Based on the search strategy and inclusion criteria, a total of 1.162 references were identified through database searching. After excluding 477 duplicates, 685 records were identified. Grey literature review did not add any further records. Another 458 articles were further excluded due to violation of eligibility criteria. The selection process led to 227 articles eligible for full-text reading. Following exclusions (n = 216), reported in Fig. 1, and the adjunct of two studies identified by citation searching, 13 studies were considered eligible for quantitative synthesis, all published between 2004 and 2022 [25,26,27,28,29,30,31,32,33,34,35,36,37]. The inter-rater reliability for reviewers had a k value exceeding 0.90.

Fig. 1

Flow chart of the study selection process according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (PRISMA 2020)

Study characteristics

Eight studies were RCTs [25, 26, 30,31,32,33,34,35], and five were prospective or retrospective n-RCTs [27,28,29, 36, 37]. All studies together analysed data from 13,709 patients (6971 who received LMWH in the Intervention group and 6738 who received standard treatment in the Control group). Limiting the analysis to RCTs, 670 patients were allocated to the Intervention group and 670 to the Control group. The study characteristics are summarised in Table 1, while the baseline characteristics of the patients included in the systematic review and meta-analysis are reported in Supplementary Table 2. Two studies were conducted in India [25, 26], two in the United States [27, 28], six studies in China [29,30,31,32,33,34], one in Turkey [35], whereas for two studies, presented as conference abstracts, no country of origin was identified [36, 37]. Two studies [32, 33] were conducted on the same population of patients. However, the reviewers included both in the pooled analysis as different outcomes were analysed. All records analysed the outcomes of prophylactic LMWH compared with standard treatment without LMWH in patients with moderately-severe and severe AP from biliary origin, alcohol, idiopathic or other causes. One study [30] included 560 children who satisfied the diagnostic and classificatory criteria for AP. Sensitivity analyses were performed to assess the impact of this study on the final results.

Table 1 General characteristics of the studies included in the systematic review and meta-analysis

Five studies focused on patients with severe AP [26, 30, 31, 33, 37], whereas three studies included patients with both moderately-severe and severe AP [25, 35, 36]. Zhou et al. [29] included patients with pancreatic necrosis without further indications of the stage of AP. For three studies, the AP stage was not specified [27, 28, 34]. Considerable heterogeneity among included studies concerning diagnostic criteria for moderately-severe and severe AP was found. Marked heterogeneity was also demonstrated in the dose and length of LMWH administration and outcomes evaluated. Overall, similar exclusion criteria were found, except for Han et al. [30], whose analysis was limited to children. The presence of chronic pancreatitis represented exclusion criteria in the study by Patil et al. [25] and Tozlu et al. [35]. Hypersensitivity to LMWH, pregnancy, lactating and coagulation disorders and end-stage chronic disease with Charlson’s Comorbidity Index ≥ five were listed as exclusion criteria in all the included studies. Type, dosage and length of LMWH treatment varied within the included studies, ranging from standard prophylactic regimens (1 mg/kg body weight LMWH subcutaneously, 12 hourly for seven days) to routine continuous, systematic anticoagulation with a therapeutic dose of LMWH (2 mg/Kg per day, or 40 mg/per day) (Table 1).

Risk of bias

Figures 2 and 3 present the risk of bias analysis. Regarding the RCTs, outcomes might have been influenced by bias arising from the randomisation process, as methods of patient allocation showed substantial variability and some concerns in the majority of included trials, deviations from intended interventions [31, 32, 34], missing outcome data [32, 34], measurement of the outcome [26, 34, 35], and selection of the reported results [26, 32]. Overall, only one RCT was considered at low risk of bias [25]. In contrast, three studies [30, 32, 35] and four studies [26, 31, 32, 34] were considered to have some concerns and at high risk of bias, respectively. Among the five n-RCTs, two were considered at moderate risk of bias [27, 29] and one at serious risk of bias [28]. For two studies [36, 37], no information was available to determine the risk of bias. Graphically, potential publication bias was not present for mortality, occurrence of organ failure, and need for any operative intervention during treatment. Funnel plots are available in Supplementary Fig. 1.

Fig. 2

ROB2 Risk of Bias assessment of randomised controlled trials included in the systematic review and meta-analysis

Fig. 3

ROBINS-I Risk of Bias assessment of non-randomised controlled trials included in the systematic review and meta-analysis

Baseline patient characteristics

Patients in the Intervention and Control groups did not significantly differ in terms of baseline characteristics regarding sex, age, aPTT, Balthazar score, Charlson’s comorbidity index, Computed Tomography Severity Index (CTSI), platelet count, serum amylase, and serum bilirubin. Conversely, albumin (SMD = 0.47, 95% CI = 0.21; 0.73, P = 0.0004, I2 = 0%) and APACHE II score (MD = 0.08, 95% CI = 0.05; 0.12, P < 0.00001, I2 = 0%) were higher in the Intervention group. PT time (MD= − 0.09, 95% CI = − 0.15; − 0.03, P = 0.002, I2 = 0%) was longer in the Control group (Supplementary Fig. 2, Supplementary Table 3, Supplementary Table 4).

Analysis of the primary outcomes

The comparison of Intervention and Control groups showed the superiority of LMWH to standard treatments in terms of overall mortality (11 studies, 13.320 patients; RR = 0.44, 95% CI = 0.31; 0.64, P < 0.0001, I2 = 51%, Test for subgroup differences P < 0.0001, I2 = 93.5%), mortality in severe AP (3 studies, 456 patients; RR = 0.33, 95% CI = 0.20; 0.54, P < 0.00001, I2 = 0%, Test for subgroup differences P = 0.94, I2 = 0%), acute necrotic collections (2 studies; 240 patients; RR = 0.24, 95% CI = 0.09; 0.62, P = 0.003, I2 = 0%), CTSI (5 studies, 1165 patients; MD = − 1.38, 95% CI= − 2.26; − 0.50, P = 0.002, I2 = 93%), organ failure (8 studies, 13,124 patients; RR = 0.67, 95% CI = 0.48; 0.93, P = 0.02, I2 = 78%, Test for subgroup differences P = 0.94, I2 = 0%), walled-off pancreatic necrosis (2 studies, 240 patients; RR = 0.24, 95% CI = 0.11; 0.53, P = 0.0004, I2 = 0%), number of patients with progress in CTSI (2 studies, 240 patients; RR = 0.24, 95% CI = 0.12; 0.47, P < 0.0001, I2 = 0%), and shock (3 studies, 12,090 patients; MD = 0.84, 95% CI = 0.72; 0.98, P = 0.02, I2 = 0%, Test for subgroup differences P = 0.42, I2 = 0%) (Fig. 4, Supplementary Table 5).

Fig. 4

Meta-analysis of primary outcomes: A Mortality; B Acute necrotic collections; C CTSI; D Organ failure; E Walled-off pancreatic necrosis; F Shock

Subgroup analyses, performed according to study design, showed RCTs yielding a pooled effect favouring the Intervention group for mortality (7 studies, 1.340 patients; RR = 0.34, 95% CI = 0.25; 0.46, P < 0.00001, I2 = 0%), whereas organ failure was not statistically different (6 studies, 1.299 patients; RR = 0.60, 95% CI = 0.35; 1.02, P = 0.06, I2 = 76%).

Analysis of the secondary outcomes

In the analysis of secondary outcomes, the Intervention group showed superior outcomes compared with the Control group for APACHE II score (4 studies, 980 patients MD = − 1.81, 95% CI = − 2.63; − 0.98, P < 0.001, I2 = 90%, Test for subgroup differences P = 0.03, I2 = 79.9%), need for ICU admission (2 studies, 11,962 patients; RR = 0.77, 95% CI = 0.67; 0.89, P = 0.0003, I2 = 0%, Test for subgroup differences P = 0.89, I2 = 0%), mean time of tolerance to diet (3 studies, 374 patients; MD = − 2.55, 95% CI = − 4.99; − 0.12, P = 0.04, I2 = 98%), and systemic complications during treatment (3 studies, 933 patients; RR = 0.42, 95% CI = 0.32; 0.55, P < 0.00001, I2 = 0%, Test for subgroup differences P = 0.70, I2 = 0%). Conversely, regarding the rate of cured illness (3 studies, 925 patients; RR = 1.20, 95% CI = 0.87; 1.66, P = 0.26, I2 = 97%) and local complications (5 studies, 12,165 patients; RR = 0.62, 95% CI = 0.25; 1.55, P = 0.31, I2 = 84%, Test for subgroup differences P = 0.64, I2 = 0%), the two study groups showed no statistically significant difference (Supplementary Fig. 3, Supplementary Table 6).

The Intervention group showed superior outcomes compared with the Control group for gastrointestinal bleeding (4 studies, 1487 patients; RR = 0.64, 95% CI = 0.44; 0.94, P = 0.02, I2 = 0%, Test for subgroup differences P = 0.46, I2 = 0%), length of hospital stay (7 studies, 13,024 patients; MD = − 6.08, 95% CI = − 10.08; − 2.07, P = 0.003, I2 = 98%, Test for subgroup differences P = 0.48, I2 = 0%), need for operative interventions (any type) (7 studies, 1354 patients; RR = 0.50, 95% CI = 0.29; 0.87, P = 0.01, I2 = 61%, Test for subgroup differences P = 0.40, I2 = 0%), need for surgical intervention (3 studies, 959 patients; RR = 0.39, 95% CI = 0.24; 0.65, P = 0.0003, I2 = 0%), and vascular thrombosis (5 studies, 1167 patients; RR = 0.43, 95% CI = 0.31; 0.61, P < 0.00001, I2 = 0%, Test for subgroup differences P = 0.10, I2 = 63.2%).

Conversely, regarding the need for endoscopic intervention (3 studies, 255 patients; RR = 0.72, 95% CI = 0.27; 1.93, P = 0.51, I2 = 57%, Test for subgroup differences P = 0.77, I2 = 0%), Balthazar score (2 studies, 240 patients; MD = 0.01, 95% CI= − 0.11; 0.12, P = 0.90, I2 = 0%) and necrosis score (2 studies, 240 patients; MD = 0.02, 95% CI= − 0.31; 0.36, P = 0.89, I2 = 0%) the two study groups showed no statistically significant difference (Supplementary Fig. 4, Supplementary Table 7).

Subgroup analyses, performed according to study design, showed that in RCTs the Intervention group was superior in terms of APACHE II score (2 studies, 825 patients; MD = − 1.10, 95% CI = − 1.39; − 0.81, P < 0.00001, I2 = 0%), systemic complications (2 studies, 660 patients; RR = 0.42, 95% CI = 0.32; 0.55, P < 0.00001, I2 = 5%), vascular thrombosis (3 studies, 505 patients; RR = 0.15, 95% CI = 0.04; 0.55, P = 0.004, I2 = 0%), and need for operative intervention (5 studies, 1199 patients; RR = 0.42, 95% CI = 0.25; 0.69, P = 0.0007, I2 = 30%). Conversely, equivalent results between the two study groups were found for local pancreatic complications (3 studies, 340 patients; RR = 0.51, 95% CI = 0.15; 1.68, P = 0.27, I2 = 89%) and gastrointestinal bleeding (2 studies, 825 patients; RR = 0.54, 95% CI = 0.25; 1.18, P = 0.12, I2 = 45%).

Analysis of the laboratory parameters after treatment

Regarding laboratory parameters after treatment, no statistically significant difference was found between the Intervention and Control groups, in terms of platelets and albumin. Conversely, PT time (3 studies, 664 patients; MD = 0.72, 95% CI = 0.17; 1.27, P = 0.01, I2 = 81%) was longer in the Intervention group (Supplementary Fig. 5, Supplementary Table 8).

Sensitivity and subgroup analyses

The results of the sensitivity analyses done using the leave-one-out method showed that mortality was not impacted by any of the included studies, similar to what happened in the analyses of organ failure and the need for any type of operative intervention during treatment. However, the benefits of LMWH were less evident, in particular for organ failure and the need for operative treatment, when the study by Han et al. [30] was excluded.

Considering the different stages of the disease (moderately-severe vs. severe AP), the results of the subgroup analysis showed superior outcomes for the Intervention group in terms of mortality (8 studies, 1454 patients; RR = 0.34, 95% CI = 0.25; 0.45, P < 0.00001, I2 = 0%, Test for subgroup differences P = 0.48, I2 = 0%). The subgroup analysis showed superior results for the Intervention group in moderately-severe (3 studies, 338 patients; RR = 0.20, 95% CI = 0.05; 0.87, P = 0.03, I2 = 0%) and severe AP (5 studies, 1116 patients; RR = 0.34, 95% CI = 0.25;0.47, P < 0.00001, I2 = 0%). Regarding the need for any intervention (7 studies, 1354 patients; RR = 0.50, 95% CI = 0.29; 0.87, P = 0.01, I2 = 61%, Test for subgroup differences P = 0.93, I2 = 0%), the Intervention group was again superior. However, the subgroup analysis showed equivalent results between the two study groups for moderately-severe (3 studies, 338 patients; RR = 0.48, 95% CI = 0.21; 1.14, P = 0.10, I2 = 44%) and severe AP (4 studies, 1.016 patients; RR = 0.51, 95% CI = 0.22; 1.18, P = 0.12, I2 = 75%). The incidence of organ failure in the two groups was equivalent (6 studies, 1229 patients; RR = 0.64, 95% CI = 0.34; 1.21, P = 0.17, I2 = 74%, Test for subgroup differences P = 0.81, I2 = 0%). The subgroup analysis based on the disease severity resulted in similar results in the two study groups for moderately-severe (2 studies, 240 patients; RR = 0.45, 95% CI = 0.03; 6.93, P = 0.57, I2 = 67%) and severe AP (4 studies, 1059 patients; RR = 0.64, 95% CI = 0.32; 1.28, P = 0.21, I2 = 82%) (Supplementary Fig. 6).

In the subgroup analysis on the duration of LMWH (< 8 days vs. 8–14 days), the Intervention