Impact of operative time on complications after PCNL in the elderly (aged ≥70 years): the Safety and Efficacy of Nephrolithotomy in Older patients undergoing PCNL (SENIOR‐PCNL) study

Abstract

Objectives

To evaluate the association between operative time (OT) and postoperative complications after percutaneous nephrolithotomy (PCNL) in elderly patients, and to identify a clinically meaningful OT threshold for risk stratification.

Patients and Methods

A retrospective international multicentre study was performed including patients aged ≥70 years who underwent PCNL between January 2014 and February 2025 across 21 centres in 15 countries. Postoperative complications within 30 days were analysed as overall, major, and infectious events. The optimal OT threshold was determined using receiver operating characteristic (ROC) analysis, whereas the association between OT and postoperative outcomes was evaluated using multivariable logistic regression and adjusted restricted cubic spline models.

Results

Among 716 patients, 120 (16.8%) developed postoperative complications; 49 (6.8%) had major and 49 (6.8%) infectious complications. ROC analysis identified a common OT threshold around 95 min. An OT ≥95 min was independently associated with overall complications (odds ratio [OR] 3.37, 95% confidence interval [CI] 2.16–5.26), major complications (OR 2.63, 95% CI 1.35–5.10), and infectious complications (OR 3.79, 95% CI 1.90–7.57). The highest discriminative performance was observed for infectious complications (multivariable model area under the curve 0.77). Restricted cubic spline analysis confirmed significant non‐linear associations between OT and all outcomes, with a steeper increase in adjusted complication probability beyond approximately 90–100 min. Non‐papillary puncture was independently associated with all complication types, while positive preoperative urine culture predicted infectious complications.

Conclusions

In elderly patients undergoing PCNL, prolonged OT was independently associated with higher odds of overall, major, and infectious postoperative complications. The relationship was non‐linear, with a clinically relevant threshold around 95 min, supporting OT as a practical intraoperative marker for risk stratification and procedural planning.

Abbreviations

AUC

area under the curve

EAU

European Association of Urology

ECIRS

endoscopic combined intrarenal surgery

IQR

interquartile range

OR

odds ratio

OT

operative time

PCNL

percutaneous nephrolithotomy

(N)(P)PV

(negative) (positive) predictive value

ROC

receiver operating characteristic

SFR

stone‐free rate

SIRS

systemic inflammatory response syndrome

Introduction

Percutaneous nephrolithotomy (PCNL) remains the standard treatment for large and complex renal calculi, offering high stone‐free rates (SFRs) despite a non‐negligible risk of serious postoperative adverse events [1]. Accordingly, the identification of perioperative factors influencing procedural risk remains clinically important. Operative time (OT) is of particular interest because, unlike many patient‐ and stone‐related characteristics, it may represent a partly modifiable procedural factor. Growing evidence from adult and paediatric populations suggests that prolonged OT is associated with worse outcomes after PCNL, including major and infectious complications [2, 3]. However, the ‘safe’ OT thresholds proposed in the literature remain inconsistent. In addition, recent studies suggest that the relationship between OT and postoperative risk may not be purely linear [2].

The ongoing global trend toward population ageing has resulted in an increasing number of elderly patients requiring active surgical treatment for kidney stones, including PCNL [4, 5]. Because elderly patients often present with a greater burden of comorbidity and reduced physiological reserve, they may be particularly vulnerable to perioperative stress and postoperative complications [6]. In this context, OT may represent an especially relevant procedural factor, and the identification of clinically meaningful safety thresholds appears important for risk stratification and treatment planning. However, data specifically addressing this issue in elderly patients remain scarce, with a particular paucity of robust international multicentre evidence.

Therefore, the aim of the present multicentre study was to evaluate the association between OT and postoperative complications after PCNL in elderly patients and to identify a clinically meaningful OT threshold for risk stratification.

Patients and Methods

Retrospective multicentre data were collected for the Safety and Efficacy of Nephrolithotomy in Older patients undergoing PCNL (SENIOR‐PCNL) study. Patients underwent PCNL for renal or upper ureteric calculi between January 2014 and February 2025 across 21 centres from 15 countries participating in the international European Association of Urology‐Young Academic Urologists (EAU‐YAU) Endourology collaborative network. Data were collected using a standardised spreadsheet template circulated to all participating centres together with written definitions for each variable, in order to ensure uniform interpretation across sites. Records were reviewed and entered exclusively by the co‐authors of the present study, all of whom are practising urologists at the participating institutions. Completed datasets were returned to the coordinating centre (Wroclaw Medical University), where they were merged and screened centrally for missing values, out‐of‐range entries, and internal inconsistencies.

Patients were excluded a priori if they had anatomical or functional abnormalities of the urinary tract predisposing to altered infectious or procedural risk, namely neurogenic bladder, urinary diversion or bladder augmentation, renal transplantation, and congenital renal anomalies including horseshoe, ectopic, or malrotated kidney. All centres were high‐volume institutions with established experience in endourological stone surgery. Ethical or institutional approval was obtained according to local regulations at each participating centre, and all data were anonymised prior to central analysis. To avoid within‐patient dependence arising from repeated procedures, only one procedure per patient was included.

Despite varying literature‐based thresholds used to define elderly populations, patients aged ≥70 years were considered elderly in the present study, as this cut‐off is consistent with contemporary population‐level ageing analyses and better reflects the stage at which comorbidity burden and reduced physiological reserve become clinically relevant [7]. Recorded variables included demographics, comorbidity burden, stone characteristics, preoperative urine culture status, preoperative hydronephrosis, preoperative urinary drainage, and other relevant perioperative parameters. Stone burden was assessed using the single largest stone diameter, cumulative stone diameter, and estimated stone volume calculated with the ellipsoid formula (length × width × depth × π × 0.167). Cumulative stone diameter was calculated as the sum of the maximal diameters of all stones identified on preoperative imaging. Procedural variables included patient positioning, tract size, lithotripsy device, and exit strategy. OT was recorded as the interval from renal puncture to completion of the procedure, including nephrostomy tube placement, ureteric stenting, or tubeless tract closure according to the selected exit strategy. Stone composition analysis was performed, where available, according to local institutional protocols. Owing to the retrospective multicentre design, analytical methods were not standardised across centres, and stones were classified according to their predominant component.

The SFR was assessed according to local imaging protocols, including CT, plain radiography of the kidney–ureter–bladder, and/or ultrasonography. In view of the retrospective multicentre design and expected variability in imaging follow‐up and reporting across centres, only zero residual fragment status on postoperative imaging was considered stone‐free for the purpose of analysis.

The primary outcomes were postoperative complications within 30 days. Complications were graded according to the modified Clavien–Dindo classification validated for PCNL [8] and analysed as overall (Clavien–Dindo Grade I–V), major (Clavien–Dindo Grade III–V), and infectious complications. Infectious complications were defined as postoperative UTI or sepsis according to standard clinical and laboratory criteria.

Statistical Analysis

All statistical analyses were performed using R software, version 4.5.2 (R Foundation for Statistical Computing, Vienna, Austria).

Normality of continuous variables was assessed using the Shapiro–Wilk test. Because all continuous variables were not normally distributed, they were summarised as median and interquartile range (IQR) and compared using the Mann–Whitney U test. Categorical variables were presented as counts and percentages and compared using the chi‐square test or Fisher's exact test, as appropriate.

The discriminatory ability of OT for each outcome was assessed using receiver operating characteristic (ROC) curve analysis. The area under the curve (AUC) with 95% CIs was calculated, and the optimal cut‐off value was determined by maximising the Youden index. Sensitivity, specificity, positive predictive value (PPV), and negative PV (NPV) were also determined. To assess the stability of the identified threshold, bootstrap re‐sampling was performed.

Associations between OT and postoperative complications were assessed using logistic regression models. Univariable logistic regression was first performed for each outcome. Subsequently, multivariable logistic regression models were constructed to estimate adjusted odds ratios (ORs) with 95% CIs, including variables with a P ≤ 0.1 in univariable analyses. OT was examined both as a continuous variable and as a categorical variable according to the ROC‐derived threshold. The discriminative performance of the multivariable models was further evaluated by calculating AUCs based on model‐predicted probabilities.

Given the potentially non‐linear relationship between OT and postoperative complication risk, adjusted restricted cubic spline models were fitted, with OT entered as a continuous variable using a four‐knot restricted cubic spline. For each outcome, spline models were adjusted for the same covariates as the corresponding multivariable logistic regression models. Overall association and departure from linearity were assessed by analysis of variance of the fitted models, with the P value for the non‐linear component used as the formal test for non‐linearity. For graphical presentation, adjusted predicted probabilities and adjusted ORs relative to the 95‐min threshold were estimated, together with their 95% CIs.

All tests were two‐sided, and a P < 0.05 was considered statistically significant.

Results

Among 716 elderly patients undergoing PCNL, 120 (16.8%) experienced at least one postoperative complication type. Table 1 presents the baseline demographic, clinical, stone‐related, and perioperative characteristics of the study cohort, stratified by overall, major, and infectious complications. In total, 136 Clavien–Dindo Grade I–V complication events were recorded (Table S1A), indicating that several patients developed more than one complication. Major and infectious complications were each observed in 49 (6.8%) patients. Patients with overall, major, and infectious complications all had significantly longer OT (median [IQR] 112.5 [85–131.25] vs 80 [55–120] min, P < 0.001; 102 [80–130] vs 86 [55–120] min, P = 0.007; and 118 [90–135] vs 85 [55–120] min, P < 0.001, respectively), longer hospital stay, and lower SFR (Table 1).

Table 1.

Baseline demographic, stone‐related, and perioperative characteristics stratified by overall, major, and infectious complications.

Variable	Overall complications			Major complications			Infectious complications
	No (n = 596)	Yes (n = 120)	P	No (n = 667)	Yes (n = 49)	P	No (n = 667)	Yes (n = 49)	P
Age, years, median (IQR)	74 (71–77)	75 (72–77.25)	0.020	74 (71–77)	75 (72–77)	0.209	74 (71–77)	75 (73–77)	0.095
Sex, male n (%)	296 (49.7)	68 (56.7)	0.193	330 (49.5)	34 (69.4)	0.008	336 (50.4)	28 (57.1)	0.378
BMI, kg/m2, median (IQR)	27 (24.26–29.6)	26.7 (24.6–29)	0.644	26.92 (24.3–29.4)	27 (24.2–29.1)	0.916	27 (24.3–29.4)	26 (24.3–29.3)	0.736
ASA grade, n (%)
I	82 (13.8)	7 (5.8)	0.035	86 (12.9)	3 (6.1)	0.401	84 (12.6)	5 (10.2)	0.440
II	304 (51.0)	66 (55.0)		345 (51.7)	25 (51.0)		346 (51.9)	24 (49.0)
III	196 (32.9)	41 (34.2)		218 (32.7)	19 (38.8)		220 (33.0)	17 (34.7)
IV	14 (2.3)	6 (5.0)		18 (2.7)	2 (4.1)		17 (2.5)	3 (6.1)
ACCI score, median (IQR)	3 (3–4)	3 (3–4.75)	0.933	3 (3–4)	3 (3–5)	0.247	3 (3–4)	3 (2–5)	0.487
AC/AP drug use, yes, n (%)	182 (30.5)	40 (33.3)	0.589	209 (31.3)	13 (26.5)	0.526	208 (31.2)	14 (28.6)	0.752
Hypertension, yes, n (%)	345 (57.9)	69 (57.5)	1	385 (57.7)	29 (59.2)	0.882	388 (58.2)	26 (53.1)	0.549
Diabetes mellitus, yes, n (%)	205 (34.4)	36 (30.0)	0.397	226 (33.9)	15 (30.6)	0.755	226 (33.9)	15 (30.6)	0.755
Preoperative urine culture, positive, n (%)	218 (36.6)	47 (39.2)	0.605	247 (37.0)	18 (36.7)	1	237 (35.5)	28 (57.1)	0.003
Duration of preoperative antibiotic therapy in patients with a positive urine culture, days, median (IQR)	2 (1–3)	2 (1–4)	0.438	2 (1–3)	2 (1–4)	0.517	2 (1–3)	2 (2–4)	0.184
Preoperative hydronephrosis, yes, n (%)	253 (42.4)	57 (47.5)	0.314	292 (43.8)	18 (36.7)	0.373	287 (43.0)	23 (46.9)	0.655
Preoperative JJ/nephrostomy stent, yes, n (%)	197 (33.1)	49 (40.8)	0.102	225 (33.7)	21 (42.9)	0.194	230 (34.5)	16 (32.7)	0.795
Single largest stone size, mm, median (IQR)	22 (17–30)	27 (17–32.5)	0.024	23 (17–30)	25 (16–32)	0.895	22.17 (17–30)	27 (19–35)	0.015
Stone density, Hounsfield units, median (IQR)	1035 (716–1312)	1003 (733.5–1256)	0.394	1031 (716–1300)	1047 (733.5–1384)	0.613	1035 (742–1305.75)	991 (541–1300)	0.260
Cumulative stone size, mm, median (IQR)	32 (26–43)	36 (29–50)	0.106	32 (26.375–44)	37.5 (29.35–50.75)	0.106	32 (26–43.25)	48.75 (32.25–55.75)	0.006
Stone volume, mm3, median (IQR)	2270 (1155–4976)	3011 (1269–5759)	0.163	2356 (1224–5041)	2368 (1011–5616)	0.822	2288 (1154–5041)	3510 (1655–8086)	0.020
Multiple stones, yes, n (%)	271 (45.5)	44 (36.7)	0.087	295 (44.2)	20 (40.8)	0.658	301 (45.1)	14 (28.6)	0.026
Staghorn stone, yes, n (%)	144 (24.2)	43 (35.8)	0.012	168 (25.2)	19 (38.8)	0.043	166 (24.9)	21 (42.9)	0.010
Operative position, n (%)
Supine	476 (79.9)	103 (85.8)	0.161	540 (81.0)	39 (79.6)	0.851	536 (80.4)	43 (87.8)	0.260
Prone	120 (20.1)	17 (14.2)		127 (19.0)	10 (20.4)		131 (19.6)	6 (12.2)
Tract size, n (%)
Mini PCNL (12–22 F)	389 (65.3)	69 (57.5)	0.118	424 (63.6)	34 (69.4)	0.445	427 (64.0)	31 (63.3)	1
Standard PCNL (24–30 F)	207 (34.7)	51 (42.5)		243 (36.4)	15 (30.6)		240 (36.0)	18 (36.7)
Puncture site, non‐papillary, n (%)	56 (9.4)	21 (17.5)	0.014	66 (9.9)	11 (22.4)	0.014	66 (9.9)	11 (22.4)	0.014
ECIRS, yes, n (%)	192 (32.2)	43 (35.8)	0.457	212 (31.8)	23 (46.9)	0.039	217 (32.5)	18 (36.7)	0.533
Fragmentation device, n (%)
Laser	353 (59.2)	58 (48.3)	0.058	388 (58.2)	23 (46.9)	0.183	386 (57.9)	25 (51.0)	0.124
Pneumatic/ultrasonic	231 (38.8)	58 (48.3)		265 (39.7)	24 (49.0)		268 (40.2)	21 (42.9)
Other	12 (2.0)	4 (3.3)		14 (2.1)	2 (4.1)		13 (1.9)	3 (6.1)
Exit strategy, n (%)
Totally tubeless	29 (4.9)	4 (3.3)	<0.001	31 (4.6)	2 (4.1)	0.405	31 (4.6)	2 (4.1)	0.019
JJ stent only	295 (49.5)	40 (33.3)		314 (47.1)	21 (42.9)		319 (47.8)	16 (32.7)
Nephrostomy tube only	111 (18.6)	19 (15.8)		124 (18.6)	6 (12.2)		124 (18.6)	6 (12.2)
Both JJ stent and nephrostomy tube	161 (27.0)	57 (47.5)		198 (29.7)	20 (40.8)		193 (28.9)	25 (51.0)
OT, min, median (IQR)	80 (55–120)	112.5 (85–131.25)	<0.001	86 (55–120)	102 (80–130)	0.007	85 (55–120)	118 (90–135)	<0.001
Length of hospital stay, days, median (IQR)	3 (2–4)	5 (3.75–8)	<0.001	3 (2–4)	5 (3–9)	<0.001	3 (2–4)	7 (4–10)	<0.001
SFR, n (%)	453 (76.0)	75 (62.5)	0.003	499 (74.8)	29 (59.2)	0.027	498 (74.7)	30 (61.2)	0.044
Postoperative imaging, n (%)
CT	364 (61.1)	74 (61.7)	0.903	404 (60.6)	34 (69.4)	0.222	408 (61.2)	30 (61.2)	0.994
KUB and/or ultrasonography	232 (38.9)	46 (38.3)		263 (39.4)	15 (30.6)		259 (38.8)	19 (38.8)

Continuous variables are presented as median (IQR).

AC/AP, anticoagulant/antiplatelet; ACCI, age‐adjusted Charlson Comorbidity Index; ASA, American Society of Anaesthesiologists; BMI, body mass index; ECIRS, endoscopic combined intrarenal surgery; KUB, plain radiography of the kidney–ureter–bladder.

Statistically significant results are presented in bold.

As a single predictor, OT demonstrated modest discrimination for overall, major, and infectious complications, with AUCs of 0.64, 0.62, and 0.65, respectively. The optimal Youden‐derived cut‐offs were highly similar across endpoints, ranging from 93 to 95 min, with sensitivities of 0.67–0.71 and specificities of 0.58–0.61 (Fig. 1). The PPVs were low across all outcomes (10.6–26.2%), whereas the NPVs were consistently high (90.6–96.7%). Detailed ROC metrics are provided in Table S2. Bootstrap analyses across all three outcomes consistently indicated an optimal OT threshold between 90 and 100 min; therefore, a common cut‐off of 95 min was selected for subsequent analyses (Fig. S1A). After stratification of the cohort by the pre‐defined 95‐min OT threshold (Table S3), procedures lasting ≥95 min (n = 311 [43.4%]) were characterised by greater stone and procedural complexity, including larger stone size (26 vs 21 mm, P < 0.001), higher stone volume (3290 vs 1900 mm³, P < 0.001), more frequent staghorn calculi (38.9% vs 16.3%, P < 0.001), and higher endoscopic combined intrarenal surgery (ECIRS) use (44.1% vs 24.2%, P < 0.001).

Fig. 1.

The ROC curves for OT as a predictor of postoperative complications after PCNL. (A) For overall complications. (B) For major complications. (C) For infectious complications. For each outcome, the AUC and the optimal Youden‐derived cut‐off are displayed.

Results of the univariable logistic regression analyses are presented in Table S4. Detailed results of all multivariable models are provided in Table 2. An OT ≥95 min was associated with increased odds of overall complications (OR 3.37, 95% CI 2.16–5.26; P < 0.001), together with non‐papillary puncture (OR 2.14, 95% CI 1.19–3.85; P = 0.018). In the model for major complications, an OT ≥95 min (OR 2.63, 95% CI 1.35–5.10; P = 0.004), male sex (OR 2.21, 95% CI 1.16–4.21; P = 0.015), and non‐papillary puncture (OR 2.76, 95% CI 1.30–5.86; P = 0.008) remained significantly associated. In the model for infectious complications, significant associations were observed for an OT ≥95 min (OR 3.79, 95% CI 1.90–7.57; P < 0.001), positive preoperative urine culture (OR 2.40, 95% CI 1.30–4.42; P = 0.005), and non‐papillary puncture (OR 3.35, 95% CI 1.56–7.19; P = 0.001). Compared with OT analysed as single parameter, the multivariable models provided better discrimination, with AUCs of 0.70, 0.73, and 0.77 for overall, major, and infectious complications, respectively (Fig. S2).

Table 2.

Multivariable logistic regression analysis for overall, major, and infectious complications using OT categorised at the 95‐min threshold.

	OR (95% CI)	P
Overall complications
OT ≥95 min (<95 min as reference)	3.37 (2.16–5.26)	<0.001
Age (per 1 year increase)	1.04 (0.99–1.09)	0.053
Single largest stone diameter (per 10 mm increase)	0.99 (0.80–1.22)	0.948
Preoperative JJ/nephrostomy stent, yes (no as reference)	1.38 (0.90–2.11)	0.130
Multiple stones, yes (no as reference)	0.74 (0.44–1.25)	0.376
Staghorn stone, yes (no as reference)	1.08 (0.60–1.95)	0.475
Tract size, mini‐PCNL (standard‐PCNL as reference)	0.93 (0.56–1.55)	0.949
Puncture site, non‐papillary (papillary as reference)	2.14 (1.19–3.85)	0.018
Fragmentation device, pneumatic/ultrasonic (laser as reference)	1.47 (0.89–2.44)	0.155
Major complications
OT ≥95 min (<95 min as reference)	2.63 (1.35–5.10)	0.004
Sex, male (female as reference)	2.21 (1.16–4.21)	0.015
Staghorn stone, yes (no as reference)	1.36 (0.72–2.60)	0.336
Puncture site, non‐papillary (papillary as reference)	2.76 (1.30–5.86)	0.008
ECIRS, yes (no as reference)	1.45 (0.78–2.67)	0.230
Fragmentation device, pneumatic/ultrasonic (laser as reference)	1.49 (0.81–2.74)	0.195
Infectious complications
OT ≥95 min (<95 min as reference)	3.79 (1.90–7.57)	<0.001
Single largest stone diameter (per 10 mm increase)	1.05 (0.80–1.39)	0.702
Preoperative urine culture, positive (negative as reference)	2.40 (1.30–4.42)	0.005
Multiple stones, yes (no as reference)	0.64 (0.29–1.40)	0.266
Staghorn stone, yes (no as reference)	1.10 (0.48–2.54)	0.808
Puncture site, non‐papillary (papillary as reference)	3.35 (1.56–7.19)	0.001

Statistically significant results are presented in bold.

In supplementary multivariable logistic regression analyses with OT entered as a continuous variable, OT was significantly associated with overall complications, but not with major or infectious complications (Table S5).

Adjusted restricted cubic spline analyses showed significant associations between OT and overall complications (overall P < 0.001), major complications (overall P = 0.029), and infectious complications (overall P < 0.001). For each outcome, the association was non‐linear (P for non‐linearity <0.05). The spline curves (Fig. 2A,C,E) indicated that the adjusted probability of complications increased with longer OT, with a steeper rise beyond approximately 90–100 min. Using 95 min as the reference value, the spline‐based OR curves (Fig. 2B,D,E) showed a consistent directional pattern across all outcomes, remaining below 1 at OTs shorter than 95 min and above 1 at longer OTs. All other covariates identified as significant in the corresponding multivariable logistic regression models remained significant in the spline‐adjusted analyses.

Fig. 2.

Adjusted restricted cubic spline curves for the association between OT and postoperative complications. Solid lines represent spline‐based estimates, shaded areas indicate 95% CIs, and the red dashed line marks the 95‐min OT threshold. (A) Adjusted predicted probability of overall complications. (B) Adjusted odds ratio for overall complications, using 95 min as the reference operative time. (C) Adjusted predicted probability of major complications. (D) Adjusted OR for major complications, using 95 min as the reference OT. (E) Adjusted predicted probability of infectious complications. (F) Adjusted OR for infectious complications, using 95 min as the reference OT.

Discussion

In this international multicentre cohort of elderly patients undergoing PCNL, OT emerged as a clinically relevant procedural factor associated with postoperative morbidity. Longer procedures were associated with higher odds of overall, major, and infectious complications, and a threshold around 95 min appeared to differentiate lower‐ from higher‐risk cases. The relatively high NPV of this cut‐off suggests that procedures completed within this time frame were generally associated with a low likelihood of postoperative complications.

As OT is defined inconsistently across the PCNL literature, ranging from puncture‐based measures limited to the percutaneous phase, through definitions extending to nephrostomy placement or skin closure, to broader measures that also include retrograde steps or even total operating‐room time [9, 10, 11]. In the present study, OT was recorded as the interval from renal puncture to completion of the procedure according to the selected exit strategy. Although this differs from broader definitions that also include preoperative or non‐surgical procedural components, it more specifically captures the active surgical phase of the PCNL and reduces the influence of non‐operative time elements that are less directly related to procedural complexity and postoperative outcomes.

In our cohort, OT showed the strongest independent association with infectious complications. In multivariable analysis, OT ≥95 min was associated with nearly fourfold higher odds of infectious events (OR 3.79). This finding was further supported by the highest discriminative performance of the final multivariable model (AUC 0.77). When analysed as an isolated predictor, OT also achieved its best ROC performance for this endpoint (AUC 0.65), with 71.4% sensitivity, 60.9% specificity, and a NPV of 96.7%. Notably, the threshold we identified converges closely with independent estimates derived in unrelated cohorts. In a single‐centre series of 802 patients with complex renal calculi, an operating time ≥100 min emerged as an independent predictor of post‐PCNL sepsis (OR 4.4) alongside positive stone culture [12]. Similarly, a recent sex‐stratified analysis identified an OT threshold of 96 min for infectious complications in women, although a considerably longer threshold of 137 min in men [13]. The consistency of these independently derived cut‐offs in the 95–100 min range, despite differing populations, endpoints and analytical approaches, lends external support to the threshold observed in our elderly cohort.

A plausible explanation is that longer PCNL may increase cumulative exposure to sustained renal pelvic pressure, pyelovenous backflow, and irrigation‐fluid absorption, thereby facilitating the systemic translocation of bacteria and endotoxins released during stone manipulation [14, 15]. In turn, these inflammatory triggers may accumulate in a threshold‐dependent manner, which could help explain the observed non‐linear relationship between OT and infectious complications. Support for this interpretation comes from the recent paediatric study by Abudurexiti et al. [2]. Notably, this is one of the very few PCNL studies to specifically evaluate non‐linearity and threshold effects. Using restricted cubic spline and two‐piecewise regression analyses, the authors showed a non‐linear association between OT and post‐PCNL systemic inflammatory response syndrome (SIRS) with an inflection point at 70 min [2]. Our findings are broadly consistent with these results, as our spline regression analysis confirmed that the effect of OT was not purely linear, with a steeper increase in the adjusted probability of infectious complications beyond approximately 90–100 min.

It has to be emphasised that intraoperative stone cultures and renal pelvic urine cultures were not systematically obtained across the participating centres. Stone composition analysis was available in 286 patients (39.9%; Table S6); however, because analytical methods and reporting practices were not standardised across centres and data were unavailable for most of the cohort, stone composition was assessed descriptively and was not included in the multivariable analyses. Although a positive preoperative bladder urine culture was independently associated with infectious complications in our cohort, bladder urine provides an incomplete representation of the upper‐tract microbiological environment. Prospective studies have demonstrated that stone and renal pelvic urine cultures more accurately identify the organisms associated with post‐PCNL infectious complications, while discordance between preoperative bladder urine and stone cultures is common [16, 17]. These findings have subsequently been confirmed in systematic reviews and meta‐analyses showing that stone culture is more informative than preoperative midstream urine culture for predicting postoperative sepsis [18]. Current EAU Guidelines therefore strongly recommend obtaining a stone culture or urine culture directly from the renal pelvis at the time of PCNL whenever feasible [1]. The absence of these data prevented us from assessing culture discordance and from determining whether the association between prolonged OT and infectious complications was mediated by an increased intraoperative microbial burden; therefore, this proposed mechanism remains inferential rather than directly demonstrated in our cohort.

Importantly, perioperative antibiotic protocols were not standardised across participating centres. Among patients with a positive preoperative urine culture, the median duration of antibiotic therapy was 2 days and did not differ significantly according to the occurrence of overall complications (median [IQR] 2 [1–3] vs 2 [1–4] days; P = 0.438), major complications (median [IQR] 2 [1–3] vs 2 [1–4] days; P = 0.517), or infectious complications (median [IQR] 2 [1–3] vs 2 [2–4] days; P = 0.184). This suggests that differences in treatment duration alone are unlikely to fully explain the observed variation in postoperative infectious outcomes, although residual confounding related to antibiotic selection, timing, susceptibility‐guided therapy, and centre‐specific protocols cannot be excluded. Current evidence indicates that extended preoperative therapy may be beneficial mainly in higher‐risk patients, but findings remain inconsistent. The 7‐day regimens have been associated with lower rates of SIRS or sepsis in some trials and meta‐analyses, particularly in patients with positive urine cultures or indwelling drainage, whereas other randomised studies found no advantage over shorter treatment [19, 20, 21, 22]. Therefore, inter‐centre variation in perioperative antibiotic management may still have contributed to residual heterogeneity in infectious outcomes.

We also found that an OT ≥95 min was associated with slightly weaker, yet still consistent, increases in the odds of overall (OR 3.37) and major (OR 2.63) complications. This may partly reflect cumulative renal parenchymal and vascular trauma during PCNL. Our ‘safe’ threshold appears less conservative than that reported by Pakdel et al. [3], who found associations between longer OT and overall and major complications (but not SIRS) and suggested lower cut‐offs of 65 and 55 min, respectively. In contrast, Liu et al. [23], in an elderly cohort, found that an OT >60 min was not independently associated with postoperative complications Thus, based on our findings, excessively conservative OT thresholds do not necessarily need to be applied even in a more vulnerable geriatric population; however, the acceptable procedure duration should always be individualised according to the patient's clinical condition, frailty status, comorbidity burden, and overall operative complexity.

We also identified non‐papillary puncture as additional risk factor for all types of analysed complications. Although non‐papillary access is often considered safe in the general PCNL literature [24], our findings suggest that in elderly patients it may be associated with less favourable outcomes, possibly due to lower physiological reserve and greater susceptibility to parenchymal trauma. In addition, positive preoperative urine culture remained associated with a higher risk of postoperative infectious complications, underscoring the importance of careful preoperative microbiological assessment and management in this population.

The OT obviously reflects stone and procedural complexity, which was also evident in our cohort (Table S3). Cases with an OT ≥95 min were characterised by greater stone volume and higher prevalence of staghorn calculi, which is consistent with previous literature [25]. However, because OT remained independently associated with postoperative complications after adjustment in our multivariable models, it seems to represent more than a simple surrogate of case complexity. Rather, particularly in higher‐risk populations such as elderly patients, OT may serve as a practical intraoperative warning sign. From a technical perspective, this supports efforts to minimise procedural duration whenever feasible, including appropriate case selection and allocation to experienced surgeons, especially given evidence that OT is prolonged during the learning curve [26]. Likewise, in patients with highly complex stone burden, a staged approach may be worth considering when further attempts at complete clearance are expected to substantially prolong the procedure, as previous studies have shown that staged PCNL can achieve comparable stone‐free outcomes while reducing bleeding and infectious complication rates [27, 28]. However, in elderly patients any potential benefit of limiting prolonged single‐session operative stress must still be balanced against the risks of repeated anaesthesia.

Of note, our complication rates merit comparison with published benchmarks. The overall complication rate of 16.8% was comparable to the 20.5% reported in the prospective Clinical Research Office of the Endourological Society (CROES) PCNL Global Study [29], whereas major complications (6.8%) and sepsis (3.1%) exceeded the pooled estimates of 5.2% and 0.6% reported for elderly patients in a recent systematic review and meta‐analysis [30]. This is consistent with the age‐stratified findings of that analysis, in which studies applying a 70‐year threshold (as in the present cohort) demonstrated a significantly higher risk of major complications than those using 60‐ or 65‐year definitions (OR 1.95, 95% CI 1.15–3.31) [30]. The transfusion rate of 4.9% fell within the contemporary range of approximately 5–7% and plausibly reflects the predominance of miniaturised access (mini‐PCNL in 64%), supine positioning in most cases, and the exclusively high‐volume nature of the participating centres, all of which have been associated with lower bleeding‐related morbidity [31].

From a practical perspective, the approximately 95‐min threshold should not be interpreted as a rigid stopping rule, but rather as an intraoperative checkpoint prompting reassessment of the balance between further stone clearance and cumulative procedural risk. The decision to continue or stage the procedure should remain individualised and should incorporate residual stone burden, proximity to completion, bleeding and visibility, haemodynamic and anaesthetic stability, microbiological risk, comorbidity, frailty, and the expected duration of additional manoeuvres. Staging may be reasonable when substantial residual treatment remains and further attempts at clearance are expected to prolong the procedure considerably, whereas continuation may be appropriate when completion is imminent and the patient remains stable. Thus, OT should complement, rather than replace, the responsible surgeon's clinical judgement.

The present study has several important strengths. It represents a large international multicentre cohort focused exclusively on patients aged ≥70 years, a population that remains underrepresented in the PCNL literature. Moreover, the association between OT and postoperative morbidity was evaluated across three clinically relevant endpoints using complementary analytical approaches, including multivariable regression, bootstrap assessment of threshold stability, and restricted cubic spline modelling. Several limitations should nevertheless be acknowledged. First, the retrospective design introduces the possibility of selection bias, incomplete data capture, and residual confounding. Second, perioperative management, surgical technique, antimicrobial protocols, and reporting practices were not fully standardised across centres. Unmeasured centre‐level effects may therefore persist, while the inclusion of exclusively high‐volume institutions may limit generalisability to lower‐volume or less specialised settings. Third, patients with urinary tract abnormalities were excluded during local data collection, but the number of excluded patients was not recorded centrally. The findings should therefore not be extrapolated to patients with neurogenic bladder, urinary diversion, or congenital renal anomalies. Fourth, intraoperative stone and renal pelvic urine cultures were not systematically obtained, preventing assessment of their concordance with preoperative bladder urine cultures and limiting evaluation of the microbiological mechanisms underlying infectious complications. Perioperative antibiotic regimens were also not standardised, and differences in antibiotic selection, susceptibility‐guided treatment, and timing may have contributed to heterogeneity in infectious outcomes. Fifth, relevant perioperative laboratory variables, including pre‐ and postoperative haemoglobin levels and renal function parameters such as estimated GFR, were not uniformly available. Consequently, we could not evaluate baseline anaemia, perioperative haemoglobin decline, renal function changes, or their contribution to transfusion requirements and postoperative outcomes. Finally, data on individual operator experience and trainee involvement were not captured. As OT may be influenced by surgeon volume and position on the learning curve, residual confounding by operator experience cannot be excluded, despite all procedures being performed at experienced high‐volume institutions.

Conclusions

In elderly patients undergoing PCNL, prolonged OT was independently associated with a higher risk of overall, major, and infectious postoperative complications. Our findings further suggest that this relationship is not simply linear, but follows a non‐linear pattern with a clinically relevant threshold effect around 95 min and a more pronounced increase in risk beyond approximately 90–100 min. As OT is one of the few potentially modifiable procedural factors, these results may have practical implications for intraoperative decision‐making and risk stratification. Although the retrospective multicentre design warrants cautious interpretation, this study provides robust international evidence supporting OT as a meaningful marker of postoperative risk in elderly patients undergoing PCNL.

Author Contributions

The authors listed below have made substantial contributions to the intellectual content of the paper in the various sections described. Conceptualisation: Łukasz Nowak, Bhaskar Somani, Begoña Ballesta Martinez. Methodology: Łukasz Nowak, Bhaskar Somani, Begoña Ballesta Martinez. Software: Łukasz Nowak. Validation: Łukasz Nowak, Bhaskar Somani, Begoña Ballesta Martinez. Formal analysis: Łukasz Nowak. Investigation: Łukasz Nowak, Bhaskar Somani, Wojciech Krajewski, Tomasz Szydełko, Vineet Gauhar, Steffi Kar Kei Yuen, Matthias Boeykens, Thomas Tailly, Ioannis Mykoniatis, Alba Sierra, Emre T. Sener, Vincent De Coninck, Rifat Burak Ergul, Alberto Olivero, Lazaros Tzelves, Francesco Esperto, Angelis Peteinaris, Yuigi Yuminaga, Qi Hou, Patrick Juliebø‐Jones, Arman Tsaturyan, Bartosz Małkiewicz, Agata Góral, Paulina Biesiadecka, Francisco Javier Lopez Vivo, Mantu Gupta, Juan Serna, Ziv Savin, Anastasios Anastasiadis, Dimitrios Memmos, Charlotte Slots, Rita Pages, Ceyhun Ozdemiroglu, Maria Florencia Frascheri, Rafael Carrasco Moreno, Eugenio Ventimiglia, Panagiotis Kallidonis, Evangelos Liatsikos, Otas Durutovic, Amelia Pietropaolo, Begoña Ballesta Martinez. Resources: Łukasz Nowak, Bhaskar Somani, Wojciech Krajewski, Tomasz Szydełko, Vineet Gauhar, Steffi Kar Kei Yuen, Matthias Boeykens, Thomas Tailly, Ioannis Mykoniatis, Alba Sierra, Emre T. Sener, Vincent De Coninck, Rifat Burak Ergul, Alberto Olivero, Lazaros Tzelves, Francesco Esperto, Angelis Peteinaris, Yuigi Yuminaga, Qi Hou, Patrick Juliebø‐Jones, Arman Tsaturyan, Bartosz Małkiewicz, Agata Góral, Paulina Biesiadecka, Francisco Javier Lopez Vivo, Mantu Gupta, Juan Serna, Ziv Savin, Anastasios Anastasiadis, Dimitrios Memmos, Charlotte Slots, Rita Pages, Ceyhun Ozdemiroglu, Maria Florencia Frascheri, Rafael Carrasco Moreno, Eugenio Ventimiglia, Panagiotis Kallidonis, Evangelos Liatsikos, Otas Durutovic, Amelia Pietropaolo, Begoña Ballesta Martinez. Data Curation: Łukasz Nowak, Begoña Ballesta Martinez. Writing – Original Draft: Łukasz Nowak, Begoña Ballesta Martinez. Writing – Review and Editing: Łukasz Nowak, Bhaskar Somani, Wojciech Krajewski, Tomasz Szydełko, Vineet Gauhar, Steffi Kar Kei Yuen, Matthias Boeykens, Thomas Tailly, Ioannis Mykoniatis, Alba Sierra, Emre T. Sener, Vincent De Coninck, Rifat Burak Ergul, Alberto Olivero, Lazaros Tzelves, Francesco Esperto, Angelis Peteinaris, Yuigi Yuminaga, Qi Hou, Patrick Juliebø‐Jones, Arman Tsaturyan, Bartosz Małkiewicz, Agata Góral, Paulina Biesiadecka, Francisco Javier Lopez Vivo, Mantu Gupta, Juan Serna, Ziv Savin, Anastasios Anastasiadis, Dimitrios Memmos, Charlotte Slots, Rita Pages, Ceyhun Ozdemiroglu, Maria Florencia Frascheri, Rafael Carrasco Moreno, Eugenio Ventimiglia, Panagiotis Kallidonis, Evangelos Liatsikos, Otas Durutovic, Amelia Pietropaolo, Begoña Ballesta Martinez. Visualisation: Łukasz Nowak, Begoña Ballesta Martinez. Supervision: Wojciech Krajewski, Bhaskar Somani. Project administration: Łukasz Nowak, Begoña Ballesta Martinez. Funding acquisition: Not applicable.

Disclosure of Interests

None declared.

Funding

None.

Ethics Approval

Ethical or institutional approval was obtained according to local regulations at each participating centre.

Clinical Trial Number

Not applicable.

Consent for Publication

Not applicable.

Supporting information

Fig. S1A. Bootstrap validation of the ROC‐derived OT cut‐off for overall complications. The figure shows the distribution of OT by complication status, the ROC curve, the bootstrap distribution of optimal cut‐off values, and out‐of‐bag Youden index estimates. Although bootstrap‐derived cut‐offs were not identical across re‐samples, most clustered around the mid‐90‐min range, supporting the use of an operative time threshold close to 95 min.

Fig. S1B. Bootstrap validation of the ROC‐derived OT cut‐off for major complications. The figure shows the distribution of operative time by complication status, the ROC curve, the bootstrap distribution of optimal cut‐off values, and out‐of‐bag Youden index estimates. Although bootstrap‐derived cut‐offs were not identical across re‐samples, most clustered around the mid‐90‐min range, supporting the use of an operative time threshold close to 95 min.

Fig. S1C. Bootstrap validation of the ROC‐derived OT cut‐off for infectious complications. The figure shows the distribution of operative time by complication status, the ROC curve, the bootstrap distribution of optimal cut‐off values, and out‐of‐bag Youden index estimates. Although bootstrap‐derived cut‐offs were not identical across re‐samples, most clustered around the mid‐90‐min range, supporting the use of an operative time threshold close to 95 min.

Fig. S2. The ROC curves of the final multivariable logistic regression models for overall, major, and infectious complications. Panel (A) shows the ROC curve for the model predicting overall complications, panel (B) for major complications, and panel (C) for infectious complications. Corresponding AUC values with 95% CIs are shown in each panel.

Table S1. Detailed distribution of postoperative complications according to the modified Clavien–Dindo classification system validated for PCNL.

Table S2. The ROC analysis of OT for overall, major, and infectious complications.

Table S3. Comparison of baseline demographic, stone‐related, and perioperative characteristics according to the OT <95‐ vs ≥95‐min groups.

Table S4. Univariable logistic regression analysis for overall, major, and infectious complications. Variables with P ≤ 0.10, considered eligible for further selection into multivariable analyses, are highlighted in bold.

Table S5. Multivariable logistic regression analysis for overall, major, and infectious complications with OT entered as a continuous variable.

Table S6. Distribution of stone composition among patients with available stone analysis.

Data Availability Statement

The data and the full statistical code are available from the corresponding author upon reasonable request.