Although some prospective cohort studies have shown that baseline BMI is positively associated with a future incident risk for hypertension, these studies do not account for weight changes during the observation period. Therefore, it is not evident whether future incident risk for hypertension in obese, non-hypertensive people increases when their weight remains stable. We examined the association between long-term weight stability and risk for developing hypertension.
A total of 5201 Japanese male workers aged 30–59 years underwent health checkups in 2002 and were followed through 2006. To consider transitions in covariates during the follow-up period, we used a time-dependent covariate Cox proportional hazard model to compute the relative risks (RRs) of incident hypertension. Furthermore, as a complementary analysis, we restricted the data to individuals whose BMI remained unchanged (±5% of baseline BMI) during the follow-up and compared the RRs between BMI categories.
During the follow-up, there were 899 newly diagnosed cases of hypertension among the 5201 men (14 888 person-years). Mean change in BMI during the follow-up period of all subjects was 0.2±1.1 kg/m2 (range: −6.6 to 6.3 kg/m2). The multivariate RRs for hypertension increased as BMI increased when we applied the time-dependent covariate Cox proportional hazard model. The complementary analysis showed that the multivariate RR (confidence interval) within the ⩾27.0 kg/m2 BMI category was 1.43 (1.16–1.77) times higher than the reference of 23.0–24.9 kg/m2, whereas the RR for the <21.0 kg/m2 BMI category was 0.63 (0.51–0.79) times lower than the reference.
A higher baseline BMI increases future incident risk for hypertension even when there has been no major weight increase. Weight management should be encouraged for obese, non-hypertensive people to prevent future hypertension.
As high blood pressure is a frequent cause of major diseases, such as coronary heart disease, cerebrovascular disease and renal diseases, hypertension is one of our most important worldwide health challenges. Its prevalence rate is especially high in the middle aged and elderly populations.1, 2, 3, 4, 5 Many studies6, 7, 8, 9, 10, 11, 12 indicate that obesity is an important risk factor for hypertension. These epidemiological studies6, 7, 8, 9, 10, 11, 12 show that body weight, or BMI, is positively associated with hypertension risk. There are also a few prospective cohort studies6, 11 using Cox proportional hazards regression analysis that show a high BMI is associated with an increased future incident risk for hypertension. Recently, Shuger et al.6 clearly showed that women with a BMI ⩾24.7 kg m−2 had a risk of developing hypertension that was over two times higher than women with a BMI of 18.5–20.0 kg m−2.
On the other hand, some recent prospective cohort studies13, 14, 15, 16, 17, 18, 19 reported that long-term weight gain increases the risk for hypertension. Those studies13, 14, 15, 16, 17, 18, 19 showed that the risk for incident hypertension in subjects whose BMI increased during a follow-up period was significantly greater compared with subjects who maintained or decreased BMI. Given this association between long-term weight gain and hypertension risk, previous prospective cohort studies6, 11 have a statistical problem as the Cox regression analysis using several fixed baseline variables as covariates cannot account for any change in BMI during the observation period.20, 21, 22 As a BMI transition during the observation period certainly influences future incident risk for hypertension, it makes determination of the actual influence of baseline BMI on future risk for hypertension less certain. Therefore, it is not evident whether future incident risk for hypertension in obese, non-hypertensive people increases even when their weight remains stable.
A study by Williams et al.14 appears to be the only published study that considered transitions in BMI during a follow-up period in their analysis of hypertension risk. Williams et al.14 excluded subjects who gained or lost >0.4 kg m−2 of their BMI during ∼8 years of follow-up and demonstrated that a higher BMI was still associated with a higher risk of future hypertension, even if the baseline BMI remained stable during the follow-up period. However, the study,14 part of the National Runners’ Health Survey, involved only extremely active individuals (runners). Moreover, the data mainly consisted of a questionnaire, which means BMIs were calculated using self-reported weights and heights along with self-reported diagnoses of hypertension. In addition, the Cox regression analysis could not be applied because of the limited frequency of measurements in their study.
In our study, we tried to clarify the association between long-term weight stability and hypertension risk using time-dependent Cox regression analysis or restricted data consisting of subjects who did not experience a major BMI change during the observation period.
Participants in this study included 8685 men 30–59 years of age working in the central region of Japan who had completed annual health checkups conducted by the Japan Labor's Culture, Health and Welfare Association in 2002. Data were collected on anthropometric values, blood pressure, blood samples and from interview questionnaires pertaining to smoking habits, daily alcohol intake and medical history.
At the baseline inquiry, we excluded 211 men with a history of stroke, coronary heart disease or diabetes mellitus, and 1978 men with hypertension defined as a systolic blood pressure (SBP)⩾140 mm Hg, a diastolic blood pressure (DBP)⩾90 mm Hg, or a past or current history of antihypertensive medication use. In addition, we excluded 18 men who did not have complete data and 1277 men who could not be followed after their first health checkups. Therefore, a total of 5201 men were followed from 2002–2006 (Figure 1). The mean follow-up period was 2.9 years. The protocol of this cohort study was approved by the Ethical Committee of the Dokkyo Medical University School of Medicine, Japan.
Assessment and criteria of hypertension
At the annual health checkups, a trained nurse measured the SBP and DBP levels of subjects on the right arm using a mercury manometer and a standard protocol after the subjects rested for at least 5 min in the sitting position. Using the first and fifth Korotkoff sounds as indicators for SBP and DBP levels, respectively, these values were obtained in duplicate. Cuff sizes were selected based on upper arm girth and length. Subjects were defined as incident cases of hypertension by meeting one of the following criteria during follow-up: (1) SBP⩾140 mm Hg, (2) DBP⩾90 mm Hg or (3) new antihypertensive medication use.
Assessment of body mass index and other covariates
At the annual health checkups, height was measured in bare feet and weight was measured in light clothing. We calculated BMI as weight (kg) divided by the square of the height (m). Venous blood was drawn for measurements after an overnight fast. The laboratory of FALCO biosystems (Kyoto, Japan), where measurements have been standardized by the College of American Pathologists (CAP), analyzed all blood specimens. We conducted interviews to ascertain disease histories, smoking status (never smoked, former smoker or current smoker) and alcohol intake (never, 1–2 times per week, 3–4 times per week, 5–6 times per week or everyday).
All statistical analyses were conducted using SAS software version 9.2. (SAS Institute, Cary, NC, USA). We classified participants into the following seven categories with regard to their baseline BMI: <18.5, 18.5–20.9, 21.0–22.9, 23.0–24.9, 25.0–26.9, 27.0–29.9 or ⩾30.0 kg m−2. We compared baseline characteristics of subjects from each BMI category using χ2 tests for categorical variables and analysis of variance for continuous variables. Person-years of follow-up were calculated from the date of the baseline survey in 2002 to the day when hypertension was diagnosed or to 2006, whichever came first. We used Cox proportional hazards regression models (PROC PHREG in the SAS procedure) to compute the relative risks (RRs) and 95% confidence intervals (CIs) of incident hypertension relative to BMI categories with a reference category of 23.0–24.9 kg m−2 (recent studies23, 24 indicate that a BMI of 23.0–24.9 kg m−2 is associated with the lowest mortality among the middle-aged Japanese male population). At first, we used the PROC PHREG procedure with several baseline (that is, non-time-dependent or fixed) variables, such as age in years, SBP (mm Hg), log-transformed triglyceride level (mg per 100 ml), hemoglobin A1c (%), smoking status (never smoked, former smoker, current smoker) and alcohol intake (never, 1–2 times per week, 3–4 times per week, 5–6 times per week or everyday). Next, to account for transitions of covariates during the follow-up period, we further analyzed the data using the time-dependent covariate Cox proportional hazard model20, 21, 22 with the BMI, age, SBP, log-transformed triglyceride level, hemoglobin A1c, smoking status and alcohol intake as time-dependent covariates. An explanatory variable is called a time-dependent covariate if its value for any given individual can change over time. The SAS procedure of PROC PHREG allows the Cox proportional hazards regression model to use both fixed (non-time-dependent) and time-dependent covariates. Details on the techniques for using time-dependent covariates in the PROC PHREG procedure are presented elsewhere.25 Briefly, the form x(t) will be used for a covariate that may vary over time. The proportional hazards regression model with time-dependent covariates can then be written: h(t, x(t))=h (t0)exp(β′x(t)), where β is a column vector of regression coefficients. Information on the value of x for all values of t must be provided. Although this looks like DATA step programming, these statements do not appear in the DATA step. They are placed after the MODEL statement that goes with the PROC PHREG statement.
Furthermore, as a complementary analysis, we excluded 1148 men who increased (767 men) or decreased (381 men) their BMI more than 5% from their baseline BMI during the follow-up period. Figure 1 shows the flow diagram of the participants in this study. From the 4053 qualified participants remaining, we calculated RRs for hypertension in the absence of any major weight change. In this complementary analysis, we used baseline (non-time-dependent or fixed) variables, such as age, SBP, log-transformed triglyceride level, hemoglobin A1c, smoking status and alcohol intake as covariates in the PROC PHREG procedure.
During the follow-up period through 2006, there were 899 newly diagnosed cases of hypertension among the 5201 men (14 888 person-years). Table 1 shows baseline characteristics, BMI at end point (2006 or the year when hypertension was diagnosed) and changes in BMI (delta of the end point measurement minus the baseline measurement) within each BMI category. Mean change in BMI of all subjects was 0.2±1.1 kg m−2 (range: −6.6 to 6.3 kg m−2) during the follow-up period. We observed significant BMI increases in the lowest four BMI categories (⩽24.9 kg m−2 ), whereas there were no significant changes in the highest three BMI categories (⩾25.0 kg m−2). Trend test results for BMI categories are also shown in Table 1.
Table 2 shows RRs using the Cox proportional hazard model with fixed (non-time-dependent) baseline variables as covariates and the number of hypertension incidents within the BMI categories. Both the age-adjusted and multivariate RRs for hypertension increased as BMI increased. The RRs for the highest two BMI categories were significantly higher than our reference category of 23.0–24.9 kg m−2 , whereas the RRs of the lowest three BMI categories were significantly lower than the reference. Table 3 shows RRs using the time-dependent covariate Cox proportional hazard model method. These results indicate an increase in RR for hypertension as BMI increases, even if transitions in BMI and other covariates during follow-up period are taken into account.
In our complementary analysis, there were 732 newly diagnosed cases of hypertension among the 4053 men (11 199 person-years). Figure 2 shows multivariate RRs and number of hypertension incidents within BMI categories. In this figure, because of the small incident number, we expanded the BMI categories to include <18.5 kg m−2 in the 18.5–20.9 kg m−2 category and ⩾30.0 kg m−2 in the 27.0–29.9 kg m−2 category. The multivariate RR (CI) within the >27.0 kg m−2 BMI category was 1.43 (1.16–1.77) times higher than our reference category of 23.0–24.9 kg m−2 , whereas the multivariate RRs (CIs) within BMI categories of <21.0 and 21.0–22.9 kg m−2 were 0.64 (0.47–0.86) and 0.63 (0.51–0.79) times lower, respectively, than the reference category of 23.0–24.9 kg m−2.
To the best of our knowledge, this is the first study demonstrating an association between long-term stability of weight and future incident risk for hypertension among a middle-aged Asian (Japanese) population.
Although in many previous epidemiological studies,6, 7, 8, 9, 10, 11, 12 a higher BMI at baseline related to an increased future incident risk for hypertension, these studies included a statistical problem whereby transitions of covariates during the follow-up period were not taken into account. Most clinical trials have found that weight loss reduced blood pressure level in hypertensive or high-normal blood pressure patients.2, 26 In contrast, many prospective cohort studies13, 14, 15, 16, 17, 18, 19 have shown that weight gain predicts future incident risk for hypertension. Thus, because there is a significant association between changes in weight during the observation period and future incident risk for hypertension, weight (BMI) changes should be considered when investigating the effect of baseline BMI on future incident risk for hypertension. A recent study by Chen et al.13 showed that even a small BMI gain during a short-term (2 years) observation is associated with increased future incident risk for hypertension in a Chinese population, especially among the male subjects: the multivariate risk ratios within the second (–0.5 to 0.2 kg m−2), third (0.2 to 0.9 kg m−2) and fourth (>0.9 kg m−2 ) quartiles of BMI changes were 1.55 (95% CI: 1.04–2.31), 1.81 (95% CI: 1.21–2.69) and 1.81 (95% CI: 1.22–2.68) times higher, respectively, than within the lowest quartile of BMI changes (⩽−0.5 kg m−2 ).
In the present study, the mean BMI of all subjects increased significantly (mean: 0.2±1.1 kg m−2, minimum:−6.6 kg m−2, maximum: 6.3 kg m−2 ) during the observation period (Table 1). In this study, we considered the effect of weight change during the follow-up period in our analyses in order to investigate the association between long-term weight stability (maintaining BMI at baseline level) and future incident risk for hypertension. To consider covariate transitions during the follow-up period, we applied the time-dependent covariate Cox regression analysis. This statistical technique is used to solve a problem in prospective cohort studies: how to use potential risk factors, which are measured repeatedly during a follow-up period, to evaluate the relationship of a risk factor to disease development.20, 21, 22 In this study, although Table 2 shows the RRs for incident hypertension by BMI category using a traditional Cox regression model, Table 3 shows the RRs using the time-dependent Cox model. Although the RRs of the two methods were not completely consistent, we observed very similar RR trends within the two Cox models, that is, the RR increased when the BMI category increased. Moreover, we tried to calculate RR for incident hypertension without using subjects who experienced major weight changes during the follow-up period (Figure 2). These results are consistent with the results of the time-dependent Cox regression analysis in this study. That is, compared with our reference BMI category (23.0–24.9 kg m−2), we observed significantly higher-future incident risk for hypertension when baseline BMI was ⩾27.0 kg m−2 and significantly lower risk when baseline BMI was ⩽22.9 kg m−2 . Consequently, our results add mounting evidence:6, 7, 8, 9, 10, 11, 12 a higher baseline BMI increases future incident risk for hypertension even when there has not been a major weight increase, whereas maintaining a leaner body weight lowers the risk for hypertension. Expanding on previous studies,13, 14, 15, 16, 17, 18, 19 our results indicate that not only does long-term weight gain increase the risk for hypertension, but so does long-term stable obesity.
The study by Williams14 restricted analysis to individuals whose weights remained relatively unchanged (±0.4 kg m−2) during the follow-up and showed that each kg m−2 increment in BMI was associated with an odds ratio for future hypertension of 1.19 (CI: 1.14–1.24) in men and 1.11 (CI: 1.02–1.20) in women. This is consistent with our results, although the BMIs in his study were calculated using self-reported height and weight and the participants were not from an Asian population.
In every calculation of RR in our analyses (Tables 2 and 3 and Figure 2), the risk for developing hypertension in subjects from the BMI category ⩾27.0 kg m−2 was significantly higher than in subjects from the lower BMI categories. The evidence that an increase in BMI beyond 27 kg m−2 related to an increased risk for hypertension is consistent with previous studies.18, 27 The US National Heart, Lung and Blood Institute27 reported that excess body weight (BMI⩾27 kg m−2) correlates closely with increased blood pressure. Similarly, in Japanese subjects, Ishikawa-Takata18 reported that the risk for hypertension greatly increased in subjects with a BMI above 27 kg m−2.
Several investigators6, 11, 17, 19, 28 have discussed the mechanisms by which higher BMI is associated with increased future incident risk for hypertension. Complex interactions among body fat, insulin resistance, the renin–angiotensin system and the sympathetic nervous system may impact the relationship between baseline BMI and developing hypertension. However, the specific biologic mechanisms of this association are not completely known. Future research is needed to clarify these mechanisms.
Our study did have some limitations. First, Fisher and Lin21 indicate that the form of a time-dependent covariate is much more complex than in Cox models with fixed (non-time-dependent) covariates and that interpreting the results tends to be difficult and misleading. However, they also state in the paper21 that a well-established statistical software package, such as the SAS system used in this study, supports the use of time-dependent covariates in Cox model analyses, which can open exciting opportunities for exploring associations and potential causal mechanisms. With these points in mind, in our analyses, we showed the association between long-term stability of weight and future incident risk for hypertension not only with a time-dependent covariate Cox proportional hazard model, but through a complementary analysis that used the traditional Cox hazard model with fixed baseline (non-time-dependent) covariates and restricted data consisting of subjects who did not experience a major BMI change during the observation period. We obtained almost the same results from these two analyses. Second, the data using adjusted variables (alcohol intake and smoking status) are based on interview questionnaires, and we have no information on the validity of the answers, although the interviews were conducted by a trained observer. Third, smoking status may influence blood pressure or incident hypertension.29 It would be difficult to exclude smoking subjects in our analyses because of the high rate of past and current smoking habits among male workers in this study, although the smoking habit was included in multivariate regression analyses as an adjusted variable. Finally, we could not obtain information on diet intake30 or physical activity,31 which might lead to residual confounding factors when evaluating the association between baseline BMI and blood pressure.
In conclusion, baseline BMI is associated with future incident risk for hypertension even when weight change during the follow-up period is taken into account. The risk for hypertension increases as baseline BMI increases among middle-aged Japanese male workers. Weight management should be encouraged for obese, non-hypertensive people to prevent future hypertension.
Staessen JA, Wang JG, Thijs L . Cardiovascular protection and blood pressure reduction: a meta-analysis. Lancet 2001; 358: 1305–1315.
Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA, Izzo Jr JL et al. Seventh report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. Hypertension 2003; 42: 1206–1252.
Ueshima H, Zhang XH, Choudhury SR . Epidemiology of hypertension in China and Japan. J Hum Hypertens 2000; 14: 765–769.
Klag MJ, Whelton PK, Randall BL, Neaton JD, Brancati FL, Ford CE et al. Blood pressure and end-stage renal disease in men. N Engl J Med 1996; 334: 13–18.
MacMahon S, Peto R, Cutler J, Collins R, Sorlie P, Neaton J et al. Blood pressure, stroke, and coronary heart disease. Part 1, Prolonged differences in blood pressure: prospective observational studies corrected for the regression dilution bias. Lancet 1990; 335: 765–774.
Shuger SL, Sui X, Church TS, Meriwether RA, Blair SN . Body mass index as a predictor of hypertension incidence among initially healthy normotensive women. Am J Hypertens 2008; 21: 613–619.
Pang W, Sun Z, Zheng L, Li J, Zhang X, Liu S et al. Body mass index and the prevalence of prehypertension and hypertension in a Chinese rural population. Intern Med 2008; 47: 893–897.
Nguyen TT, Adair LS, He K, Popkin BM . Optimal cutoff values for overweight: using body mass index to predict incidence of hypertension in 18- to 65-year-old Chinese adults. J Nutr 2008; 138: 1377–1382.
Nemesure B, Wu SY, Hennis A, Leske MC . The relationship of body mass index and waist-hip ratio on the 9-year incidence of diabetes and hypertension in a predominantly African-origin population. Ann Epidemiol 2008; 18: 657–663.
Kawada T, Morihashi M, Ueda H, Sirato T . Body mass index of 23 or more is a risk factor for hypertension and hyperlipidemia in Japanese workers. Percept Mot Skills 2007; 104 (3 Part 1): 733–738.
Gelber RP, Gaziano JM, Manson JE, Buring JE, Sesso HD . A prospective study of body mass index and the risk of developing hypertension in men. Am J Hypertens 2007; 20: 370–377.
Kawada T . Body mass index is a good predictor of hypertension and hyperlipidemia in a rural Japanese population. Int J Obes Relat Metab Disord 2002; 26: 725–729.
Chen PC, Sung FC, Su TC, Chien KL, Hsu HC, Lee YT . Two-year change in body mass index and subsequent risk of hypertension among men and women in a Taiwan community. J Hypertens 2009; 27: 1370–1376.
Williams PT . Increases in weight and body size increase the odds for hypertension during 7 years of follow-up. Obesity (Silver Spring) 2008; 16: 2541–2548.
Truesdale KP, Stevens J, Cai J . Effect of 3-year weight history on blood pressure: the atherosclerosis risk in communities study. Obesity (Silver Spring) 2008; 16: 1112–1119.
Yang G, Xiang YB, Zheng W, Xu WH, Zhang X, Li HL et al. Body weight and weight change in relation to blood pressure in normotensive men. J Hum Hypertens 2007; 21: 45–52.
Juhaeri, Stevens J, Chambless LE, Tyroler HA, Rosamond W, Nieto FJ et al. Associations between weight gain and incident hypertension in a bi-ethnic cohort: the Atherosclerosis Risk in Communities Study. Int J Obes Relat Metab Disord 2002; 26: 58–64.
Ishikawa-Takata K, Ohta T, Moritaki K, Gotou T, Inoue S . Obesity, weight change and risks for hypertension, diabetes and hypercholesterolemia in Japanese men. Eur J Clin Nutr 2002; 56: 601–607.
Huang Z, Willett WC, Manson JE, Rosner B, Stampfer MJ, Speizer FE et al. Body weight, weight change, and risk for hypertension in women. Ann Intern Med 1998; 128: 81–88.
D’Agostino RB, Lee ML, Belanger AJ, Cupples LA, Anderson K, Kannel WB . Relation of pooled logistic regression to time dependent Cox regression analysis: the Framingham Heart Study. Stat Med 1990; 9: 1501–1515.
Fisher LD, Lin DY . Time-dependent covariates in the Cox proportional-hazards regression model. Annu Rev Public Health 1999; 20: 145–157.
Cox DR, Oakes D (eds). Analysis of Survival Data. Chapman and Hall: New York, 1984.
Matsuo T, Sairenchi T, Iso H, Irie F, Tanaka K, Fukasawa N et al. Age- and gender-specific BMI in terms of the lowest mortality in Japanese general population. Obesity (Silver Spring) 2008; 16: 2348–2355.
Hozawa A, Okamura T, Oki I, Murakami Y, Kadowaki T, Nakamura K et al. Relationship between BMI and all-cause mortality in Japan: NIPPON DATA80. Obesity (Silver Spring) 2008; 16: 1714–1717.
Cantor AB . SAS Survival Analysis Techniques for Medical Research, 2nd edn. SAS Institute Inc: Cary, NC, 2003.
Neter JE, Stam BE, Kok FJ, Grobbee DE, Geleijnse JM . Influence of weight reduction on blood pressure: a meta-analysis of randomized controlled trials. Hypertension 2003; 42: 878–884.
National Heart L and Blood Institute. The sixth report of the joint national committee on prevention, detection, evaluation, and treatment of high blood pressure. Arch Intern Med 1997; 157: 2413–2446.
Pi-Sunyer X . Body mass index as a risk factor for incident hypertension. Nat Clin Pract Endocrinol Metab 2007; 3: 742–743.
Omvik P . How smoking affects blood pressure. Blood Press 1996; 5: 71–77.
Stamler R, Stamler J, Gosch FC, Civinelli J, Fishman J, McKeever P et al. Primary prevention of hypertension by nutritional-hygienic means. Final report of a randomized, controlled trial. JAMA 1989; 262: 1801–1807.
Hu G, Barengo NC, Tuomilehto J, Lakka TA, Nissinen A, Jousilahti P . Relationship of physical activity and body mass index to the risk of hypertension: a prospective study in Finland. Hypertension 2004; 43: 25–30.
The authors declare no conflict of interest.
About this article
Cite this article
Matsuo, T., Sairenchi, T., Suzuki, K. et al. Long-term stable obesity increases risk of hypertension. Int J Obes 35, 1056–1062 (2011). https://doi.org/10.1038/ijo.2010.226
- body mass index
- blood pressure
- weight change
- prospective cohort study
- time-dependent covariate
World Journal of Hypertension (2019)
Body mass index, abdominal fatness, and hypertension incidence: a dose-response meta-analysis of prospective studies
Journal of Human Hypertension (2018)
Prevalence and trends of overweight and obesity in older adults from 10 European countries from 2005 to 2013
Scandinavian Journal of Public Health (2018)
European Journal of Sport Science (2018)
Body mass index, abdominal adiposity, weight gain and risk of developing hypertension: a systematic review and dose-response meta-analysis of more than 2.3 million participants
Obesity Reviews (2018)