Performance of the Primary Osteoporosis Screening Tool (POST) in osteoporosis screening: a cross-sectional study in China

Abstract

Objectives

This study aimed to evaluate the efficacy of the Primary Osteoporosis Screening Tool (POST) in identifying osteoporosis among Chinese adults aged 50 years and older.

Methods

This retrospective cross-sectional study consecutively included patients from the University-Town Hospital of Chongqing Medical University. Osteoporosis was defined based on the T-scores of femoral neck bone mineral density (BMD) or lumbar spine BMD by dual-energy X-ray absorptiometry examinations. The POST scores were computed according to the subject information on age, sex, and weight. In addition, the performance of the POST was assessed and compared with the Osteoporosis Self-assessment Tool for Asians (OSTA) by area under the receiver operating characteristic curve, net reclassification improvement (NRI), integrated discrimination improvement (IDI).

Results

A total of 2861 patients were included in the final analysis, of whom 78.43% were women, with a mean age of 65.67 ± 9.81 years. The POST exhibited superior performance compared to OSTA in osteoporosis screening, as evidenced by a higher area under curve, enhanced NRI/IDI metrics, and remarkable sensitivity. Subgroup analysis further highlighted the superior efficacy of POST in individuals aged 65 years and older. Additionally, sensitivity analysis confirmed that the POST maintained robust performance in identifying individuals with osteoporosis, even after the exclusion of underweight patients.

Conclusions

The POST has been shown to be an effective instrument for osteoporosis screening among Chinese adults. Nonetheless, this study has limitations including its single-center retrospective design and imbalanced gender distribution, which may affect generalizability.

Introduction

Osteoporosis, marked by diminished bone mineral density (BMD) and an elevated risk of fractures, represents a significant public health issue, particularly among older populations [1]. The occurrence of osteoporotic fractures contributes to considerable morbidity, mortality, and economic costs, highlighting the critical need for early screening and intervention [2–4]. Research indicates that osteoporosis affects over 200 million individuals globally, leading to approximately 9 million fractures each year [5]. In China, the rapid aging of the population has resulted in an increasing prevalence of osteoporosis, affecting over 30% of adults aged 50 years and older [6]. This trend aligns with results from a nationwide, multicenter dual X-ray absorptiometry (DXA) survey, which underscored the widespread prevalence of osteoporosis within the Chinese population [7]. Therefore, there is an increasing emphasis on the importance of implementing effective early screening strategies in the management of osteoporosis.

Despite the availability of validated osteoporosis screening tools such as the Osteoporosis Self-assessment Tool for Asians (OSTA), Simple Calculated Osteoporosis Risk Estimation (SCORE), Osteoporosis Risk Assessment Instrument (ORAI), and Fracture Risk Assessment Tool (FRAX) significant limitations persist in clinical practice [8, 9]. First, these tools exhibit suboptimal diagnostic accuracy in diverse populations. For instance, the OSTA demonstrates variable screening performances across ethnic groups, with particularly diminished performance in Asian cohorts [10–16]. Second, the existing osteoporosis screening models or tools indeed have room for further improvement and optimization in terms of practicality. Several osteoporosis screening tools, while effective in predicting osteoporosis risk, may have a certain level of complexity that affects the willingness of participants to cooperate. For example, the FRAX tool requires the collection of various variables, including age, gender, weight, height, history of fractures, family fracture history, smoking, and alcohol consumption habits. Collecting this information can be cumbersome for some participants, thereby reducing their willingness to participate [17]. Therefore, these limitations mentioned above underscore the urgent need for regionally adapted, simplified screening paradigms balancing accuracy and practicality.

In 2023, Tang et al. developed a novel screening tool named Primary Osteoporosis Screening Tool (POST) [18]. This innovative screening instrument was devised as a scoring scale derived from the cohort aged 50 years and older, utilizing data from the National Health and Nutrition Examination Survey (NHANES). The formulation of this tool employed logistic regression models, incorporating variables including age, sex, and weight [18]. Although FRAX is extensively utilized, it was commonly used to calculate the 10-year fracture probability rather than directly assessing densitometric osteoporosis and necessitates multiple clinical inputs that are not consistently accessible in primary care or community screening settings in China. Conversely, the POST algorithm relies solely on age, sex, and weight, offering a pragmatic balance between simplicity and predictive efficacy for identifying individuals with osteoporosis as defined by BMD. Moreover, the POST showed good performances in identifying individuals with osteoporosis in a lager sample population from the NHANES and a small sample population from a single center in China, which suggests that the POST might have great potential in osteoporosis screening, especially for Chinese populations [18]. However, on the one hand, the POST was developed based on the general United States (US) population from the NHANES [18]. On the other hand, the single-center sample size of Chinese population enrolled in the previous study was relatively small [18]. Therefore, its applicability to Chinese populations still warrants further investigation, as ethnic differences in body composition, lifestyle, and fracture risk may influence screening accuracy. Notably, previous research on the applicability of POST within Chinese populations has been constrained by small, single-center samples. This study will address this limitation by providing a significantly larger, independent evaluation of POST within a Chinese cohort, thereby assessing its applicability and scalability for real-world screening programs.

Building upon the aforementioned background, this study sought to evaluate the efficacy of the POST in identifying osteoporosis among Chinese adults aged 50 years and older. Furthermore, the study aimed to compare the effectiveness of the POST and the OSTA in estimating osteoporosis risk, with the objective of determining the most suitable assessment tool for osteoporosis screening.

Methods

Study design and population

This retrospective cross-sectional section consecutively included patients who underwent DXA examinations from the medical examination center of the University-Town Hospital of Chongqing Medical University from March 2023 to November 2024. Furthermore, the inclusion and exclusion criteria for participants were listed as follows. Inclusion criteria: (i) aged 50 and older; (ii) with data on lumbar spine BMD (LS-BMD) or femoral neck BMD (FN-BMD). Moreover, patients with missing data on weight or height were excluded from this study. Furthermore, the present study has been granted approval by the ethics review board of the University-Town Hospital of Chongqing Medical University (No. IIT-LL-2025011). In addition, a waiver of informed consent was granted for this retrospective study. Finally, data from 2867 patients aged 50 years or older with BMD measurements at the lumbar spine or femoral neck were collected. After excluding 6 participants with missing data on body weight or height, a total of 2861 patients were included in the final analysis.

Osteoporosis and BMD

All participants in this study underwent DXA scanning for the assessment of BMD, conducted by qualified medical personnel using a Hologic Horizon DXA scanner (Hologic Inc., Marlborough, MA). The anatomical sites assessed included the lumbar vertebrae (L1-4) for LS-BMD and the total hip for FN-BMD. Osteoporosis was defined as a T-score of less than or equal to − 2.5 standard deviations at either the LS-BMD or FN-BMD.

POST

The prior study elaborated on the comprehensive methodology employed to compute the POST scores [18]. In summary, the POST, articulated as a scoring scale, was formulated utilizing logistic regression models that incorporated variables such as age, sex, and weight. For example, a 65 year-old female weighing 75.3 kg was assigned a POST score of 11.

OSTA

The OSTA scores were computed based on age and weight. The calculation followed the algorithm: OSTA score = 0.2 × [weight (kg)−age (years)], with the resulting values truncated to produce an integer. For instance, for a woman aged 65 years with a weight of 75.3 kg, the OSTA score was determined to be 2, as calculated by 0.2 × (75.3−65) = 2.06, which was truncated to 2.

Statistical analysis

Baseline characteristics are reported as means ± standard errors (SD) for continuous variables and as frequencies (percentages) for categorical variables. The study conducted comparisons of baseline characteristics between individuals with and without osteoporosis utilizing chi-squared tests for categorical variables and t-tests for continuous variables. Correlations between POST scores, OSTA scores, and BMD were evaluated using Pearson's correlation coefficient. Receiver operating characteristic (ROC) curves were generated to determine the area under the curve (AUC) of the ROC. Additionally, the study assessed the efficacy of identifying patients with osteoporosis through the DeLong test, net reclassification improvement (NRI), and integrated discrimination improvement (IDI). Sensitivities and specificities of the POST (≥ 7 points) and OSTA (≤ − 1 or < 2 points) at a specified threshold were calculated, and the McNemar test was used to compare these metrics between the POST and OSTA. Furthermore, the subgroup analysis was stratified by age groups (50–64 years and 65 years and older) and by sex (male and female). Additionally, this study assessed the efficacy of the POST in osteoporosis screening by conducting a sensitivity analysis that excluded underweight patients. All data analyses and visualizations were conducted using R software (version 4.4.0; https://www.R-project.org) and GraphPad software (version 10.4.0; https://www.graphpad.com/), with statistical significance defined as p < 0.05. For subgroup analyses, P-values were adjusted for multiple comparisons using the Bonferroni method. Given four subgroup comparisons (age-stratified and sex-stratified groups), the significance threshold was set at 0.0125 (0.05/4).

Results

Baseline characteristics

In total, 2861 patients were included in the final analysis, of whom 78.43% were women, with a mean age of 65.67 ± 9.81 years. The study population had mean values for weight, height, and BMI of 58.40 ± 9.98 kg, 157.05 ± 7.61 cm, and 23.63 ± 3.33 kg/m², respectively. Detailed baseline characteristics are presented in Table 1. Furthermore, patients diagnosed with osteoporosis exhibited lower OSTA scores and higher POST scores compared to those without osteoporosis, as illustrated in Fig. 1.

Table 1.

Baseline characteristics of the study population

Characteristics	Total	Non-osteoporosis	Osteoporosis	P-value
	(N = 2861)	(N = 1301)	(N = 1560)
Age (years)	65.67 ± 9.81	63.16 ± 9.44	67.76 ± 9.62	< 0.001
Sex, n (%)				< 0.001
Men	617 (21.57%)	399 (30.67%)	218 (13.97%)
Women	2244 (78.43%)	902 (69.33%)	1342 (86.03%)
Weight (kg)	58.40 ± 9.98	62.13 ± 9.95	55.29 ± 8.88	< 0.001
Height (cm)	157.05 ± 7.61	159.31 ± 7.54	155.16 ± 7.15	< 0.001
BMI (kg/m2)	23.63 ± 3.33	24.44 ± 3.26	22.95 ± 3.24	< 0.001

BMI: body mass index

Fig. 1.

Violin plot for comparisons of the OSTA and POST scores. A OSTA; (B) POST. OSTA: Osteoporosis Self-Assessment Tool for Asians; POST: Primary Osteoporosis Screening Tool

Correlation between POST scores and BMD

The results of Pearson correlation analysis (Fig. 2) revealed that OSTA scores had a positive correlation with FN-BMD (R = 0.47, p < 0.001) and LS-BMD (R = 0.39, p < 0.001), while POST scores were negatively correlated with FN-BMD (R = − 0.45, p < 0.001) and LS-BMD (R =  − 0.45, p < 0.001).

Fig. 2.

Correlation between POST (or OSTA) scores and BMD. A OSTA score and FN-BMD; (B) POST score and FN-BMD; (C) OSTA score and LS-BMD; (D) POST score and LS-BMD. BMD: bone mineral density; FN: femoral neck; LS: lumbar spine; OSTA: Osteoporosis Self-Assessment Tool for Asians; POST: Primary Osteoporosis Screening Tool

Performance of the POST

The ROC analysis results (Fig. 3) indicated that the AUC values for POST and OSTA in identifying patients with osteoporosis were 0.737 (95% CI 0.719–0.755) and 0.718 (95% CI 0.700–0.737), respectively. Furthermore, DeLong's test (Fig. 3) revealed that POST was significantly more effective than OSTA in identifying patients with osteoporosis (p = 0.001). Additionally, both the IDI (0.0226, 95%CI 0.0147–0.0304, p < 0.001) and the NRI (0.1568, 95%CI 0.0856–0.2280, p < 0.001) metrics favored POST over OSTA, as detailed in Table 2.

Fig. 3.

ROC curves. AUC: area under curve; NPV: negative prediction value; OP: osteoporosis; OSTA: Osteoporosis Self-Assessment Tool for Asians; POST: Primary Osteoporosis Screening Tool; PPV: positive prediction value; ROC: receiver operating characteristic

Table 2.

NRI and IDI between the POST and the OSTA

Population	Index	Value	95%CI	p
Overall	IDI	0.0226	0.0147–0.0304	< 0.001
	NRI	0.1568	0.0856–0.2280	< 0.001
50–64 years	IDI	− 0.0035	– 0.0117–0.0047	0.404
	NRI	0.0752	– 0.0255–0.1758	0.143
65 and older	IDI	0.0516	0.0394–0.0637	< 0.001
	NRI	0.4049	0.3028–0.5070	< 0.001
Men	IDI	– 0.0018	– 0.0087–0.0050	0.604
	NRI	– 0.0559	– 0.2210–0.1091	0.506
Women	IDI	0.0071	0.0025–0.0117	0.002
	NRI	0.1028	0.0194–0.1861	0.016
Non-underweight patients	IDI	0.0273	0.0195–0.0350	< 0.001
	NRI	0.2156	0.1437–0.2876	< 0.001

Bold font indicates significance (Bonferroni correction for multiple comparisons in subgroup analyses, α = 0.0125 for 4 comparisons)

IDI: integrated discrimination improvement; NRI: net reclassification improvement; OSTA: Osteoporosis Self-Assessment Tool for Asians; POST: Primary Osteoporosis Screening Tool

Sensitivity and specificity

The results presented in Table 3 indicate that the sensitivities of the POST, OSTA (threshold: < 2), and OSTA (threshold: ≤ − 1) were 98.53% (95% CI 97.76–99.04%), 95.00% (95% CI 93.77–96.00%), and 70.19% (95% CI 67.84–72.44%), respectively. In contrast, the specificities for these measures were 8.53% (95% CI 7.10–10.22%), 19.83% (95% CI 17.72–22.12%), and 62.95% (95% CI 60.25–65.57%), respectively. Furthermore, the POST exhibited a significantly higher sensitivity but a notably lower specificity compared to the OSTA at both threshold levels (< 2 and ≤ − 1).

Table 3.

Comparison of the performances between POST and OSTA

	POST	OSTA (threshold: < 2)	p	OSTA (threshold: ≤ – 1)	p
Sensitivity, % (95% CI)	98.53 (97.76, 99.04)	95.00 (93.77, 96.00)	< 0.001	70.19 (67.84, 72.44)	< 0.001
Specificity, % (95% CI)	8.53 (7.10, 10.22)	19.83 (17.72, 22.12)	< 0.001	62.95 (60.25, 65.57)	< 0.001

CI confidence interval; OSTA: Osteoporosis Self-Assessment Tool for Asians; POST: Primary Osteoporosis Screening Tool

Subgroup analysis

The results of subgroup analysis and DeLong's test (Fig. 4) showed that the POST demonstrated significantly superior performance compared to the OSTA in identifying osteoporosis in individuals aged 65 years and older (p < 0.001). Conversely, no significant differences were observed between the POST and the OSTA in detecting osteoporosis among individuals under 65 years of age, as well as within male and female subgroups. Furthermore, both the IDI (0.0516, 95%CI 0.0394–0.0637, p < 0.001) and the NRI (0.4049, 95%CI 0.3028–0.5070, p < 0.001) metrics indicated a preference for the POST over the OSTA in patients aged 65 years and older. Additionally, the IDI (0.0071, 95%CI 0.0025–0.0117, p = 0.002) but not NRI demonstrated a preference for the POST over the OSTA specifically among female patients (Table 2).

Fig. 4.

Subgroup analysis. A Patients aged 50 to 64 years; (B) Patients aged 65 and older; (C) Men; (D) Women. AUC: area under curve; NPV: negative prediction value; OP: osteoporosis; OSTA: Osteoporosis Self-Assessment Tool for Asians; POST: Primary Osteoporosis Screening Tool; PPV: positive prediction value; ROC: receiver operating characteristic

Sensitivity analysis

Upon excluding individuals classified as underweight (BMI < 18.5 kg/m²), the sensitivity analysis (Fig. 5) revealed that the AUC values for POST and OSTA in the identification of osteoporosis were 0.730 (95% CI 0.711–0.748) and 0.708 (95% CI 0.689–0.727), respectively. Furthermore, DeLong's test results (Fig. 5) indicated that POST was significantly more effective than OSTA in identifying patients with osteoporosis (p < 0.001). In addition, the IDI (0.0273, 95%CI 0.0195–0.0350, p < 0.001) and NRI (0.2156, 95%CI 0.1437–0.2876, p < 0.001) also favored the POST compared with the OSTA (Table 2).

Fig. 5.

Sensitivity analysis. AUC: area under curve; NPV: negative prediction value; OP: osteoporosis; OSTA: Osteoporosis Self-Assessment Tool for Asians; POST: Primary Osteoporosis Screening Tool; PPV: positive prediction value; ROC: receiver operating characteristic

Discussion

This study comprehensively evaluated the efficacy of the POST in osteoporosis screening within a substantial sample of the Chinese population. The principal findings demonstrated that the POST showed higher AUC than OSTA in our cohort in identifying osteoporosis among Chinese adults aged 50 years and above. Furthermore, the POST presented distinct advantages over the OSTA, particularly in specific subgroups, such as older adults and female patients. However, the clinical utility of the POST requires validation in multi-center studies, particularly given the retrospective design and male underrepresentation.

This study highlighted the superior performance of the POST in comparison to the OSTA for identifying osteoporosis among Chinese adults aged 50 years and older. The elevated AUC values further substantiate POST's enhanced discriminative capacity, likely due to its inclusion of age, sex, and weight, which collectively encompass key risk factors for osteoporosis. Additionally, the findings suggest sex-specific differences in BMD trajectories related to age and body weight [19, 20]. Furthermore, the significantly higher sensitivity of POST underscores its potential to reduce false-negative results, a critical advantage in clinical settings where early detection is essential for preventing fractures and initiating timely interventions. Furthermore, the enhanced NRI and IDI metrics substantiate the clinical utility of POST. These sophisticated statistical measures not only demonstrate its superior capacity to reclassify individuals into appropriate risk categories but also indicate a more nuanced integration of predictive information. Nonetheless, these findings should be considered in light of the study's limitations. The retrospective and single-center design associated with medical examination cohorts may limit the generalizability of the results.

In this study, we adopted a threshold value of ≥ 7 points for the POST to identify patients with osteoporosis, based on the initial study that developed the POST [18]. The determination of appropriate thresholds for disease screening remains a contentious issue. Screening tests for osteoporosis generally prioritize sensitivity over specificity, as the primary objective is to identify as many individuals at risk for the disease as possible. Early identification of high-risk individuals can facilitate timely preventive and therapeutic interventions, thereby potentially reducing the incidence of fractures. However, the emphasis on high sensitivity in screening tests can result in an increased rate of false-positive outcomes, a phenomenon corroborated by multiple studies [21, 22]. a high rate of false positives inevitably leads to increased costs associated with follow-up examinations, such as DXA. Consequently, traditional statistical methods, such as the maximum Youden index, may not be the most effective approach for determining the threshold in osteoporosis screening, given that the relative importance of sensitivity and specificity can vary in this context. Previous research has suggested various methodologies to address this issue. For instance, the weighted Youden index serves as a tool to assess diagnostic tests by allowing for the adjustment of the importance of sensitivity and specificity based on user-defined weights [23]. Nonetheless, determining appropriate weight values for adjusting the significance of sensitivity and specificity remains challenging. In this study, we propose a preliminary framework for establishing the threshold in osteoporosis screening. This approach incorporates the probability of subsequent fractures and the associated costs as the weight for the false negative rate (1 - sensitivity), while utilizing the cost of DXA as the weight for the false positive rate (1 - specificity). When the sum of the two parts reaches its minimum, the corresponding threshold is considered the optimal solution. However, the specific parameter values within the preliminary framework, including the probability of subsequent fractures and the cost of DXA, may exhibit significant regional variations due to differences in socio-economic and sanitary conditions. Consequently, further cohort studies with larger sample sizes and prospective follow-up are essential to substantiate this hypothesis.

Numerous instruments have been proposed for case finding. The OSTA, which utilizes age and weight, has been extensively studied in Asian populations; however, its accuracy varies among different subgroups [24, 25]. The ORAI and the SCORE incorporate additional variables, thereby increasing complexity and data requirements [26]. While the FRAX is clinically valuable for estimating the 10-year fracture probability [27], it is not designed to diagnose densitometric osteoporosis and necessitates multiple clinical inputs that may not always be readily available in primary care settings. Given our objective of screening for BMD-defined osteoporosis in a large Chinese cohort, and considering the variables available in our database, OSTA was deemed the most suitable comparator. Future prospective studies with more comprehensive covariates could facilitate direct comparisons of the performance of the POST against ORAI, SCORE, and FRAX-based strategies in identifying individuals with a T-score of ≤  − 2.5.

The findings of this study have substantial implications for clinical practice and future scientific research. Clinically, we propose a vision for the implementation of the POST screening tool in future clinical practice. In resource-limited settings, the POST can serve as the foundation for a streamlined two-step pathway: (i) calculate the POST score at the time of patient intake, utilizing age, sex, and weight, which are typically available in most clinical encounters; (ii) refer patients for DXA scanning if their POST score meets or exceeds a predetermined threshold. Our findings, which demonstrate high overall sensitivity and particularly strong performance in adults aged 65 and older, suggest that primary care clinics and community screening programs should prioritize DXA referrals for older adults and individuals with lower body weight. In clinics with limited capacity, a tiered approach may be implemented: employ a high-sensitivity threshold to reduce the likelihood of missed cases, and, if necessary, adjust the threshold using a cost-aware framework that balances the downstream implications of false negatives (such as fracture risk and associated costs) against the expenses of additional DXA scans. Integrating an automated POST calculator into electronic health records or intake forms, providing brief training for staff, and monitoring process metrics (such as the DXA completion rate following a positive screen, the proportion of DXA-confirmed osteoporosis cases, and the initiation of treatment) can support scalable adoption and ongoing quality improvement. From a research standpoint, this study emphasizes the necessity for culturally tailored screening instruments within diverse populations. The robust correlation between POST scores and BMD underscores the significance of incorporating demographic variables into risk prediction models. Future research should prioritize multicenter prospective studies to assess the generalizability of POST across heterogeneous populations, including males and younger age groups that are underrepresented in this cohort. Furthermore, investigating the integration of POST with other clinical information, biochemical markers (such as serum bone turnover markers), or emerging technologies (such as artificial intelligence-based imaging analysis) may enhance the precision of screening accuracy [28–31].

This study is subject to several limitations. Firstly, this study was a single center retrospective study. Furthermore, the sample was drawn from a single medical center, which may introduce selection bias and restrict the generalizability of the findings to broader populations, especially those in different geographic or healthcare contexts. Therefore, multicenter prospective studies based on large sample size are required in the future. Secondly, although the overall sample size was substantial (n = 2861), the male subgroup comprised only 21.57% (n = 617) of the cohort. This demographic imbalance may lead to insufficient statistical power for drawing sex-specific conclusions. Consequently, further research is necessary to validate the performance of the POST in male participants and to ascertain potential differences in its efficacy between male and female subjects. Thirdly, after Bonferroni adjustment, the apparent advantage of POST in female participants became non-significant. This may reflect limited statistical power due to the retrospective design rather than true biological differences, underscoring the need for larger prospective studies to clarify sex-specific performance. Fourthly, while the exclusion of underweight patients in the sensitivity analysis enhances internal validity, it may diminish the tool's applicability in clinical scenarios involving such individuals. Addressing these limitations in future research will enhance the robustness and clinical relevance of the findings. Finally, beyond considerations related to study design and sample composition, the possibility of residual confounding cannot be overlooked. The clinical data collected lacked several variables that could influence BMD and screening efficacy, such as physical activity levels, calcium and vitamin D intake, smoking habits, alcohol consumption, glucocorticoid therapy, and comorbidities including thyroid, renal, and metabolic diseases. This deficiency hinders the ability to adjust for these factors. Additionally, selection bias may be a concern, given that the study population was drawn from individuals attending a medical examination center, and degenerative changes could potentially lead to artificially elevated lumbar spine BMD in some older adults. Although our sensitivity analysis, which excluded underweight participants, produced robust results, there remains a need for prospective multicenter studies with more comprehensive covariate data to accurately quantify and address these potential biases.

Conclusion

The POST has been shown to be an effective instrument for osteoporosis screening among Chinese adults. Nonetheless, this study has limitations including its single-center retrospective design and imbalanced gender distribution, which may affect generalizability.

Author contributions

Yuchen Tang: conceptualization, methodology, formal analysis, data curation, writing—original draft, writing—review & editing; Xin Hu: conceptualization, methodology, formal analysis, data curation, writing—original draft, writing—review & editing; Wen Dong: data curation, methodology, validation, investigation, writing—original draft, writing—review & editing; Jun Qin: data curation, software, validation, visualization, writing—review & editing; Jianchuan Shu: data curation, investigation, writing—review & editing; Jiaxin Kuang: data curation, writing—review & editing; Qiufu Wang: writing—review & editing; Guanyin Jiang: writing—review & editing; Miao Lei: writing—review & editing; Yongle Wu: conceptualization, data curation, methodology, writing—review & editing, supervision; Jie Hao: conceptualization, methodology, funding acquisition, writing—review & editing, supervision; Zhenming Hu: conceptualization, methodology, writing—review & editing, supervision.

Funding

This study was supported by the Special Project for the Discipline Summit Program of the First Clinical College, Chongqing Medical University (No. CYYY-XKDFJH-DSTD-202404).

Data availability

The data that support the findings of this study are not openly available due to reasons of sensitivity and are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

The present study has been granted approval by the ethics review board of the University-Town Hospital of Chongqing Medical University (No. IIT-LL-2025011). The requirement for informed consent was waived because of the retrospective nature of the study. All study protocols were done in accordance with the Declaration of Helsinki.

Consent for publication

All the authors consented for publication.

Competing interests

The authors declare no competing interests.