Page 72 - Read Online

Sugihara et al. Mini-invasive Surg 2018;2:11 I http://dx.doi.org/10.20517/2574-1225.2018.14 Page 3 of 7

performed once or twice per month on average. Second, the range of hospital volume needed to be limited to
regulate the hospital volume effect, a well-known factor affecting laparoscopic skill quality. If no restrictions
were placed on hospital volume in the study, the group for interval of ≤ 7 days would be mainly occupied
by cases from high-volume hospitals (e.g., ≥ 50 cases per year) and the group for interval of ≥ 57 days would
be filled by cases from low-volume facilities (e.g., ≤ 5 cases per year). Such a large inconsistency in hospital
volume among the interval categories would lead to a wide difference in laparoscopic skills at baseline.

Ethical approval
All procedures performed in studies involving human participants were in accordance with the ethical
standards of the institutional and national research committee and with the 1964 Helsinki declaration and
its later amendments or comparable ethical standards. The present study was approved by the Institutional
Review Board and Ethics Committee of The University of Tokyo (No. 3501).

Informed consent
Because the clinical data in the database were thoroughly de-identified and the study design was a secondary
analysis of administrative claims data, informed consent from individual patients was not required.

Statistical analysis
First, the interval period was defined as the duration between the current case and the last laparoscopic case
in the hospital. For example, when a laparoscopic nephrectomy was performed on April 30 and the previous
procedure was carried out on April 20, the interval was 10 days. Tables showing the patient background data
and distribution of laparoscopic time were constructed.

Second, the relationship between interval days and laparoscopic time was illustrated by adopting a method
for restricted cubic spline curves, as a technique that allows flexible descriptions of non-linear relationships
[5,6]
among variables .
Finally, multivariable linear regression analyses for laparoscopic time were performed with interval days
and other adjusting variables including age, gender, comorbidity (in the form of the Charlson comorbidity
[7]
[8]
index ), oncological stage (according to the International Union Against Cancer ), type of surgery
(nephrectomy or nephroureterectomy), hospital academic status, and hospital volume. The hospital
[9]
clustering effect was adjusted by a general equation estimation method . In the multivariable analyses,
missing values for oncological stage were regulated to avoid bias caused by incomplete data. This was
achieved by performing multiple imputations to replace the missing values with a set of substituted plausible
[10]
values by creating five filled-in copies using a method for polytomous regressions .

All statistical analyses were conducted using SPSS version 23.0 (IBM SPSS, Armonk, NY) and R version
3.0.2 software (R Foundation for Statistical Computing, Vienna, Austria) with the RMS 4.0-0 package [11,12] .
2
Univariable comparisons were analyzed by the χ test and Kruskal-Wallis test as appropriate. The threshold
for significance was P < 0.05.

RESULTS
Overall, 6849 laparoscopic nephrectomy/nephroureterectomy cases were included from 222 hospitals.
From the 222 hospitals, 750 laparoscopic adrenalectomy, 151 laparoscopic pyeloplasty and 816 laparoscopic
prostatectomy cases were also identified. Intervals to laparoscopic surgery was calculated based on the total
of 8566 cases, and the baseline characteristics in the interval groups are shown in Table 1. The long interval
groups (≥ 29 days) were slightly biased toward advanced oncological stage, frequent nephrectomy, low
hospital volume, low rate of academic hospitals, and longer laparoscopic time.