Journal of Critical Care 43 (2018) 7–12

Contents lists available at ScienceDirect

Journal of Critical Care

journal homepage: www.jccjournal.org

Sepsis/Infection

Initial fluid resuscitation following adjusted body weight dosing is

associated with improved mortality in obese patients with suspected
septic shock
Stephanie Parks Taylor, MD a,⁎, Colleen H. Karvetski, PhD b, Megan A. Templin, MPH, MS c,
Alan C. Heffner, MD, FCCM d,e, Brice T. Taylor, MD d
a
Department of Internal Medicine, Carolinas Medical Center, 1000 Blythe Blvd MEB 5th floor, Charlotte, NC, 28203, USA
b
Information and Analytics Services, Carolinas Medical Center, Charlotte, NC, USA
c
Center for Outcomes Research and Evaluation, Carolinas HealthCare System, Charlotte, NC, USA
d
Department of Internal Medicine, Division of Critical Care, Carolinas Medical Center, Charlotte, NC, USA
e
Department of Emergency Medicine Carolinas Medical Center, Charlotte, NC, USA

a r t i c l e i n f o a b s t r a c t

Keywords: Objective: The optimal initial fluid resuscitation strategy for obese patients with septic shock is unknown. We
Body mass index evaluated fluid resuscitation strategies across BMI groups.
Fluid therapy Materials and methods: Retrospective analysis of 4157 patients in a multicenter activation pathway for treatment
Hospital mortality of septic shock between 2014 and 2016.
Obesity Results: 1293 (31.3%) patients were obese (BMI ≥ 30). Overall, higher BMI was associated with lower mortality,
Sepsis
however this survival advantage was eliminated in adjusted analyses. Patients with higher BMI received signif-
Septic shock
icantly less fluid per kilogram at 3 h than did patients with lower BMI (p ≤0.001). In obese patients, fluid given at
3 h mimicked a dosing strategy based on actual body weight (ABW) in 780 (72.2%), adjusted body weight
(AdjBW) in 95 (8.8%), and ideal body weight (IBW) in 205 (19.0%). After adjusting for condition- and treat-
ment-related variables, dosing based on AdjBW was associated with improved mortality compared to ABW
(OR 0.45; 95% CI [0.19, 1.07]) and IBW (OR 0.29; 95% CI [0.11,0.74]).
Conclusions: Using AdjBW to calculate initial fluid resuscitation volume for obese patients with suspected shock
may improve outcomes compared to other weight-based dosing strategies. The optimal fluid dosing strategy for
obese patients should be a focus of future prospective research.
© 2017 Elsevier Inc. All rights reserved.

1. Introduction Despite Surviving Sepsis Campaign recommendations for standard-

Although obesity is a well-recognized risk factor for a considerable
proportion of human disease, the association between obesity and ad-
verse outcomes is not straightforward. A paradoxical survival benefit
of obesity has been demonstrated in critically ill patients [1,2] and spe-
cifically in those with severe sepsis and septic shock [3-5]. This phenom-
enon of reduced mortality with increasing body mass index (BMI) has
been termed the obesity paradox. The mechanism underlying this effect
is unclear. Several theories have been postulated, including favorable
endocrine [6] and inflammatory [7-10] milieu, availability of nutritional
reserves [1,10], lower relative fluid administration [3-4], more vigilant
care [11-12], and unaccounted confounds.
⁎ Corresponding author.
E-mail address: stephanie.p.taylor@carolinashealthcare.org (S.P. Taylor).
ized initial fluid resuscitation (i.e. 30 ml/kg) [13], a few studies have
shown that obese patients receive relatively lower fluid volumes than
non-obese patients [3-4,14]. Although some patients with septic shock
may be hypovolemic and require fluid to restore intravascular volume,
the primary rationale for fluid administration in distributive shock is
to augment stroke volume in patients who are fluid responsive. It is
unclear what role obesity has on fluid responsiveness for patients
with septic shock. Because excessive fluid administration is known to
be associated with worse outcomes [15-17], it is plausible that more
conservative fluid administration for patients with higher BMI contrib-
uted to lower mortality in these groups [3,4]. The optimal dosing
strategy for obese patients is unknown — guidelines recommending
weight-based fluid administration do not clearly specify whether actual,
ideal, or adjusted body weight should be used to calculate total fluid
volumes and there are no robust data to guide this decision.
Previous studies evaluating fluid resuscitation in obese septic pa-
tients found that patients across all BMI ranges received an equivalent

http://dx.doi.org/10.1016/j.jcrc.2017.08.025
0883-9441/© 2017 Elsevier Inc. All rights reserved.
8 S.P. Taylor et al. / Journal of Critical Care 43 (2018) 7–12
total fluid bolus, [3-4,14] which represents either a non-weight-based
strategy or a strategy based on ideal body weight. The optimal fluid
dosing strategy is unknown. We hypothesized that dosing based on
IBW or AdjBW would be associated with improved outcomes compared
to ABW.
2. Methods
The Carolinas Healthcare System institutional review board
approved the study protocol. We performed a retrospective analysis of
patients enrolled in a septic shock registry from 9 facilities within Caro-
linas HealthCare System between January 1, 2014 and June 30, 2016. All
patients at least 18 years old with weight and height documented were
included. Patients were included in the registry if they were admitted
through the emergency department as code sepsis, defined as a
suspected infection with either refractory hypotension or an initial
lactate level greater than or equal to 4 mmol/L. Once patients were iden-
tified as code sepsis, providers use a standardized computer provider
order entry protocol which includes the following elements for comple-
tion within the first 3 h from the time of code sepsis activation: 1) anti-
biotics administered within 1 h, 2) lactate result returned, 3) blood
cultures (prior to antibiotics), and 4) infusion of 30 ml/kg of crystalloid.
2.1. Weight classifications
BMI was calculated using the height and body weight documented
in the medical record at the time of hospital admission. Weight catego-
ries were defined based on the World Health Organization (WHO)
criteria: underweight (BMI b 18.5 kg/m2), normal range (BMI = 18.5
to 24.9 kg/m2), overweight (BMI = 25.0 to 29.9 kg/m2), obese (BMI =
30 to 39.9 kg/m2) and obese class III (BMI N 40 kg/m2).
2.2. Clinical characteristics
Baseline demographic, physiologic, and laboratory parameters at
presentation were recorded. We specifically chose to use individual
markers of illness acuity (e.g. systolic blood pressure, mean arterial
pressure, heart rate, oxygen saturation, lactate level, serum creatinine,
temperature, and white blood count) rather than scoring systems such
as the Acute Physiology and Chronic Health Evaluation or Simplified
Acute Physiology Score to avoid the covariation of the chronic illness
components of these scores with our comorbidity variables. Comorbid
conditions diabetes, cirrhosis, chronic kidney disease, dementia, malig-
nancy, heart failure, COPD, acute kidney injury were identified by ICD-9
or ICD-10 discharge diagnosis codes present within the past 24 months
of the index inpatient encounter. We calculated ideal body weight
(IBW) as recommended for males and females [20] and adjusted body
weight (AdjBW) based on the recommended formula for medication
dosing [21].
2.3. Sepsis treatments
We recorded the following information regarding interventions for
patients with severe sepsis: compliance with 3-h sepsis bundle, fluid
volume administered at 3 and 6 h, mechanical ventilation, and central
line placement. We recorded lactate normalization (lactate b2 mmol/l)
at 6 h.
2.4. Outcomes
The primary outcome was in-hospital mortality. Secondary out-
comes were ICU mortality, ICU length of stay, and hospital length of stay.
2.5. Statistical analysis
Patients were compared across demographic, comorbid, clinical con-
dition/acuity, and treatment variables by BMI group. For the univariate
analyses comparing BMI groups, continuous variables were analyzed
using ANOVAs or Kruskal-Wallis tests where appropriate and categori-
cal variables were compared using chi-square tests. We used multiple
logistic regressions to assess the set of confounding factors associated
with in-hospital mortality in 2 separate models: (1) for the overall pop-
ulation, with BMI group as an independent predictor and (2) for the
obese population only (BMI ≥ 30), where we use categorical dosing
strategy (ABW, IBW, or AdjBW) as an independent predictor. For both
models, any variable significant using a backward elimination method
was considered for inclusion in the reduced model using a hierarchical
approach (demographic, comorbidities, clinical condition, and treatment
models were each generated). Variables were dropped if p N 0.15. In
model (1), the reduced logistic regression model was generated using
a stepwise elimination method including variables with p ≤ 0.05, forcing
the variable BMI group (reference set at normal BMI); in model (2), the
stepwise elimination method was also applied, with dosing strategy
(ABW, AdjBW, or IBW) as a forced independent predictor (reference
set at ABW). ORs and 95% CIs are reported for the crude, full, and reduced
models.
A separate multiple linear regression model was used to determine
association between normalized fluid volume delivered in 3 h and
BMI group, while controlling for clinically relevant confounding
variables. We conducted analyses using SAS Enterprise Guide software
version 6.1 (SAS Institute Inc., Cary, NC, USA).
3. Results
3.1. Baseline characteristics
Of the 4157 patients included during the study period, 4126 (99.3%)
had height and weight documented. Overall, 7.6% of the study popula-
tion were underweight, 34.0% were normal weight, 27.0% were over-
weight, 22.4% were obese and 8.9% were very obese. Baseline
characteristics among BMI groups are shown in Table 1. Increasing
BMI was associated with younger age, male sex, higher initial systolic
blood pressure (SBP), higher initial temperature, lower initial oxygen
saturation and lower white blood cell count. Patients with higher BMI
had greater frequency of comorbid congestive heart failure, cirrhosis,
and type 2 diabetes, but lower frequency of dementia and malignancy.
Soft tissue infection was more common source of sepsis in patients
with higher BMI. Increasing BMI was associated with lower frequency
of do-not-resuscitate (DNR) orders present on admission.
3.2. Sepsis interventions
Table 2 shows the pattern of sepsis interventions and fluid dosing
among BMI groups. Patients with higher BMI were more likely to be ad-
mitted to an intensive care unit than were lower BMI patients (p =
0.009). Increasing BMI was associated with a higher frequency of me-
chanical ventilation, central line placement, and a trend towards greater
use of vasopressor agents.
During the initial resuscitation, patients with higher BMI received
significantly less fluid per kilogram at 3 h than did patients with lower
BMI (p ≤0.0001). Effective dosing strategy was determined by dividing
total fluid received at 3 h by 30 c/kg, and determining whether this
value was closest to the patient's ABW, AdjBW, or IBW. In obese pa-
tients, fluid given at 3 h mimicked a dosing strategy based on actual
body weight (ABW) in 780 (72.2%), adjusted body weight (AdjBW) in
95 (8.8%), and ideal body weight (IBW) in 205 (19.0%) across 1080 pa-
tients. Characteristics of patients in each of these dosing groups is
shown in Table 3.
 S.P. Taylor et al. / Journal of Critical Care 43 (2018) 7–12 9

Table 1
Baseline Characteristics by Body Mass Index Category.

Body mass index (kg/m2)

Demographic and clinical data b18.50 18.50–24.99 25.00–29.99 30.00–39.99 N40.00 P-value
(n = 315) (n = 1401) (n = 1117) (n = 925) (=368)

Age (years) mean (SD) 65.5 (17.6) 64.9 (18.8) 64.5 (16.5) 61.6 (15.5) 57.0 (12.8) b0.001
Sex (female) number (%) 160 (50.8) 766 (54.7) 582 (52.1) 399 (43.1) 111 (30.2) b0.001
Actual body weight (kg), mean (SD) 47.6 (8.0) 63.6 (9.5) 78.9 (10.5) 96.0 (14.7) 136.1 (31.1) b0.001
Ideal body weight (kg), mean (SD) 63.4 (8.4) 63.3 (8.1) 63.3 (7.6) 62.4 (8.0) 62.5 (8.6) 0.02
BMI (kg/m2), mean (SD) 16.5 (1.6) 22.0 (1.8) 27.4 (1.4) 33.9 (2.8) 48.3 (8.1) b0.001

Comorbidities, number (%)

Cirrhosis 26 (8.3) 145 (10.3) 138 (12.4) 124 (13.4) 53 (14.4) 0.02
Congestive heart failure 52 (16.5) 313 (22.3) 277 (24.8) 216 (23.4) 114 (31) b0.001
COPD 100 (31.7) 375 (26.8) 280 (25.1) 253 (27.4) 111 (30.2) 0.11
Chronic kidney disease 62 (19.7) 315 (22.5) 256 (22.9) 216 (23.4) 106 (28.8) 0.06
Diabetes mellitus (type 2) 83 (26.3) 441 (31.5) 443 (39.7) 475 (51.4) 220 (59.8) b0.001
Malignancy 69 (21.9) 314 (22.4) 239 (21.4) 166 (17.9) 36 (9.8) b0.001
Dementia 97 (30.8) 370 (26.4) 223 (20) 131 (14.2) 26 (7.1) b0.001
Infection source, number (%)
Urinary tract 86 (27.3) 343 (24.5) 275 (24.6) 244 (26.4) 96 (26.1) 0.71
Pneumonia 97 (30.8) 373 (26.6) 300 (26.9) 231 (25) 89 (24.2) 0.27
Cellulitis 15 (4.8) 56 (4) 57 (5.1) 70 (7.6) 80 (19) b0.001
C. difficile colitis 16 (5.1) 69 (4.9) 42 (3.8) 36 (3.9) 13 (3.5) 0.48

Initial clinical parameters

SBP (mm Hg), median (SD) 101 (27.2) 105 (29.3) 108 (31) 110 (31) 114 (30.1) b0.001
MAP (mm Hg), median (SD) 73.5 (19.6) 70 (21.4) 71 (22.0) 69 (21.2) 70 (24.6) 0.57
Heart rate (bpm), median (SD) 107 (23.5) 106 (23.6) 104 (23.5) 104 (24.3) 105 (23.3) 0.28
Oxygen saturation, %, median (SD) 97 (9.2) 97 (9) 96 (6.3) 96 (6.2) 96 (8) 0.03
Temperature, F, mean (SD) 98.2 (2.7) 98.7 (2.9) 98.9 (2.6) 99.0 (2.4) 99.2 (2.8) b0.001
Serum creatinine, mean (SD) 1.8 (1.7) 2.0 (2.2) 2.0 (2.3) 2.2 (2.1) 2.2 (1.9) b0.001
White blood cell count, mean (SD) 16.2 (11.1) 15.3 (11.8) 14.6 (11.2) 15.5 (9.4) 15.8 (7.7) b0.001
Initial lactate, mmol/L, mean (SD) 5.1 (3.1) 4.8 (3.3) 4.8 (3.1) 4.6 (2.8) 4.5 (2.7) 0.14
Multidrug resistant organism status, number (%)
CRE 12 (3.8) 27 (1.9) 27 (2.4) 24 (2.6) 8 (2.2) 0.37
MRSA 18 (5.7) 85 (6.1) 64 (5.7) 59 (6.4) 24 (6.5) 0.96
Do-not resuscitate order (present), number (%) 44 (14.0) 164 (11.7) 91 (8.1) 60 (6.5) 24 (6.5) b0.001

Crude outcomes
Hospital mortality, N (%) 69 (21.9) 227 (16.2) 167 (15.0) 142 (15.4) 50 (13.6) 0.026
ICU mortality, N (%) 65 (28.4) 196 (18.1) 153 (17.5) 131 (18.0) 46 (14.9) 0.001
Hospital length of stay, median (IQR) 5 (45) 5 (122) 5 (68) 5 (68) 6 (88) 0.001
ICU length of stay, median (IQR) 1.7 (16.4) 1.9 (30.0) 1.9 (30.0) 1.9 (35.1) 2.1 (38.6) 0.15

BMI: Body mass index, COPD: chronic obstructive pulmonary disease, SBP: Systolic blood pressure, MAP: Mean arterial pressure, CRE: Carbapenem-resistant Enterobacteriaceae, MRSA:
Methicillin-resistant Staphylococcus aureus.

Fluid per kilogram at 6 h was also lower for patients with higher
BMI than for patients with lower BMI (p b 0.0001). Among the
3072 (74.5% of 4126) subjects with elevated lactate (N 2) at triage,
lactate normalization rate was significantly different by BMI group
(p = 0.004). Specifically, lactate normalization rates were lower for
patients with obese patients compared to non-obese patients
(58.1% vs 52.1%; p = 0.002).
3.3. Clinical outcomes
As shown in Table 1, crude hospital mortality, ICU mortality, and
hospital LOS differed among BMI groups. The logistic regression
modeling for mortality outcome over the entire population (n =
3746) with BMI group forced as an independent predictor revealed
significant independent predictors of mortality: patient age (OR 1.04;

Table 2
Sepsis related interventions among BMI groups.

Body mass index (kg/m2)

Interventions b18.50 18.50–24.99 25.00–29.99 30.00–39.99 ≥ 40.00 P-value

(n = 315) (n = 1401) (n = 1117) (n = 925) (=368)

Total fluid at 3 h, (L), mean (SD) 2.5 (1.3) 2.6 (1.6) 2.8 (1.7) 3.0 (1.9) 3.5 (2.3) b0.0001
Fluid at 3 h (ml/kg actual body weight), median (IQR) 49.8 (202.8) 40.4 (177.0) 35.0 (126.9) 32.7 (95.0) 29.8 (73.3) b0.0001
Fluid at 3 h (ml/kg ideal body weight), median (IQR) 36.7 (100.8) 40.6 (159.3) 43.8 (157.7) 50.6 (138.0) 60.1 (202.0) b0.0001
Fluid at 3 h (ml/kg adjusted body weight), median (IQR) 40.7 (126.2) 40.5 (166.0) 39.8 (141.5) 41.5 (111.8) 42.4 (117.8) 0.21
Fluid at 6 h (ml/kg actual body weight), median (IQR) 62.5 (207.3) 47.8 (224.1) 41.1 (169.9) 37.6 (120.5) 30.8 (101.1) b0.0001
Intensive Care Unit admission, number (%) 229 (72.7%) 1081 (77.2%) 876 (78.4%) 726 (78.5%) 309 (84.0%) 0.009
Central line placed, N (%) 184 (58.4) 872 (62.6) 682 (61.1) 609 (65.8) 256 (69.6) 0.005
Vasopressors used, N (%) 136 (43.2) 674 (48.1) 532 (47.6) 464 (50.2) 196 (53.3) 0.079
Mechanical ventilation, N (%) 56 (17.8) 224 (16.6) 206 (18.4) 171 (18.5) 84 (22.8) 0.041
Lactate Normalization Rate (%) 145 (58.2) 622 (60.5) 455 (55.1) 355 (51.6) 149 (53.2) 0.004
Sepsis bundle compliant, N (%) 205 (65.1) 854 (61.0) 606 (54.3) 457 (49.4) 144 (39.1) b0.0001
Antibiotics compliant, N (%) 288 (91.4) 1257 (89.7) 1007 (90.2) 808 (87.4) 323 (87.8) 0.13
Fluid goal compliant, N (%) 262 (83.2) 999 (71.3) 719 (64.4) 531 (57.4) 183 (49.7) b0.0001
10 S.P. Taylor et al. / Journal of Critical Care 43 (2018) 7–12

Table 3
Patient characteristics and outcomes by dosing strategy.

Demographic and clinical data Ideal body weight Adjusted body weight Actual BW P-value
(n = 205) (n = 95) (n = 780)

Age (years), mean (SD) 60.6 (14.3) 61.5 (14.2) 61.0 (14.8) 0.87
Sex (female), N (%) 87 (42.4) 40 (42.1) 292 (37.4) 0.33
Actual body weight (kg), mean (SD) 110.4 (28.8) 120.6 (32.7) 105.4 (26.1) b0.001
Ideal body weight (kg), mean (SD) 63.0 (8.1) 63.2 (7.5) 62.2 (8.4) 0.30
BMI (kg/m2), mean (SD) 38.6 (8.5) 42.3 (11.4) 37.5 (7.5) b0.001

Comorbidities, N (%)
Cirrhosis 35 (17.1) 16 (16.8) 105 (13.5) 0.33
Congestive heart failure 61 (29.8) 27 (28.4) 203 (26.0) 0.53
COPD 62 (30.2) 33 (34.7) 226 (29.0) 0.50
Chronic kidney disease 68 (33.2) 25 (26.3) 179 (22.9) 0.011
Diabetes mellitus (type 2) 111 (54.1) 63 (66.3) 419 (53.7) 0.064
Malignancy 36 (17.6) 18 (18.9) 118 (15.1) 0.49
Dementia 26 (12.7) 7 (7.4) 102 (13.1) 0.28
Infection source, N (%)
Urinary tract 49 (23.9) 29 (30.5) 214 (27.4) 0.43
Pneumonia 53 (25.9) 24 (25.3) 195 (25.0) 0.97
Cellulitis 25 (12.2) 15 (15.8) 81 (10.4) 0.25
C. difficile colitis 6 (2.9) 3 (3.2) 31 (4.0) 0.75
Initial clinical parameters
SBP (mm Hg), mean (SD) 117.5 (34.5) 117.0(28.6) 112.6 (30.4) 0.098
MAP (mm Hg), mean (SD) 76.3 (23.6) 71.6 (19.5) 72.0 (21.7) 0.052
Heart rate (bpm), mean (SD) 105.9 (23.3) 101.4 (22.7) 105.1 (23.9) 0.29
Oxygen saturation, median (IQR) 96 (37) 96 (46) 96 (98) 0.15
Temperature, F, mean (SD) 99.1 (2.4) 99.4 (2.4) 99.1 (2.6) 0.68
Serum Creatinine, median (IQR) 1.37 (17.6) 1.46 (15.2) 1.5 (13.3) 0.82
White blood cell count, median (IQR) 14.9 (114.3) 14.9 (37.7) 13.7 (58.6) 0.041
Initial lactate, mmol/L, median (IQR) 4.0 (9.9) 4.1 (11.6) 4.4 (19.7) 0.002

Multidrug resistant organism status, N (%)

CRE 4 (2) 3 (3.2) 24 (3.1) 0.68
MRSA 14 (6.8) 12 (12.6) 55 (7.1) 0.14
Do-not resuscitate order (present), N (%) 15 (7.3) 6 (6.3) 57 (7.3) 0.94

Crude outcomes
Hospital mortality, number (%) 33 (16.1) 8 (8.4) 119 (15.3) 0.18
ICU mortality, number (%) 29 (18.1) 7 (10.0) 109 (17.1) 0.28
Hospital length of stay, median (IQR) 5 (68) 4 (40) 5 (88) 0.35
ICU length of stay, median (IQR) 1.8 (28.0) 1.7 (38.5) 2.0 (35.1) 0.17

BMI: Body mass index, COPD: chronic obstructive pulmonary disease, SBP: Systolic blood pressure, MAP: Mean arterial pressure, CRE: Carbapenem-resistant Enterobacteriaceae, MRSA:
Methicillin-resistant Staphylococcus aureus; ICU: Intensive care unit.

95% CI [1.03, 1.04]), history of cirrhosis (OR 1.64; 95% CI [1.24, 2.17]),
malignancy (OR 1.79; 95% CI [1.43, 2.25]), and dementia (OR 0.74;
95% CI [0.58, 0.96]), DNR status (OR 1.79; 95% CI [1.33, 2.41]), pneumo-
nia infection site (OR 1.43; 95% CI [1.17, 1.74]), initial lactate (OR 1.21;
95% CI [1.18, 1.25]), SBP (per 10 mm Hg) (OR 0.93; 95% CI [0.90,
0.96]), and temperature (OR 0.94; 95% CI [0.91, 0.97]). Sepsis interven-
tions that were significant independent variables associated with
mortality were central line placement (OR 2.47; 95% CI [1.93, 3.15])
and mechanical ventilation (OR 2.35; 95% CI [1.87, 2.95]). Model AUC
was 0.79. BMI grouping was not an independent predictor of mortality
in the adjusted analysis.
Table 4 shows the unadjusted, full, and reduced results of Model (2),
analyzing only obese patients (BMI ≥ 30). Although crude mortality was
not statistically different among fluid dosing strategies (p = 0.18) (8.4%,
15.3%, 16.1% in the adjusted BW, actual BW, and ideal BW strategies,
respectively), after adjusting for condition- and treatment-related vari-
ables using a hierarchical approach, dosing based on AdjBW was associ-
ated with improved mortality compared to ABW (OR 0.45; 95% CI [0.19,
1.07]) and IBW (OR 0.29; 95% CI [0.11,0.74]). Significant predictors of
mortality in obese patients included comorbid conditions of malignancy
(OR 1.88; 95% CI [1.16,3.24]) and cirrhosis (OR 1.95; 95% CI [1.16,3.24]),
age (OR 1.40; 95% CI [1.20,1.65]), pneumonia infection site (OR 1.99;
95% CI [1.30,3.04]), DNR status (OR 2.21; 95% CI [1.13,4.17]), central
line placement (OR 7.43; 95% CI [4.10,14.63]), triage oxygen saturation
(OR 0.97; 95% CI [0.95,0.99]), and triage lactate (OR 1.38; 95% CI
[1.29,1.48]). Fluid dosing strategy was not independently associated
with ICU mortality or hospital length of stay.
The linear regression model predicting volume of total fluid
delivered (mls/kg ABW) showed that BMI group was a significant
independent predictor (p b 0.0001), as were comorbid conditions
heart failure (p = 0.01) and chronic kidney disease (p = 0.008), triage
SBP (p = 0.0006), SI (p = 0.002), and lactate (p b 0.0001).
4. Discussion
In this study, the obesity paradox was eliminated after adjusting for
patient characteristics and sepsis interventions. A major finding in this
study is that, after adjusting for clinical factors that influence fluid admin-
istration, patients whose initial fluid resuscitation simulated dosing based
on an Adjusted Body Weight (IBW plus 40% of the difference of actual and
ideal body weights) experienced 55.2% lower odds of mortality compared
to those whose dosing simulated Actual Body Weight and 71.0% lower
odds of mortality compared to those whose dosing simulated Ideal Body
Weight. It seems improbable that physicians explicitly calculated the vol-
ume for AdjBW, but rather implicitly adjusted down the recommended
weight-based volume to a value that was greater than what would be
predicted for ideal body weight but less than actual body weight. This
hedge-your-bets strategy makes sense physiologically–blood volume
increases with increasing BMI, but not in a 1:1 relationship [22].
Another possibility is that fluid volume differences reflected con-
founding by indication, i.e., obese patients received less fluid because
they had fewer physiologic derangements that influence clinicians to
administer more fluid. However, there was no difference in initial
serum lactate level between BMI groups, and patients with higher BMI
 S.P. Taylor et al. / Journal of Critical Care 43 (2018) 7–12 11

Table 4
Predictors of mortality in obese patients with suspected shock.

Adjusted Model ORs (95% CI)

Predictor Model with dosing strategy alone (AUC 0.53) Full model (AUC 0.85) Reduced model with dosing strategy (AUC 0.84)
(n = 1080) (n = 1008) (n = 1013)

Dosing Strategy (ABW, AdjBW, IBW)

ABW vs AdjBW 0.511 (0.241, 1.081) 0.389 (0.144, 1.053) 0.448 (0.188, 1.065)
IBW vs AdjBW 0.479 (0.212, 1.082) 0.300 (0.111, 0.813) 0.289 (0.112, 0.740)

Demographics
Gender 1.222 (0.795, 1.878)
AgeR (per 10 yrs) 1.041 (1.023, 1.058) 1.400 (1.199, 1.645)

Comorbidities
Heart failure 1.093 (0.653, 1.829)
CKD 1.085 (0.628, 1.877)
CirrhosisR 1.924 (1.121, 3.300) 1.954 (1.164, 3.240)
COPD 0.808 (0.487, 1.341)
Diabetes 0.915 (0.592, 1.415)
MalignancyR 2.240 (1.364, 3.679) 1.884 (1.163, 3.022)
Dementia 0.890 (0.478, 1.654)

Triage clinical condition

Serum creatinine at triage 1.107 (0.998, 1.228)
Lactate at TRIAGER 1.351 (1.252, 1.457) 1.379 (1.287, 1.481)
WBC at triage 0.994 (0.975, 1.015)
SBP at triage (per 10 mm Hg) 0.936 (0.772, 1.118)
Temperature at triage 0.927 (0.857, 1.003)
O2 saturation at triager 0.982 (0.957, 1.008) 0.971 (0.950, 0.994)
MAP at triage (per 10 mm Hg) 0.909 (0.809, 1.015)
HR at triage (per 10 beats/min) 1.140 (0.945, 1.392)

Infection site
Pneumonia 1.638 (1.050, 2.556) 1.994 (1.303, 3.041)
UTI 0.919 (0.569, 1.485)
Skin and soft tissue 1.079 (0.534, 2.182)
C Diff 0.821 (0.291, 2.318)

Treatment
Mechanical ventilation status 1.548 (0.954, 2.514)
DNR statusR 1.933 (0.959, 3.897) 2.206 (1.134,4.165)
Central line placedR 5.015 (2.592, 9.701) 7.428 (4.095, 14.627)
3 hour bundle compliance goal met 1.073 (0.657, 1.752)
Fluid volume in 3 h (/actual BW) (per 5 mls/kg) 0.992 (0.974, 1.011)
Antibiotics received within 3 h from triage 0.564 (0.145, 2.194)
R
- p b 0.05 for inclusion in the reduced model
ABW: Actual Body Weight, AdjBW: Adjusted Body Weight, IBW: Ideal Body Weight. CHF: Congestive heart failure, CKD: Chronic kidney disease, COPD: chronic obstructive pulmonary
disease, WBC: White blood cell, SBP: Systolic blood pressure, MAP: Mean arterial pressure, DNR: Do-not-resuscitate.

had higher initial SBP than did lower BMI patients. Furthermore, BMI There are several important limitations to this study. Foremost, we
category remained independently associated with fluid volume at 3 h acknowledge that BMI is an imperfect measure of obesity. Nonetheless,
after adjusting for patient- and acuity-related covariates. it is the most widely used surrogate both clinically and in research.
Currently there are no robust data or even expert opinions to guide
dosing of initial fluid resuscitation for obese patients. Unfortunately,
obese patients comprise a significant proportion of the septic shock pop-
ulation — nearly one-third of patients treated through the septic shock
pathway in this study had BMI N 30 kg/m2. This represents a substantial
population for which there is clinical uncertainty surrounding an essen-
tial component of sepsis treatment. Previous studies speculated that a
conservative fluid strategy following an ideal-body-weight pattern
might have contributed to a survival advantage for obese patients [3,4].
In our study, fluid administration following a pattern of adjusted body-
weight was associated with improved mortality. Given the potential
harms associated with both persistent hypovolemia [23] and overzealous
volume administration [15-17], further study into the optimal fluid dos-
ing strategy for obese patients is imperative. Until then, performance as-
sessments of compliance with initial fluid resuscitation goals might
consider using ideal or adjusted weight for fluid goals to avoid penalizing
physicians for appropriately under-dosing obese patients. Fig. 1 illustrates
the effect of different interpretations of the weight-based fluid goal on
compliance in obese patients. In the extreme example of obesity class
III, the proportion of cases considered compliant with the quality metric
of 30 cm3/kg administered within 3 h decreases from 75% to 50% when Fig. 1. Proportion of cases compliant with 30 cm3/kg fluid goal based on different weight
calculated based on ideal versus actual body weight, respectively. calculations.
12 S.P. Taylor et al. / Journal of Critical Care 43 (2018) 7–12
Future studies may consider more precise measures of obesity; such as
sagittal abdominal diameter [23], or adipose tissue quantification by
computed tomography scans [24]. Additionally, the population studied
included patients with suspected septic shock rather than confirmed
septic shock. Certainly, a proportion of these patients were ultimately
revealed to have alternate diagnoses. However, we felt that using a pop-
ulation of patients selected for treatment via a septic shock treatment
pathway was the most clinically relevant to assess early fluid resuscita-
tion strategies. Other limitations are those inherent in a retrospective
study, including inaccuracies in source data and lack of access to certain
useful information such as type of nutritional support, microbiological
data, and fluid administered beyond the initial resuscitation period.
5. Conclusion
Despite standardized protocols for weight-based fluid administra-
tion, we found differences in initial fluid administration among BMI cat-
egories in patients with suspected septic shock. Patients whose dosing
strategy mimicked dosing based on adjusted body weight appeared to
have improved mortality compared to other dosing strategies. The opti-
mal fluid dosing strategy for obese patients should be a focus of future
prospective research.
Funding
No financial support was provided for the study
References
[1] Oliveros H, Villamor E. Obesity and mortality in critically ill adults: a systematic re-
view and meta-analysis. Obesity 2008:515–21.
[2] Hutagalung R, Marques J, Kobylka K, et al. The obesity paradox in surgical intensive
care unit patients. Intensive Care Med 2011;37:17939.
[3] Wacharasint P, Boyd JH, Russell, et al. One size does not fit all in severe infection:
obesity alters outcome, susceptibility, treatment, and inflammatory response. Crit
Care 2013;17:R122.
[4] Arabi YM, Dara SI, Tamim HM, et al. Clinical characteristics, sepsis interventions and
outcomes in the obese patients with septic shock: an international multicenter co-
hort study. Crit Care 2013;17:R72.
[5] Pepper DJ, Sun J, Welsh J, et al. Increased body mass index and adjusted mortality in
ICU patients with sepsis or septic shock: a systematic review and meta-analysis. Crit
Care 2016;15(20):181.
[6] Kershaw EE, Flier JS. Adipose tissue as an endocrine organ. J Clin Endocrinol Metab
2004;89:2548–56.
[7] Bornstein SR, Licinio J, Tauchnitz R, et al. Plasma leptin levels are increased in survi-
vors of acute sepsis: associated loss of diurnal rhythm, in cortisol and leptin secre-
tion. J Clin Endocr Metab 1998;83:280–3.
[8] Stapleton RD, Dixon AE, Parsons PE, et al. The association between BMI and plasma
cytokine levels in patients with acute lung injury. Chest 2010;138:568577.
[9] Thompson PA, Kitchens RL. Native high-density lipoprotein augments monocyte re-
sponses to lipopolysaccharide (LPS) by suppressing the inhibitory activity of LPS-
binding protein. J Immunol 2006;177:4880–7.
[10] Robinson MK, Mogensen KM, Casey JD, et al. The relationship among obesity, nutri-
tional status, and mortality in the critically ill. Crit Care Med 2015;43:87100.
[11] Nasraway Jr SA, Hudson-Jinks TM, Kelleher RM. Multidisciplinary care of the obese
patient with chronic critical illness after surgery. Crit Care Clin 2002;17:643657.
[12] El-Sohl AA. Clinical approach to the critically ill, morbidly obese patient. Am J Respir
Crit Care Med 2004;17:557561.
[13] Rhodes A, Evans L, Alhazzani W, et al. Surviving sepsis campaign: international
guidelines for management of sepsis and septic shock: 2016. Crit Care Med 2017;
45:381–565.
[14] Adams C, Tucker C, Allen B, et al. Disparities in hemodynamic resuscitation of the
obese critically ill septic shock patient. J Crit Care 2017;37:219–23.
[15] Boyd JH, Forbes J, Nakada TA, et al. Fluid resuscitation in septic shock: a positive fluid
balance and elevated central venous pressure are associated with increased mortal-
ity. Crit Care Med 2011;39:259265.
[16] Marik PE, Linde-Zwirble WT, Bittner EA, Sahatjian J, Hansell D. Fluid administration
in severe sepsis and septic shock, patterns and outcomes: an analysis of a large na-
tional database. Intensive Care Med 2017 Jan;27 [Epub ahead of print].
[17] Maitland K, Kiguli S, Opoka RO, et al. Mortality after fluid bolus in African children
with severe infection. N Engl J Med 2011;364:2483–95.
[20] Devine BJ. Gentamicin Therapy. DICP, 8; 1974 6505.
[21] Winter ME. Basic Pharmacokinetics. London: Lippincott Williams and Williams;
2004.
[22] Nelson J, Billeter AT, Seifert B, et al. Obese trauma patients are at increased risk of
early hypovolemic shock: a retrospective cohort analysis of 1084 severely injured
patients. Crit Care 2012;16:R77.
[23] Paolini JB, Mancini J, Genestal M, et al. Predictive value of abdominal obesity vs. body
mass index for determining risk of intensive care unit mortality. Crit Care Med 2010;
38:1308–14.
[24] Pisitsak C, Lee JG, Boyd JH, et al. Increased ratio of visceral to subcutaneous adipose
tissue in septic patients is associated with adverse outcomes. Crit Care Med 2016;44:
1966–73.