A comparative meta-analysis of stress shielding between short-stem and conventional femoral stem in total hip arthroplasty

Abstract

Background

Stress shielding-related complications remain a key clinical challenge in total hip arthroplasty (THA). While short-stem prostheses have shown potential bone-preserving benefits in preclinical and single-center studies, pooled high-quality evidence remains inconsistent. This study aimed to compare periprosthetic bone mineral density (BMD) changes between short-stem and conventional straight-stem THA, and evaluate the stress-shielding benefits of short-stem designs.

Methods

We systematically searched PubMed, Embase, Web of Science, and the Cochrane Library from database inception to 31 March 2026 for eligible randomized controlled trials (RCTs). Meta-analysis was performed using standardized mean differences with 95% confidence intervals. Two reviewers independently assessed study quality via the GRADE criteria, and a minimum clinically important difference framework was applied to interpret BMD changes.

Results

Six RCTs were included, involving 398 patients (202 hips with short-stem THA, 225 with conventional THA). Within 12 months postoperatively, short-stem THA showed significantly better BMD preservation in Gruen zone 1 at 3 (SMD = 0.22, P = 0.04), 6 (SMD = 0.51, P = 0.02) and 12 months (SMD = 0.59, P = 0.002), and in Gruen zone 7 at 6 months (SMD = 0.65, P < 0.00001). Subgroup analysis confirmed significant BMD preservation benefits with novel-design short stems (12-month Gruen zone 1: SMD = 0.94, P < 0.00001), while traditional tapered short stems showed no significant difference from conventional stems. No between-group differences were found in other Gruen zones.

Conclusion

This meta-analysis provides moderate-quality evidence that novel-design short-stem THA is associated with better short-term proximal femoral BMD preservation compared with conventional stems within 1 year postoperatively, with benefits highly specific to prosthesis design subtypes. It must be emphasized that periprosthetic BMD is a surrogate endpoint, and the observed short-term bone preservation advantage has not yet been definitively linked to reduced revision rates or decreased aseptic loosening in large-scale registry data. Stress shielding is a chronic, ongoing process, and the clinical implications of BMD differences measured at 12 months require validation through long-term follow-up studies. Clinical decision-making should be individualized, and further large-sample, long-term RCTs are needed to determine whether the observed BMD benefits translate into improved hard clinical outcomes.

Trial registrations

PROSPERO registration number: CRD420250653026.

Introduction

In total hip arthroplasty (THA), biomechanical optimization of femoral stem design has always aimed to reduce stress–shielding–related complications [1–3]. Traditional straight (long) stems transfer load along their full length in the femur, often leading to proximal stress shielding from the stiffness mismatch; this results in decreased femoral BMD around the implant and increases the risk of aseptic loosening and early revision [4, 5]. Short-stem hip arthroplasty (SHA), defined as using femoral stems with a center-of-head-to-tip (CT) length ≤ 120 mm, represents an approximate 30–40% reduction in length compared with conventional straight stems, which typically measure approximately 140–170 mm [6]. In addition, according to the design principle, SHA can be divided into four categories: pure femoral neck support type, metaphyseal load type, metaphyseal load type with lateral expansion, and short tapered traditional improved type. The core differences lie in the geometry of the stem (such as taper, surface texture) and the matching of material elastic modulus [7].

According to Wolff's law [8, 9],bone adapts to mechanical loads by remodeling trabecular architecture. Dual-energy X-ray absorptiometry (DEXA) is the gold standard for quantifying stress shielding via serial BMD measurements in predefined Gruen zones [10–12].Previous studies suggest SHA reduces distal load transfer and bone ingrowth, thereby preserving proximal bone compared to straight stems. If confirmed by follow-up, this may provide better bone stock for future revisions [13]. However, clinical results have been mixed. Freitag et al. [13]reported clearer proximal load transfer with SHA than with straight stems, and Bieger et al. [14] via in vitro analysis found short stems yield more physiological strain distribution than straight stems. In contrast, a 5-year RCT by Hirao et al. [15]found no difference in stress-shielding(SS) bone loss between short and straight stems. Additionally, concentrating load proximally could increase bone loss at the greater trochanter.For example, Zweymüller stems have shown 32.8% BMD loss at Gruen zone 7 at 1 year, versus minimal loss at distal zones [16]. These discrepancies highlight ongoing debate on whether SHA truly outperforms standard THA in mitigating stress shielding. Existing evidence has not yet fully resolved this clinical controversy.Therefore, we conducted this meta-analysis based on RCTs to investigate the periprosthetic stress shielding benefits of SHA compared with standard THA by evaluating changes in periprosthetic BMD, aiming to guide clinical practice and prosthesis design.

Materials and methods

This study was performed in line with the recommendations of the Cochrane Handbook, and reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. This meta-analysis was registered in PROSPERO (Registration No. CRD420250653026).

Search strategy

We Conducted a systematic literature search in PubMed, Embase, Web of Science, and the Cochrane Library to identify relevant randomized controlled trials (RCTs) from database inception to 31 March 2026. The search was not restricted by language or publication status,and the Cochrane Highly Sensitive Search Strategy for randomized controlled trials (RCT filter) was applied to ensure only randomized controlled trials were included.

We screened the database using the following terms: ("THA"[Title/Abstract] OR "Total Hip Arthroplasty"[Title/Abstract] OR "Arthroplasty, Replacement, Hip"[Mesh]) AND.

("short stem"[Title/Abstract]OR"conventional stem"[Title/Abstract] OR "standard stem"[Title/Abstract]) AND.

("Stress Shielding"[Title/Abstract] OR "BMD"[Title/Abstract] OR "Bone Density"[Mesh]).

Search strategy.

#1Total hip arthroplasty.

#2 short stem or conventional stem Short handle.

#3 Stress shielding or Bone Density.

#4#1 and #2 and #3.

Inclusion and exclusion criteria

All RCTs comparing short-stem versus standard-stem prostheses in total hip arthroplasty (THA). Only articles with level of evidence I or II according to the Oxford center for Evidence-Based Medicine criteria were included. For inclusion, relevant studies were carefully selected according to the following criteria—Participants: only studies of patients assigned to primary cementless THA were included, without age, sex and ethnicity restrictions; Intervention and Comparative Intervention: THA with SHA versus standard THA clearly documented, with no restrictions on the type of implant details; Outcome: provided sufficient data to calculate postoperative periprosthetic BMD changes; Study Design: only randomised controlled trials were included.

Studies that did not meet the above inclusion criteria were excluded. Studies without extractable BMD data were excluded. Animal studies, cadaver studies, and biomechanical studies were excluded. Letters, editorials, case series, case reports, and other revision studies were excluded. Articles involving hip resurfacing or hemiarthroplasty were also excluded. Only articles reporting quantitative data were included.

Data extraction

Two investigators independently extracted general characteristics and results. If there was a disagreement between the two investigators that could not be resolved through discussion, the study was judged by another investigator.

The following information was extracted from each study and entered into a standardized Excel file: lead author, year of publication, inclusion period, number and characteristics of participants, interventions, and follow-up. Mean, standard deviation (SD), and sample size of BMD change were extracted from the original studies. Data presented only in the form of tables and graphs were extracted as numerical values whenever possible, but were only included if the two reviewers agreed. Unpublished data were obtained by contacting the original researchers, and if this failed, calculations were performed using available data.

Quality assessment

We assessed risk of bias using the Cochrane risk of bias tool, which included selection bias, performance bias, detection bias, attrition bias, reporting bias, and other bias. Risk of bias summaries and figures were obtained from Review Manager 5.4 [Cochrane]. The strength of evidence for each primary outcome was assessed according to the GRADE criteria.

Definition of short-stem prostheses

For this meta-analysis, a "short-stem femoral prosthesis" was defined as a cementless femoral stem with a center-of-head-to-tip (CT) length of < 120 mm [49]—representing an approximately 30–40% length reduction compared with conventional straight stems (typically 150–180 mm in length [6].To account for clinically relevant design heterogeneity, we a priori categorized the included short stems into two subgroups based on the international classification system proposed by Khanuja et al. [7], original RCT reports, and manufacturer official technical manuals:

• Traditional tapered short stems: Characterized by a conical taper design and distal fixation with contact to the femoral isthmus, without specialized metaphyseal expansion structure;

• Novel-design short stems: Including metaphyseal loading type, neck-supporting type, and ultra-short type, all designed for dominant proximal load transfer and minimal medullary invasion.

The detailed verification process for each included prosthesis is as follows:

Freitag et al. 2015 [13]reported the Fitmore short stem (Zimmer). Described as a trochanter-sparing short stem in the original study, its compliance with the short-stem CT length definition was confirmed via Zimmer’s official surgical technical manual, and it was classified as a novel-design short stem (metaphyseal loading type).Hirao et al. 2021 [15]reported the Taperloc Complete Microplasty short stem (Zimmer). Clearly defined as a short stem with a 35 mm shorter length than the standard stem in the original study, its CT length < 120 mm was verified via manufacturer specifications, and it was classified as a traditional tapered short stem.Sluimer et al. 2006 [17]reported the Omnifit-HA 1090 short stem (Stryker). Described as a modified proximal fixation short stem in the original study, its CT length of 108 mm was confirmed via Stryker’s official product catalog, and it was classified as a traditional tapered short stem.Lacko et al. 2021 [18] reported the Metha short stem (Aesculap/B.Braun). Clearly defined as a short-stem design in the original study, its CT length of approximately 80 mm was verified via manufacturer specifications, and it was classified as a novel-design short stem (neck-supporting type).Salemyr et al. 2015 [19] reported the Proxima ultra-short stem (DePuy Synthes). Clearly defined as an ultra-short stem in the original study, its CT length < 80 mm was confirmed via DePuy’s official specifications, and it was classified as a novel-design short stem (metaphyseal loading type).Samy et al. 2018 [20] reported the Mini-Hip short stem (Corin). Described as a short stem in the original study, its CT length < 115 mm was verified via manufacturer specifications and the Khanuja classification system, and it was classified as a novel-design short stem (ultra-short type).The above verification process ensured that all included short stems strictly met the predefined CT length < 120 mm definition, with consistent classification across all subgroup analyses.Detailed prosthesis information and Khanuja classification are presented in Table 1.

Table 1.

Characteristics of included short-stem prostheses

Study ID	Short-stem prosthesis	Manufacturer	CT length (mm)	Khanuja classification	Design subtype in this meta-analysis	Conventional stem used as control
Freitag et al. 2015	Fitmore	Zimmer	< 120	Type 2 (metaphyseal loading)	Novel-design short stem	Spotorno CLS
Hirao et al. 2021	Taperloc Complete Microplasty	Zimmer	< 120	Type 4 (traditional tapered)	Traditional tapered short stem	Taperloc Complete
Sluimer et al. 2006	Omnifit-HA 1090	Stryker	108	Type 4 (traditional tapered)	Traditional tapered short stem	Omnifit-HA
Lacko et al. 2021	Metha	Aesculap/B.Braun	80	Type 1 (neck-supporting)	Novel-design short stem	Conventional straight stem
Salemyr et al. 2015	Proxima	DePuy Synthes	< 80	Type 3 (metaphyseal loading with lateral flare)	Novel-design short stem	Corail
Samy et al. 2018	Mini-Hip	Corin	< 115	Type 2 (metaphyseal loading/ultra-short)	Novel-design short stem	Spotorno CLS

CT length = center-to-tip length. Short stem was defined as CT length < 120 mm

Khanuja classification:

Type 1: Pure femoral neck-supporting

Type 2: Metaphyseal loading

Type 3: Metaphyseal loading with lateral expansion

Type 4: Traditional tapered short stem

Novel-design short stems included Types 1, 2, and 3

Traditional tapered short stems corresponded to Type 4

Statistical analysis

Continuous outcomes (postoperative periprosthetic bone mineral density [BMD] changes) were pooled as Standardized Mean Difference (SMD) with corresponding 95% Confidence Interval (95% CI), calculated using Hedges' g formula. Statistical heterogeneity was quantified using the I² statistic: an I² value > 50% was considered to indicate substantial heterogeneity, and a random-effects model was applied for pooled analysis; otherwise, a fixed-effects model was used. Subgroup analyses were performed based on short-stem design subtypes, to explore the sources of heterogeneity and investigate the impact of implant design on periprosthetic BMD changes.

Sensitivity analyses were conducted to assess the stability of the pooled results and identify potential sources of heterogeneity. Given the small number of included trials (n = 6) and repeated measurements across Gruen zones and follow-up time points, two additional sensitivity analyses were performed: (1) leave-one-out sensitivity analysis, by omitting one included study at a time to recalculate the pooled effect size; (2) model selection sensitivity analysis, where the fixed-effects model and random-effects model were forcibly applied respectively to recalculate the SMD and 95% CI for the primary outcomes (BMD changes in Gruen zone 1 at 3, 6, and 12 months postoperatively, and BMD changes in Gruen zone 7 at 6 months postoperatively), to verify the impact of model selection strategy on the core conclusions of this study. The Minimum Clinically Important Difference (MCID) normalization framework was applied to interpret the clinical significance of BMD changes in the results [21]. Referring to similar studies on surrogate endpoints in the orthopedic field, the mean and standard deviation of BMD changes were converted into "MCID units", where an MCID unit ≥ 1 indicated a clinically important difference, and an MCID unit < 1 indicated limited clinical significance. Consistent with the latest dual-energy X-ray absorptiometry (DEXA) study by Werneburg F et al. [22, 23], a BMD change of ≥ 3% was set as the clinically important threshold for this surrogate endpoint. Egger’s test was performed using Stata version 17.0 (StataCorp LLC, College Station, TX, USA) to detect the presence of publication bias. A two-sided P-value < 0.05 was considered statistically significant. All data pooling and forest plot generation were performed using Review Manager 5.4 (Cochrane Collaboration) and Stata 17.0.

Results

Search results

We identified 342 potentially relevant studies through electronic searches; of these, 160 articles were removed due to duplicate reporting, and 136 articles were excluded because the title and abstract were irrelevant or only met the abstract requirements. Of the remaining 24 articles, the full texts were retrieved and screened, of which 17 were excluded due to lack of relevant data and 1 was excluded because it mainly studied the content related to short-stem anteversion. Finally, 6 randomized controlled trials met our selection criteria. Figure 1 shows the results of the selection process for the literature search.

Fig. 1.

Preferred Reporting items for Systematic Reviews and Meta-Analyses flowchart for the inclusion of the eligible studies

Study characteristics

Table 2 summarizes the characteristics of the six included RCTs. These studies were published between 2006 and 2022, with four from Europe and two from Asia. The individual sample sizes of these studies ranged from 29 to 144 patients. These studies included a total of 202 patients who underwent SHA and 225 patients who underwent THA. For SHA, four studies used a traditional tapered short-stem design, while three studies used Metha, Proxima, and Mini-Hip special-design short-stems, respectively. In all studies, BMD (g/cm²) data or BMD change rate (%) were obtained using DEXA, and data were collected at 3, 6, and 12 months after surgery.

Table 2.

Characteristics of the included studies

Year	Prosthesis Stem philosophy	Stem design	Publication	Study design	Demographics Number of hips SS/CS	Age (mean) years	BMI	Gender (n) M:F	Follow Up (mean),years	Pathologytreated	Outcomes Clinical 0utcome measures	Radiographid outcome Measure
2006	HFSt	Omnifit-HA 1017 (regular)/Omnifit-HA 1090 (short)	Slumier et al	RCT	40/40	56/53	/	17 M:23F/15 M:25F	2	OA、AVN	HHS、MPS、PAL	BMD、DEXA、ES、RBA、RL
2015	TS	Fitmore short stem/CLS straight stem	Freitag et al	RCT	57/81	56.8/59.1	29.7/28.3	21 M:36F/31 M:52F	1	ESPOA	HHS,WOMAC	BMD、RS
2015	TS	Ultra-short stem (Proxima)/Conventional tapered stem (Bi-metric)	Salemyr et al	RCT	26/25	62/62	27/28	11 M:15F/11 M:14F	2	POA	HHS, WOMAC, EQ-5D	BMD、RSM
2018	TS	Cementless standard femoral stems (Spotorno CLS Hip Stem/Zimmer Inc.)/Short cementless stems (Mini-Hip, Corin)	Samy et al	RCT	25/25	48.7/45.4	27/28	14 M:11F/13 M:12F	1.78	OA、AVN、Post-traumatic、DDH	HHS、VAS	BMD、DEXA
2020	TS	Taperloc Complete Microplasty/Taperloc Complete	Hirao et al	RCT	29/29	58.3/58.3	23.4	4 M: 25F	5	OA、ON	JOA、Thigh pain	BMD、RPL、stem subsidence
2021	TS	Metha short stem (Ti6Al4V)/Bicontact straight stem (Ti6Al4V)	Marek et al	RCT	25/25	52.9/55.8	27.1/28.5	29 M/21F	1	OA	HHS	BMD、OI、SS

This table summarizes the key characteristics of the 6 RCTs included in this meta-analysis, including publication year, prosthesis design, sample size, patient demographics, follow-up duration, primary pathology, and outcome measures. Abbreviations: SS = short stem, CS = conventional stem, OA = osteoarthritis, AVN = avascular necrosis, HHS = Harris Hip Score, BMD = bone mineral density, DEXA = dual-energy X-ray absorptiometry

One week after surgery was used as the baseline value. Comparison of BMD changes was performed in Gruen zones 1 to 7 in 5 studies and in Gruen zones 1 and 7 in 1 study. The mean patient age in the 6 studies was 55.6 years (range 45.4–62. 8) and the mean follow-up period was 2.3 years (range 1.0–5.0). Only 1 study had a follow-up period of more than 5 years, and the rest had a maximum follow-up period of 1 year. To reduce heterogeneity, BMD in this study was extracted at 3, 6, and 12 months.

Bias analysis

Six studies were assessed as having low to moderate risk of bias, and specific bias risk information for each study is shown in Fig. 2. Most of the six studies used appropriate randomization and allocation concealment methods, with only Freitag2015 [13] and Hirao2020 [15]lacking information on allocation concealment. For blinding, only Freitag2015 [13], Samy2018 [20], and Hirao2020 [15] had unclear blinding, and the remaining studies were blinded to the treatment regimen. In terms of blinding of outcome assessment, most studies were blinded, with only Freitag2015 [13] not reporting this. All studies showed a low risk of bias in terms of incomplete results, selective outcome reporting, and other biases, but Freitag2015 [13] had a high risk of bias. In addition, Hirao2020 [15] reported patient attrition, but the attrition rate was less than 10%. Figure 2 shows the quality of evidence and strength of recommendation according to the GRADE system.

Fig. 2.

Risk of bias assessment of each included study. A Risk of bias summary and B risk of bias graph

Meta-analysis results

Changes in BMD in Gruen zones 1

A total of 6 studies were included in the analysis of Gruen zone 1. Drawing on the MCID normalization framework (Ramadanov N, 2025), BMD changes were converted into MCID units for clinical significance interpretation, using the formula: Value_MCID = Observed Mean/MCID. At 3 months postoperatively,data from 5 studies showed that the short-stem hip arthroplasty (SHA) group had a statistically significant advantage in bone mineral density (BMD) preservation compared with the conventional total hip arthroplasty (THA) group (SMD = 0.22, 95% CI 0.01–0.44, P = 0.04; I² = 14%), with a corresponding MCID unit of approximately 0.73,which did not reach the predefined clinically important threshold in this study [13, 15, 17–19].At 6 months postoperatively,data from 4 studiesshowed that the SHA group had a significant BMD preservation advantage(SMD = 0.51, 95% CI 0.10–0.93, P = 0.02; I² = 56%), with a corresponding MCID unit of approximately 1.70,which reached the predefined clinically important threshold in this study [17–20].At 12 months postoperatively, all 6 studies confirmed the significant BMD preservation advantage of the SHA group (SMD = 0.59, 95% CI 0.21–0.98, P = 0.002; I² = 66%), with a corresponding MCID unit of approximately 1.97, which reached the predefined clinically important threshold in this study(Fig. 3) [13, 15, 17–20].

Fig. 3.

Forest plots comparing periprosthetic bone mineral density (BMD) changes in Gruen zone 1 between short-stem (SHA) and traditional hip arthroplasty (THA) at postoperative intervals: A 3-month follow-up (fixed-effects model), B 6-month follow-up (random-effects model), and C 12-month follow-up (random-effects model). Pooled standardized mean differences (SMD) reflect temporal progression of BMD preservation, with model selection guided by quantified heterogeneity (I.² thresholds)

Subgroup analysis stratified by short-stem design subtypes revealed divergent patterns of periprosthetic BMD preservation between implant categories, with no statistically significant difference in BMD changes between the SHA and THA groups at 6 and 12 months postoperatively in the traditional tapered short stem subgroup [13, 17](6 months: SMD = 0.26, 95% CI, −0.18–0.70; P = 0.24; 12 months: SMD = 0.23, 95% CI, −0.04–0.50; P = 0.10), where the corresponding MCID units for both time points were < 1 and did not reach the predefined clinically important threshold in this study; in contrast, the novel-design short stem subgroup [18–20]showed significantly better BMD preservation in the SHA group compared with the THA group at both 6 and 12 months postoperatively (6 months: SMD = 0.62, 95% CI 0.06–1.18, P = 0.03; 12 months: SMD = 0.94, 95% CI 0.59–1.29, P < 0.00001), with corresponding MCID units of 2.07 and 3.13 respectively, both of which reached or exceeded the predefined clinically important threshold in this study(Fig. 4).

Fig. 4.

Forest plots of subgroup analyzes comparing periprosthetic bone mineral density (BMD) changes in Gruen zone 1 between shortened tapered conventional stem short hip arthroplasty (SHA) and other short-stem designs versus standard total hip arthroplasty (THA): A 6-month follow-up, B 12-month follow-up. Model selection followed heterogeneity thresholds (I.² criteria)

Changes in BMD in Gruen zones 2–6

There were no statistically significant differences in BMD changes between the SHA group and the THA group in Gruen zones 2–6 at 3, 6, and 12 months postoperatively (all P > 0.05), with corresponding MCID units < 1 for all time points, which did not reach the predefined clinically important threshold in this study. The level of heterogeneity varied across different time points(Figs. 5 and 6) [13, 15, 17–20].

Fig. 5.

Forest plots comparing periprosthetic bone mineral density (BMD) changes between short-stem hip arthroplasty (SHA) and conventional total hip arthroplasty (THA) in Gruen zones 2–4 at A 3, B 6, and C 12 months. Across all intervals, pooled standardized mean differences (SMD) were close to zero with no statistically significant between-group differences (all P > 0.05).Model selection followed heterogeneity thresholds (I.² criteria)

Fig. 6.

Forest plots comparing periprosthetic BMD changes between SHA and THA in Gruen zones 5–6 at A 3, B 6, and C 12 months. No significant differences were detected at any follow-up (all P > 0.05). Model selection followed heterogeneity thresholds (I² criteria)

Changes in BMD in Gruen zones 7

At 3 months postoperatively, there was no statistically significant difference in BMD changes between the SHA group and the THA group (SMD = 0.18, 95% CI −0.03–0.39, P = 0.10; I² = 38%), with a corresponding MCID unit of approximately 0.60, which did not reach the predefined clinically important threshold in this study [13, 15, 17–19]. At 6 months postoperatively, the SHA group showed a significant BMD preservation advantage (SMD = 0.65, 95% CI 0.38–0.92, P < 0.00001; I² = 0%), with a corresponding MCID unit of approximately 2.17, which reached.

the predefined clinically important threshold in this study [17–20]. At 12 months postoperatively, there was no statistically significant difference between the two groups (SMD = 0.43, 95% CI −0.09–0.94, P = 0.10; I² = 82%), with a corresponding MCID unit of approximately 1.43, which did not reach the predefined clinically important threshold in this study(Fig. 7) [13, 17–20].

Fig. 7.

Forest plots comparing periprosthetic bone mineral density (BMD) changes in Gruen zone 7 (greater trochanter region) between short-stem (SHA) and traditional hip arthroplasty (THA): A 3-month follow-up (fixed-effects model), B 6-month follow-up (fixed-effects model), C 12-month follow-up (random-effects model). Model selection followed heterogeneity thresholds (I.² criteria)

Subgroup analysis showed no significant difference in BMD changes in Gruen zone 7 at 12 months between the traditional tapered short stem subgroup [13, 17] and the conventional THA group (SMD = −0.03, 95% CI −0.70–0.64, P = 0.93), while the novel-design short stem subgroup [18–20] had significantly less BMD loss (SMD = 0.85, 95% CI 0.51–1.19, P < 0.00001) (Fig. 8).

Fig. 8.

This forest plot shows subgroup analysis of 12-month BMD changes in Gruen zone 7, stratified by short-stem design subtypes (traditional tapered vs. novel-design) versus conventional THA. Pooled SMD with 95% CI are presented, with a formal test for subgroup differences performed

Publication bias and sensitivity analysis

Funnel plots and Egger’s linear regression tests were applied to assess potential publication bias across all included studies, with results presented in Fig. 9.At 3 months postoperatively, the P-values of Egger’s tests for all Gruen zones 1 to 7 were greater than 0.05, indicating no significant publication bias in these regions.At 6 months postoperatively, the P-values of Egger’s tests remained > 0.05 for Gruen zones 1 to 5 and 7, while potential publication bias was detected in Gruen zone 6 (P = 0.0102). At 12 months postoperatively, the P-values of Egger’s tests were all > 0.05 for Gruen zones 2 to 5, whereas potential publication bias was observed in Gruen zones 1, 6 and 7, with specific P-values of 0.0342 for zone 1, 0.0040 for zone 6, and < 0.0001 for zone 7.

Fig. 9.

Funnel plots for publication bias assessment across Gruen zones at postoperative intervals. A 12-month follow-up (Gruen zone 1), B 12-month follow-up (Gruen zone 6), C 12-month follow-up (Gruen zone 7), D 6-month follow-up (Gruen zone 6).Plot asymmetry indicates potential publication bias

Leave-one-out sensitivity analysis was performed to evaluate the stability of the pooled results. The analysis showed that no single included study altered the final pooled results for Gruen zones 1 to 7 at 3 months postoperatively, confirming the stability and reliability of the meta-analysis results for these early follow-up outcomes. However, significant changes in the overall effect size were observed under specific conditions: omitting the study by Sluimer et al. [17] altered the 12-month result in Gruen zone 1; excluding the study by Lacko et al. [18] led to a significant change in the overall effect size of Gruen zones 1 and 6 at 12 months postoperatively; removal of the study by Freitag et al. [13] significantly changed the 12-month result in Gruen zone 7; and omitting the study by Salemyr et al. [19] altered the overall effect size of Gruen zone 1 at 6 months postoperatively. The results of the sensitivity analysis for these regions are presented in Fig. 10.

Fig. 10.

Annotation: forest plot of sensitivity analyses, (A) 12-month follow-up (Gruen zone 7), B (B) 12-month follow-up (Gruen zone 1), C 6- month follow-up (Gruen zone 1), D 12- month follow-up (Gruen zone 6)

To further verify the impact of model selection on the robustness of the study results, additional model selection sensitivity analysis was performed for the primary outcomes, including BMD changes in Gruen zone 1 at 3, 6 and 12 months postoperatively, and BMD changes in Gruen zone 7 at 6 months postoperatively. The pooled standardized mean difference (SMD) and corresponding 95% confidence interval (CI) were recalculated by forcing the application of a fixed-effects model and random-effects model respectively, with the following results: for Gruen zone 1 at 3 months, the fixed-effects model yielded an SMD of 0.22 (95% CI 0.01–0.44, P = 0.04), while the random-effects model yielded an SMD of 0.22 (95% CI −0.02–0.46, P = 0.32); for Gruen zone 1 at 6 months, the fixed-effects model yielded an SMD of 0.48 (95% CI 0.21–0.75, P = 0.0005), while the random-effects model yielded an SMD of 0.51 (95% CI 0.10–0.93, P = 0.02); for Gruen zone 1 at 12 months, the fixed-effects model yielded an SMD of 0.49 (95% CI 0.28–0.71, P < 0.00001), while the random-effects model yielded an SMD of 0.59 (95% CI 0.21–0.98, P = 0.002); for Gruen zone 7 at 6 months, both the fixed-effects and random-effects models yielded an SMD of 0.65 (95% CI 0.38–0.92, P < 0.00001). With the exception of a slight change in the P-value under the random-effects model for Gruen zone 1 at 3 months, all other key outcomes maintained consistent statistical significance and direction of effect under both models, indicating that the I² threshold-based model selection strategy had minimal impact on the core conclusions of this study.

Discussion

Bone mass preservation and stress shielding regulation mediated by design heterogeneity of short-stem prostheses

The results based on short-term follow-up data within 12 months postoperatively from the cohort included in this study showed that the average BMD loss rate of conventional stems in Gruen zones 1 and 7 was 17.9% in this pooled analysis, with the Spotorno CLS Hip Stem reported by Samy et al. [20] reaching a BMD loss rate as high as 30.9% in Gruen zone 1 in its single-center study cohort. In contrast, the average BMD loss rate of short stems in Gruen zones 1 and 7 was 10.5% at 12 months postoperatively in this study; specifically, the Metha short stem and Proxima ultra-short stem reported by Marek et al.and Salemyr et al. [18, 19] had BMD loss rates as low as 2.9% and 4% in Gruen zone 1, respectively, corresponding to MCID units of approximately 0.97 to 1.33 (all ≥ 1 MCID unit, indicating potential short-term clinical importance of this bone-preserving effect only based on the predefined MCID threshold in this study). Conversely, the traditional tapered short stem (e.g., Omnifit-HA 1090 reported by Sluimer et al. [17]) still had BMD loss rates of 14.3% to 14.4% in Gruen zones 1 and 7 in this pooled analysis, corresponding to MCID units of approximately 4.77 to 4.80, with no significant difference from conventional stems within the 12-month follow-up period of this study.

Because the heterogeneity of the results in Greun areas 1, 2, 3, 6, and 7 at 12 months and Gruen areas 1 and 6 at 6 months was greater than 50%, we performed a subgroup analysis of the short stems.Subgroup analysis based on the included studies further confirmed that, compared with conventional stems, the traditional tapered short stem subgroup had MCID units < 1 for Gruen zone 1 (6 months: SMD = 0.26, P = 0.24; 12 months: SMD = 0.23, P = 0.10) and Gruen zone 7 (12 months: SMD = −0.03, P = 0.93), which did not reach the predefined clinically important threshold in this study; in contrast, the novel-design short stem subgroup had MCID units ≥ 2 for Gruen zone 1 (6 months: SMD = 0.62, P = 0.03; 12 months: SMD = 0.94, P < 0.00001) and Gruen zone 7 (12 months: SMD = 0.85, P < 0.00001), which reached the predefined clinically important threshold in this study within the limited follow-up period.Heterogeneity was substantial in Gruen zone 1 at 6 months (I² = 56%) and 12 months (I² = 66%). In addition to stem design (already explored via subgroup analysis), potential sources of heterogeneity include surgical approach (e.g., direct anterior approach [DAA] versus posterior approach) and patient activity levels. The DAA may preserve more soft tissue and allow more physiological load transfer compared with the posterior approach, potentially influencing proximal femoral BMD changes. Patient activity level, commonly assessed by the UCLA Activity Score, is another important confounder, as higher postoperative activity may accelerate bone remodeling according to Wolff’s law. Only a few included studies reported these factors in detail; future trials should stratify analyses by surgical approach and activity level to reduce heterogeneity. [24].

These results are highly consistent with preclinical biomechanical evidence and published clinical observation data. This is in line with the findings of finite element analysis (FEA) and DEXA measurements of the Metha short stem by Lerch et al. [25], where the bone preservation advantage of anatomical designs observed in short-term follow-up is supported by biomechanical simulation results. Theoretically and only in the preclinical simulation setting, these designs may have better long-term bone preservation potential, but this potential has not been confirmed by high-quality, large-sample long-term clinical follow-up studies to date. Similarly, this short-term bone preservation trend is consistent with the findings from the 9.2-year average follow-up study of 36 patients with bilateral THA by Koyano et al. [26], which found significantly higher BMD in the proximal Gruen zones on the anatomical short stem side compared with the straight stem side, and a narrower area of stress shielding-related bone atrophy, confirming that anatomical designs achieve better proximal bone preservation through physiological load transfer in its small-sample cohort. However, the clinical performance of traditional tapered short stems is similar to that of conventional stems in this study, suggesting that simply shortening the stem length is not a sufficient condition for optimizing stress distribution in the short-term postoperative period, and a comprehensive design combining surface coating and stem geometry is still required for potential long-term optimization.

It is also worth noting that the changes in BMD in Gruen zones 1 and 7 of the short-stem hip arthroplasty (SHA) in this study cohort showed significant design heterogeneity as well as notable temporal variability. At 3 months postoperatively, there was a statistically significant difference in the mean BMD change in Gruen zone 1 in the SHA group (SMD = 0.22, 95% CI: 0.01–0.44, P = 0.04), but the difference in mean BMD change in Gruen zone 7 did not reach statistical significance (SMD = 0.18, 95% CI: −0.03–0.39, P = 0.10) in this pooled analysis. At 6 months postoperatively, a significant difference was observed in the mean BMD change in Gruen zone 7 (SMD = 0.65, 95% CI: 0.38–0.92, P < 0.00001). However, at 12 months postoperatively, the difference in mean BMD change in Gruen zone 7 did not reach statistical significance again, and substantial heterogeneity was observed in Gruen zone 1 (I² = 66%) in this study. Subgroup analysis at 12 months postoperatively showed that the BMD loss rate in Gruen zone 1 was not significantly different between the traditional tapered short stem group and the conventional stem group (SMD = 0.23, P = 0.10), while the novel-design short stem group still maintained a statistically significant advantage (SMD = 0.94, 95% CI: 0.59–1.29, P < 0.00001), along with a sustained advantage in Gruen zone 7 (SMD = 0.85, 95% CI: 0.51–1.19, P < 0.00001) in this subgroup analysis.The observed temporal pattern of BMD change in Gruen zone 7 in this study can be theoretically and reasonably explained by three biomechanical phases based on published preclinical and clinical studies. In the early postoperative phase (≈3 months), the bone–implant interface of cementless stems is often not yet fully osseointegrated in most clinical cases, so functional load transfer remains unstable and any proximal-load advantage of short-stem designs may not be manifested in individual patients [27]. In the mid-term phase (≈6 months), many clinical DEXA and imaging studies report [28–30] progressive osseointegration and relative interface stability; at this point, metaphyseal-loading short stems with increased proximal contact (for example, laterally-flared designs such as Proxima [31]) can more effectively transmit load to the proximal femur, thereby reducing stress shielding and preserving BMD in preclinical simulation settings.. By the 12-month endpoint, however, heterogeneity in implant geometry and fixation strategy becomes important: a study by Zhen et al. [32] reported that some novel metaphyseal-anchoring short stems continue to demonstrate proximal bone preservation, whereas certain traditional tapered or diaphyseal-engaging designs behave more like conventional stems with respect to distal load transfer — a mixture that attenuates the pooled effect when these designs are combined in this meta-analysis.Finally, finite element analysis studies by Yan et al. [33] and Burchardet al. [34] also support the view that only stems with optimized metaphyseal contact and appropriate stiffness can consistently maintain proximal loading and reduce stress shielding in simulation models, which is consistent with our subgroup findings; however, it should be explicitly clarified that the above biomechanical simulation results are only theoretical mechanistic explanations from preclinical studies, and cannot be directly equated to long-term bone remodeling outcomes in real clinical settings.

Based on 12-month postoperative short-term follow-up data from the randomized controlled trials (RCTs) included in this meta-analysis, quantitative analysis identified significant design subtype heterogeneity in the proximal femoral bone preservation effect of short-stem prostheses within our pooled cohort. Novel non-traditional tapered short stems exhibited superior short-term proximal bone preservation as early as the immediate postoperative phase, with an approximately fourfold lower BMD loss rate in Gruen zones 1 and 7 versus conventional stems; this finding may be closely linked to their anatomical geometric design, optimized surface texture, and elastic modulus matching, as supported by preclinical biomechanical studies. Notably, traditional tapered short stems — which only featured shortened length without optimized load transfer patterns — showed no statistically significant improvement in stress shielding over the 12-month follow-up, with a BMD change pattern comparable to conventional stems. This suggests that implant size reduction alone is insufficient to optimize physiological load transfer in the proximal femur in the short term, and its long-term biomechanical and clinical effects remain unvalidated to date. These findings provide preliminary, hypothesis-generating reference for the preclinical biomechanical optimization of femoral prostheses. Theoretically, optimal stress shielding mitigation may be achieved through integrated multi-dimensional innovations, including three-dimensional anatomical matching, surface functionalized coating, and material mechanical adaptation, rather than sole reliance on stem length shortening. While this design concept may theoretically support dynamic bone remodeling balance via precise biomechanical regulation, its associated long-term clinical benefits — including reduced risks of aseptic loosening, periprosthetic fracture, and revision surgery — remain unproven, and require further validation in high-quality, large-sample, long-term clinical studies.

Association and translational value of bone mineral density changes with core clinical outcomes in patients

As a classic surrogate marker for stress shielding, bone mineral density (BMD) requires explicit clarification: dynamic changes in BMD across Gruen zones only reflect short-term postoperative bone remodeling trends, and these changes cannot be directly equated to or linearly predict patients’ long-term hard clinical endpoints [35–37].Within the predefined scope of this study, the short-term clinical significance of these BMD changes may be interpreted in a standardized manner using the MCID normalization methodological framework.This framework converts BMD change values into unified "MCID units" by dividing them by the predefined MCID threshold, which can address the clinical interpretability of heterogeneous study outcomes to a certain extent (Value_MCID = Observed Mean/MCID; Δ_MCID = (Postop − Preop)/MCID) [21]. Based on the short-term follow-up data included in this study, the average BMD loss rate of conventional stems in Gruen zones 1 and 7 reached 17.9% at 12 months postoperatively in this pooled analysis, while that of the included novel-design short stems (Metha, Proxima) was only 2.9% to 4.0% (corresponding to MCID units of approximately 0.97 to 1.33), and that of traditional tapered short stems was 14.3% to 14.4% (corresponding to MCID units of approximately 4.77 to 4.80). This design subtype-specific gradient difference is only a quantitative result of short-term BMD changes from the included studies, which can provide preliminary quantitative reference for subsequent clinical translation research, but cannot directly confirm its long-term clinical benefits in real-world clinical practice.

Association between the degree of BMD preservation and aseptic loosening as well as revision rate

Aseptic loosening is the primary cause of mid-to-long-term revision after total hip arthroplasty (THA), with its pathological basis being proximal bone loss induced by stress shielding and increased micromotion at the prosthesis-bone interface in long-term clinical follow-up. [38–40] In this study, the standardized mean difference (SMD = 0.59, P = 0.002) of novel-design short stems in Gruen zone 1 at 12 months corresponds to a relative BMD preservation advantage of approximately 7.4% in the short-term follow-up.This short-term bone preservation effect only provides a potential theoretical basis for improving the long-term proximal bone bed condition, and there is no direct clinical evidence to confirm that it can reduce the risk of mid-to-long-term aseptic loosening, prosthesis subsidence and revision surgery. Koyano et al. [26], in their 9.2-year follow-up of 36 patients with bilateral THA, did confirm the advantage of anatomical short stems in proximal bone preservation. However, these studies do not directly provide a linear mapping of "every X% BMD preservation → Y% reduction in revision risk." In other words, the correlation between significant and sustained proximal bone loss and the risk of late revision has been more consistently reported. In contrast, current evidence remains insufficient to determine whether "short-term differences of a few percentage points" can be directly converted into a significant decrease in revision rate, which requires quantification based on registry/cohort data with larger sample sizes and longer follow-up periods.

Temporal association between gradational differences in BMD and postoperative pain as well as limb function

Postoperative pain relief and limb function recovery are more dependent on early implant stability, soft tissue status, and changes in local stress distribution. Both clinical studies and the present research suggest that cases with significant proximal BMD preservation within 6–12 months are often accompanied by faster functional recovery and pain relief. [27, 41] However, most of these differences are on the order of "a few points" and are influenced by the minimum clinically important difference (MCID) of the assessment tools. [42] Therefore, directly interpreting "a few percentage points of BMD difference" as " ≥ 1-point or ≥ 5-point improvement in HHS" is not always valid, which needs to be confirmed in studies with sufficient statistical power using patient-reported outcomes (PROs) as prespecified endpoints. [43].

Long-term translation of proximal BMD reserve and postoperative fracture risk

Bone mass reserve of the proximal femur is crucial for preventing postoperative fragility fractures. [44, 45] The substantial BMD loss in Gruen zones 1 and 7 of conventional stems at 12 months postoperatively in this study suggests a potential theoretical increase in the risk of long-term fragility fractures.

Individual randomized controlled trials (RCTs) and long-term follow-up studies have shown that in series with minimal proximal BMD loss—such as the ultra-short stem (Proxima) studied by Salemyr et al. [19]—long-term proximal fracture events are extremely rare, and some cohorts have reported favorable survival rates. [46] However, due to the inherent rarity of fracture events, individual small-sample trials struggle to provide robust risk ratio estimates for differences.

In addition, although finite element analysis (FEA) by Burchard et al. [34] has demonstrated that increasing proximal bone mass or optimizing proximal contact can reduce local stress concentration, providing biomechanical support for the directional conclusion of reduced fracture risk, FEA outputs should not be regarded as direct clinical risk values.

Interpretation framework for clinical significance of BMD changes

Based on the predefined MCID threshold and normalization framework of this study, the SMD of 0.94 in Gruen zone 1 at 12 months for novel short stems corresponds to approximately 3.13 MCID units, indicating that its BMD preservation effect compared with conventional stems preliminarily reaches the clinical importance threshold within the predefined framework of this study. This short-term bone preservation advantage may theoretically generate potential benefits of reducing the risk of mid-to-long-term aseptic loosening and revision by reducing micromotion at the prosthesis-bone interface and maintaining proximal bone mass reserve, but there is no direct clinical evidence to support this potential benefit. In contrast, the MCID units of traditional tapered short stems at all time points were < 1, suggesting that although they showed statistical trends at some time points, they did not achieve clinically meaningful bone preservation effect within the predefined MCID framework of this study, which also provides a certain explanation for the controversies on the bone preservation advantages of short-stem prostheses in previous similar studies.

It should be specially noted that there is currently no unified and recognized MCID threshold for periprosthetic BMD changes after THA worldwide. In this study, a BMD change of ≥ 3% was set as the threshold for clinical relevance with reference to relevant literatures on similar surrogate endpoints in the orthopedic field, and the universality and clinical guidance value of this threshold remain to be further verified. Meanwhile, when comparing the absolute BMD difference in Gruen zone 1 at 12 months (approximately 0.05–0.08 g/cm²) in this study with the MCID of similar surrogate endpoints in the field (mostly reported > 0.10 g/cm²), the results show that this difference may not have reached the recognized clinical importance threshold in mainstream clinical practice. Therefore, the clinical translation value of the short-term BMD advantage observed in this study must be interpreted with extreme caution, and its clinical significance should not be overstated. In the future, a standardized MCID normalization interpretation system for periprosthetic BMD changes can be established based on large-sample long-term follow-up data to further quantify its clinical improvement value.Statistically significant but sub-MCID differences (e.g., 3-month Gruen zone 1) were observed in the present study. While these may indicate early biological signals of remodeling, their predictive value for long-term implant stability remains uncertain. As a surrogate endpoint, BMD changes have not been definitively linked to reduced revision rates or aseptic loosening in large-scale registry data, and further long-term studies are required to establish such associations. [47]This concern is supported by emerging long-term clinical evidence. The 10-year follow-up RCT by Axenhus et al. [48] found that despite significantly less periprosthetic BMD loss in Gruen zone 1 with the Proxima ultra-short stem at both 6 and 10 years (mean difference at 10 years: − 18.3%, 95% CI − 28.0 to − 9.0; p < 0.001), adverse events and clinical outcome scores did not differ between groups, demonstrating that a sustained BMD preservation advantage does not necessarily translate into measurable clinical benefit. Furthermore, a systematic review examining the relationship between stress shielding, BMD changes and implant outcomes concluded that no direct correlation was identified between BMD changes caused by stress shielding and implant migration, loosening, or fracture. From a registry perspective, Steinbrück et al. [6] reported comparable 5-year cumulative revision probabilities between short-stem and conventional-stem groups (2.9% vs. 3.1%) in a propensity-matched cohort of over 131,580 cementless THAs, noting that further analyses are required to rule out late mechanical failure of short stems. Aseptic loosening — the leading cause of THA revision (approximately 35% of revisions across multiple national registries) — typically emerges at a mean of approximately 118 months post-surgery, well beyond the current follow-up horizon of the included RCTs. Accordingly, the short-term BMD differences documented in this meta-analysis should be regarded as hypothesis-generating observations warranting long-term validation, rather than as evidence of clinical superiority of short-stem prostheses.

Limitations of BMD translation to clinical outcomes and clinical implications

It should be clearly clarified that the main limitation of this study is that most of the included trials have a follow-up duration ≤ 1 year (only a few studies have follow-up > 5 years), with a moderate sample size and significant heterogeneity among different implant designs. Therefore, the conclusions of this study are only applicable to the assessment of short-term bone remodeling effects within 1 year after THA in the included cohort, and cannot be extrapolated to mid-to-long-term clinical outcomes, nor can they be generalized to all types of short-stem prostheses on the market. Based on the above clear evidence boundary, we recommend that the results of this study should only be used as a reference in clinical practice, and individualized stratified decision-making should be adopted: for young patients with long life expectancy and high concerns about future revision, or patients with insufficient bone mass reserve, the short-term proximal bone preservation potential of novel proximally loaded/anatomical short stems in this study can be cautiously referenced on the premise of fully informing the patient of the limitations of existing evidence, potential benefits and unknown long-term risks; for elderly patients with low activity levels, the results of this study show no significant difference in short-term BMD changes between traditional tapered short stems and conventional long stems, so the minimally invasive advantages and short-term stress shielding risks should be individually weighed according to the patient's individual conditions, and short-stem prostheses should not be blindly recommended.

Limitations of this study

This meta-analysis has several limitations. With only 6 RCTs and a total of 398 patients, the statistical power of this study is limited. First, despite using a random-effects model, included RCTs varied in follow-up duration (primarily 1–12 months), prosthesis design, and measurement methods, which may limit generalizability. Second, some trials did not fully report randomization concealment or blinding procedures, and a few individual studies carried a high risk of bias (for example, Freitag 2015), which may reduce confidence in pooled estimates. Most studies had short follow-up (≤ 1 year), preventing assessment of long-term stress-shielding outcomes such as bone resorption, aseptic loosening, and revision rates at 5–10 years; longer observation is needed to verify the robustness of conclusions. In addition, we did not perform subgroup analyses for individual patient factors such as age, sex, body mass index (BMI), or baseline bone metabolic status, each of which can affect bone-remodeling dynamics and thus may confound interpretation. Finally, the spectrum of short-stem types included in the analysis was limited; some implants were essentially geometrically shortened versions of conventional stems and therefore behaved biomechanically similar to THA, potentially underestimating the advantages of novel short-stem designs.

In conclusion

This meta-analysis, based on short-term follow-up data (≤ 12 months) from 6 randomized controlled trials (RCTs), preliminarily investigated the differences in stress shielding-related bone mineral density (BMD) between cementless short-stem hip arthroplasty (SHA) and conventional straight-stem total hip arthroplasty (THA). The results indicate that SHA shows better short-term bone mass preservation in Gruen zones 1 and 7 compared with conventional THA, and this advantage is highly dependent on prosthesis design subtypes — non-traditional tapered novel short stems (e.g., Metha, Proxima) exhibit more significant benefits in proximal bone mass preservation, while traditional tapered short stems show no significant difference from conventional THA.Notably, stress shielding is a chronic process. BMD is a surrogate endpoint, and preserved BMD at 12 months has yet to be definitively linked to lower revision rates or reduced aseptic loosening in large-scale registry data.Therefore, in clinical practice, the proximal bone preservation potential of novel-design short stems may be referenced for young patients with longer life expectancy and concerns about future revision surgery, or patients with insufficient bone mass reserve, based on the current limited short-term evidence. For elderly patients with low activity levels, the minimally invasive advantages and short-term stress shielding risks should be individually weighed, as no significant short-term difference was observed between traditional tapered short stems and conventional THA in this study.Future studies with long-term follow-up and standardized assessment systems are needed to further clarify the impact of prosthesis design on mid-to-long-term clinical prognosis, to provide more robust evidence-based support for clinical decision-making and prosthesis optimization.

Abbreviations

BMD

Bone Mineral Density

CI

Confidence Interval

DEXA

Dual-energy X-ray Absorptiometry

GRADE

Grading of Recommendations Assessment, Development and Evaluation

HHS

Harris Hip Score

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

RCT

Randomized Controlled Trial

ROI

Region of Interest

SD

Standard Deviation

SED

Strain Energy Density

SHA

Short-stem Hip Arthroplasty

SMD

Standardized Mean Difference

SS

Stress Shielding

THA

Total Hip Arthroplasty

Authors’ contributions

Y.J. J.C. and X.W. designed the research study and extracted the data. Y.H. X.C. and H.C. extracted the data. Z.W. W.W. and Y.Z. conducted the data analysis and quality evaluation. Y.J. and X.W. determined the final results and wrote the manuscript. All the authors have read and approved the final manuscript.

Funding

Jiangxi Provincial Health Commission Science and Technology Program Project (No.202510090).

Data availability

All data generated or analysed during this study are included in this published article and its supplementary information files.

Declarations

Ethics approval and consent to participate

Not applicable for this study.

Consent for publication

Not applicable for this study.

Competing interests

The authors declare no competing interests.