Epidemiology

Relation Between Body Mass Index, Waist Circumference,

and Death After Acute Myocardial Infarction
Marianne Zeller, PhD; Philippe Gabriel Steg, MD, PhD; Jack Ravisy, MD; Luc Lorgis, MD;
Yves Laurent, MD; Pierre Sicard, PhD; Luc Janin-Manificat, MD; Jean-Claude Beer, MD;
Hamid Makki, MD; Anne-Ccile Lagrost, MSc; Luc Rochette, PharmD, PhD;
Yves Cottin, MD, PhD; for the RICO Survey Working Group

BackgroundAn elevated body mass index (BMI) has been reported to be associated with a lower rate of death after acute
myocardial infarction (AMI). However, waist circumference (WC) may be a better marker of cardiovascular risk than
BMI. We used data from a contemporary French population-based cohort of patients with AMI to analyze the impact
of WC and BMI on death rates.
Methods and ResultsWe evaluated 2229 consecutive patients with AMI. Patients were classified according to BMI as
normal, overweight, obese, and very obese (BMI 25, 25 to 29.9, 30 to 34.5, and 35 kg/m2, respectively) and as
increased waistline (WC 88/102 cm for women/men) or normal. Half of the patients were overweight (n1044), and
one quarter were obese (n397) or very obese (n128). Increased WC was present in half of the patients (n1110).
Increased BMI was associated with a reduced death rate, with a 5% risk reduction for each unit increase in BMI (hazard
ratio, 0.95; 95% CI, 0.93 to 0.98; P0.001). In contrast, WC as a continuous variable had no impact on all-cause death
(P0.20). After adjustment for baseline predictors of death, BMI was not independently predictive of death. The group
of patients with high WC but low BMI had increased 1-year death rate.
ConclusionsNeither BMI nor WC independently predicts death after AMI. Much of the inverse relationship between
BMI and the rate of death after AMI is due to confounding by characteristics associated with survival. This study
emphasizes the need to measure both BMI and WC because patients with a high WC and low BMI are at high risk of
death. (Circulation. 2008;118:482-490.)
Key Words: body mass index death myocardial infarction obesity

O besity is a major risk factor for the development of fatal

and nonfatal cardiovascular (CV) events.1 In patients
with established CV disease, its impact on risk is less clear.
body mass and may thus help to explain the controversy
known as the obesity paradox.9 Abdominal obesity, which
reflects central body fat distribution, has been suggested as a
Increased body mass index (BMI) is associated with a higher better marker of the obesity risk.10
risk of acute myocardial infarction (AMI).2 Several studies,
however, have reported an apparent paradoxical effect of Editorial p 469
BMI on outcomes: In patients undergoing elective percuta- Clinical Perspective p 490
neous coronary intervention (PCI), BMI was inversely asso- Waist circumference (WC) is an anthropometric index
ciated with worse outcomes, with obese and overweight usually considered a surrogate marker of abdominal fat mass
patients having improved survival compared with those with (subcutaneous and intraabdominal).11 High WC values are
normal BMI.3,4 In patients with AMI, high BMI appears to associated with CV complications, with a predictive value
have an unexplained protective effect on survival.57 The role superior to that of BMI.12,13 Despite the high prevalence of
of increased BMI as an independent CV risk factor is abdominal obesity in patients admitted with AMI,14 there is a
controversial.8 In patients with coronary artery disease, BMI paucity of data on its impact on prognosis after the acute event.
does not adequately discriminate between body fat and lean Whether increased waistline is involved in the obesity paradox

Received November 19, 2007; accepted May 16, 2008.

From the Laboratory of Experimental Cardiovascular Pathophysiology and Pharmacology, IFR sant-STIC, Faculty of Medicine, University of
Burgundy, Dijon, France (M.Z., P.S., L.R.); INSERM U-698, Universite Paris VII, AP-HP, Centre Hospitalier Bichat-Claude Bernard, France (P.G.S.);
Service de Cardiologie, Clinique de Fontaine, Fontaine les Dijon, France (J.R.); Service de Cardiologie, CHU Bocage, Dijon, France (J.-C.B., A.-C.L.,
L.L., Y.C.); Service de Cardiologie, Centre Hospitalier, Semur en Auxois, France (Y.L.); Service de Cardiologie, Centre Hospitalier, Beaune, France
(L.J.-M.); and Service de Cardiologie, Centre Hospitalier, Chtillon sur Seine, France (H.M.).
The online Data Supplement, which contains a supplemental Methods section, can be found with this article at http://circ.ahajournals.org/cgi/
content/full/CIRCULATIONAHA.107.753483/DC1.
Correspondence to Dr Marianne Zeller, Laboratory of Experimental and Cardiovascular Physiopathology and Pharmacology, IFR sant-STIC, Faculty
of Medicine, 7 Bd Jeanne dArc, 21000 Dijon, France. E-mail [email protected]
2008 American Heart Association, Inc.
Circulation is available at http://circ.ahajournals.org DOI: 10.1161/CIRCULATIONAHA.107.753483

482
Downloaded from http://circ.ahajournals.org/ at UNIVERSITY OF BIRMINGHAM on May 29, 2015
 Zeller et al Obesity in Acute Myocardial Infarction 483

remains to be determined. Thus, from a contemporary prospec- tions between continuous variables. The cumulative incidence of
tive study of unselected consecutive patients, we compared the all-cause death was estimated according to the KaplanMeier
method, and the log-rank test was used to evaluate differences
prognostic impact of obesity, assessed by both BMI and WC, on
between groups. Cox regression analysis was performed to deter-
short- and intermediate-term mortality after AMI. mine the effects of WC or BMI as continuous variables on the rate
of death in unadjusted models (models 1 and 2, respectively). BMI
Methods was then tested by multivariate analysis. Age, sex (female as
reference), and NT-proBNP were individually tested as covariates
Patients given the strong correlation between BMI and age or NT-proBNP
The design and methods of the Observatoire des Infarctus de Cte and because age and NT-proBNP are major contributors to prognosis
dOr (RICO) Survey have been published.14 Patients presenting with after MI. The other variables tested in univariate analysis were those
diagnosed AMI15 were included in the present study (see the online known to potentially affect the outcome after MI and variables
Data Supplement). The present study complied with the Declaration showing a variation according to BMI tertiles. Overall, the variables
of Helsinki and was approved by the ethics committee of University tested were the following: age, female gender, diabetes, hyperten-
Hospital of Dijon. Each patient gave written consent before sion, dyslipidemia, smoking, prior MI, acute therapies (statin,
participation. -blockers, angiotensin-converting enzyme inhibitors), biological
parameters (NT-proBNP, C-reactive protein, creatinine, high-density
Data Collection lipoprotein cholesterol, and triglycerides), Killip class I, ST-
Data on demographics, CV risk factors (history of hypertension, elevation myocardial infarction (STEMI), and LVEF. Among these,
diabetes, treated hyperlipidemia, current smoking), and prior myocardial age, sex, NT-proBNP, acute therapy, Killip class I, prior MI,
infarction were collected prospectively, along with admission charac- hypertension, diabetes, hyperlipidemia, smoking, C-reactive protein,
teristics and hemodynamic parameters. Blood samples were drawn at STEMI, LVEF, and creatinine were predictors of prognosis and
admission. Plasma creatinine levels were measured on a Vitros 950 therefore were included as covariates in multivariate analysis.
analyzer (Ortho Clinical Diagnostics, Rochester, NY). High-sensitivity Because smoking is a potential confounder of the relationship
C-reactive protein was measured on Dimension Xpand (Dade Behring, between BMI and the outcome, its interaction was tested in univar-
Newark, Neb) with an immunonephelometry assay. Plasma N-terminal iate analysis and introduced as a covariate in multivariate analysis.
pro B-type natriuretic peptide (NT-proBNP) was determined by ELISA Linearity of the continuous variables with respect to the response
with an Elecsys NT-proBNP sandwich immunoassay on Elecsys 2010 variable was assessed by determining the quartile of their distribu-
(Roche Diagnostics, Basel, Switzerland). Overnight fasting blood sam- tion. Subsequently, hazard ratios (HRs) for each quartile were
ples were collected on the morning after admission for blood lipid calculated. All variables showed a linear trend in the estimated HRs
measurements. High-density lipoprotein cholesterol and triglyceride and thus were introduced into the model as continuous. The
concentrations were measured on a Dimension analyzer (Dade Be- proportional-hazards assumption in the Cox models was assessed
hring). The level of low-density lipoprotein cholesterol was calculated with graphical methods (log-log plots) and with models including
from the Friedewald formula. time-by-covariate interactions. No violations of the proportional-
Echocardiography was performed at day 31 by a local investi- hazards assumption were identified. Statistical analyses were per-
gator according to the Simpson method using the apical views to formed with SPSS software (SPSS, Inc, Chicago, Ill).
calculate left ventricular ejection fraction (LVEF). Data on acute The authors had full access to and take full responsibility for the
(24 hour) reperfusion procedures (thrombolysis or PCI) were integrity of the data. All authors have read and agree to the
collected. Heart failure, defined as rales over more than half of the manuscript as written.
lung field (Killip class II), pulmonary edema (Killip class III), or
cardiogenic shock (Killip class IV), was evaluated on admission. Results
Duration of hospital stay in the coronary care unit also was collected.
After hospital discharge, 30-day or 1-year information was acquired Study Population
by contacting either each patient individually, their relatives, or A total of 2229 patients were included in the study. Fewer
treating physician and by reviewing the hospital records if the patient than one third (n660) had a normal BMI; half of the patients
had been rehospitalized. No patient was lost to follow-up (see the were overweight (46.8%, n1044); and one quarter were
Data Supplement).
obese (n397) or very obese (n128). Half of the patients
Group Definition (49.8%, n1110) had increased WCs.
Anthropometric parameters were measured within 48 hours of The patients characteristics are summarized in Table 1 for
admission. BMI was categorized according to the World Health women and Table 2 for men. Time to admission, MI location,
Organization16 standards as normal (25 kg/m2), overweight (25 to and LVEF were similar across BMI subgroups for both sexes.
29.9 kg/m2), obese (30 to 34.9 kg/m2), and very obese (35 kg/m2). In elevated WC tertiles, heart failure as assessed by Killip
WC was measured with a nonelastic tape at the mid distance between
the top of the iliac crest and the bottom of the rib cage and as the
class on admission was more frequent in men. For both sexes,
average of 1 measurement taken after inspiration and 1 taken after elevated systolic blood pressure was found across the BMI or
expiration. Increased WC was defined using the National Cholesterol WC tertiles. In men, acute PCI was used less frequently in the
Education Program Adult Treatment Panel III (NCEP ATP) cutoff of high BMI groups, whereas statins and angiotensin-converting
WC 102 cm in men and 88 cm in women.17,18 Data for both BMI enzyme inhibitors were prescribed more frequently.
and WC were split into sex-specific tertiles for analysis.
C-reactive protein, creatinine, and triglyceride levels mark-
Statistical Analysis edly increased with elevated WC, whereas high-density
Data are presented as median (25th to 75th percentile) or meanSD lipoprotein cholesterol decreased. In women, NT-proBNP
as appropriate or as proportion. For continuous variables, a levels showed no significant changes across anthropometric
Kolmogorov-Smirnov analysis was performed to test for normality. categories, whereas in men, the propeptide levels were
For the tests across tertiles, we performed the Kruskal-Wallis 1-way reduced across BMI tertiles.
ANOVA by rank for nonnormally distributed values or 1-way
ANOVA for normally distributed values. Qualitative variables, BMI as a continuous variable was inversely correlated with
expressed as numbers and percents, were compared by the 2 test for age and NT-proBNP (r0.13, P0.001 and r0.11,
trends. Spearmans rank correlation was applied to test for associa- P0.001). In contrast, WC as a continuous variable was
Downloaded from http://circ.ahajournals.org/ at UNIVERSITY OF BIRMINGHAM on May 29, 2015
484 Circulation July 29, 2008

Table 1. Characteristics of the Study Population According to BMI and WC on Admission in Women (n593)
BMI Tertiles, kg/m WC Tertiles, cm

1 2 3 P 1 2 3 P
Median (25th75th) 21 (2023) 1524 26 (2527) 2428 31 (2934) 2850 82 (1688) 6093 98 (95101) 93104 112 (108120) 104165
minimummaximum
Risk factors
Age, y* 77 (6583) 76 (7082) 72 (6079) 0.001 73 (5882) 77 (6982) 74 (6481) 0.009
Current smoking, 34 (17.2) 18 (9.1) 29 (14.7) 0.06 36 (18.2) 25 (12.6) 20 (10.2) 0.06
n (%)
Prior MI, n (%) 24 (12.1) 26 (13.1) 24 (12.2) 0.94 29 (14.6) 20 (10.0) 25 (12.7) 0.39
Hypertension, 113 (57.1) 129 (65.2) 148 (75.1) 0.001 104 (52.5) 135 (68.2) 151 (76.6) 0.001
n (%)
Diabetes, n (%) 34 (17.2) 46 (23.2) 74 (37.6) 0.001 31 (15.7) 39 (19.7) 84 (42.6) 0.001
Dyslipidemia, 67 (33.8) 105 (53.0) 96 (48.7) 0.001 79 (39.9) 96 (48.5) 93 (47.2) 0.18
n (%)
Clinical data
Time from 225 (117600) 210 (92475) 210 (110582) 0.44 180 (90535) 215 (111540) 240 (120571) 0.08
symptom onset to
admission, min*
Killip class I on 59 (29.8) 52 (26.3) 55 (27.9) 0.73 55 (27.8) 51 (25.8) 60 (30.5) 0.58
admission, n (%)
Heart rate, bpm* 80 (68100) 78 (6895) 80 (6995) 0.64 80 (68100) 78 (6792) 80 (6897) 0.38
Systolic blood 130 (110157) 140 (120160) 144 (129168) 0.002 130 (112158) 140 (120165) 140 (125162) 0.009
pressure, mm Hg*
Diastolic blood 76 (6690) 76 (6790) 80 (7090) 0.25 77 (6890) 80 (7090) 78 (6690) 0.63
pressure, mm Hg*
STEMI, n (%) 116 (58.6) 108 (54.5) 101 (51.3) 0.34 117 (59.1) 105 (53.0) 103 (52.3) 0.33
Anterior wall 89 (44.9) 95 (48.0) 73 (37.1) 0.08 89 (44.9) 85 (42.9) 83 (42.1) 0.84
location, n (%)
LVEF, %* 55 (40-65) 51 (43-60) 55 (45-63) 0.24 55 (43-65) 55 (45-63) 50 (42-62) 0.31
Biological data
hs-CRP, mg/L* 4.2 (1.69.8) 5.5 (2.211.8) 7.1 (3.612.7) 0.001 4.0 (1.48.3) 5.7 (2.311.8) 7.5 (3.715.1) 0.001
Creatinine, 80 (6798) 82 (6999) 80 (6698) 0.81 78 (6594) 82 (7199) 84 (67106) 0.015
mol/L*
NT-proBNP, 3002 (8848907) 1871 (6946207) 1712 (6265812) 0.05 2067 (6336026) 2002 (7327200) 2272 (7847465) 0.79
pg/mL*
LDL-C, mg/dL 12235 12339 12541 0.79 12436 12637 12142 0.42
HDL-C, mg/dL* 45 (3856) 44 (3655) 38 (3150) 0.001 47 (3956) 42 (3555) 38 (3149) 0.001
Triglycerides, 107 (77138) 109 (86148) 136 (96190) 0.001 100 (71132) 113 (86151) 138 (97185) 0.001
mg/dL*
Acute treatments
48 h, n (%)
Primary PCI 36 (18.2) 38 (19.2) 29 (14.7) 0.47 44 (22.2) 27 (13.6) 32 (16.2) 0.07
Thrombolysis 25 (12.6) 36 (18.2) 25 (12.7) 0.20 33 (16.7) 29 (14.6) 24 (12.2) 0.45
-Blocker 138 (69.7) 152 (76.8) 144 (73.1) 0.28 147 (74.2) 150 (75.8) 137 (69.5) 0.35
ACE inhibitor 116 (58.6) 123 (62.1) 117 (59.4) 0.75 123 (62.1) 119 (60.1) 114 (57.9) 0.69
Statin 136 (68.7) 141 (71.2) 153 (77.7) 0.12 141 (71.2) 143 (72.2) 146 (74.1) 0.81
Duration of stay 4 (36) 4 (35) 4 (36) 0.36 4 (36) 4 (36) 4 (36) 0.41
in CCU, d*

hs-CRP indicates high-sensitivity C-reactive protein; LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol; ACE, angiotensin-
converting enzyme; and CCU, coronary care unit.
*Median (25th to 75th percentile).

positively correlated with age (r0.07, P0.001) but not of 346 days (25th to 75th percentile, 338 to 354 days).
with NT-proBNP (r0.03, P0.155). See the online Data Sex-specific death rates were reported in an additional table
Supplement for sex-specific correlations. in the Data Supplement. The rate of death at 1 year was
significantly reduced in men with increasing BMI tertiles
Outcomes (P0.016). No significant variation in the rate of death was
At 1 year, there were 253 CV deaths (11.4%) and 301 found across WC tertiles for both sexes. These findings were
all-cause deaths (13.5%), including 30-day CV (167 [7.5%]) confirmed by KaplanMeier analysis of cumulative survival
and all-cause (186 [8.3%]) deaths during a median follow-up curves. Death rates across BMI tertiles were not significant in
Downloaded from http://circ.ahajournals.org/ at UNIVERSITY OF BIRMINGHAM on May 29, 2015
 Zeller et al Obesity in Acute Myocardial Infarction 485

Table 2. Characteristics of the Study Population According to BMI and WC on Admission in Men (n1636)
BMI Tertiles, kg/m WC Tertiles, cm

T1 T2 T3 P T1 T2 T3 P
Median (25th75th) 24 (2224) 1625 27 (2627) 2528 31 (2933) 2849 89 (8492) 5695 100 (97102) 95-105 111 (108118) 105158
minimummaximum
Risk factors
Age, y* 67 (5477) 67 (5576) 61 (5173) 0.001 62 (5174) 66 (5475) 66 (5575) 0.002
Current smoking, 208 (38.2) 153 (28.1) 158 (28.9) 0.001 212 (38.9) 160 (29.4) 147 (26.9) 0.001
n (%)
Prior MI, n (%) 88 (16.1) 87 (16.0) 79 (14.5) 0.70 69 (12.7) 96 (17.6) 89 (16.3) 0.06
Hypertension, 223 (40.9) 279 (51.2) 323 (59.2) 0.001 202 (37.1) 274 (50.3) 349 (63.9) 0.001
n (%)
Diabetes, n (%) 93 (17.1) 100 (18.3) 168 (30.8) 0.001 77 (14.1) 108 (19.8) 176 (32.2) 0.001
Dyslipidemia, 222 (40.7) 263 (48.3) 285 (52.2) 0.001 229 (42.0) 266 (48.8) 275 (50.4) 0.013
n (%)
Clinical data
Time from 180 (90420) 180 (90492) 180 (101481) 0.46 180 (90420) 180 (95500) 195 (100487) 0.31
symptom onset to
admission, min*
Killip class I on 101 (18.5) 103 (18.9) 103 (18.9) 0.98 76 (13.9) 107 (19.6) 124 (22.7) 0.001
admission
Heart rate, bpm* 75 (6290) 75 (6489) 77 (6790) 0.09 73 (6187) 75 (6490) 78 (6790) 0.001
Systolic blood 130 (116152) 139 (121160) 140 (125165) 0.001 131 (116150) 138 (120160) 141 (126167) 0.001
pressure, mm Hg*
Diastolic blood 80 (7090) 80 (7092) 80 (7095) 0.001 80 (7090) 80 (7093) 80 (7093) 0.015
pressure, mm Hg*
STEMI, (%) 331 (60.7) 334 (61.3) 296 (54.2) 0.031 338 (62.0) 328 (60.2) 295 (54.0) 0.019
Anterior wall 201 (36.9) 198 (36.3) 191 (35.0) 0.80 208 (38.2) 189 (34.7) 193 (35.3) 0.44
location , (%)
LVEF, %* 55 (4564) 55 (4364) 55 (4563) 0.48 55 (4565) 54 (4363) 55 (4463) 0.09
Biological data
hs-CRP, mg/L* 3.4 (1.610.1) 3.8 (1.67.2) 4.5 (2.210.0) 0.015 3.2 (1.58.0) 3.8 (1.78.2) 4.7 (2.310.2) 0.001
Creatinine, 90 (80111) 94 (80115) 96 (80108) 0.30 89 (80106) 94 (80115) 98 (80115) 0.001
mol/L*
NT-proBNP, 826 (2122962) 766 (2292589) 624 (1692042) 0.008 629 (1662134) 735 (1922683) 832 (2432651) 0.05
pg/mL*
LDL-C, mg/dL 12238 12138 12336 0.84 12737 12037 11936 0.006
HDL-C, mg/dL* 41 (3451) 40 (3148) 35 (2744) 0.001 41 (3351) 38 (3148) 36 (2845) 0.001
Triglycerides, 100 (73135) 109 (83159) 145 (100222) 0.001 99 (74145) 116 (82162) 136 (94206) 0.001
mg/dL*
Acute treatments
48 h, (%)
Primary PCI 118 (21.7) 84 (15.4) 91 (16.7) 0.018 111 (20.4) 95 (17.4) 87 (15.9) 0.15
Thrombolysis 132 (24.2) 134 (24.6) 132 (24.2) 0.98 141 (25.9) 141 (25.9) 116 (21.2) 0.12
-Blocker 431 (79.1) 436 (80.0) 454 (83.2) 0.20 451 (82.8) 438 (80.4) 432 (79.1) 0.30
ACE inhibitor 327 (60.0) 319 (58.5) 384 (70.3) 0.001 335 (61.5) 330 (60.6) 365 (66.8) 0.07
Statin 417 (76.5) 445 (81.7) 466 (85.3) 0.001 428 (78.5) 444 (81.5) 456 (83.5) 0.11
Duration of stay 4 (35) 4 (36) 4 (35) 0.13 4 (35) 4 (35) 4 (35) 0.73
in CCU, d*

Abbreviations as in Table 1.
*Median (25th to 75 percentile).

women (Figure 1A) (P0.158). In men, a lower death rate impact on all-cause death in the whole study population (HR,
was observed with increasing BMI tertiles (log-rank 1.00; 95% CI, 1.00 to 1.01; P0.20) (Figure 3, model 1), men
P0.019) (Figure 1B). No increased risk was found across alone (HR, 1.00; 95% CI, 1.00 to 1.01; P0.20), or women
WC tertiles in either women (Figure 2A) (P0.321) or men alone (HR, 1.01; 95% CI, 1.00 to 1.02; P0.11). Multivariate
(Figure 2B) (P0.547). When analyzed as a continuous analysis showed that when age (model 3), sex (model 4),
variable, BMI also was significantly associated with death, NT-proBNP (model 5), or covariates that had a significant
with a 5% reduction in the risk of death for each unit increase impact on outcomes (model 6) were added to the models,
in BMI (HR, 0.95; 95% CI, 0.93 to 0.98; P0.001) (Figure 3, BMI was not an independent predictor of outcome (Figure 3),
model 2). In contrast, WC as a continuous variable had no Moreover, no significant interaction between C-reactive pro-
Downloaded from http://circ.ahajournals.org/ at UNIVERSITY OF BIRMINGHAM on May 29, 2015
486 Circulation July 29, 2008

Figure 1. KaplanMeier survival curves for BMI tertiles (T) in

women (A) (n593) and men (B) (n1636). Figure 2. KaplanMeier survival curves for WC tertiles (T) in
women (A) (n593) and men (B) (n1636).

tein and either BMI (P0.53) or WC (P0.61) was found for

low (inframedian) and high (supramedian) BMI groups. The
outcome.
median WC was 100 cm (25th to 75th percentile, 96 to 103
Stratification of death rates according to both BMI and WC
cm) for both groups. The median BMI was 24 kg/m2 (25th to
tertiles showed that for both sexes, a high WC is associated
with increased rate of death in the 2 lowest BMI tertiles, 75th percentile, 23 to 25 kg/m2) for the low BMI and 27 kg/m2
which identified a subgroup of high-risk patients (Figure 4A (25th to 75th percentile, 27 to 28 kg/m2) for the high BMI
and 4B). In male subjects, a high WC but low BMI is group. In this new data set, age and NT-proBNP were
observed in almost one quarter of patients (2 of 9) and is significantly different in the low versus high BMI group (75
associated with a high death rate (almost 1 in 5) (Figure 4B). years [25th to 75th percentile, 65 to 81 years] versus 67 years
To further investigate the capacity of BMI to predict death [25th to 75th percentile, 54 to 76 years], P0.001; and 1781
independently of WC, we performed a subgroup analysis on pg/mL [25th to 75th percentile, 431 to 6308 pg/mL] versus
832 waist-matched patients comparing the outcome of low 872 [25th to 75th percentile, 190 to 2666 pg/mL], P0.001,
BMI (n416) and high BMI (n416). The classification of respectively). In addition, as expected, the rate of death
BMI was based on a median BMI value (26 kg/m2) defining remained markedly increased in the low versus high BMI
Downloaded from http://circ.ahajournals.org/ at UNIVERSITY OF BIRMINGHAM on May 29, 2015
 Zeller et al Obesity in Acute Myocardial Infarction 487

Models Hazard ratios (95%CI) for 1 year mortality

Per cm increase
Model 1: WC (unadjusted) 1.00(1.00-1.01) 0.20
Per kg/m increase
Model 2: BMI (unadjusted) 0.95(0.93-0.98) <0.001

Model 3: model 2+age 0.98(0.96-1.01) 0.21

Model 4: model 3+female 0.98(0.95-1.01) 0.18

Model 5: model 4+NT-proBNP 1.02(0.99-1.05) 0.18

1.01(0.97-1.06) 0.60
Model 6: model 5+covariates*

0.90 0.95 1.00 1.05 1.10

1 year mortality decreased 1 year mortality increased

* Covariates : acute therapy, Killip>I, prior MI, hypertension, diab etes, hyperlipidemia, smoking, CRP, STEMI, LVEF,
creatinine, BMI*smoking.

Figure 3. HRs for each unit increase in BMI (per kg/m2) or WC (per cm) for 1-year death rates in unadjusted and adjusted models.

group (91 [22%] versus 55 [13%], respectively; P0.001). Discussion

Analysis of this subgroup of patients matched for WC In the present study, based on a large, unselected cohort of
confirmed that for the same WC, BMI was still relevant to AMI patients, neither BMI nor WC independently predicted
discriminate patients at high risk. death after acute MI. Much of the apparent obesity paradox

A 31,8%

One-year 24,2%
mortality
24,2%
30%
15,4%
25%
15,2%
20%
10,6%
13,6%
15%
10,6%
10% 7,6%
T3
5% T2
iles
0% T1 t ert
T1 T2 T3 WC
BMI tertiles Figure 4. Stratification of 1-year death
rates according to both BMI and WC ter-
B tiles in women (A) and men (B).
One-year
mortality 20,9%
30% 18,2%

25% 12,7%

20%
10,4% 11,5%
15% 11,5%
10,4% 6,6%
10% 7,7%
T3
5% T2
s
er tile
Ct
0% T1
W
BMI tertiles

Downloaded from http://circ.ahajournals.org/ at UNIVERSITY OF BIRMINGHAM on May 29, 2015

488 Circulation July 29, 2008

(the inverse relationship between BMI and death after MI) is was performed. However, the data at 30 days, although not
related to confounding by differences in baseline character- significant (P0.096), showed the same trend as the 1-year
istics that predict survival after MI. Moreover, we identified follow-up, ie, a decrease in the rate of death with increasing
a group of high-risk patients, ie, those with a high WC and a BMI. Our findings are confirmed by major contemporary
low BMI, compared with patients with both high WC and studies in acute coronary syndromes that address the differ-
BMI values. ential impact of BMI-based obesity at the acute phase (ie, in
hospital) versus midterm (ie, 6 months) or 1 year.6,7,21 These
BMI-Based Obesity studies found the same trend of obesity whatever the time
In this contemporary, population-based French registry of delay of follow-up. Overall, these data strongly suggest that
patients with AMI, the prevalence of overweight and obesity the apparent obesity paradox is already present at the early
was very high (70%), reflecting the epidemic nature of phase of an acute MI.
obesity in Western Europe. This finding underlines the In acute MI, the level of NT-proBNP, an indicator of the
importance of obesity as a CV risk factor and underscores the hemodynamic severity of MI, as well as systolic and/or
need to improve our understanding of the relationship be- diastolic dysfunction, peaks at the time of admission and is
tween excess weight and CV outcomes. We found a marked considered one of the most powerful predictors of death.24 26
association between increasing BMI and younger age in men Low proBNP levels in patients with high BMI, with a pattern
that was even stronger in women. This association, which similar to that of heart failure, may therefore participate in the
illustrates the premature occurrence of AMI, has been re- more favorable outcome reported in the present study.27
ported in previous studies.57,19 21 In accordance with our Surprisingly, low propeptide levels, similar time to admis-
results, an inverse independent relationship between NT- sion, and LVEF were not associated with lower hemodynam-
proBNP and lean mass, rather than fat mass, has been ic severity in patients with high BMI. These findings suggest
demonstrated in the Framingham Heart Study and the Dallas that hemodynamic severity and left ventricular dysfunction
Heart Study.22,23 The mechanisms linking low natriuretic may not be the only major contributors of admission propep-
peptide levels and obesity remain to be elucidated but tide levels in patients with a high BMI.
probably involve decreased release of natriuretic peptides
from the heart rather than increased clearance.23 Excess WC-Based Obesity
weight was associated with an elevated incidence of the WC is a surrogate measure of abdominal fat, with which it
obesity-associated CV risk factors of hypertension, diabetes, has been strongly correlated by computed tomography or
and dyslipidemia, also consistent with a high-risk profile. magnetic resonance imaging.28 High WC, as a component of
Rates of primary PCI were decreased in higher BMI catego- metabolic syndrome and insulin resistancerelated disorders,
ries, a finding that differs from the results of a randomized is associated with CV death.29 In healthy subjects, WC is a
trial showing a more aggressive use of PCI in patients with better predictor of acute coronary events than BMI.30 Para-
higher BMI.20 The use of acute medications was similar doxically, few studies have analyzed abdominal obesity in the
across the BMI subgroups, except for statins and angioten- setting of AMI. We found that an elevated WC (as defined by
sin-converting enzyme inhibitors, which were prescribed the NCEP ATP III threshold) was very frequent in patients
more often in patients with a high BMI, a group with a higher with AMI because it is present in half of the patients. In
prevalence of dyslipidemia at baseline. patients with CV disease, a high prevalence of abdominal
obesity also was reported.31 One of the major findings of our
Obesity Paradox study is that, in contrast to the strong negative relationship
An apparent protective effect of high BMI on mortality has observed between BMI and age, WC was poorly but posi-
been found in randomized trials of patients with unstable tively correlated with age. These findings suggest a weaker
angina and non-STEMI,5,7 STEMI,6,19,21 or both.20 In con- impact of WC as a risk factor for AMI than BMI. CV risk has
trast, our data, from a registry study of unselected patients, been shown to derive more from abdominal obesityassoci-
showed that after adjustment, BMI was not a prognostic ated risk factors (high blood pressure, diabetes, and dyslip-
factor after myocardial infarction, a finding that is consistent idemia) than from abdominal obesity per se.30 Moreover,
with registry-based studies in AMI.19,20 A recent study in recent findings from the Dallas Heart Study suggested that
patients with non-STEMI found that obesity was signifi- WC was not independently associated with prevalent athero-
cantly, albeit weakly, correlated with improved outcomes sclerosis after adjustment for standard risk factors.32
after adjustment for confounding prognostic factors No significant increased risk in intermediate-term death
(P0.036).5 The major impact of younger age in the apparent was associated with WC. These findings extend our previous
protection conferred by obesity also has been reported in results that showed that NCEP ATP III defined elevated WC
most recent studies.6,19,21 Moreover, young age may drive an was not an independent predictor of in-hospital outcomes.14
increased use of medications and procedures, a factor that They also extend to acute MI recent findings showing no
may have a favorable impact on outcomes. Interesting data increased WC-associated risk for coronary artery disease in
from the Mayo Clinic registry on MI patients 80 years of men and women with prevalent disease.33 However, further
age included between 1998 and 2001 suggested that the larger prospective studies in acute MI are needed to confirm
obesity paradox, which is present in the short term, disap- these trends. In men, a lower BMI at a given waist girth has
pears in the long term.19 In our study, no data were available been suggested to be associated with higher levels of visceral
on in-hospital death rates because only a 30-day follow-up adipose tissue.13 In the present study, a strikingly worse
Downloaded from http://circ.ahajournals.org/ at UNIVERSITY OF BIRMINGHAM on May 29, 2015
 Zeller et al Obesity in Acute Myocardial Infarction 489

outcome was observed in patients with high WC but normal prevention, particularly in patients with high WC but without
BMI, which presumably reflects the presence of visceral overall obesity. Our results strongly suggest that after MI,
obesity with low muscle mass and a lack of functional neither WC nor BMI has an independent impact on death,
subcutaneous adipose tissue. The identification of such high- taking into account traditional risk factors and other con-
risk subgroups has potential major clinical implications and founding variables. These results appear robust regardless of
may warrant more aggressive lifestyle and therapeutic inter- the type of adjustment. However, as in any observational
ventions for secondary prevention after AMI. study, residual confounding cannot be excluded, and these
results should be confirmed by large, independent, prospec-
Study Limitations tive studies.
In the present study, the same kinds of tape measure and
scales were used to assess WC and BMI at all of the Acknowledgments
centers either public or privately-funded hospitalspartic- We are grateful to Philip Bastable for his help in the preparation of
ipating in the RICO Survey and throughout the inclusion this report and to Sophie Rushton-Smith, PhD, for editorial services.
period. Anthropometric measurements were carried out by
nurses trained in measuring WC using a standardized proce- Sources of Funding
dure for all the centers. However, we cannot exclude the This work was supported by the Association de Cardiologie de
Bourgogne, Centre Hospitalier Universitaire de Dijon, and Conseil
possibility that the impact of WC and BMI might have been Rgional de Bourgogne and by grants from Union Rgionale des
blurred by differences in measurement techniques at the Caisses dAssurance Maladie and Agence Rgionale de
different sites. No data were available on the angiographic lHospitalization de Bourgogne.
features of the study population, a variable known to influ-
ence outcome after MI. However, in patients with AMI, the Disclosures
angiographic extent of coronary artery disease was found to Dr Zeller has received a research grant from Servier. Dr Steg has
received a research grant from sanofi-aventis; served on the speakers
be similar across obesity categories, suggesting that it may bureau for Boehringer-Ingelheim, BMS, GSK, Medtronic, Nycomed,
similarly affect prognosis in all of the groups.5,7,19,21 More- sanofi-aventis, and Servier; and served as a consultant or on the
over, although BMI- and WC-based obesity classifications advisory board for Astellas, AstraZeneca, Bayer, Boehringer-
were unable to independently predict 1-year death rates, we Ingelheim, BMS, Endotis, GSK, Medtronic, MSD, Nycomed, sanofi-
cannot exclude the possibility that obesity may affect the very aventis, Servier, and The Medicines Company. Dr Rochette has
received research grants from sanofi-aventis, Servier, and AstraZen-
long-term follow-up because obesity may have a delayed eca. Dr Cottin has served as a consultant or on the advisory board for
influence on the progression of coronary artery disease. After Servier. The remaining authors report no conflicts.
elective PCI,3,4,34 a U-shaped relation between BMI and the
post-PCI risk of death also has been reported, with the lowest References
risk observed in class I and II obesity and the highest risk in 1. Barrett-Connor EL. Obesity, atherosclerosis, and coronary artery disease.
extreme groups, ie, very lean (BMI 18.5 kg/m2) and very Ann Intern Med. 1985;103:1010 1019.
2. Wolk R, Berger P, Lennon RJ, Brilakis ES, Somers VK. Body mass
severely obese (BMI 40 kg/m2) individuals. However, in index: a risk factor for unstable angina and myocardial infarction in
the present work, the analysis of the extreme groups was patients with angiographically confirmed coronary artery disease.
limited by the small number of subjects. Our study enrolled Circulation. 2003;108:2206 2211.
3. Gruberg L, Weissman NJ, Waksman R, Fuchs S, Deible R, Pinnow EE,
almost exclusively white patients, and it has been suggested
Ahmed LM, Kent KM, Pichard AD, Suddath WO, Satler LF, Lindsay J
that thresholds for defining obesity correlated to CV risk in Jr. The impact of obesity on the short-term and long-term outcomes after
Asian patients may differ from those in white patients percutaneous coronary intervention: the obesity paradox? J Am Coll
(French law prohibits collecting race or ethnicity-related Cardiol. 2002;39:578 584.
4. Gurm HS, Whitlow PL, Kip KE. The impact of body mass index on short-
information). Finally, our study is limited to patients arriving and long-term outcomes inpatients undergoing coronary revasculariza-
at the hospital alive and able to provide consent and undergo tion: insights from the Bypass Angioplasty Revascularization Investi-
anthropometric measurements. Patients who die in the pre- gation (BARI). J Am Coll Cardiol. 2002;39:834 840.
hospital setting and those with an extremely severe clinical 5. Buettner HJ, Mueller C, Gick M, Ferenc M, Allgeier J, Comberg T,
Werner KD, Schindler C, Neumann FJ. The impact of obesity on mor-
status on arrival may differ from the average survivor of the tality in UA/non-ST-segment elevation myocardial infarction. Eur
early phase studied here. Heart J. 2007;28:1694 1701.
6. Mehta L, Devlin W, McCullough PA, ONeill WW, Skelding KA, Stone
Conclusions GW, Boura JA, Grines CL. Impact of body mass index on outcomes after
The present study underscores the high prevalence of in- percutaneous coronary intervention in patients with acute myocardial
creased BMI and WC in AMI patients, present in one quarter infarction. Am J Cardiol. 2007;99:906 910.
7. Diercks DB, Roe MT, Mulgund J, Pollack CV Jr, Kirk JD, Gibler WB,
and one half of the patients, respectively. Until the epidemic Ohman EM, Smith SC Jr, Boden WE, Peterson ED. The obesity paradox
progression of obesity is confronted, cardiologists will be in non-ST-segment elevation acute coronary syndromes: results from the
faced with a growing prevalence of obesity in patients with Can Rapid Risk Stratification of Unstable Angina Patients Suppress
AMI. Given this high prevalence, the characterization of CV Adverse Outcomes With Early Implementation of the American College
of Cardiology/American Heart Association Guidelines Quality
risk associated with obesity after the index event is of Improvement Initiative. Am Heart J. 2006;152:140 148.
importance. We found that much of the apparent obesity 8. Poirier P. Adiposity and cardiovascular disease: are we using the right
paradox can be explained by confounding. Our findings also definition of obesity? Eur Heart J. 2007;28:20472048.
9. Romero-Corral A, Somers VK, Sierra-Johnson J, Jensen MD, Thomas RJ,
emphasize the need to measure both BMI and WC for risk
Squires RW, Allison TG, Korinek J, Lopez-Jimenez F. Diagnostic per-
stratification after MI and suggest the importance of manag- formance of body mass index to detect obesity in patients with coronary
ing high WC-related modifiable risk factors in secondary artery disease. Eur Heart J. 2007;28:20872093.

Downloaded from http://circ.ahajournals.org/ at UNIVERSITY OF BIRMINGHAM on May 29, 2015

490 Circulation July 29, 2008

10. Despres JP, Lemieux I. Abdominal obesity and metabolic syndrome. 23. Das SR, Drazner MH, Dries DL, Vega GL, Stanek HG, Abdullah SM,
Nature. 2006;444:881 887. Canham RM, Chung AK, Leonard D, Wians FH Jr, de Lemos JA. Impact
11. Pouliot MC, Despres JP, Lemieux S, Moorjani S, Bouchard C, Tremblay of body mass and body composition on circulating levels of natriuretic
A, Nadeau A, Lupien PJ. Waist circumference and abdominal sagittal peptides: results from the Dallas Heart Study. Circulation. 2005;112:
diameter: best simple anthropometric indexes of abdominal visceral 21632168.
adipose tissue accumulation and related cardiovascular risk in men and 24. de Lemos JA, Morrow DA, Bentley JH, Omland T, Sabatine MS,
women. Am J Cardiol. 1994;73:460 468. McCabe CH, Hall C, Cannon CP, Braunwald E. The prognostic value of
12. Kissebah AH, Vydelingum N, Murray R, Evans DJ, Hartz AJ, Kalkhoff B-type natriuretic peptide in patients with acute coronary syndromes.
RK, Adams PW. Relation of body fat distribution to metabolic compli- N Engl J Med. 2001;345:1014 1021.
cations of obesity. J Clin Endocrinol Metab. 1982;54:254 260. 25. Jernberg T, Lindahl B, Siegbahn A, Andren B, Frostfeldt G, Lagerqvist B,
13. Onat A, Avci GS, Barlan MM, Uyarel H, Uzunlar B, Sansoy V. Measures Stridsberg M, Venge P, Wallentin L. N-terminal pro-brain natriuretic
of abdominal obesity assessed for visceral adiposity and relation to peptide in relation to inflammation, myocardial necrosis, and the effect of
coronary risk. Int J Obes Relat Metab Disord. 2004;28:1018 1025. an invasive strategy in unstable coronary artery disease. J Am Coll
14. Zeller M, Steg PG, Ravisy J, Laurent Y, Janin-Manificat L, LHuillier I, Cardiol. 2003;42:1909 1916.
Beer JC, Oudot A, Rioufol G, Makki H, Farnier M, Rochette L, Verges 26. Lindahl B, Lindbck J, Jernberg T, Johnston N, Stridsberg M, Venge P,
B, Cottin Y. Prevalence and impact of metabolic syndrome on hospital Wallentin L. Serial analyses of N-terminal pro-B-type natriuretic peptide
outcomes in acute myocardial infarction. Arch Intern Med. 2005;165: in patients with non-ST-segment elevation acute coronary syndromes: a
11921198. Fragmin and Fast Revascularisation During Instability in Coronary Artery
15. Alpert JS, Thygesen K, Antman E, Bassand JP. Myocardial infarction Disease (FRISC)-II substudy. J Am Coll Cardiol. 2005;45:533541.
redefined: a consensus document of the Joint European Society of Car- 27. McCord J, Mundy BJ, Hudson MP, Maisel AS, Hollander JE, Abraham
diology/American College of Cardiology Committee for the Redefinition WT, Steg PG, Omland T, Knudsen CW, Sandberg KR, McCullough PA.
of Myocardial Infarction. J Am Coll Cardiol. 2000;36:959 969. Relationship between obesity and B-type natriuretic peptide levels. Arch
16. World Health Organization. Obesity: Preventing and managing the Intern Med. 2004;164:22472252.
global epidemic. Report on a WHO Consultation. Geneva, Switzerland: 28. Williams SR, Jones E, Bell W, Davies B, Bourne MW. Body habitus and
World Health Organization; 2000. coronary heart disease in men: a review with reference to methods of
17. Executive summary of the Third Report of the National Cholesterol body habitus assessment. Eur Heart J. 1997;18:376 393.
Education Program (NCEP) Expert Panel on Detection, Evaluation, and 29. Isomaa B, Almgren P, Tuomi T, Forsen B, Lahti K, Nissen M, Taskinen
Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel MR, Groop L. Cardiovascular morbidity and mortality associated with the
III). JAMA. 2001;285:2486 2497. metabolic syndrome. Diabetes Care. 2001;24:683 689.
18. Wang Y, Rimm EB, Stampfer MJ, Willett WC, Hu FB. Comparison of 30. Lakka HM, Lakka TA, Tuomilehto J, Salonen JT. Abdominal obesity is
abdominal adiposity and overall obesity in predicting risk of type 2 associated with increased risk of acute coronary events in men. Eur
diabetes among men. Am J Clin Nutr. 2005;81:555563. Heart J. 2002;23:706 713.
19. Nigam A, Wright RS, Allison TG, Williams BA, Kopecky SL, Reeder 31. Dagenais GR, Yi Q, Mann JF, Bosch J, Pogue J, Yusuf S. Prognostic
GS, Murphy JG, Jaffe AS. Excess weight at time of presentation of impact of body weight and abdominal obesity in women and men with
myocardial infarction is associated with lower initial mortality risks but cardiovascular disease. Am Heart J. 2005;149:54 60.
higher long-term risks including recurrent re-infarction and cardiac death. 32. See R, Abdullah SM, McGuire DK, Khera A, Patel MJ, Lindsey JB,
Int J Cardiol. 2006;110:153159. Grundy SM, de Lemos JA. The association of differing measures of
20. Eisenstein EL, McGuire DK, Bhapkar MV, Kristinsson A, Hochman JS, overweight and obesity with prevalent atherosclerosis: the Dallas Heart
Kong DF, Califf RM, Van de Werf F, Yancy WS Jr, Newby LK. Elevated Study. J Am Coll Cardiol. 2007;50:752759.
body mass index and intermediate-term clinical outcomes after acute 33. Canoy D, Boekholdt SM, Wareham N, Luben R, Welch A, Bingham S,
coronary syndromes. Am J Med. 2005;118:981990. Buchan I, Day N, Khaw KT. Body fat distribution and risk of coronary
21. Nikolsky E, Stone GW, Grines CL, Cox DA, Garcia E, Tcheng JE, heart disease in men and women in the European Prospective Investi-
Griffin JJ, Guagliumi G, Stuckey T, Turco M, Negoita M, Lansky AJ, gation Into Cancer and Nutrition in Norfolk cohort: a population-based
Mehran R. Impact of body mass index on outcomes after primary angio- prospective study. Circulation. 2007;116:29332943.
plasty in acute myocardial infarction. Am Heart J. 2006;151:168 175. 34. Minutello RM, Chou ET, Hong MK, Bergman G, Parikh M, Iacovone F,
22. Wang TJ, Larson MG, Levy D, Benjamin EJ, Leip EP, Wilson PW, Wong SC. Impact of body mass index on in-hospital outcomes following
Vasan RS. Impact of obesity on plasma natriuretic peptide levels. percutaneous coronary intervention (report from the New York State
Circulation. 2004;109:594 600. Angioplasty Registry). Am J Cardiol. 2004;93:1229 1232.

CLINICAL PERSPECTIVE
The present study underscores the high prevalence of increased body mass index and waist circumference in patients with
acute myocardial infarction, present in one quarter and one half of the patients, respectively. Until the epidemic progression
of obesity is confronted, cardiologists will be faced with a growing prevalence of obesity in patients with acute myocardial
infarction. Given this high prevalence, the characterization of cardiovascular risk associated with obesity after the index
event is important. Most studies have reported a lower rate of death after myocardial infarction for patients with increased
body mass index (the obesity paradox). We found that much of this apparent obesity paradox is related to confounding by
baseline characteristics associated with survival. Neither body mass index nor waist circumference was an independent
predictor of survival. However, in both men and women, a high waist circumference with low body mass index
(presumably reflecting visceral obesity with low muscle mass and lack of functional subcutaneous adipose tissue) was
predictive of increased 1-year death rate. This emphasizes the need to measure both body mass index and waistline in
patients with myocardial infarction, particularly to identify this sizable fraction of the patient population at high risk of
death.

Downloaded from http://circ.ahajournals.org/ at UNIVERSITY OF BIRMINGHAM on May 29, 2015

 Relation Between Body Mass Index, Waist Circumference, and Death After Acute
Myocardial Infarction
Marianne Zeller, Philippe Gabriel Steg, Jack Ravisy, Luc Lorgis, Yves Laurent, Pierre Sicard,
Luc Janin-Manificat, Jean-Claude Beer, Hamid Makki, Anne-Ccile Lagrost, Luc Rochette and
Yves Cottin
for the RICO Survey Working Group

Circulation. 2008;118:482-490; originally published online July 14, 2008;

doi: 10.1161/CIRCULATIONAHA.107.753483
Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright 2008 American Heart Association, Inc. All rights reserved.
Print ISSN: 0009-7322. Online ISSN: 1524-4539

The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://circ.ahajournals.org/content/118/5/482

Data Supplement (unedited) at:

http://circ.ahajournals.org/content/suppl/2008/07/25/CIRCULATIONAHA.107.753483.DC1.html

Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published
in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial
Office. Once the online version of the published article for which permission is being requested is located,
click Request Permissions in the middle column of the Web page under Services. Further information about
this process is available in the Permissions and Rights Question and Answer document.

Reprints: Information about reprints can be found online at:

http://www.lww.com/reprints

Subscriptions: Information about subscribing to Circulation is online at:

http://circ.ahajournals.org//subscriptions/

Downloaded from http://circ.ahajournals.org/ at UNIVERSITY OF BIRMINGHAM on May 29, 2015