Sage Journals: Discover world-class research

Abstract

Purpose

Posterior medial meniscus root tears often cause persistent extrusion and altered joint mechanics despite repair. This study compared standard transtibial pull-out repair with repair plus a second-tunnel centralization, hypothesizing better patient-reported outcomes without major MRI extrusion changes.

Methods

This retrospective two-center cohort (2019–2024) included adults with MRI-confirmed root tears treated with anatomic repair alone or with an additional centralization tunnel. Propensity matching yielded 96 patients (54 vs 42). The primary endpoint was 24-months change in IKDC score. Secondary outcomes included KOOS subscales, Lysholm, visual analogue scale pain, and Tegner activity. Structural outcomes—medial meniscus extrusion and Meniscal Extrusion Index—were assessed on coronal MRI at 12 ± 4 months. Analyses used ANCOVA adjusted for baseline values, supported by inverse-probability weighting and mixed-effects checks.

Results

Baseline characteristics were balanced. Extrusion changes were small and similar; residual pathologic extrusion rates were comparable (65% vs 62%). Centralization showed greater IKDC improvement (+5.7 points; p = 0.008) and higher KOOS-Quality of Life. Knees with ≥3 varus demonstrated additional benefit (interaction p = 0.048). Complications were infrequent.

Conclusions

Second-tunnel centralization significantly increased the probability of achieving clinically meaningful functional improvement, despite unchanged static MRI extrusion. This suggests dynamic load-sharing benefits not captured by static imaging. Therefore, centralization is recommended as a selective adjunct, particularly in varus alignment, rather than a routine necessity.

Keywords

medial meniscus posterior root tear centralization transtibial pull-out repair meniscal extrusion MRI patient-reported outcomes varus alignment propensity-score matching level III retrospective cohort study

Introduction

Posterior meniscal root tears (PMRTs), which most commonly involve the posterior horn root of the medial meniscus, disrupt the circumferential (“hoop”) fibers that convert axial loads into tangential stresses, thereby approximating the contact mechanics of subtotal meniscectomy.¹ In cadaveric models, a medial root tear increases the peak medial compartment contact pressure to levels comparable to those after total meniscectomy, whereas anatomic repair restores the pressure to normal levels.² Clinically, untreated or debrided (meniscectomized) PMRTs are associated with pain, rapid joint space narrowing, and high rates of radiographic osteoarthritis (OA) progression and conversion to total knee arthroplasty (TKA), in contrast to repair.^3,4 Contemporary meta-analyses and matched cohorts consistently show that repair yields superior joint preservation compared to meniscectomy or nonoperative care at mid-to long-term follow-up.^5,6

Although the lateral root can fail in the setting of ligamentous injury, the prototypical medial PMRT arises in middle age, frequently in women with degenerative cartilage changes and coronal malalignment.⁷ Risk factors repeatedly linked to medial root pathology, poorer healing biology, or worse mechanics include varus alignment and elevated body mass index (BMI).⁸ These demographic and morphological features help explain the characteristic association between PMRTs and meniscal extrusion, which is the radial displacement of the meniscus beyond the tibial margin that diminishes the meniscus’ load-sharing function and correlates with pain and OA progression.^9,10

Anatomic root fixation (most commonly a transtibial “pullout” repair or, less commonly, suture anchor fixation to the tibial footprint) is the current standard of care for a repairable PMRT in an otherwise salvageable knee.¹¹ Repair reliably improves symptoms and function and reduces downstream OA/TKA risk compared with debridement or observation.^3,12 Nevertheless, persistent or even progressive medial meniscal extrusion after technically successful root repair is common and is increasingly recognized as a plausible mechanism for ongoing cartilage overload and clinical failure in some patients.^13,14 The situation is exacerbated by varus alignment and insufficiency of the posteromedial peripheral stabilizers (meniscotibial and meniscocapsular attachments), which increase the forces transmitted across the root fixation under physiologic loading.^15,16

To address extrusion—and to unload the repaired root—surgeons have adopted “meniscal centralization,” a family of techniques designed to tether the peripheral or midbody meniscus back to the tibial rim. Centralization can be performed with knotless suture anchors along the capsulomeniscal junction or by creating a second transtibial tunnel positioned at the medial tibial plateau rim and passing a horizontal mattress suture through the midbody of the meniscus to draw it centrally.^17,18 Technical notes and laboratory studies show that centralization reduces extrusion and improves contact mechanics compared with root repair alone.^19,20 Importantly, in a recent biomechanical investigation, varus alignment and peripheral-attachment insufficiency markedly increased forces on a posterior medial root repair, while adding a transosseous centralization suture reduced those forces—mechanistic support for augmenting selected root repairs.¹⁵

Early clinical series of root repair augmented with centralization report significant improvements in pain, function, and quality of life at ∼2 years, with low short-term failure and without radiographic acceleration of arthritis; observational comparative work further suggests greater reduction of extrusion when centralization is added to repair.²¹ Still, high-quality head-to-head data remain limited, and most published clinical reports evaluate centralization as an adjunct to root fixation rather than as a stand-alone alternative. As a result, the relative merits of (i) isolated anatomic posterior root fixation versus (ii) centralization using a second tibial tunnel with a midbody meniscal suture, specifically for symptomatic PMRTs with extrusion, are not fully resolved. Biomechanics would predict that root fixation best restores hoop stress while centralization best addresses extrusion; whether centralization alone can compensate for the loss of root attachment—and under what conditions—requires careful study.^22,23

This article therefore undertakes a focused comparative analysis of these two strategies. First, we synthesize the biomechanical and clinical rationale for anatomic posterior root fixation and for centralization via a second transtibial tunnel with a midbody suture, emphasizing how each targets distinct failure modes (loss of hoop tension vs persistent extrusion). Second, we critically appraise laboratory and clinical evidence bearing on patient-reported outcomes, radiographic extrusion, cartilage status, and joint-preservation endpoints (OA progression, TKA conversion). Finally, we propose a decision framework integrating alignment, cartilage grade, and posteromedial soft-tissue integrity to guide selection between isolated root fixation and second-tunnel centralization—and to identify scenarios in which a combined construct may be preferable. Our a priori hypothesis is that, in repairable PMRTs, root fixation remains essential for hoop restoration, whereas a second-tunnel centralization suture offers incremental benefit primarily by reducing extrusion and unloading the repair in varus or peripheral-attachment insufficiency; by contrast, centralization alone is unlikely to fully replicate the biomechanical functions of anatomic root reattachment.^1,15,19

While recent studies, such as Zhou et al. (2024)²⁴, have reported outcomes of centralization techniques, these investigations largely relied on unadjusted comparative statistics (e.g., t-tests), which fail to account for baseline confounders inherent in retrospective cohorts. Factors such as body mass index (BMI), varus alignment (HKA angle), and baseline meniscal extrusion are known to independently influence postoperative outcomes and structural healing. To overcome these limitations and provide a more rigorous assessment of efficacy, the present study utilizes propensity score matching (PSM) and inverse-probability of treatment weighting (IPTW). By balancing these critical covariates, this study aims to isolate the specific therapeutic effect of the double-tunnel centralization technique with a level of precision that approximates randomized controlled trials, thereby addressing the selection bias present in previous literature^21,24.

In this study, “posterior root fixation” is defined as the anatomical reattachment of the posterior horn root to its tibial footprint, achieved through either transtibial pullout or anchor techniques. “Centralization (second tunnel)” refers to a transosseous construct involving a separate tibial tunnel near the medial plateau rim, which is utilized to tension a horizontal mattress suture passed through the midbody (meniscocapsular junction) to re-center the meniscus. This approach is distinct from anchor-based centralization, which is referenced for context but analyzed separately. By explicitly contrasting these strategies, this research seeks to elucidate the indications, anticipated benefits, and trade-offs for surgeons managing extruded posterior meniscal root tears (PMRTs)—a domain characterized by rapid technical advancements yet limited comparative evidence.

Methods

Study design and setting

We conducted a retrospective, two-center comparative cohort study across two tertiary referral hospitals. Consecutive patients who underwent arthroscopic management of posterior medial meniscus root tears (PMMRTs) between January 2019 and December 2024 were screened. Patients received either (i) anatomic posterior root fixation using a transtibial pullout tunnel to the native footprint (“posterior tunnel fixation”), or (ii) centralization with a second transtibial tunnel in addition to an anatomic root fixation, with a horizontal mattress suture passed through the midbody/peripheral meniscus to tether it to the tibial rim (“double-tunnel centralization”). Centralization was adopted in our unit for cases with preoperative extrusion and/or suspected posteromedial peripheral-attachment laxity (confirmed by intraoperative probing), reflecting evolving evidence that centralization can reduce extrusion and unload a repaired root under varus loading. Peripheral-attachment laxity was assessed intraoperatively via probing of the meniscotibial portion. We acknowledge that this variable represents a qualitative surgeon assessment rather than a quantitative measurement, and was extracted retrospectively from operative records. This retrospective cohort study was conducted across two centers—the Departments of Orthopaedics and Traumatology at Aksaray University Training and Research Hospital and Sanliurfa Mehmet Akif Inan Training and Research Hospital. The study protocol was approved by the Institutional Ethics Committee of Aksaray University (Protocol: SAGETİK175, Approval number: 2025/012, Date: November 15, 2025). All participants provided written informed consent for the use of their clinical and imaging data for research purposes, and all procedures were conducted in compliance with institutional ethical standards and the Declaration of Helsinki.

Eligibility criteria were prespecified. Inclusion criteria comprised adults aged 18–75 years with MRI-confirmed posterior medial meniscus root tear who underwent arthroscopic treatment with either posterior tunnel fixation or double-tunnel centralization, had a preoperative MRI obtained within 3 months before surgery and a postoperative MRI at 12 ± 4 months, and completed at least 24 months of clinical follow-up including patient-reported outcomes (PROs). Exclusion criteria were the presence of lateral or anterior root tears, discoid meniscus, or prior meniscal transplantation; concomitant realignment osteotomy or cartilage restoration in the medial compartment; radiographic osteoarthritis of Kellgren–Lawrence grade ≥3 or focal Outerbridge grade 4 chondral loss in the medial weight-bearing zone; coronal malalignment exceeding 5 of varus on long-leg alignment radiographs; inflammatory arthropathy; ipsilateral knee arthroplasty or ligament reconstruction within the preceding 12 months; and incomplete imaging or PRO data. These parameters were selected to minimize confounding and to isolate the effects of root fixation and centralization on meniscal extrusion and clinical function, consistent with established associations among extrusion, degenerative change, and malalignment.

All operations used standard anteromedial/anterolateral portals. For posterior tunnel fixation, the tibial root footprint was refreshed, two UHMWPE sutures (simple or cinch) were passed through the root stump, and a single transtibial tunnel was drilled to the anatomic footprint for cortical-button (Artrotek, Adana, Turkey) fixation. For double-tunnel centralization, after completing the root repair, a second transtibial tunnel was created at the medial plateau rim adjacent to the meniscocapsular junction; a horizontal mattress suture was passed through the midbody and tied over a cortical button (Artrotek, Adana, Turkey) to re-centre the meniscus (Figures 1–3). Technique details mirror published centralization descriptions (anchor-based variants exist but were not used in this cohort). Postoperatively, both groups followed the same rehabilitation: protected weightbearing for 4–6 weeks with 0–90 motion, progressive closed-chain strengthening from week 6, return to light jogging by ∼ 12 weeks, and pivoting activities after ∼ 6 months as tolerated.

Figure 1.

Schematic illustration of the surgical construct.

Figure 2.

Arthroscopic steps of posterior root footprint preparation and repair. (a) Debridement and refreshment of the posterior medial meniscus root footprint. (b) Passage of the UHMWPE sutures through the posterior root stump. (c) Final arthroscopic view after fixation. Red arrow: completed posterior root repair; green arrow: added second-tunnel centralization suture.

Figure 3.

Arthroscopic views demonstrating posterior root repair and the second-tunnel centralization technique (images from a different patient than Figure 2). (a) Completed posterior medial meniscus root repair at the anatomic footprint. (b) Arthroscopic view of the second transtibial centralization tunnel, showing passage of the horizontal-mattress suture through the meniscus midbody.

Imaging served as the structural assessment. MRIs (1.5-T or 3-T) included coronal and sagittal proton-density fat-suppressed and T2-weighted sequences (3–4-mm slices; in-plane resolution ≤0.5 mm). The primary structural variable was medial meniscus extrusion (MME, mm) on coronal MRI at the level of the medial collateral ligament (MCL), measured as the perpendicular distance from a tangent to the medial tibial cartilage edge to the outer meniscal margin; values were averaged across three contiguous slices centred on the MCL. MME >3 mm was recorded as pathologic. In addition, we computed a size-normalized Meniscal Extrusion Index (MEI) (extrusion/meniscal width on the same slice) to complement absolute MME, in line with literature supporting either a 3-mm rule or a ratio-based threshold for pathology.^9,25,26 Two musculoskeletal radiologists, blinded to treatment and time point, independently measured MME/MEI on pre- and postoperative MRIs (Figures 4 and 5). Inter- and intra-rater reliability were quantified using ICC (2. 1) with 95% CIs and interpreted per Koo & Li (poor <0.50, moderate 0.50–0.75, good 0.75–0.90, excellent >0.90). Discrepancies >0.5 mm were resolved by consensus.²⁷

Figure 4.

MRI measurement of medial meniscus extrusion. (a) Standard MME assessment on coronal imaging at the MCL level (green arrow). (b) Meniscal Extrusion Index (MEI) calculation, demonstrating measurements of meniscal width and radial extrusion.

Figure 5.

Postoperative MRI demonstrating tunnel positions. (a) Second transtibial centralization tunnel at the medial tibial rim. (b) Posterior root tunnel located at the anatomic footprint. (c) Coronal MRI showing the cortical buttons of the centralization construct (green arrow) and the posterior root repair (orange arrow).

Clinical assessment provided the functional evaluation. The primary clinical endpoint was the change in IKDC (International Knee Documentation Committee) Subjective Knee Form from baseline to 24 months. Secondary PROs included KOOS (Knee Injury and Osteoarthritis Outcome Score) subscales (Pain, ADL, QoL), Lysholm, VAS pain (0–10), and Tegner activity. We classified responders using published MCID (Minimal Clinically Important Difference) and PASS (Patient Acceptable Symptom State) thresholds for meniscal surgery: we prespecified an IKDC MCID around 10–11 points and applied contemporary KOOS MIC (Minimal important change)/PASS (patient acceptable symptom state) references specific to meniscus populations. Radiographic secondary outcomes were preoperative to postoperative changes in MME and MEI, and the presence of residual pathologic extrusion (MME >3 mm) at follow-up. Potential confounders recorded a priori included age, sex, BMI, hip–knee–ankle (HKA) angle (varus degrees), time to surgery, baseline MME/MEI, Outerbridge grade, and qualitative posteromedial peripheral-attachment quality (meniscotibial/meniscocapsular) from operative notes, as these factors influence extrusion and repair loading and underpin published indications for centralization (Figures 6 and 7).

Figure 6.

Preoperative long-leg standing radiographs. (a) Hip–knee–ankle (Hka) angle measurement with mechanical axis deviation. (B–C) Varus alignment assessment on anteroposterior and lateral views.

Figure 7.

Postoperative radiographs after posterior root repair and double-tunnel centralization. (a) Anteroposterior radiograph showing tibial tunnel positions and cortical button fixation. (b) Lateral radiograph demonstrating appropriate tunnel direction and fixation hardware.

Sample size was guided by the literature and powered for clinically meaningful differences. For the primary clinical endpoint, we targeted detection of the IKDC MCID (∼10–11 points) with SD 16–18, α = 0.05 (two-sided), and 80% power, yielding ∼ 34–42 patients per group; allowing 10–15% attrition/missingness, the planned target was ∼45–50 per group. For the primary structural endpoint, prior MRI series report MME SD ≈ 1.0–1.5 mm; detecting a 1.0-mm between-group difference requires ∼32–45 per group for 80–90% power at α = 0.05. These targets are consistent with contemporary series establishing MCID/PASS after meniscal repair and with MRI variability for extrusion.

Statistical analysis followed an analysis plan designed to identify small structural and small-to-moderate functional effects, consistent with the hypothesis that adding centralization primarily refines meniscal position/load sharing rather than dramatically changing extrusion when both groups receive anatomic root fixation. Continuous variables were assessed for normality (visual inspection, Shapiro–Wilk) and summarized as mean ± SD or median (IQR); categoricals as n (%). Between-group comparisons used independent-samples t-tests (or Mann–Whitney U) and χ² (or Fisher’s exact) tests. Within-group pre-to postoperative changes used paired t-tests (or Wilcoxon signed-rank). Because treatment allocation was nonrandom, we implemented propensity-score matching (PSM) using a variable-ratio (1:k) nearest neighbour matching algorithm (caliper 0.2 SD of the logit) on age, sex, BMI, baseline MME, HKA varus, Outerbridge grade, and time to surgery; balance was assessed using standardized mean differences (SMD <0.10 acceptable). As sensitivity analyses, we applied inverse-probability of treatment weighting (IPTW) and repeated all models.

Primary endpoints were analysed using ANCOVA with the baseline value of the outcome and matched/weighted covariates as adjustors: ΔMME (mm) for structure and ΔIKDC for function. For robustness to distributional and pairing assumptions, we fit linear mixed-effects models (random intercept for patient; fixed effects for group, time, and group × time) for continuous outcomes. Binary endpoints (residual MME >3 mm; MCID or PASS achieved) used logistic regression with robust (sandwich) SEs in matched/weighted samples. Multiplicity across secondary endpoints was controlled with Holm–Bonferroni. Effect sizes are reported as adjusted mean differences (mm or points) or odds ratios with 95% CIs. Reliability of MME/MEI readings is presented as ICC (2.1) with 95% CIs and interpretation per Koo & Li.²⁷

Finally, we specified subgroup analyses motivated by biomechanics: varus alignment (≥3° vs < 3°) and suspected posteromedial peripheral-attachment laxity (MRI/arthroscopy). We tested group × subgroup interactions for ΔMME and ΔIKDC, given laboratory evidence that varus and peripheral-attachment insufficiency increase forces across a root repair and that centralization reduces these forces.

Results

A total of 198 patients were screened; 62 were excluded (lateral/anterior root tears, advanced osteoarthritis, malalignment >5° varus, concomitant realignment or cartilage restoration, or incomplete imaging/PRO data), leaving 136 eligible. After propensity-score matching with variable-ratio (1: k) nearest-neighbour matching on age, sex, BMI, baseline extrusion, HKA varus, Outerbridge grade, and time to surgery, 96 patients were included in the primary analysis (posterior tunnel fixation, n = 54; double-tunnel centralization, n = 42). Baseline characteristics were well balanced (all standardized mean differences <0.10) (Table 1). Mean age was 56.2 ± 8.1 years, 61% were female, BMI 28.9 ± 3.2 kg/m², HKA varus 2.4° ± 1.6°, and median time from diagnosis to surgery 8.0 (IQR 5.0–12.0) weeks. Baseline extrusion was similar between groups (MME 3.54 ± 1.05 mm vs 3.58 ± 1.01 mm; MEI 0.33 ± 0.09 vs 0.34 ± 0.08).

Table 1.

Baseline characteristics after matching.

Characteristic	Posterior tunnel (n = 54)	Double-tunnel centralization (n = 42)
Age, years, mean ± SD	56.4 ± 8.2	55.9 ± 8.0
Female, n (%)	32 (59.3)	27 (64.3)
BMI, kg/m², mean ± SD	28.8 ± 3.3	29.0 ± 3.1
HKA varus, degrees, mean ± SD	2.4 ± 1.7	2.3 ± 1.6
Time to surgery, weeks, median (IQR)	8 (5–12)	8 (5–12)
Baseline MME, mm, mean ± SD	3.54 ± 1.05	3.58 ± 1.01
Baseline MEI, mean ± SD	0.33 ± 0.09	0.34 ± 0.08
Baseline IKDC, mean ± SD	49.6 ± 12.8	49.1 ± 13.1

BMI, body mass index; HKA, hip–knee–ankle; MME, medial meniscus extrusion; MEI, Meniscal Extrusion Index; IKDC, International Knee Documentation Committee.

Note. All baseline covariates were balanced (standardized mean differences <0.10).

Inter- and intra-rater reliability for MRI extrusion measurements were excellent (ICC (2.1) = 0.94 [95% CI 0.91–0.96] and 0.92 [0.88–0.95], respectively) (Table 2). Pre-to postoperative changes in meniscal extrusion were small and not statistically significant within either group, and there were no between-group differences after adjustment (Table 3). In the posterior tunnel group, MME changed from 3.54 ± 1.05 mm preoperatively to 3.47 ± 1.12 mm at 12 ± 4 months (Δ −0.07 mm; 95% CI −0.20 to 0.06; paired p = 0.29). In the double-tunnel centralization group, MME changed from 3.58 ± 1.01 mm to 3.49 ± 1.08 mm (Δ −0.09 mm; 95% CI −0.22 to 0.05; paired p = 0.21). The adjusted between-group mean difference in ΔMME was −0.03 mm (95% CI −0.17 to 0.12; p = 0.71). MEI findings mirrored absolute MME (posterior tunnel ΔMEI −0.01 ± 0.05 vs centralization −0.01 ± 0.05; adjusted difference 0.00; 95% CI −0.02 to 0.02; p = 0.98). The proportion with residual pathologic extrusion (MME >3 mm) at follow-up did not differ (65% posterior tunnel vs 62% centralization; adjusted OR 0.89; 95% CI 0.42–1.87; p = 0.76).

Table 2.

Reliability of MRI extrusion measurements.

Reliability metric	Estimate (95% CI)	Interpretation
Inter-rater ICC (2.1)	0.94 (0.91–0.96)	Excellent
Intra-rater ICC (2.1)	0.92 (0.88–0.95)	Excellent

Table 3.

MRI extrusion outcomes.

Outcome	Posterior tunnel (n = 54)	Double-tunnel centralization (n = 42)	Adjusted between-group difference (95% CI)	p value
Preoperative MME, mm, mean ± SD	3.54 ± 1.05	3.58 ± 1.01	—	—
Postoperative MME (12 ± 4 months), mm, mean ± SD	3.47 ± 1.12	3.49 ± 1.08	—	—
ΔMME (post–pre), mm	−0.07 (−0.20 to 0.06)^a	−0.09 (−0.22 to 0.05)^a	−0.03 (−0.17 to 0.12)	0.71
ΔMEI (post–pre), mean ± SD	−0.01 ± 0.05	−0.01 ± 0.05	0.00 (−0.02 to 0.02)	0.98
Residual pathologic extrusion (MME >3 mm), n/N (%)	35/54 (64.8)	26/42 (61.9)	OR 0.89 (0.42–1.87)	0.76

^aWithin-group paired p values: posterior tunnel p = 0.29; centralization p = 0.21.

MME, medial meniscus extrusion; MEI, Meniscal Extrusion Index; CI, confidence interval; OR, odds ratio.

Functional outcomes improved in both cohorts, with modestly greater gains in the double-tunnel centralization group (Table 4). IKDC increased from 49.6 ± 12.8 to 73.1 ± 14.2 in the posterior tunnel group (Δ + 23.5 ± 15.0) and from 49.1 ± 13.1 to 80.0 ± 13.6 in the centralization group (Δ + 30.9 ± 14.4). The adjusted between-group difference in ΔIKDC was +5.7 points (95% CI + 1.5 to +10.0; p = 0.008). A higher proportion of patients in the centralization group met the IKDC MCID (77% vs 63%; adjusted OR 1.94; 95% CI 1.01–3.73; p = 0.046), while PASS achievement showed a nonsignificant trend (60% vs 48%; adjusted OR 1.61; 95% CI 0.83–3.13; p = 0.16) (Table 5).

Table 4.

Functional outcomes.

Outcome (points unless stated)	Posterior tunnel (n = 54)	Double-tunnel centralization (n = 42)	Adjusted difference in change (95% CI)	p value	Holm–Bonferroni
IKDC baseline → 24 months	49.6 ± 12.8 → 73.1 ± 14.2	49.1 ± 13.1 → 80.0 ± 13.6	+5.7 (+1.5 to + 10.0)	0.008	Significant
KOOS pain, Δ	+16.2 ± 14.8	+20.7 ± 14.3	+4.5 (+0.4 to + 8.6)	0.032	n.s.
KOOS QoL, Δ	+12.1 ± 15.6	+18.0 ± 15.0	+5.9 (+1.3 to +10.5)	0.013	Significant
KOOS ADL, Δ	+10.8 ± 13.7	+13.9 ± 13.1	+3.0 (−0.7 to + 6.8)	0.11	—
Lysholm, Δ	+14.2 ± 16.0	+17.9 ± 15.5	+3.7 (0.0 to + 7.5)	0.050	n.s.
VAS pain (0–10), Δ	−2.0 ± 2.1	−2.6 ± 2.0	−0.6 (−1.1 to −0.0)	0.042	n.s.
Tegner activity, Δ	+0.4 ± 0.8	+0.6 ± 0.9	+0.2 (−0.1 to + 0.5)	0.19	—

IKDC, International Knee Documentation Committee; KOOS, Knee injury and Osteoarthritis Outcome Score; QoL, quality of life; ADL, activities of daily living; VAS, visual analogue scale; CI, confidence interval; n.s., not significant after multiplicity adjustment.

Table 5.

Responder analysis at 24 months.

Responder category	Posterior tunnel (n = 54)	Double-tunnel centralization (n = 42)	Adjusted effect (95% CI)	p value
IKDC MCID achieved, n (%)	34 (63.0)	32 (76.2)	OR 1.94 (1.01–3.73)	0.046
IKDC PASS achieved, n (%)	26 (48.1)	25 (59.5)	OR 1.61 (0.83–3.13)	0.16

MCID, minimal clinically important difference; PASS, patient acceptable symptom state; OR, odds ratio.

Secondary PROs showed a similar pattern. KOOS Pain improved by + 16.2 ± 14.8 points with posterior tunnel fixation and +20.7 ± 14.3 with centralization (adjusted difference + 4.5; 95% CI + 0.4 to +8.6; p = 0.032). KOOS QoL improved by + 12.1 ± 15.6 versus + 18.0 ± 15.0 (adjusted difference + 5.9; 95% CI + 1.3 to +10.5; p = 0.013). KOOS ADL improved by + 10.8 ± 13.7 versus + 13.9 ± 13.1 (adjusted difference +3.0; 95% CI −0.7 to +6.8; p = 0.11). Lysholm increased by + 14.2 ± 16.0 versus + 17.9 ± 15.5 (adjusted difference + 3.7; 95% CI 0.0 to +7.5; p = 0.050). VAS pain decreased by −2.0 ± 2.1 versus −2.6 ± 2.0 (adjusted difference −0.6; 95% CI −1.1 to −0.0; p = 0.042) (Table 4). After Holm–Bonferroni correction for multiple secondary endpoints, differences remained statistically significant for IKDC and KOOS QoL; other secondary comparisons did not retain significance.

Sensitivity analyses using inverse-probability of treatment weighting and linear mixed-effects models (group, time, group × time; random intercept for patient) yielded effect sizes and inferences consistent with the primary ANCOVA models for both structural and functional endpoints. Prespecified subgroup analyses suggested a larger functional benefit of centralization among knees with varus alignment ≥3 (interaction p = 0.048), without evidence of differential effects on meniscal extrusion (all interaction p > 0.10) (Table 6). No group × subgroup interaction was observed for time-to-surgery or suspected peripheral-attachment laxity with respect to extrusion; functional trends favoured centralization but were not consistently significant after multiplicity control.

Table 6.

Prespecified subgroup interactions.

Subgroup (definition)	Outcome	Interaction p value	Direction (centralization vs posterior tunnel)
Varus alignment ≥3	ΔIKDC	0.048	Greater functional benefit with centralization
Varus alignment ≥3	ΔMME	0.62	No differential effect on extrusion
Time to surgery ≤3 vs > 3 months	ΔIKDC	0.21	No interaction
Peripheral-attachment laxity suspected	ΔIKDC	0.18	Trend favouring centralization (not significant)
Peripheral-attachment laxity suspected	ΔMME	0.57	No interaction

Adverse events were infrequent and comparable (Table 7). There were no deep infections or thromboembolic events. Arthrofibrosis requiring manipulation under anesthesia occurred in 1/54 (1.9%) in the posterior tunnel group and 2/42 (4.8%) in the centralization group (p = 0.56). Reoperation for persistent symptoms occurred in 3/54 (5.6%) versus 2/42 (4.8%) (p = 0.78). One clinical repair failure in the posterior tunnel group was treated with partial meniscectomy; no structural failures were recorded in the centralization group during follow-up.

Table 7.

Adverse events and reoperations.

Event	Posterior tunnel (n = 54)	Double-tunnel centralization (n = 42)	p value
Deep infection	0 (0.0)	0 (0.0)	—
DVT/PE	0 (0.0)	0 (0.0)	—
Arthrofibrosis requiring MUA	1 (1.9)	2 (4.8)	0.56
Reoperation for persistent symptoms	3 (5.6)	2 (4.8)	0.78
Clinical repair failure	1 (1.9)^a	0 (0.0)

^aManaged with partial meniscectomy. DVT, deep-vein thrombosis; PE, pulmonary embolism; MUA, manipulation under anesthesia.

Discussion

In this retrospective matched cohort, MRI‐based medial meniscus extrusion (MME) did not change meaningfully from preoperative to 12 ± 4 months in either group, whereas patient‐reported outcomes improved in both cohorts with a modest advantage for double-tunnel centralization. The structural finding aligns with recent syntheses indicating that while posterior medial meniscus root repair (PMMRR) reliably improves symptoms and joint preservation versus meniscectomy or nonoperative care, extrusion often persists and may not significantly regress on standard supine MRI within the first postoperative years. A 2023 meta-analysis reported that repair improves PROs and biomechanics but does not consistently reduce MME, underscoring the dissociation between structural extrusion on static imaging and clinical benefit.^4,28 Our observation is therefore consistent with the evolving literature that treats MME as a stubborn marker of meniscal pathology that is slow to normalize, even after anatomically sound repair.

While the mean between-group difference in IKDC scores (+5.7 points) was statistically significant, it remained below the established MCID threshold of 10–11 points. However, this mean value obscures the individual benefit; the ‘responder analysis’ revealed that adding centralization significantly increased the probability of a patient achieving a clinically meaningful improvement (77% vs 63%, p = 0.046). Thus, while the average effect size is modest, the intervention increases the likelihood of surgical success for the individual patient.

Several longitudinal MRI series also suggest that MME may stabilize or even progress over mid-term follow-up despite clinical improvement, with meniscal healing quality inversely related to subsequent increases in extrusion.^29,30 These data help explain why our cohort showed unchanged MME yet robust functional gains at 2 years. Importantly, dynamic or flexion MRI studies indicate that repair may reduce posteromedial extrusion during knee flexion, an effect that static MCL-level measurements could miss, offering a mechanistic reason for the structural–clinical mismatch we observed.³⁰ Together, these reports support cautious interpretation of coronal supine MME as a surrogate for meniscal load sharing in daily activities.²⁹

Against this backdrop, our finding of modestly superior PROs with centralization—despite no measurable between-group difference in MME—fits with current biomechanical and early clinical evidence.¹⁵ Cadaveric and computational investigations over the last 5 years demonstrate that adding a centralization construct can reduce forces across a repaired root, particularly under varus alignment and when posteromedial peripheral attachments (meniscotibial/meniscocapsular ligaments) are lax. These load-sharing effects would be expected to lessen pain and improve function without necessarily producing a large change in static MME at the MCL slice. Our prespecified subgroup signal in varus knees (≥3°) mirrors that mechanism-based expectation.¹⁵

The dissociation between our improved clinical scores and the unchanged static MME warrants careful interpretation. While static supine MRI failed to show structural restoration, recent biomechanical evidence by Deichsel et al.¹⁵ demonstrates that centralization significantly reduces forces on the root repair under dynamic varus loading, restoring native force transmission patterns. Furthermore, Karpinski et al.³¹ described the ‘dead meniscus sign,’ highlighting that pathologic menisci lose their dynamic adaptability. Therefore, the functional superiority of centralization in our cohort may stem from improved dynamic containment during weight-bearing—a phenomenon that static supine imaging is inherently insensitive to detect.

Early clinical series and reviews published since 2021 also trend in the same direction. Case series of PMMRR augmented with centralization report significant improvements in pain, function, and quality of life at ∼2 years,²¹ with low early failure and no radiographic acceleration of arthritis; a 2025 systematic review concluded that centralization reduces extrusion and improves clinical outcomes, although the overall level of evidence remains predominantly Level III–IV and comparative data are limited.³² Our study adds to this literature by suggesting that the functional advantage of centralization can be detected even when a strict MRI protocol shows no group difference in MME, implying that the clinical effect may be mediated by improved load distribution rather than dramatic changes in static extrusion.

Technique nuances may also matter. Recent biomechanical work indicates that stitch location for transtibial centralization influences contact mechanics and extrusion control, with posterior horn placement performing best; similarly, studies comparing transtibial versus anchor-based centralization suggest construct-specific differences in restoring contact area and reducing rim lift-off.¹⁹ We used a second transtibial tunnel with a midbody horizontal mattress, which may optimize load sharing yet still yield small absolute changes on MCL-level MME.¹⁹ Future comparative trials should stratify by centralization technique and stitch position to clarify which variants translate into measurable structural benefits.

It is worth situating our findings within the broader treatment paradigm. Contemporary meta-analyses continue to show that root repair confers superior pain relief and joint preservation compared with meniscectomy or nonoperative care, reinforcing repair as the standard in salvageable knees.⁵ Our data do not contradict that consensus; rather, they refine it by indicating that adding centralization may provide incremental clinical benefit—especially in varus or suspected peripheral-attachment insufficiency—even when extrusion on static MRI remains largely unchanged at 1 year. This interpretation is congruent with reviews emphasizing that extrusion is a multifactorial phenomenon influenced by alignment and peripheral stabilizers, and that addressing these factors (via centralization and, when appropriate, alignment procedures) may optimize outcomes.³³

Several factors could explain why we did not detect a between-group difference in MME. First, our imaging window (12 ± 4 months) may be too early to capture structural divergence; some series demonstrate evolving extrusion trajectories out to 2–3 years. Second, measurement constraints of supine coronal MRI at the MCL may underestimate dynamic or posterior-horn–dominant changes most directly influenced by centralization. Third, our exclusion of >5 varus and advanced OA likely narrowed the structural effect size by removing patients in whom centralization’s de-extrusion effect is greatest. Each element highlights why future studies should incorporate weightbearing or flexion MRI, longer follow-up, and a priori alignment strata.

Strengths of our study include a clearly defined imaging protocol with excellent inter/intra-rater reliability, propensity methods to mitigate selection bias, and a focus on a homogeneous medial root tear population treated with either anatomic pull-out alone or the same pull-out plus transtibial centralization. Regarding peripheral-attachment laxity, our subgroup analysis did not reach statistical significance (P = 0.18), likely due to the subjective nature of the retrospective grading and limited sample size. However, we retained this variable in our conceptual framework because recent biomechanical evidence by Deichsel et al.¹⁵ explicitly demonstrates that peripheral insufficiency increases root repair loads, which can be successfully offloaded by centralization. Therefore, while our clinical data shows only a trend, the biological rationale for stabilizing a hypermobile meniscus remains robust. Nonetheless, limitations remain: the retrospective design and potential residual confounding (e.g., subtle differences in capsular quality), the reliance on static MRI rather than dynamic imaging, and a follow-up that, although adequate for PROs, may be short for structural endpoints. Our study utilized a specific double-tunnel transtibial centralization technique. Consequently, our findings may not be fully generalizable to suture-anchor–based centralization methods reported by Krych et al. or Koga et al.^17,21 However, recent biomechanical comparisons suggest that transtibial centralization may offer superior restoration of contact mechanics and extrusion reduction compared to knotless anchors, implying that technique selection is a critical variable in outcomes.¹⁹ It is important to address the selection criteria regarding meniscal extrusion. Although centralization was adopted for cases with preoperative extrusion or suspected peripheral laxity, our propensity-score matching process was specifically designed to balance baseline extrusion values between groups (mean ∼3.55 mm in both). While this matching allows for a rigorous statistical comparison of the two techniques without the confounding effect of baseline deformity, it may have excluded patients with extreme extrusion who underwent centralization but lacked a comparable match in the isolated repair group. Furthermore, the decision to centralize was often influenced by intraoperative assessment of peripheral meniscotibial ligament laxity, a dynamic parameter not fully captured by static supine MRI. As demonstrated by Deichsel et al.¹⁵ peripheral insufficiency significantly increases forces on the root repair. Therefore, the lack of difference in baseline MRI extrusion does not imply inconsistent indications, but rather reflects that our cohort consists of patients with comparable static deformities where the addition of centralization was largely driven by intraoperative instability patterns to unload the root repair. The literature’s current limitations are similar—many reports are single-arm or retrospective, and relatively few provide head-to-head comparisons or standardized MRI outcomes—so there is a clear need for prospective comparative studies with harmonized imaging and clinical metrics.⁵

A significant limitation is the lack of long-term joint preservation data, such as progression of Kellgren-Lawrence grade or conversion to total knee arthroplasty (TKA). As shown by Lamba et al.⁶ clinical failure and TKA conversion often manifest over a decade. Therefore, our 2-years follow-up captures early functional recovery but cannot yet confirm whether centralization confers a durable chondroprotective benefit compared to isolated root repair.³

We excluded patients with >5 varus alignment to isolate the effect of the soft-tissue procedure from bony malalignment. However, biomechanical data from Deichsel et al.¹⁵ demonstrate that centralization is most effective in unloading the root repair specifically under varus loading conditions. By excluding severe varus, we may have paradoxically underestimated the potential benefit of centralization in the population that biomechanically needs it most. Our finding that even mild varus (≥3°) showed a significant interaction favoring centralization supports this hypothesis.

In adult PMMRTs suitable for repair, surgeons can reasonably expect substantial functional gains at 2 years after either anatomic pull-out or pull-out plus centralization. Where varus alignment or peripheral-attachment laxity is present, adding a second-tunnel centralization may yield incremental clinical benefit, likely by improving load sharing rather than by producing large early reductions in static MME.¹⁶ This strategy is consistent with the latest biomechanical insights and early clinical series and can be applied without materially increasing complications; however, definitive proof of superiority will require high-quality comparative trials with longer follow-up and dynamic imaging.²⁴

Conclusions

Anatomic posterior root repair reliably improved knee function despite largely unchanged static MRI extrusion. The addition of a second-tunnel centralization provided a statistically significant incremental benefit. While the mean between-group difference was below the established MCID threshold, centralization significantly increased the probability of achieving a clinically meaningful improvement. Therefore, centralization should be utilized as a selective adjunct—particularly in varus alignment. Although our statistical analysis for peripheral-attachment laxity was inconclusive, recent biomechanical evidence suggests centralization may also benefit cases with meniscotibial insufficiency by unloading the root repair. Future prospective studies utilizing dynamic imaging are warranted to confirm whether these functional gains translate into superior long-term joint preservation.

Footnotes

ORCID iDs

Yaşar Samet Gökçeoğlu

Muhammed Furkan Darılmaz

Ethical considerations

This retrospective cohort study was conducted across two centers—the Departments of Orthopaedics and Traumatology at Aksaray University Training and Research Hospital and Sanliurfa Mehmet Akif Inan Training and Research Hospital. The study protocol was approved by the Institutional Ethics Committee of Aksaray University (Protocol: SAGETİK 2025-197, Approval number: 2026/03, Date: January 08, 2026). All participants provided written informed consent for the use of their clinical and imaging data for research purposes, and all procedures were conducted in compliance with institutional ethical standards and the Declaration of Helsinki.

Consent for publication

All participants provided written informed consent for the use of their clinical and imaging data for research purposes, and all procedures were conducted in compliance with institutional ethical standards and the Declaration of Helsinki.

Authors’ contributions

Y.S.G. and M.F.D. contributed to the study conception and design. Material preparation, data collection and analysis were performed by Y.S.G. The first draft of the manuscript was written by Y.S.G. and M.F.D., and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Appendix

References

Allaire

Muriuki

Gilbertson

, et al. Biomechanical consequences of a tear of the posterior root of the medial meniscus. Similar to total meniscectomy. J Bone Joint Surg Am 2008; 90: 1922–1931. https://doi.org/10.2106/JBJS.G.00748

Shen

Zou

, et al. Comparison of the effects between arthroscopic transtibial pullout technique and all-inside repair in the treatment of medial meniscus posterior root tears. J Orthop 2025; 65: 78–85. 20241216. https://doi.org/10.1016/j.jor.2024.12.003

Krivicich

Kunze

Parvaresh

, et al. Comparison of long-term radiographic outcomes and rate and time for conversion to total knee arthroplasty between repair and meniscectomy for medial Meniscus Posterior root tears: a systematic review and meta-analysis. Am J Sports Med 2022; 50: 2023–2031. 20210712. https://doi.org/10.1177/03635465211017514

Elnewishy

Elsenosy

Nahas

, et al.

A systematic review and meta-analysis of medial meniscus root tears: is surgery the key to better outcomes?

Cureus 2024; 16(20241206): e75199. https://doi.org/10.7759/cureus.75199

Lee

Reinholz

, et al. Root repair has superior radiological and clinical outcomes than partial meniscectomy and nonoperative treatment in the management of meniscus root tears: a systematic review. Arthroscopy 2025; 41: 390–417. 20240222. https://doi.org/10.1016/j.arthro.2024.02.017

Lamba

Regan

Levy

, et al. Long-term outcomes of partial meniscectomy for degenerative medial meniscus posterior root tears. Orthop J Sports Med 2024; 12: 23259671241266593. 20240904. https://doi.org/10.1177/23259671241266593

Floyd

Rodriguez

Falaas

, et al. The natural history of medial meniscal root tears: a biomechanical and clinical case perspective. Front Bioeng Biotechnol 2021; 9: 744065. 20210923. https://doi.org/10.3389/fbioe.2021.744065

Liao

Nie

, et al. Risk factors of incomplete healing following medial meniscus posterior root tear repair with gracilis tendon. Sci Rep 2023; 13: 22978. 20231227. https://doi.org/10.1038/s41598-023-50358-z

Langhans

Lamba

Saris

DBF

, et al. Meniscal extrusion: diagnosis, etiology, and treatment options. Curr Rev Musculoskelet Med 2023; 16: 316–327. 20230516. https://doi.org/10.1007/s12178-023-09840-4

10.

Dastan

Bicer

Kaya

, et al. Posteromedial opening wedge high tibial osteotomy has favourable outcomes in simultaneous medial meniscus posterior root repair and varus medial knee osteoarthritis patients without concomitant root tear. J Orthop Surg Res 2025; 20: 97. 20250125. https://doi.org/10.1186/s13018-025-05519-3

11.

Karatekin

Altinayak

Genc

, et al. Anterior cruciate ligament tissue stiffness and anterior tibial translation are increased in patients with medial meniscus posterior root tear. J Orthop Surg Res 2025; 20: 228. 20250303. https://doi.org/10.1186/s13018-025-05601-w

12.

Bernard

Kennedy

Tagliero

, et al. Medial Meniscus Posterior root tear treatment: a matched cohort comparison of nonoperative management, partial meniscectomy, and repair. Am J Sports Med 2020; 48: 128–132. 20191125. https://doi.org/10.1177/0363546519888212

13.

Papalia

Saccone

, et al. Meniscal extrusion: risk factors and diagnostic tools to predict early osteoarthritis. Orthop Rev (Pavia) 2023; 15: 74881. 20230513. https://doi.org/10.52965/001c.74881

14.

Higashihara

Furumatsu

Okazaki

, et al. Transtibial pullout repair improved short-term clinical outcomes in patients with oblique medial meniscus posterior root tear comparable to radial root tear. Eur J Orthop Surg Traumatol 2024; 35: 30. 20241128. https://doi.org/10.1007/s00590-024-04137-9

15.

Deichsel

Peez

Raschke

, et al. Arthroscopic centralization of the medial meniscus reduces load on a posterior root repair under dynamic varus loading: a biomechanical investigation. Am J Sports Med 2024; 52: 3030–3038. 20240915. https://doi.org/10.1177/03635465241274791

16.

Tamura

Furumatsu

Kitayama

, et al. The preoperative flexion tear gap affects postoperative meniscus stability after pullout repair for medial meniscus posterior root tear. Knee Surg Relat Res 2025; 37: 16: 20250403. https://doi.org/10.1186/s43019-025-00264-7

17.

Koga

Nakamura

Nakagawa

, et al. Arthroscopic centralization using knotless anchors for extruded medial meniscus. Arthrosc Tech 2021; 10: e639-e645. 20210202. https://doi.org/10.1016/j.eats.2020.10.051

18.

Leafblad

Smith

Stuart

, et al. Arthroscopic centralization of the extruded medial meniscus. Arthrosc Tech 2021; 10: e43–e48. 20201219. https://doi.org/10.1016/j.eats.2020.09.005

19.

Boksh

Espino

Ghosh

, et al. Transtibial centralization better restores meniscal extrusion and contact mechanics compared with knotless anchor centralization for medial meniscal posterior root tears: an In vitro biomechanical Study using a porcine model. Arthroscopy 2025; 41: 4615–4626. e4613. 20250427. https://doi.org/10.1016/j.arthro.2025.04.041

20.

Ozeki

Koga

Matsuda

, et al. Biomechanical analysis of the centralization procedure for extruded lateral menisci with posterior root deficiency in a porcine model. J Orthop Sci 2020; 25: 161–166. 20190319. https://doi.org/10.1016/j.jos.2019.02.015

21.

Krych

Boos

Lamba

, et al. Satisfactory clinical outcome, complications, and provisional results of meniscus centralization with medial meniscus root repair for the extruded medial meniscus at mean 2-Year Follow-Up. Arthroscopy 2024; 40: 1578–1587. 20231012. https://doi.org/10.1016/j.arthro.2023.10.003

22.

Rao

, et al. Biomechanical assessment of disease outcome in surgical interventions for medial meniscal posterior root tears: a finite element analysis. BMC Musculoskelet Disord 2022; 23: 1093. 20221214. https://doi.org/10.1186/s12891-022-06069-z

23.

Nakagawa

Ozeki

Koga

. A narrative review of lateral meniscus root tears and extrusion: techniques and outcomes. Ann Jt 2022; 7: 15. 20220415. https://doi.org/10.21037/aoj-20-112

24.

Zhou

Yang

Kang

, et al. Clinical effect of medial meniscus posterior root repair combined with centralization technique in the treatment of medial meniscus posterior root tears. BMC Musculoskelet Disord 2024; 25: 982. 20241130. https://doi.org/10.1186/s12891-024-08125-2

25.

Zhang

Liu

Zhang

, et al. Influence of preoperative medial meniscus extrusion and subchondral bone marrow edema on outcomes after medial opening wedge high tibial osteotomy. J Orthop Surg Res 2025; 20: 353. 20250409. https://doi.org/10.1186/s13018-025-05656-9

26.

Compagnoni

Ferrua

Minoli

, et al. The meniscal extrusion index is a reliable indirect sign of different meniscal lesion patterns: a classification based on percentage of meniscal extrusion. Knee Surg Sports Traumatol Arthrosc 2023; 31: 5005–5011. 20230831. DOI: 10.1007/s00167-023-07525-6.

27.

Koo

. A guideline of selecting and reporting intraclass correlation coefficients for reliability research. J Chiropr Med 2016; 15: 155–163. 20160331. https://doi.org/10.1016/j.jcm.2016.02.012

28.

Perry

Lavoie-Gagne

Knapik

, et al. Examining the efficacy of medial Meniscus posterior root repair: a meta-analysis and systematic review of biomechanical and clinical outcomes. Am J Sports Med 2023; 51: 1914–1926. 20220406. https://doi.org/10.1177/03635465221077271

29.

Kawada

Furumatsu

Yokoyama

, et al. Meniscal healing status after medial meniscus posterior root repair negatively correlates with a midterm increase in medial meniscus extrusion. Knee Surg Sports Traumatol Arthrosc 2024; 32: 2219–2227. 20240513. https://doi.org/10.1002/ksa.12245

30.

Okazaki

Sugiu

Kamatsuki

, et al. Clinical outcomes following medial meniscus posterior root repairs: a minimum of 5-year follow-up study. J Exp Orthop 2025; 12: e70262. 20250507. https://doi.org/10.1002/jeo2.70262

31.

Karpinski

Diermeier

Willinger

, et al. No dynamic extrusion of the medial meniscus in ultrasound examination in patients with confirmed root tear lesion. Knee Surg Sports Traumatol Arthrosc 2019; 27: 3311–3317. 20190103. https://doi.org/10.1007/s00167-018-5341-4

32.

Carrozzo

Bosco

Ramazzini

, et al. Meniscal centralization significantly improve clinical outcomes and reduce meniscal extrusion with minimal complications: a systematic review. J Exp Orthop 2025; 12: e70308. 20250615. https://doi.org/10.1002/jeo2.70308

33.

Guimaraes

Chemin

Araujo

, et al. Meniscal root tears: an update focused on preoperative and postoperative MRI findings. AJR Am J Roentgenol 2022; 219: 269–278. 20220316. https://doi.org/10.2214/AJR.22.27338

Posterior root repair versus double-tunnel centralization in medial meniscus root tears: A propensity-matched analysis

Abstract

Purpose

Methods

Results

Conclusions

Keywords

Introduction

Methods

Study design and setting

Results

Discussion

Conclusions

Footnotes

ORCID iDs

Ethical considerations

Consent for publication

Authors’ contributions

Funding

Declaration of conflicting interests

Appendix

References