0
Original Investigation |

Relation of Low Body Mass to Death and Stroke in the Systolic Hypertension in the Elderly Program FREE

Sylvia Wassertheil-Smoller, PhD; Cathy Fann, PhD; Richard M. Allman, MD; Henry R. Black, MD; Greta H. Camel, MD; Barry Davis, MD, PhD; Kamal Masaki, MD; Sarah Pressel, MS; Ronald J. Prineas, MB, BS, PhD; Jeremiah Stamler, MD; Thomas M. Vogt, MD, MPH; for the SHEP Cooperative Research Group
[+] Author Affiliations

From the Department of Epidemiology and Social Medicine, Albert Einstein College of Medicine of Yeshiva University, Bronx, NY (Drs Wassertheil-Smoller and Fann); Division of Gerontology, Department of Medicine, the University of Alabama at Birmingham, and the Veterans Affairs Medical Center, Birmingham (Dr Allman); Departments of Preventive Medicine, Rush–Presbyterian–St Luke's Medical Center (Dr Black) and Northwestern University Medical School (Dr Stamler), Chicago, Ill; Hypertension Division, Washington University School of Medicine, St Louis, Mo (Dr Camel); Department of Biometry, University of Texas Health Science Center, Houston (Dr Davis and Ms Pressel); Division of Geriatric Medicine, the John A. Burns School of Medicine (Dr Masaki) and the Center for Health Research, Kaiser Permanente (Dr Vogt), Honolulu, Hawaii; and the Department of Public Health Sciences, Wake Forest University School of Medicine, Winston Salem, NC (Dr Prineas). A complete list of the members of the SHEP Cooperative Research Group appears in JAMA (1991;265:3255-3264).


Arch Intern Med. 2000;160(4):494-500. doi:10.1001/archinte.160.4.494.
Text Size: A A A
Published online

Background  There are scant data on the effect of body mass index (BMI) (calculated as weight in kilograms divided by the square of height in meters) on cardiovascular events and death in older patients with hypertension.

Objective  To determine if low body mass in older patients with hypertension confers an increased risk of death or stroke.

Patients  Participants were 3975 men and women (mean age, 71 years) enrolled in 17 US centers in the Systolic Hypertension in the Elderly Program trial, a randomized, double-blind, placebo-controlled clinical trial of low-dose antihypertensive therapy, with follow-up for 5 years.

Main Outcome Measures  Five-year adjusted mortality and stroke rates from Cox proportional hazards analyses.

Results  There was no statistically significant relation of death or stroke with BMI in the placebo group (P = .47), and there was a U- or J-shaped relation in the treatment group. The J-shaped relation of death with BMI in the treated group (P = .03) showed that the lowest probability of death for men was associated with a BMI of 26.0 and for women with a BMI of 29.6; the curve was quite flat for women across a wide range of BMIs. For stroke, men and women did not differ, and the BMI nadir for both sexes combined was 29, with risk increasing steeply at BMIs below 24. Those in active treatment, however, had lower death and stroke rates compared with those taking placebo.

Conclusions  Among older patients with hypertension, a wide range of BMIs was associated with a similar risk of death and stroke; a low BMI was associated with increased risk. Lean, older patients with hypertension in treatment should be monitored carefully for additional risk factors.

Figures in this Article

THERE HAS been much emphasis on obesity as a risk factor for cardiovascular disease and mortality,14 but there has been no comparable focus on possible risks associated with leanness. Many studies58 have found a U- or J-shaped relation between body mass index (BMI) (calculated as weight in kilograms divided by the square of height in meters) and mortality. Others,2,9,10 however, have found a strong and direct relation between BMI and cardiovascular disease and mortality. These studies have helped to disseminate the widespread belief that losing weight is universally desirable. The association between low levels of BMI and mortality has often been explained as being due to confounding, by smoking, for example,8 or to preexisting illness. However, studies2,3 that control for confounders and that exclude events occurring during the first period of follow-up still find a residual J- or U-shaped relation. Whether early or chronic obesity confers the excess risk observed at high levels of BMI and whether weight loss or a low BMI per se confers the excess risk at low levels of BMI are not established. Little attention has been devoted to BMI and risk in older persons or in those with an existing recognized comorbid condition. This article presents data on the relation of BMI with death and stroke in a cohort of older persons enrolled in the Systolic Hypertension in the Elderly Program (SHEP), a placebo-controlled, randomized, clinical trial of antihypertensive therapy for isolated systolic hypertension.

Estimates of the prevalence of elevations in SBP of 160 mm Hg or higher with a normal diastolic blood pressure range from 6% to 18% depending on age,11 and prevalence is higher in women vs men and in blacks vs whites.12 This report includes information on the effect of antihypertensive treatment in thin vs obese older patients with hypertension and on whether there is a difference in the relation of BMI to event rates in women vs men.

The primary objective of SHEP was to determine if drug treatment for isolated systolic hypertension in older people reduced the 5-year stroke incidence. The design is described in detail elsewhere.13 In summary, the clinical trial enrolled 4736 men and women aged 60 years and older who had SBPs of 160 mm Hg or higher and a diastolic blood pressure lower than 90 mm Hg. Participants were randomized to active treatment or matching placebo and followed up for 5 years, with a blood pressure measurement obtained every 4 months. The drugs used in active treatment were chlorthalidone, 12.5 mg, increased to 25 mg, and atenolol if necessary, 25 or 50 mg, to achieve a goal SBP of 160 mm Hg or 20 mm Hg less than the blood pressure at study entry. The final results of SHEP indicated a significant 37% reduction in stroke risk with treatment.14

The cohort analyzed to address the question of the relation of BMI to SHEP end points included all participants with complete data except those who died or had a stroke or myocardial infarction (MI) within 6 months of the baseline visit and those who died of cancer or had a cancer diagnosis at any time during the trial, to minimize the possible effect of underlying illness. There were 4022 persons who met those criteria. There were 47 persons with missing BMI values who were also excluded, resulting in 3975 persons in the analysis cohort.

First, the cohort was divided into sex-specific quintiles of BMI to determine crude stroke and death rates in each quintile. Five-year cumulative probabilities of event rates, adjusted for age, race, sex, and educational level, were obtained from Cox proportional hazards analyses for each quintile. The baseline characteristics of participants in the lowest quintile were compared with those of participants in the other quintiles and were tested for significance by the χ2 test for discrete variables and the t test for continuous variables. Multiple comparisons were adjusted for by the Bonferroni procedure.15,16

To adjust for additional confounders and to calculate relative risk, multivariate Cox proportional hazards analyses were done, with covariates being age, sex, race, and educational level; history of stroke, MI, or diabetes; and activities of daily living,17 smoking status, serum cholesterol level, SBP attained in treatment as a time-dependent variable and its square term, and baseline BMI and its square term. In such an analysis, if the quadratic or square term is significant, it indicates that the relation between the independent variable of interest and the outcome variable is not linear but can better be fitted by a J or U shape. The reason that the square term of attained in-treatment SBP was included is that in a previous analysis a quadratic (J-shaped) relation was found between SBP, as a time-dependent variable, and death and cardiovascular disease within the treatment group. The lowest risk of death was at a treated SBP of 142 mm Hg and of cardiovascular disease at a treated SBP of 146 mm Hg. The risk relative to the nadir of attained blood pressure was slightly higher at blood pressures lower than the nadir and much higher at blood pressures higher than 160 mm Hg (S.W.-S., B.D., unpublished data, 1996). Hence, we controlled for that relation. In the present analysis of the restricted cohort that excluded persons who died within the first 6 months and those who had a cancer death or diagnosis at any time, the J-shaped SBP relation to death had a nadir at 145 mm Hg but was not statistically significant (P = .18) for men and women combined.

When the square term of BMI was significant (indicating a J or U relation), we calculated the value of BMI associated with the lowest risk (the nadir) and obtained the risk of event for different values of BMI relative to the nadir (the formula is available from the authors). To determine whether the BMI-mortality relation was different for control and treatment groups, several Cox proportional hazards models were run, including models with interaction terms for this variable. These analyses indicated that the BMI-mortality relation was different in the control and treatment groups, and these 2 groups were, therefore, analyzed separately. We performed similar tests to see whether the BMI-stroke relation was different for the control and treatment groups. These analyses indicated that the BMI-stroke relation was also different for the control and treatment groups; hence, these 2 groups were analyzed separately. Furthermore, for the outcome of death within the treatment group, the interaction term between sex and BMI was significant in the survival analysis (P = .04). Thus, it was appropriate to do separate analyses of men and women for death within the treatment group. Sex was not a significant (P = .46) covariate for stroke; hence, analyses of the end point of stroke were done for both sexes combined.

RELATION OF BMI TO DEATH AND STROKE

Table 1 shows the crude mortality and fatal plus nonfatal stroke rates by BMI quintiles and the rates adjusted for sex, age, race, and educational level. Mortality rates were higher in the lowest quintile of BMI than in the middle quintile for the total group and within the control and treatment groups separately. For stroke, rates were higher in the lowest quintile than in either the middle or the highest quintile of BMI in the control and treatment groups and lower for the treatment than the control group within each quintile.

Table Graphic Jump LocationTable 1. Mortality and Stroke Rates by BMI Quintile*

Since the lowest BMI was associated with greater risk, we examined the baseline characteristics of those in the lowest quintile compared with those in the middle quintile (Table 2). In the lowest quintile of BMI compared with the middle quintile, persons tended to be older and there was a statistically significant higher proportion of smokers and of persons with a history of stroke and a significantly lower proportion of persons with diabetes. There was no difference in systolic or diastolic blood pressure, and the proportion of persons with a history of MI was similar in all 5 quintiles. To adjust for possible confounders, multivariate Cox proportional hazards analyses were done, with control for the multiple covariates noted previously and including the square term of BMI. Since the BMI and BMI2 randomization group interaction was significant for mortality and stroke, analyses were done separately for the treatment and control groups.

Table Graphic Jump LocationTable 2. Baseline Characteristics by BMI Quintile*

In the control group, the relation between BMI and death was inverse and significant, with control for age, sex, race, and educational level (P = .05), and became nonsignificant when smoking was added to the model and when all the other covariates were also included. These findings were unchanged in an analysis of the subcohort of never smokers only (n = 1852) or of current and past smokers; the quadratic term of BMI was not significant in the control group, for all participants, for never smokers only, or for current and past smokers. Thus, with control of multiple confounders, there was no significant relation of BMI to death in the placebo group, although the crude and age, sex, race, and educational level adjusted rates indicate that in the control and the treatment groups, the highest mortality was in the lowest BMI quintile. The same findings were true for stroke, with no statistically significant relation of BMI to stroke in the placebo group, with control for all covariates, but with crude and age, sex, race, and educational level adjusted rates indicating higher stroke rates in the lowest BMI quintile (Table 1).

For the active treatment group, Table 3 shows multivariate Cox regression coefficients and their SEs for death and stroke. The BMI quadratic term for both events was significant, with the point of lowest risk (nadir) for death occurring at a BMI of 27.7 and for stroke at a BMI of 29.3 (Figure 1). We repeated these analyses for never and for current and past smokers combined, with similar results, indicating a U shape in the treatment. The nadir for never smokers was 27.9 for death (P = .05) and 30.7 for stroke (P = .25), which was not significant; for current and past smokers, it was 29.1 for death (P = .007) and 28.9 for stroke (P = .06).

Table Graphic Jump LocationTable 3. Cox Regression Analyses' Coefficients for Death and Stroke in the Active Treatment Group
Place holder to copy figure label and caption
Figure 1.

Adjusted relative risk of death and fatal and nonfatal stroke within the active treatment group by body mass index (BMI) (calculated as weight in kilograms divided by the square of height in meters). The variables included educational level, history of diabetes, history of myocardial infarction, history of stroke, age, activity level, sex, race, smoking status, cholesterol level, BMI, BMI2, systolic blood pressure (SBP), and SBP2 (time dependent).

Graphic Jump Location

We also examined specific causes of the 215 deaths in this cohort (data not shown). The pattern of crude higher mortality in the lowest BMI quintile was true for deaths due to MI, stroke, infections, and indeterminate causes and sudden or rapid death (sudden death is within 1 hour, and rapid death is within 24 hours). There were 13 deaths due to indeterminate causes in the placebo group and 9 in the active treatment group. The number of deaths in each of these categories was too small to draw any statistical inferences; however, deaths due to coronary heart disease (CHD) and stroke in the active treatment group were lower in the lowest quintile of BMI (26.2/1000 compared with 38.4/1000 in the highest BMI quintile and 17.7/1000 in the middle quintile). In contrast, in the placebo group, while there was a markedly higher rate of deaths in the lowest BMI quintile (49.8/1000), there was not the expected increase in deaths due to CHD or stroke in the highest BMI quintile, which had a rate of 15.1 per 1000. An analysis of the 154 deaths due to cancer excluded from these analyses indicated that 25.3% occurred in the lowest, 21.4% in the middle, and 18.2% in the highest quintile of baseline BMI. The mean baseline BMI among the 4227 persons in the entire SHEP group who lived for the next 5 years was 27.6 (95% confidence interval, 27.4-27.7), while for those who had a subsequent stroke it was 25.7 (95% confidence interval, 23.5-28.0) and for those who had a subsequent MI it was 25.5 (95% confidence interval, 24.3-26.7) (data not shown).

RELATION OF BMI TO DEATH BY SEX

Since the interaction between sex and BMI was significant within the treatment group for death, analyses were done for men and women separately. Figure 2 shows that the relation of BMI to death was U shaped in each sex group. The nadir of BMI for men in the treatment group was 25.8; for women, 29.6. Among men in the placebo group (not shown), there was also a U-shaped relation of BMI and death (P = .06), with the nadir at 30.3.

Place holder to copy figure label and caption
Figure 2.

The adjusted relative risk of death within the treatment group by body mass index (BMI) (calculated as weight in kilograms divided by the square of height in meters). The variables included educational level, history of diabetes, history of myocardial infarction, history of stroke, age, activity level, race, smoking status, cholesterol level, BMI, BMI2, systolic blood pressure (SBP), and SBP2.

Graphic Jump Location

The curve for women is relatively flat for a BMI of 27 to 34, indicating that the risk of death is nearly the same across a wide range of BMIs. The curve is markedly steeper to the right for men, but for men also a range of BMIs from 24 to 28 is associated with a similar risk of death. The U-shaped relation between BMI and death was similar when the analysis was done among the group of nonsmoking men and similar when done when those with a history of diabetes were excluded (data not shown).

As an example, in the treatment group at a BMI of 20, compared with the optimal BMI value, there is about a 72% excess risk of death for a woman and 36% for a man. Such a BMI value for a 173-cm-tall (5'8") man corresponds to a weight of 60 kg (132 lbs); for a 163-cm-tall (5'4") woman, 53 kg (117 lbs).

We have found, in a study of 4000 persons aged 60 years and older who were participants in a placebo-controlled clinical trial (SHEP), that there was increased risk of death among lean, elderly patients with hypertension compared with those in the midrange of BMI due to all causes, stroke, CHD, and noncardiovascular causes. In the actively treated group, there was a marked U-shaped relation of BMI to death in men and women, and those who had a low BMI had a greater risk of death and stroke than those with more intermediate weights, with the lowest risk of death at a BMI of 27.7 and of stroke at a BMI of 29.3. The lowest risk of death for men was associated with a BMI of 25.8; for women, a BMI of 29.6. These results persisted after controlling for multiple covariates. There was a wide range of BMIs within which the probability of death or stroke was virtually the same, especially for women. For an elderly man who is 173 cm tall (5'8"), the lowest risk was at a weight of 78 kg (172 lbs), with a statistically equivalent risk between 74 and 80 kg (165-178 lbs). For an elderly woman who is 163 cm tall (5'4"), the lowest risk was at a weight of 71.7 kg (158 lbs).

An analysis of the causes of death indicated that crude rates were higher for the lowest BMI quintile from cardiovascular and noncardiovascular causes, in the placebo and active treatment groups. It may be that elderly persons who are excessively thin are less able to withstand physiological insults in general. Low body weight in older persons may also be associated with poor nutritional status, which may contribute to disease.18

The findings reported herein, although based on data for persons with hypertension, sound a note of caution for weight loss recommendations for older persons who are only moderately overweight. Data from the National Health and Nutrition Examination Survey Epidemiologic Follow-up Study and the National Health Interview Survey of people6 aged 30 years and older found consistent U-shaped relations across all sex-race groups, with a steeply increasing relative risk of mortality at low BMI values and a broad range of BMI values in the middle of the distribution being associated with low relative mortality. The BMI associated with minimum mortality ranged from 25 to 28 for whites, depending on the cohort analyzed and on sex, and was higher for blacks, ranging from 27 to 30. A more recent analysis of those data from the same group,19 which modeled the relation between BMI and mortality compared with the minimum, found that the interval of BMI over which all-cause mortality did not increase by more than 20% was 9 BMI units wide. Hypertension, diabetes, and hypercholesterolemia were not controlled for in this analysis. The mean age in this cohort ranged from 48 years among men of both races to 54 years among black men, substantially younger than our cohort, whose mean age was older than 70 years.

An important study of the association of BMI to mortality among younger persons with hypertension (aged 30-69 years), in the Hypertension Detection and Follow-up Program, found a U-shaped relation to all-cause, cardiovascular, and noncardiovascular mortality in men and women.7,8 Persons with a BMI below 22 had higher adjusted 8-year death rates than those with a BMI in the median range of 26 to 29. Among men, smokers and nonsmokers had higher death rates at a low BMI; deaths related to lifestyle factors, such as smoking and alcohol intake, contributed to the excess risk, particularly among lean persons with hypertension.8 Our results are similar to those found in the younger nonhypertensive population samples, and in younger persons with hypertension, and extend the findings that there is a wide range of BMIs associated with similar risk to treated hypertensive elderly people.

Even studies2,10 that have been widely quoted as reporting a direct effect of high BMI on mortality have found that at a low BMI (<20) there is an excess risk of death, but these findings have not been emphasized and have largely been ignored in the popularly held belief that "thinner is better" regardless of age. A study in Germany20 of 6193 obese participants (BMI, >25) aged 18 to 75 years followed up for an average of 14 years indicated that moderate obesity (BMI, 25-32) was not associated with excess mortality. Similar to our findings, the impact of excess obesity (BMI, ≥32) was less pronounced in women than in men; gross obesity (BMI, ≥40) conferred a 3-fold excess risk in men and a 2-fold excess risk in women. The American Cancer Society's Cancer Prevention Study21 of 62 116 nonsmoking men and 262 019 nonsmoking women reported that 12-year mortality increased with increasing BMI up to age 75 years, but the relative risk associated with a greater BMI declined with age. The curves of relative risk of deaths due to all causes and deaths due to cardiovascular disease for women aged 65 years and older are quite flat across a range of BMIs from 19 to more than 32. For younger men and women, relative risk increases after a BMI of 25 compared with the reference category of BMI of 19.0 to 21.9. These relative risks were adjusted for age, educational level, physical activity, and alcohol consumption.

The Buffalo Health Study22 analyzed the experience of 611 white men and 697 white women aged 20 to 96 years with an average length of follow-up of 29 years. They found that for older men (aged ≥65 years), there was no effect of BMI on all-cause, cardiovascular, or CHD mortality. For those men younger than 65 years, however, risk increased with successive quartiles of BMI compared with the lowest quartile. Among women of all ages combined, and among nonsmoking women analyzed separately, BMI was not related to all-cause mortality when adjusted for age, but there was increased relative risk for mortality due to CHD in the upper quartiles. The researchers did not present data for older women separately. In the Epidemiologic Studies of the Elderly, participants aged 65 years and older were followed up prospectively for 10 years starting in 1981; the decedent cohort was compared with older survivors (men >80 years and women >85 years). Survivors had a higher BMI at baseline than their counterparts who died during the follow-up period. Overweight, however, was significantly related to disability before death.23 However, BMI may be a more important risk factor for CHD in younger than in older people. A report from the Atherosclerosis Risk in Communities Study24 indicates that in a cohort of 14 040 participants aged 45 to 64 years and free of CHD at baseline with an average follow-up of 6.2 years, there was a trend of increasing relative risks of heart disease across increasing quartiles of BMI overall for men (P = .06 for trend), but not significantly so for never smokers (P = .17 for trend); among women, risks increased for the total group (P = .02 for trend) and for never smokers (P = .05 for trend), although CHD incidence was relatively flat across the lower 2 quartiles of BMI. However, the waist-hip ratio may be more predictive than BMI. The Atherosclerosis Risk in Communities study found that the relation of increasing waist-hip ratio with increasing CHD incidence was stronger for men and women than the relation with BMI. In older people, this may be an especially important factor, since there is loss of lean body mass in older individuals and BMI may not fully reflect the level of adiposity in these persons. We did not have waist-hip measures available for analysis in this cohort and so cannot say if there would be a different relation between waist-hip ratio and mortality than what we found with BMI. Thus, despite the fact that most studies indicate that there is little, if any, effect of BMI on mortality in older women, until grossly high levels of obesity are reached, the prevailing conclusions drawn from the reports of these studies imply that the lower the relative weight, the better for older women, and for younger ones.

We could not determine from our data set if the baseline BMI was stable in the years before our study or if there had been a weight gain or loss from a younger age. However, we ran Cox proportional hazards models that included BMI as a time-dependent covariate to adjust for changes in BMI during the 5-year follow-up period. In these analyses, we also found the quadratic term of BMI to be statistically significant, indicating a J- or U-shaped relation for death and stroke in the active treatment group.

It has been postulated that the effects of preexisting illness may account for these U-shaped relations of weight to mortality. We cannot rule that out completely; however, we have lessened that possibility by excluding from our analysis cohort those persons who had a diagnosis of cancer or a death due to cancer at any time during the 5-year follow-up and those who had a stroke or MI or died before 6 months of follow-up. We repeated these analyses by excluding those who died or who had a history of stroke before 1 year, or who had diabetes, or who were never smokers, with the same results in each case.

In the treatment group, the lowest death risk was at a BMI of 26 for men and 30 for women. However, the risk of death and stroke remained similar across a wide range of BMIs (the curves are relatively flat). We cannot explain the different shape of the relation between BMI and death or stroke that we found in the treatment vs the control group. In the treatment group, there was increased risk at a low BMI, while in the control group there was not a significant quadratic relation (an inverse linear trend in the control group crude mortality rate was not significant when smoking was controlled for). There is no U or J shape in the control group because there is not the expected increase in risk at the high BMI end and not because there is a lower risk at a low BMI. In contrast, in the treatment group, there is an increase in risk at the high BMI end. However, it is especially important to note that those in active treatment for hypertension compared with the placebo group had a lower risk of stroke within each BMI quintile and a lower risk of death within each quintile, except the highest BMI quintile, which had aberrantly low death rates in the control group, as shown in Table 1.

While the treatment and placebo groups were similar in baseline characteristics that could be potential confounders, as shown in Table 2, and we controlled for the known and potential confounders shown in Table 3, it is possible that, despite randomization, the control and treatment groups differed in important unknown confounders or that our statistical adjustment did not provide sufficient control. In particular, the confounding of smoking and obesity could present a problem. Obesity may increase the risk of mortality through its effect on hypertension. Smoking, in turn, is associated with lower weight. We found, however, that in never smokers there was still a J- or U-shaped relation of mortality and stroke with BMI in the active treatment group.

In conclusion, elderly persons with systolic hypertension who are thin should be especially carefully monitored for cardiovascular risk, nutritional status, and other comorbidities, including depression, which is a risk factor for death and may itself be related to poor nutritional status.18,25 Nevertheless, although these results suggest that excessive leanness in elderly persons confers increased risk, it should not be forgotten that excessive overweight also confers increased risk, especially among men; continued attention should be given to the well-established relations between obesity and the known risk factors for cardiovascular disease (hypertension, hypercholesterolemia, and diabetes).

Accepted for publication April 22, 1999.

This study was supported by grant NO1-HC-35130 from the National Institutes of Health, Bethesda, Md.

We thank Victor Kamensky, MS, for his work on the progamming of the statistical analyses.

Reprints: Sylvia Wassertheil-Smoller, PhD, Department of Epidemiology and Social Medicine, Albert Einstein College of Medicine of Yeshiva University, 1300 Morris Park Ave, Room 1312 Belfer, Bronx, NY 10461 (e-mail: smoller@aecom.yu.edu).

Lew  EAGarfinkel  L Variations in mortality by weight among 750,000 men and women. J Chronic Dis. 1979;32563- 576
Manson  JWillett  WStampfer  M  et al.  Body weight and mortality among women. N Engl J Med. 1995;333677- 685
Sorlie  PGordon  TKannel  W Body build and mortality: the Framingham Study. JAMA. 1980;2431828- 1831
Dyer  ARStamler  JBerkson  DMLindberg  HA Relationship of relative weight and body mass index to 14-year mortality in the Chicago Peoples Gas Company Study. J Chronic Dis. 1975;28109- 123
Folsom  AKaye  SSellers  T  et al.  Body fat distribution and 5-year risk of death in older women. JAMA. 1993;269483- 487
Durazo-Arvizu  RCooper  RLuke  APrewitt  TLiao  YMcGee  D Relative weight and mortality in U.S. blacks and whites: findings from representative national population samples. Ann Epidemiol. 1997;7383- 395
Langford  HGStamler  JWassertheil-Smoller  SPrineas  RJ All-cause mortality in the Hypertension Detection and Follow-up Program: findings for the whole cohort and for persons with less severe hypertension, with and without other traits related to risk of mortality. Prog Cardiovasc Dis. 1986;29 ((suppl 1)) 29- 54
Stamler  ROFord  CEStamler  J Why do lean hypertensives have higher mortality rates than other hypertensives? findings of the Hypertension Detection and Follow-up Program. Hypertension. 1991;17553- 564
Mattila  KHaavisto  MRajala  S Body mass index and mortality in the elderly. Br Med J (Clin Res Ed). 1986;292867- 868
Harris  TCook  EGarrison  R  et al.  Body mass index and mortality among non-smoking older persons: the Framingham Heart Study. JAMA. 1988;2591520- 1524
Curb  JBorhani  NEntwistle  G  et al.  Isolated systolic hypertension in 14 communities. Am J Epidemiol. 1985;121362- 370
Not Available, Blood Pressure Levels in Persons 18-74 Years of Age in 1976-1980 and Trends in Blood Pressure From 1960 to 1980 in the United States: Data From the National Health Survey.  Hyattsville Md US Dept of Health and Human Services, Public Health Service1986;DHHS publication (PHS) 86-1684. Vital and Health Statistics 11, No. 234
Borhani  NOApplegate  WBCutler  JA  et al.  Part I: rationale and design. Hypertension. 1991;17 ((suppl 2)) II-2- II-15
Group  SCR Prevention of stroke by antihypertensive drug treatment in older persons with isolated systolic hypertension: final results of the Systolic Hypertension in the Elderly Program (SHEP). JAMA. 1991;2653255- 3264
O'Brien  PCFleming  TR A multiple testing procedure for clinical trials. Biometrics. 1979;35549- 556
Wassertheil-Smoller  S Biostatistics and Epidemiology: A Primer for Health Professionals. 2nd ed. New York, NY Springer-Verlag NY Inc1995;
Branch  LKatz  SKniepman  K A prospective study of functional status among community elders. Am J Public Health. 1984;74266- 268
Fischer  JJohnson  M Low body weight and weight loss in the aged. J Am Diet Assoc. 1990;901697- 1706
Durazo-Arvizu  RMcGee  DCooper  RLiao  YLuke  A Mortality and optimal body mass index in a sample of the US population. Am J Epidemiol. 1998;147739- 749
Bender  RTrautner  CSpraul  MBerger  M Assessment of excess mortality in obesity. Am J Epidemiol. 1998;14742- 48
Stevens  JCai  JPamuk  ERWilliamson  DFThun  MJWood  JL The effect of age on the association between body-mass index and mortality. N Engl J Med. 1998;3381- 7
Dorn  JMSchisterman  EFWinkelstein  WTrevisan  M Body mass index and mortality in a general population sample of men and women: the Buffalo Health Study. Am J Epidemiol. 1997;146919- 931
Leveille  SGuralnik  JFerrucci  LLanglois  J Aging successfully until death in old age: opportunities for increasing active life expectancy. Am J Epidemiol. 1999;149654- 664
Folsom  AStevens  JSchreiner  PMcGovern  PAtherosclerosis Risk in Communities Study Investigators, Body mass index, waist/hip ratio, and coronary heart disease incidence in African Americans and whites. Am J Epidemiol. 1998;1481187- 1194
Wassertheil-Smoller  SApplegate  WBBerge  K  et al.  Change in depression as a precursor of cardiovascular events. Arch Intern Med. 1996;156553- 561

Figures

Place holder to copy figure label and caption
Figure 1.

Adjusted relative risk of death and fatal and nonfatal stroke within the active treatment group by body mass index (BMI) (calculated as weight in kilograms divided by the square of height in meters). The variables included educational level, history of diabetes, history of myocardial infarction, history of stroke, age, activity level, sex, race, smoking status, cholesterol level, BMI, BMI2, systolic blood pressure (SBP), and SBP2 (time dependent).

Graphic Jump Location
Place holder to copy figure label and caption
Figure 2.

The adjusted relative risk of death within the treatment group by body mass index (BMI) (calculated as weight in kilograms divided by the square of height in meters). The variables included educational level, history of diabetes, history of myocardial infarction, history of stroke, age, activity level, race, smoking status, cholesterol level, BMI, BMI2, systolic blood pressure (SBP), and SBP2.

Graphic Jump Location

Tables

Table Graphic Jump LocationTable 1. Mortality and Stroke Rates by BMI Quintile*
Table Graphic Jump LocationTable 2. Baseline Characteristics by BMI Quintile*
Table Graphic Jump LocationTable 3. Cox Regression Analyses' Coefficients for Death and Stroke in the Active Treatment Group

References

Lew  EAGarfinkel  L Variations in mortality by weight among 750,000 men and women. J Chronic Dis. 1979;32563- 576
Manson  JWillett  WStampfer  M  et al.  Body weight and mortality among women. N Engl J Med. 1995;333677- 685
Sorlie  PGordon  TKannel  W Body build and mortality: the Framingham Study. JAMA. 1980;2431828- 1831
Dyer  ARStamler  JBerkson  DMLindberg  HA Relationship of relative weight and body mass index to 14-year mortality in the Chicago Peoples Gas Company Study. J Chronic Dis. 1975;28109- 123
Folsom  AKaye  SSellers  T  et al.  Body fat distribution and 5-year risk of death in older women. JAMA. 1993;269483- 487
Durazo-Arvizu  RCooper  RLuke  APrewitt  TLiao  YMcGee  D Relative weight and mortality in U.S. blacks and whites: findings from representative national population samples. Ann Epidemiol. 1997;7383- 395
Langford  HGStamler  JWassertheil-Smoller  SPrineas  RJ All-cause mortality in the Hypertension Detection and Follow-up Program: findings for the whole cohort and for persons with less severe hypertension, with and without other traits related to risk of mortality. Prog Cardiovasc Dis. 1986;29 ((suppl 1)) 29- 54
Stamler  ROFord  CEStamler  J Why do lean hypertensives have higher mortality rates than other hypertensives? findings of the Hypertension Detection and Follow-up Program. Hypertension. 1991;17553- 564
Mattila  KHaavisto  MRajala  S Body mass index and mortality in the elderly. Br Med J (Clin Res Ed). 1986;292867- 868
Harris  TCook  EGarrison  R  et al.  Body mass index and mortality among non-smoking older persons: the Framingham Heart Study. JAMA. 1988;2591520- 1524
Curb  JBorhani  NEntwistle  G  et al.  Isolated systolic hypertension in 14 communities. Am J Epidemiol. 1985;121362- 370
Not Available, Blood Pressure Levels in Persons 18-74 Years of Age in 1976-1980 and Trends in Blood Pressure From 1960 to 1980 in the United States: Data From the National Health Survey.  Hyattsville Md US Dept of Health and Human Services, Public Health Service1986;DHHS publication (PHS) 86-1684. Vital and Health Statistics 11, No. 234
Borhani  NOApplegate  WBCutler  JA  et al.  Part I: rationale and design. Hypertension. 1991;17 ((suppl 2)) II-2- II-15
Group  SCR Prevention of stroke by antihypertensive drug treatment in older persons with isolated systolic hypertension: final results of the Systolic Hypertension in the Elderly Program (SHEP). JAMA. 1991;2653255- 3264
O'Brien  PCFleming  TR A multiple testing procedure for clinical trials. Biometrics. 1979;35549- 556
Wassertheil-Smoller  S Biostatistics and Epidemiology: A Primer for Health Professionals. 2nd ed. New York, NY Springer-Verlag NY Inc1995;
Branch  LKatz  SKniepman  K A prospective study of functional status among community elders. Am J Public Health. 1984;74266- 268
Fischer  JJohnson  M Low body weight and weight loss in the aged. J Am Diet Assoc. 1990;901697- 1706
Durazo-Arvizu  RMcGee  DCooper  RLiao  YLuke  A Mortality and optimal body mass index in a sample of the US population. Am J Epidemiol. 1998;147739- 749
Bender  RTrautner  CSpraul  MBerger  M Assessment of excess mortality in obesity. Am J Epidemiol. 1998;14742- 48
Stevens  JCai  JPamuk  ERWilliamson  DFThun  MJWood  JL The effect of age on the association between body-mass index and mortality. N Engl J Med. 1998;3381- 7
Dorn  JMSchisterman  EFWinkelstein  WTrevisan  M Body mass index and mortality in a general population sample of men and women: the Buffalo Health Study. Am J Epidemiol. 1997;146919- 931
Leveille  SGuralnik  JFerrucci  LLanglois  J Aging successfully until death in old age: opportunities for increasing active life expectancy. Am J Epidemiol. 1999;149654- 664
Folsom  AStevens  JSchreiner  PMcGovern  PAtherosclerosis Risk in Communities Study Investigators, Body mass index, waist/hip ratio, and coronary heart disease incidence in African Americans and whites. Am J Epidemiol. 1998;1481187- 1194
Wassertheil-Smoller  SApplegate  WBBerge  K  et al.  Change in depression as a precursor of cardiovascular events. Arch Intern Med. 1996;156553- 561

Correspondence

CME
Meets CME requirements for:
Browse CME for all U.S. States
Accreditation Information
The American Medical Association is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. The AMA designates this journal-based CME activity for a maximum of 1 AMA PRA Category 1 CreditTM per course. Physicians should claim only the credit commensurate with the extent of their participation in the activity. Physicians who complete the CME course and score at least 80% correct on the quiz are eligible for AMA PRA Category 1 CreditTM.
Note: You must get at least of the answers correct to pass this quiz.
You have not filled in all the answers to complete this quiz
The following questions were not answered:
Sorry, you have unsuccessfully completed this CME quiz with a score of
The following questions were not answered correctly:
Commitment to Change (optional):
Indicate what change(s) you will implement in your practice, if any, based on this CME course.
Your quiz results:
The filled radio buttons indicate your responses. The preferred responses are highlighted
For CME Course: A Proposed Model for Initial Assessment and Management of Acute Heart Failure Syndromes
Indicate what changes(s) you will implement in your practice, if any, based on this CME course.
NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).
Submit a Comment

Multimedia

Some tools below are only available to our subscribers or users with an online account.

Web of Science® Times Cited: 48

Related Content

Customize your page view by dragging & repositioning the boxes below.

Articles Related By Topic
Related Topics
PubMed Articles