Intelligent versus flexible negative-pressure aspiration sheath for ureteroscopic lithotripsy: a single-center retrospective study with systematic review and meta-analysis

Abstract

Background

Negative-pressure aspiration ureteral access sheaths (NP-UAS) outperform traditional sheaths in retrograde intrarenal surgery (RIRS), but whether the two designs—the intelligent pressure-controlled type (real-time intrarenal pressure monitoring) and the flexible negative-pressure type (steerable tip)—differ in clinical outcomes is unknown. Three head-to-head comparisons exist, all using clinically insignificant residual fragment (CIRF)-based stone-free rate (SFR) definitions, with no systematic synthesis. This study aimed to determine whether the two designs differ in clinically relevant outcomes after RIRS.

Methods

This study combined a single-center retrospective cohort (264 patients; 147 intelligent, 117 flexible NP-UAS) with a meta-analysis of all published head-to-head comparisons (4 studies, 539 patients). The primary outcome was SFR, defined as zero residual fragments on 1-mm computed tomography (CT) within 48 hours to maximize sensitivity to between-group differences. Confounding by indication was addressed with multivariable regression, propensity score matching (PSM), inverse probability of treatment weighting (IPTW), Firth penalized likelihood, and E-value analysis; the meta-analysis used a restricted maximum likelihood (REML) random-effects model. Certainty was graded using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach.

Results

Crude SFR was 24.5% (intelligent) versus 14.5% (flexible) [odds ratio (OR) 1.91, 95% confidence interval (CI): 1.02–3.68, P=0.047], attenuating to non-significance after sequential adjustment (operative time excluded: OR 1.72, 95% CI: 0.83–3.68, P=0.15), with consistent results across PSM (OR 1.67), IPTW (OR 1.49), and Firth regression (OR 1.65); E-values were 3.23 (crude) and 2.83 (adjusted). Operative time was shorter with the flexible sheath (crude −25.9 min), but mediation analysis showed it did not mediate the sheath-SFR relationship [average causal mediation effect (ACME) P=0.30]. Postoperative infection did not differ between groups. In the exploratory meta-analysis, subgrouping by SFR definition reduced I² from 75.7% to 0.7%, though confounded with study center. GRADE certainty was very low for all outcomes.

Conclusions

Across five analytical methods and meta-analysis, the evidence is insufficient to establish a clinically meaningful SFR difference between intelligent and flexible NP-UAS. The flexible sheath is associated with shorter operative time, but without differential stone clearance, and GRADE certainty is very low. A non-inferiority randomized controlled trial (RCT) (approximately 500 patients) with standardized zero-fragment CT SFR is required.

Highlight box.

Key findings

• Five complementary methods (multivariable regression, propensity score matching, inverse probability of treatment weighting, Firth penalized likelihood, E-value analysis) and an exploratory meta-analysis (4 studies, 539 patients) found insufficient evidence of a clinically meaningful difference in stone-free rate (SFR) between intelligent pressure-controlled and flexible negative-pressure aspiration ureteral access sheaths (NP-UAS); Grading of Recommendations Assessment, Development and Evaluation certainty of evidence is very low.

• SFR definition was completely confounded with study center; stratification by definition reduced between-study heterogeneity (I² from 75.7% to 0.7%).

• The flexible sheath was associated with approximately 26 minutes shorter operative time; this finding is hypothesis-generating.

What is known and what is new?

• NP-UAS improve stone clearance over traditional sheaths, but head-to-head evidence between the two designs was lacking.

• This first systematic synthesis demonstrates that current evidence cannot establish a difference in stone clearance or infection outcomes.

What is the implication, and what should change now?

• Sheath selection should be individualized on the basis of patient factors and local expertise until a multicenter non-inferiority randomized controlled trial (approximately 500 patients) with standardized SFR definitions is completed; SFR reporting should be standardized across future studies.

Introduction

Background

Ureteroscopic lithotripsy with retrograde intrarenal surgery (RIRS) is a first-line treatment for renal and ureteral stones (1). The ureteral access sheath (UAS) facilitates repeated passage of the ureteroscope, improves irrigation outflow, and reduces intrarenal pressure—a key determinant of postoperative infectious complications (2,3). Recent technological advances have introduced negative-pressure aspiration UAS, which actively evacuates stone fragments during lithotripsy through an integrated suction channel, potentially improving stone clearance while maintaining low intrarenal pressure (4).

Two major types of negative-pressure UAS are currently in clinical use: the intelligent pressure-controlled aspiration sheath, which incorporates real-time intrarenal pressure monitoring with automated irrigation-suction feedback regulation; and the flexible negative-pressure aspiration sheath, which features a steerable bendable tip for direct calyceal access (5,6). The class-effect superiority of negative-pressure sheaths over traditional non-aspiration UAS is well established (7).

Rationale and knowledge gap

Whether the two negative-pressure sheath types—which differ substantially in design philosophy (electronic monitoring versus mechanical flexibility)—yield different clinical outcomes remains an open question. The intelligent sheath’s real-time pressure monitoring is hypothesized to improve stone clearance by maintaining a stable, low-pressure irrigation environment that optimizes fragment evacuation, whereas the flexible sheath’s steerable tip is hypothesized to improve access to lower-pole calyces; whether either mechanism translates into a superior stone-free rate (SFR) is unproven. Three head-to-head comparative studies have been published (5,6,8), but all used a clinically insignificant residual fragment (CIRF)-based SFR definition, and all three originated from a single center (Zigong, China), limiting generalizability. No systematic evidence synthesis has addressed this clinically relevant comparison.

Objective

This study had two complementary objectives. First, we conducted a single-center retrospective analysis of 264 consecutive patients comparing intelligent and flexible negative-pressure sheaths under a strict zero-fragment computed tomography (CT)-based SFR definition, employing five complementary analytical methods to address confounding by indication. Second, we conducted a systematic review and meta-analysis of all published head-to-head comparisons to contextualize these findings and to test whether SFR-definition heterogeneity could explain the discrepancy in reported outcomes. We present this article in accordance with the STROBE and PRISMA reporting checklists (available at https://tau.amegroups.com/article/view/10.21037/tau-2026-0611/rc).

Methods

Study design and patients

We retrospectively reviewed consecutive patients who underwent RIRS at the Department of Urology, Xiangxi Tujia and Miao Autonomous People’s Hospital between January 2022 and December 2024. Inclusion criteria were: (I) age ≥18 years; (II) renal or ureteral stones treated with flexible ureteroscopy (FURS); (III) use of either intelligent pressure-controlled or flexible negative-pressure aspiration sheath. Exclusion criteria were missing sheath type or key outcome data. After excluding one patient with missing sheath type, 264 patients were included for analysis (147 intelligent, 117 flexible). The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Ethics Committee of Xiangxi Tujia and Miao Autonomous People’s Hospital (No. EC-LCKY2026055) and the requirement for informed consent was waived owing to the retrospective, anonymized design of the study.

Sheath selection and confounding by indication

All procedures were performed by two experienced endourologists. Both sheath types were available throughout the study period, and sheath selection was at the discretion of the operating surgeon. Because the rationale for sheath selection was not prospectively recorded, the selection pattern described below is inferred from the observed baseline differences and clinical plausibility rather than from documented decision logs. The intelligent pressure-controlled sheath was preferentially selected for patients with preoperative infection, large stone burden, or anatomical complexity, given its real-time intrarenal pressure monitoring and automated irrigation-suction regulation, which were perceived to enhance safety in high-risk cases. The flexible sheath was typically selected for cases requiring direct calyceal access (e.g., lower-pole stones) in patients without significant infectious risk, owing to its steerable bendable tip. This clinical selection logic, while rational, introduces confounding by indication: the two groups differ systematically in baseline risk, and crude outcome comparisons are therefore biased. We addressed this through the multi-method analytical framework described below. However, we acknowledge that undocumented factors—including device availability on the day of surgery, operating room scheduling constraints, and individual surgeon preference—may also have influenced sheath assignment, and the direction of any resulting residual confounding cannot be determined with certainty.

Surgical technique

All procedures were performed by the two endourologists using a flexible ureteroscope (Wolf; Richard Wolf GmbH, Knittlingen, Germany) and a holmium:YAG laser (Raykeen, Shanghai, China; 20 W, 0.5–1.0 J, 10–20 Hz, dusting mode). In the intelligent group, a 12/14 Fr intelligent pressure-controlled negative-pressure aspiration ureteral access sheaths (NP-UAS; Inventor Technology, Ganzhou, China) was used with an Inventor Technology intelligent irrigation-suction monitoring system that maintained a continuous irrigation flow of 80–120 mL/min while monitoring intrarenal pressure against a safety threshold of 30 mmHg. In the flexible group, an 11/13 Fr flexible negative-pressure NP-UAS (Well Lead, Guangzhou, China) was used with a single-use irrigation pump set at 120 mmHg. No patient in either group underwent pre-stenting. A double-J ureteral stent was placed postoperatively in all patients and was removed after 3 weeks in both groups.

Variable framework

Variables were classified into four categories: preoperative baseline (sex, age, total hospital stay, height, weight, body mass index (BMI), first-episode symptoms, ureteral surgery history, hypertension, diabetes, other comorbidities); stone complexity [stone multiplicity, stone size, stone CT attenuation value, stone location, preoperative infection]; core grouping (sheath type); and outcomes (postoperative infection, postoperative hospital stay, stone-free status, operative time). All binary variables used 0= no /1= yes encoding. A complete variable dictionary is provided in Table S1.

Outcome definitions

The primary outcome was SFR, defined as complete absence of any residual fragment on postoperative non-contrast CT (1-mm slice thickness) performed within 48 hours of surgery. This zero-fragment definition was selected for several reasons. First, it provides the highest sensitivity for detecting between-group differences by eliminating the classification ambiguity inherent in CIRF thresholds (<2, <3, or <4 mm). Second, residual fragments of any size may serve as nidus for stone regrowth and have been associated with higher rates of stone-related events (9); even CIRF (≤4 mm) carry substantial risks of complications (27%) and re-intervention (21%) (10). Third, the use of a strict definition in a retrospective study enables more conservative and defensible comparisons: if no difference is detected under the most sensitive definition, it is unlikely that a clinically meaningful difference exists under a more lenient one. We acknowledge that the SFR observed under this definition (20.1%) is substantially lower than rates reported under CIRF-based definitions (typically 75–95%), and that imaging at 48 hours may not capture delayed fragment passage. Published benchmarks employing comparably strict SFR criteria are compiled in Table S2.

Secondary outcomes included operative time (minutes from ureteroscope insertion to removal), postoperative infection, and postoperative hospital stay. The multi-dimensional outcome profile across SFR, operative time, postoperative infection, and hospital stay is visualized in Figure S1. Postoperative infection was defined as fever ≥38 ℃ (any measurement within 30 days post-surgery) or positive urine culture (≥10⁵ CFU/mL) collected within 30 days of surgery. This definition was adapted from the Centers for Disease Control and Prevention (CDC), National Healthcare Safety Network (NHSN) criteria for surgical site infection and urinary tract infection (11), with the 30-day surveillance window corresponding to the standard deep/organ-space infection monitoring period. We note that inclusion of positive urine culture may capture cases of asymptomatic bacteriuria in addition to clinically significant infection, and that the observed infection rate (37.5%) should be interpreted with this caveat. Preoperative infection was defined using the same criteria applied postoperatively (fever ≥38 ℃ within 30 days before surgery, or a positive preoperative urine culture ≥10⁵ CFU/mL), to ensure definitional consistency between the baseline covariate and the outcome.

Statistical analysis

Baseline characteristics were compared using standardized mean differences (SMDs), with SMD >0.20 indicating meaningful imbalance. Multivariable logistic regression for SFR was performed in four sequential models: (A) unadjusted; (B) adjusted for clinical variables (age, sex, BMI, preoperative infection); (C) additionally adjusted for stone complexity markers (size, CT value, multiplicity, location); (D) fully adjusted including comorbidities (hypertension, diabetes, ureteral surgery history). Because formal mediation analysis demonstrated that operative time was a post-treatment variable rather than a confounder (see results), operative time was deliberately excluded from the primary models B–D to avoid post-treatment adjustment; versions of models B and D that additionally included operative time were computed and reported for transparency.

Given the non-randomized design and documented baseline imbalance, three complementary approaches were employed to address confounding by indication. First, 1:1 propensity score matching (PSM) was performed using nearest-neighbor matching with a caliper of 0.2 standard deviations of the logit of the propensity score. The propensity score was estimated via logistic regression including age, sex, BMI, stone size, stone CT value, stone multiplicity, stone location, preoperative infection, hypertension, diabetes, and ureteral surgery history. Post-matching balance was assessed using SMD with a threshold of <0.10. Conditional logistic regression stratified by matched pair was used to estimate the sheath effect. Second, stabilized inverse probability of treatment weighting (IPTW) with weight truncation at the 99th percentile was applied. Third, Firth penalized maximum likelihood estimation (logistf package) was used for the fully adjusted model to address small-sample bias [events per variable (EPV) ≈3.5 in the 15-parameter model D].

E-value analyses were performed to quantify the minimum strength of association that an unmeasured confounder would need to have with both sheath selection and SFR to fully explain away the observed association; E-values were computed for both the crude odds ratio (OR) and the fully adjusted model D OR, together with the confidence-interval-based E-value (12). We pre-specified a non-inferiority margin of 10 percentage points absolute SFR difference for clinical interpretation. Sensitivity analyses for the diabetes covariate included a sheath × diabetes interaction test, Firth-penalized estimation of the diabetes effect, and exclusion of all diabetic patients.

The average hospital stay of approximately 10 days in this cohort reflects the standard care pathway at a prefectural-level hospital in western China, where patients are typically admitted for comprehensive preoperative evaluation (laboratory testing, imaging, infection control if indicated), surgical intervention, and postoperative observation. The median postoperative stay was 4 days.

All analyses were performed in R 4.5.3 (R Foundation for Statistical Computing, Vienna, Austria) using the MatchIt, survey, survival, and logistf packages. Certainty of evidence was rated using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach (Table S3).

Supplementary systematic review and meta-analysis

A systematic review and meta-analysis of all published head-to-head comparisons between intelligent and flexible NP-UAS was conducted as an exploratory synthesis, given the small number of available studies. Five databases (PubMed, Embase, Cochrane CENTRAL, CNKI, WanFang) were searched from inception to May 2026. Four studies (539 patients) met inclusion criteria, three of which originated from a single center (Zigong, China). The study selection process is summarized in the PRISMA 2020 flow diagram (Figure S2). Random-effects meta-analysis [restricted maximum likelihood (REML) with Hartung-Knapp-Sidik-Jonkman adjustment] was performed, with a pre-specified subgroup analysis by SFR definition (zero-fragment CT versus CIRF <2–3 mm). Risk of bias was assessed using RoB 2 [for the single randomized controlled trial (RCT)] and ROBINS-I (for non-randomized studies). Here we provide a brief summary of the meta-analytic results relevant to the interpretation of the present single-center findings.

Results

Baseline characteristics

Patient demographics and stone complexity markers were largely balanced between groups (Table 1). Stone size (24.6±4.4 versus 24.3±4.8 mm, P=0.64) and CT attenuation (1,075±330 versus 1,047±314 HU, P=0.49) were well matched (SMD <0.10). Multiple stones were more common in the intelligent group (85.0% versus 74.4%, P=0.04, SMD =0.27).

Table 1. Baseline characteristics of the study cohort stratified by sheath type.

Variable	Intelligent (n=147)	Flexible (n=117)	SMD	P value
Demographics
Age (years)	54.9±12.8	51.7±11.5	0.26	0.04
Sex (male)	70 (47.6)	72 (61.5)	0.28	0.03
Height (cm)	158.8±8.5	159.8±8.3	0.12	0.33
Weight (kg)	61.7±11.6	64.7±12.8	0.25	0.04
BMI (kg/m2)	24.4±3.9	25.2±3.8	0.21	0.10
Hospital stay (days)	11.4±6.1	9.6±2.8	0.37	0.004
Comorbidities
Hypertension	45 (30.6)	29 (24.8)
Diabetes	22 (15.0)	8 (6.8)
Other comorbidities	9 (6.1)	5 (4.3)
Prior ureteral surgery	38 (25.9)	31 (26.5)
Stone complexity
Stone size (mm)	24.6±4.4	24.3±4.8	0.06	0.64
Stone CT value (HU)	1,074.8±329.1	1,047.2±312.5	0.09	0.49
Single stone	22 (15.0)	30 (25.6)	0.27	0.04
Preoperative infection	97 (66.0)	47 (40.2)	0.54	<0.001

Data are presented as mean ± SD for continuous variables and n (%) for categorical variables. SMD >0.20 indicates meaningful between-group imbalance. BMI, body mass index; CT, computed tomography; HU, Hounsfield unit; SD, standard deviation; SMD, standardized mean difference.

Consistent with the clinical selection logic described above, the intelligent sheath group had a significantly higher rate of preoperative infection (66.0% versus 40.2%, SMD =0.54, P<0.001), longer total hospital stay (11.4±6.1 versus 9.6±2.8 days, SMD =0.37, P=0.004), and longer operative time (126.0±44.8 versus 100.1±34.9 min, SMD =0.65, P<0.001). These imbalances were clinically expected consequences of the deliberate selection of the intelligent sheath for higher-risk patients.

Primary outcome: SFR

The overall SFR was 20.1% (53/264) under the strict zero-fragment 1-mm CT criterion at 48 hours postoperatively. While this appears low relative to the 75–95% SFR commonly cited in the RIRS literature, it is consistent with published series employing comparably strict criteria for large stone burdens. Ulvik et al. reported a true SFR of 54.2% in 1,019 all-comer ureteroscopy (URS) procedures using zero-fragment CT at 3 months (13), and 49% for renal stones treated with Ho:YAG laser in a subsequent RCT (14). In a 25-center prospective study using suction UAS [flexible and navigable suction ureteral access sheath (FANS)], Gauhar et al. reported 57.4% zero-fragment SFR on CT within 30 days (15). The European Association of Urology (EAU) Section of Urolithiasis has stated that “true SFR based on post-operative CT scan are under 50% with FURS without suction” (16). The lower SFR in the present cohort (20.1%) relative to these benchmarks reflects the combined effect of three factors: (I) the strictest possible criterion (1-mm CT slices versus standard 3–5 mm); (II) very early assessment (48 hours versus 3 months, precluding delayed fragment passage); and (III) mean stone size of 24.5 mm, which is larger than the populations in the Ulvik and Gauhar series. A formal comparison of SFR definitions and stone sizes across published RIRS series is provided in Table S2.

Crude SFR was 24.5% (36/147) in the intelligent group versus 14.5% (17/117) in the flexible group, corresponding to an unadjusted OR of 1.91 [95% confidence interval (CI): 1.02–3.68, P=0.047]. The apparent advantage of the intelligent sheath progressively attenuated and lost significance with sequential adjustment (Figure 1, Table 2).

Figure 1.

Sequential multivariable adjustment demonstrating progressive attenuation of the sheath-type effect on SFR. ORs (OR >1 favors intelligent sheath) plotted on logarithmic scale across four nested logistic regression models. Dashed line at OR =1 indicates no difference. OR, odds ratio; SFR, stone-free rate.

Table 2. Multivariable logistic regression and sensitivity analyses for SFR.

Model	OR (intelligent versus flexible)	95% CI	P value
A: unadjusted	1.91	1.02–3.68	0.047
B: + clinical variables	1.86	0.96–3.73	0.07
C: + stone complexity	1.74	0.87–3.60	0.13
D: + comorbidities (full)	1.72	0.83–3.68	0.15

Operative time is excluded from the primary models because it is a post-treatment variable (see Methods). The full model D is adjusted for age, sex, BMI, preoperative infection, stone size, stone CT value, stone multiplicity, stone location, hypertension, diabetes, and prior ureteral surgery. Cross-method comparison demonstrates consistent attenuation of the sheath effect. BMI, body mass index; CI, confidence interval; CT, computed tomography; OR, odds ratio; SFR, stone-free rate.

Operative time was excluded from all models because mediation analysis identified it as a post-treatment variable (see Methods); for transparency, versions of models B and D that additionally included operative time yielded OR 1.43 (95% CI: 0.70–2.99) and OR 1.51 (95% CI: 0.69–3.38), respectively, consistent with the primary estimates.

Propensity score-based analyses confirmed the multivariable results. After 1:1 PSM (97 matched pairs), SFR was 24.7% versus 16.5% (conditional OR 1.67, 95% CI: 0.81–3.41, P=0.16). Post-matching balance was adequate (SMD ≤0.10 for all covariates). IPTW analysis yielded an OR of 1.49 (95% CI: 0.74–3.01, P=0.27). Firth penalized maximum likelihood estimation, addressing the overfitting concern from fitting 15 parameters to 53 events (EPV ≈3.5), produced consistent results (OR 1.65, 95% CI: 0.81–3.42, P=0.17). All five methods converged on non-significance (Figure 1).

The E-value for the crude OR of 1.91 was 3.23, indicating that an unmeasured confounder would need to be associated with both sheath selection and SFR by an OR of at least 3.2—above and beyond all measured covariates—to fully explain the crude association. The E-value for the fully adjusted model D OR (1.72) was 2.83, and for the Firth-penalized OR (1.65) was 2.69. Because the fully adjusted 95% CI (0.83–3.68) already includes the null, the confidence-interval-based E-value is 1.0, indicating that no bound on unmeasured confounding can be established for the adjusted estimate.

In the fully adjusted model, diabetes was the only variable nominally associated with SFR (OR 3.82, 95% CI: 1.47–9.94, P=0.006), and the Firth-penalized estimate was consistent (OR 3.51, 95% CI: 1.40–8.72, P=0.008). Given the small number of diabetic patients (n=30, 11.4%) and the counterintuitive direction (diabetic patients typically have worse surgical outcomes), this association is likely attributable to sparse-data bias and should not be overinterpreted. The sheath effect was robust to how diabetes was handled: excluding the 30 diabetic patients left the point estimate essentially unchanged and non-significant (intelligent-vs-flexible OR 1.64, 95% CI: 0.75–3.71; Firth OR 1.58, 95% CI: 0.74–3.45), and the sheath × diabetes interaction was not significant (P=0.33).

Secondary outcomes

Operative time was consistently shorter with the flexible sheath across all analytical approaches: crude difference −25.9 min (95% CI: −35.8 to −16.0, P<0.001); PSM paired difference −25.9 min (P<0.001); IPTW weighted difference −27.3 min (P<0.001). The multi-covariate adjusted difference was 27.3 min (P<0.001). To examine whether this difference primarily reflected differential case complexity, we stratified by median stone size (24.0 mm). The operative time difference persisted in both strata: large stones (>24 mm, n=109): −15.3 min (95% CI: −30.3 to −0.4, P=0.045); small stones (≤24 mm, n=155): −33.0 min (95% CI: −45.7 to −20.2, P<0.001). The larger difference in the small-stone stratum argued against the hypothesis that the overall operative time difference was driven by selective deployment of the intelligent sheath in more complex cases; if case complexity were the primary explanation, the difference would be expected to be larger—not smaller—in the large-stone stratum.

A formal mediation analysis [1,000 bootstrap simulations; mediation package (17)] was conducted to determine whether operative time lies on the causal pathway between sheath selection and SFR (Figure S3). The indirect effect via operative time, the average causal mediation effect (ACME), was 0.016 (95% CI: −0.017 to 0.051, P=0.30), and the direct effect was 0.055 (95% CI: −0.054 to 0.162, P=0.34). The proportion mediated was 22.8%. These findings demonstrated that operative time did not mediate the sheath-SFR relationship. Consequently, including operative time as a covariate in regression models constituted overadjustment: the further attenuation when operative time was added to model B (from OR 1.86 to 1.43) reflected adjustment for a post-treatment variable rather than confounding control. We therefore excluded operative time from the primary models (models B–D) and present the operative-time-included versions for transparency only.

Postoperative infection rates did not differ between groups (intelligent 35.4% versus flexible 40.2%; adjusted OR 1.41, 95% CI: 0.79–2.52, P=0.24). However, the overall infection rate of 37.5% reflects a composite definition that includes asymptomatic bacteriuria, and fever and culture-positive subtypes were not recorded separately; a sheath-specific difference in clinically significant infection (e.g., fever ≥38 ℃ or sepsis) therefore could not be isolated, and this comparison should be interpreted cautiously. Postoperative hospital stay was not significantly different after adjustment (−0.42 days, 95% CI: −1.23 to 0.39, P=0.31).

Supplementary meta-analysis summary

The systematic review identified four head-to-head studies (539 patients; Table S4). The meta-analytic displays comprise the same-center exclusion sensitivity analysis (Figure S4), the primary forest plot (Figure S5), the subgroup analysis by SFR definition (Figure S6), and the contour-enhanced funnel plot (Figure S7); all pooled estimates are summarized in Table S5. In the primary random-effects analysis, the pooled OR was 1.29 (95% CI: 0.21–7.94, P=0.69), with substantial heterogeneity (I²=75.7%, 95% CI: 33.0–91.2%). A pre-specified subgroup analysis by SFR definition reduced this heterogeneity: the three CIRF-based studies (threshold <2–3 mm)—all from a single center (Zigong, China)—showed near-zero between-study variance (I²=0.7%) with a pooled OR of 1.90 favoring the flexible sheath, while the present zero-fragment CT-based study (a single study, k=1) showed the opposite direction (OR 0.52). However, SFR definition was completely confounded with study center, so the apparent resolution of heterogeneity cannot be attributed solely to the definitional difference. In a leave-two-of-three same-center sensitivity analysis retaining the present study plus each Zigong study in turn (k=2), the pooled OR ranged from 0.54 to 1.43 and was not statistically significant in two of three combinations, confirming that the meta-analytic estimate is fragile and highly sensitive to which center’s studies are retained. The subgroup difference approached but did not reach statistical significance (P=0.09). The complete methods, results, and discussion of this systematic review, including sensitivity analyses and risk of bias assessment, are presented in Appendix 1 as a standalone synthesis.

Discussion

Key findings

This single-center retrospective study, employing five complementary analytical methods, found no statistically significant difference in SFR between intelligent pressure-controlled and flexible negative-pressure aspiration sheaths. The consistent convergence of multivariable regression, PSM, IPTW, and Firth penalized likelihood on non-significant ORs between 1.49 and 1.72—together with an E-value of 3.23 for the crude association (2.83 for the fully adjusted estimate)—provides a more robust basis for clinical interpretation than the crude association alone would suggest. The supplementary systematic review and meta-analysis (4 studies, 539 patients) provides the first formal synthesis of head-to-head evidence on this comparison, with findings consistent with the single-center results: the pooled effect estimate is imprecise, heterogeneity is high, and no definitive conclusion regarding SFR can be drawn from the existing literature.

We emphasize the distinction between absence of evidence and evidence of absence. The fully adjusted 95% CI (0.83–3.68) is compatible with the intelligent sheath conferring either a clinically meaningful advantage or disadvantage relative to the pre-specified 10-percentage-point non-inferiority margin. The correct interpretation is that the available evidence is insufficient to determine whether SFR differs between sheath types. This finding does not establish equivalence; it establishes uncertainty. This interpretation is consistent with clinical experience: the two sheath types share the same fundamental mechanism of action—negative-pressure aspiration—and differ primarily in adjunctive features (pressure monitoring versus tip flexibility). It is biologically and mechanically plausible that these adjunctive features influence procedural efficiency and safety more than the ultimate outcome of fragment clearance, which is primarily determined by laser lithotripsy completeness, irrigation flow, and stone composition.

Strengths and limitations

Strengths

This study has several methodological strengths. First, the employment of five complementary analytical methods with consistent convergence provides a more robust evidence base than any single analytical approach alone. Second, the formal mediation analysis clarifies the role of operative time as a non-mediator of SFR, with direct implications for covariate selection in observational RIRS research. Third, the pre-specified subgroup analysis in the meta-analysis identified SFR definition as the primary driver of between-study heterogeneity (I² from 75.7% to 0.7%), a finding with implications for the broader endourology literature. Fourth, the integration of GRADE evidence assessment and sample-size benchmarking provides clinicians with transparent estimates of current evidentiary uncertainty and the scale of evidence required for definitive conclusions.

Limitations

Several limitations should be considered. (I) Confounding by indication: the non-randomized, single-center, retrospective design is the most important constraint: despite comprehensive statistical adjustment, the structural imbalance in baseline risk (SMD =0.54 for preoperative infection) means that residual confounding cannot be excluded, and the E-value of 3.23 provides moderate but not definitive protection. (II) Low absolute SFR: the 20.1% SFR reflects a deliberate methodological choice—strictest-possible CT criterion at the earliest time point—and, as discussed below, is consistent with the broader literature. However, it limits statistical power (53 events) and may not reflect the SFR that would be observed at clinically standard 3-month follow-up. (III) Operative time confounding: surgery dates were not recorded in the retrospective dataset, precluding formal learning-curve or temporal-trend analysis. Because the flexible sheath became available later in the study period, part of the observed operative-time difference may reflect operator maturation and workflow refinement over time rather than a device-specific effect. The operative time difference should therefore be regarded as hypothesis-generating. (IV) Meta-analysis fragility: with only four studies (k=4), three of which originated from a single center (Zigong, China), meta-analytic estimates are sensitive to study removal. A leave-two-of-three sensitivity analysis (retaining the present study plus each Zigong study in turn) produced pooled ORs ranging from 0.54 to 1.43 and was non-significant in two of three combinations, demonstrating this fragility (Figure S4). (V) Unmeasured confounders and unmeasured outcomes: stone composition, anatomical complexity metrics (Guy’s stone score, S.T.O.N.E. nephrolithometry, infundibulopelvic angle), and sheath-specific adverse events were not systematically recorded. Intrarenal pressure—the purported core advantage of the intelligent sheath—was not directly measured as an outcome, so any pressure-related benefit remains unverified. In addition, postoperative infection was recorded only as a single composite endpoint, precluding analysis of fever- or culture-defined subtypes. (VI) PROSPERO registration: the systematic review was not prospectively registered. To mitigate potential selective reporting, all analytical methods and the pre-specified SFR-definition subgroup analysis were documented before conducting the review and are publicly available in the analysis repository. (VII) Generalizability: the single-center, single-country design, procedures performed by two surgeons, and 10-day mean hospital stay (reflecting prefectural-level hospital care pathways in western China) may limit applicability to centers with different practice patterns.

Comparison with similar research

The three published head-to-head comparisons (Cui et al. 2024, Ma et al. 2024 RCT, Gong et al. 2025) used CIRF-based SFR definitions (<2–3 mm) and reported SFR in the 75–90% range, substantially higher than the 20.1% SFR in the present cohort (5,6,8). This discrepancy is primarily attributable to the strict zero-fragment CT definition applied at 48 hours postoperatively in the present study, rather than to inferior surgical quality. The Ulvik et al. benchmark study of 1,019 URS procedures reported 54.2% zero-fragment SFR on CT at 3 months (13), declining to 49% for renal stones treated with Ho:YAG laser (14). The 25-center Global FANS study using suction UAS achieved 57.4% zero-fragment SFR on CT within 30 days (15), while the EAU Endourology Section consensus states that true CT-based SFR is “under 50% with FURS without suction” (16). The lower SFR in the present cohort (20.1%) relative to these benchmarks reflects the combined effect of (I) 1-mm CT slice thickness versus standard 3–5 mm; (II) assessment at 48 hours versus 3 months; and (III) mean stone size of 24.5 mm treated in a single-session paradigm. Table S2 provides a systematic comparison across published series.

The supplementary meta-analysis provides critical external context. The three CIRF-based studies, all from a single center (Zigong, China), show consistent direction (pooled OR 1.90 favoring flexible) with near-zero heterogeneity (I²=0.7%). The present zero-fragment CT study shows the opposite direction (OR 0.52). Because SFR definition is completely confounded with study center, the apparent resolution of heterogeneity cannot be uniquely attributed to the definitional difference; both the SFR definition and center-specific patient selection or technique may contribute. This finding underscores the urgent need for a consensus SFR definition—and for multi-center studies—in endourology research.

Explanations of findings

Stone-free rate

The null SFR finding has a biologically plausible explanation. The two sheath types share the same fundamental mechanism—negative-pressure aspiration—and differ in adjunctive features (electronic pressure monitoring versus mechanical tip flexibility). Fragment clearance is primarily determined by laser lithotripsy efficacy, irrigation flow dynamics, and stone composition—factors that are largely independent of sheath type. It is therefore mechanistically expected that the adjunctive features would influence secondary outcomes (operative efficiency, safety monitoring) more than the primary outcome of stone clearance. Notably, the intelligent sheath’s core differentiating feature—real-time intrarenal pressure monitoring—was not directly evaluated in this study, because intrarenal pressure was not measured as an outcome; any safety benefit of pressure control therefore remains mechanistically plausible but unverified.

Operative time

The 26-minute operative time advantage of the flexible sheath was robust to stone-size stratification (larger difference in small-stone stratum: −33.0 versus −15.3 min in large-stone stratum) and adjustment for measured confounders. Potential mechanisms include: (I) the flexible sheath’s steerable tip enabling faster calyceal access; (II) the intelligent sheath’s pressure-monitoring system introducing setup and calibration time; and (III) unmeasured anatomical factors. However, formal mediation analysis demonstrated that operative time does not mediate the sheath-SFR relationship (proportion mediated 22.8%, ACME P=0.30). This null mediation has an important methodological implication: including operative time as a covariate in SFR regression models constitutes overadjustment for a post-treatment variable [a directed acyclic graph of this causal model is shown in Figure S3 (18)]. Models C and D (without operative time, OR ≈1.72–1.74) should be regarded as the primary adjusted estimates. The GRADE certainty for operative time is very low (Table S3), and this finding should be regarded as hypothesis-generating.

Postoperative infection

The 37.5% overall infection rate, which did not differ between sheath types, is consistent with the broad CDC NHSN-based definition (fever ≥38 ℃ or positive urine culture within 30 days). The inclusion of microbiological criteria likely captures asymptomatic bacteriuria, which is common after ureteroscopic instrumentation. The absence of a between-group difference suggests that the intelligent sheath’s theoretical infection-prevention advantage, if it exists, is clinically modest.

Implications and actions needed

Clinical implications

The available evidence does not establish SFR superiority of either sheath type, and the multi-method convergence on non-significance provides only a limited basis for individualized sheath selection. We tentatively propose a three-factor clinical decision framework; because every factor rests on mechanistic reasoning, indirect comparisons, or non-significant observational estimates, this framework is speculative and is not directly supported by comparative outcome data:

Infectious risk (evidence strength: mechanistic, ⊕◯◯◯): for patients with preoperative infection, positive urine culture, or suspected infectious stones, the intelligent sheath’s real-time intrarenal pressure monitoring may provide a safety advantage by maintaining pressure below the pyelovenous reflux threshold. This rationale is mechanistically plausible but has not been tested in comparative studies. Notably, in our data, the 47 patients with preoperative infection who received the flexible sheath did not experience higher infection rates (14.9% versus 21.3%, P=0.41).

Stone anatomy (evidence strength: indirect comparative, ⊕◯◯◯): for patients with lower-pole stones or difficult calyceal anatomy, the flexible sheath’s steerable tip may facilitate direct access. The meta-analytic CIRF subgroup showed a pooled OR of 1.90 favoring the flexible sheath, though the CI was wide (0.11–34.31).

Other cases (evidence strength: convergent observational, ⊕◯◯◯): for patients without strong indications under the above factors, five analytical methods converge on the absence of a clinically meaningful SFR difference. Selection may be based on surgeon experience, device availability, and institutional cost.

The GRADE certainty of evidence is very low for all outcomes (Table S3). Clinicians should discuss this uncertainty with patients when sheath selection involves trade-offs between pressure-monitoring safety and procedural efficiency.

Actions needed

A definitive non-inferiority RCT would require approximately 200–300 patients per arm depending on expected SFR (Appendix 2). Under realistic assumptions [57% SFR with thulium fiber laser (TFL) and suction UAS], approximately 500–670 patients would be needed with dropout inflation—exceeding any published single-center RIRS device comparison. Multicenter collaboration across 5–10 sites could complete enrollment within 2–3 years. We recommend that such a trial: (I) pre-specify a 10-percentage-point non-inferiority margin; (II) employ zero-fragment CT at 3 months as the primary SFR endpoint, with CIRF <2 mm as a secondary endpoint; (III) stratify randomization by stone size, location, and preoperative infection; and (IV) include operative time and postoperative infection as pre-specified secondary endpoints. Standardized SFR reporting—using both zero-fragment and CIRF criteria at clinically meaningful time points—is essential for the meta-analyzability of future endourology research.

Conclusions

Across five complementary analytical methods (multivariable regression, PSM, IPTW, Firth penalized likelihood, and E-value analysis) and an exploratory meta-analytic synthesis (4 studies, 539 patients), the available evidence is insufficient to establish a clinically meaningful difference in SFR between intelligent pressure-controlled and flexible NP-UAS; GRADE certainty of evidence is very low for all outcomes.

The flexible sheath is associated with approximately 26 minutes shorter operative time. This difference persists after stone-size stratification and is not explained by case complexity alone, but it may partly reflect later device availability and operator maturation, formal mediation analysis demonstrates that operative time does not translate into differential stone clearance, and GRADE certainty is very low. The finding is hypothesis-generating and warrants confirmation as a pre-specified endpoint in future randomized comparisons.

SFR definition (zero-fragment CT versus CIRF <2–3 mm) was completely confounded with study center in the available literature, so the apparent resolution of between-study heterogeneity (I² 75.7%→0.7% after stratification) cannot be uniquely attributed to the definitional difference. Standardized SFR reporting—using both zero-fragment and CIRF criteria at clinically meaningful time points, across multiple centers—is essential for the meta-analyzability of future endourology research.

A multicenter non-inferiority RCT with standardized zero-fragment CT SFR at 3 months as the primary endpoint (estimated approximately 500 patients) is required to provide definitive clinical guidance; until then, sheath selection should be guided by individual patient factors and local expertise rather than by evidence of superiority of either device.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Ethics Committee of Xiangxi Tujia and Miao Autonomous People’s Hospital (No. EC-LCKY2026055) and the requirement for informed consent was waived owing to the retrospective, anonymized design of the study.

Data Sharing Statement

Available at https://tau.amegroups.com/article/view/10.21037/tau-2026-0611/dss

tau-15-10-374-dss.pdf ^{(70.7KB, pdf)}

DOI: 10.21037/tau-2026-0611