Introduction

Coagulation parameters are essential for assessing hemostatic function and guiding the clinical management of coagulation disorders. Their diagnostic utility extends to identifying specific hematologic abnormalities, including hemophilia A (factor VIII deficiency), hemophilia B (factor IX deficiency), von Willebrand disease (vWD), and acquired vitamin K-dependent coagulopathies1. Beyond diagnostic applications, precise interpretation of these parameters enhances resource utilization and minimizes the risks associated with unnecessary transfusions. By identifying patients at elevated risk of bleeding complications, clinicians can implement targeted preventive strategies to improve safety during surgical interventions2. From initial diagnostic evaluation to longitudinal treatment monitoring, these parameters provide a dynamic framework for clinical decision-making, facilitating personalized treatment approaches that enhance procedural safety and improve long-term patient prognoses.

Emerging evidence underscores various demographic factors, including ethnicity, sex, and aging trajectories, can influence coagulation homeostasis. For instance, East Asians are more susceptible to bleeding events, especially gastrointestinal bleeding and intracranial hemorrhage, after receiving antithrombotic therapy because of their lower thrombotic tendency3. Specifically, Chinese populations exhibiting statistically lower circulating levels of protein C (PC, a vitamin K-dependent serine protease) and protein S (PS, a critical cofactor for activated protein C) compared to Caucasian counterparts4. Sex differences in coagulation parameters have also been observed, likely influenced by hormonal variations5,6. Additionally, aging is accompanied by changes in fibrinolytic activity and an increase in procoagulant factor levels7,8. These findings collectively highlight the importance of considering demographic differences when assessing coagulation health and managing anticoagulation therapy.

The establishment of region-specific reference intervals (RIs) for coagulation assays constitutes a diagnostic imperative for precise hemostatic evaluation, yet remains constrained by pervasive operational challenges including methodological barriers, resource limitations, and technological disparities across laboratories9. Current practice continues to rely on manufacturer- or literature-based RIs, which are often derived from non-representative populations, as most vendor protocols insufficiently address demographic confounders driving interpopulation variability in coagulation parameters. Studies emphasize the need for local research to create accurate RIs that consider the unique demographic characteristics10,11. To address this need, we assessed coagulation parameters by sex and age in both southern (Shenzhen) and northern (Beijing) China to develop population-specific reference values for Chinese adults. Following dual standardization guidelines (CLSI C28-A3 and WS/T 402-2012), we analyzed six coagulation parameters: lupus anticoagulant (LA) normalized ratio, protein C (PC), protein S (PS), factor VIII (FVIII), factor IX (FIX), and von Willebrand factor antigen (vWF Ag). A sufficiently large sample size (n ≥ 120 healthy donors per group) was employed to minimize variability in the results12. By characterizing sex- and age-related changes in coagulation homeostasis, this study establishes key baselines to support tailored diagnostics and anticoagulation management.

Materials and methods

Study population and selection of samples

This study was reviewed and approved by the Ethics Committee of Shenzhen People’s Hospital (No. LL-KY-2023198-02) and the Ethics Committee of Beijing Huaxin Hospital (Nos. 2023-06-R01 and 2023-07-R01). All methods were carried out in accordance with relevant guidelines (CLSI C28-A3 and WS/T 402-2012).

Participants were recruited through advertisements posted on platform of WeChat and provided blood samples at the health examination departments of either Shenzhen People’s Hospital or Beijing Huaxin Hospital in China. They were compensated for their participation. A health questionnaire was used to assess each participant’s eligibility, ensuring they met the study’s reference population criteria. To participate, individuals had to voluntarily sign an informed consent form, be between 18 and 65 years of age, and have a body mass index (BMI) between 18.5 and 28 kg/m2. Exclusion criteria included a history of cardiovascular or cerebrovascular diseases, thrombotic disorders, or liver and kidney diseases related to hemorrhagic conditions. Cases involving the use of antithrombotic medications, direct oral anticoagulants (DOACs), heparin, or vitamin K antagonists (VKAs) were also excluded. Other exclusion criteria included smoking, alcohol abuse, pregnancy, or breastfeeding, as well as recent surgical procedures or intense physical activity.

The recruitment period lasted two weeks in December 2023, with a target of approximately 600 healthy participants. This ensures that each analysis subgroup will include more than 120 samples. Samples from individuals with hyperlipidemia, hemolysis, or jaundice were not accepted. Additionally, participants who withdrew consent, declined to complete the health questionnaire, or refused to provide additional blood samples were excluded. Duplicate samples from the same donor, as well as those that did not meet testing requirements—such as insufficient volume, contamination, or visible clotting—were also excluded from the study.

Laboratory analysis

To ensure analytical reliability and consistency across the two study sites, all laboratory procedures were performed using harmonized protocols. Both Shenzhen People’s Hospital and Beijing Huaxin Hospital utilized identical analytical platforms (Mindray, CX-9000) and matched reagent lots (Mindray, Lot No. 1723080111 for LA; 1223110311 for PS; 1123080211 for PC; 1323080311 for FVIII; 1423080111 for FIX; and 1623090211 for vWF Ag). Prior to sample testing, trueness verification was performed using in-house quality controls traceable to WHO international reference materials (NIBSC code: 13/172 for LA; 03/228 for PS; 02/342 for PC; 07/316 for FVIII and vWF Ag; 09/172 for FIX). For the five quantitative assays (PS, PC, FVIII, FIX, and vWF Ag), each analyte was tested in three independent replicates at two concentration levels (normal and pathological). All results met predefined accuracy criteria. For the qualitative LA assay, sample consistency was assessed using 10 independently prepared positive controls and 10 negative controls. Both the positive and negative consistency rates were 100%, supporting the reliability of the assay’s categorical classification.

Fresh venous blood was mixed with 0.109 M sodium citrate in a 9:1 ratio and then centrifuged at 2000 × g for 15 min to isolate the upper plasma layer. Most samples were tested immediately after collection, with analyses completed within 4 h. Samples that could not be tested on the same day were stored in a − 80 °C freezer. Frozen samples were rapidly thawed in a 37 °C water bath for 5 min and thoroughly mixed before analysis. After testing, all samples were stored at − 80 °C for a maximum of 2 months.

Following standard operating procedures, the automatic coagulation analyzer was prepared and stabilized. Reagents were loaded and calibrated according to the kit instructions, after which sample testing commenced. To ensure the reliability of test results, all levels of quality control materials (Mindray) were tested at least once every 24 h. Results were recorded for the LA normalized ratio, determined using dilute Russell’s viper venom time (dRVVT) (LA normalized ratio = dRVVT screen ratio / dRVVT confirm ratio), as well as for the activities of PC, PS, FVIII, FIX, and the level of vWF Ag.

Data visualizations and statistical analysis

All sample analyses and visualizations for this study were conducted using R software (version 4.3.0). The reference population was designed to encompass a diverse range of sex and age groups, and samples were filtered accordingly to ensure appropriate proportions for each subgroup, reflecting demographic trends. We also analyzed the differences between various parameters across the two locations to confirm that the data from the multicenter study could be effectively combined for analysis.

For continuous variables, results are presented using a combination of violin plots and box plots. The violin plots illustrate the distribution patterns of each group, with the width indicating the density of the data. Embedded within the violin plots, the box represents the interquartile range (IQR), encompassing the 25th (First quartile, Q1) to 75th (Third Quartile, Q3) centiles, while the line inside the box indicates the median. The whiskers extend to Q1 − 1.5*IQR and Q3 + 1.5*IQR, with outliers beyond the whiskers represented as individual points. These outliers were retained in the analytical dataset and visually highlighted in the violin plots, preserving the complete biological variation. For binary categorical variables, the frequency distribution of each group is displayed using 100% stacked bar charts. “N” represents the count of each group, while “%N” indicates the percentage contribution of each group to the total. “Mean” refers to the average value for each group, and “SD” denotes the standard deviation. Additionally, “Male” indicates the count of males within the group, while “%Male” represents the percentage of males in that group.

Shapiro–Wilk tests were conducted to determine whether the data followed a normal distribution. For normally distributed data, a t-test (for two groups) or ANOVA (for three or more groups) was performed, with Tukey’s HSD post hoc test and Bonferroni correction applied after ANOVA for multiple comparisons. For data exhibiting a skewed distribution, a Mann–Whitney U test (for two groups) or Kruskal–Wallis test (for three or more groups) was used, followed by pairwise Wilcoxon rank-sum tests with Bonferroni correction after the Kruskal–Wallis test. A chi-square trend test was conducted to assess the variation trends of each parameter by sex. Results were considered statistically significant at a two-tailed P value of < 0.05.

Based on age and sex differences for each parameter, kernel density estimation is used to visualize the distribution of various parameters, while the percentile method (a non-parametric method) will be applied to determine the 99th centile cut-off values or the 95% RIs. To assess the robustness of the RIs, comprehensive sensitivity analyses were performed to compare RIs with and without outlier removal (Supplementary Table 1).

Results

We recruited a total of 599 eligible healthy adults from both Shenzhen (N = 399) and Beijing (N = 200). To support subgroup analyses by sex and age, we initially characterized the overall study population. After sex stratification, the age distribution remained consistent, with participant density peaking around ages 30 and 50 and an average age of approximately 42 years (Fig. 1A). We found that females made up a larger proportion of the recruited population, accounting for 64% (Fig. 1A). This higher proportion of females was consistent across all three age groups: young adults (18–35 years), middle-aged adults (36–50 years), and older adults (51–65 years) (Fig. 1B). As a result, an excessive proportion of females may introduce bias into the experimental results.

Fig. 1
figure 1

Demographic characteristics and distribution of six anticoagulation parameters. (A) Age distribution of participants across sex groups. (B) Sex proportions of participants across three age groups (young adults, middle-aged adults, and older adults). (C) Left panel: sex proportions of participants in Sample Set I (for LA testing) by region. (D) Left panel: sex proportions in Sample Set II (for PS, PC, FVIII, FIX, and vWF Ag testing) by region. Right panel: age distribution of participants in Sample Set II by region. (E–J) Levels of six anticoagulation parameters across regions. (n.s.: no significant; *: p < 0.05; **: p < 0.01).

Full size image

Before conducting the anticoagulation parameter tests, we selectively excluded some samples to achieve a balanced sex distribution. We divided the participants into two sample sets: Sample Set I for testing LA (Fig. 1C), and Sample Set II for testing PS, PC, FVIII, FIX, and vWF Ag (Fig. 1D). In the screened sample sets, the sex imbalance was corrected, with the proportion of males exceeding 45%. We also conducted regional subgroup analysis to account for potential geographic influences. No significant sex differences were observed between the two regions, but in Sample Set II, there was a discrepancy in age distribution between Shenzhen and Beijing (Fig. 1C and D). Additionally, we analyzed the distribution of six coagulation parameters and found that only vWF Ag levels were elevated in the Shenzhen group, while the other five parameters showed no significant differences (Fig. 1E–J). The elevated vWF Ag levels observed in the Shenzhen region may be attributed to the higher proportion of elderly individuals in the population. Based on these results, we conclude that the populations from both regions meet the consistency criteria, allowing for data to be combined.

In sample sets I and II, the age distribution was consistent across sex groups, and there were no significant differences in sex proportions across age groups (Fig. 2A and B). Cross-analysis confirmed that sex and age were independent factors in this population, enabling separate subgroup analyses.

Fig. 2
figure 2

Distribution of six coagulation parameters stratified by sex and age group. (A–B) Comparison of age distribution across sex (left panel) and sex proportions across age groups (right panel) in sample sets I (A) and II (B). (C–H) Levels of six coagulation parameters stratified by sex. (I–N) Levels of six coagulation parameters stratified by age groups. (n.s.: no significant; *: p < 0.05; **: p < 0.01; ****: p < 0.0001).

Full size image

In the sex-based subgroups, we observed that the LA normalized ratio and PS activity were significantly higher in males, suggesting that sex may influence the RIs for these two parameters (Fig. 2C and D). No sex-based differences were found for the other four parameters (Fig. 2E–H). In age group comparisons, PC activity and FVIII activity were notably elevated in the older group, while vWF Ag levels were significantly lower in the younger group; the other three parameters remained unchanged (Fig. 2I–N). Additionally, we observed a progressive trend in FVIII activity and vWF Ag levels with increasing age (based on mean or median values). Based on these results, we infer that sex and age are independently associated with five coagulation parameters, while FIX activity is unaffected by either of the two demographic factors.

Based on the distribution characteristics of six coagulation parameters across sex and age, we established different grouping schemes for RIs. For the LA normalized ratio, the 99th centile is typically used as a cut-off value to determine whether a sample is positive. Although there is a notable sex difference for this parameter, the cut-off values for males (1.189) and females (1.194) are similar to that of the overall population (1.198) (Fig. 3A). The instructions from several manufacturers (Stago, Werfen, Sysmex, and Beijing Strong) shows a consistent positive cut-off value of > 1.2, with no subgrouping. Therefore, we conclude that it is unnecessary to set sex-specific RIs for the LA normalized ratio in the Chinese adults.

Fig. 3
figure 3

The cut-off value or RIs of six coagulation parameters by sex or age group. (A–B) The distribution of LA normalized ratio (A) or PS activity (B) across sex. The blue and orange shading represent females and males, respectively, while the gray shading reflects the overall distribution across all groups combined. The dashed line indicating the cut-off value or RIs. (C–F) The distribution of PC activity (C), FVIII activity (D), FIX activity (E), or VWF Ag levels (F) across age groups. The blue, orange, and red shading representing the age subgroups, while the gray shading reflects the overall distribution across all groups combined. The dashed line indicating the RIs.

Full size image

In clinical practice, assessing activity of PS is typically aimed at evaluating potential thrombotic risk. In contrast, elevated PS activities are rarely seen in clinical settings and generally do not lead to pathological conditions. Therefore, the 2.5th centile is crucial for identifying PS deficiency. In this study, the 2.5th percentile values for men and women were 72.97% and 57.68%, respectively, showing a marked difference (Fig. 3B). Previous studies have reported sex differences in PS activity among healthy adults, consistent with our findings. Manufacturer instructions (Stago, Siemens) also recommend establishing sex-specific RIs for this parameter. The RI provided by Werfen is for the overall population (63.5–149.0%) without sex-specific grouping, with a note indicating that age and hormones may affect levels in females. Based on our data, we suggest sex-specific RI for PS activity in the Chinese adult population: 73–136% for men and 58–123% for women.

Similar to PS, the primary clinical focus for PC is on identifying reduced activities, so we concentrated on the variation at the 2.5th centiles. Previous analyses indicated that the activity of PC tend to be higher in the elderly group, so we divided the population into two age groups: young to middle-aged adults (18–50 years) and older adults (51–65 years). The 2.5th percentile values for the young to middle-aged and elderly groups were 68.46% and 67.33%, respectively, close to the overall population value of 68.13% (Fig. 3C). However, the 97.5th centiles showed a 7.5% difference between the two age groups, suggesting that age may mainly impact the upper reference limit (Fig. 3C). Manufacturer instructions from Stago, Werfen, and Siemens recommend RIs of 70–130%, 70–140%, and 70–140% for PC activities in healthy adults, respectively. Stago additionally notes that PC activities are not influenced by age or sex, which slightly differs from our findings. Based on our data, we recommend setting the RI for PC activity in the overall population at 68–143%.

The deficiency and severity of individual coagulation factors can be used to screen for congenital or acquired coagulation factor deficiencies. Elevated FVIII and FIX activities are typically associated with hypercoagulable states and thrombotic disorders. Our study found that FVIII activity increases with age, prompting a subgroup analysis based on age. In this study, the 95% RIs for the young, middle-aged, and elderly groups were 54.40–151.36%, 63.62–155.98%, and 71.17–158.08%, respectively (Fig. 3D). RIs provided by different manufacturers for FVIII assays vary significantly: Stago 60–150%, Werfen 50–150%, Siemens 70–150%. Based on our findings, we recommend establishing age-specific RI for FVIII activity.

Similarly, the RI for FIX provided by different manufacturers also show considerable variation: Stago 60–150%, Werfen 65–150%, Siemens 70–120%. Our findings suggest that FIX activity is independent of age and sex, supporting the establishment of a single RI of 71–152% for the overall adult population (Fig. 3E).

A reduction in the quantity or abnormality in the quality of plasma vWF is the primary pathogenic mechanism of von Willebrand disease. Additionally, vWF serves as a sensitive marker for endothelial injury, making it useful in the prevention and treatment of thrombotic diseases. Our data revealed a positive correlation between vWF Ag levels and advancing age. After dividing participants into young, middle-aged, and elderly groups, the 95% RIs were 45–163%, 60–176%, and 69–182%, respectively (Fig. 3F). The RIs provided by Stago and Siemens are both 50–160%, accompanied by specific notes highlighting the physiological increase in values with age. Individuals with blood type O tend to have slightly lower vWF Ag levels compared to non-O blood types. Werfen’s instructions provide RIs based on blood type: 42.0–140.8% for type O individuals and 66.1–176.3% for non-O individuals. Based on our findings, we recommend establishing age-specific RI for vWF Ag in Chinese adults.

We reviewed RI data for coagulation tests from various manufacturers (Table 1) and identified several issues, including unclear target populations, small sample sizes, and undefined criteria for RIs. Additionally, we compiled the RIs and grouping recommendations for six coagulation parameters in Table 1. This study allows us to provide more precise RI data for Chinese adults, stratified by sex and age.

Table 1 Summary of cut-off value or RIs for six coagulation parameters.
Full size table

Discussion

The findings of this study underscore the critical influence of demographic factors, particularly sex and age, on coagulation parameters in the Chinese adult population. In our study population, LA levels were generally higher in men compared to women, despite no differences in the positive cut-off values. Notably, a clear association between LA and sex has been observed in various disease settings. For example, in patients with systemic lupus erythematosus, LA levels are significantly higher in women, with studies reporting an odds ratio of 1.98 (95% CI: 1.53–2.57), indicating that the prevalence of LA in female patients is nearly twice that in males13. Conversely, in patients with primary thrombotic antiphospholipid syndrome, men with isolated LA positivity are at greater risk of thrombotic recurrence14. The observed sex-specific differences in PS activity corroborate earlier reports of hormonal modulation on coagulation factors15, where estrogen in females may suppress hepatic synthesis of PS, leading to lower baseline levels. This sex disparity necessitates tailored diagnostic thresholds to avoid misclassifying healthy women as PS-deficient, which could lead to unnecessary interventions or oversight of true thrombotic risks in men. Furthermore, the age-dependent elevation in PC, FVIII and vWF Ag levels aligns with the prothrombotic shift associated with aging, likely driven by endothelial dysfunction and chronic low-grade inflammation16,17. These findings emphasize the importance of age-stratified RIs, particularly for older adults, who may otherwise be misdiagnosed with hypercoagulable states if uniform thresholds are applied.

However, this study has certain limitations. Although we rigorously analyzed healthy adults aged 18–65 years, the exclusion of pediatric, adolescent, and elderly populations (> 65 years) limits the generalizability of our reference intervals across the full lifespan. While our data suggest that age does not significantly affect the RIs for PS activity and FIX activity within the 18–65 age group, previous studies have reported markedly lower levels in infants and young children18. In addition, with the ongoing aging of China’s population, coagulation-related disorders in the elderly (> 65 years) are emerging as a significant clinical concern. Elevated levels of vWF, FVIII, and FIX were identified as common risk factors for venous thromboembolism19. The underrepresentation of these subgroups may hinder the identification of clinically meaningful variations in reference intervals, especially for parameters influenced by developmental or age-related physiological changes. Future studies should aim to include a broader and more diverse population to establish the precision and clinical utility of RIs across all age groups.

Factors such as genetic polymorphisms or dietary habits were also not explored in our study, which could further refine RIs. For instance, vWF Ag exhibited more noticeable differences after outlier exclusion, particularly in the 18–35 years age subgroup, where the lower reference limit increased from 45.4 to 53.6% (Supplementary Table 1). This suggests that retaining outliers preserved a broader range of physiological variation in younger individuals. It is well-established that individuals with blood type O typically exhibit lower vWF Ag levels compared to those with non-O blood types20. As such, the lower reference values observed in this age group may be partly attributed to the underlying distribution of ABO blood groups within the population. Future studies that incorporate both age and ABO blood group information are warranted to more accurately define vWF Ag RIs and enhance their clinical applicability across diverse populations.

Furthermore, the recruitment was restricted to two urban centers (Beijing and Shenzhen), potentially limiting the generalizability to rural or regionally diverse populations. In this study, participants from Shenzhen were marginally older than those from Beijing (mean age: 42.98 vs. 40.31 years, p < 0.05). Concurrently, the overall mean vWF Ag levels were significantly higher in Shenzhen compared to Beijing (121.9% vs. 113.5%, p < 0.05). Age-stratified analyses, however, revealed that this geographical disparity was predominantly attributable to variations in the elderly population. Notably, within the 51–65 years subgroup, Shenzhen participants demonstrated markedly elevated vWF Ag levels relative to their Beijing counterparts (131.7% vs. 119.9%, p = 0.02). In contrast, no statistically significant regional variations were observed in younger cohorts (18–35 years: 112.0% vs. 105.8%, p = 0.13; 36–50 years: 121.7% vs. 118.2%, p = 0.26).

Clinically, our findings advocate for laboratory-specific RIs in China, moving beyond the “One-Size-Fits-All” approach of manufacturer guidelines. Implementing sex- and age-adjusted thresholds could enhance diagnostic accuracy for conditions like von Willebrand disease or thrombophilia, reducing both overdiagnosis and underdiagnosis. For instance, using a uniform PS cutoff (e.g., 63.5–149.0% as per Werfen) might fail to detect deficiency in men or overestimate it in women, whereas our stratified intervals account for biological variability. Additionally, the rising prevalence of aging-related cardiovascular diseases in China underscores the urgency of adopting age-specific reference standards to optimize anticoagulant therapy and surgical safety21,22.

In conclusion, this study establishes a foundational framework for personalized RIs of coagulation parameters in Chinese adults. Sex and age information can be readily incorporated into laboratory information systems, enabling the automated delivery of individualized reference ranges. This approach allows clinicians to achieve greater accuracy in diagnosing and managing hemostatic disorders, ultimately enhancing clinical outcomes in a population with distinct coagulation characteristics.