Predictive value of the Ranson and BISAP scoring systems for the severity and prognosis of acute pancreatitis: A systematic review and meta-analysis.

Acute pancreatitis (AP), an inflammatory disease of the pancreas, is one of the most common gastrointestinal diseases which requires acute hospital admission, with a mounting incidence, significant morbidity and succeeding mortality [ 1 – 3 ]. This disorder exhibits varying severity [ 4 ]. The revised Atlanta Classification and Definitions in 2012 classifies AP as mild, moderately severe and severe acute [ 5 ]. The prognosis of AP patients primarily depends on the progress of organ failure and secondary infection of pancreatic or peripancreatic necrosis [ 6 ]. Besides, patients with AP usually need intensive care unit (ICU) admission, particularly as signs of multi-organ failure appear [ 7 ]. Early severity stratification and prognostic prediction are essential for lowering the mortality rate of AP patients [ 8 ]. There are various scoring systems to evaluate the severity and outcomes of AP. The Ranson score, developed in 1974, is the first scoring system to predict AP, which has been criticized for its low predictive ability and delayed management despite continuous wide application [ 9 , 10 ]. The Bedside Index of Severity in Acute Pancreatitis (BISAP) is a commonly used scoring system at present, and compared with the Ranson, the BISAP can be adopted at admission and has fewer parameters [ 11 ]. A review by Ong and Shelat shows that the Ranson score consistently exhibits similar predictive accuracy to the BISAP scoring system, advocating for the sustained clinical practicability of the Ranson score in modern times [ 9 ]. The Acute Physiology, and Chronic Health Examination II (APACHE II) has been reported to be the most accurate scoring system for predicting mortality [ 12 ]. It is the most widely utilized mortality prediction score among critically ill patients, but it had 12 items with many clinical parameters, so its application may be cumbersome which limits its widespread use. Besides, the APACHE II is devised for patients admitted to the ICU and is therefore not suitable for early prediction of the severity of AP. The BISAP and Ranson were shown to have overlapped AUCs with the APACHE II [ 13 ]. The BISAP and Ranson are determined upon admission or within 48 hours of admission, but in the computed tomography severity index (CTSI) prediction, the recommended timing for CT examination is 72 to 96 hours after symptom onset [ 14 , 15 ]. This may limit the early predictive ability of CTSI, as the BISAP and Ranson can predict severity or mortality early with similar performances [ 12 ]. Additionally, the presence of inter-observer variability can affect the accuracy of CTSI score calculation [ 9 ]. For these reasons, this study paid attention to the early prediction of AP severity and outcomes with the Ranson and BISAP. Several studies have comprehensively evaluated the predictive performance of the BISAP for severity and prognosis in AP [ 16 – 18 ], whereas no specific meta-analysis is conducted for direct comparison between the Ranson and BISAP. Given that the Ranson does not consider imaging data, uses multiple parameters, and misses possibly valuable window for early treatment, whether it is still applicable requires comprehensive quantitative evaluation. This systematic review and meta-analysis aimed to systematically assess and compare the predictive value of the Ranson and BISAP scoring systems for the severity and prognosis of AP based on the revised Atlanta Classification and Definitions in 2012.

A total of 3878 studies were identified, with 908 from PubMed, 946 from Embase, 606 Cochrane Library, and 1418 from Web of Science. After removing duplicates, 1815 studies were included for screening based on titles and abstracts. Then remaining 55 studies were subject to full-text screening. Ultimately, 17 [ 21 – 37 ] studies of 5476 AP patients were eligible for this systematic review and meta-analysis, with 15 studies included for quantitative analysis. Fig 1 presents the selection process of qualified studies. Among the included studies, 4 studies came from China, 4 from India, 4 from Korea, 3 from Pakistan, 1 from Serbia, and 1 from Singapore. The year of publication ranged from 2013 to 2022. Eight articles were designed as prospective studies, and 9 as retrospective studies. The characteristics of the included studies are shown in Table 1 . N, sample size; M/F, male/female; ICU, intensive care unit. For the risk of bias, 16 studies [ 22 – 37 ] had low risks, and 1 study [ 21 ] exhibited a high risk in the patient selection; most studies had unclear risks in the index test; 6 studies [ 24 – 26 , 28 , 32 , 37 ] showed high risks, and 11 [ 21 – 23 , 27 , 29 – 31 , 33 – 36 ] had low risks in the reference standard; all the included studies had low risks in the flow and timing. For clinical applicability, all studies except 1 study [ 35 ] had low risks in the patient selection; 6 studies [ 23 , 26 , 29 , 30 , 34 , 36 ] had high risks, and 11 studies [ 21 , 22 , 24 , 25 , 27 , 28 , 31 – 33 , 35 , 37 ] had low risks in the index test; 8 studies [ 24 – 26 , 28 , 30 , 32 , 35 , 37 ] had high risks, and 9 [ 21 – 23 , 27 , 29 , 31 , 33 , 34 , 36 ] showed low risks in the reference standard ( Table 2 ). L, low risk; H, high risk; U, unclear risk. Four studies [ 21 , 22 , 31 , 33 ] provided data on the Ranson and BISAP for severity. The SROC curves of the Ranson and BISAP did not present “shoulder-arm” distributions, indicating no threshold effects ( Fig 2 ). Further, the Spearman correlation coefficient for the Ranson and BISAP was -0.8 ( P = 0.2) and 0.6 ( P = 0.4), respectively, which confirmed the absence of threshold effects. SROC, Summary receiver operating characteristic; BISAP, Bedside Index of Severity in Acute Pancreatitis; AP, acute pancreatitis. For the Ranson, the pooled sensitivity was 0.95 (95%CI: 0.87, 0.98); the pooled specificity was 0.74 (0.52, 0.88); the pooled PLR was 3.64 (95%CI: 1.74, 7.62); the pooled NLR was 0.07 (95%CI: 0.02, 0.21); the pooled DOR was 51.66 (95%CI: 9.28, 287.66), and the pooled AUC was 0.95 (95%CI: 0.93, 0.97). For the BISAP, the pooled sensitivity was 0.67 (95%CI: 0.27, 0.92); the pooled specificity was 0.95 (95%CI: 0.85, 0.98); the pooled PLR was 12.71 (95%CI: 4.07, 39.69); the pooled NLR was 0.35 (95%CI: 0.11, 1.09); the pooled DOR was 36.50 (95%CI: 5.57, 239.39), and the pooled AUC was 0.94 (95%CI: 0.92, 0.96). No significant difference was found in the pooled AUC between the Ranson and BISAP ( P = 0.480) ( Table 3 , Fig 2 , S1 Fig ). P (AUC): P value for AUC. AP, acute pancreatitis; AUC, area under curve; DOR, diagnostic odds ratio; NLR, negative likelihood ratio; PLR, positive likelihood ratio; SEN, sensitivity; SPE, specificity; BISAP, Bedside Index of Severity in Acute Pancreatitis; ICU, intensive care unit. Cho et al. [ 27 ] showed that the AUC of the Ranson and BISAP was 0.848 (95%CI: 0.77–0.92) and 0.826 (95%CI: 0.74–0.92), respectively. In the study of Lee et al. [ 27 ], the AUC of the Ranson and BISAP was 0.76 (95%CI: 0.64–0.87) and 0.66 (95%CI: 0.50–0.82), respectively. Comparison of the Ranson and BISAP for mortality was assessed by 12 studies [ 22 , 23 , 26 , 28 – 33 , 35 – 37 ]. No threshold effects were shown according to no “shoulder-arm” distributions in the SROC curves ( Fig 3 ). The Spearman correlation coefficient for the Ranson and BISAP was 0.516 ( P = 0.086) and 0.427 ( P = 0.167), respectively, further suggesting the absence of threshold effects. SROC, Summary receiver operating characteristic; BISAP, Bedside Index of Severity in Acute Pancreatitis; AP, acute pancreatitis. For the Ranson, the pooled sensitivity was 0.89 (95%CI: 0.73, 0.96); the pooled specificity was 0.79 (95%CI: 0.68, 0.87); the pooled PLR was 4.22 (95%CI: 2.74, 6.50); the pooled NLR was 0.14 (95%CI: 0.05, 0.36); the pooled DOR was 30.37 (95%CI: 10.69, 86.29), and the pooled AUC was 0.91 (95%CI: 0.88, 0.93). For the BISAP, the pooled sensitivity was 0.77 (95%CI: 0.58, 0.89); the pooled specificity was 0.90 (95%CI: 0.86, 0.93); the pooled PLR was 8.00 (95%CI: 5.56, 11.50); the pooled NLR was 0.25 (95%CI: 0.13, 0.50); the pooled DOR was 31.59 (95%CI: 13.56, 73.58), and the pooled AUC was 0.92 (95%CI: 0.90, 0.94). No significant difference was found in the pooled AUC between the Ranson and BISAP ( P = 0.480) ( Table 3 , Fig 3 , S2 Fig ). Information about organ failure was shown in 4 studies [ 24 , 25 , 28 , 34 ]. No “shoulder-arm” distributions in the SROC curve illustrated that there were no threshold effects ( Fig 4 ). The Spearman correlation coefficient for the Ranson and BISAP was 0.000 ( P = 1.000) and 0.949 ( P = 0.051), respectively, confirming no threshold effects. SROC, Summary receiver operating characteristic; BISAP, Bedside Index of Severity in Acute Pancreatitis; AP, acute pancreatitis. For the Ranson, the pooled sensitivity was 0.84 (95%CI: 0.76, 0.90); the pooled specificity was 0.84 (95%CI: 0.63, 0.94); the pooled PLR was 5.18 (95%CI: 1.99, 13.53); the pooled NLR was 0.19 (95%CI: 0.11, 0.31); the pooled DOR was 27.40 (95%CI: 7.41, 101.33), and the pooled AUC was 0.86 (95%CI: 0.82, 0.88). For the BISAP, the pooled sensitivity was 0.78 (95%CI: 0.60, 0.90); the pooled specificity was 0.90 (95%CI: 0.72, 0.97); the pooled PLR was 7.64 (95%CI: 3.01, 19.41); the pooled NLR was 0.24 (95%CI: 0.13, 0.44); the pooled DOR was 31.64 (95%CI: 15.69, 63.83), and the pooled AUC was 0.90 (95%CI: 0.87, 0.93). No significant difference was found in the pooled AUC between the Ranson and BISAP ( P = 0.110) ( Table 3 , Fig 4 , S3 Fig ). Five studies [ 24 , 25 , 28 , 33 , 37 ] evaluated the Ranson and BISAP for pancreatic necrosis. The SROC curves did not show “shoulder-arm” distributions, suggesting no threshold effects ( Fig 5 ). The Spearman correlation coefficient for the Ranson and BISAP was -0.8 ( P = 0.104) and -0.6 ( P = 0.285), respectively, further indicating the absence of threshold effects. SROC, Summary receiver operating characteristic; BISAP, Bedside Index of Severity in Acute Pancreatitis; AP, acute pancreatitis. For the Ranson, the pooled sensitivity was 0.63 (95%CI: 0.35, 0.84); the pooled specificity was 0.90 (95%CI: 0.77, 0.96); the pooled PLR was 6.04 (95%CI: 1.78, 20.54); the pooled NLR was 0.42 (95%CI: 0.19, 0.92); the pooled DOR was 14.48 (95%CI: 2.02, 103.94), and the pooled AUC was 0.87 (95%CI: 0.84, 0.90). For the BISAP, the pooled sensitivity was 0.63 (95%CI: 0.23, 0.90); the pooled specificity was 0.93 (95%CI: 0.89, 0.96); the pooled PLR was 8.97 (95%CI: 3.98, 20.19); the pooled NLR was 0.40 (95%CI: 0.13, 1.19); the pooled DOR was 22.39 (95%CI: 3.64, 137.79), and the pooled AUC was 0.93 (95%CI: 0.91, 0.95). A significant difference was found in the pooled AUC between the Ranson and BISAP ( P = 0.001) ( Table 3 , Fig 5 , S4 Fig ). Three studies [ 25 , 31 , 34 ] were identified for ICU admission. A “shoulder-arm” distribution was demonstrated in the SROC curve for the Ranson, indicating the existence of a threshold effect. The Spearman correlation coefficient for the Ranson was 1.000 ( P = 0.000), which further confirmed the presence of threshold effects. No “shoulder-arm” distribution and a Spearman correlation coefficient of 0.5 ( P = 0.667) for the BISAP suggested that no threshold effect existed ( Fig 6 ). SROC, Summary receiver operating characteristic; BISAP, Bedside Index of Severity in Acute Pancreatitis; ICU, intensive care unit; AP, acute pancreatitis. For the Ranson, the pooled sensitivity was 0.86 (95%CI: 0.77, 0.92); the pooled specificity was 0.58 (95%CI: 0.55, 0.61); the pooled PLR was 2.93 (95%CI: 1.43, 6.00); the pooled NLR was 0.23 (95%CI: 0.14, 0.38); the pooled DOR was 26.80 (95%CI: 5.31, 135.23), and the pooled AUC was 0.92 (95%CI: 0.81, 1.00). For the BISAP, the pooled sensitivity was 0.63 (95%CI: 0.52, 0.73); the pooled specificity was 0.84 (95%CI: 0.81, 0.86); the pooled PLR was 3.50 (95%CI: 1.70, 7.24); the pooled NLR was 0.47 (95%CI: 0.21, 1.07); the pooled DOR was 7.68 (95%CI: 2.49, 23.74), and the pooled AUC was 0.86 (95%CI: 0.67, 1.00). No significant difference was found in the pooled AUC between the Ranson and BISAP ( P = 0.592) ( Table 3 , Fig 6 , S5 Fig ). Publication bias was evaluated for the outcome morality. The Deeks’ funnel plot asymmetry test showed that there was no publication bias for the Ranson (T = -0.66, P = 0.525) ( S6 Fig ), and for the BISAP, publication bias may also not exist (T = 2.23, P = 0.05) ( S7 Fig ).

The Ranson score exhibited a similar predictive performance to the BISAP, and had greater sensitivity but lower specificity than the BISAP, indicating the applicability and accuracy of the Ranson in predicting severity and prognosis of AP patients under limited resources and time. More large-scale studies are warranted to confirm these findings.

PubMed, Embase, Cochrane Library, and Web of Science were systematically searched by two independent authors (Y Wang, MD Fang). The last search was performed on February 15, 2023. English search terms included “Pancreatitis” OR “Pancreatitides” OR “Acute Pancreatitis” OR “Acute Pancreatitides” OR “Pancreatic Parenchymal Edema” OR “Pancreatic Parenchyma with Edema” OR “Peripancreatic Fat Necros*” AND “Ranson” OR “BISAP” OR “Bedside Index for Severity In Acute Pancreatitis”. Endnote X9 (Clarivate, Philadelphia, PA, USA) was applied for primary screening based on titles and abstracts of the retrieved studies, followed by screening according to full texts. Disagreements were settled by another author (JP Zhu) to reach a consensus. This systematic review and meta-analysis was conducted following the reporting guidelines of Meta-analysis of Observational Studies in Epidemiology (MOOSE). (a) studies on patients with AP of varying severity [ 5 ]; (b) studies reporting the BISAP versus Ranson with a cutoff point at 3 [ 17 , 19 ]; (c) studies on at least one of the following outcomes: severity, mortality, organ failure, pancreatic necrosis, and ICU admission; (d) observational studies; (e) studies providing relevant data to calculate indicators such as sensitivity and specificity or area under the curve (AUC) values; (f) English literature; (g) studies published since 2012. (a) studies on patients with chronic or recurrent pancreatitis, pregnant or lactating patients, or patients who had been hospitalized for less than 48 hours; (b) studies reporting the BISAP versus Ranson with an unclear cutoff point or a cutoff point that was not 3; (c) meta-analyses, reviews, meeting abstracts, animal experiments, case reports, letters, or comments. Outcomes in this analysis included severity and prognosis (mortality, organ failure, pancreatic necrosis, and ICU admission). Severity was defined as persistent single or multiple organ failure for more than 48 h. Organ failure was defined as two or more points in one of the cardiovascular, renal and respiratory systems in the modified Marshall scoring system. Pancreatic necrosis was defined as the absence of enhanced pancreatic parenchyma on contrast-enhanced computed tomography (CECT). Two authors (Y Wang, MD Fang) independently collected data from eligible studies, including the first author, year of publication, author’s country, study period, study design, sample size (N), sex (male/female), age (years), etiology, and endpoints. Disagreements were resolved by discussion with a third author (LF Wu). The revised Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool was used to evaluate the quality of diagnostic accuracy studies, based on the risk of bias and clinical applicability [ 20 ]. The risk of bias includes patient selection, index test, reference standard, and flow and timing. Clinical applicability includes patient selection, index test, and reference standard. Each item was classified as high (risk), low (risk), or unclear (risk). Statistical analysis was carried out by Meta-disc 1.4 (Clinical Biostatistics, Ramony Cajal Hospital, Madrid, Spain), Stata 15.1 (Stata Corporation, College Station, Texas, USA), and Revman 5.4 (The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark). Results were obtained through direct extraction or indirect calculation. Meta-disc 1.4 was adopted to determine whether there was a threshold effect. When the Spearman correlation coefficient between the logarithm of sensitivity and the logarithm of 1-specificity showed a strong positive correlation, it indicated the existence of a threshold effect. To measure the predictive performance of the Ranson and BISAP, Stata 15.1 was employed to assess the sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), and diagnostic odds ratio (DOR) as well as 95% confidence intervals (CI) for clinical outcomes using a bivariate model. Summary receiver operating characteristic (SROC) curves were generated, and the AUC was calculated with 95%CI. The DeLong test was used for AUC comparisons. For the outcome evaluated by over 9 studies, publication bias was assessed using the Deeks’ funnel plot asymmetry test via Stata 15.1. Publication bias was neglected when a funnel plot was symmetric. Revman 5.4 was applied to draw a quality assessment chart for the included studies. P <0.05 was deemed as statistically significant differences.

BISAP, Bedside Index of Severity in Acute Pancreatitis; AP, acute pancreatitis. (TIF) BISAP, Bedside Index of Severity in Acute Pancreatitis; AP, acute pancreatitis. (TIF) BISAP, Bedside Index of Severity in Acute Pancreatitis; AP, acute pancreatitis. (TIF) BISAP, Bedside Index of Severity in Acute Pancreatitis; AP, acute pancreatitis. (TIF) (a) Sensitivity of the Ranson; (b) Specificity of the Ranson; (c) Sensitivity of the BISAP; (d) Specificity of the BISAP. BISAP, Bedside Index of Severity in Acute Pancreatitis; ICU, intensive care unit; AP, acute pancreatitis. (TIF) AP, acute pancreatitis. (TIF) AP, acute pancreatitis. (TIF) (DOCX) (DOCX)