222 320
retreatment using the same modality [ie, more than one
PNL session] or other modality such as shockwave
lithotripsy [SWL] or transurethral endoscopic lithotripsy);
need for adjunctive procedure (a procedure to deal with a
complication [eg, nephrostomy or double-J stent insertion
for obstruction] and procedures incidental to the stone
removal process such as stent insertion and/or removal);
duration of hospital stay; hospital readmission; and
emergency department visit.
2.3.
Data extraction
A data extraction form was developed a priori to collect
information on study design, characteristics of participants
and interventions, and primary and secondary outcome
measures (benefits and harms).
2.4.
Risk-of-bias assessment
Two reviewers (Y.R. and A.T.) independently assessed the
risk of bias (RoB) for individual studies. Any disagreements
were resolved by discussion or by consulting a third author
(T.K.). The RoB in RCTs was assessed using the recom-
mended tool in the
Cochrane Handbook for Systematic
Reviews of Interventions
[13]. This included assessment of
random sequence generation, allocation concealment,
blinding of participants and personnel, blinding of outcome
assessment, incomplete outcome data, selective reporting,
and other sources of bias.
RoB in NRCSs was assessed using all seven domains
above and an extra item to assess the risk of findings being
explained by confounding. This is a pragmatic approach
informed by methodological literature pertaining to asses-
sing RoB in NRCSs
[14,15]. A list of the most important
potential confounders for harm and benefit outcomes was
developed a priori with clinical content experts (European
Association of Urology Urolithiasis Guidelines Panel). The
potential confounding factors were stone size, stone
location, obesity, previous open surgery on a kidney (eg,
open nephrolithotomy), subcostal versus intercostal punc-
ture, and calyceal access (upper vs middle vs lower calyx).
For each study, an algorithmic approach was used to
assess the risk of confounding bias. The following were
considered in sequence. (1) Was the prognostic confounder
considered? If ‘‘no’’, the study was deemed to be at ‘‘high’’
RoB for this confounder. If ‘‘yes’’, go to the next question. (2)
Was the confounder balanced between the intervention(s)
and control group (s)? If ‘‘yes’’, the study was at ‘‘low’’ RoB. If
‘‘no’’, go to the next questions. (3) Was the confounder
controlled for in the analysis, for example via statistical
adjustment such as univariate or multivariate analysis or
propensity score matching? If ‘‘yes’’, the study was at ‘‘low’’
RoB. If ‘‘no’’, the study was at ‘‘high’’ RoB.
The approach described above cannot be used to assess
RoB in NRCSs. To address the external validity (ie,
applicability of the results to different people, places, or
time) of the NRCSs, we assessed whether study participants
were selected consecutively or representative of a wider
patient population, and whether the specified confounding
factors were comparable across studies reporting on the
same intervention. This too was a pragmatic approach
informed by the methodological literature
[16,17]. Further
details on the approach outlined above are available on
PROSPERO (CRD42015023766;
www.crd.york.ac.uk/ PROSPERO).
The key RoB and confounder assessments from the tools
described above were summarized and presented graphi-
cally
( Fig. 2 ).
2.5.
Data analysis
For the purposes of this review, we refer to all procedures
using smaller tract sizes ( 22 Fr) as mPNL when compared
with procedures using larger instruments (unless otherwise
specified). Ameta-analysis for the two RCTs was not feasible
because of heterogeneity in study design. Forest plots were
created for the RCTs and NRCSs reporting SFRs to show the
direction and magnitude of effects.
3.
Evidence synthesis
3.1.
Search results
The search returned 2945 abstracts
( Fig. 1), of which
240 were scrutinized for eligibility. Four articles in
languages other than English were excluded, as were an
additional 218 studies that did not meet the remaining
inclusion criteria. A total of 18 studies were eligible for final
inclusion, of which 13 were full-text articles
[8,11,18–28]and five were conference abstracts
[29–33] .3.2.
Study and patient characteristics
Of the 18 studies included, two were RCTs
[19,28], six were
NRCSs
[11,21,22,24,26,27], and ten were single-arm case
series on mPNL only
[8,18,20,23,25,29–33]. There was one
study on PNL using tract size 22 Fr
[21]and one on tract
sizes
<
20 Fr
[29] .Three studies evaluated tract size 18 Fr
[18,22,23] ,two studies 16 Fr
[19,26], and two 14 Fr
[11,25]. One study reported on tracts sized 14–18 Fr
[24],
and another on tracts 18 Fr
[27]. There were three studies
on tract sizes 11–13 Fr
[8,20,30]and one on 4.8 Fr
[28] .Three
studies presented as conference abstracts did not explicitly
state the tract size, but the procedures were termed mPNL
[31–33] .No studies reported QOL outcomes. The baseline
characteristics of the participants in the studies included
are outlined in
Table 1and the findings are summarized in
Table 2.
3.3.
RCTs
Two RCTs were identified, both of which were full-text
articles. Only one study investigated the benefits and harms
of mPNL using ‘‘regular’’ (16 Fr) compared to standard PNL
(24 Fr)
[19] .Both blood loss and the need for blood
transfusion were significantly lower in the mPNL group
(both
p
<
0.05), although the types of stone were not fully
comparable. The SFR was only significantly different for
E U R O P E A N U R O L O G Y 7 2 ( 2 0 1 7 ) 2 2 0 – 2 3 5
222