Comparison of complication rates for unilateral and bilateral percutaneous nephrolithotomy (PCNL) using a modified Clavien grading system


Adam Kadlec, Loyola University Medical Center Department of Urology, Fahey Center, Bldg 54, Maywood, IL 60153, USA. e-mail:


What's known on the subject? and What does the study add?

Standardised complication reporting for urological procedures has been encouraged. Percutaneous nephrolithotomy (PCNL) has an acceptably low complication rate, with high-grade complications being rare. A recent review shows that complication rates for PCNL vary from 15% to 60%, but the Clavien system has not been used in all studies. Also, no study has directly compared unilateral and bilateral procedures.

In a direct comparison retrospective study, we showed that bilateral procedures had a higher overall complication rate and suggest that this is probably due to an increased number of percutaneous tracts.


  • • To compare complication rates of unilateral vs bilateral percutaneous nephrolithotomy (PCNL) using the modified Clavien grading system.


  • • Single-institution retrospective chart review.
  • • Identified patients who had synchronous bilateral PCNL (B-PCNL) over an 11-year period (2000–2011).
  • • B-PCNL patients were matched by age, gender, and stone burden per renal unit with patients who underwent unilateral PCNL (U-PCNL) over the same period.
  • • Complications were recorded and classified using the modified Clavien classification system, and then compared with a two-sided chi-square test of proportions.


  • • In all, 47 B-PCNL patients [mean (range) age 54 (14–84) years] and 78 U-PCNL patients [mean (range) age 54 (17–91 years] were compared. The mean (range) stone burden per renal unit was 2.8 (0.7–7.0) cm and 2.9 (0.7–9.0) cm, respectively.
  • • The overall complication rate was higher for B-PCNL (53.2% vs 30.8%, P= 0.013). Rates of Grade I and Grade II complications trended toward being higher in the B-PCNL group. Low-grade complications (Grade I and II combined) occurred more often in the B-PCNL group (P= 0.011); the most common low-grade complication was fever.
  • • Two procedures terminated for poor visualisation due to bleeding were successfully completed in delayed fashion with no transfusion and were classified as Grade I complications.
  • • There were equivalently low rates of Grade III complications. There were no Grade IV or V complications.


  • • B-PCNL carries a higher overall complication rate than U-PCNL when the modified Clavien system is used for classification.
  • • Patients undergoing U-PCNL who have more than one tract dilatation have a higher complication rate than those who have only one tract dilatation.
  • • High-grade complications are uncommon for both procedures.

body mass index


length of stay


(bilateral) (unilateral) percutaneous nephrolithotomy.


Standardised grading of complications is an essential component in the process of reporting surgical outcomes. The Clavien classification system, which was originally described in 1992 [1] and more recently modified in 2004 by Dindo et al. [2], is well-described and widely used in the field of urology.

Percutaneous nephrolithotomy (PCNL) is a minimally invasive endourological operation that is first-line treatment for large, multiple, or >1-cm lower-pole renal stones. In the past several years, there has been a drive to apply the Clavien system to newly-published and existing case series of PCNL [3], such that meaningful conclusions can be drawn when evaluating different surgical approaches, new technology, and/or separate patient populations. A recent comprehensive review by Seitz et al. [4] applied the modified Clavien classification to multiple international series of PCNL and concluded that the procedure carries an acceptably low complication rate in experienced hands.

Patients with large-volume bilateral nephrolithiasis represent a high-risk subset of patients with urolithiasis and may be candidates for synchronous bilateral PCNL (B-PCNL). To our knowledge, the present study is the first to use the modified Clavien system to directly compare complication rates between B-PCNL and the more commonly performed unilateral PCNL (U-PCNL).


We reviewed our Institutional Review Board-approved institutional database of endourological procedures and identified consecutive patients with urolithiasis who underwent planned synchronous B-PCNL over an 11-year period (January 2000–April 2011). A rank-list format was used to manually match these patients by age, gender, and stone burden per renal unit with a cohort of patients who underwent U-PCNL over the same period. Only initial procedures (‘first-look’) were analysed.

All patients completed standard preoperative evaluation including office history and physical examination, laboratory work including urine culture, and CT. Stone burden per renal unit was estimated by adding the lengths of all stones on transverse-plane CT images. Patients with positive urine cultures were treated with culture-appropriate antibiotics up to the day of the procedure. Perioperative i.v. antibiotics typically consisted of a β-lactam combined with an aminoglycoside and were started at the time of nephrostomy access.

Renal accesses were obtained by an interventional radiologist one day before each procedure. A Foley catheter was placed at the start of each procedure and the patient was turned prone. Dilatation was performed with a balloon dilator, over which a 30-F access sheath was placed. Using a combination of rigid and flexible nephroscopy, lithotripsy was performed with a combination of ultrasonic, holmium laser, and pneumatic energy, depending on stone size, location, and hardness. When appropriate, antegrade ureteroscopy was performed to clear the ureter of significant stone fragments. A 20-F Councill-tip catheter over a 5-F double open-ended ureteric catheter was left as a nephrostomy tube at the conclusion of all procedures. Patients confirmed to be stone-free by postoperative imaging (see below) were given a capping trial before hospital discharge and sent home tube-free, if the capping trial was tolerated. No tubeless procedures were performed.

Patients had postoperative imaging with either CT or plain abdominal X-ray (radiograph of the kidneys, ureters and bladder). The stone-free rate was determined based on the result of the initial procedure and was calculated by dividing patients with residual stone burden of <4 mm by the total number of patients treated in each group. Because second-look procedures were excluded, the stone-free rate only reflects that of the initial procedure. An additional procedure was recorded if any further endourological intervention was done, including second-look renal endoscopy, ureteroscopy, shockwave lithotripsy, insertion of ureteric stent, or insertion of percutaneous nephrostomy tube. Additional procedures were not recorded as complications unless they were unplanned and/or represented an acute change in management. Therefore, a second-look PCNL or planned ureteroscopy was not considered a complication.

Complications were recorded and classified using the modified Clavien grading system [2]. A two-sided chi-square test of proportions was used to compare complication rates between groups. The means of continuous variables were compared by a two-sided t-test of independent samples.


In all, 47 patients who had B-PCNL were eligible for study inclusion; most of B-PCNLs (34/47, 72%) were done in the latter 4 years of the study period. Appropriate matching was confirmed by the fact that age, gender, stone burden per renal unit, body mass index (BMI), and diabetic status did not differ between groups (Table 1). Operative duration was ≈50 min longer and hospital length stay was 1 day greater in the B-PCNL group. Of all stones, 32% were lower pole alone, while 16.9% and 19.1% were partial or complete staghorn stones in the U-PCNL and B-PCNL groups, respectively. The stone-free status after the ‘first look’ was equivalent between groups, as was the rate of additional procedures, which included second looks (Table 2).

Table 1. The clinical characteristics of the patients
Mean (sd) age, years53 (14)54 (17)0.82
Gender, n/N (%male)40/78 (51)26/47 (55)0.66
Mean (sd) stone burden, cm/renal unit2.8 (1.4)2.8 (1.3)0.85
Mean (sd) BMI, kg/m232 (9)32 (11)0.80
Diabetic, n/N (%)14/60 (23)12/40 (30)0.55
Table 2. The clinical outcomes. The stone-free rate reflects the number of patients with a stone burden of <4 mm on their first trip to the operating room. Additional procedures included planned endourological procedures
Mean (sd, range) operative duration, min80 (33, 20–176)131 (62, 49–375)<0.001
Mean (sd, range) LOS, days4 (2, 1–9)5 (2, 2–15)<0.001
Stone-free after initial procedure, n/N (%)37/78 (47)23/47 (49)0.87
Additional procedure performed, n/N (%)21/78 (27)15/47 (32)0.55

The modified Clavien classification scheme is reviewed in Table 3. A list of all recorded complications, sub-divided by grade, appears in Table 4. The total number of complications exceeds the number of patients because some patients had more than one complication. If so, the complication of highest grade was used for analysis, consistent with previous methodology in the urological literature [5].

Table 3. Modified Clavien classification system
  1. ICU, Intensive Care Unit.

Grade IAny deviation from the normal postoperative course without the need for pharmacological treatment or surgical, endoscopic, and radiological interventions.
Allowed therapeutic regimens are: antiemetics, antipyretics, analgesics, diuretics, electrolytes, and physiotherapy.
Grade IIRequiring pharmacological treatment with drugs other than such allowed for grade I complications.
Grade IIIaRequiring surgical, endoscopic, or radiological interventions not under general anaesthesia.
Grade IIIbRequiring surgical, endoscopic, or radiological interventions under general anaesthesia.
Grade IVaLife-threatening complication requiring ICU management (single organ dysfunction).
Grade IVbLife-threatening complication requiring ICU management (multiple organ dysfunction).
Grade VDeath of a patient
Table 4. List of all recorded complications, in both groups, subdivided by grade. For any given patient, only the complication of highest grade was analysed
Grade I (n)Grade II (n)Grade IIIa (n)Grade IIIb (n)
Fever (18)Acute cystitis (3)Ureteric obstruction (3)Pneumatosis intestinalis (1)
Persistent pain (3)Acute pyelonephritis (2)Hydrothorax (2) 
Transient confusion (4)Sepsis (4)  
Bradycardia (1)Pulmonary embolism (3)  
Tachycardia (5)Atrial fibrillation (2)  
Nausea/vomiting (4)Pneumonia (2)  
Acute urinary retention (1)Transfusion (5)  
Acute renal failure (5)   
Blurry vision (1)   
Narcotic withdrawal (1)   
Intraoperative bleeding (2)   
Pulmonary oedema (1)   

Febrile and infectious complications were most commonly seen, accounting for 18 of the 49 recorded complications in the total cohort. All patients with a fever of >38.6 °C had blood and urine cultures collected. If fever occurred in the absence of clinical signs of cystitis or pyelonephritis (e.g. dysuria, flank pain, nausea) or positive bacterial urine and/or blood cultures, a Grade I complication was recorded. If clinical signs of UTI were present, or if bacterial urine and/or blood cultures were positive, a Grade II complication was recorded, as these patients were treated with empiric antibiotics. Grade II complications were also recorded if antibiotic coverage was broadened due to a febrile episode, even if the patient was asymptomatic and cultures were negative. The complication of sepsis occurred in four patients who had both positive blood cultures and an alteration in mental status or haemodynamics.

There were only two intraoperative complications, both of which were bleeding that led to early termination of the procedure due to suboptimal visualisation. Neither of the patients was transfused and both were successfully treated with second-look PCNL during the same inpatient stay. These two complications were classified as Grade I and are discussed in more detail in the Discussion section.

Two pulmonary embolisms occurred, both in the U-PCNL group, and were diagnosed via spiral CT and treated with heparinization. Although pulmonary embolisms can be a life-threatening complication, both events were classified as Grade II complications, as the patients did not have haemodynamic changes and were effectively managed in a non-Intensive Care Unit setting.

High-grade complications were relatively rare and included three urinary tract decompressions for febrile obstruction, two chest tubes for hydrothorax (both were diagnosed and treated in postoperative recovery area), and one exploratory laparotomy performed by the general surgery service for acute abdominal pain with pneumatosis intestinalis. In that case, all bowel segments appeared viable and no bowel resection was performed.

Comparison of groups showed a higher overall complication rate (53% vs 31%, P< 0.001) in the B-PCNL group. Although there were no statistically significant differences between groups within each complication grade (Fig. 1), low-grade complications (Grade I and II combined) were more common in the B-PCNL group (Table 5). The six patients with a high-grade complication had an average length of stay (LOS) of 8.5 days. Patients with any complication had an average LOS of 5 days vs those with no complication who had an average LOS of 3 days (P≤<0.05).

Figure 1.

Frequency of complications divided by grade.

Table 5. Comparison of complications between groups
 U-PCNL, n/N (%)B-PCNL, n/N (%)P
Any24/78 (31)25/47 (53)0.01
Low-grade20/78 (26)23/47 (49)0.01
Grade I13/78 (17)14/47 (30)0.08
Grade II7/78 (9)9/47 (19)0.10
Grade III4/78 (5)2/47 (4)0.83

Our final analysis compared U-PCNL patients who had more than one tract dilated (14 patients) with U-PCNL patients who had only one tract dilated (64; Table 6). The group with multiple tract dilatations had a significantly higher overall complication rate (57% vs 25%, P= 0.018).

Table 6. Number of tracts dilated in each group
 U-PCNL, n (%)B-PCNL, n (%)
1 64 (82.1)0
2 13 (16.7)38 (80.9)
3 1 (1.3)6 (12.8)
4 03 (6.4)


There are multiple published series that show the efficacy and safety of B-PCNL [6–15]. In the comprehensive review by Seitz et al. [4], a complication (Grade ≥ 1) occurred in 23.3% of PCNLs, with a range of 14.2–60.3%, most of which were low-grade. Labate et al. [15] published a large multicentre series with a 20.5% complication rate, again noting that most complications were minor. The present series was comparable to other series in both overall complication rate as well as the skew toward low-grade complications. It also was consistent with previous literature in that stone-free and auxiliary procedure rates were similar between U-PCNL and B-PCNL.

Complication rates of B-PCNL vs U-PCNL have been reported elsewhere, but none have used the Clavien classification system. In 2002, Holman et al. [10] published a large series comparing the two procedures. B-PCNLs were synchronous and had a complication rate of 14.3%, which was statistically equivalent to U-PCNL. Silverstein et al. [12] reported a comparison of staged (which might be considered two U-PCNLs) vs synchronous B-PCNL and again found an equivalent complication rate. However, the Clavien system was not used in those studies. Conort et al. [7] presented a series of 60 bilateral cases in which the high-grade complication rate, using the Clavien classification, was 6.7%. Another large series used the Clavien classification to record complication rates of PCNL, but a distinction between B-PCNL and U-PCNL was not made [8,16].

In the present series, B-PCNL had a higher overall and low-grade complication rate than U-PCNL. There was a trend toward a higher complication rate in the B-PCNL for Grade I and Grade II complications (taken separately) and we hypothesise that those trends would be statistically significant in a larger series. This poses the question as to whether the higher complication rate is due to patient factors (e.g. anatomical abnormalities or comorbidities) or renal factors (e.g. number of access tracts or results of urine cultures from the renal pelvis). We think the present data suggests that renal factors drive the higher complication rate of B-PCNL, as our cohorts were matched for patient factors (i.e. age, BMI, diabetes) that are felt to have an effect on complication rates.

It is possible that the higher complication rate was driven primarily by the additional access points inherent to bilateral procedures, as was suggested in the Muslumanoglu et al. [17] series in 2006. In the present series, patients who had U-PCNL with more than one tract dilatation had a significantly higher complication rate; in fact the complication rate was similar to the B-PCNL patients, which further suggests that the cost of additional tract dilatation is an increase in the rate of complications. With regard to access, it would be interesting to repeat the Watterson et al. [18] study, which showed a lower complication rate when access was obtained by a urologist, but did not use the Clavien system to record complications.

Febrile episodes and infections represented the most common complications in both groups, with 10 (12.8%) in the U-PCNL group and eight (16.7%) in the B-PCNL group, which falls within the range of other studies [4]. Although a formal analysis on infectious complications is beyond the scope of the present study, the frequency of such complications does merit further discussion. Fevers can coexist with higher-grade complications and therefore some patients with a fever may have been analysed as a higher-grade complication. For instance, a patient with sepsis requiring antibiotics who had a fever is analysed only as a Grade II complication, not as a Grade I (for the fever) and a Grade II (for the sepsis). This is a feature inherent to the Clavien classification system, the cornerstone of which is the eventual therapeutic method (and its attendant risk) required to treat a given complication [2]. In that light, Table 4 lists all complications that occurred, but the present analysis of any given patient used the complication of highest grade.

All episodes of sepsis, wherein an alteration in haemodynamics or mental status was noted, occurred in the B-PCNL group. One patient had predominant (>50%) struvite component to his or her stones and four patients had a predominant calcium phosphate component; composition of stone was unknown in four patients. All of our patients received perioperative parenteral antibiotics, but only those with positive urine cultures were treated with >24 h of preoperative antibiotic therapy.

Variation in clinical practice may lead to differences in Clavien grading for febrile episodes, as not all clinicians would obtain blood and/or urine cultures for an isolated overnight fever (thereby ‘missing’ a potentially positive culture), while others may initiate empiric parenteral antibiotics for fever alone, which would create a Grade II complication even in the absence of positive cultures. In the presence of fever, in addition to blood cultures, we typically order bladder and renal pelvic urine cultures, as renal pelvic cultures have been shown to be more clinically informative than bladder urine cultures [19].

Bleeding issues represented another common complication. Severe bleeding complications of PCNL are somewhat rare (1.4% of 1854 patients required angiographic intervention in one large series [20]), but can have life-threatening implications. Although no life-threatening episodes occurred in the present cohort, the bleeding complications that did occur happened more often in the B-PCNL group, with three transfusions and two terminated procedures vs one transfusion and no terminated procedures in the U-PCNL group. Again, this was probably due to the additional access obtained in the B-PCNL patients. Other authors have classified a procedure terminated due to bleeding as a Grade III complication, as it necessitates a trip back to either the operating room for second-look PCNL or the sedation suite for shockwave lithotripsy [21]. However, we felt this to be over-grading, as termination of a procedure is a subjective decision that is usually made to prevent a more serious complication from occurring. Furthermore, the trip back to the operating room is not a ‘treatment’ of the bleeding but is rather further treatment of the stone; the treatment of the bleeding is observation as the bleeding self-resolves. The present patients did not have haemodynamically significant haemorrhage and neither patient required transfusion. Therefore, considering the circumstances, we classified the terminated procedures as deviations from the normal course, or a Grade I complication.

A reviewer noted a wide range in LOS and asked for further clarification regarding same. The longest LOS was 15 days in a patient with multiple medical problems who had B-PCNL for a 6.5-cm total stone burden who required a second-look as well as an unplanned stent for a retained ureteric stone fragment. Patients with a high-grade complication had a longer LOS than those who had no complications. Likewise, patients with any complication had a longer LOS than those who did not have a complication. This data further validates the utility of the Clavien system for PCNL and shows that recorded complications do translate to longer LOS, increased cost, and the potential for increased morbidity. Tubeless PCNL has been rapidly gaining popularity and is perhaps becoming the standard of care for certain patients, as it appears to lessen pain, LOS, and cost [22]. A purely tubeless cohort would be susceptible to sample bias toward smaller stones, uncomplicated anatomy, and cases with minimal bleeding; by extension, a non-tubeless cohort from an institution that often performs tubeless procedures may be biased toward larger stones and more complicated anatomical cases. As we do not perform tubeless procedures at our institution, the present cohort should give an accurate picture of the complication rates for all-comers.

The retrospective nature of the present study may have limited our ability to capture low-grade (especially Grade 1) complications, e.g. early dislodgement of nephrostomy tube or failure of capping trial due to flank pain. This limitation was hopefully minimised by use of an electronic medical record as the source of all data for the bulk of the study period.

Although the Clavien system clearly allows improvement in complication reporting, it has some limitations with regard to PCNL and other endoscopic urological procedures. It is imperfect for grading events such as second-look procedures or, as discussed earlier, early termination of a first-look procedure for bleeding issues. There is also the issue of tracking febrile and infectious events, where considerable overlap between Grade I and Grade II might be seen, depending on both practice patterns and methods used to identify and categorise complications. The ideal method of complication reporting for PCNL would allow for differences in surgical approach and patient management philosophy by only capturing events that should truly be regarded as complications, as well as categorising those events at a grade level that is appropriate for their severity.

In conclusion, when the modified Clavien system was used for reporting of surgical complications, B-PCNL had a higher overall complication rate than U-PCNL, which was driven by a higher rate of low-grade complications. U-PCNL patients who had more than one tract dilatation had a complication rate similar to those patients undergoing B-PCNL. The rate of high-grade complications for either procedure is low.


Nothing to disclose.