One Session or Two? A Meta-Analysis of Same-Session Versus Staged Bilateral Percutaneous Nephrolithotomy

Eligibility criterion

The studies included in this comparative meta-analysis adhered to the following inclusion criterion: (1) comparison between bilateral same-session percutaneous nephrolithotomy and staged percutaneous nephrolithotomy for the management of bilateral intrarenal stones, and (2) randomized or nonrandomized studies reporting relevant clinical outcomes such as SFR, complications, Hb decrease, transfusion need, hospital stay, and operation duration. Exclusion criterion included studies that focused solely on unilateral PCNL, lacked a control group for comparison, provided insufficient stratified data, or were case reports, letters, editorials, or comments.

Search strategy and data extraction

A systematic search was performed in PubMed, Scopus, and Cochrane databases from inception until November 2024. The search strategy included the following terms: (“Percutaneous nephrolithotomy” OR “PCNL” OR “PNL” OR “Percutaneous Nephrolithotripsy”) AND (“Bilateral” OR “Simultaneous” OR “Same session” OR “Same-session” OR “Same day” OR “Same-day” OR “Staged” OR “Staged Approach”). In addition, the reference lists of all included studies were manually searched to identify any potentially eligible studies. Data extraction was conducted independently by two authors (L.V.S. and C.M.F.), with disagreements resolved through a third author (H.L.L.).

End points and subanalyses

We included studies of (P) patients with bilateral intrarenal stones undergoing bilateral PCNL. The (I) intervention of interest was ssPCNL, which was compared against (C) sPCNL. The primary end points (O) of this meta-analysis were SFR, which was defined and analyzed according to the criterion used in each study, minor complications (Clavien-Dindo grades I and II), and hospital stay. Secondary outcomes included need for blood transfusions, changes in Hb, and operative time. Subgroup analyses were performed to account for potential variations in study design. The first subgroup consisted of randomized clinical studies, whereas the second subgroup included retrospective studies. For outcomes exhibiting significant heterogeneity, as determined by Cochran’s Q test and I² statistics, a Baujat analysis was conducted to identify the studies contributing most to the observed heterogeneity.

Quality assessment

The quality assessment of included randomized studies was performed using the Cochrane Collaboration tool for assessing the risk of bias in randomized trials (RoB-2). ⁷ For nonrandomized studies, the quality assessment was conducted using the Risk of Bias in Nonrandomized Studies of Interventions (ROBINS-I). ⁸ Two independent authors (A.H. and C.M.F.) assessed the risk of bias, resolving disagreements through discussion until a consensus was reached.

Statistical analyses

This systematic review and meta-analysis was conducted in accordance with the PRISMA guidelines. ⁹ For dichotomous outcomes, the results were expressed as odds ratios (ORs) with 95% confidence intervals (CIs). For continuous outcomes, MDs with 95% CI were used. Heterogeneity was assessed using the Cochran Q test and the I² statistic, with p-values <0.05 and I² >25% considered indicative of significant heterogeneity. A random-effects model was applied for all pooled analyses.

All statistical analyses were conducted using RStudio software. The Baujat plot and leave-one-out (LOO) analysis were performed to address significant heterogeneity.

Results reported as “per-procedure” for the staged group were adjusted for a comparison of cumulative results. The means were doubled. Cumulative standard deviation was calculated by summing the variances of the two procedures and taking the square root of the sum.

Study selection

We have done a search providing an overall of 1068 results (Fig. 1): 451 of them were in PubMed, 556 in Scopus, and 61 in Cochrane Library. We removed 428 duplicated articles, leaving 640 studies eligible for screening based on their titles and abstracts. Thirty-nine studies were then selected for full-text reading and exclusion screening. Finally, five studies were included in our final meta-analysis.

OPEN IN VIEWER

FIG. 1.

PRISMA flow diagram. This PRISMA flow diagram illustrates our process of screening and selecting studies. As a result, five studies were included in the meta-analysis.

Baseline characteristics of studies and population

The baseline characteristics of this meta-analysis are presented in Table 1. In the five studies analyzed, there was a total of 749 patients, with 441 undergoing sPCNL and 308 undergoing ssPCNL. The table separates each study into bilateral or unilateral groups and contains all the information referring to sample size, male or female division, age, body mass index, and follow-up of the population in the included articles. In addition, the table presents the characteristics of the stones and important information about the studies for this meta-analysis: the number of staghorn calculi, stones’ burden and size, and the preoperative Hb.

OPEN IN VIEWER

Table 1.

Studies’ Baseline Characteristics and Patient Information

Author (year)	Study design	Groups	Sample size	Male/Female (n)	Age	Body mass index (kg/m²)	Follow-up in months	N° staghorn calculi (%)	Stone burden (cm²)	Stone size (cm)	Preoperative hemoglobin (mg/dL)
ElSheemy et al. 14	P	Bilateral Unilateral	45 55	28/17 30/25	*38.95 ± 13.21 *35 ± 11.83	NA	41	13.3 8.2	*3.36 ± 1.61 *3.38 ± 1.18	NA	*12.66 ± 1.24 *12.76 ± 1.73
Holman et al. 12	R	Bilateral Unilateral	198 300	163/35 NA	**39 (1.25–70) NA	NA	NA	3.03 4.67	NA	**1–5 NA	NA
Silverstein et al. 6	R	Bilateral Unilateral	7 19	3/4 9/10	**44 (17–61) **56 (40–77)	NA	84	NA	8.61 11.16	2.94 3.46	NA
Torricelli et al. 5	P	Bilateral Unilateral	8 18	2/6 5/13	*33.3 ± 12.3 *45.9 ± 10.6	*31.6 ± 5.1 *29.4 ± 8 .3	95	NA	NA	NA	*13.1 ± 1.4 *13.1 ± 1.4
Wang et al. (2010)	Randomized Controlled Trial (RCT)	Bilateral Unilateral	50 49	37/13 38/11	57.1 58.1	*24.94 ± 2.62 *27.71 ± 2.84	*18.04 ± 4.62 *18.31 ± 4.66	100	*4624.64 ± 1425.1 mm² *4656.90 ± 1567.8 mm²	NA	NA

*Mean.

**Median.

Quality assessment

Every study incorporated in this analysis was measured for risk of bias in agreement with the Cochrane Handbook ¹⁰ and is presented in Figure 2.

OPEN IN VIEWER

FIG. 2.

Quality assessment of the included articles in this meta-analysis, according to the RoB2 tool, for the randomized trials (A) and ROBINS-I, for nonrandomized studies of intervention (B). RoB2 = risk of bias in randomized trials; ROBIN-I = risk of bias in nonrandomized studies of intervention.

The overall quality of the included studies varied, with the RCT demonstrating a low risk of bias, whereas the nonrandomized studies showed a moderate-to-serious risk of bias, primarily because of potential confounding and selection issues inherent in their designs.

The study by Wang and colleagues, being an RCT, was assessed using the RoB2 tool. The randomization process was adequately described, and although blinding is not feasible, outcomes were objective measurements. The study reported a clear protocol with minimal loss to follow-up.

The remaining nonrandomized studies were assessed using the ROBINS-I tool. The studies by ElSheemy and associates and Torricelli and coworkers were judged to carry moderate risk of bias. Both were prospective but did not randomize patients. For instance, there were strict inclusion/exclusion criterion in the study by ElSheemy; however, patients that declined ssPCNLs were included in the staged group. The studies by Holman and colleagues and Silverstein and associates were considered to carry serious risk of bias. Both were retrospective and less recent. For example, Silverstein and colleagues did not report if patients failed prior treatments before PCNL or if there were none.

Stone-free rate

The SFR forest plot for this outcome is presented in Figure 3. There was no difference in the overall pooled result (OR: 0.80 [0.47; 1.36]) as well as in subgroup analyses. There was also no heterogeneity in this comparison.

OPEN IN VIEWER

FIG. 3.

(A) Forest plot’s comparative analysis of the stone-free rate for same-session vs staged nephrolithotomy.

Minor complications

The minor complications’ forest plot analysis is exhibited in Figure 4. There was no difference in the overall pooled result (OR: 0.88 [0.44; 1.73]). The prospective subgroup was neutral (OR: 0.90 [0.45; 1.81]). The retrospective subgroup contained only one study (OR: 0.47 [0.02; 10.94]). There was also no heterogeneity in this comparison.

OPEN IN VIEWER

FIG. 4.

Minor complications comparative analysis between same-session vs staged nephrolithotomy in forest plot analysis.

Hospital stay

The forest plot for hospital stay is presented in Figure 5. The overall pooled analysis using a random effects model showed a statistically significant reduction in hospital stay for the same-session group (MD = −2.92 days; [−3.60; −2.24]; p ≤ 0.0001; I² = 84.7%). No heterogeneity was found within each subgroup (I² = 0% for both). Subgroup analysis revealed difference between prospective (MD = −2.51; [−3.00; −2.03]) and retrospective studies (MD = −3.70; [−3.88; −3.52]), as confirmed by the test for subgroup differences (χ² = 20.29, p ≤ 0.0001). The Baujat plot identified Holman and associates as the main contributor to heterogeneity and overall influence. Sensitivity analysis showed that removal of Holman and coworkers reduced heterogeneity to I² = 18.8% and maintained the significant effect size (MD = −2.62; [−3.04; −2.21]; p ≤ 0.0001).

OPEN IN VIEWER

FIG. 5.

Forest plot’s comparative analysis of the hospital stay for same-session vs staged nephrolithotomy.

Transfusion need

The forest plot for transfusion need is presented in Figure 6. There was no difference in overall (OR: 0.73 [0.24; 2.18]) and subgroup results. There was moderate heterogeneity for the overall pooled result (I² = 41.3%, τ² = 0.41, p = 0.16). The Baujat plot showed that the study by ElSheemy and associates was an outlier. After LOO, the omission of ElSheemy and coworkers led to the complete resolution of heterogeneity. The result remained neutral (OR: 1.18 [0.58; 2.37]).

OPEN IN VIEWER

FIG. 6.

Transfusion need comparative analysis between same-session vs staged nephrolithotomy in forest plot.

Hb decrease

The forest plot for this outcome is presented in Figure 7. There was no overall difference between staged and same-session (MD: 0.35 g/dL [−0.02; 0.72]). The Baujat plot showed two outliers—ElSheemy and associates and Wang and coworkers. After LOO, the omission of Wang and associates led to a significant result that favored the staged approach (MD: 0.52 [0.23; 0.81]) while reducing heterogeneity (I² = 89.1%–35.9%).

OPEN IN VIEWER

FIG. 7.

Hemoglobin decrease comparative analysis between same-session vs staged nephrolithotomy in forest plot analysis.

Operative time

The forest plot is presented in Figure 8. Overall difference favored ssPCNL (MD −39.18 minutes [−60.86; −17.51], I² = 55.6%). The Baujat plot evidenced Silverstein and colleagues as the most important outlier. After the omission of this study, the effect size was maintained, and heterogeneity was virtually removed (MD: −45.18 [−58.89; −31.46], I² = 0.9%).

OPEN IN VIEWER

FIG. 8.

Forest plot’s comparative analysis of the operation duration for same-session vs staged nephrolithotomy.

Summarization of results

Table 2 summarizes the results, considering the discussed sensitivity analysis.

OPEN IN VIEWER

Table 2.

Summarization of Results

Outcome	Effect size	Heterogeneity	Interpretation
Stone-free rates	OR 0.80 [0.47; 1.36]	I2 = 0%	No difference
Minor complications	OR 0.88 [0.44; 1.73]	I2 = 0%	No difference
*Hospital stay	MD −2.62; [−3.04; −2.21]	I2 = 18.8%; τ2 = 0.003	Favors ssPCNL
Transfusion need	OR 0.73 [0.24; 2.18]	I2 = 41.3%; τ2 = 0.51; p < 0.16	No difference
*Hemoglobin decrease	MD 0.52 [0.23; 0.81]	I2 = 35.9%; τ2 = 0.03	Favors sPCNL
*Operative time	MD −45.18 [−58.89; −31.56]	I2 = 0.9%	Favors ssPCNL

*After leave-one-out analysis and interpretation.

OR = odds ratio; MD = mean difference; ssPCNL = same-session bilateral percutaneous nephrolithotomy; sPCNL = staged bilateral PCNL.

Stone-free rates

The effectiveness of both methods was comparable. Most of the studies utilized a computed tomography scan for SFR evaluation; however, intravenous pyelograms were also used. The population of the studies had mixed profiles. They ranged from patients with a stone burden of less than 10 cm², in the case of ElSheemy and associates to a median greater than 46 cm² in the Wang and colleagues study.^11,14 The latter only included patients with staghorn calculi, differing from studies in which staghorn calculi were the minority of stones. This showed that staged and same-session PCNLs were both effective in clearing stones, regardless of clinical scenario, with similar results. Regardless of baseline differences, the absence of heterogeneity and publication bias was reassuring for this end point.

Minor complications

There was no difference between approaches. Complications were relatively infrequent considering the size of the procedures. The choice to compare only minor complications was because of the absence of major complications in studies. The only study that reported Clavien-Dindo III or IV complications was Silverstein and associates. ⁶

The low rate of complications in ssPCNLs could be attributed to the safety checkpoint that was observed in every study, converting ssPCNLs to staged PCNLs if serious complications occurred during the first side clearance.^5,6,11,12,14 Nevertheless, the studies by Holman and coworkers and ElSheemy and associates reported a conversion rate of 1% and 0%, respectively. Furthermore, major complications could appear after the completion of the second side, which was rarely seen.

The most common minor complication was bleeding. That is because of the inherent need to create tracts, as per the technique and patient profiles. In experienced centers, there should be replicability of these results.

Hospital stay

This end point favored ssPCNLs. There was a reduction in hospital stay when operating on both sides within one session. Moreover, the sole fact that ssPCNLs did not cause increased stay points toward increased efficiency. Hospital stay can be a surrogate indicator of complications after PCNLs, as they are closely associated with prolonged hospitalization.^13,15 The reduction can be explained by a single admission with a similar postoperative recovery. Since both approaches yield similar effectiveness with comparable SFRs, the advantage of ssPCNL lies in achieving a satisfactory outcome with shorter hospitalization.

A limitation of this finding is the difference between subgroups (i.e., prospective vs retrospective). A possible explanation is the discharge criterion, since the retrospective studies were also the least recent. Some protocols suggest waiting for the removal of drains, whereas some discharge patients with them in place. For instance, in the trial by Wang and colleagues ¹¹ a structured approach to the postoperative management could have led to more predictable discharge times. Nevertheless, both groups powerfully show a clear advantage of ssPCNL.

Secondary outcomes

Since bleeding was the major concern regarding complications in the included studies, transfusion need was included in our secondary analysis. No difference was found. However, there was moderate overall heterogeneity, caused by the study of ElSheemy and associates. ¹⁴ After omission by the LOO analysis, the result remained similar with a complete resolution of heterogeneity. This is consistent with the previous minor complications comparison.

Hb decrease was another factor related to intraoperative hemorrhage that was reported by studies and, therefore, included in this review. The result was neutral before sensitivity analysis. After the omission of the study by Wang and colleagues, ¹¹ the staged approach showed a smaller Hb decrease. However, the difference was small and likely clinically insignificant.

Operative time was included as a secondary outcome because several factors can directly impact this end point. For instance, positioning choices (supine vs prone), techniques, and surgeon preferences can significantly alter the operative time. ¹⁶ Initial heterogeneity was moderate; however, it was resolved after sensitivity analysis. After the omission of the study by Silverstein and associates, ⁶ the result favored the same-session approach. Patient positioning was not the same across the included studies. The faster completion of ssPCNLs can be explained by the need for only one anesthetic session and the possibility of draping both sides and operating in tandem. The technique of avoiding redraping in ssPCNLs has long been described and allows for a faster procedure. ¹⁷

Creatinine change was not statistically assessed because of reporting variations. However, the vast majority of patients did not experience a long-term, significant decline in kidney function. The RCT by Wang and coworkers reported pain through the Visual Analog Scale for pain. The results favored ssPCNL. ¹¹ Holman and colleagues mentioned that analgesia requirements were similar between groups. ¹²

Secondary outcomes were considered less central because of different confounding factors and single-study influence. Since evidence was provided, we pooled the results in our analysis. However, even after sensitivity analysis, these results should be interpreted with caution.

Cost-effectiveness

Our study did not include cost-effectiveness as an outcome. However, it is a pertinent addition to the discussion, and we want to underpin its necessity. As our meta-analysis provides a comprehensive clinical comparison, there is still a need to assess it through a resource management lens.

The study by Murray and associates ¹⁸ through a decision-analytic Markov model found that ssPCNLs were more cost-effective and offered increased quality-adjusted life years (QALY). However, they found slightly higher complication rates and a reduction in SFR for ssPCNLs. This highlights that even in suboptimal results of ssPCNLs, there are significant advantages. It was also stated that SFRs had to increase by 30% for staged procedures to be more cost-effective than simultaneous bilateral PCNLs. Moreover, considering the actual SFR found, the cost for staged PCNLs had to decrease by 50% to reach the cost of ssPCNLs. ¹⁸

This can be explained by a number of factors. The trial by Wang and colleagues ¹¹ also found a decreased cost for same-session PCNLs. They highlight fewer instruments used and decreased hospital stay as major reasons for decreased cost in ssPCNLs. Nevertheless, even if hospitals had considerable benefits, simultaneous procedures could be discouraged because of decreased physician reimbursement. ¹¹ Similarly, that possibility was also introduced in the work of Bagrodia and colleagues. ¹⁹ There were financial disadvantages for physicians performing ssPCNLs, but a decreased overall cost for these procedures.

All considered, the choice is complex and requires further studies assessing QALY, cost-effectiveness, and cost-reduction solutions.

Limitations

This study provides important insights but nevertheless carries significant limitations. Some studies had younger than usual patients and different percentages of staghorn calculi. Younger patients and less staghorn calculi could, in theory, decrease the overall incidence of complications. Also, ssPCNLs could have been especially favored because of lower single procedural difficulty and healthier patients for a larger procedure. However, in practice, such a trend was not evident. For instance, in the included randomized controlled trial, the rate of complications was comparable, and no Clavien-Dindo III or IV complication occurred at all, even with bilateral staghorn patients included. ¹¹

As ssPCNLs can be converted to sPCNL after a complicated first side, this is a potential source of latent bias. Retrospective studies carry the higher risk because analysis was done after procedures were performed. The studies that provided conversion rates—Holman and associates and ElSheemy and colleagues—reported 1% and 0%, respectively.^12,14

More recent series report shorter hospital stays. Shorter hospital stay for both arms favors sPCNLs because the mean difference tends to decrease. Practices like leaving nephrostomy tubes in for longer periods, as was common in the earlier studies, contrast with modern tubeless or early-removal techniques designed to expedite recovery. Also, smaller tracts with mini-PCNLs are increasingly more common.

SFRs differed between studies. There were two studies that included residual fragments in the stone-free status. The thresholds for Holman and coworkers and ElSheemy and colleagues were, respectively, <3 mm and <2 mm.^12,14 However, it was only used as the ratio between arms instead of the actual SFRs. Furthermore, tract sizes were not the same. Smaller tracts may reduce blood loss because of less trauma and add operative time, especially for larger stones.

Finally, paucity of well-designed studies led to the inclusion of nonrandomized studies, providing more evidence. Retrospective studies may inherently introduce bias. We performed subgroup analyses for every outcome and thorough investigation. As there were less than 10 studies included, publication bias assessment was not reliable and remains a latent issue. Embase and Web of Science were not searched directly, since Scopus has excellent breadth and overlaps with the others. Even though the reference lists of all included studies were manually searched to identify any potentially eligible studies, this could be a limitation.