypoalbuminemia further increased the accuracy signif- icantly to 0.74 (95% CI, 0.70-0.78) (P = .005).

Principal conclusions: Severe thinness (BMI b 16 kg/m²) was associated with 30-day mortality in patients with CAP, showing improved prognostic performance when combined with PSI. We propose that physicians consider a pa- tient’s nutritional state using BMI when predicting mortality in CAP.

(C) 2014

Introduction

Community-acquired pneumonia is a leading cause of morbid- ity and mortality worldwide, and many organizations have attempted to improve the results of current treatment options [1,2]. To improve the Treatment outcome of patients with CAP, factors that increase mortality and morbidity resulting from CAP need to be identified. The Pneumonia severity index is widely used to predict mortality in patients with pneumonia and, in our experience, is the factor on which the decision to admit and discharge patients with CAP is based [3,4]. Nutrition is one of the most important clinical factors influencing mortality and morbidity resulting from infectious disease [5-9]. However, the PSI does not include any clinical factor on malnutrition. A previous study showed that

Abbreviations: AUC, area under the receiver operating characteristic curve; BMI, body mass index; CAP, community-acquired pneumonia; ED, emergency department; ICU, inten- sive care unit; HR, heart rate; PSI, pneumonia severity index; SBP, systolic blood pressure.

? Prior presentations: no.

?? Funding sources: This study was partly supported by grant 11-2014-061 from the Seoul National Universty Bundang Hospital research fund.

* Corresponding author. Department of Emergency Medicine, Seoul National University, Bundang Hospital, 300 Gumi-dong, Bundang-gu, Sungnam-si, Gyeonggi-do, 463-707, Republic of Korea. Tel.: +82 31 787-7572; fax: +82 31 787 4055.

E-mail address: [email protected] (K. Kim).

mortality predictability increases when hypoalbuminemia is factored into PSI [10]. Body mass index (BMI) is widely used to define baseline nu- tritional status in adults [11]. Several studies showed that being under- weight is associated with mortality in patients with pneumonia [12-14]. However, these previous studies did not sufficiently consider nutritional factors that may influence mortality in pneumonia patients. Furthermore, it is unclear whether factoring of being underweight increases the predict- ability of mortality when combined with a severity scale such as PSI [3].

The primary objectives of this study were to investigate 30-day mor- tality predictability of BMI assessment in patients with CAP and to de- termine whether being underweight combined with PSI and other clinical factors is an independent risk factor for 30-day mortality. The secondary objectives were to investigate the association of BMI, inten- sive care unit (ICU) admission rates, and length of hospital stay among patients with CAP.

Methods

Study design and setting

All patients presenting with pneumonia who were admitted to our emergency department (ED) since 2008 have been registered in a

http://dx.doi.org/10.1016/j.ajem.2014.11.019

0735-6757/(C) 2014

prospectively collected pneumonia database [4,10,15]. We extracted and retrospectively analyzed information from this database. The study hospital is a 950-bed tertiary academic hospital with an annual ED census of 80000. This study was approved by the institutional re- view board of the study hospital, and the requirement for patient con- sent was waived.

Data collection and follow-up

The pneumonia database has 130 variables, including demographic factors, clinical factors, laboratory data, and therapeutic results. Vital signs and BMI were recorded at the triage stage in the ED. Initial labora- tory data after ED admission were recorded. We extracted data on the following clinical factors: age, sex, BMI, diabetes mellitus , hyper- tension (HTN), chronic obstructive pulmonary disease, heart failure, ce- rebrovascular disease, liver disease, neoplastic disease, renal disease, systolic blood pressure (SBP), heart rate (HR), respiratory rate (RR), body temperature (BT), white blood cell count, hematocrit, platelet count, and the levels of glucose, albumin, Blood urea nitrogen , creatinine, sodium, and C-reactive protein (CRP). In terms of therapeutic results, we investigated the 30-day mortality, ICU admis- sion, and length of hospital stay (days). Patients who were discharged before 30 days were followed up by phone counseling. In cases of miss- ing patient data, our research team reinvestigated the electronic medi- cal records or obtained additional information from patients by phone or personal interview.

Inclusion and exclusion criteria

Eligible patients were older than 15 years and were hospitalized be- tween November 2009 and September 2013 following a diagnosis of CAP. Patients who were discharged from the ED, transferred to another hospital, or diagnosed with Hospital-acquired pneumonia or health care-associated pneumonia were excluded. Hospital-acquired pneumo- nia was defined as pneumonia that occurred more than 48 hours after hospitalization [16-18]. Patients who were hospitalized in an acute care hospital for more than 2 days within 90 days of the infection, resid- ed in a nursing home or long-term care facility, attended a hemodialysis clinic, recently received Intravenous antibiotics or chemotherapy, or sought wound care within 30 days of the infection were classified as having health care-associated pneumonia [16,18-20].

BMI classification

BMI was calculated as weight (kilogram)/height (meter)². The BMI groups were defined according to the basic criteria of the World Health Organization guidelines with a modified cutoff value for the definition of overweight and obese patients in an Asian population (underweight, b 18.5 kg/m²; normal, 18.5-22.99 kg/m²; overweight, 23-24.99 kg/m²; obese, >=25 kg/m²). The underweight population was divided into 3 subcategories: severe thinness (b 16 kg/m²), moderate thinness (16-16.99 kg/m²), and mild thinness (17-18.49 kg/m²) [11,21-24].

Statistical analysis

Continuous variables were dichotomized by clinical relevance or ref- erence range. Cutoff values were as follows: age more than 65 years, SBP less than 90 mm Hg, HR greater than 25 beats per minute, RR greater than 30 cycles per minute, BT greater than 40?C or BT less than 35?C, WBC less than 4000 or WBC greater than 12000 cells per microliter, he- matocrit less than 30%, platelets less than 130000/mm³, BUN greater than 30 mg/dL, creatinine greater than 1.5 mg/dL, glucose greater than 250 mg/dL, albumin less than 3.3 mg/dL, sodium less than 130 mmol/L, and CRP greater than 0.5 mg/dL. Patients with missing data were excluded from the analysis. Continuous variables are present- ed as mean +- standard deviation and were compared using analysis of

variance. Binominal variables are shown as frequency (percentage) and were analyzed using the ?² or Fisher exact test. Univariable analysis was conducted after dichotomizing variables to compare 30-day mor- tality and clinical factors in each BMI group. Subsequent multivariable logistic regression analysis was performed to determine whether being underweight is a risk factor for mortality and to identify other in- dependent predictors for 30-day mortality by adjusting for PSI and other variables found significant in univariable analysis. Tests for trend analysis and the restricted cubic spline curve were used to deter- mine the trend in 30-day mortality based on BMI [25]. Mortality pre- dictability using only PSI and PSI combined with BMI was calculated in terms of 30-day mortality by computing a receiver operating charac- teristic curve and the Area Under the Receiver Operating Characteristic Curve . The AUCs were compared using the method of Hanley and McNeil [26]. Kaplan-Meier survival analysis and log-rank tests were performed to estimate survival functions according to risk factors. A 2-sided test was used with a 5% level of significance. All calculations were conducted using STATA version 12.0 (StataCorp, College Station, TX).

Results

A total of 4169 patients were admitted to our ED and diagnosed with CAP during the study period. Of these patients, 1421 were hospitalized. Patients with missing patient data were excluded from the analysis, and the remaining 1403 patients were enrolled in this study. Enrolled pa- tients were divided into 6 groups according to BMI (Fig. 1). We followed up enrolled patients for 30 days after the ED visit by electronic medical record review or phone counseling. The mean +- standard deviation age of the enrolled patients was 71.6 +- 14.6 years, and 70.5% were men. The overall 30-day mortality was 11.2%, and the median length of hospital stay was 14.3 days. The mean BMI was 20.9 +- 3.9 kg/m². The cubic spline curve showed that the probability of 30-day mortality increased as the BMI decreased (Fig. 2). Patients with lower BMI tended to have a higher 30-day mortality rate in test for trend analysis (P = .005). In univariable analysis, the following variables were found to differ be- tween BMI groups: age more than 65 years, DM, HTN, heart failure, ce- rebrovascular disease, renal failure, SBP less than 90 mm Hg, HR greater than 125 beats per minute, albumin less than 3.3 mg/dL, creatinine less than 1.5 mg/dL, and PSI. The ICU admission rate and length of hospital stay did not differ significantly between BMI groups (Table 1). Of the

Fig. 1. Study population. A total of 4169 patients were admitted to our ED and diagnosed with CAP during study period. Of these patients, 1421 were hospitalized. Patients with missing value were excluded from the analysis, and the remaining 1403 patients were enrolled in this study. Enrolled patients were divided into 6 groups according to BMI.

Fig. 2. BMI and probability of 30-day mortality. The probability of 30-day mortality was in- creased as the BMI decreased. Patients with lower BMI tended to have a higher 30-day mortality in test for trend analysis (P = .005).

variables that were found significant in univariable analysis, the follow- ing 6 variables were the composition of the PSI: age, renal failure, heart failure, cerebrovascular disease, SBP, and HR. Therefore, these variables

were not included in the multivariate regression model [3]. We per- formed multivariable logistic regression analysis to investigate whether being underweight was an independent risk factor for mortality by adjusting for PSI and the other variables that significantly differed be- tween BMI groups. In the multivariable regression model, the following 3 variables were associated with 30-day mortality in patients with CAP: severe thinness (BMI b 16 kg/m²), PSI IV/V, and hypoalbuminemia (albumin b 3.3 mg/dL) (Table 2). In terms of mortality prediction, the ac- curacy of PSI was 0.67 (95% CI, 0.63-0.71) as measured by the AUC. When hypoalbuminemia was combined with PSI, the prediction accura- cy significantly increased to 0.71 (95% CI, 0.66-0.75; P = .02). Finally, the accuracy further increased significantly to 0.74 (95% CI, 0.70-0.78) when severe thinness was assessed in combination with PSI and hypo- albuminemia (P = .005) (Fig. 3). In Kaplan-Meier survival curves anal- ysis, the probability of 30-day mortality as assessed with PSI was increased in patients with hypoalbuminemia and severe thinness (P b .001) (Fig. 4).

Discussion

Previous studies have shown that being underweight is associated with malnutrition and may increase the mortality and morbidity risk in patients with CAP [12-14]. We wanted to know what subgroup of the underweight population has a mortality risk in CAP and whether

Table 1

Baseline characteristics of underweight, normal, overweight, and obesity groups and univariable analysis

	Group (no. of patients)				Underweight (385)				Normal (574)	Overweight (233)	Obesity (211)	P
	BMI (kg/m2)		Severe thinness (155) b 16		Moderate thinness (75) 16-16.99		Mild thinness (155) 17-18.49		18.5-22.99	23-24.99	>=25
	Demographics (%)
	Age N 65 y		124(80)		63(84)		133(85.8)		434(75.6)	171(73.3)	145(68.7)	.002
	Male sex		105(67.7)		57(76)		120(77.4)		398(69.3)	160(68.6)	150(71)	.305
	Comorbidities (%)
DM		28(18)		14(18.6)		42(27.1)		164(28.5)		81(34.7)	83(39.3)	b.001
HTN		41(26.4)		31(41.3)		72(46.4)		276(48)		135(57.9)	130(61.6)	b.001
COPD		20(12.9)		15(20.0)		18(11.6)		98(17.0)		31(13.3)	33(15.6)	.335
Heart failure		26(16.7)		8(10)		31(20)		155(27)		56(24)	53(25)	.007
Cerebrovascular disease		50(32.2)		27(36)		60(38.7)		213(37.1)		67(28.7)	47(22.2)	.001
Liver disease		6(3.8)		0(0)		13(8.3)		22(3.8)		13(5.5)	13(6.1)	.055
Neoplastic disease		26(16.7)		15(20)		42(27.1)		131(22.8)		57(24.4)	52(24.6)	.321
Renal disease		6(3.8)		9(12)		10(6.4)		63(10)		29(12.4)	34(16.1)	.003
	Vital signs (%)
SBP b90 mm Hg		25(16.1)		5(6.6)		17(10.9)		47(8.1)		18(7.7)	10(4.7)	.005
HR N 125 beats/min		33(21.2)		11(14.6)		29(18.7)		72(12.5)		26(11.1)	23(10.9)	.015
RR N 30 cycles/min		18(11.6)		11(14.6)		24(15.4)		61(10.6)		21(9.0)	17(8.1)	.205
BT N 40?C or b35?C		2(1.2)		0(0)		1(0.6)		7(1.2)		0(0)	6(2.8)	.095
Laboratory findings (%)
WBC b4000 or 12000 cells/uL		76(49.0)		34(45.3)		92(59.3)		277(48.2)		118(50.6)	118(55.9)	.097
Hematocrit b30%		35(22.5)		18(24)		32(20.6)		114(19.8)		38(16.3)	39(18.4)	.596
Platelet b130000 cells/uL		21(13.5)		9(12)		19(12.2)		106(18.4)		39(16.7)	39(18.4)	.279
Glucose N 250 mg/dL		11(7.1)		4(5.3)		11(7.1)		62(10.8)		22(9.4)	30(14.2)	.092
Albumin b3.3 mg/dL		71(56.8)		31(50.8)		70(53)		200(42.1)		66(35.2)	51(29.8)	b.001
BUN N 30 mg/dL		41(26.4)		15(20)		34(21.9)		126(21.9)		47(20.1)	58(27.4)	.376
Creatinine N 1.5 mg/dL		17(10.9)		9(12)		22(14.1)		115(20)		50(21.4)	62(29.3)	b.001
Sodium b130 mmol/L		31(20)		15(20)		31(20)		75(13)		29(12.4)	30(14.2)	.062
CRP N 0.5 mg/dL		153(98.7)		72(96)		151(97.4)		556(96.8)		221(94.8)	208(98.5)	.188
PSI (%)
I, II		21(13.5)		7(9.3)		10(6.4)		92(16)		36(15.4)	33(15.6)	.001
III		23(14.8)		14(18.6)		20(12.9)		92(16)		49(21)	43(20.3)
IV		54(34.8)		34(45.3)		67(43.2)		232(40.4)		98(42)	95(45)
V		57(36.7)		20(26.6)		58(37.4)		158(27.5)		50(21.4)	40(18.9)
	Result (%)
ICU admission (%)		23(14.8)		7(9.3)		27(17.4)		69(12)		24(10.3)	25(11.8)	.293
Hospital stay (d) +- SD		17.1(16.1)		14.9(16.9)		16.3(20)		14.1(16.6)		13(14.7)	12.8(19.1)	.093
30-d mortality (%)		32(20.6)		14(18.6)		23(14.8)		62(10.8)		13(5.5)	14(6.6)	.011

Categorical data are presented as number (percentage) of patients and analyzed by ?² test. Continuous data are presented as mean +- SD and analyzed by analysis of variance. Age more than 65 years, DM, HTN, heart failure, cerebrovascular disease, renal failure, SBP less than 90 mm Hg, HR greater than 125 beats per minute, albumin less than 3.3 mg/dL, creatinine less than 1.5 mg/dL, and PSI were different among each BMI groups. Patients with lower BMI had a tendency to higher 30-day mortality in test for trend (P = .011). Abbreviations: COPD, chronic obstructive pulmonary disease; SD, standard deviation.

Table 2

Multivariate logistic regression analysis of risk factors for 30-day mortality

	Odds ratio	95% CI
BMI (kg/m2)
Severe thinness (b16)	1.89	1.09-3.26
Moderate thinness (16-16.99)	1.66	0.79-3.49
Mild thinness (17-18.49)	1.11	0.61-2
Normal (18.5-22.99)	Reference
Overweight (23-24.99)	0.53	0.25-1.08
Obesity (>=25)	0.91	0.47-1.72
PSI I, II	Reference
III	1.80	0.60-5.36
IV	2.93	1.12-7.65
V	6.04	2.28-15.9
DM	0.87	0.55-1.35
HTN	0.95	0.62-1.44
Albumin b3.3 mg/dl	2.24	1.49-3.35
Creatinine N 1.5 mg/dL	1.06	0.65-1.69

Different variables in univariable analysis were used in multivariable analysis. Severe thin- ness (BMI b16 kg/m²), PSI IV/V, and albumin b 3.3 mg/dL were associated with 30-day mortality.

being underweight increased the predictability of mortality when assessed with PSI and Albumin level. The cubic spline curve showed that the probability of 30-day mortality and the slope of the curve in- creased as BMI decreased. The increase in mortality rate was greater in patients with lower BMI (Fig. 2), and the result corresponded with multivariable logistic regression analysis. Patients with lower BMI tended to have a higher 30-day mortality rate in test for trend analysis (P = .011). However, only severe thinness (BMI b 16 kg/m²) correlated significantly with prediction of 30-day mortality in multivariablete analysis (Table 2). Thus, we showed that severe thinness in the under- weight population is a risk factor for mortality and can be used as a pre- dictive factor for mortality when combined with PSI and the albumin level in patients with CAP. To the best of our knowledge, this is the first study to describe clinical outcomes in a subgroup of underweight patients with CAP. Corrales-Medina et al [12] previously showed that patients with lower BMI had a trend towards an increased mortality rate in pneumococcal, Haemophilus CAP. However, the authors did not compare each BMI group in terms of mortality rate. This study did not sufficiently consider the clinical information of underweight patients, and the number of underweight patients was only 36. LaCroix et al

[13] proved that underweight men who were diagnosed with pneumo-

nia had a high mortality rate; however, the results were not adjusted for

Fig. 3. Mortality predictability of PSI, albumin, and BMI. In terms of mortality prediction, PSI showed an accuracy of 0.67 (95% CI, 0.63-0.71) as measured by the AUC. When hypo- albuminemia (albumin b3.3 mg/dL) was combined with PSI, the prediction accuracy was significantly increased to 0.71 (95% CI, 0.66-0.75; P = .02). Finally, the AUC was increased significantly to 0.74 (95% CI, 0.70-0.78) when severe thinness (BMI b 16 kg/m²) was assessed with PSI and hypoalbuminemia (P = .005).

Fig. 4. Kaplan-Meier survival curves and log-rank test. The probability of 30-day mortality was increased in patients with severe thinness (BMI b16 kg/m²) and hypoalbuminemia (albumin b 3.3 mg/dL) when assessed with PSI.

clinical factors that may influence mortality in pneumonia. King et al

[27] also investigated BMI and mortality in pneumonia but did not con- sider clinical factors such as laboratory results that reflect the severity of pneumonia. Therefore, based on previous studies, we could not conclu- sively determine whether being underweight increases the mortality predictability of PSI in patients with CAP. In terms of obesity and pneu- monia, several studies have showed that obesity is associated with a low mortality rate in patients with CAP. In these studies, the authors insisted that obesity was protective in patients with pneumonia [12,27-30]. Al- though patients in the higher BMI group had low mortality rate in our study, we did not observe a significant advantage for obese patients based on multivariate analysis. In our study, hypoalbuminemia was as- sociated with mortality in patients with CAP, a finding that is supported by a previous study [10]. Hypoalbuminemia, as well as BMI, reflects mal- nutrition and could be a predictive factor in patients with CAP.

This study had several limitations. Firstly, this study was conducted in Asia, which might limit the applicability of these results to other pop- ulations. Asians have different body compositions and obesity complica- tion rates from Europeans and other populations. Therefore, we used BMI modified cutoff values for an Asian population when defining over- weight and obese participants based on World Health Organization guidelines and several studies [11,21-24]. However, cutoff values for defining underweight for Asians are the same as the international clas- sification. Therefore, the results regarding underweight may be applica- ble to other populations. Secondly, this study was conducted in a single institute and included patients who were admitted to the ED. A multi- center study is needed to confirm whether severe thinness could be considered a mortality risk factor in patients with CAP. Finally, potential bias may exist despite our use of a prospectively collected database.

In conclusion, severe thinness (BMI b 16 kg/m²) is associated with an increased 30-day mortality risk in patients with CAP and improved the prognostic performance of PSI. We propose that physicians consider the patient’s nutritional state using BMI when predicting mortality in pa- tients with CAP.

Acknowledgment

No authors declare a conflict of interest.

References

1. Mandell LA, Wunderink RG, Anzueto A, Bartlett JG, Campbell GD, Dean NC, et al. In- fectious Diseases Society of America/American Thoracic Society consensus guide- lines on the management of community-acquired pneumonia in adults. Clin Infect Dis 2007;44(Suppl 2):S27-72.
2. Jackson ML, Neuzil KM, Thompson WW, Shay DK, Yu O, Hanson CA, et al. The burden of community-acquired pneumonia in seniors: results of a population-based study. Clin Infect Dis 2004;39(11):1642-50.
3. Fine MJ, Auble TE, Yealy DM, Hanusa BH, Weissfeld LA, Singer DE, et al. A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med 1997;336(4):243-50.
4. Jo S, Kim K, Jung K, Rhee JE, Cho IS, Lee CC, et al. The effects of incorporating a pneu- monia severity index into the admission protocol for community-acquired pneumo- nia. J Emerg Med 2012;42(2):133-8.
5. Baqui AH, Sack RB, Black RE, Chowdhury HR, Yunus M, Siddique AK. Cell-mediated immune deficiency and malnutrition are independent risk factors for persistent diarrhea in Bangladeshi children. Am J Clin Nutr 1993;58(4):543-8.
6. Van den Broeck J, Eeckels R. Effect of malnutrition on child mortality. Lancet 1994; 344(8917):273.
7. Beisel WR. History of nutritional immunology: introduction and overview. J Nutr 1992;122(3 Suppl.):591-6.
8. Tomkins A, Behrens R, Roy S. The role of zinc and vitamin A deficiency in diarrhoeal syndromes in Developing countries. Proc Nutr Soc 1993;52(1):131-42.
9. Beisel WR. Single nutrients and immunity. Am J Clin Nutr 1982;35(2 Suppl.):417-68.
10. Lee JH, Kim J, Kim K, Jo YH, Rhee J, Kim TY, et al. Albumin and C-reactive protein have prognostic significance in patients with community-acquired pneumonia. J Crit Care 2011;26(3):287-94.
11. Physical status: the use and interpretation of anthropometry. Report of a WHO Expert Committee. World Health Organ Tech Rep Ser 1995;854:1-452.
12. Corrales-Medina VF, Valayam J, Serpa JA, Rueda AM, Musher DM. The obesity para- dox in community-acquired Bacterial pneumonia. Int J Infect Dis 2011;15(1):e54-7.
13. LaCroix AZ, Lipson S, Miles TP, White L. Prospective study of pneumonia hospitaliza- tions and mortality of U.S. older people: the role of chronic conditions, health behav- iors, and Nutritional status. Public Health Rep 1989;104(4):350-60.
14. Lange P, Vestbo J, Nyboe J. Risk factors for death and hospitalization from pneumo- nia. A prospective study of a general population. Eur Respir J 1995;8(10):1694-8.
15. Lee JH, Chung HJ, Kim K, Jo YH, Rhee JE, Kim YJ, et al. red cell distribution width as a prognostic marker in patients with community-acquired pneumonia. Am J Emerg Med 2013;31(1):72-9.
16. Tablan OC, Anderson LJ, Besser R, Bridges C, Hajjeh R. Cdc, Healthcare Infection Control Practices Advisory C. Guidelines for preventing health-care-associated

    pneumonia, 2003: recommendations of CDC and the Healthcare Infection Control Practices Advisory Committee. MMWR Recomm Rep 2004;53(RR-3):1-36.

    Niederman MS. Guidelines for the management of respiratory infection: why do we need them, how should they be developed, and can they be useful? Curr Opin Pulm Med 1996;2(3):161-5.

17. American Thoracic S, Infectious Diseases Society of A. Guidelines for the manage- ment of adults with hospital-acquired, ventilator-associated, and healthcare- associated pneumonia. Am J Respir Crit Care Med 2005;171(4):388-416.
18. Hutt E, Kramer AM. Evidence-based guidelines for management of nursing home- acquired pneumonia. J Fam Pract 2002;51(8):709-16.
19. Mylotte JM. Nursing home-acquired pneumonia. Clin Infect Dis 2002;35(10): 1205-11.
20. Weisell RC. Body mass index as an indicator of obesity. Asia Pac J Clin Nutr 2002;11 (Suppl. 8):S681-4.
21. James WP, Chunming C, Inoue S. Appropriate Asian body mass indices? Obes Rev 2002;3(3):139.
22. Misra A. Redefining obesity in Asians: more definitive action is required from the WHO. Natl Med J India 2004;17(1):1-4.
23. Kim DM, Ahn CW, Nam SY. Prevalence of obesity in Korea. Obes Rev 2005;6(2): 117-21.
24. Cleophas TJ. Clinical trials: spline modeling is wonderful for nonlinear effects. Am J Ther 2013;20(5):519-24.
25. Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983;148(3): 839-43.
26. King P, Mortensen EM, Bollinger M, Restrepo MI, Copeland LA, Pugh MJ, et al. Impact of obesity on outcomes for patients hospitalised with pneumonia. Eur Respir J 2013; 41(4):929-34.
27. Singanayagam A, Singanayagam A, Chalmers JD. Obesity is associated with improved survival in community-acquired pneumonia. Eur Respir J 2013;42(1):180-7. http:// dx.doi.org/10.1183/09031936.00115312.
28. Nie W, Zhang Y, Jee SH, Jung KJ, Li B, Xiu Q. Obesity survival paradox in pneumonia: a meta-analysis. BMC Med 2014;12:61.
29. Kahlon S, Eurich DT, Padwal RS, Malhotra A, Minhas-Sandhu JK, Marrie TJ, et al. Obe- sity and outcomes in patients hospitalized with pneumonia. Clin Microbiol Infect 2013;19(8):709-16.