Analysis of effect of amount of weight lost. The x’s indicate that a weight measurement was recorded.
Analysis of effect of maintenance of weight loss. The x’s indicate that a weight measurement was recorded.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Moore LL, Visioni AJ, Qureshi MM, Bradlee ML, Ellison RC, D’Agostino R. Weight Loss in Overweight Adults and the Long-term Risk of Hypertension: The Framingham Study. Arch Intern Med. 2005;165(11):1298–1303. doi:10.1001/archinte.165.11.1298
Few studies address long-term effects of weight loss on risk of incident hypertension among overweight adults.
We evaluated weight loss among 623 overweight (body mass index [calculated as weight in kilograms divided by the square of height in meters] ≥25) middle-aged (aged 30-49 years) and 605 overweight older (aged 50-65 years) adults in Framingham, Mass. Subjects were classified first according to amount of weight lost over 4 years: (1) weight changed by less than 1.8 kg (stable weight), (2) lost 1.8 to less than 3.6 kg, (3) lost 3.6 to less than 6.8 kg, and (4) lost 6.8 kg or more. We also classified weight loss according to whether it was sustained during the next 4 years.
After adjusting for age, sex, education, baseline body mass index, height, activity, smoking, and alcohol intake, weight loss of 6.8 kg or more led to a 21% to 29% reduction in long-term hypertension risk. After adjusting for cancer or cardiovascular disease occurring during follow-up, weight loss of 6.8 kg or more led to a 28% reduction in risk (relative risk [95% confidence interval], 0.72 [0.49-1.05]) for middle-aged adults and a 37% reduction (0.63 [0.42-0.95]) for older adults. Sustained weight loss led to a 22% reduction in hypertension risk (0.78 [0.60-1.03]) among middle-aged and a 26% reduction (0.74 [0.56-0.97]) in older adults. This risk reduction was strengthened by adjustment for prevalent cancer or cardiovascular disease during follow-up.
A modest weight loss, particularly when sustained, substantially lowers the long-term risk of hypertension in overweight adults.
An estimated 50 million adults in the United States have high blood pressure.1 The strongest known risk factor for this disease is excess body weight,2-4 and its adverse effects have been shown to start in childhood.5,6 The treatment and primary prevention of hypertension have received a great deal of attention in recent years.1,7-9 Most trials of hypertension treatment have included a weight-loss component, and these studies have for the most part shown that weight loss, whether alone or in combination with antihypertensive medication use, has a beneficial effect on blood pressure control.10-15 A few studies examined the effects of weight loss among those with borderline high blood pressure or those who were otherwise prone to hypertension.16-18 In all of these studies, weight loss reduced both systolic and diastolic blood pressure, although in 2 of the 3 studies examined, the effects diminished over the 3- to 4-year follow-up.17,18 One study16 demonstrated a 50% reduction in incident hypertension over 5 years associated with a lifestyle intervention that included exercise, weight loss, and other dietary changes. A prospective epidemiologic study19 examined self-reported weight loss from age 18 years to middle adulthood and found an overall reduction in the risk of subsequent hypertension. In that study, the authors were unable to compare sustained and nonsustained weight loss and did not specifically examine the potential benefits of weight loss among overweight individuals.
Obesity is notoriously difficult to treat; 90% of those who lose weight have been reported to regain it.20 The weight-loss goals of many overweight adults may be extreme, making it even more difficult to achieve and maintain weight loss over time. Data on the long-term health implications of modest weight loss are sparse, and data evaluating the separate effects of sustained and nonsustained weight loss are needed.
Earlier analyses from the Framingham Study showed that change in weight was associated with a linear change in blood pressure2,21 and that sustained weight loss was associated with a 30% reduction in the risk of diabetes mellitus in overweight adults.22 The goal of the current study was to estimate the effects of both the amount of weight lost and the persistence of the weight loss on the risk of incident hypertension among already-overweight adults.
The surviving members of the original 5209 subjects in the Framingham cohort have been examined every 2 years since 1948. At each clinical examination, subjects completed a structured interview that included a detailed medical history and risk-behavior assessment, a physical examination, and laboratory and other measurements. All subjects provided yearly written informed consent to participate, as mandated by the institutional review board of Boston Medical Center, Boston, Mass.
Blood pressure was measured by an examining physician using a standard mercury sphygmomanometer and an appropriate-sized cuff, after the subject had had a brief period of rest in the seated position. Two measurements were taken on the left arm, 2 minutes apart. Systolic blood pressure was defined as the first appearance of sound (Korotkoff, phase 1), and diastolic blood pressure as the disappearance of sound (Korotkoff, phase 5). The mean of the 2 physician measurements was taken to reflect the systolic and diastolic blood pressures at each examination.
The outcome of interest for all analyses was incident hypertension, using criteria modified slightly from the seventh report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 7).23 All subjects with a mean systolic blood pressure of 140 mm Hg or higher and/or a mean diastolic blood pressure of 90 mm Hg or higher on 2 separate examination visits were considered to be hypertensive. The JNC 7 criteria require that the subject have elevated blood pressure readings on 2 separate visits approximately 6 weeks apart. The Framingham Study examinations, however, are at 2-year intervals. We were concerned that the use of measurements from a single examination visit to define subjects as being hypertensive using the lower JNC 7 standards would lead to unacceptably high numbers of false-positive cases. Therefore, we required subjects to meet the diagnostic criteria at 2 consecutive Framingham Study examinations. However, subjects taking antihypertensive medication (for the purpose of blood pressure lowering) were considered to have hypertension at the examination at which such medication use was first reported. In addition, those having frank hypertension based on a mean systolic blood pressure of 160 mm Hg and/or a mean diastolic blood pressure of 95 mm Hg or higher from the 2 separate physician measurements at a single examination visit were also considered to be hypertensive.
The following potential confounding variables were examined in these analyses: age (in years), sex, height (in meters), baseline body mass index (BMI) (calculated as weight in kilograms divided by the square of height in meters), education level (dichotomized as >high school vs ≤high school), alcohol intake (mean number of drinks per day during the weight-change period), cigarette smoking (mean number of cigarettes per day during the weight-change period), and physical activity (mean activity index score during the weight-change period). In a final model, we also adjusted for any occurrence of cancer or cardiovascular disease during the follow-up period to determine whether such comorbid conditions associated with weight loss might confound the results.
Height and weight were measured without shoes at each visit, with subjects wearing an examination gown; a standard balance beam scale was used. To reduce the error associated with measurement of height, we calculated the mean of all available adult height measurements through age 60 years; mean height was combined with examination-specific weights to estimate BMI at each visit. Physical activity and alcohol intake were assessed at periodic examinations. Subjects reported the number of hours spent each day sleeping and in sedentary, light, moderate, and vigorous physical activity. An index of moderate plus vigorous physical activity was created by summing the number of self-reported hours per day spent participating in each level of activity multiplied by a numeric weight derived from the estimated oxygen consumption required (in liters per minute) for that activity level.24 The self-reported daily or weekly consumption of beer, wine, and spirits was used to estimate the average number of drinks per day. Cigarette smoking (average number of cigarettes per day and information on starting and stopping) was assessed at each examination. For those variables not assessed at every examination, values at the intermediate examinations were imputed by averaging reported values from adjacent examinations.
Of the 3630 men and women who were followed up in the Framingham Study at some time during their middle-adult years (aged 30-49 years), 2188 were overweight (BMI ≥25) at some time during that age interval. We excluded 279 subjects who had incomplete weight-change data and 55 individuals who had missing covariate information. Of the remaining 1854 subjects, 21 died before the start of follow-up, 850 were excluded owing to prevalent hypertension, and 2 were lost to follow-up. To reduce the possibility of confounding by a preexisting disease, we excluded an additional 10 subjects with prevalent diabetes, 30 with prevalent cardiovascular disease, and 17 with prevalent cancer. Of the remaining 924 overweight disease-free subjects, 623 had either stable weight or lost weight during the next 4 years and were included in our analyses (Figure 1).
In a secondary analysis, we classified those subjects who lost weight over 4 years according to whether they kept the weight off during the next 4 years. In this manner, we classified each of the weight losers from the first analysis as having sustained or nonsustained weight loss. Finally, we excluded subjects from the referent category (stable weight) if their weight did not remain stable during the weight maintenance period (Figure 2). This yielded a final sample of 417 subjects for the analysis of the effects of sustained and nonsustained weight loss.
We carried out the same analyses for overweight older adults (aged 50-65 years). Applying the same inclusion/exclusion criteria yielded a sample of 605 overweight subjects for the analysis of the amount of weight lost and 424 subjects for the analysis of sustained and nonsustained weight loss.
We selected subjects for the 2 age groups of interest (aged 30-49 years and 50-65 years) by examining each subject’s BMI measures at sequential examination visits. On the first occasion at which the subject reached a BMI of 25 or higher, he or she was selected for the age cohort and that examination served as the baseline examination. Thus, the baseline examination visit varies among subjects.
Figure 1 shows the timing of the exposure and follow-up periods for the analysis of the amount of weight lost. We used 3 weight measurements from a 4-year period to calculate subject-specific weight slopes during that interval, using a simple linear regression model with sequential measures of weight regressed on age. The estimated weight loss from the linear regression model was used to assign subjects to 1 of the 4 weight loss categories (stable weight, lost 1.8 to <3.6 kg, lost 3.6 to <6.8 kg, lost ≥6.8 kg). Subjects who gained 1.8 kg or more were excluded from this analysis. We excluded the first 4 years of person-time after the end of the weight-change period to remove the possibility that serious preexisting illnesses might have been responsible for the weight loss. Thus, follow-up for incident hypertension began 8 years after the baseline weight measurement.
Hypertension incidence was calculated as the number of incident cases of high blood pressure divided by the total number of person-years of follow-up. Total follow-up time for each subject continued until the occurrence of 1 of the following censoring events: incident hypertension diagnosis, death, loss to follow-up, or examination 24 (the end of our follow-up period). Hypertension incidence was calculated separately for those whose weight was stable and for each category of weight loss. We used Cox proportional hazards analyses to estimate the adjusted relative risk (RR) and 95% confidence intervals (CIs) associated with the amount of weight lost.25
For those subjects who lost 1.8 kg or more over 4 years, we then examined weight change during the next 4 years (the weight maintenance period) to allow for the assessment of the effects of sustained and nonsustained weight loss (Figure 2). Those who kept the weight off or who lost more weight during the weight maintenance period were considered to have had sustained weight loss. Those who regained the weight (ie, regained ≥1.8 kg) were considered to have had nonsustained weight loss. For this secondary analysis, subjects whose weight remained within 1.8 kg of the baseline value throughout the entire 8-year exposure period were considered to have stable weight (referent category). Follow-up for hypertension occurrence began at the end of the weight maintenance period.
Table 1 provides baseline characteristics of the 623 overweight middle-aged adults and 605 overweight older adults who either lost weight or had stable weight over 4 years. In both age groups, those who lost the most weight were heaviest at baseline and more frequently female.
In Table 2, we examine the effects of the amount of weight lost over 4 years. In both age groups, the crude rates of hypertension were highest among those overweight men and women who did not lose weight despite the fact that their BMI values were lowest at baseline. As expected, after 4 years of follow-up, the final BMI for those with stable weight was unchanged from the baseline BMI. Those who had lost 6.8 kg or more, however, reduced their mean BMI by more than 3 units, giving them a final BMI that was about 2 units lower than the final BMI of those who had stable weight. As would also be expected, older men and women had higher rates of hypertension than did middle-aged subjects regardless of weight-change category.
In Table 2, we present 2 adjusted models. In the first, we adjusted for age, sex, height, baseline BMI, education level, alcohol intake, cigarette smoking, and physical activity. Here we see that middle-aged overweight adults who lost 6.8 kg or more reduced their long-term risk of hypertension by 21%, while older adults reduced their risk by 29%. In the final model, after accounting for the occurrence of cancer or cardiovascular disease during the follow-up period, the results were strengthened. We see that there was a trend toward a reduction in the long-term risk of hypertension even among those who lost more modest amounts of weight.
In Table 3, we examined the effects of sustained and nonsustained weight loss. Sustained weight loss in middle-aged overweight subjects resulted in 7 fewer cases of hypertension per 1000 person-years than did maintaining a stable weight (45.1 − 38.1 cases per 1000 person-years). Sustained weight loss in older subjects resulted in 13.3 fewer cases per 1000 person-years. In our first adjusted model, we found that middle-aged men and women who had a sustained weight loss of 1.8 kg or more had a 22% reduction in the long-term risk of hypertension (RR, 0.78; 95% CI, 0.60-1.03), whereas older subjects had a 26% reduction in risk (RR, 0.74; 95% CI, 0.56-0.97). Accounting for the occurrence of comorbidities during the follow-up period strengthened the results for older adults, where sustained weight loss was associated with a 36% reduction in the long-term risk of hypertension.
In this study, we examined the effect of weight loss among overweight middle-aged and older adults on the long-term risk of hypertension. The results of this study suggest that a weight loss of 6.8 kg or more can reduce the long-term risk of hypertension by 21% to 29%. Weight loss of even smaller amounts is effective, particularly if the weight loss is sustained. The results of these analyses also suggest that there may be some benefit to weight loss even when it is not sustained; there is no evidence that regaining weight has an adverse effect on hypertension risk.
This study adds important information to the existing literature and suggests that there may be significant long-term blood pressure benefits of modest weight loss in middle-aged and older men and women. Although earlier studies have demonstrated that weight loss reduces blood pressure in those with prevalent hypertension as well as those at high risk for developing hypertension,10-18 there are few data on weight loss and the long-term risk of primary hypertension among normotensive adults.19 These findings may provide useful data for clinicians contemplating weight-loss recommendations for their overweight patients.
An important strength of the Framingham Study data set is the availability of repeated measures of height and weight over many years, allowing us to classify subjects more accurately with respect to their BMI, as well as their change in weight over time. Second, the routine monitoring and replicate standardized measurement of blood pressure provide for more accurate detection of incident hypertension. Both factors enhanced our ability to obtain a more accurate and precise estimate of the effect of weight loss on long-term hypertension risk. Finally, the repeated measures of weight enabled us to separate the effect of weight loss that was sustained from that which was not sustained.
Unfortunately, this observational study does not allow us to separate voluntary from involuntary weight loss. Because we were concerned that a large weight loss or persistent weight loss might be involuntary and signal the presence of undetected preclinical illnesses, we created a final multivariable model in which we adjusted for serious comorbidities occurring during the follow-up period. In this way, we hoped to reduce some of the confounding associated with involuntary weight loss. The comorbid condition that resulted in the greatest change in the effect estimates was cancer and, in fact, a final model including cancer alone was very similar to that including both cancer and cardiovascular disease.
There are a number of mechanisms by which weight loss may affect the risk of hypertension. Although it is not clear how obesity may cause high blood pressure, there are a number of important physiological changes that accompany increased body weight. Obesity is associated with higher levels of insulin resistance, as well as hyperinsulinemia, rises in cardiac output, increases in triglyceride and cholesterol levels, and increases in sympathetic nervous system activity.26,27 Most of these changes have also been associated with increases in blood pressure. In recent years, there has been a great deal of focus on the roles of hyperinsulemia and insulin resistance in the development of hypertension.27-30 These factors may increase blood pressure in a number of ways: via an increase in circulating epinephrine or norepinephrine, increased vascular sensitivity to vasoconstrictor or vasodilator substances, enhanced sodium retention by the distal renal tubules, and other mechanisms associated with intracellular retention of free calcium or sodium (in turn, stimulating vascular smooth muscle contraction).
In addition, weight loss and concurrent changes in diet and physical activity may have direct and indirect effects on blood pressure. For example, dietary changes directed at weight loss may lead to an overall reduction in sodium intake, while exercise, even in the absence of weight loss, decreases insulin resistance in obese individuals. Together, these factors may help explain some of the long-term reduction in hypertension risk associated with weight loss in this study.
The results of our study show that those who lost the most weight had the lowest BMI values at the end of the weight-change period, and the mean BMI of individuals who sustained their weight loss was more than 2 units lower than their mean baseline BMI. Thus, it is plausible that the beneficial effect of weight loss simply derives from that lower BMI at the end of follow-up. Subjects with a nonsustained weight loss had a final BMI (after 8 years) that was similar to their baseline BMI. Although not statistically significant, there was a trend toward a lower risk of hypertension in the long term for those whose weight loss was not sustained. We postulate that reducing the BMI, even for a few years, may have beneficial physiological effects.
Hypertension is a remarkably common condition among overweight individuals and frequently leads to cardiovascular sequelae, including stroke and myocardial infarction. Prevention or even delay in the onset of hypertension may reduce the occurrence of these devastating diseases. The results of this study suggest that at least 15% of the cases of hypertension in overweight middle-aged adults and 22% of the cases occurring in overweight older adults could be prevented by a modest amount of sustained weight loss.
Correspondence: Lynn L. Moore, DSc, MPH, Section of Preventive Medicine and Epidemiology, Boston University School of Medicine, 715 Albany St, Boston, MA 02118 (email@example.com).
Accepted for Publication: January 5, 2005.
Financial Disclosure: None.
Funding/Support: This work was supported by National Heart, Lung, and Blood Institute’s Framingham Heart Study (Contract No. N01-HC-25195).