Efficacy of gastric decompression after pancreatic surgery: a systematic review and meta-analysis

A systematic literature search was conducted in PubMed, EMBASE, Cochrane Central Register of Controlled Trials (CENTRAL), and Web of Science for eligible studies comparing RGD with no gastric decompression (NGD) after pancreatic resection. References of relevant articles were also reviewed to identify potentially eligible studies.

The search strategy was (“pancreaticoduodenectomy” OR “pancreatoduodenectomy” OR “Whipple” OR “pancreatic resection” OR “pancreatectomy” OR “pancreatic surgery”) AND (“nasogastric tube” OR “gastrostomy” OR “gastric decompression”). The search strategy was adapted to the databases accordingly. The last search was conducted on Nov 3, 2019.

All types of original study articles that performed a comparison between RGD and NGD after pancreatic resections were considered. No restrictions were made regarding the methods of gastric decompression (nasogastric tube versus gastrostomy).

Publications were excluded if they meet any of the following criteria: (1) articles were published as case reports, conference abstracts, letters to the editor, or reviews; (2) articles were published in any language other than English; (3) articles compared routine gastric decompression to selective gastric decompression, rather than to no-gastric decompression.

The primary outcomes of this meta-analysis were the incidence of overall complications and major complications after pancreatic resections. Secondary outcomes included delayed gastric emptying (DGE), postoperative pancreatic fistula (POPF), and postoperative mortality. Secondary gastric decompression and the length of hospital stay were also analyzed.

The extracted data included study characteristics (study design, study period, sample size, and investigated surgical procedure), patient characteristics (age, sex, body mass index), and outcome measures [POPF), DGE, hemorrhage, intra-abdominal fluid collection/abscess, bile leakage, wound infection, pneumonia, overall morbidity, mortality, reoperations, length of hospital stay].

For the outcomes of interest, when the continuous variable was reported only as medians and ranges, the methods of Hozo et al. [18] and Wan et al. [19] were applied to calculate means and standard deviations.

When data were reported in more than one article or were analyzed using two statistical methods to analyze patients from the same database, the one with larger data sets were included.

The quality of included studies was assessed using the Newcastle–Ottawa Scale for cohort studies (NOS). Included studies were ranked with a maximum of 9 points, including three parts: “selection” (four elements), “comparability” (one element) and “outcome” (three elements). Cohort studies with an NOS score < 6 were considered of moderate or low quality.

Statistical analyses were performed using Review Manager 5.3 for Windows. For continuous outcomes, weighted mean differences (MD) and corresponding 95% confidence intervals (CI) were calculated by the inverse variance method. For dichotomous outcomes, odds ratios (OR) and the corresponding 95% CI were calculated by the Mantel–Haenszel (MH) method. For the assessment of statistical heterogeneity, I² statistics were performed. When I² > 50%, statistical heterogeneity was considered high. Due to the clinical heterogeneity and for a relatively conservative perspective, a random-effects model was chosen for the meta-analyses regardless of the absence of statistical heterogeneity. Publication bias for the primary outcome was analyzed using funnel plots and Egger’s test. P < 0.05 was considered to be statistically significant, and the 95% CI was set for efficiency measures.

Figure 1 depicts the screening and selection process of the literature in accordance with PRISMA guidelines. Initially, a total of 232 studies were identified from the databases. After discarding duplicates an unrelated study according to the exclusion criteria, seven full-text articles were reviewed to assess further eligibility. In addition, one study was excluded because some patients in the NGD group underwent nasogastric tube insertion (NGT) [20]. Finally, six studies were included in the systematic review and meta-analysis [5, 15, 16, 21‐23]. The final analysis included a total 940 patients, with 484 patients in the RGD group, and 456 patients in the NGD group.

Study quality was assessed using the modification of the Newcastle-Ottawa scale for the cohort study. The median quality score of the studies according to the Newcastle-Ottawa Scale was 7.

All six studies were non-randomized studies, four of which used history as the control. Two studies were conducted in the USA, two in Korea, one in France, and one in Norway. Sample sizes ranged from 41 to 231. The studies were published between 2011 and 2019, and the study period ranged from 1994 to 2016. The baseline characteristics of the six included studies are summarized in Table 1.

This meta-analysis has shown that there was no difference in terms of overall complications, major complications, incidence of secondary gastric decompression, incidence of overall DGE, incidence of CR-POPF and incidence of POPF between RGD group and NGD group following pancreatic surgery. RGD was associated with a higher incidence of CR-DGE, a higher rate of mortality, and a longer length of hospital stay.

These results are similar to those of previous meta-analyses, which showed that nasogastric decompression brings no benefit in non-pancreatic abdominal surgery, such as esophagectomy [7], gastrectomy [8, 10], or colorectal resection [12, 30]. In esophagectomy, in a systematic analysis of 608 patients, Weijs et al. showed no significant difference in adverse outcomes between nasogastric decompression or no nasogastric decompression following esophagectomy [7]. In gastrectomy for gastric cancer, Yang et al., with a meta-analysis of 717 patients from five RCTs, found that time to oral diet was significantly shortened in the no-decompression group, while time to flatus, anastomotic leakage, pulmonary complications, length of hospital stay, morbidity, and mortality were similar in both groups [8]. The finding was further confirmed by Wei et al. [10]. In a meta-analysis of 1141 patients, which found that nasogastric or nasojejunal decompression neither facilitated the recovery of bowel function nor reduced the risk of postoperative complications after gastrectomy for gastric cancer. Although the absence of routine placement of RGD has been clearly proved in other digestive surgeries and is now recommended after pancreatic surgery (including PD) by ERAS [31], routine nasogastric tube decompression is still practiced by many surgeons treating pancreatic cancer. This phenomenon can be attributed to several reasons. First, previous studies on the necessity of RGD after pancreatic resections were single-institution, retrospective studies with relatively small sample sizes. Therefore, the ERAS recommendation is based only on moderate evidence. Second, the high morbidity after pancreatic resection contributes to this practice. DGE is one of the most common complications after pancreatic surgery, especially following PD, which negatively impacts the quality of life, prolongs the hospital stay, and increases hospital costs. Although its pathophysiology remains unclear, it has discouraged many surgeons from abandoning this practice. Routine nasogastric tube placement after abdominal surgery is thought to prevent postoperative nausea and vomiting and abdominal distention by gastric decompression; these are the core symptoms of DGE. Third, because NG tube has been used following gastrointestinal anastomoses for several decades, it is difficult to change the clinical habit and radically stop using routine gastric decompression [32, 33].

Instead of absolutely prohibiting RGD after pancreatic surgery, some surgeons preferred a more conservative method, namely selective NGT usage, such as when they unable to extubate the patient postoperatively [20]. In their retrospective study with 250 patients, Kunstman et al. found that patients in the selective use of RGD had decreased incidence of delayed gastric emptying, length of stay, and time to dietary tolerance [20]. Nevertheless, the authors agreed that RGD could be omitted in many cases.

Previous studies in non-pancreatic surgery have found that pulmonary complications, such as atelectasis and pneumonia, occur more frequently in patients with a nasogastric tube than in those without. These findings were also confirmed in pancreatic resections; however, because only two studies reported this complication, a meta-analysis was not done in this study.

The results of this meta-analysis should be interpreted with caution due to several reasons. First, all six included studies employed a non-randomized design, which carries the potential for selection bias. However, four of the studies used historical controls, which may mitigate the selection bias. Second, there was heterogeneity between the two groups in terms of surgical procedures, histological grades, as well as tumor stage. Third, secondary outcomes were not reported by all the studies. Therefore, many important outcomes, such as pulmonary complications and time to dietary tolerance, were not analyzed, or only a limited number of patients were included for the meta-analysis of secondary outcomes, which might affect the reliability of the results. Finally, some studies did not directly provide means and SDs, and the Hozo algorithm was adopted to estimate means and SDs based on median and range, which may have introduced bias.