Association between hemoglobin A1c to HDL cholesterol ratio and hypertension in middle-aged and older Chinese adults: a cross-sectional study Authors: Wenqing Si, Weicheng Lai, Liangxiu Wu, Shenshen Du, Yupeng Wu, Xin Li, Wenlong Zhao, Liangyan Wu, Boda Zhou Published: 2026 ABSTRACT IntroductionThe hemoglobin A1c to high-density lipoprotein cholesterol ratio (HHR) is a novel composite index reflecting both chronic glycemic exposure and lipid metabolism. However, its association with hypertension remains unclear. This study aimed to investigate the independent association between HHR and hypertension prevalence in middle-aged and older Chinese adults.MethodsThis cross-sectional study analyzed data from the 2011 and 2015 biomarker waves of the China Health and Retirement Longitudinal Study (CHARLS). Among 25,873 initially identified participants, 11,797 were ultimately included after applying the inclusion and exclusion criteria. Multivariable logistic regression and restricted cubic spline analyses were conducted to assess the association between HHR and hypertension prevalence, adjusting for sociodemographic, lifestyle, and metabolic confounding factors.ResultsHHR was positively associated with hypertension. In the fully adjusted model, each 0.1-unit increase in HHR was associated with higher odds of hypertension (OR = 1.02, 95% CI: 1.01–1.04). Participants in the high HHR group had significantly higher odds of hypertension than those in the low HHR group (OR = 1.38, 95% CI: 1.24–1.54). Restricted cubic spline analysis showed a significant nonlinear association between HHR and hypertension.ConclusionElevated HHR was associated with hypertension in a nonlinear pattern among middle-aged and older Chinese adults. Given the cross-sectional design, these findings should be interpreted as evidence of association rather than prediction or causality. Further prospective studies are warranted to determine whether HHR provides incremental value for hypertension risk assessment.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12872-026-06174-0. Introduction Hypertension remains a major global health concern and a key modifiable risk factor for cardiovascular disease, chronic kidney disease, and premature mortality [1, 2]. In China, the prevalence of hypertension has risen steadily in recent decades, driven by rapid urbanization and population aging, yet control rates remain suboptimal [3]. The pathogenesis of hypertension is multifactorial, with growing evidence implicating metabolic dysfunction, particularly disturbances in glucose and lipid metabolism, as a central contributor to vascular remodeling and endothelial injury [4]. Accordingly, identifying accessible and integrative biomarkers that capture these metabolic abnormalities is important for improving the understanding of cardiometabolic risk in hypertension. Hyperglycemia and dyslipidemia, although distinct, frequently co-occur and exert synergistic deleterious effects on vascular health. Hemoglobin A1c (HbA1c), a reliable indicator of chronic glycemic exposure, has been positively associated with arterial stiffness and endothelial dysfunction, even in non-diabetic individuals [5]. Conversely, high-density lipoprotein cholesterol (HDL-C) supports vascular health by promoting reverse cholesterol transport and exerting anti-inflammatory and antioxidative effects [6]. While these markers are valuable individually, single biomarkers may not adequately reflect the complex interaction between glucotoxicity and impaired lipid-mediated vascular protection. In recent years, composite metabolic indices have been increasingly used to characterize cardiometabolic risk across diverse populations [7, 8]. Among these, the HbA1c/HDL-C ratio (HHR) has emerged as a novel index integrating glycemic and lipid status [9]. Elevated HHR has been associated with diabetic retinopathy and coronary artery disease severity, suggesting its sensitivity to vascular injury [10]. However, population-based evidence regarding the association between HHR and hypertension remains limited. Specifically, it remains unclear whether HHR is independently associated with hypertension in the general population or whether this association exhibits a nonlinear pattern. Therefore, the present study utilized data from the China Health and Retirement Longitudinal Study (CHARLS), a large-scale, nationally representative cohort, to investigate the relationship between HHR and the prevalence of hypertension in middle-aged and older Chinese adults. We aimed to assess the independent association between HHR and hypertension, explore potential nonlinear trends using restricted cubic spline (RCS) models, and evaluate the consistency of this relationship across key demographic and metabolic subgroups. Methods Study design This study utilized data from the CHARLS, which covers participants from 28 provinces, 150 counties or districts, and 450 villages across China. The dataset includes demographic, socioeconomic, health status, blood tests, and functional information. Baseline data were collected in Wave 1(2011), with follow-up surveys conducted every two years in Wave 2 (2013), Wave 3 (2015), Wave 4 (2018) and Wave 5 (2020). Blood tests were only carried out in 2011 and 2015. Information was obtained through structured face-to-face interviews, physical examinations, and laboratory testing by trained health professionals. In this study, we used biomarker data from the 2011 and 2015 CHARLS waves to examine the association between HHR and hypertension prevalence, because blood biomarkers required for calculating HHR were available only in these two waves. To avoid repeated inclusion of the same individual, each participant was included only once. For participants with eligible data in both 2011 and 2015, the 2011 record was preferentially used; for participants without complete eligible data in 2011 but with available data in 2015, the 2015 record was used. Thus, no participant was counted more than once in the final analytic sample. The CHARLS dataset is available at its official website: charls.ccer.edu.cn/en. The Ethics Committee of Peking University approved the CHARLS study (IRB00001052-11015), and all participants signed informed consent forms. All methods were performed in accordance with the relevant guidelines and regulations. Study population This cross-sectional study was based on data from the CHARLS, a nationally representative survey, using eligible participants from the 2011 and 2015 biomarker waves, with each participant included only once. A total of 25,873 participants were initially identified. As illustrated in Fig. 1, a series of exclusion criteria were applied to ensure data completeness and eliminate potential confounders. First, participants with missing HHR values or lacking self-reported hypertension information were excluded (n = 11,109). Subsequently, individuals younger than 45 years or with incomplete demographic data were removed (n = 483). We further excluded those with missing systolic blood pressure (SBP) or diastolic blood pressure (DBP) measurements (n = 1,764). Finally, participants who were receiving antihyperglycemic therapy were excluded (n = 720) to avoid the influence of pharmacologic glucose control on HHR. After these exclusions, a total of 11,797 eligible participants were included in the final analysis. Fig. 1Flowchart of participant selection Flowchart of participant selection Calculate the HHR The HHR was derived by calculating the ratio of HbA1c (%) to HDL-C (mg/dL) [11]. Data collection and potential covariates Data collection for CHARLS was conducted by trained staff using standardized protocols and structured questionnaires. Sociodemographic information, including age, gender, and education level, was collected through face-to-face interviews. Health-related behaviors and medical history were also recorded, including smoking status, alcohol consumption, current antihypertensive medication use, and history of cardiovascular or metabolic conditions such as stroke, dyslipidemia, and coronary heart disease. Trained medical personnel performed anthropometric and blood pressure (BP) assessments according to standardized operating procedures. BP was measured using an Omron HEM-7200 electronic monitor. Participants were required to rest in a seated position for at least 5 min before measurement. Three readings were taken at 1-minute intervals, and the mean value was used for analysis. Body mass index (BMI) was calculated from measured weight (kg) and height (m) while participants wore light clothing and no shoes. Fasting venous blood samples were obtained in the morning after an overnight fast of at least 8 h. These specimens were analyzed in certified laboratories to assess fasting blood glucose (FBG), HbA1c, C-reactive protein (CRP), serum creatinine (Scr), blood urea nitrogen (BUN), serum uric acid (SUA), and lipid profiles, including triglycerides (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), and HDL-C. The HHR was calculated as HbA1c (%) divided by HDL-C (mg/dL). Covariates were selected based on clinical relevance and prior literature. These included sociodemographic factors (age, gender, and education level), anthropometric status (BMI), lifestyle factors (smoking status and alcohol consumption), and metabolic and biochemical markers, including FBG, Scr, BUN, SUA, CRP, LDL-C, TC, and TG. SBP and DBP were not included as covariates because they were components of the hypertension definition and directly determined the outcome status; therefore, including them in the primary models could lead to overadjustment. HbA1c and HDL-C were not included as additional covariates because they were components of HHR. Definition of hypertension Hypertension was defined as having a SBP ≥ 140 mmHg or DBP ≥ 90 mmHg in individuals not currently taking antihypertensive medication. In addition to measured blood pressure, participants were also classified as hypertensive if they answered “yes” to either of the following self-reported questions: (1) “Are you currently being treated for hypertension with either traditional Chinese medicine or Western medicine?” or (2) “Have you ever been diagnosed with hypertension by a doctor?” Individuals who met any one of these criteria were considered to have hypertension [12]. This composite definition was used consistently in all tables, figures, and regression models involving hypertension. Statistical analysis Statistical analyses were performed using Empower (version 2.0) and R software (version 4.3.1). Continuous variables were expressed as mean ± standard deviation, while categorical variables were presented as frequencies and percentages. One-way analysis of variance (ANOVA), Kruskal-Wallis H test, or chi-square test were used to compare differences in variables among different HHR groups. Multivariate logistic regression models were constructed to assess the association between HHR and hypertension prevalence. Three progressive models were established: Model 1 was unadjusted; Model 2 adjusted for gender, age, education level, smoking status, drinking status, and BMI; Model 3 further adjusted for FBG, Scr, BUN, SUA, CRP, LDL-C, TC, and TG on the basis of Model 2. SBP and DBP were not included in the multivariable models because they were components of the hypertension definition and directly determined the outcome status; therefore, adjustment for SBP and DBP could result in overadjustment. HbA1c and HDL-C were not included as covariates because they were the numerator and denominator of HHR, respectively. To assess potential multicollinearity among covariates in the fully adjusted model, generalized variance inflation factors (GVIFs) and adjusted GVIFs [GVIF1/(2 × Df)] were calculated. Results are presented as odds ratios (ORs) with 95% confidence intervals (CIs). HHR was analyzed both as a continuous variable and as a categorical variable according to the receiver operating characteristic (ROC) derived threshold. The optimal cutoff value was determined using the Youden index, and the corresponding area under the curve (AUC), sensitivity, specificity, and Youden index were reported. The low HHR group (HHR ≤ 0.12) served as the reference group. As a sensitivity analysis, HHR was also categorized into quartiles, with the lowest quartile used as the reference group. Comparative ROC analyses were further conducted to compare the discriminatory performance of HHR with that of HbA1c, HDL-C, the triglyceride-glucose index, and the LDL-C/HDL-C ratio for prevalent hypertension. RCS analysis with four knots was performed to explore the potential nonlinear association between HHR and the prevalence of hypertension. The knots were placed at the 5th, 35th, 65th, and 95th percentiles of the HHR distribution, corresponding to HHR values of 0.066, 0.096, 0.119, and 0.174, respectively. These knot locations were selected to flexibly model potential nonlinearity while ensuring sufficient observations across the exposure range. The overall association and nonlinearity were tested using likelihood ratio tests. Survey-weighted logistic regression was additionally performed as a sensitivity analysis using the available CHARLS biomarker sampling weights and community-level clustering. The 2011 biomarker weight was used for participants whose biomarker data were derived from the 2011 wave, and the 2015 biomarker weight was used for participants whose biomarker data were derived from the 2015 wave. Formal stratification variables were not available in the analytic dataset. Subgroup analyses were conducted stratified by age, gender, education level, dyslipidemia status, drinking status, smoking status, BMI categories, and FBG levels. Interaction effects between HHR and these stratification variables were evaluated using multiplicative interaction terms in the regression models. All statistical tests were two-sided, and P-values < 0.05 were considered statistically significant. Participants with missing data on HHR, hypertension status, age or other required demographic variables, SBP, DBP, or covariates required for the main analyses were excluded. No imputation was performed, and the primary analyses were conducted using a complete case approach. To assess potential selection bias, baseline characteristics were compared between participants included in the final analysis and those excluded from the analysis. For included participants, the index wave corresponded to the wave used in the final analytic dataset. For excluded participants, characteristics from the 2011 wave were used when available; if 2011 information was unavailable but 2015 information was available, 2015 characteristics were used instead. Variables derived from multiple questionnaire items, such as smoking status, drinking status, and hypertension status, were constructed using the same definitions as those applied in the main analytic dataset whenever possible. Results Characteristics of the participants Initially, 25,873 individuals from the 2011 and 2015 CHARLS biomarker waves were evaluated. Participants were excluded for the following reasons: missing HHR or hypertension data (n = 11,109), age under 45 years or incomplete demographic information (n = 483), missing SBP or DBP measurements (n = 1,764), and use of antihyperglycemic medications (n = 720). After these exclusions, 11,797 participants remained for analysis (Fig. 1). Table 1 presents the baseline characteristics stratified by HHR into two groups: the low HHR group (HHR ≤ 0.12, N = 7,850) and the high HHR group (HHR > 0.12, N = 3,947). In the final analytic sample, the overall prevalence of hypertension was 27.53%, and the prevalence was higher in the high HHR group than in the low HHR group (34.89% vs. 23.83%). Overall, participants with higher HHR (> 0.12) demonstrated poorer cardiovascular and metabolic parameters. Compared with the low HHR group, individuals in the high HHR group exhibited significantly higher prevalence of hypertension, diabetes, cardiovascular disease, and stroke. They also had significantly elevated levels of SBP, DBP, TG, FBG, SUA, CRP, and HbA1c, while demonstrating significantly lower HDL-C concentrations and a higher proportion of participants with BMI > = 25 kg/m² (all P < 0.001). Additionally, the high HHR group had a higher proportion of men, lower education levels, and higher smoking rates, whereas age and marital status distributions were comparable between groups. Collectively, these findings indicate that elevated HHR was associated with the clustering of multiple adverse metabolic and cardiovascular risk factors. Baseline characteristics of included and excluded participants were compared to assess potential selection bias (Supplementary Table 6). Differences in several baseline characteristics were observed between the two groups, suggesting that potential selection bias due to analytic exclusions cannot be completely excluded. Table 1Baseline characteristics of participants HHR in cross-sectional studyVariableTotalLow HHR≤ 0.12High HHR> 0.12P-valueParticipants, sample size (N)11,79778503947NAAge, years59.30 ± 9.4559.36 ± 9.5459.18 ± 9.290.325Gender (%)< 0.001 Male5587 (47.36)3511 (44.73)2076 (52.60) Female6210 (52.64)4339 (55.27)1871 (47.40)Education level (%)< 0.001 Below high school or vocational school10,626 (90.07)7147 (91.04)3479 (88.14) High school or vocational school1025 (8.69)627 (7.99)398 (10.08) Above high school or vocational school146 (1.24)76 (0.97)70 (1.77)Marital status (%)0.145 Married9785 (82.94)6466 (82.37)3319 (84.09) Married separation504 (4.27)349 (4.45)155 (3.93) Separated53 (0.45)38 (0.48)15 (0.38) Divorced93 (0.79)62 (0.79)31 (0.79) Widowed1277 (10.82)883 (11.25)394 (9.98) Never married85 (0.72)52 (0.66)33 (0.84)Hypertension (%)< 0.001 yes3248 (27.53)1871 (23.83)1377 (34.89) no8549 (72.47)5979 (76.17)2570 (65.11)DM (%)< 0.001 yes286 (2.42)117 (1.49)169 (4.28) no11,509 (97.58)7731 (98.51)3778 (95.72)Cardiovascular Disease (%)< 0.001 yes1427 (12.16)845 (10.81)582 (14.84) no10,312 (87.84)6971 (89.19)3341 (85.16)Stroke (%)< 0.001 yes299 (2.54)160 (2.04)139 (3.53) no11,464 (97.46)7668 (97.96)3796 (96.47)SBP, mmHg128.61 ± 21.06127.64 ± 21.04130.53 ± 20.97< 0.001DBP, mmHg75.16 ± 12.1674.50 ± 12.1276.47 ± 12.13< 0.001BMI, Kg/m2< 0.001 < 18.5779 (6.69)653 (8.42)126 (3.24) >=18.5, < 234867 (41.81)3692 (47.59)1175 (30.25) >=23, < 252393 (20.55)1574 (20.29)819 (21.09) >=253603 (30.95)1839 (23.70)1764 (45.42)Antihypertensive medication (%)< 0.001 yes2284 (19.49)1253 (16.06)1031 (26.34) no9432 (80.51)6549 (83.94)2883 (73.66)Drinking (%)0.026 yes5028 (42.64)3402 (43.36)1626 (41.21) no6764 (57.36)4444 (56.64)2320 (58.79)Smoking (%)< 0.001 yes4816 (40.83)3032(38.64)1784 (45.20) no6978 (59.17)4815 (61.36)2163 (54.80)TC, mg/dL190.79 ± 38.08193.75 ± 36.40184.89 ± 40.58< 0.001TG, mg/dL132.92 ± 101.42106.17 ± 57.38186.23 ± 141.27< 0.001LDL-C, mg/dL113.03 ± 33.92115.77 ± 32.83107.57 ± 35.35< 0.001HDL-C, mg/dL51.32 ± 14.1657.97 ± 12.0438.09 ± 6.90< 0.001FBG, mg/dL105.33 ± 29.14101.18 ± 20.28113.59 ± 40.26< 0.001BUN, mg/dL15.67 ± 4.6315.83 ± 4.4915.35 ± 4.50< 0.001SUA, mg/dL4.58 ± 1.304.45 ± 1.244.85 ± 1.38< 0.001Scr, mg/dL0.79 ± 0.260.78 ± 0.280.81 ± 0.23< 0.001CRP, mg/dL2.69 ± 7.162.30 ± 6.273.46 ± 8.62< 0.001HbA1c (%)5.35 ± 0.675.21 ± 0.485.62 ± 0.88< 0.001Annotation: Data are presented as mean ± standard deviation or number (%) Table 1Baseline characteristics of participants HHR in cross-sectional studyVariableTotalLow HHR≤ 0.12High HHR> 0.12P-valueParticipants, sample size (N)11,79778503947NAAge, years59.30 ± 9.4559.36 ± 9.5459.18 ± 9.290.325Gender (%)< 0.001 Male5587 (47.36)3511 (44.73)2076 (52.60) Female6210 (52.64)4339 (55.27)1871 (47.40)Education level (%)< 0.001 Below high school or vocational school10,626 (90.07)7147 (91.04)3479 (88.14) High school or vocational school1025 (8.69)627 (7.99)398 (10.08) Above high school or vocational school146 (1.24)76 (0.97)70 (1.77)Marital status (%)0.145 Married9785 (82.94)6466 (82.37)3319 (84.09) Married separation504 (4.27)349 (4.45)155 (3.93) Separated53 (0.45)38 (0.48)15 (0.38) Divorced93 (0.79)62 (0.79)31 (0.79) Widowed1277 (10.82)883 (11.25)394 (9.98) Never married85 (0.72)52 (0.66)33 (0.84)Hypertension (%)< 0.001 yes3248 (27.53)1871 (23.83)1377 (34.89) no8549 (72.47)5979 (76.17)2570 (65.11)DM (%)< 0.001 yes286 (2.42)117 (1.49)169 (4.28) no11,509 (97.58)7731 (98.51)3778 (95.72)Cardiovascular Disease (%)< 0.001 yes1427 (12.16)845 (10.81)582 (14.84) no10,312 (87.84)6971 (89.19)3341 (85.16)Stroke (%)< 0.001 yes299 (2.54)160 (2.04)139 (3.53) no11,464 (97.46)7668 (97.96)3796 (96.47)SBP, mmHg128.61 ± 21.06127.64 ± 21.04130.53 ± 20.97< 0.001DBP, mmHg75.16 ± 12.1674.50 ± 12.1276.47 ± 12.13< 0.001BMI, Kg/m2< 0.001 < 18.5779 (6.69)653 (8.42)126 (3.24) >=18.5, < 234867 (41.81)3692 (47.59)1175 (30.25) >=23, < 252393 (20.55)1574 (20.29)819 (21.09) >=253603 (30.95)1839 (23.70)1764 (45.42)Antihypertensive medication (%)< 0.001 yes2284 (19.49)1253 (16.06)1031 (26.34) no9432 (80.51)6549 (83.94)2883 (73.66)Drinking (%)0.026 yes5028 (42.64)3402 (43.36)1626 (41.21) no6764 (57.36)4444 (56.64)2320 (58.79)Smoking (%)< 0.001 yes4816 (40.83)3032(38.64)1784 (45.20) no6978 (59.17)4815 (61.36)2163 (54.80)TC, mg/dL190.79 ± 38.08193.75 ± 36.40184.89 ± 40.58< 0.001TG, mg/dL132.92 ± 101.42106.17 ± 57.38186.23 ± 141.27< 0.001LDL-C, mg/dL113.03 ± 33.92115.77 ± 32.83107.57 ± 35.35< 0.001HDL-C, mg/dL51.32 ± 14.1657.97 ± 12.0438.09 ± 6.90< 0.001FBG, mg/dL105.33 ± 29.14101.18 ± 20.28113.59 ± 40.26< 0.001BUN, mg/dL15.67 ± 4.6315.83 ± 4.4915.35 ± 4.50< 0.001SUA, mg/dL4.58 ± 1.304.45 ± 1.244.85 ± 1.38< 0.001Scr, mg/dL0.79 ± 0.260.78 ± 0.280.81 ± 0.23< 0.001CRP, mg/dL2.69 ± 7.162.30 ± 6.273.46 ± 8.62< 0.001HbA1c (%)5.35 ± 0.675.21 ± 0.485.62 ± 0.88< 0.001Annotation: Data are presented as mean ± standard deviation or number (%) Baseline characteristics of participants HHR in cross-sectional study Annotation: Data are presented as mean ± standard deviation or number (%) Association between HHR and hypertension prevalence Multivariable logistic regression analysis was performed to assess the association between HHR and the prevalence of hypertension (Table 2). When analyzed as a continuous variable, HHR was significantly and positively associated with the odds of hypertension when expressed per 0.1-unit increase. In Model 1, each 0.1-unit increase in HHR was associated with higher odds of hypertension (OR = 1.06, 95% CI: 1.05–1.07, P < 0.0001). This association remained significant after adjusting for demographic and clinical factors in Model 2 (OR = 1.07, 95% CI: 1.05–1.08, P < 0.0001). After further adjustment for laboratory parameters in Model 3, the association was attenuated but remained statistically significant (OR = 1.02, 95% CI: 1.01–1.04, P = 0.0018). The ROC analysis was performed to determine the HHR cutoff for categorical grouping. Based on the maximum Youden index, the corresponding HHR cutoff value was 0.121, which was rounded to 0.12 for subsequent categorical analyses. The AUC was 0.581 (95% CI: 0.569–0.592), with a sensitivity of 0.425, specificity of 0.698, and Youden index of 0.123 (Supplementary Table 2). This cutoff was applied solely for categorical analysis and was not intended to establish the clinical screening utility of HHR. Similarly, when HHR was categorized based on the ROC curve threshold of 0.12, participants in the high HHR group had significantly higher odds of hypertension compared with those in the low HHR group. The odds ratio was 1.71 (95% CI: 1.57–1.86, P < 0.0001) in Model 1 and remained significant in Model 2 (OR = 1.77, 95% CI: 1.62–1.93, P < 0.0001) and Model 3 (OR = 1.38, 95% CI: 1.24–1.54, P < 0.0001). In the quartile-based sensitivity analysis, higher HHR categories were generally associated with increased odds of hypertension. In the fully adjusted model, compared with participants in the lowest HHR quartile, those in Q2, Q3, and Q4 had higher odds of hypertension, with ORs of 1.225 (95% CI: 1.072–1.401, P = 0.003), 1.255 (95% CI: 1.086–1.452, P = 0.002), and 1.631 (95% CI: 1.383–1.923, P < 0.001), respectively (Supplementary Table 3). Collectively, these results indicate a consistent positive association between elevated HHR and hypertension prevalence, which persisted even after adjusting for potential demographic, clinical, and biochemical confounding factors. Collinearity diagnostics showed no evidence of severe multicollinearity in the fully adjusted model. All adjusted GVIF values were below 3.0, ranging from 1.013 to 2.956, with the highest adjusted GVIF observed for TC (Supplementary Table 1). Survey-weighted sensitivity analyses using the available CHARLS biomarker sampling weights and community-level clustering yielded results generally consistent with the primary unweighted models. Among the 11,797 participants included in the primary analysis, 11,598 were included in the weighted sensitivity analysis. In the fully adjusted weighted model, HHR remained significantly associated with hypertension both as a continuous variable per 0.1-unit increase and as a categorical variable using the 0.12 cutoff (Supplementary Table 4). In addition, comparative ROC analyses were performed to evaluate the discriminatory performance of HHR and related metabolic indicators for prevalent hypertension. The AUCs were 0.580 for HHR, 0.541 for HbA1c, 0.568 for HDL-C, 0.591 for TyG index, and 0.566 for the LDL-C/HDL-C ratio. These results are presented in Supplementary Tables 5 and were considered supplementary descriptive analyses rather than evidence of clinical screening utility. Table 2Multivariate regression analysis of the association between HHR and hypertension prevalenceVariableModel 1Model 2Model 3OR (95% CI)P- valueOR (95% CI)P- valueOR (95% CI)P- valueHHR, per 0.1-unit increase1.06 (1.05, 1.07)< 0.00011.07 (1.05, 1.08)< 0.00011.02 (1.01, 1.04)0.0018Low HHR (≤ 0.12)Ref.Ref.Ref.High HHR (> 0.12)1.71 (1.57, 1.86)< 0.00011.77 (1.62, 1.93)< 0.00011.38 (1.24, 1.54)< 0.0001Model 1 was unadjusted. Model 2 was adjusted for gender, age, education level, smoking status, drinking status, and BMI. Model 3 was further adjusted for FBG, Scr, BUN, SUA, CRP, LDL-C, TC, and TG on the basis of Model 2. HHR was analyzed as a continuous variable per 0.1-unit increase and as a categorical variable according to the ROC-derived threshold. HHR, hemoglobin A1c to high-density lipoprotein cholesterol ratio Table 2Multivariate regression analysis of the association between HHR and hypertension prevalenceVariableModel 1Model 2Model 3OR (95% CI)P- valueOR (95% CI)P- valueOR (95% CI)P- valueHHR, per 0.1-unit increase1.06 (1.05, 1.07)< 0.00011.07 (1.05, 1.08)< 0.00011.02 (1.01, 1.04)0.0018Low HHR (≤ 0.12)Ref.Ref.Ref.High HHR (> 0.12)1.71 (1.57, 1.86)< 0.00011.77 (1.62, 1.93)< 0.00011.38 (1.24, 1.54)< 0.0001Model 1 was unadjusted. Model 2 was adjusted for gender, age, education level, smoking status, drinking status, and BMI. Model 3 was further adjusted for FBG, Scr, BUN, SUA, CRP, LDL-C, TC, and TG on the basis of Model 2. HHR was analyzed as a continuous variable per 0.1-unit increase and as a categorical variable according to the ROC-derived threshold. HHR, hemoglobin A1c to high-density lipoprotein cholesterol ratio Multivariate regression analysis of the association between HHR and hypertension prevalence Model 1 was unadjusted. Model 2 was adjusted for gender, age, education level, smoking status, drinking status, and BMI. Model 3 was further adjusted for FBG, Scr, BUN, SUA, CRP, LDL-C, TC, and TG on the basis of Model 2. HHR was analyzed as a continuous variable per 0.1-unit increase and as a categorical variable according to the ROC-derived threshold. HHR, hemoglobin A1c to high-density lipoprotein cholesterol ratio Nonlinear association between HHR and the prevalence of hypertension Restricted cubic spline analysis demonstrated a significant nonlinear association between HHR and hypertension prevalence (P for overall < 0.001; P for nonlinear < 0.001). As shown in Fig. 2, the adjusted odds of hypertension increased markedly when HHR exceeded approximately 0.10 and reached the highest level around 0.15. Thereafter, the curve gradually attenuated rather than showing a continuously increasing trend. However, this attenuation in the upper HHR range should be interpreted cautiously because observations were relatively sparse at higher HHR values, resulting in wider confidence intervals and less stable spline estimates. Overall, these findings indicate a nonlinear relationship between HHR and hypertension, with the most prominent increase in hypertension odds occurring at moderately elevated HHR levels. Fig. 2Restricted cubic spline analysis of the nonlinear association between HHR and hypertension prevalence Restricted cubic spline analysis of the nonlinear association between HHR and hypertension prevalence Subgroup analyses of the association between HHR and hypertension Figure 3 presents subgroup analyses evaluating the association between high HHR and hypertension across demographic, lifestyle, and clinical strata. In the overall population, high HHR was associated with significantly higher odds of hypertension after full adjustment (OR = 1.38, 95% CI: 1.24–1.54, P < 0.001). This positive association was generally consistent across age, gender, education, dyslipidemia, drinking, smoking, BMI, and FBG subgroups. No statistically significant interactions were observed across the predefined subgroups. Although the association appeared numerically stronger among participants with FBG ≥ 126 mg/dL, the interaction did not reach statistical significance (P for interaction = 0.080). Overall, the forest plot supports a generally consistent association between high HHR and higher odds of hypertension across major population subgroups. Fig. 3Adjusted subgroup analysis of the association between HHR and hypertension prevalence Adjusted subgroup analysis of the association between HHR and hypertension prevalence Discussion In this large-scale, nationally representative cross-sectional study based on data from the CHARLS cohort, we identified a significant positive association between the HHR and the prevalence of hypertension in Chinese adults aged 45 years and older. Participants with elevated HHR had significantly higher odds of hypertension across all models, even after adjusting for demographic, lifestyle, and metabolic confounders. This relationship remained significant whether HHR was treated as a continuous variable or dichotomized based on the optimal ROC-derived threshold. Furthermore, restricted cubic spline analysis revealed a nonlinear association, with a marked increase in hypertension risk when HHR exceeded approximately 0.10. Subgroup analyses showed that this association was generally consistent across sex, age, education level, and cardiometabolic status. Although the association appeared numerically stronger among individuals with elevated fasting blood glucose, the interaction did not reach statistical significance and should therefore be interpreted cautiously. These findings suggest that HHR may serve as a simple composite biomarker associated with prevalent hypertension, reflecting the combined burden of dysglycemia and dyslipidemia. Although the association between HHR and hypertension has not been fully characterized, its pathophysiological basis is well established, and our study expands upon existing concepts in this field [13]. Previous evidence suggests that chronic glycemic exposure and lipid abnormalities are closely related to endothelial dysfunction, arterial stiffness, vascular inflammation, and cardiometabolic risk [14–16]. In addition, clinical evidence regarding the individual components indicates that elevated HbA1c is not only a marker of diabetes but also a major driver of microvascular complications such as diabetic retinopathy [17]; conversely, reduced HDL-C levels are closely associated with the severity of coronary artery disease and endothelial dysfunction [18]. Furthermore, metabolic ratios, such as the TyG index, may provide complementary information beyond single markers in cardiovascular risk assessment [19]. While prior research has confirmed the individual associations of HbA1c and HDL-C with blood pressure and cardiovascular risk, few studies have assessed the combined effect of their ratio [20, 21]. The rationale for integrating these biomarkers lies in their complementary roles in cardiovascular metabolic regulation. HbA1c reflects chronic glucose exposure and is associated with endothelial dysfunction, vascular inflammation, and increased arterial stiffness [22, 23]. In contrast, HDL-C exerts vascular protective effects, including anti-inflammatory, antioxidant, and endothelial support [24]. Therefore, a high HHR may reflect a state of simultaneous hyperglycemia and HDL deficiency, an atherogenic metabolic profile that may predispose individuals to vascular dysfunction and hypertension. Several biological mechanisms may explain the observed association between elevated HHR and hypertension. Higher HbA1c reflects chronic glycemic exposure, which may promote oxidative stress, advanced glycation end-product formation, endothelial dysfunction, and vascular inflammation. These processes can impair nitric oxide bioavailability, increase arterial stiffness, and contribute to vascular remodeling [25]. In parallel, lower HDL-C may weaken antioxidative, anti-inflammatory, and endothelial-protective effects, thereby further aggravating vascular injury [26]. Elevated HHR may therefore represent a combined metabolic state characterized by dysglycemia and impaired lipid-related vascular protection. Insulin resistance may also link elevated HHR to hypertension through sympathetic nervous system activation, sodium retention, vascular smooth muscle proliferation, and activation of the renin-angiotensin-aldosterone system [27]. Collectively, these mechanisms provide biological plausibility for the association between elevated HHR and higher odds of prevalent hypertension. The observed nonlinear relationship between HHR and hypertension is particularly noteworthy. Our restricted cubic spline analysis revealed a relatively stable risk profile at lower HHR levels, but a sharp rise in the odds of hypertension once HHR exceeded 0.10. The curve subsequently attenuated at higher HHR levels, suggesting that the association was nonlinear rather than continuously increasing. Nevertheless, this attenuation above approximately 0.15 should not be overinterpreted, because the number of participants in the upper HHR range was relatively small and the corresponding spline estimates were less stable. These findings indicate that moderately elevated HHR levels may be associated with a higher likelihood of prevalent hypertension, although the cross-sectional design precludes any inference regarding prediction or causality. Subgroup analyses showed that the association between HHR and hypertension was generally consistent across the predefined strata. Although the association appeared numerically stronger among individuals with elevated FBG (≥ 126 mg/dL), the interaction did not reach statistical significance. Therefore, this subgroup finding should be interpreted cautiously and considered exploratory. These results are consistent with prior evidence highlighting the complex interplay between insulin resistance, lipid metabolism, and vascular tone regulation [28, 29]. Our findings may have public health relevance. HHR is derived from two routine, standardized, and widely available laboratory measurements. Unlike more complex insulin resistance indices requiring insulin measurements, HHR can be easily calculated in routine clinical or epidemiological settings. However, because this study was cross-sectional and formal prediction or reclassification analyses were not the primary focus, HHR should be interpreted as a biomarker associated with prevalent hypertension rather than as an established screening or predictive tool. Future prospective studies are needed to determine whether HHR provides incremental value for hypertension risk assessment beyond established clinical and metabolic risk factors. This study has several notable strengths. First, the use of data from the nationally representative CHARLS enhances the generalizability of our findings to middle-aged and older Chinese adults. Second, the large sample size provided sufficient statistical power to detect associations and examine interactions across diverse subgroups. Third, adjustment for a wide range of demographic, clinical, and biochemical covariates improves the stability of our results regarding the association between HHR and hypertension. Fourth, the use of RCS analysis allowed for characterization of the nonlinear association between HHR and hypertension. However, several limitations warrant consideration. First, the cross-sectional design precludes causal inference; thus, we cannot determine whether elevated HHR precedes hypertension or results from shared metabolic disturbances. Future longitudinal studies are needed to establish temporal relationships. Second, despite adjusting for numerous confounders, residual confounding from unmeasured variables cannot be excluded. Third, hypertension diagnosis relied partially on self-reported data, which is subject to recall bias; however, the inclusion of objective blood pressure measurements and medication history likely mitigated this concern. Fourth, as our cohort consisted of Chinese adults aged ≥ 45 years, findings may not be generalizable to younger populations or other ethnic groups. Fifth, single measurements of blood biomarkers precluded the assessment of intra-individual variability. Sixth, formal discrimination and reclassification analyses, such as net reclassification improvement or integrated discrimination improvement, were not performed; therefore, the clinical screening utility of HHR cannot be established based on the present study. Finally, more than half of the initially identified participants were excluded because of missing biomarker, blood pressure, demographic, or medication-related information. Although we compared baseline characteristics between included and excluded participants, potential selection bias due to missing data and analytic exclusions cannot be completely excluded. Excluding participants receiving antihyperglycemic therapy, while necessary to minimize confounding from pharmacologic glucose control, may also have selected a relatively healthier subset of participants. Conclusion In conclusion, HHR was significantly and nonlinearly associated with the prevalence of hypertension among middle-aged and older Chinese adults. These findings suggest that HHR may reflect an adverse metabolic profile characterized by concurrent dysglycemia and dyslipidemia. Given the cross-sectional design, the present findings should be interpreted as evidence of association rather than prediction or causality. Further prospective studies are warranted to validate these findings and to determine whether HHR provides incremental value for hypertension risk assessment. The green solid line represents the adjusted odds ratio for hypertension, and the light green shaded area represents the 95% confidence interval. The model was adjusted for gender, age, education level, smoking status, drinking status, BMI, FBG, Scr, BUN, SUA, CRP, LDL-C, TC, and TG. Odds ratios were estimated for high HHR versus low HHR. The model was adjusted for gender, age, education level, smoking status, drinking status, BMI categories, FBG, Scr, BUN, SUA, CRP, LDL-C, TC, and TG. In each subgroup analysis, the stratification variable itself was not included as an adjustment covariate. Supplementary Information Supplementary Material 1. Supplementary Material 1. Supplementary Material 2. Supplementary Material 2. Supplementary Material 3. Supplementary Material 3. Supplementary Material 4. Supplementary Material 4. Supplementary Material 5. Supplementary Material 5. Supplementary Material 6. Supplementary Material 6.