Association of high-sensitivity C-reactive protein-to-hemoglobin ratio with stroke risk: A study based on the China Health and Retirement Longitudinal Study and English Longitudinal Study of Ageing cohorts

Abstract

Objective

To investigate the association between the high-sensitivity C-reactive protein-to-hemoglobin ratio and stroke risk among middle-aged and older adults.

Methods

A prospective follow-up study was conducted using cohort data from the China Health and Retirement Longitudinal Study (CHARLS) and the English Longitudinal Study of Ageing (ELSA). Participants without a history of stroke at baseline were included, comprising 5368 individuals from CHARLS and 1422 from ELSA. Cox proportional hazards regression models were used to analyze the association between C-reactive protein-to-hemoglobin ratio levels and stroke risk, along with dose–response analyses and subgroup interaction testing.

Results

After adjusting for sex, age, lifestyle factors, and multiple clinical biochemical indicators, the participants in Q2, Q3, and Q4 demonstrated a significantly higher risk of stroke than those in Q1 (reference group). The hazard ratios were 2.23 (1.35–3.69), 2.24 (1.36–3.71), and 3.14 (1.93–5.12), respectively (all P < 0.01). Dose–response analysis revealed an approximately linear positive association between C-reactive protein-to-hemoglobin ratio and stroke risk.

Conclusions

C-reactive protein-to-hemoglobin ratio levels are significantly and positively associated with stroke risk in middle-aged and older adults.

Keywords

High-sensitivity C-reactive protein hemoglobin high-sensitivity C-reactive protein-to-hemoglobin ratio stroke cohort study

Introduction

Stroke is an acute cerebrovascular disease characterized primarily by vascular occlusion or hemorrhage, leading to ischemia, hypoxia, tissue necrosis, and subsequent neurologic deficits.¹ In China, population aging, changing lifestyles, and the increasing prevalence of risk factors such as hypertension and diabetes have contributed to a continuous rise in stroke incidence, disability, and recurrence rates. This trend places a substantial burden on individuals, family caregivers, and healthcare systems, making stroke a major chronic disease and a serious threat to public health in China.² Extensive research has demonstrated that stroke results from the combined effects of genetic, environmental, and lifestyle factors. Early identification and intervention among high-risk populations are crucial for reducing stroke incidence and mortality. Currently, commonly recognized stroke risk factors include age, sex, hypertension, diabetes, dyslipidemia, obesity, smoking, and alcohol consumption.³ However, these traditional risk factors are limited in scope and do not comprehensively identify all individuals at high risk. Some individuals without apparent conventional risk factors may still experience a stroke. Therefore, there is an urgent need to identify novel, simple and practical biomarkers to improve stroke risk assessment.

High-sensitivity C-reactive protein (hsCRP) is a liver-derived protein produced in response to inflammation and serves as a sensitive biomarker of systemic low-grade inflammation. Elevated hsCRP levels are closely associated with vascular endothelial injury, atherosclerotic plaque formation, and plaque instability.⁴ Multiple studies have shown that elevated hsCRP levels are an independent risk factor for stroke, with higher concentrations associated with an increased risk of both incident and recurrent stroke.^5,6 Hemoglobin (Hb), is the primary protein in red blood cells, is responsible for transporting oxygen and carbon dioxide. Changes in hemoglobin concentration directly affect the oxygen-carrying capacity of blood and tissue perfusion. Anemia can result in inadequate oxygen delivery to brain tissue, leading to vascular endothelial dysfunction, increased blood viscosity, and a greater risk of thrombosis. In addition, anemia increases cardiac workload, thereby indirectly increasing the risk of cerebrovascular disease. Previous studies have identified anemia as an important risk factor for stroke among middle-aged and older adults.^7,8

The high-sensitivity C-reactive protein-to-hemoglobin ratio (CHR) has emerged as a novel composite biomaker because it integrates information from two key pathophysiological processes: systemic inflammation and blood oxygen-carrying capacity. As such, it may provide a more comprehensive reflection of an individual's pathological status. However, studies investigating the association between CHR and stroke risk remain limited, and prospective evidence from large, multicohort studies is lacking. Consequently, the clinical utility of CHR for stroke risk prediction remains unclear.

In this study, we used data from two independent large-scale community-based cohorts—the China Health and Retirement Longitudinal Study (CHARLS) and the English Longitudinal Study of Ageing (ELSA)—to conduct a prospective investigation. We aimed to examine the association between CHR and incident stroke among community-dwelling middle-aged and older adults using two independent prospective cohorts.

Methods

Data source and participants

The present study utilized data from the CHARLS dataset, which includes Chinese adults aged >45 years. Data from Wave 1 (2011) to Wave 5 (2020) were used, with Wave 1 designated as the baseline. The study was conducted in accordance with the guidelines established by the Peking University Medical Ethics Committee, and all participants provided written informed consent. A total of 23,962 participants were screened. Participants with a history of stroke at baseline were excluded. Participants with missing data were also excluded from the analysis. The final sample comprised 5368 elderly participants who completed follow-up (Figure 1). For the ELSA study, data from Wave 2 (2004–2005) to Wave 10 (2020–2021) were utilized, with Wave 2 serving as the baseline. The objectives of ELSA are similar to those of CHARLS, namely to investigate health status, socioeconomic conditions, and associated determinants among individuals aged ≥50 years in the UK. Complete data from Waves 2 to 10 were considered for analysis. Using the same study design and screening criteria as those applied to the CHARLS cohort, participants with a prior diagnosis of stroke at baseline, those without the key indicators required to calculate CHR, and those with substantial missing data were excluded. The final analysis included 1422 participants without a history of stroke for external validation and synchronous analysis (Figure 1). This study was conducted in accordance with the principles of the Declaration of Helsinki (1975, revised 2024). The reporting of this study conforms to Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.⁹ The baseline interview date was used as the starting point for follow-up. Follow-up time was calculated from baseline to the first occurrence of stroke, death, loss to follow-up, or the end of the study, whichever came first. All participant data were de-identified prior to analysis.

Figure 1.

Flowchart of the population included in our study.

Data collection

The study dataset contained basic demographic characteristics of respondents collected through household interviews. Confounding variables included sex, age, marital status, educational level, and smoking and drinking habits. Furthermore, respondents were asked to report previously diagnosed chronic conditions, including hypertension, stroke, diabetes, kidney disease, and cardiovascular disease. A comprehensive set of physical measurements was recorded, including height, weight, and body mass index (BMI). This study focused on a series of hematological and biochemical indicators, including fasting blood glucose (FBG), glycated hemoglobin (HbA1c), total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), blood urea nitrogen (BUN), uric acid (UA), creatinine, Hb, and hsCRP. All participants fasted for a minimum of 8 h, and the relevant parameters were measured using automated blood analysis equipment.

Definition of stroke

The primary outcome of this study was incident stroke, as determined by self-reported physician diagnosis. Participants asked whether they had ever received a diagnosis of stroke from a healthcare professional. Those who responded affirmatively were classified as having a history of stroke. Participants without a history of stroke at baseline were included, and the first reported physician diagnosis of stroke during follow-up was considered the endpoint event. The specific questions used to ascertain stroke diagnosis were as follows: in CHARLS, “Has a doctor ever told you that you have had a stroke?”; and in ELSA, “Has a doctor ever diagnosed you with a stroke?.”^10–12

CHR

CHR was calculated as follows: CHR = hsCRP (mg/dL)/Hb (g/L). The hsCRP values were expressed in mg/dL and the Hb values in g/L. No unit conversion was performed in this study; CHR was calculated directly using the units recorded in the original database. Participants were categorized into four quartiles (Q1–Q4) according to CHR levels, with Q1 representing the lowest quartile and serving as the reference group.

Statistical analysis

All statistical analyses were performed using R version 4.2. Normally distributed continuous variables are presented as mean ± standard deviation (x ± s), whereas non-normally distributed variables are presented as median (interquartile range). Categorical variables are presented as frequencies and percentages. Analysis of variance or rank-sum tests was performed on the baseline data for each group. In the event that the variables satisfy the assumptions of normality and homogeneity of variance, a one-way analysis of variance should be employed to compare the four CHR quantiles (Q1–Q4). In the case of non-normally distributed data, the Kruskal–Wallis test is to be used. Several Cox proportional hazards regression models were constructed to investigate the association between CHR and incident stroke. The proportional hazards assumption was assessed using Schoenfeld residual tests for CHR and all major covariates. Analysis of the rank correlation between residuals and survival time showed that all p-values were >0.05. Confounding factors were selected based on previous research and clinical causal pathways. The findings of the multicollinearity analysis showed that all variance inflation factor values were <5, indicating no significant multicollinearity (Table S1). Restricted cubic spline (RCS) analyses were performed to further evaluate the nonlinear association between CHR and stroke incidence. The RCS uses four nodes, located at the 5th, 25th, 75th, and 95th percentiles, respectively. In addition, subgroup analyses were also conducted in this study. In the context of all statistical analyses, a p value <0.05 was considered to indicate statistical significance.

Results

CHARLS cohort baseline characteristics

The present study included a total of 5368 participants from the CHARLS cohort, comprising 5157 individuals in the normal group and 211 comprising the stroke group. The findings of the study demonstrated that the stroke group was significantly older (p < 0.001), with considerably higher levels of TC, LDL-C, TG, HbA1c, creatinine, and UA in comparison to the normal group (all p < 0.05). The distribution of CHR quantiles differed markedly (p < 0.001), with the highest proportion (37.91%) in the stroke group's CHR quartile 4 (Q4), whereas the normal group exhibited uniform distribution across quantiles. With regard to comorbidities, the stroke group had a significantly higher prevalence of kidney disease and cardiovascular disease (both p < 0.05). FBG, educational level, and hypertension prevalence approached statistical significance, whereas sex, alcohol consumption, smoking status, marital status, and BMI showed no significant differences between groups (all p > 0.05) (Table 1).

Table 1.

Baseline characteristics of included participants in CHARLS.

Variables	CHARLS
	Total(n = 5368)	Stroke(n = 211)	Normal(n = 5157)	p
Age (years)	57.78 ± 8.51	60.48 ± 8.21	57.67 ± 8.50	<0.001
BMI (kg/m²)	24.29 ± 3.11	24.18 ± 3.93	24.29 ± 3.90	0.971
FBG (mg/dL)	107.53 ± 29.08	111.36 ± 37.35	107.37 ± 28.69	0.051
TC (mg/dL)	193.58 ± 37.74	200.31 ± 36.77	193.30 ± 37.76	0.008
HDL-C (mg/dL)	51.86 ± 15.06	51.01 ± 14.99	51.89 ± 15.06	0.405
LDL-C (mg/dL)	117.87 ± 34.14	124.02 ± 34.02	117.61 ± 34.12	0.008
TG (mg/dL)	124.75 ± 74.16	135.43 ± 76.09	124.32 ± 74.06	0.033
BUN (mg/dL)	15.64 ± 4.41	15.71 ± 4.13	15.64 ± 4.42	0.832
Creatinine (mg/dL)	0.77 ± 0.18	0.80 ± 0.20	0.77 ± 0.18	0.011
UA (mg/dL)	4.38 ± 1.21	4.60 ± 1.29	4.37 ± 1.21	0.008
HbA1c (%)	5.23 ± 0.75	5.41 ± 1.00	5.22 ± 0.74	0.007
Hemoglobin (g/L)	14.45 ± 2.22	14.57 ± 2.22	14.45 ± 2.22	0.457
hsCRP (mg/dL)	2.00 ± 5.13	2.53 ± 4.48	1.98 ± 5.15	0.128
CHR quantile, n (%)				<0.001
Q1	1342 (25.00)	22 (10.43)	1320 (25.60)
Q2	1342 (25.00)	53 (25.12)	1289 (25.00)
Q3	1342 (25.00)	56 (26.54)	1286 (24.94)
Q4	1342 (25.00)	80 (37.91)	1262 (24.47)
Sex, n (%)				0.251
Male	2446 (45.57)	88 (41.71)	2358 (45.72)
Female	2922 (54.43)	123 (58.29)	2799 (54.28)
Drinking, n (%)				0.217
No	3580 (66.69)	149 (70.62)	3431 (66.53)
Yes	1788 (33.31)	62 (29.38)	1726 (33.47)
Smoking, n (%)				0.375
Never smokers	3797 (70.73)	155 (73.46)	3642 (70.62)
Ever smokers	1571 (29.27)	56 (26.54)	1515 (29.38)
Marital status, n (%)				0.795
Married or partnered	4587 (85.45)	179 (84.83)	4408 (85.48)
Other status	781 (14.55)	32 (15.17)	749 (14.52)
Educational level, n (%)				0.076
Below high school	2475 (46.11)	109 (51.66)	2366 (45.88)
High school	1191 (22.19)	50 (23.70)	1141 (22.13)
College or above	1702 (31.71)	52 (24.64)	1650 (32.00)
Hypertension, n (%)				0.066
No	3470 (64.69)	124 (58.77)	3346 (64.93)
Yes	1894 (35.31)	87 (41.23)	1807 (35.07)
Diabetes, n (%)				0.303
No	4654 (87.60)	180 (85.31)	4474 (87.69)
Yes	659 (12.40)	31 (14.69)	628 (12.31)
Kidney, n (%)				0.013
Yes	339 (6.34)	22 (10.43)	317 (6.18)
No	5005 (93.66)	189 (89.57)	4816 (93.82)
Cardiovascular disease, n (%)				0.007
Yes	562 (10.51)	34 (16.11)	528 (10.28)
No	4783 (89.49)	177 (83.89)	4606 (89.72)

Comparison of baseline clinical characteristics between stroke and nonstroke patients at the end of follow-up.

BMI: body mass index; BUN: blood urea nitrogen; CHARLS: China Health and Retirement Longitudinal Study; CHR: C-reactive protein-to-hemoglobin ratio; FBG: fasting blood glucose; HbA1c: glycated hemoglobin; HDL-C: high-density lipoprotein cholesterol; hsCRP: high-sensitivity C-reactive protein; LDL-C: low-density lipoprotein cholesterol; Q: quartile; TC: total cholesterol; TG: triglyceride; UA: uric acid.

Participants were divided into four groups according to CHR quartiles (1342 individuals per group). In the context of continuous variables, it was observed that age, FBG, TC, LDL-C, TG, UA, HbA1c, and hsCRP exhibited significant increases with rising CHR quartiles (all p < 0.001). In contrast, HDL-C and Hb levels demonstrated a significant decrease (all p < 0.001). Creatinine showed a statistically significant yet modest discrepancy (p = 0.019), whereas BMI and BUN demonstrated no intergroup variations (p > 0.05). Among the categorical variables, only alcohol consumption status demonstrated significant differences (p = 0.003), while demographic characteristics such as sex and smoking status exhibited no intergroup variations. With regard to comorbidities, the prevalence of hypertension, diabetes mellitus, cardiovascular disease, and stroke significantly increased with rising CHR (all p < 0.05), whereas the prevalence of kidney disease showed no difference (p = 0.792) (Table 2).

Table 2.

Baseline characteristics of included participants in CHARLS.

Variables	CHR
Variables	Q1(n = 1342)	Q2(n = 1342)	Q3(n = 1342)	Q4(n = 1342)	p
Age (years)	56.36 ± 8.20	57.46 ± 8.39	58.34 ± 8.50	58.94 ± 8.71	<0.001
BMI (kg/m²)	22.19 ± 3.18	24.29 ± 5.28	24.15 ± 3.66	26.48 ± 3.63	0.064
FBG (mg/dL)	102.81 ± 20.77	105.66 ± 25.96	110.40 ± 32.52	111.23 ± 34.27	<0.001
TC (mg/dL)	189.99 ± 34.66	191.60 ± 36.61	196.81 ± 38.11	195.90 ± 40.90	<0.001
HDL-C (mg/dL)	56.46 ± 15.66	52.36 ± 14.61	50.08 ± 14.20	48.53 ± 14.55	<0.001
LDL-C (mg/dL)	113.45 ± 31.01	116.28 ± 32.74	121.23 ± 34.53	120.50 ± 37.39	<0.001
TG (mg/dL)	107.07 ± 61.05	121.42 ± 70.86	135.26 ± 78.75	135.27 ± 80.73	<0.001
BUN (mg/dL)	15.67 ± 4.64	15.59 ± 4.35	15.79 ± 4.23	15.52 ± 4.42	0.436
Creatinine (mg/dL)	0.76 ± 0.17	0.78 ± 0.18	0.78 ± 0.19	0.78 ± 0.18	0.019
UA (mg/dL)	4.06 ± 1.10	4.38 ± 1.21	4.44 ± 1.20	4.65 ± 1.26	<0.001
HbA1c (%)	5.13 ± 0.57	5.15 ± 0.65	5.30 ± 0.81	5.33 ± 0.90	<0.001
Hemoglobin (g/L)	14.74 ± 2.40	14.39 ± 2.18	14.52 ± 2.19	14.16 ± 2.09	<0.001
hsCRP (mg/dL)	0.32 (0.26,0.39)	0.63 (0.50,0.77)	1.28 (1.09,1.51)	3.13 (2.27,5.13)	<0.001
Sex, n (%)					0.155
Male	640 (47.69)	623 (46.42)	591 (44.04)	592 (44.11)
Female	702 (52.31)	719 (53.58)	751 (55.96)	750 (55.89)
Drinking, n (%)					0.003
No	865 (64.46)	861 (64.16)	925 (68.93)	929 (69.23)
Yes	477 (35.54)	481 (35.84)	417 (31.07)	413 (30.77)
Smoke, n (%)					0.786
Never smokers	950 (70.79)	935 (69.67)	956 (71.24)	956 (71.24)
Ever smokers	392 (29.21)	407 (30.33)	386 (28.76)	386 (28.76)
Marital status, n (%)					0.264
Married or partnered	1138 (84.80)	1140 (84.95)	1169 (87.11)	1140 (84.95)
Other status	204 (15.20)	202 (15.05)	173 (12.89)	202 (15.05)
Educational level, n (%)					0.262
Below high school	590 (43.96)	638 (47.54)	622 (46.35)	625 (46.57)
High school	307 (22.88)	312 (23.25)	286 (21.31)	286 (21.31)
College or above	445 (33.16)	392 (29.21)	434 (32.34)	431 (32.12)
Hypertension, n (%)					<0.001
No	965 (71.91)	899 (67.04)	846 (63.13)	760 (56.67)
Yes	377 (28.09)	442 (32.96)	494 (36.87)	581 (43.33)
Diabetes, n (%)					<0.001
No	1224 (91.75)	1177 (88.63)	1134 (85.84)	1119 (84.14)
Yes	110 (8.25)	151 (11.37)	187 (14.16)	211 (15.86)
Kidney, n (%)					0.792
Yes	82 (6.13)	79 (5.93)	87 (6.51)	91 (6.81)
No	1256 (93.87)	1254 (94.07)	1249 (93.49)	1246 (93.19)
Cardiovascular disease, n (%)					0.003
Yes	117 (8.74)	124 (9.30)	152 (11.38)	169 (12.65)
No	1222 (91.26)	1210 (90.70)	1184 (88.62)	1167 (87.35)
Stroke, n (%)					<0.001
Yes	22 (1.64)	53 (3.95)	56 (4.17)	80 (5.96)
No	1320 (98.36)	1289 (96.05)	1286 (95.83)	1262 (94.04)

Comparison of baseline clinical characteristics between stroke and nonstroke patients at the end of follow-up.

CHR and stroke risk

In the CHARLS cohort, CHR was significantly and positively associated with incident stroke across all multivariable models (Table 3). In Model 4, after adjustment for all potential confounders, Q2, Q3, and Q4 groups had significantly higher stroke risks than those in comparison to Q1, with hazard ratios (HRs) (95% confidence interval (CI)) of 2.23 (1.35–3.69), 2.24 (1.36–3.71), and 3.14 (1.93–5.12), respectively (all p < 0.01), thus indicating a clear dose–response relationship.

Table 3.

Cox regression analysis of CHR in relation to new-onset stroke in CHARLS.

Variables	Model 1		Model 2		Model 3		Model 4
Variables	HR (95% CI)	p	HR (95% CI)	p	HR (95% CI)	p	HR (95% CI)	p
CHR
Q1	1.00 (Reference)		1.00 (Reference)		1.00 (Reference)		1.00 (Reference)
Q2	2.42 (1.47–3.98)	<0.001	2.33 (1.42–3.84)	<0.001	2.34 (1.42–3.84)	<0.001	2.23 (1.35–3.69)	0.002
Q3	2.59 (1.58–4.24)	<0.001	2.43 (1.48–3.98)	<0.001	2.40 (1.46–3.94)	<0.001	2.24 (1.36–3.71)	0.002
Q4	3.75 (2.34–6.01)	<0.001	3.47 (2.16–5.57)	<0.001	3.39 (2.11–5.46)	<0.001	3.14 (1.93–5.12)	<0.001

Model 1: Crude. Model 2: Adjust: sex, drinking, smoking, educational level, and age. Model 3: Adjust: sex, drinking, smoking, hypertension, diabetes, marital status, educational level, cardiovascular disease, kidney, and age. Model 4: Adjust: sex, drinking, smoking, hypertension, diabetes, marital status, educational level, cardiovascular disease, kidney, age, HDL-C, LDL-C, TG, TC, BUN, creatinine, UA, FBG, and HbA1c.

CHR: C-reactive protein-to-hemoglobin ratio; CI: confidence interval; HR: hazard ratio; Q: quartile.

The results of the Cox regression analysis in the ELSA cohort were consistent with those observed in the CHARLS cohort (Table 4). After adjustment for all potential confounding variables, the Q2, Q3, and Q4 groups demonstrated a marked increase in stroke risk compared with those in the Q1 group, with HRs (95% CIs) of 2.66 (1.54–4.57), 3.91 (2.31–5.78), and 4.22 (1.81–7.67), respectively (all p < 0.001). This finding serves to further substantiate the established positive association between CHR and the risk of stroke.

Table 4.

Cox regression analysis of CHR in relation to new-onset stroke in ELSA.

Variables	Model 1		Model 2		Model 3		Model 4
Variables	HR (95% CI)	p	HR (95% CI)	p	HR (95% CI)	p	HR (95% CI)	p
CHR
Q1	1.00 (Reference)		1.00 (Reference)		1.00 (Reference)		1.00 (Reference)
Q2	2.83 (1.65–4.85)	<0.001	2.63 (1.53–4.51)	<0.001	2.62 (1.53–4.50)	<0.001	2.66 (1.54–4.57)	<0.001
Q3	4.39 (2.62–7.35)	<0.001	3.97 (2.36–6.68)	<0.001	3.93 (2.33–6.61)	<0.001	3.91 (2.31–6.61)	<0.001
Q4	4.67 (2.79–7.84)	<0.001	4.05 (2.39–6.85)	<0.001	3.97 (2.34–6.74)	<0.001	3.82 (2.22–6.58)	<0.001

CHR: C-reactive protein-to-hemoglobin ratio; CI: confidence interval; HR: hazard ratio; Q: quartile.

Subgroup analysis

Subgroup analyses showed that the positive association between CHR and stroke risk remained consistent across subgroups defined by sex, age (<60 years/≥60 years), BMI (<28 kg/m²/≥28 kg/m²), smoking status, alcohol consumption, marital status, educational level, hypertension, diabetes, heart disease, and kidney disease (Figure 2). Interaction analyses revealed no significant interactions between these factors and CHR (all p for interaction > 0.05), indicating that the association between CHR and stroke risk was consistent across different populations.

Figure 2.

Subgroup analyses of the association between CHR and stroke.

Dose–response relationship

RCS analysis indicates that there is an approximately linear positive association between CHR and the risk of stroke, with no significant nonlinear relationship (Figure 3). The p value for overall association was 0.010, whereas the p value for nonlinearity was 0.086 (Figure 3(a)). After adjustment for multiple confounding factors, the p value for overall association was 0.015, and the p value for nonlinearity was 0.080 (Figure 3(b)), further supporting a linear increase in stroke risk with increasing CHR levels.

Figure 3.

Results of nonlinear correlation analysis. (a) Nonlinear correlation analysis between CHR and stroke, unadjusted confounding factors. (b) Nonlinear correlation analysis between CHR and stroke, adjusted confounding factors. The adjusted confounding factors include sex, drinking, smoking, hypertension, diabetes, marital status, educational level, cardiovascular disease, kidney, age, HDL-C, LDL-C, TG, TC, BUN, creatinine, UA, FBG, and HbA1c.

Survival curve analysis

The survival curves showed that the cumulative probability of stroke occurrence among participants progressively increases with rising CHR quartiles (Figure 4). The probability of stroke occurrence in the Q4 group was significantly higher than that in the Q1 group (p < 0.001), further supporting the association between CHR levels and stroke risk.

Figure 4.

Kaplan–Meier event-free survival curve. Kaplan–Meier analysis of incident stroke based on CHR quartiles (log-rank, p < 0.001).

Discussion

This prospective study, based on two independent large-scale community cohorts (CHARLS and ELSA), examined the association between CHR and incident stroke among middle-aged and older adults. The study established a dose–response relationship and validated the applicability of CHR across different subpopulations. The findings demonstrated a significant positive linear association between CHR levels and stroke risk among middle-aged and older adults. This association remained robust after adjustment for multiple potential confounders and showed good stability following validation in the ELSA cohort. Subgroup analyses revealed that there were no significant differences in the association between CHR and stroke risk across different subgroups defined by sex, age, smoking status, alcohol consumption, hypertension, and diabetes. The findings of the dose–response analyses demonstrated a linear relationship, thereby indicating that individuals with elevated CHR levels are predisposed to a significantly increased risk of stroke.

Elevated hsCRP levels are associated with higher stroke risk through systemic inflammation and impaired oxygen transport. As a marker of chronic low-grade inflammation,^13–15 hsCRP contributes to atherosclerosis by inducing endothelial damage, promoting vascular smooth muscle cell proliferation and migration, accelerating lipid deposition, and destabilizing plaques, thereby increasing the risk of plaque rupture, thrombosis, and ischemic stroke.^4,16,17 Consistent with previous evidence, elevated hsCRP may serve as a plausible independent biomarker associated with stroke incidence and recurrence, with higher hsCRP levels corresponding to a greater likelihood of stroke events. However, the selection of confounding factors in this study was based on previous literature and clinical pathophysiological considerations. Model 4 was the fully adjusted model and included numerous biochemical markers closely related to the inflammatory and metabolic pathways involved in stroke pathogenesis. Therefore, the possibility of overadjustment cannot be discounted, and a cautious interpretation is warranted. Although Model 4 might not be regarded as the sole optimal model, the consistent direction and significance of the associations across all adjusted models lend support to the robustness of the primary findings.

Conversely, Hb is the principal oxygen-carrying protein in the circulation. Lower Hb levels may impair the oxygen-carrying capacity of blood and potentially result in inadequate cerebral oxygen delivery. Such pathophysiological alterations are biologically plausible to trigger a cascade of physiological perturbations that may correlate with an elevated likelihood of ischemic stroke.^18–20 First, cerebral hypoxia associated with reduced Hb might disturb vascular endothelial function, potentially upregulating endothelin-1 release and concurrently lowering nitric oxide bioavailability. This imbalance may promote vasoconstriction and vascular spasm, increase peripheral vascular resistance, and facilitate a prothrombotic state. Second, hypoxemia may stimulate compensatory overactive bone marrow hematopoiesis, which could raise red blood cell counts and blood viscosity while reducing cerebral perfusion. These changes may further aggravate cerebral ischemia and hypoperfusion, thereby increasing thrombotic risk. Anemia could also increase cardiac burden, with possible compensatory tachycardia and elevated cardiac output. Chronic anemic status may act as a predisposing factor for the development of cardiovascular comorbidities such as hypertension and heart failure, which are well-established stroke risk factors and may indirectly increase the susceptibility to stroke events.^7,21,22 Moreover, anemia may influence systemic inflammatory activity and contribute to elevated circulating hsCRP concentration levels. Collectively, the interaction between chronic inflammation and anemia may create a self-perpetuating vicious cycle that further increases stroke risk at the population level.

As a composite indicator integrating hsCRP and Hb, CHR simultaneously reflects both the body's chronic inflammatory status and blood oxygen-carrying capacity. Compared with a single indicator, it provides a more comprehensive and accurate reflection of the underlying pathophysiological processes associated with stroke risk. In this study, the association strength between CHR and stroke risk was greater than that reported in previous single-marker studies. Furthermore, following validation across multiple cohorts and sensitivity analyses, this association remained consistent, further supporting the reliability of the observed relationship between CHR and stroke risk. Nevertheless, in this long-term follow-up cohort of older adults, mortality from causes other than stroke represents an important competing event. It is evident that traditional Cox proportional hazards regression models are not equipped to account for the occurrence of competing risks. This inherent limitation may result in the overestimation or underestimation of the true association between the CHR and the risk of stroke, consequently leading to biased effect estimates. The present study did not employ a Fine-Gray competing risk model, limiting the ability to account for the potential influence of nonstroke mortality. It is recommended that future studies utilize a competing risks model to validate the observed associations.

The present study has several limitations. First, the study employed a complete-case analysis approach, resulting in a substantial reduction in sample size from the initially screened population to the final analytical cohort. A considerable reduction in sample size may result in the introduction of selection bias, as participants with complete data may exhibit systemic differences compared to those excluded because of missing data or other inclusion criteria. This may limit the generalizability of the research findings. Second, although the study adjusted for several potential confounding factors, unmeasured factors may still have an impact, and no validation of competing models was performed. Third, because both cohorts consisted primarily of middle-aged and older adults, selection bias may be present, and findings may not be generalizable to younger populations. Further studies in younger community-based cohorts to validate the value of CHR in stroke prevention. Finally, this study focused exclusively on the association between CHR and the risk of stroke; self-reporting is subject to misclassification and underreporting, and no distinction was made between ischemic and hemorrhagic stroke. Further investigations are warranted to enhance the clinical applicability of CHR in stroke research.

Conclusion

The present study demonstrates a significant positive linear association between CHR levels and stroke risk among middle-aged and older adults and shows that this relationship is consistent across different subpopulations.

Supplemental Material

sj-docx-1-imr-10.1177_03000605261459294 - Supplemental material for Association of high-sensitivity C-reactive protein-to-hemoglobin ratio with stroke risk: A study based on the China Health and Retirement Longitudinal Study and English Longitudinal Study of Ageing cohorts

Supplemental material, sj-docx-1-imr-10.1177_03000605261459294 for Association of high-sensitivity C-reactive protein-to-hemoglobin ratio with stroke risk: A study based on the China Health and Retirement Longitudinal Study and English Longitudinal Study of Ageing cohorts by Mingyue Ding and Bingyu Qin in Journal of International Medical Research

Footnotes

Acknowledgments

The authors sincerely thank all participants and staff involved in the CHARLS and ELSA studies for their valuable contributions and dedication.

ORCID iD

Mingyue Ding

Ethical considerations

This study was conducted in accordance with the guidelines established by the Peking University and London Multicentre Research Ethics Committees and was approved by both institutions. Informed consent was obtained from all participants.

Author contributions

Mingyue Ding and Bingyu Qin were responsible for the study's conception and design of the study. Mingyue Ding analyzed the data. The initial manuscript was drafted by Mingyue Ding and subsequently revised by Bingyu Qin. All authors reviewed and approved the final manuscript.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Henan Provincial Natural Science Foundation (202300410458).

Declaration of conflicting interests

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Data availability statement

The CHARLS and ELSA datasets used in this study are publicly available open-access cohort databases. Researchers may access the original data through the official websites of CHARLS (http://charls.pku.edu.cn/) and ELSA (). The processed analytical dataset supporting the findings of this study is available from the corresponding author upon reasonable academic request.

Supplemental material

Supplemental material for this article is available online.

References

Huo

Tang

Liu

, et al. Association between C-reactive protein-triglyceride glucose index and stroke risk in different glycemic status: insights from the China Health and Retirement Longitudinal Study (CHARLS). Cardiovasc Diabetol 2025; 24: 142.

Fang

Lan

, et al. Association between atherogenic index of plasma and new-onset stroke in individuals with different glucose metabolism status: insights from CHARLS. Cardiovasc Diabetol 2024; 23: 215.

Huo

Liao

Zhai

, et al. Interacting and joint effects of triglyceride-glucose index (TyG) and body mass index on stroke risk and the mediating role of TyG in middle-aged and older Chinese adults: a nationwide prospective cohort study. Cardiovasc Diabetol 2024; 23: 30.

Chen

Liu

, et al. Predictive value of remnant cholesterol inflammatory index for stroke risk: evidence from the China health and retirement longitudinal study. J Adv Res 2025; 76: 543–552.

Gao

Wang

, et al. The predictive value of the hs-CRP/HDL-C ratio, an inflammation-lipid composite marker, for cardiovascular disease in middle-aged and elderly people: evidence from a large national cohort study. Lipids Health Dis 2024; 23: 66.

Wan

Wang

, et al. The comprehensive effects of high-sensitivity C-reactive protein and triglyceride glucose index on cardiometabolic multimorbidity. Front Endocrinol (Lausanne) 2025; 16: 1511319.

Chang

Lee

Kim

, et al. Influence of hemoglobin concentration on stroke recurrence and composite vascular events. Stroke 2020; 51: 1309–1312.

Sato

Nakamura

Kayaba

, et al. Hemoglobin concentration and the incidence of stroke in the general Japanese population: the Jichi Medical School cohort study. J Epidemiol 2022; 32: 125–130.

von Elm

Altman

Egger

, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med 2007; 147: 573–577.

10.

Zhang

Hong

, et al. Joint association of triglyceride glucose index (TyG) and a body shape index (ABSI) with stroke incidence: a nationwide prospective cohort study. Cardiovasc Diabetol 2025; 24: 7.

11.

Zhao

Cheng

. Dynamic patterns of depression and their association with stroke: insights from CHARLS, ELSA, and HRS. Int J Surg 2026; 112: 2341–2350.

12.

Xiong

Gao

Zhang

, et al. Investigating the impact of remnant cholesterol on new-onset stroke across diverse inflammation levels: insights from the China Health and Retirement Longitudinal Study (CHARLS). Int J Cardiol 2024; 405: 131946.

13.

Cui

Liu

, et al. Joint association of TyG index and high sensitivity C-reactive protein with cardiovascular disease: a national cohort study. Cardiovasc Diabetol 2024; 23: 156.

14.

Guay

Paquette

Blais

, et al. High hsCRP concentration is associated with acute pancreatitis in multifactorial chylomicronemia syndrome. J Clin Endocrinol Metab 2024; 109: e2255–e2260.

15.

Zheng

Liao

, et al. Frailty, high-sensitivity C-reactive protein and cardiovascular disease: a nationwide prospective cohort study. Aging Clin Exp Res 2025; 37: 58.

16.

Yang

Xue

, et al. Efficacy of dual antiplatelet therapy after ischemic stroke according to hsCRP levels and CYP2C19 genotype. Am Heart J 2025; 280: 89–97.

17.

McCabe

Walsh

Gorey

, et al. Interleukin-6, C-reactive protein, and recurrence after stroke: a time-course analysis of individual-participant data. Stroke 2024; 55: 2825–2834.

18.

Xiong

Xie

Yao

, et al. Hemoglobin-to-red blood cell distribution width ratio is negatively associated with stroke: a cross-sectional study from NHANES. Sci Rep 2024; 14: 28098.

19.

Wei

Liu

Zhang

, et al. Prognostic ability of the haemoglobin-to-red blood cell distribution width ratio in predicting in-hospital mortality: a retrospective, longitudinal, observational study among elderly patients with acute ischaemic stroke. BMJ Open 2025; 15: e097075.

20.

Zuo

Dong

Liao

, et al. Low HALP (Hemoglobin, Albumin, Lymphocyte, and platelet) score increases the risk of post-stroke cognitive impairment: a multicenter cohort study. Clin Interv Aging 2024; 19: 81–92.

21.

Le Gall

Harambat

Combe

, et al. Haemoglobin trajectories in chronic kidney disease and risk of major adverse cardiovascular events. Nephrol Dial Transplant 2024; 39: 669–682.

22.

Gatti

Ahmad

Gardner Yelton

, et al. Relative anemia and perioperative stroke in children with moyamoya. J Stroke Cerebrovasc Dis 2024; 33: 107476.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.01 MB