Association of body mass index and waist circumference with high blood pressure in older adults

The study population consisted of participants of a comprehensive health check-up program conducted at fourteen medical examination centers (Supplementary 1). Generally, the Central People’s Government of China demands that residents aged ≥60 years participate in health examinations annually to promote good health by enabling early detection of chronic diseases and associated risk factors. The study collected clinical, demographic, and lifestyle information from all participants by face-to-face interviews, physical examinations and blood biochemical examinations. Cross-sectional study data from Xinzheng from January to December 2019 were combined for analyses. For this study, 1969 people were excluded from the current study due to missing physical examination data (n = 117) or biochemical test data (n = 1852). Finally, we had data for 58,115 men and 68,008 women resident ≥60 years of age, who were enrolled to assess the association between BMI, WC and HBP (Fig. 1). Written informed consent was obtained from each participant before data collection. The research ethics committee of Zheng Zhou University approved the current study methodology, protocol, and procedures. (Reference Number: ZZUIRB2019–019).

Height, weight, and WC were measured twice by trained nurses following rigorous protocols. Body height was measured without shoes with a stadiometer, and body weight was measured with participants in light clothing and without shoes by electronic scales. WC was measured with gentle breathing at the midpoint between the lowest rib and the iliac crest to the nearest 0.1 cm. BMI was calculated as weight (kg) divided by the square of height (m). BMI was categorized into quintiles (Male: < 22.20, 22.20 ~ 24.02, 24.02 ~ 25.73, 25.73 ~ 27.78, ≥27.78 Female: < 22.43, 22.43 ~ 24.42, 24.4 ~ 26.27, 26.27 ~ 28.58, ≥28.58). And WC was categorized into quintiles (Male:< 81,81 ~ 86,86 ~ 90,90 ~ 96,≥96Female:< 79,79 ~ 84,84 ~ 89,89 ~ 95,≥95).

Systolic blood pressure (SBP) and diastolic blood pressure (DBP) were collected according to the World Health Organization definition [15]. After sitting quietly for five minutes, certified nurses measured each participant’s seated blood pressure three times using a mercury sphygmomanometer (Omron HEM-7125, Kyoto, Japan). We calculated the means of SBP and DBP. Diagnosis of HBP was defined as SBP ≥ 140 mmHg/or DBP ≥90 mmHg or use of antihypertensive medication within 2 weeks [15].

Information on demographic characteristics (age, sex, marital status, and place of residence) and behavioral measures (smoking, physical exercise and alcohol consumption) were obtained by a standardized questionnaire, which was in strict accordance with the National Standards for Basic Public Health Services (2011). Marital status was categorized as unmarried, married, divorced and death of a spouse. Smoking status was defined as current smoker, former smokers, or never smoker [16]; Alcohol consumption and physical exercise status were categorized as never, once in a while, more than once a week and every day. Place of residence (rural and urban) was defined according to country-specific definitions. Stroke was defined as sudden onset of a focal, non-convulsive neurological deficit persisting longer than 24 h. Diagnosis of psychotic illnesses was met Diagnostic and Statistical Manual of Mental Disorders (DSM − 5) criteria for a diagnosis within the spectrum of primary psychotic illnesses. Diagnosis of cancer based on the International Classification of Diseases. Overnight fasting blood samples were collected into vacuum tubes for assessing serum levels of blood glucose using standard methods.

Continuous data were expressed as the mean ± standard deviation (SD) and categorical data as the number (percentage) for quintiles of BMI or WC, using different values for men and women. Comparisons of the basic characteristics of the quintiles were performed with the χ²test, ANOVA, and Kruskal-Wallis test.

Logistic regression models were used to estimate the adjusted odds ratios (ORs) with 95% confidence intervals (CIs) for BMI, WC and HBP, taking the quintile with the lowest baseline BMI or WC values as the reference. ORs and 95% CIs for HBP were estimated for each group with adjustment of multiple confounders. One obesity parameter was introduced at a time in each model to avoid the collinear effect. Model 1 was unadjusted. Model 2 was adjusted for sex and age. Model 3 was adjusted for Model 2 and marital status, alcohol consumption, smoking, physical activity, place of residence, cancer, stroke, psychotic illness and blood glucose. A linear trend test was performed by modeling the median value of each exposure category as a continuous variable in the models. Fully adjusted restricted cubic spline analyses were used to characterize the dose-response association and explore the potential linear or nolinear relationship of BMI, and WC with HBP. The knots were placed at the 5th, 25th, 50th, 75th and 95th percentiles. The test result for overall association was checked first. If the test for overall association was significant, the test result for nonlinearity and linearity were checked, and the P-value for non-linear association < 0.05 indicated a significant result indicating the linear association. We also evaluated the additive interaction between BMI and WC for HBP with BMI and WC analyzed as continuous variable in two categories [(BMI: BMI < 25 and BMI ≥ 25. WC: WC < 102 for males, WC < 88 for females and WC ≥ 102 for males, WC ≥ 88 for females) or (WHO proposed cut-off points [17])]. We applied three indicators to evaluate the additive interaction: relative excess risk due to interaction (RERI), attributable proportion due to interaction (AP) or synergy index (S). RERI > 0, AP > 0, or S > 1 was regarded as a significant additive interaction.

Statistical analyses involved the use of SAS V .9.1 (SAS Institute) and R × 64 4.0.0.All reported P values were two-sided, with P < 0.05 considered statistically significant.

A total of 126,123 participants were eligible for inclusion in this study. The mean (SD) age was 70.29 (6.94) years. The basic demographic characteristics of the study population according to BMI and WC quintiles are shown in Tables 1 and 2. The prevalence of HBP and stroke, levels of alcohol consumption, physical exercise and blood glucose showed a significant progressive increase from the quintile with the lowest BMI to the quintile with the highest BMI. This was also the case for the WC quintiles.

Table 3 presents the results from the logistic regression that estimated the association between the levels of BMI, WC and HBP. The multivariable adjusted OR (95% CI) per 1 kg/m² increase in BMI was 1.084 (1.08 to 1.087). In all three models, the ORs for HBP increased significantly with increasing BMI quintiles (P for trend < 0.01). In Model 3, the multivariate-adjusted OR (95% CIs) for HBP with the highest BMI quintile group compared with the lowest quintile group was 2.300(2.217 to 2.386). The multivariable adjusted OR (95% CI) per 1 cm increase in WC was 1.025 (1.024 to 1.027). The highest quintile group had a greater HBP prevalence than the other quintile group, and the crude ORs (95% CIs) for HBP compared with the lowest quintile group was 1.879 (1.814 to 1.947). After adjustment for sex, age, marital status, alcohol consumption, smoking, physical activity, place of residence and blood glucose, the ORs were enhanced, and the multivariate-adjusted ORs (95% CIs) for HBP with the highest WC quintile group compared with the lowest quintile group was 1.977(1.906 to 2.050).

Multivariable adjusted restricted cubic spline analyses showed the nonlinear relationships of BMI with HBP (all P < 0.001; Fig. 2a). The risk of HBP increased with increasing BMI. As BMI increased, the ORs increased from 0.31(0.28 to 0.35) to 2.08(1.82 to 2.39) in the 15–42 kg/m² range. As a result, the ORs were inversely associated with HBP when BMI was below 25 kg/m², but presented a significant risk effect above this value. Subgroup analyses on men-women did not show significant differences (Fig. 2b; c).

The results of the dose-response relationship analysis between WC and HBP are shown in Fig. 3a. A nonlinear association (all P < 0.001) between WC and HBP was detected. As WC increased, the ORs increased from 0.32(0.27 to 0.39) to 2.57(2.24 to 2.95) in the 40–130 cm range. When stratified by sex, the ORs increased from 0.27(0.20 to 0.37) to 3.22(2.62 to 3.97) in the male population. With an increase in WC, when WC was over 88 cm and 86 cm for males and females, respectively, WC was more steeply positively associated with the risk of HBP (Fig. 3b; c).

Table 4 presents the results from additive interaction analysis. We observed a significant additive interaction between higher BMI and WC such that the prevalence of HBP increased (RERI = 1.28, 95% CI: 1.13–1.43; AP = 0.43, 95% CI: 0.41–0.45; S = 2.88, 95% CI: 2.79–2.97). If BMI < 25.0 kg/m² and WC < 88 cm for males and WC < 86 cm for females were used as the reference, BMI ≥ 25 alone and WC ≥ 88 for males or WC ≥ 86 for females alone were both associated with increased risks of HBP. The copresence of both factors greatly enhanced the adjusted ORs of higher BMI alone 1.476(1.418 to 1.536) and higher WC alone 1.230(1.186 to 1.275) to 1.729(1.685 to 1.775) for HBP, with significant additive interactions. When BMI and WC were classified by the proposed cut-off points [17] on the waist circumference continuum (BMI: 25 kg/m², WC: 88 cm for females and 102 cm for males), there was still significant additive interaction (RERI = 1.68, 95% CI: 1.48–1.87; AP = 0.48, 95% CI: 0.46–0.50; S = 3.00, 95% CI: 2.91–3.10).