Table 2 Risk of bias in individual studies evaluating the effect of glucocorticosteroids for acute pancreatitis
Study | Study species | Study design | LOE I–IV | Group similarity (good, fair, poor) | Estimated risk of bias (high, moderate, low) | Support for Risk of bias judgement | Group size |
|---|---|---|---|---|---|---|---|
Okanishi et al. [57] | Dogs | Non-randomised controlled trial | III | Good (age, sex, breed, diagnosis incl clinical signs + spec-cPLI or plasma lipase, ultrasound, CRP. Baseline variables were similar between groups | Moderate | Selection bias: Non-concealed, non-randomised allocation of participants to GC group 2011–2014, non-GC group 2015–2016. Breeds and ages in groups were not matched. Performance bias: No blinding. V-LIP used in diagnosis is inferior to spec-cPLI. Detection bias: No blinding of outcome assessment. Statistical analysis described. 3 drop-outs excluded from relevant analysis | Moderate (45/20) |
Studley and Schenk [39] Experiment 1 and 2 | Dogs | Randomised controlled trial | II | Poor, age/sex/breed not stated. Diagnosis ok as histopathology confirmed induced pancreatitis | Low | Selection bias: Method of randomisation not described. Performance bias: No blinding, detection bias: No blinding of outcome assessment, but as outcome was survival, the lack of blinding was less worrying. No drop-outs. Stat analysis, but not described | Very small (5–8) |
Attix et al. [58] | Dogs | Controlled trial | III | Poor, age not stated, division of which sex into which group not stated, all “mixed-breed”, induced AP diagnosis was confirmed by histopathology | High | Selection bias: No randomisation/concealment. Performance bias: No blinding of personnel performing subjective clinical evaluation and histopathology. Detection bias: no blinding of outcome assessment. Reporting bias: clinical and pathologic features were to be compared for GC treated and non-GC treated, no results for histopathology were reported. No drop-outs, no statistics | Very small (4/4) |
Stewart et al. [59] | Dogs | Controlled trial | III | Poor, age not stated, division of which sex into which group not stated, all “mixed-breed”, induced AP diagnosis was confirmed by histopathology | High | Selection bias: no randomisation/concealment. Performance: No blinding. Detection bias: No blinding of outcome assessment, however, survival is an objective parameter, histopathology less so. 3 cortisone treated died of GI ulceration biasing the mortality rates. No drop-outs, no statistics | Moderate/small (25/16) |
Imahori et al. [60] | Dogs | Controlled trial | III | Poor, age and sex not stated, all “mixed-breed”, induced AP diagnosis was confirmed by histopathology | Low (High for electron microscopy) | Selection bias: No randomisation/concealment. Blinded histopathologist. No drop-outs, statistics but not explained | Study 1 small (14/14), study 2 very small (5/6) |
Kaplan et al. [65] | Humans | Retrospective case–control study | III | Fair, AP diagnosis by clinical signs and plasma amylase, age and sex stated | High | Selection bias: No randomisation/concealment. No standardisation among groups in diagnostics or treatments. Performance bias: No blinding of personnel or participants. Variation in dose and timing of GC and other treatment. High incidence of gall bladder disease. Detection bias: no blinding of outcome assessment. No drop-outs, no statistics | Small (15/48) |
Liu et al. [40] | Humans | Randomised controlled trial | II | Poor, AP diagnosis by clinical diagnosis (not further elaborated), age and sex of patients not mentioned | Moderate | Selection bias: Method of randomisation not described. Detection bias: Lack of blinding of outcome assessment seems irrelevant, as outcome is mortality. Other bias: Salviae Miltiorrhiza and Dextran was given with GC. In 2 of the 3 deaths in GC group, early preventive treatment was delayed till 3 days after onset of disease. No drop-outs, statistics explained | Moderate (26/53) |
Eklund et al. [37] | Humans | Retrospective case–control study | III | Good, AP diagnosis by clinical signs, amylase and CT, age and sex distribution similar among groups | Low | Selection bias: Only patients with severe acute pancreatitis requiring norepinephrine support for hemodynamic shock were included. Performance bias and detection bias: No blinding, but outcomes CRP and mortality are objective parameters. Other comments: Cause of AP: Alcohol 7, gallstones 3, unusual causes for dogs, should be considered when results are extrapolated for dogs. No drop-outs, statistics | Small (10/11) |
Wang et al. [66] | Humans | Case-series | IV | Fair: AP diagnosis by clinical signs, amylase and ultrasound, CT, age and sex stated | High | Performance and detection bias: No blinding of participants or personnel, who evaluated pain. Reporting bias: Comparison of mortality with literature is questionable. Other bias: Dextran 40 was given along with GC. No drop-outs mentioned, no statistics | Moderate (32) |
Wan et al. [38] | Humans | Randomised controlled trial | II | Good, AP diagnosis by clinical signs, lipase, CT, no statistical difference among groups according to sex, age, disease severity, etiology, complications | Low | Randomisation by SPSS software, drop-out rate not statistically significant different between groups, Performance and detection bias: no blinding stated but all parameters objective. Statistics explained | Moderate (43/38) |
Zhang et al. [41] | Rats | Randomised controlled trial | II | Good, healthy, all male, similar weight rats | Moderate | Selection bias: Method of randomisation not described. Performance and detection bias: No blinding of histopathologist stated, but semiquantative grading of histopathologic findings. Attrition bias: histopathology score of 2 rats from control group was not available due to their death at 12 h. No other drop-outs, statistics explained | Moderate (45/45) |
Schulz et al. [42] | Rats | Randomised controlled trial | II | Good, all male, similar weight rats | Low | Selection bias: Method of randomisation not described. Blinded histopathologist, semiquantative grading of histopathology findings, no drop-outs, statistics explained | Small (15/15) |
Gloor et al. [61] | Rats | Controlled trials | III | Good, all female, similar weight rats | Low | Selection bias: No randomisation stated, blinded histopathologists, semiquantative grading of histopathology findings, no drop-outs, no statistics on histopathology, but statistics on mortality | Moderate (32/32) Moderate (40/40) Moderate/very small (21/8) |
Cosen-Binker et al. [43] | Rats | Randomised controlled trial | II | Good, pathogen-free, all male similar weight rats | Moderate | Selection bias: Method of randomisation not described. Blinded histopathologists, semiquantative grading of histopathological findings, no drop-outs. Reporting bias: Unclear description of results: GC “did not present much improvement”, although results are marked by P < 0,001. Results and statistics not adequately explained | Small (16/16/16) |
Cosen-Binker et al. [44] | Rats | Randomised controlled trial | II | Good, pathogen-free, all male similar weight rats | Low | Selection bias: Method of randomisation not described. Blinded histopathologists, semiquantative grading of histopathology findings, no drop-outs. Reporting bias: Authors state that treatment 4 h post induction is harmful, but results put these in the same disease category as controls. Statistics | Small (16/16/16/16) |
Jha et al. [45] | Rats | Randomised controlled trial | II | Good, all male similar weight rats | High | Selection bias: Method of randomisation not described. Blinded histopathologist. Reporting bias: No results of histopathology score or TEM shown, only photo examples with conclusions, no statistics on histopathology score of GC group or TEM. No drop-outs | Moderate (24/24) |
Melo et al. [46] | Rats | Randomised controlled trial | III | Good, all male similar weight rats | Low | Selection bias: Randomisation not stated. Blinded histopathologist, semiquantative grading of histopathology findings, no drop-outs, statistics | Very small (8/8) |
Ramudo et al. [62] | Rats | Controlled trial | III | Good, all male similar weight rats | Moderate | Selection bias: No randomisation stated. Blinded histopathologists, semiquantative grading of histopathology findings, no drop-outs, statistics | Very small (6/6) |
Ramudo et al. [47] | Rats | Randomised controlled trial | II | Good, all male similar weight rats | Low | Selection bias: Method of randomisation not described. Blinded histopathologists, semiquantative grading of histopathology findings, no drop-outs, statistics | Unknown, group sizes not stated |
Ou et al. [48] | Rats | Randomised controlled trial | II | Good, clean grade, all male similar weight rats | High | Selection bias: Method of randomisation not described. Performance and detection bias: Blinding not stated. Reporting bias: Results of “pathological severity score” are concluded upon, but their definition is not explained. No drop-outs, statistics | Moderate (36/36) |
Zhao et al. [49] | Rats | Randomised controlled trial | II | Good, all male similar weight and age | High | Selection bias: Method of randomisation not described. Performance, detection and reporting bias: Blinding of histopathologist(s) was not stated, no scoring technique described, histopathology results not shown, but conclusions were stated. No research question about histopathology, no statistics on histopathology, no drop-outs | Small (15/15) |
Foitzik et al. [50] | Rats | Randomised controlled trial | II | Good, all male similar weight rats | Moderate | Selection bias: Method of randomisation not described. Performance and detection bias: Blinding of histopathologist(s) was not stated, semiquantitative histopathology scoring was done, Possible attrition bias: histopathological analysis included only those animals that survived the whole experiment. Statistics | Small (10/10/12/10) |
Biradar and Veeresh 2012 [51] | Rats | Randomised controlled trial | II | Good, all male similar weight rats | High | Selection bias: Method of randomisation not described. Blinded histopathologist. Reporting bias: GC not in the research question, used as positive control, no results or conclusion for CRP shown, no results for histopathology score shown, but conclusions stated. No drop-outs, no statistics | Very small (6/6) |
Biradar and Veeresh [52] | Rats | Randomised controlled trial | II | Good, all male similar weight rats | High | Selection bias: Method of randomisation not described. Blinded histopathologist. Reporting bias: GC not in the research question, used as pos control, no results or conclusion for CRP shown, no results for histopathology score shown, but conclusions stated. No drop-outs, no statistics | Very small (8/8) |
Duan et al. [63] | Rats | Controlled trial | III | Good, all male similar weight and age rats | Moderate | Selection bias: Randomisation not stated, only relevant parameter was mortality, which makes the lack of blinding less worrying. No drop-outs, no statistics | Moderate (21/24) |
Wang et al. [36] | Rats | Randomised controlled trial | II | Good, all male similar weight rats | Moderate | Selection bias: Method of randomisation not described. Blinded histopathologists. Reporting bias: Results for histopathology were not shown, but concluded upon. No drop-outs, statistics | Moderate (25/25) |
Liu et al. [53] | Rats | Randomised controlled trial | II | Fair, rats of similar weight, otherwise not described | Low | Selection bias: Method of randomisation not described. Blinded histopathologists. Performance bias: The type of glucocorticoid was not specified making the dose hard to evaluate. No drop-outs, statistics | Small (16/16) |
Sha et al. [54] | Rats | Randomised controlled trial | II | Good, all male similar weight rats | Low | Selection bias: Method of randomisation not described. Blinded histopathologist. Reporting bias: Results are shown for histopathology score and there are conclusions and statistical analysis. For TEM no results are shown, but conclusions are made. No drop-outs | Moderate (24/24) |
Kilic et al. [55] | Rats | Randomised controlled trial | II | Good, rats aged 7–8 months, similar weight | Low | Selection bias: Method of randomisation not described. Blinding, histopathology scoring system, clear results and statistical evaluation, research question was answered. No drop-outs | Very small (8/8) |
Cui et al. [56] | Rats | Randomised controlled trial | II | Good, all male, similar age and weight | High | Selection bias: Method of randomisation not described. Performance, detection and reporting bias: No blinding, no stated histopathology scoring system, no statistical evaluation, seems to be conclusions based on subjective evaluation. No drop-outs | Very small (6/6) |
Zhao et al. [64] | Mice | Controlled trial | III | Good, all male, similar age mice of 2 genotypes | High | Selection bias: No randomisation stated. Performance, detection and reporting bias: No blinding stated, histopathological changes of the pancreas was not included in the aims, but shown by photos and not commented on. Mortality rate was included in the aims, results provided, but no statistical evaluation was done. No drop-outs | Small (15/20/34) |