Non-liquid as initial meal in mild acute pancreatitis: Renewed meta-analysis.

In this study, we first evaluated that all of the studies included were randomized controlled trials (RCTs). Second, the number of patients in the present meta-analysis is larger than before, so the conclusion is more convincing.

We have reviewed the article titled, “Three initial diets for management of mild acute pancreatitis: a meta‐analysis”,1 which strongly attracted our interest. This meta‐analysis indicated that the non‐liquid, soft, solid diets did not increase pain recurrence after re‐feeding when compared with the clear‐liquid diet. The non‐liquid diets could reduce the length of hospitalization. While these results differ greatly from the common clinical practice, the time and method of resumption of feeding after mild acute pancreatitis (MAP) is based upon resolution of the signs and symptoms of acute pancreatitis, and patients are typically placed on a clear liquid diet. If the diet is tolerated (such as no recurrence of pain or vomiting), the diets of patients are extended to full liquids or low‐fat solids. The previous meta‐analysis included three papers,2, 3, 4 considering unclear subgroups, significant heterogeneity, a small sample size, and a confusing conclusion. We performed a renewed meta‐analysis to take the non‐liquid as the initial meal in MAP included in the latest paper.5

We identified all randomized controlled trials (RCTs) concerning the diets for managing MAP by searching the PubMed/Medicine, Embase, Cochrane Central Register of Controlled Trials (CENTRAL) databases up to February 2017; data were analyzed using Stata12.0 (StataCorp, College Station, TX, USA) and pooled for summary estimates. We expressed results of dichotomous outcomes as relative risk (RR) and mean difference (MD) with 95% CIs for continuous outcomes. Heterogeneity assessment was performed using the I 2 index and chi‐square test. We used a randomized‐effect model for calculations of summary estimates. The primary outcome was the post‐re‐feeding length of hospitalization (PRLOH) and the secondary outcomes were recurrence of pain and total length of hospitalization (TLOH).

Four RCTs and 492 patients were included in this meta‐analysis. All papers had a Jadad scoring of 5 (Table 1), and non‐liquid diets significantly decreased the PRLOH and TLOH when compared to liquid diets (−0.76 [−1.33, −0.19], I 2 = 87.2%, P < 0.05; −0.70 [−1.27, −0.13], I 2 = 81.9%, P < 0.05) (Fig. 1). But the subgroup showed no difference in PRLOH and TLOH for soft diets compared to liquid diets (−0.12 [−1.38, 0.13], I 2 = 89.6%, P < 0.05; −0.63 [−1.35,−0.08], I 2 = 88.5%, P < 0.05). There was no difference in the recurrence of pain after re‐feeding, when comparing the non‐liquid diets with clear‐liquid diets (RR = 1.09 [0.70, 1.70] I 2 = 0, P = 0.804].

Table 1

Results on length of hospitalization, TLOH, and recurrence of pain

Type	Moraes et al.2			Jacobson et al.3		Sathiaraj et al.4		Rajkumar et al.5
	Solid	Soft	Liquid	Solid	Liquid	Soft	Liquid	Soft	Liquid
Number	70	70	70	55	66	49	52	30	30
PRLOH (days)	5.8 ± 1.1	7.4 ± 1.5	7.3 ± 1.6	1.71 ± 2.04	1.68 ± 1.85	4.18 ± 2.86	6.75 ± 3.37	4.23 ± 2.08	6.91 ± 2.43
TLOH (days)	7.5 ± 3.5	8.2 ± 2.4	8.2 ± 2.6	4 (3–6)	4 (3–5)	5.92 ± 2.978	8.71 ± 4.995	1.96 ± 1.63	4.10 ± 1.64
Recurrence of pain (n)	15	12	14	6	4	4	3	6	6

PRLOH, post‐re‐feeding length of hospitalization; TLOH, total length of hospitalization.

Figure 1

Outcomes in non‐liquid diet versus liquid diet with length of hospitalization and total length of hospitalization.

The results of the renewed meta‐analysis were similar to the former, as this result subverts our traditional diets.6, 7, 8 Some people believe that eating too early can increase the burden on the pancreas and worsen the symptoms.9, 10 Because the sample size was larger than before, these data are also useful. Before the non‐liquid diets can be recommended as the preferred re‐feeding treatment option for MAP, some aspects need to be considered. First, the types and dosages of no‐liquid diets used in trials have been varied and it is currently unclear which specific species, strain, dose, and regimens are the most efficient, meanwhile these different diets may be the major causes of significant heterogeneity. Second, once treatments with no‐liquid diets are initiated, the duration and end point of treatment remain unclear. Third, the pathophysiological mechanisms of the aforementioned results are unclear.

Future research should focus on these aspects to better elucidate how diets best fit into the treatment algorithm for MAP.