With great interest, we have read the recent network meta-analysis by Njei et al [1] comparing the effectiveness of endoscopic and pharmacological interventions in terms of the prevention of post-ERCP pancreatitis (PEP). We applaud the authors for their work to answer this clinically relevant question by performing network meta-analysis, which is a very intricate type of analysis to compare multiple interventions in the setting of paucity of head-to-head clinical trials. This study shows pancreatic stent to be the most effective intervention for high-risk patients followed by Ringer’s lactate in combination with rectal nonsteroidal anti-inflammatory drugs (NSAIDs). However, by looking at data presented in their manuscript and thorough review of articles included for analysis, we have some concerns.
First of all, the authors stated that they only included randomized controlled trials (RCTs) with high-risk patients. While that is true of all the studies evaluating the role of the pancreatic stent as shown in Fig. 5 of network meta-analysis [1], it’s not true for studies evaluating rectal NSAIDs and Ringer’s lactate. All the studies using Ringer’s lactate except Mok et al 2017 shown in Fig. 4 of the manuscript [1] do not seem to be limited to high-risk patients only. In addition, studies using rectal NSAIDs shown in Fig. 3 of the manuscript [1] have the following issues: 1) No studies except for Elmunzer et al 2012 and Murray et al 2003 mention that only high-risk patients were included, on contrary, some of them mention including average to high-risk patients; 2) A closer look at those studies shows that the number of patients with PEP mentioned in Fig. 3 of the study [1] are in fact numbers of patients with only moderate to severe pancreatitis, instead of the number of PEP in high-risk patients; 4) We can’t identify a way to isolate the number of high-risk patients and the incidence of PEP from the study articles themselves.
We also noticed an error in the number of patients included in the analysis not meeting inclusion criteria, in [Table 1] and [Table 2]. The patients in RCTs by Sotoudehmanesh et al 2007 and Otsuka et al 2012 were stratified as average risk in previous network meta-analysis by Akbar et al published in 2013 [2] in contrast to current network meta-analysis by Njei et al [1]. Thus, as per the authors’ inclusion criteria ideally, only 15 studies would have data on PEP incidence in high-risk patients instead of the 29 studies that the authors incorporated for this analysis. Finally, we speculate that the data were included in the network meta-analysis to arrive at a conclusion. If that is true, it might affect overall results of the study, and if not, it might at least affect the validity of this analysis.
Table 1
Issues with studies of Ringer’s lactate for PEP prevention included in the network as high risk.
Study
|
Availability of PEP incidence data from high-risk patients in original studies
|
Issues with current network-analysis
|
Buxbaum el al 2014
|
No separate incidence data based on risk
|
Incidence data from all patients were included regardless of risk class
|
Shaygan-Nejad et al 2015
|
No separate incidence data based on risk
|
Incidence data from all patients were included regardless of risk class
|
Chuankrekkul et al 2015
|
Abstract; with no mention of including only high-risk patients
|
Incidence data from all patients were included regardless of risk class
|
NCT0250049 2016
|
Incomplete RCT; no mention of including only high-risk patients
|
Incidence data from all patients were included regardless of risk class
|
Rosa et al 2016
|
Abstract; no mention of including only high-risk patients
|
Incidence data from all patients were included regardless of risk class
|
Choi et al 2016
|
Separate data available which were included appropriately
|
|
Chang et al 2016
|
Abstract; no mention of including only high-risk patients
|
Incidence data from all patients were included regardless of risk class
|
Mok et al 2017
|
Only high-risk patients included
|
Incidence data from moderate-severe pancreatitis used instead of total numbers
|
PEP, post-endoscopic retrograde cholangiopancreatography pancreatitis; RCT, randomized controlled trial
Table 2
Issues with studies of rectal NSAIDS for PEP prevention included in the network as high risk.
Study
|
Availability of PEP incidence data from high-risk patients in original studies
|
Issues with current network-analysis
|
Murray et al 2003
|
Only high-risk patients included
|
Numbers included in current network metanalysis are inaccurate
|
Sotoudehmanesh et al 2007
|
No separate incidence data based on risk
|
Incidence data from moderate-severe pancreatitis used instead
|
Otsuka et al 2012
|
No separate incidence data based on risk
|
Incidence data from moderate-severe pancreatitis used instead
|
Elmunzer et al 2012
|
Only high-risk patients included
|
Incidence data from moderate-severe pancreatitis used instead of total numbers
|
Doborante et al 2012
|
No separate incidence data based on risk
|
Incidence data from moderate-severe pancreatitis used instead
|
Alabd et al 2013
|
Abstract; not able to be accessed for review
|
|
Doborante et al 2014
|
No separate incidence data based on risk
|
Incidence data from moderate-severe pancreatitis used instead
|
Andrade-Davilla et al 2015
|
Only high-risk patients included
|
Incidence data from only moderate-severe pancreatitis used instead of total numbers
|
Patai et al 2015
|
No separate incidence data based on risk
|
Incidence data from moderate-severe pancreatitis used instead
|
Lua et al 2015
|
Only high-risk patients included
|
Incidence data from only moderate-severe pancreatitis used instead of total numbers
|
Luo et al 2016
|
Study designs compare preprocedural diclofenac with post-procedural for average to high-risk patients
|
Numbers of moderate-severe pancreatitis in average-risk patients are taken (See Table 3 of original study)
|
Levenick et al 2016
|
No separate incidence data based on risk
|
Incidence data from moderate-severe pancreatitis used instead
|
Ucar et al 2016
|
No separate incidence data based on risk
|
Incidence data from moderate-severe pancreatitis used instead
|
NSAID, nonsteroidal anti-inflammatory drug; PEP, post-endoscopic retrograde cholangiopancreatography pancreatitis