Aggressive Versus Conservative Fluid Treatment for Acute Pancreatitis in Adult Patients: A Meta-Analysis of 3,127 Cases

Background: The advantages of aggressive uid treatment (AFT) compared to conservative uid treatment (CFT) within 24 h for acute pancreatitis (AP) remain controversial in adult patients. A meta-analysis was undertaken to investigate whether aggressive strategies are more benecial. Methods: We searched (on February 1, 2021) PubMed, Embase, and the Cochrane Library for eligible trials that assessed the two therapies and performed a meta-analysis. The primary endpoint was in-hospital mortality. Secondary outcomes were adverse events (e.g., renal failure and pancreatic necrosis) within 24 h of treatment. Results: Five randomized controlled trials (RCTs) and 8 observational studies involving 3,127 patients were identied. There was a signicant difference in in-hospital mortality for AFT compared to CFT (OR, 1.66; P = 0.0001). The incidences of renal failure (OR, 2.38; P < 0.00001) and pancreatic necrosis (OR, 2.34; P < 0.0001) were similar and signicantly different between the two groups. Patients aged > 50 years had a potentially higher utilization of mechanical ventilation and incidence of respiratory failure (OR, 4.88; P < 0.00001). Persistent organ failure, systemic inammatory response syndrome (SIRS) and length of hospital stay did not differ signicantly between the two groups. Sensitivity analysis identied two signicant changes: one in persistent SIRS (OR, 2.37; P = 0.02) in patients aged > 50 years and one in the overall incidence of persistent organ failure (OR, 1.81; P = 0.02). Conclusions: Compared to CFT, AFT increases in-hospital mortality and the incidence of renal failure, pancreatic necrosis and respiratory failure with relatively strong evidence. trials. Trials with events in both arms were included. Funnel plots were used to screen for publication bias.

that investigated the in uence of uid type, especially for lactated ringer's solution, were also appropriate for inclusion in the present study. Therefore, the MeSH term lactated ringer's solution and combinations of the individual words were added to our electronic search. The meta-analysis was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Metaanalyses (PRISMA) guidelines for systematic reviews and meta-analyses [25]. The latest search was performed on February 1, 2021.
All comparative trials that compared AFT with CFT, had at least one of the following comparable outcomes and enrolled adult patients with AP, were included. AFT was de ned as uid administration at a rate greater than 10 ml/kg/h or 250 ml/h for the rst 24 h. Alternatively, with signi cant variability in the medical record over time and di culty with retrieving data on uid volume (FV) every hour per kilogram from a case, patients receiving FV > 4,000 ml or more than 1/3 of the total 72 h FV within the rst 24 h in observational studies were also considered to have ful lled the de nition of AFT. Meanwhile, uid therapy at a lower rate or FV than AFT was de ned as CFT. Trials in which differentiation between aggressive and conservative uid administration was not clearly prede ned were excluded, as were trials relevant to patients undergoing uid resuscitation for AP induced by endoscopic retrograde cholangiopancreatography (ERCP) or treated perioperatively. Editorials, letters, review articles, conference abstracts, case reports and animal experiments were not included.

Data Extraction
Extractable data from the included trials were summarized individually by two of the authors (Zhan and Wu). Any discrepancies were resolved by the adjudicative senior authors (Liao and Lai).
The following data were extracted from the included studies: rst author, year of publication, study design, characteristics of patients, inclusion and exclusion criteria, matching criteria, and number of enrolled cases for AFT and CFT. In-hospital mortality was regarded as the primary safety outcome. Secondary outcomes comprised the posttreatment incidence of persistent systemic in ammatory response syndrome (SIRS), pancreatic necrosis, persistent organ failure, renal failure, respiratory failure, and length of hospital stay (LOS). In addition to the primary endpoints, all eligible data were obtained within the rst 24 h.

Quality Assessment and Statistical Analysis
Trials were assessed using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) system to rate the level of evidence[26] with the GRADE pro ler (GRADEpro, version 3.6). With respect to the quality of evidence, the included studies were classi ed separately as RCTs and observational studies into 4 categories as follows: very low, low, moderate, or high.
To assess risk of bias, the Cochrane Collaboration tool was used to assess RCTs, and the Newcastle-Ottawa scale (NOS) was used to assess observational studies. The risk of bias in RCTs was rated according to seven factors: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other bias. To evaluate the dynamic change in uenced by the complex intervention, blinding of patients and clinicians was extremely di cult and generally not feasible, and prior to this study, we determined that in-hospital mortality was less prone to be in uenced by the absence of blinding. Trials were regarded as having a high risk of bias if at least one of the factors, other than blinding, was judged as high risk. The details of the quality assessment for the observational trials were summarized according to three factors: method of patient selection, comparability of study groups, and assessment of exposure. A score of 0-9 was allocated (as stars) to each observational study, and studies achieving seven or more stars were considered to be of high quality.
Data were combined for meta-analysis using Review Manager version 5.3, and the statistical tests were two-sided. The odds ratios (ORs) for dichotomous variables or weighted mean differences (WMDs) for continuous variables were calculated using the chisquare test with signi cance set at P < 0.05. Pooled data were estimated with 95% con dence intervals (CIs). Statistical heterogeneity was quanti ed using the I 2 statistic, with the level of signi cance set at I 2 > 40%. If the heterogeneity was considered to be substantial, we performed the analysis with a random-effects model; otherwise, a xed-effects model was used. Additionally, continuous data were presented as median and range values, and standard deviations (SDs) were calculated using the technique described by Hozo et al [27]. Prespeci ed subgroup analyses were performed to examine the safety of AFT compared to CFT with respect to the average age of patients. To test the robustness of the outcomes, sensitivity analysis was performed by omitting one trial at a time and analyzing the remaining trials. Trials with events in both arms were included. Funnel plots were used to screen for potential publication bias.

Results
Thirteen studies [12][13][14][15][16][17][18][19][20][28][29][30][31] published from 2009 to 2020 that included 3,127 patients ful lled the selection criteria and were included in the nal meta-analysis (Fig. 1). Analysis was performed on 1,472 patients treated with AFT and 1,655 patients treated with CFT. An expanded search of the reference lists of the included studies and reviews did not yield any additional studies to evaluate. Some overlap in patients was potentially present in two trials [15,17]; however, a discrepancy between the study period and endpoints was displayed as well, and both studies were therefore included in the nal analysis.

Characteristics and Quality Assessment of the Included Studies
The study characteristics are shown in Table 1. Among the 5 included RCTs [12,13,15,17,28], 3 [12,13,28] were judged to be of high methodological quality (Additional le 1: Fig. S1) and had similar therapeutic protocols for uid rate (FR) that were adjusted for the patient's weight. The remaining 2 RCTs [15,17] were judged to be of low quality with a high risk of bias in terms of selective reporting and other bias (Additional le 1: Fig. S1). These 2 studies presented a high risk of bias due to a lack of con ict of interest statement, although the protocols were supported by academic ndings, and unclear de nitions of outcomes without differentiation of primarysecondary outcomes. One of the RCTs [15] that was published early used a clear de nition of FR that was consistent with that of prior high-quality RCTs; another RCT [17] used a different method of assignment that was in line with investigating the in uence of rapid versus slow uid therapy. Among the 8 included observational studies [14,16,[18][19][20][29][30][31], 5 studies [14,18,19,29,31] were assessed as high quality using the modi ed NOS, and another 3 studies[16, 20,30] were assessed as low quality (Additional le 1: Table S2). In terms of the therapeutic protocol for the observational studies, 6 studies [14,18,20,[29][30][31] directly de ned AFT with a clear FV, but in 2 studies [16,19], AFT was de ned as the use of a percentage of the cumulative 72-h FR within the initial 24 h. However, all of the included observational studies ful lled the prede ned criteria for AFT with a FV of no less than 4,000 ml, except for one trial [19]. Categorization of uid type seems to be impossible due to the controversial protocol, but most of them selected a crystal solution as the main resuscitation uid. * Overlapping patients were potentially presented in these studies, but different outcomes were displayed as well.
# Fluid volume was not display directly, but the cumulative total uid volume in rst 24 hour was higher than 4 L. & Fluid rate was lower than 5ml/kg/h, but the cumulative total uid volume in rst 24 hour was higher than 4 L.

Sensitivity Analysis and Publication Bias
The sensitivity analysis results are shown in Table 2. The minimum and maximum values of WMD or OR for every outcome are shown after the sensitivity analysis. With respect to the results for mortality, renal failure, pancreatic necrosis, respiratory failure and persistent SIRS within 24 h, omitting 1 study did not change the overall conclusion of this meta-analysis. However, the overall estimates for organ failure (OR, 1.81; 95% CI, 1.08-3.02; P = 0.02; I 2 = 41%) and LOS (WMD, 2.29; 95% CI, 0.52-4.07; P = 0.01; I 2 = 92%) were changed after omitting the study published by Warndorf et al [19]. In addition, omitting the same trial in the subgroup analysis of average age > 50 years for persistent SIRS yielded a signi cant change (OR, 2.37; 95% CI, 1.12-5.02; P = 0.02; I 2 = 0%). Compared to the initial results for the organ failure and LOS, signi cant differences were changed after sensitivity analysis.
The publication bias for reporting in-hospital morality in this meta-analysis is shown in Fig. 5. There was no obvious publication bias, with a homogeneous distribution around the vertical direction, and the 95% CI included all of the included trials.

Quality of Evidence
The GRADE system was used to rate the level of evidence by study type: RCT and observational (Additional le 1: Table S3 and Table  S4). For the RCTs, there was only one outcome that assessed in-hospital mortality, and the rating was high. The evidence for each of the following outcomes was rated as moderate: incidence of renal failure, pancreatic necrosis, and respiratory failure. The remaining evidences of outcomes for persistent SIRS, organ failure and LOS were rated as low to very low. For the observational studies, there were two evidences for renal failure and pancreatic necrosis were rated as moderate, and the remaining evidences were rated as low to very low.

Discussion
In this meta-analysis of 13 trials incorporating 3,127 patients, treatment with AFT compared with CFT indicated a signi cant disadvantage in terms of in-hospital mortality, renal failure and pancreatic necrosis with relatively strong evidence. Additionally, the incidence of respiratory failure seems to be higher in the AFT subgroup with average age > 50 years. In the sensitivity analysis, omitting one study at a time yielded no signi cant change in the above outcomes with the remaining studies.
Simultaneously, we found no signi cant difference in organ failure, persistent SIRS within 24 h or LOS; however, pooled estimates from the sensitivity analysis were changed after excluding the study conducted by Warndorf et al [19]. Although this study ful lled the inclusion criteria and the risk of bias assessment indicated that the study was high quality, there were several potential factors relevant to the statistical alteration. Two important differences displayed in this study were an FV less than 4,000 ml and unbalanced sample sizes in both study arms. In addition, the calculated standard deviation of LOS may not be normally distributed. As a result, if the factors were indeed shown, a signi cant difference in estimates for organ failure and persistent SIRS could be found, and AFT was potentially related to the increased incidence of organ failure and persistent SIRS in the subgroup of patients aged > 50 years. Nevertheless, the LOS data remained inconclusive, possibly due to the nonnormal distribution published by two other studies [16,30].
In the evolution of AP, there are multiple factors, including vomiting, uid loss in the third space, and reduced oral intake, that could induce hypovolemia. Following the progression of hypovolemia, pancreatic microcirculation could be disturbed, with the result being the release of multiple in ammatory cytokines and pancreatic ischemia and multisystem organ failure. Therefore, improved splanchnic hypovolemia and inhibition of the early in ammatory cascade are the fundamental goals of AFT. In particular, the study published by Buxbaum et al. [13] showed obvious clinical improvement and a decreased incidence of persistent SIRS in the AFT group. However, these ndings should be viewed with caution due to the limited or unbalanced sample sizes and the mild severity of AP.
Broader implementation of aggressive uid treatment should be more cautious according to this meta-analysis. The recommendations for AFT from previous guidelines [5,6] were mainly based on observational trials, expert opinions and experimental ndings. A number of studies [12,13,16,19] reported important outcomes in line with the recommendations in recent years. However, AFT is commonly accompanied by visceral edema [32] and excessive chloride [33], which may contribute to intra-abdominal hypertension and impaired renal function, and both of these results can affect each other simultaneously [34,35]. In addition, rapidly increasing intravenous volume did not change oxygen delivery and oxygen consumption, even though cardiac output was remarkably improved [36]. This type of pathological event might be caused by the hemodilution-induced decrease in hematocrit, which was noted as a risk factor the development of organ failure and pancreatic necrosis [37]. Therefore, clinical advocation for CFT has received general attention because of the non-neglectful improvements in pulmonary/renal failure, pancreatic necrosis, and mortality [15,17,18,29,31,38,39]. Adverse events, which decreased as described above, are potentially related to prior underlying pathology, and the study reported by Warndorf et al [19] might be the best proof of this. Overall, given the estimated homogeneity and signi cant of the results, our ndings strengthen the hypothesis that CFT is not inferior to AFT.
Since the rst meta-analysis on aggressive uid resuscitation for AP by Gad et al. [24], additional evidence to support the bene t of uid therapy has not been reported, even though the report indicated potential inferiority of AFT without exact statistical signi cance. Although we agree with the standpoints from the study, some de ciencies remain that are worthy of attention. First, a lower uid rate threshold of 3-5 ml/kg/h in the rst 24 h was de ned as aggressive uid resuscitation, which was inconsistent and not speci c to prior guidelines [5] that declared aggressive FR as being higher than 10 ml/kg/h or FV as being higher than 4000 ml within 24 h. More than two-thirds of the included studies were observational studies, which means that is impossible to re ne an immutable FV to an exact FR per hour per kilogram. Second, the included studies enrolled two potentially overlapping trials published by Buxbaum et al. [13,40] and Mao et al. [15,17] without de nite reason. Exaggerative outcomes were possibly generated due to this de ciency. Third, for the meta-analysis of SIRS, data were extracted without differentiation of the stage prior to and posttreatment. The study tried to analyze the in uence of uid therapy; however, data [14,31] extracted from the stage prior to treatment should not be regarded as the therapeutic effect. Consequently, our group was motivated to further investigate the bene t of uid therapy.
The following limitations of our meta-analysis should be taken into account. The main limitation is that more than half of the included trials were observational studies, and the remaining 5 RCTs were conducted with small sample sizes. Although in-hospital mortality is unlikely to be in uenced by a lack of blinding [41] due to the complex dynamic evaluation, this evidence, which was rated as having a high risk of bias, should be interpreted with caution due to the small-study effect [42]. The subgroup analysis yielded some signi cant outcomes compared to the original analysis, indicating unstable endpoints because of a speci c study. In addition, most of the present studies did not separate the severity of AP to investigate the bene t of AFT versus CFT. Some previous studies [43][44][45] suggested that the type of uid also played an important role in the prognosis of pancreatitis; however, our research could not completely distinguish the types of liquids used and perform a subgroup analysis. Future systematic analyses should evaluate the severity of AP and uid type when su cient data are available.
Overall, this meta-analysis is the rst report of the disadvantage of AFT and found a timely, conclusive outcome as data have accumulated and become available for inspection by meta-analytical theory. A broad strategy was used to identify eligible studies to increase the sensitivity of the study. Detailed sensitivity and subgroup analysis were performed to minimize the heterogeneity of every outcome. Therefore, up-to-date information was provided in our research.

Conclusion
In adults with AP, compared with CFT, AFT is associated with increased in-hospital mortality and an increased incidence of renal failure and pancreatic necrosis. In patients with an average age > 50 years, AFT is potentially related to an increase in the incidence of respiratory failure. Meanwhile, the increase in persistent organ failure and SIRS within 24 h is potentially in uenced by controversial practices. Nevertheless, several limitations presented above prevent us from drawing de nitive conclusions. Further well-designed, properly powered trials will be required to con rm and update the ndings of this meta-analysis.

Abbreviations
AFT: aggressive uid treatment; CFT: conservative uid treatment; AP: acute pancreatitis; RCT: randomized control trial; PRISMA: Preferred Reporting Items for Systematic Reviews and Meta-analyses; FR: uid rate; FV: uid volume; ERCP: endoscopic retrograde cholangiopancreatography; SIRS: systemic in ammatory response syndrome; LOS: length of hospital stay; GRADE: Grading of Recommendations Assessment, Development, and Evaluation; NOS: Newcastle-Ottawa scale; OR: odds ratio; WMD: weighted mean difference; CI: con dence intervals. SD: standard deviations.

Declarations
Ethics approval and consent to participate: Not applicable. Consent for publication: Not applicable.
Availability of data and materials: All data generated or analysed during this study are included in this published article and its supplementary information les.
Competing interests: The authors declare that they have no competing interests.
Funding: Not applicable.
Authors' contributions: JYL, YZ and JBL contributed to the study conception and design. Literature search, data collection and analysis were performed by YZ, HCW and XP. The rst draft and review of the manuscript was prepared by JYL and YZ. Supervision, material support and project administration were performed by JBL and ZJY. All authors commented on previous versions of the manuscript. All authors read and approved the nal manuscript.    Forest plot and meta-analysis of pancreatic necrosis; CI con dence interval, M-H Mantel-Haenszel, AFT aggressive uid therapy, CFT conservative uid therapy