ORIGINAL INVESTIGATION Overweight and Obesity as Determinants of Cardiovascular Risk The Framingham Experience Peter W. F. Wilson, MD; Ralph B. D’Agostino, PhD; Lisa Sullivan, PhD; Helen Parise, PhD; William B. Kannel, MD Background: To our knowledge, no single investiga- tus (men: RR, 1.46; women: RR, 1.75). New hypercho- tion concerning the long-term effects of overweight sta- lesterolemia and diabetes mellitus were less highly tus on the risk for hypertension, hypercholesterolemia, associated with excess adiposity. The age-adjusted RR diabetes mellitus, and cardiovascular sequelae has been (confidence interval [CI]) for cardiovascular disease was reported. increased among those who were overweight (men: 1.21 [1.05-1.40]; women: 1.20 [1.03-1.41]) and the obese Methods: Relations between categories of body mass in- (men: 1.46 [1.20-1.77]; women: 1.64 [1.37-1.98]). High dex (BMI), cardiovascular disease risk factors, and vascu- population attributable risks were related to excess weight lar disease end points were examined prospectively in (BMI Ն25) for the outcomes hypertension (26% men; Framingham Heart Study participants aged 35 to 75 years, 28% women), angina pectoris (26% men; 22% women), who were followed up to 44 years. The primary outcome and coronary heart disease (23% men; 15% women). was new cardiovascular disease, which included angina pectoris, myocardial infarction, coronary heart disease, or Conclusions: The overweight category is associated with stroke. Analyses compared overweight (BMI [calculated increased relative and population attributable risk for hy- as weight in kilograms divided by the square of height in pertension and cardiovascular sequelae. Interventions to meters], 25.0-29.9) and obese persons (BMI Ն30) to a ref- reduce adiposity and avoid excess weight may have large erent group of normal-weight persons (BMI, 18.5-24.9). effects on the development of risk factors and cardiovas- cular disease at an individual and population level. Results: The age-adjusted relative risk (RR) for new hy- pertension was highly associated with overweight sta- Arch Intern Med. 2002;162:1867-1872 BESITY COEXISTS with a va- population sample. This setting provided the riety of cardiovascular opportunity to estimate the relative risks risk factors and has been (RRs) and the population attributable risk related to greater cardio- percentage associated with adiposity. Ear- vascular risk in a variety lier research has emphasized the health risks Oof observational studies.1,2 Less well stud- of obesity and analyzed the effects of vari- ied is the relation between categories of adi- ables across a continuum, using relative posity, the development of metabolic risk weights, skin folds, waist girths, or other factors, and cardiovascular sequelae in a measures of adiposity as the factors under population setting, focusing on individu- study.4,5 A reliable referent weight has been als with average weight for height.2,3 Such missing from such an approach, and refer- an approach requires data with repeated ence standards have been moving targets6 measures of weight, risk factor status, and as criteria have changed over time.7 We cardiovascular disease (CVD) incidence. strike a medium stance, using “normal” lev- The long-term experience of the Framing- els of weight now recommended by obe- ham Heart Study cohort met these require- sity experts, a category that includes about ments and included up to 44 years of fol- 20% of the population.8 With the recently low-up to study the relation between body adopted broad reference standard of nor- mass index (BMI) and the development of mal weight (BMI [calculated as weight in From the Boston University CVD and intervening risk factors in middle- kilograms divided by the square of height School of Medicine (Drs Wilson and Kannel), and the aged men and women. in meters], 18.5-24.9) it is likely the RRs of Department of Mathematics, Although some investigations have fo- the health hazards we report can be ex- Boston University cused on extreme levels of obesity and their tended to other population groups. There (Drs D’Agostino, Sullivan, and sequelae, the Framingham Heart Study ex- are constraints with this approach, and what Parise), Boston, Mass. perience represents a community-based to do with underweight persons (BMI (REPRINTED) ARCH INTERN MED/ VOL 162, SEP 9, 2002 WWW.ARCHINTERNMED.COM 1867 ©2002 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/30/2021 mmol/L) or greater or reported use of lipid-lowering therapy Table 1. Characteristics of Participants* since the last clinic examination.15,16 Women who reported no menstrual periods during the 12 months prior to a clinic ex- BMI Category, kg/m2 amination were considered menopausal from that time for- ward. Data from subsequent examinations were used to reclas- Characteristic 18.5-24.9 25.0-29.9 Ն30.0 sify subjects according to their current age, BMI, diabetes mellitus Age distribution, y status, serum cholesterol level, cigarette smoking, and hyper- 35-45 23/24 20/14 19/12 tension status. 46-55 30/30 31/28 30/27 Participants were monitored over 2 years after each ex- 56-65 29/28 31/34 32/36 amination for the occurrence of an initial CVD event, and par- 66-75 18/19 18/25 19/26 ticipants were followed up to 44 years for changes in risk fac- Prevalent conditions tor status. The coronary heart disease (CHD) end points Diabetes mellitus 7/9 9/10 13/13 considered were angina pectoris, myocardial infarction, and Hypertension 15/17 23/29 36/45 coronary death. Myocardial infarction and coronary death were Elevated cholesterol 32/45 38/53 40/51 grouped together as “hard CHD,” and the initial occurrence of Cigarette smoker 58/43 43/28 43/23 any of the 3 CHD end points was labeled “total CHD.” Cere- No. of person-years 15 484/30 888 23 026/21 806 5950/9366 of observation brovascular disease included the occurrence of new strokes and transient ischemic attacks, and the end point total CVD was used for persons who developed either CHD or cerebrovascu- *Data are percentage of participants (men/women) unless otherwise specified. BMI indicates body mass index. lar disease. The end points for cardiovascular and cerebrovas- cular disease were adjudicated by senior Framingham Study sci- entists during follow-up, and the diagnostic criteria for the Ͻ18.5) is problematic. We elected to exclude them be- clinical events have been published elsewhere.17 cause previous experience has shown that this category was Age-specific incidence rates for the first occurrence of hy- likely to include heavy cigarette consumers, those with se- pertension, hypercholesterolemia, diabetes, myocardial infarc- 9-12 tion, angina pectoris, cerebrovascular disease, total CHD, and total vere chronic diseases, and persons with malignancies. CVD were calculated according to the most recently determined Our observations spanned several decades from 1948 BMI category. Analyses were age-adjusted and multivariable ad- onward and focused on multiple measures of BMI. Un- justed, including the variables age, hypertension, hypercholes- fortunately, abdominal girth, triglyceride and lipopro- terolemia, and cigarette smoking. Adjustment for menopausal sta- tein cholesterol levels, and other variables now com- tus was also included for women in multivariable analyses. monly used in vascular risk assessment were not The ␤ coefficients in the pooled logistic regression analy- determined at these 2-year intervals. Age was taken into ses were exponentiated to estimate the RR. The SE of the ␤ co- account in bivariate statistical models and included in efficients was used to calculate the 95% confidence intervals 18 all of the multivariable models, and in some instances (CIs) of the RR estimates using published methods. Popula- age categories were used to investigate the development tion attributable risk (PAR) was calculated from the category- specific RR estimate and the prevalence of the factor using the of risk factors. following equation: PARTICIPANTS AND METHODS PAR=Proportion of Cases Exposed to the Factor ϫ100ϫ(RR−1)/RR.19,20 Members of the original Framingham cohort were eligible for the present study, and the original population sample in- When the RR estimate was less than 1.0, the PAR estimate was 19-21 cluded 5209 participants aged 30 to 62 years at the initial ex- negative, representing a potentially preventive effect. The amination in 1948-1951. We used the technique of pooled re- PAR values for the overweight and obese categories were added peated measures,13 an approach that allowed individuals to together to estimate the effects of a composite overweight cat- contribute multiple person examinations to the analysis as long egory on risk factor development and the occurrence of CVD. as they met the inclusion criteria at the beginning of each ob- servation interval. Specifically, persons free of CVD at an ex- RESULTS amination and with a BMI of 18.5 or greater were eligible for 9-12 the next period of observation. Weight, blood pressure, se- Characteristics of the participants are given in Table 1 for rum cholesterol, cigarette smoking in the previous year, meno- men and women according to categories of BMI. The person- pausal status in women, and interim medication use were as- certained for each participant at each biennial clinic examination. years of experience (bottom row of Table 1) reflect the cross- Weight was determined to the nearest pound on a physi- sectional pooling of the data over the 44 years of follow- cian’s scale. Stature, measured to the nearest quarter inch at the up. The age entries within the columns give the percentage initial Framingham Heart Study examination, was used. Height distribution of the follow-up experience within a given BMI and weight were converted to meters and kilograms, allowing category. For instance, men aged 56 to 65 years com- the calculation of BMI for participants at each examination.