Perioperative Outcome Comparisons Between Open and Laparoscopic Nephroureterectomy Among a Population-Based Cohort from 2010 to 2012

Abstract

Purpose:

To compare the perioperative outcomes and costs between open and laparoscopic nephroureterectomy for malignant diseases on a contemporary population-based level.

Patients and Methods:

Based on the Japanese Diagnosis Procedure Combination database for 2010 to 2012, we compared six end points of in-hospital mortality, intraoperative and postoperative complications, transfusion, anesthesia time, postoperative length of stay, and costs between open and laparoscopic nephroureterectomy under one-to-one matching based on the propensity scores. Multivariate analyses included sex, age, Charlson comorbidity index, body mass index, oncologic stage, hospital volume, and hospital academic status. Missing values were filled in by five-copy multiple imputations.

Results:

Among 3595 open and 3349 laparoscopic nephroureterectomies, an average of 2902 matched pairs were generated by the imputation and matching process. The outcomes showing significantly favorable association with the laparoscopic approach over the open approach were in-hospital mortality (0.3% vs 0.7%; odds ratio [OR], 0.41 [95% confidence interval, CI, 0.17 to 0.99]), postoperative complications (9.4% vs 12.6%; OR, 0.73 [0.58 to 0.91]), transfusion (12.9% vs 20.6%; OR, 0.54 [0.46 to 0.64]), postoperative length of stay (median, 11 vs 12 days; Beta, −0.041 [−0.059 to −0.023]), and costs without the operating room (median, $6607 vs $7077; Beta, −0.030 [−0.048 to −0.013]), while significantly longer anesthesia time (median, 278 vs 245 min; Beta, 0.057 [0.041 to 0.074]) and higher total costs (median, $15691 vs $12846; Beta, 0.078 [0.068 to 0.088]) for laparoscopic than for open nephroureterectomies were noted. There was no difference in intraoperative complications (P=0.774).

Conclusion:

Several favorable perioperative outcomes including low mortality were observed in laparoscopic nephroureterectomy compared with open nephroureterectomy.

Introduction

Since Clayman and colleagues ¹ successfully introduced a laparoscopic technique to nephrectomy in 1990, this minimally invasive approach has gradually taken the place of traditional open approaches in nephroureterectomy and is recognized as one of the standard options for surgery of upper tract urothelial carcinoma (UTUC).² In Japan, it is estimated that a laparoscopic approach had already been applied to about 45% of nephroureterectomy procedures as of 2010.³

UTUC accounts for fewer than 5% of urinary system malignancies,⁴ and this low incidence has led to limited evidence from centers of excellence regarding the safety and feasibility of laparoscopic nephroureterectomy (LNU) compared with open nephroureterectomy (ONU). A recent systematic review documented that there were no significant differences in intraoperative complications, postoperative complications, and perioperative mortality.⁵

Conversely, Hanna and associates⁶ performed a population-based analysis comparing ONU and LNU using U.S. nationwide inpatient samples and showed several points of superiority for LNU over ONU. Their study, however, lacked oncologic stage and body mass index (BMI) data, which could cause critical potential bias in the selection of ONU and LNU. In addition, more than half of the nephroureterectomies in that study were performed before 2001, and recent cases (2006–2009) only accounted for about 10%.

Taking the current dramatic advances in laparoscopic technology into consideration, higher quality evidence for the safety of LNU and ONU on a contemporary population-based background should be obtained. In the present study, we performed a nationwide analysis to compare the perioperative outcomes between ONU and LNU using a recent cohort from 2010 to 2012, incorporating detailed background adjustments including oncologic status and BMI.

Patients and Methods

Data source for the study

In the present study, the Diagnosis Procedure Combination (DPC) database was used as a data source. The DPC database is a Japanese nationwide inpatient administrative claims database developed under a project to establish an original case-mix classification linked with a lump-sum payment system, which is partially similar to the Medicare payment system in the United States.⁷ The lump sum payment covers basic hospital fees, and some inexpensive charges for medications and procedures, while surgeries, anesthesia, and expensive pharmaceuticals and interventions were under a fee-for-service system.⁸ The database collected data from July 2010 to March 2012 and has about 45% representativeness in terms of acute inpatient hospitalization throughout Japan.

The main diagnoses, preexisting comorbidities, and complications after admission were coded by International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes. The surgical procedures were recoded according to the Japanese original coding system. In addition, the database included oncologic stage and BMI data. Because the design of the present study was a secondary analysis of the administrative claims data and the collected data were thoroughly deidentified, informed consent was not needed. Internal data audit for the DPC database were not available; however, appropriate reports for the database were required by the Japanese healthcare authority. Study approval was obtained from the Institutional Review Boards and Ethics Committee of The University of Tokyo.

Data sampling and end points

We selected patients who underwent ONU (Japanese codes: K773 and K773-2) or LNU (Japanese code: K773-1) for pelvic or ureter malignancy (ICD-10 codes: C65 and C66) between July 2010 and March 2012. Robot-assisted surgery was not included.

As the baseline patient and hospital characteristics, we extracted information for sex, age, comorbidities at admission, BMI, oncologic T, N, and M categories (according to the International Union Against Cancer),⁹ hospital volume (annual nephroureterectomy caseload at each hospital), and hospital academic status (academic or nonacademic). Comorbidities were converted into the Charlson comorbidity index (CCI) according to Quan and coworkers.¹⁰

The end points evaluated were: (1) In-hospital mortality; (2) incidence of intraoperative and in-hospital postoperative complications; (3) frequencies of blood transfusion; (4) anesthesia time; (5) postoperative length of stay; and (6) total costs with or without surgery and anesthesia. The values of anesthesia time, postoperative length of stay, and costs were transformed into log10 values for analyses because of their skewed distributions. Total costs were converted at the currency rate of 100 Japanese yen to 1 US dollar.

The presence of in-hospital complications was identified as follows: In-hospital mortality; sepsis (ICD-10 codes: A32.7, A40.x, A41.x, A42.7, B37.7, and T81.4); disseminated intravascular coagulation (D65); pulmonary embolism (I26.x); cardiac events (ischemic heart disease [I20.x–24.x]; heart failure [I11.0 and I50.x]); other vascular complications (I71.x, I73.9, I74.x, I77.x, I80–I83.x, I89.x, K55.0, K55.9, T79.0, T79.1, and T81.7); stroke (I60.x–64); pneumonia or flu (J10.x–18.x); other respiratory complications (J46, J69.0, J70.x, J80–J86.x, J93.x, J94.2, J95.x, and J96.x); peritonitis or peritoneal abscess (K65.x, N73.3, and N73.5); ileus (K56.x and K91.3); acute renal failure (N17x, N28.0, and N99.0); genitourinary infection (N10, N30.x, N41.x, N45.x, N70.x–N73.x, and N76); other genitourinary complications (N13.x, N32.x, N35.x, 36.x, N82.x, and N99.x); disruption of operation wound (T81.3); and intraoperative complications (S34.x–S39.x and T81.2).

Statistical analysis

The threshold for significance was set at P<0.05. Univariable comparisons were performed by the chi-square test and the Mann–Whitney U test as appropriate.

To make the most of the available data, we filled in any missing values of BMI and oncologic TNM categories by the multiple imputation method and generated five complemented copies. Multiple imputation is a procedure used to replace missing values with other plausible values by creating multiple filling-in patterns to avert bias caused by missing data and is recognized as an alternative approach to analyzing incomplete data.¹¹As imputation techniques, a predictive mean matching and polytomous regressions were used for BMI and TNM categories, respectively.

After the multiple imputation, we performed one-to-one propensity-score matching to adjust for baseline differences between the ONU and LNU groups in each copy.¹² This matching methodology mimics randomized selections between cases and controls, and consequently strengthened the comparability between the ONU and LNU groups. A propensity score was defined as the log odds of the probability of ONU or LNU that calculated based on a dichotomous logistic model including sex, age, CCI, BMI, TNM categories, type of hospital, and hospital volume. The matching was performed under the nearest neighborhood rule using calipers with a width equal to 0.6 of the standard deviation of the propensity score.¹³ After matching, multivariable linear or logistic regression analyses were applied to the continuous or binary end points as appropriate.

In these models, hospital clustering effects were adjusted by generalized estimating equations.¹⁴ The covariates were sex, age, CCI, BMI, oncological stage, type of hospital, and hospital volume. We conducted the propensity score matching and built a separate model on each copy of the five imputed data files first, then we combined the results of the five imputed copies into one model for each end point, from which the statistical inference was taken. Statistical analyses were conducted using R version 4.0.0 (R Foundation for Statistical Computing, Vienna, Austria) with rms 4.0-0, Zelig 4.1–3, mice 2.17, and MatchIt 2.4–21 packages.^15

–20

Results

In total, 3595 ONU and 3349 LNU cases were identified in the DPC database during 2010 to 2012. Table 1 shows the patient baseline characteristics before and after the multiple imputation and propensity-score matching. The crude baseline characteristics differed substantially between the two groups. The LNU group was characterized by lower CCI, higher BMI, earlier oncologic status, higher hospital volume, and higher frequency of academic hospitals than the ONU group. Missing data accounted for 1.3% in BMI, 16.3% in the T category, 15.3% in the N category, and 14.9% in the M category.

Table 1.

Patient Baseline Characteristics for Open or Laparoscopic Nephroureterectomy Before and After Multiple Imputation and Propensity-Score Matching

	Before propensity-score matching n (%) or median (IQR)			After propensity-score matching ^† n (%) or median (IQR)
	ONU	LNU	P	ONU	LNU	P
Total	3595	3349		2902	2902
Female	1074 (29.9)	997 (29.8)	0.924	829 (28.6)	927 (31.9)	0.005
Age, year	73 (66–79)	73 (66–78)	0.687	73 (66–79)	73 (66–79)	0.729
CCI			<0.001
0	2074 (57.7)	1968 (58.8)		1705 (58.8)	1743 (60.1)	0.536
1	808 (22.5)	766 (22.9)		661 (22.8)	651 (22.4)
≥2	713 (19.8)	615 (18.3)		536 (18.5)	508 (17.5)
BMI	22.8 (20.6–25.0)	22.9 (20.8–25.1)	0.036	22.8 (20.6–25.1)	22.9 (20.7–25.0)	0.573
Missing	51 (1.4)	41 (1.2)		–	–
T category			<0.001			0.024
T₁	1117 (31.1)	1307 (39)		1223 (42.1)	1334 (46.0)
T₂	663 (18.4)	714 (21.3)		741 (25.5)	667 (23.0)
T₃	1071 (29.8)	748 (22.3)		866 (29.8)	830 (28.6)
T₄	135 (3.8)	55 (1.6)		72 (2.5)	71 (2.4)
Missing	609 (16.9)	525 (15.7)		–	–
N category			<0.001			0.001
N₀	2602 (72.4)	2742 (81.9)		2692 (92.8)	2716 (93.6)
N₁	236 (6.6)	115 (3.4)		128 (4.4)	143 (4.9)
N_2–3	149 (4.2)	36 (1.1)		82 (2.8)	43 (1.5)
Missing	608 (16.9)	456 (13.6)		–
M category			<0.001			0.292
M₀	2883 (80.2)	2846 (85.0)		2841 (97.9)	2852 (98.3)
M₁	113 (3.1)	32 (1.0)		61 (2.1)	50 (1.7)
Missing	599 (16.7)	471 (14.1)		–
Type of hospital			<0.001			0.662
Academic	927 (25.8)	1157 (34.5)		828 (28.5)	813 (28.0)
Nonacademic	2668 (74.2)	2192 (65.5)		2074 (71.5)	2089 (72.0)
Hospital volume	8 (5–12)	11 (7–15)	<0.001	9 (5–14)	9 (6–13)	0.042

†

One of five imputed models is displayed.

IQR=interquartile range; LNU=laparoscopic nephroureterectomy; ONU=open nephroureterectomy; CCI=Charlson comorbidity index; BMI=body mass index.

To make the most of the available data, we filled in any missing values of BMI and oncologic TNM categories by the multiple imputation method and generated five complemented copies. Propensity-score matching resulted in five dataset copies with an average of 2902 matched pairs, and the background variations became closely balanced except T and N categories and hospital volume.

Table 2 shows the univariable comparisons of the detailed in-hospital complications and other end points after the propensity-score matching. The in-hospital mortality, in-hospital postoperative complications, frequencies of transfusion, postoperative length of stay, and costs excluding surgery and anesthesia were significantly favorable for the LNU group compared with the ONU group, while the LNU group recorded significantly longer anesthesia time and higher total costs than the ONU group. No significant difference was observed in the intraoperative complications.

Table 2.

Comparisons of the Propensity-Score-Adjusted Outcomes Between Open and Laparoscopic Nephroureterectomy

	After propensity-score matching n (%) or median (IQR)^*
	ONU n=2902	LNU n=2902	P
In-hospital mortality	20 (0.7)	8 (0.3)	0.023
Postoperative complications	365 (12.6)	274 (9.4)	<0.001
Sepsis/DIC	79 (2.7)	51 (1.8)	0.013
Pulmonary embolism	2 (0.1)	4 (0.1)	0.414
Cardiac events	88 (3.0)	62 (2.1)	0.031
Vascular complications	23 (0.8)	12 (0.4)	0.062
Stroke	10 (0.3)	5 (0.2)	0.196
Pneumonia or flu	15 (0.5)	6 (0.2)	0.049
Respiratory complications	34 (1.2)	23 (0.8)	0.143
Peritonitis or peritoneal Abscess	7 (0.2)	6 (0.2)	0.781
Ileus	3 (0.1)	9 (0.3)	0.083
Acute renal failure	32 (1.1)	12 (0.4)	0.002
Genitourinary infection	64 (2.2)	56 (1.9)	0.461
Genitourinary complications	32 (1.1)	47 (1.6)	0.089
Disruption of operation wound	15 (0.5)	14 (0.5)	0.852
Intraoperative complications	9 (0.3)	7 (0.2)	0.617
Transfusion	599 (20.6)	375 (12.9)	<0.001
Anesthesia time, min^†	245 (185–322)	278 (222–350)	<0.001
Postoperative length of stay^†	12 (10–17)	11 (9–14)	<0.001
Total costs, US$^{† ‡}	12846	15691	<0.001
	(11261–15638)	(14377–17814)
Costs without	7077	6607	<0.001
surgery/anesthesia, US$^{† ‡}	(5618–9505)	(5521–8516)

One of five imputed models is displayed.

†

Compared using the Mann-Whitney U test. The chi-square test was used for other comparisons. The value was median and IQR. Other values were n and percentage.

‡

1 US$=100 Japanese yen.

DIC, disseminated intravascular coagulation.

Table 3 shows the results of the multivariable regression analyses for the end points. The obtained results were similar to those of the univariable comparisons. Regarding hospital academic status, academic hospitals were significantly associated with more frequent transfusion (odds ratio [OR], 1.34, P<0.01), longer anesthesia time (D, +9.2%, P<0.01), and shorter length of stay (D, −6.3%, P=0.01).

Table 3.

Multivariable Regression Analyses for Propensity-Score-Adjusted Outcomes of Laparoscopic Nephroureterectomy Compared with Open Nephroureterectomy, Adjusting for Sex, Age, Charlson CCI, BMI, Tumor, Node, Metastasis Categories, Type of Hospital, and Hospital Volume

	Coefficient	95% CI	P
In-hospital mortality	OR=0.41	(0.17 to 0.99)	0.049
Postoperative complications	OR=0.73	(0.58 to 0.91)	0.006
Intraoperative complications	OR=0.84	(0.25 to 2.73)	0.774
Transfusion	OR=0.54	(0.46 to 0.64)	<0.001
Anesthesia time^†	Beta=0.057D=+14.1%	(0.041 to 0.074) (9.8 to 18.5)	<0.001
Postoperative length of stay^†	Beta=−0.041D=–8.9%	(−0.059 to −0.023) (−12.6 to −5.1)	<0.001
Total costs^†	Beta=0.078D=+19.7%	(0.068 to 0.088) (16.8 to 22.5)	<0.001
Costs without surgery/anesthesia^†	Beta=−0.030D=−6.7%	(−0.048 to −0.013) (−10.3 to −2.8)	<0.001

†

The values were transformed into log₁₀ values for the models because of their skewed distributions. Betas were unstandardized beta values and Ds were normal values back form logarithmic conversion.

CI=confidence interval; OR=odds ratio.

Discussion

In the present population-based study, we analyzed about 3000 propensity-matched pairs of ONU and LNU cases derived from a highly representative nationwide database and observed that the laparoscopic approach was associated with several favorable outcomes including mortality, complications, transfusion, postoperative length of stay, and costs excluding surgery and anesthesia, while having some disadvantages in anesthesia time and total costs.

This is the first observational study to show the beneficial association of LNU over ONU with postoperative complications and in-hospital mortality in the contemporary era. Today, even though laparoscopic techniques have been increasing in popularity worldwide,^3,21 evidence regarding the beneficial merits of LNU over ONU remains limited. The reason is that the rarity of UTUC makes it difficult to establish large series without including cases that are out of date for evaluating current minimally invasive technology. For example, a report from a French multicenter group including 150 LNU and 459 ONU cases between 1995 and 2010 showed no significant differences between ONU and LNU in terms of intraoperative and postoperative complications.²² Further, a recent systematic review by Ni and associates.⁵ found similar results that LNU is not superior to ONU regarding these complications. These findings were in strong contrast to those for radical nephrectomy, wherein several studies found that the laparoscopic approach had a strong association with fewer complications.^23,24

As a retrospective nationwide population-based study similar to ours, Hanna and colleagues⁶ compared 3014 ONU and 754 LNU cases based on U.S. nationwide inpatient samples during 1998 to 2009 and found fewer intraoperative complications in LNU than in ONU, but no differences in mortality and postoperative complications. Although that study was well designed and involved the largest previous population, about 55% and 88% were cases before 2001 and 2005, respectively, and we think that their results do not reflect contemporary laparoscopic techniques. In addition, the lack of oncologic status and BMI data could cause critical bias in the study baseline characteristics. The present study overcame these drawbacks by adopting a highly representative database that included oncologic stage and BMI data, and limiting the cases to only contemporary cases between 2010 and 2012.

The most remarkable points in our findings were substantially lower mortality (OR=0.41; p=0.049), postoperative complications (OR=0.73; p=0.006), and transfusion (OR=0.54; P<0.001) observed in LNU compared with ONU. Although we cannot reach definitive conclusions because of data limitations, we think that several technologic advances and the cumulative knowledge regarding safety in minimally invasive surgery during the previous decade would contribute to these results.

Emerging new technologies, including vessel sealing systems and clipping devices, such as the Hem-o-lok, allow safer, faster, and more secure ligation and transection.^25,26 Video technology had also been rapidly advancing, and we can currently enjoy high-definition scopes, flexible scopes, high optical resolution, and bright images, which were not available a decade ago.²⁷ Constant efforts to brush up surgeons' skill levels comprise another important factor. For example, in Japan, a laparoscopic skill qualification system was introduced in 2003 and the quality of surgeons' laparoscopic skills is evaluated by peer review under uniform criteria.²⁸ Lower transfusion rates in LNU compared with ONU were frequently reported in previous studies of nephroureterectomy and other surgeries, and pneumoperitoneum pressure during the laparoscopic procedure would contribute to the lower blood loss.^5,6,23

Conversely, it was observed that patients with LNU tended to be exposed to longer anesthesia (+14.1%; P<0.001) and had to defray more costs (+19.7%; P<0.001) than those with ONU. These results are plausible, because LNU necessitates more scrupulous surgical manipulation and prudent control of anesthesia.^5,25 The favorable laparoscopic features in decreasing the chances of complications observed in the current study, however, would offset these negative aspects. Shorter hospitalization (−8.9%; P<0.001) and lower costs out of the operating room (−6.7%; P<0.001) in the laparoscopic group than in the open group would also support the concept of the less invasiveness and quicker recovery for LNU than for ONU.

It is also interesting to note that favorable associations with LNU were observed for sepsis, cardiac events, pneumonia, and acute renal failure when each complication was overviewed. It would be mainly explained by the less invasive nature of LNU; however, unintended selection of the surgeon to the patient could not be ruled out as a confounding factor. These findings are quite different from the preceding study by Hanna and coworkers,⁶ which found little advantage of LNU over ONU in terms of postoperative complications, except for a significant reduction in respiratory problems. On the other hand, we did not find any significant difference in intraoperative complications, while a significant difference was highlighted by Hanna and coworkers.⁶ Despite the difficulties associated with direct comparisons between the two studies because of the differences in coding rules and healthcare systems, we interpret this result to mean that today's medical safety management has worked equally well to reduce intraoperative problems, such as unintentional organ damage, in both open and laparoscopic surgeries.

Some better outcomes at academic hospitals than nonacademic hospitals were conceivable because academic institutions may well involve high-quality subspecialty practices and undergo professional peer reviews of their work.²⁹

The biggest drawback of the present study is that it was not a randomized controlled trial. Patients were assigned to ONU or LNU on a clinical-practice basis. Physicians and facilities that could provide LNU may be more likely to catch up with up-to-date medical progress than those that could not, and such differences in quality and experience related to physicians and facilities could represent hidden confounders influencing the observed outcomes. We also failed to include several variables that can potentially influence patient selection for ONU vs LNU, such as tumor size and history of abdominal surgery or radiation therapy.

We believe that we did the best we could, subject to these deficiencies, to minimize the inherent selection bias by adopting propensity-score matching, multiple imputation, and adjustment using generalized estimating equations.^11

–15 Some latent confounders, such as those noted above, were not available from the database, however. Therefore, the results of the current study must be interpreted cautiously. In addition, although we believe multiple imputation was a useful method to manage many missing data settings, there was a limitation to show validity of a “missing at random” premise that the method required.

Other limitations must be mentioned. First, we lacked detailed surgical information about distal ureter management and the extent of lymphadenectomy, which would be selection criteria for the surgical approach. Second, the analysis of an administrative claims database intrinsically involves the problem of miscoding, which could cause underestimation or overestimation of diseases and events. Although we believe the data quality of the DPC database was acceptable because it was directly linked with the insurance payment system, lacks of data audit and external validation of the DPC database focusing on research use should be pointed out as a limitation.

Third, hospitalization duration and costs strongly depend on the healthcare system of a particular country and would limit the generalizability of the results. For example, Japan is known to have the longest duration of hospital stay among developed countries.²³ Traditionally, in Japan, long-lasting nursing care after surgery is continuously provided at the hospital. Regarding nephroureterectomy, most of these patients would stay at hospitals until successful removal of their urinary catheters and good adhesion of skin incisions.

Fourth, interpretation of the result in in-hospital mortality should be cautious because its P value was borderline (P=0.049), and it could result in insignificant value if a multiple comparison adjustment was adopted. Finally, the participating facilities in the DPC database are biased toward large bed-volume facilities, which could be linked to potential sampling bias.³⁰ Despite these limitations, our analyses suggested worthwhile information for the effectiveness of LNU over ONU on a contemporary population-based level.

Conclusion

In our population-based comparison, we observed that LNU was significantly associated with favorable effectiveness for in-hospital mortality, postoperative complications, transfusion, postoperative length of stay, and costs excluding operating room, even though there were some disadvantages regarding anesthesia time and total costs. These findings suggest the further beneficial features of LNU over ONU on a nationwide level.

Footnotes

Acknowledgments

This study was funded by a ITO Genboku and SAGARA Chian Memorial Scholarship from Saga Prefecture, Japan, a Grant-in-Aid for Research on Policy Planning and Evaluation from the Ministry of Health, Labour and Welfare, Japan (No. H22-Policy-031), a Grant-in-Aid for Scientific Research B (No. 22390131) from the Ministry of Education, Culture, Sports, Science and Technology, and a grant from the Funding Program for World-Leading Innovative R&D on Science and Technology (FIRST Program) from the Council for Science and Technology Policy, Japan (No. 0301002001001).

Author Disclosure Statement

No competing financial interests exist.

Abbreviations Used

References

Clayman

, Kavoussi

, Soper

, et al. Laparoscopic nephrectomy. N Engl J Med, 1991; 324:1370–1371.

Rouprêt

, Babjuk

, Compérat

, et al. European guidelines on upper tract urothelial carcinomas: 2013 update. Eur Urol, 2013; 63:1059–1071.

Sugihara

, Yasunaga

, Horiguchi

, et al. Regional, institutional and individual factors affecting selection of minimally invasive nephroureterectomy in Japan: A national database analysis. Int J Urol, 2013; 20:695–700.

Siegel

, Naishadham

, Jemal

. Cancer statistics, 2012. CA Cancer J Clin, 2012; 62:10–29.

, Tao

, Chen

, et al. Laparoscopic versus open nephroureterectomy for the treatment of upper urinary tract urothelial carcinoma: A systematic review and cumulative analysis of comparative studies. Eur Urol, 2012; 61:1142–1153.

Hanna

, Sun

, Trinh

, et al. Propensity-score-matched comparison of perioperative outcomes between open and laparoscopic nephroureterectomy: A national series. Eur Urol, 2012; 61:715–721.

Sugihara

, Yasunaga

, Horiguchi

, et al. Impact of hospital volume and laser use on postoperative complications and in-hospital mortality in cases of benign prostate hyperplasia. J Urol, 2011; 185:2248–2253.

Yasunaga

, Horiguchi

. The Diagnosis Procedure Combination (DPC) database. Available at: http://plaza.umin.ac.jp/∼hmp/cgi-bin/wiki/wiki.cgi?page=DPCDatabase Accessed November 10, 2014 .

Sobin

, Wittekind

. TNM Classification of Malignant Tumours. New York: Wiley, 2002.

10.

Quan

, Sundararajan

, Halfon

, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care, 2005; 43:1130–1139.

11.

Rubin

, Schenker

. Multiple imputation in health-care databases: An overview and some applications. Stat Med, 1991; 10:585–598.

12.

Griswold

, Localio

, Mulrow

. Propensity score adjustment with multilevel data: Setting your sites on decreasing selection bias. Ann Intern Med, 2010; 152:393–395.

13.

Rosenbaum

, Rubin

. Constructing a control group using multivariate matched sampling methods that incorporate the propensity score. Am Stat, 1985; 39:33–38.

14.

Panageas

, Schrag

, Riedel

, et al. The effect of clustering of outcomes on the association of procedure volume and surgical outcomes. Ann Intern Med, 2003; 139:6586–6565.

15.

Buuren

, Groothuis-Oudshoorn

. MICE: Multivariate imputation by chained equations in R. J Stat Softw, 2011; 45:1–67.

16.

Harrell

Jr.

rms: regression modeling strategies. R package version 4.0-0. 2013. Available at: http://CRAN.R-project.org/package=rms Accessed April 10, 2014 .

17.

, Imai

, King

, et al. MatchIt: Nonparametric preprocessing for parametric causal inference. J Stat Softw, 2011; 42:1–28.

18.

Imai

, King

, Lau

. Toward a common framework for statistical analysis and development. J Comput Graph Stat, 2008; 17:892–913.

19.

Imai

, King

, Lau

. Zelig: Everyone's statistical software. 2009. Available at: http://gking.harvard.edu/zelig Accessed April 10, 2014 .

20.

R Core Team. R: A language and environment for statistical computing. 2013. Available at: http://www.R-project.org/ Accessed April 10, 2014 .

21.

Xylinas

, Shariat

, Zerbib

. Uptake of laparoscopic radical nephroureterectomy in France: A 2003-2011 national practice report. Eur Urol, 2012; 62:940–942.

22.

Ariane

, Colin

, Ouzzane

, et al. Assessment of oncologic control obtained after open versus laparoscopic nephroureterectomy for upper urinary tract urothelial carcinomas (UUT-UCs): Results from a large French multicenter collaborative study. Ann Surg Oncol, 2012; 19:301–308.

23.

Sugihara

, Yasunaga

, Horiguchi

, et al. Comparisons of perioperative outcomes and costs between open and laparoscopic radical prostatectomy: A propensity-score matching analysis based on the Japanese Diagnosis Procedure Combination database. Int J Urol, 2013; 20:349–353.

24.

Sun

, Abdollah

, Shariat

, et al. Propensity-score matched comparison of complications, blood transfusions, length of stay, and in-hospital mortality between open and laparoscopic partial nephrectomy: A national series. Eur J Surg Oncol, 2012; 38:80–87.

25.

Janetschek

, Bagheri

, Abdelmaksoud

, et al. Ligation of the renal vein during laparoscopic nephrectomy: An effective and reliable method to replace vascular staplers. J Urol, 2003; 170:1295–1297.

26.

Landman

, Kerbl

, Rehman

, et al. Evaluation of a vessel sealing system, bipolar electrosurgery, harmonic scalpel, titanium clips, endoscopic gastrointestinal anastomosis vascular staples and sutures for arterial and venous ligation in a porcine model. J Urol, 2003; 169:697–700.

27.

Rassweiler

, Rassweiler

, Kenngott

, et al. The past, present and future of minimally invasive therapy in urology: A review and speculative outlook. Minim Invasive Ther Allied Technol, 2013; 22:200–209.

28.

Habuchi

, Terachi

, Mimata

, et al. Evaluation of 2,590 urological laparoscopic surgeries undertaken by urological surgeons accredited by an endoscopic surgical skill qualification system in urological laparoscopy in Japan. Surg Endosc, 2012; 26:1656–1663.

29.

Trinh

, Schmitges

, Sun

, et al. Radical prostatectomy at academic versus nonacademic institutions: A population based analysis. J Urol, 2011; 186:1849–1854.

30.

Sugihara

, Yasunaga

, Horiguchi

, et al. Admissions related to interstitial cystitis in Japan: An estimation based on the Japanese Diagnosis Procedure Combination database. Int J Urol, 2012; 19:86–89.