Impact of Overweight on the Risk of Developing Common Chronic Diseases During a 10-Year Period FREE

IN THE UNITED STATES, obesity has become a serious public health problem. According to the third National Health and Nutrition Examination Survey, 32% of adults in the United States are overweight and an additional 22.5% are obese.¹ Moreover, the prevalence of obesity has been increasing sharply among children and adults during the past 3 decades, and the trend is expected to continue.²

Overweight and obesity are risk factors for cardiovascular disease,³ certain cancers,^4- 6 diabetes,⁷ and mortality.^8- 9 In addition, overweight also exacerbates many other chronic diseases, such as hypertension,¹⁰ osteoarthritis,^11- 12 gallstones,¹³ dyslipidemia, and musculoskeletal problems.^14- 15 Unfortunately, comparison of results across epidemiologic studies has been difficult because of the lack of consensus on weight categories. Further complicating the issue is that the new US dietary guidelines and the World Health Organization now define overweight as a body mass index (BMI) (calculated as weight in kilograms divided by the square of height in meters) of 25 or more, a major shift downward from the old US cutoffs of 27.3 for women and 27.8 for men. The US dietary guidelines classify BMI as follows: less than 18.5 is underweight, 18.5 to 24.9 is the healthy weight range, 25.0 to 29.9 is moderately overweight, and 30.0 or more is severely overweight or obese.¹⁶ Since a relatively large proportion of the population has a BMI between 25.0 and 27.0, the change in cutoff for overweight has resulted in a substantial increase in the prevalence of overweight. Moreover, the change has resulted in some confusion regarding the health risks associated with overweight (ie, BMI of 25.0-29.9), since a sizable proportion of this group was previously considered to have a healthy weight.

The majority of articles with original data, as opposed to review articles, on the health risks of obesity have limited themselves to a few health outcomes. Although this is the standard approach for epidemiologic articles, providing data on multiple outcomes is the most efficient way to illustrate the impact of overweight on physical health. To assess the adverse impact of overweight, we have analyzed the 10-year associated risks of developing high cholesterol level, hypertension, gallstones, type 2 diabetes, heart disease, stroke, and colon cancer among women in the Nurses' Health Study and men in the Health Professionals Follow-up Study, 2 large ongoing prospective cohort studies.

The Nurses' Health Study was established in 1976, when 121 701 female registered nurses from across the United States, aged 30 to 55 years, answered a mailed questionnaire on risk factors for cancer and heart disease. Questionnaires, mailed to these women every other year since 1976, ask about diagnosis during the past 2 years, as well as a wide variety of lifestyle factors including diet, physical activity, smoking, and contraception.

For diseases of particular interest, we write to the nurse to obtain permission to review the medical records pertaining to the diagnosis. In addition, all women reporting diabetes are sent a supplementary questionnaire to obtain additional information regarding the symptoms and diagnosis. Several repeated mailings are sent to nonrespondents, and these are followed by telephone interviews. In instances where we cannot obtain the medical records, we attempt to acquire as much information as possible to code a disease as probable, even if definite confirmation cannot be achieved. For deceased participants, we write to the next of kin to obtain permission to review the medical records. The records are reviewed by medically trained personnel according to established criteria.

The Health Professionals Follow-up Study is a prospective study of 51 529 men, aged 40 to 75 years when the study began in 1986. Follow-up questionnaires are sent biennially to update information on exposure and disease. The follow-up has been greater than 90% through the 1994 questionnaire cycle. The follow-up procedures for medical diagnoses are similar to those used in the Nurses' Health Study.

Diagnoses of hypertension, high cholesterol level, and diabetes are not confirmed by medical record review. However, we have conducted validation studies of persons in the Nurses' Health Study and Health Professionals Follow-up Study cohorts and found that the diagnoses are accurately reported. Among a sample of women participating in the Nurses' Health Study who reported a diagnosis of hypertension and participated in the validation study, all had blood pressure greater than 140/90 mm Hg.¹⁷ In an age-stratified sample of 161 nurses in the Nurses' Health Study who lived in the greater Boston area and had never reported a diagnosis of hypertension, only 7% had blood pressure greater than 140/90 mm Hg, but none was greater than 160/95 mm Hg. In the same validation study, there was an approximately 14% false-positive rate for hypercholesterolemia. The false-negative rate was estimated to be approximately 15.5% among a sample of the 181 nurses in the Boston, Mass, area without a previous self-reported diagnosis of hypercholesterolemia.

The validity of a self-reported diagnosis of diabetes was investigated in a sample of 61 women participating in the Nurses' Health Study. The diagnosis of type 2 diabetes mellitus was confirmed by medical records in 60 (98.4%) of the women. In another substudy to assess the prevalence of undiagnosed diabetes, fasting plasma glucose and plasma fructosamine were measured in a random sample of participants who did not report a previous diagnosis of diabetes. Only 1 (0.5%) of the women had an elevated fasting plasma glucose or plasma fructosamine level in the diabetic range, and her levels were barely above the diagnostic cutoffs. By confirming all self-reported cases of diabetes, we exclude false-positive results, and the Nurses' Health Study results suggest that the false-negative rate is low. Therefore, we feel confident that the women in the sample are correctly classified by the self-report questionnaire and subsequent follow-up.

The validity of self-report of diabetes has been tested in a subsample of 71 men from the Health Professionals Follow-up Study who reported newly diagnosed diabetes between 1996 and 1998. A physician blinded to the information reported on the supplementary questionnaire reviewed the records according to the revised diagnostic criteria. Of the 71 patients, 12 had incomplete records, eg, absent laboratory data (n = 3) or 1 set of laboratory data only (n = 9). All 12 had records that were strongly suggestive of diabetes mellitus. For example, despite missing 1 or more days of laboratory data, a necessary piece of information for confirming the diagnosis, 7 men had a physician's diagnosis and 3 reported a history of using diabetic medication. Among the remaining 59 cases with complete data, the diagnosis of type 2 diabetes was confirmed in 57 (97%).

Women in the Nurses' Health Study and men in the Health Professionals Follow-up Study were excluded from the analysis if they reported a diagnosis of cancer before 1986 (3127 women and 2189 men); were missing information on demographics (31 women and 52 men), smoking (235 women and 1941 men), or BMI in 1986 (39 404 women and 1103 men); were underweight (BMI in 1986 <18.5; 1131 women and 151 men); or had outlying BMI (83 women and 33 men), leaving 77 690 women and 46 060 men for analysis. In addition, for all incident analyses, prevalent cases of the specific outcome or outcomes were excluded (eg, prevalent cases of diabetes were excluded from analyses of the incidence of diabetes).

All analyses were conducted with SAS software.¹⁸ Logistic regression was used to assess the association between overweight and the 10-year risk of developing high cholesterol level, hypertension, gallstones, type 2 diabetes, heart disease, stroke, or colon cancer. All logistic regression models adjusted for age (in 10-year strata), smoking (never, past, or current), and race or ethnic group (white, African American, Hispanic, Asian, other, or missing). The BMI in 1986 was entered as a series of dichotomous variables, with 18.5 to 24.9 serving as the referent. In the primary analyses, the other BMI categories were 25.0 to 29.9, 30.0 to 34.9, and 35.0 or more. To investigate whether the risk was uniformly low in the referent group, subanalyses were run with the referent group restricted to a BMI of 18.5 to 21.9, and an additional dichotomous term for BMI of 22.0 to 24.9 was added to the logistic regression models. To assess the risk of developing more than 1 condition, the sample was stratified by the presence of a diagnosis of hypertension or high cholesterol level, and the proportion of participants with 0, 1, or more than 1 diagnosis was calculated within the category of BMI (ie, 18.5-24.9, 25.0-29.9, 30.0-34.9, and ≥35.0). All P values are 2-sided, with P<.05 considered statistically significant.

The mean age in the 2 cohorts was similar (52.9 vs 54.5 years for women and men, respectively), and both cohorts were predominantly white (≥93%); however, the prevalences of obesity (BMI in 1986, ≥30.0) (14.8% vs 8.2%) and current smoking (21.1% vs 9.9%) were higher among the women than the men (Table 1).

During 10 years of follow-up, more than half of the men and women were diagnosed as having high blood pressure (16% of women and 19% of men) or high cholesterol level (49% of women and 48% of men) (Table 2). In addition, approximately 2% of the women and 3% of the men developed diabetes. Gallstones were diagnosed more often during a 10-year period among the women (7.8% vs 3.5%), whereas heart disease was more common among the men (5.2% vs 3.4%).

In both cohorts, the risk of developing diabetes, gallstones, hypertension, heart disease, and stroke increased with severity of overweight. Among both women and men, those with a BMI of 35.0 or more were approximately 20 times more likely to develop diabetes (relative risk [RR], 17.0; 95% confidence interval [CI], 14.2-20.5 for women; RR, 23.4; 95% CI, 19.4-33.2 for men) than their same-sex peers with a BMI between 18.5 and 24.9. Moreover, adults who were overweight but not obese (ie, BMI between 25.0 and 29.9) were more than 3 times as likely as their leaner peers to develop diabetes during 10 years (Table 3).

Women who were overweight but not obese were also significantly more likely than their leaner peers to develop gallstones (RR, 1.9; 95% CI, 1.7-2.0), hypertension (RR, 1.7; 95% CI, 1.6-1.7), high cholesterol level (RR, 1.1; 95% CI, 1.1-1.2), and heart disease (RR, 1.4; 95% CI, 1.2-1.5). The associations with the development of colon cancer and stroke were elevated but not significant. The results were similar in men, with the exception that obese men were significantly more likely to have a stroke (RR, ≥2.0) during the 10 years of follow-up. Among both women and men, the risks of developing diabetes, gallstones, hypertension, and heart disease increased with severity of overweight.

To test whether the risk of developing an outcome was uniformly low in the healthy weight range, the reference group was redefined as having a BMI between 18.5 and 21.9, and the risks associated with a BMI between 22.0 and 24.9 were assessed in the logistic regression model. Women and men (Table 4) with a BMI between 22.0 and 24.9 were significantly more likely than their leaner peers to develop at least 1 of the morbidities. In addition, the risks associated with higher BMIs (ie, ≥25.0) increased as a result of restricting the reference group. For example, women with a BMI of 35.0 or more were at a 17-fold increase in risk of being diagnosed as having diabetes when the reference category was 18.5 to 24.9; that risk increased to 30-fold when the reference group was redefined as 18.5 to 21.9.

Not only did the risk of developing diabetes, gallstones, hypertension, colon cancer, heart disease, and stroke increase with degree of overweight (Table 3), but also the risk of developing more than 1 outcome increased with weight category. Among women and men with a diagnosis of hypertension or high cholesterol level, the risk of developing additional morbidities increased across categories of BMI (Table 5). For example, the proportion of women with 1 or more incident morbidities increased from approximately 11% among the leanest women to 25% among the heaviest (P<.001) (Table 5). A similar pattern was seen among women and men without a history of hypertension or high cholesterol level. Among the men, but not the women, the risk of developing morbidities was lower in participants without a history of hypertension or high cholesterol level. Nevertheless, the proportion who developed 1 or more diseases increased with BMI.

During 10 years of follow-up, the incidence of diabetes, gallstones, hypertension, and heart disease in both men and women, and of colon cancer (women only) and stroke (men only), increased with BMI in 1986. Even adults who were overweight but not obese (ie, 25.0≤BMI≤29.9) were significantly more likely than their leaner peers to develop 1 or more of these diseases.

Although there has been debate about whether the relationship between weight and risk of death is linear, J-shaped, or U-shaped, the results have consistently shown that adults with BMI greater than 30.0 are at increased risk of death.^8,19- 21 Among 45- to 75-year-old men and women in the American Cancer Society's Cancer Prevention Study I, the risk of death increased linearly with BMI among the never smokers. The risk was particularly pronounced for death from cardiovascular disease and among men.²¹ Manson et al⁸ also observed a linear association between BMI and mortality during 16 years of follow-up among 15 195 women who were never smokers in the Nurses' Health Study. However, several other studies have observed that the risk was elevated in low-weight, as well as in overweight, adults²⁰ or did not increase significantly until BMI was greater than 27.0.²²

Although mortality is a clearly defined outcome, the results of mortality analyses can be difficult to interpret. Except for diseases that are almost always fatal regardless of treatment, mortality is a function of incidence of disease, stage of illness at diagnosis, and effectiveness of treatment. Many forms of cancer and cardiovascular disease are treatable by either pharmacotherapy or intervention (ie, angioplasty or surgery); thus, the relationship between excess weight and death from cancer or cardiovascular disease does not necessarily translate to the same relationship with the development of cancer or cardiovascular disease. Moreover, several chronic diseases, such as hypertension and hypercholesterolemia, increase the risk of cardiovascular disease, but these conditions rarely are the direct cause of death. Nevertheless, both conditions have substantial economic and emotional costs because of their high prevalence. Because of the focus on mortality rather than morbidity as the outcome, less lethal diseases that may have important costs associated with them have a minimal impact on the results, thus highlighting why morbidity is an important outcome when the effects of obesity are studied.

Our group previously reported on the health consequences of obesity observed in the Nurses' Health Study^3- 4,7- 8 and the Health Professionals Follow-up Study.^23- 25 Other investigators have also observed that overweight is predictive of developing type 2 diabetes,^26- 27 hypertension,^28- 29 dyslipidemia,³⁰ coronary heart disease,³¹ stroke,³² gallbladder disease,³³ osteoarthritis,^34- 35 sleep apnea and respiratory problems,³⁶ and certain cancers.³⁷ Although an increase in risk is found across studies, the magnitude of the risk is difficult to compare because it has been classified in a variety of ways. Because there is a near-linear relationship between BMI and risk of developing type 2 diabetes, hypertension, and gallstones, the choice of how the reference group is defined can have a large impact on the results. The higher the BMI cutoff is for the reference group, the lower the risks will appear to be in the higher weight groups because people at increased risk have been placed in the reference group. One strength of our study is that the same BMI cutoffs were used as predictors of 7 morbidities. Morever, the weight classification we used was that espoused by the US dietary guidelines,¹⁶ so the categories have inherent meaning.

Our observation that women and men who have a BMI between 25.0 and 29.9 are at significantly increased risk of developing numerous health conditions offers support for lowering the overweight cutoff from a BMI of 27.0 to that of 25.0 in the recent weight guidelines from the US Department of Agriculture. Moreover, our finding that men and women in the upper half of the healthy weight category (ie, BMI between 22.0 and 24.9) are significantly more likely than their leaner peers to develop health problems highlights the necessity of counseling patients who are in the healthy weight range to maintain their weight or increase their physical activity levels. Because few people are able to lose weight and maintain the weight loss, it is important to intervene early, before the person has developed a serious weight problem. Although treatment for obesity is recommended only for overweight individuals with a BMI of 30.0 or more, or of 27.0 or more if they have other risk factors for disease,³⁸ it may be prudent to counsel slightly leaner patients if they are gaining weight, as well as patients who are overweight but do not meet the treatment cutoffs (ie, BMI of 25.0-26.9), on how to modify their diet and physical activity to control their weight.

Accepted for publication December 4, 2000.

This study was supported by Roche Pharmaceutical, Basel, Switzerland, and Nutley, NJ, and by research grants CA40356, HL35464, and CA55075 from the National Institutes of Health, Bethesda, Md. In addition, Drs Field, Laird, and Colditz were partially supported by the Boston Obesity Nutrition Research Center (grant DK 46200).

Corresponding author and reprints: Alison E. Field, ScD, Channing Laboratory, 181 Longwood Ave, Boston, MA 02115 (e-mail: Alison.Field@channing.harvard.edu).