November 2004, VOLUME 183
NUMBER 5

Recommend & Share

November 2004, Volume 183, Number 5

Abdominal Imaging

A Modified CT Severity Index for Evaluating Acute Pancreatitis: Improved Correlation with Patient Outcome

+ Affiliations:
1Department of Radiology, Division of Abdominal Imaging and Intervention, Brigham and Women's Hospital, 75 Francis St., Boston, MA 02115.

2Department of Health Care Policy, Harvard Medical School, Boston, MA 02115.

3Department of Medicine, Division of Gastroenterology, Brigham and Women's Hospital, Harvard Medical School, Boston, MA 02115.

Citation: American Journal of Roentgenology. 2004;183: 1261-1265. 10.2214/ajr.183.5.1831261

ABSTRACT
Next section

OBJECTIVE. This study was conducted to assess the correlation with patient outcome and interobserver variability of a modified CT severity index in the evaluation of patients with acute pancreatitis compared with the currently accepted CT severity index.

MATERIALS AND METHODS. Of 266 consecutive patients diagnosed with acute pancreatitis during a 1-year period, 66 underwent contrast-enhanced MDCT within 1 week of the onset of symptoms. Three radiologists who were blinded to patient outcome independently scored the severity of the pancreatitis using both the currently accepted and modified CT severity indexes. The modified index included a simplified assessment of pancreatic inflammation and necrosis as well as an assessment of extrapancreatic complications. Outcome parameters included the length of hospital stay; the need for surgery or percutaneous intervention; and the occurrences of infection, organ failure, and death. For both the current and modified indexes, correlation between the severity of the pancreatitis and patient outcome was estimated using the Wilcoxon's rank sum test and Fisher's exact test. Interobserver agreement for both indexes was calculated using the kappa statistic.

RESULTS. When applying the modified index, the severity of pancreatitis and the following parameters correlated more closely than when the currently accepted index was applied: the length of the hospital stay (0-34 days) (modified index [p = 0.0054-0.0714] vs current index [p = 0.0052-0.3008]); the need for surgical or percutaneous procedures (10/66 patients) (modified index [p = 0.0112] vs current index [p = 0.0324]); and the occurrence of infection (21/66 patients) (modified index [p < 1e-10] vs current index [p < 1e-04]). Significant correlation between the severity of pancreatitis and the development of organ failure (9/66 patients) was seen only using the modified index (p = 0.0024), not the current index (p = 0.0513). The interobserver agreement was similar with the modified (κ range, 0.71-0.85) and the current (κ range, 0.63-0.86) indexes.

CONCLUSION. The modified CT severity index correlates more closely with patient outcome measures than the currently accepted CT severity index, with similar interobserver variability.

Introduction
Previous sectionNext section

Acute pancreatitis is generally classified into mild and severe forms: mild pancreatitis, so-called inter-stitial or edematous pancreatitis, is associated with minimal organ failure and an uneventful recovery. Severe pancreatitis, also referred to as necrotizing pancreatitis, occurs in approximately 20% of patients and is associated with organ failure or local complications, including necrosis, infection, or pseudocyst formation [1]. The rationale for assessing the severity of acute pancreatitis is mainly practical: mild pancreatitis responds well to supportive therapy, whereas severe pancreatitis requires intensive monitoring and specific therapies and has a more guarded prognosis [2, 3].

An ideal prognostic method that allows differentiation between patients with mild and those with severe pancreatitis should be accurate, easy to use, and widely available and should have low interobserver variability. It also should be applicable early in the disease process, so that patients who could potentially develop complications can be monitored more closely or empirically treated, for example with fluid resuscitation [4, 5]. Since 1974, several clinical and radiologic scoring systems have been developed for this purpose, including Ranson's criteria [6], the acute physiology and chronic health evaluation (APACHE II) scoring system [7], and the CT severity index [8, 9].

The CT severity index, developed by Balthazar and colleagues in 1994 [8, 9], was a significant advance because it helps clinicians to discriminate among mild, moderate, and severe forms of pancreatitis. The index focuses on the presence and degree of pancreatic inflammation and necrosis. On a 10-point severity scale, points are awarded for the presence or absence of fluid collections, in combination with an assessment of the presence and degree of pancreatic necrosis [8, 9]. Although this system has been successfully used to predict overall morbidity and mortality in patients with acute pancreatitis, it has limitations. First, the score obtained with the index does not significantly correlate with the subsequent development of organ failure [10], extrapancreatic parenchymal complications [11, 12], or peripancreatic vascular complications [13]. We postulated that if these complications were incorporated into a modified CT severity index, clinical outcomes may be predicted more accurately. Second, as documented in two independent studies [11, 14], the interobserver agreement for scoring CT scans using the CT severity index is only moderate, with the reported percentage of agreement approximating 75%. The source of this interobserver variability possibly relates to the complex categorization of the assessment of pancreatic inflammation and necrosis (four or five categories each). Therefore, we evaluated a novel modified and simplified CT scoring system in patients with acute pancreatitis to determine whether the scores obtained with that index could be used to predict clinical outcomes better than the scores obtained with the current scoring system while maintaining or improving the interobserver agreement.

Materials and Methods
Previous sectionNext section
Subjects

We performed a retrospective review of 266 consecutive patients who presented to our hospital over a 1-year period with a primary diagnosis of acute pancreatitis (International Classification of Diseases-9 [15] codes: 577.0, 577.1, 577.2, and 577.8). We obtained institutional review board approval for this study. Because this was a retrospective study, patient informed consent was not required. Of these 266 patients, 66 patients (37 women, 29 men; mean age, 49.2 years; age range, 19-87 years) underwent contrast-enhanced CT within 1 week of onset of symptoms. Of the remainder of patients, 105 patients underwent no cross-sectional imaging; eight patients underwent MRI; 68 patients underwent un-enhanced CT imaging; and 19 patients underwent their first contrast-enhanced CT more than one week after initiation of symptoms.

CT Technique

All examinations were performed on a MDCT scanner (Volume Zoom, Siemens Medical Solutions). Contrast-enhanced CT scans (collimation, 4 × 2.5 mm; reconstruction section thickness, 5 mm; reconstruction intervals, 5 mm) were obtained 40 sec after IV administration of 100 mL of iopromide 300 mg I/mL (Ultravist 300, Berlex Laboratories), injected at a rate of 3.0 mL/sec using a mechanical power injector. Opacification of the digestive tract was achieved with oral administration of 900 mL of barium sulfate suspension (Readi-Cat, E-Z-EM).

Image Analysis

CT scans were retrospectively and independently reviewed on PACS workstations (IMPAX, Agfa) by three experienced abdominal radiologists who were unaware of presenting signs and symptoms or of patient outcomes. The severity of the pancreatitis for each case was assessed by each observer using the CT severity index developed by Balthazar et al. [9] (Table 1), and the severity of pancreatitis was categorized as mild (score, 0-3 points), moderate (4-6 points), or severe (7-10 points). Subsequently, the severity of the pancreatitis was assessed by each observer using a modified CT severity index (Table 2). This index differs from the currently accepted index in the addition of a simplified evaluation of the presence and number of fluid collections and the extent of pancreatic necrosis and the assessment, with different weighting factors, of the presence of extrapancreatic findings, such as pleural fluid, ascites, extrapancreatic parenchymal abnormalities (infarction, hemorrhage, or subcapsular fluid collection), vascular complications (venous thrombosis, arterial hemorrhage, or pseudoaneurysm formation), and involvement of the gastrointestinal tract (inflammation, perforation, or intramural fluid collection). Using this modified index, the severity of pancreatitis for each patient was then categorized as mild (0-2 points), moderate (4-6 points), or severe (8-10 points).

TABLE 1 CT Severity Index [ 9 ]

TABLE 2 Modified CT Severity Index

Because the intention of the study was to compare the accepted CT severity index with the modified index on the basis on the existing and newly added features (with different points assigned), we opted to score the degree of inflammation and necrosis for both the indexes during the same interpretation session. In this way, we minimized the potential bias that could have been introduced if the same reviewer scored the same amount of necrosis and inflammation differently with the two scoring systems.

Outcome Parameters

Outcome parameters were collected from the hospital and radiology information systems and included the length of the hospital stay (in days), need for surgical intervention, need for percutaneous intervention (aspiration and drainage), evidence of infection in any organ system (positive results on a Gram stain or culture or the combination of a fever >100°F and an elevated WBC > 15,000/mm3), and evidence of organ failure. Patient records were retrospectively reviewed for the presence or absence of dysfunction in six separate organ systems as defined by Fagon et al. [16]. Respiratory failure was defined as a PaO2 of less than 60 mm Hg or by the need for ventilatory support. Cardiovascular system failure was defined as a systolic blood pressure of less than 90 mm Hg in the absence of hypovolemia with signs of peripheral hypoperfusion or by the need for continuous infusion of vasopressor or inotropic agents to maintain a systolic blood pressure of more than 90 mm Hg. Renal failure was defined as either a serum creatinine level that exceeded 300 μmol/L or urine output of less than 500 mL/24 hr or less than 180 mL/8 hr, or by the need for hemo- or peritoneal dialysis. Central nervous system failure was defined as a Glasgow Coma Scale score greater than 6 in the absence of sedation or by the sudden onset of confusion or psychosis. Hepatic failure was defined as serum bilirubin levels greater than 100 μmol/L or alkaline phosphatase levels greater than three times the upper limit of the normal range. Hematologic system failure was defined as a hematocrit level of less than 20%, WBC of less than 2,000/mm3, or platelet count of less than 40,000/mm3.

Data Analysis

Similar to the methods used in prior studies evaluating the CT severity index, the interobserver agreement for assessing the severity of pancreatitis (mild, moderate, or severe) with both the accepted CT severity index and the modified CT severity index was expressed for each pair of reviewers (observer 1 and observer 2; observer 1 and observer 3; observer 2 and observer 3) by means of the exact percentage agreement, along with the kappa statistic, which is used to estimate the proportion of interrater agreement above that expected by chance. A weighted kappa statistic of 0.41-0.60 was considered to indicate moderate agreement, 0.61-0.80 was considered to indicate good agreement, and 0.81-1.00 was considered to indicate excellent agreement [17].

A consensus score was obtained so that the severity of the pancreatitis (for both indexes) could be correlated with patient outcome. Consensus was determined if the results of two or all three of the three interpreters agreed. Cases in which all three interpreters scored the severity of the pancreatitis differently were excluded from the analysis. Correlation between the severity of the pancreatitis and the patient outcome measures was obtained with the one-sided Wilcoxon's rank sum test and Fisher's exact test.

Results
Previous sectionNext section
Severity of Pancreatitis

Using the currently accepted CT severity index, the observers graded the severity of the acute pancreatitis as mild in 42, moderate in 19, and severe in five of the 66 patients in whom CT scans were obtained. Interobserver agreement on these observations ranged from 74% to 83% (mean, 78%). The kappa statistic for each pair of observers ranged from 0.48 to 0.70, indicating moderate to good agreement (Figs. 1A, 1B). In none of the cases was the severity of the pancreatitis scored differently by all three observers.

figure
View larger version (75K)
Fig. 1A. 74-year-old man with acute pancreatitis. Axial contrast-enhanced CT scan shows presence of bilateral pleural effusions.

figure
View larger version (63K)
Fig. 1B. 74-year-old man with acute pancreatitis. Axial contrast-enhanced CT scan shows one fluid collection in anterior pararenal space and minimal necrosis (< 30%). Score was 5 (moderate pancreatitis) on currently accepted CT severity index and 8 (severe pancreatitis) on modified CT severity index.

Using the modified CT severity index, the observers graded severity of the acute pancreatitis as mild in 34, moderate in 22, or severe in 10 of the 66 patients with CT scans. Observers agreed on these grades in 70-79% of the cases (mean, 74%). The kappa statistic for each pair of observers ranged from 0.52 to 0.65, indicating moderate to good agreement (Figs. 1A, 1B). In none of the cases was the severity of the pancreatitis scored differently by all three observers.

Patient Outcome

The length of the hospital stay ranged from 0 to 34 days (mean, 7.2 days). Figure 2 outlines the mean length of stay for the different severity subgroups for both indexes. A total of 10 (15%) of the 66 patients underwent surgical or percutaneous interventions. Surgical débridement was performed in four patients; percutaneous CT-guided catheter drainage of pancreatic fluid collections was performed in four patients; and percutaneous CT-guided needle aspiration of pancreatic fluid collections was performed in seven patients. Evidently, some patients underwent a combination of procedures (aspiration, drainage, and surgery). Evidence of infection was present in 21 (32%) of the 66 patients. Organ system failure was present in nine patients (14%), including failure of the heart (n = 7), pulmonary system (n = 4), central nervous system (n = 3), and kidneys (n = 1). In five of these patients, more than one organ system failed. One patient died.

figure
View larger version (19K)
Fig. 2. Graph shows correlation between length of hospital stay and CT severity index score for pancreatitis using currently accepted (represented by black bars) and modified (represented by white bars) CT severity indexes. Although significant correlation (p = 0.035) was seen between modified severity index score and length of hospital stay for all severity groups (mean, 3 days for mild pancreatitis, 8 days for moderate pancreatitis, and 12 days for severe pancreatitis), no significant correlation (p = 0.15) was seen between currently accepted index and length of hospital stay in moderate and severe pancreatitis groups (mean, 9 and 11 days, respectively).

Correlation of Scoring Indexes with Patient Outcome

For the correlation between the two indexes and the length of the hospital stay (Figure 2 and Tables 3 and 4), we found a statistically significant correlation between the modified index and the length of hospital stay (p = 0.035) for all severity groups. However, when using the current CT severity index, no significant difference (p = 0.15) in the length of the hospital stay was seen between the moderate and severe pancreatitis group.

TABLE 3 Patient Outcomes Using CT Severity Index

TABLE 4 Patient Outcomes Using Modified CT Severity Index

Significant correlation between the need for surgical or percutaneous interventions and the severity of pancreatitis was seen with both the accepted CT severity index (p = 0.0324) and the modified CT severity index (p = 0.0112). Similarly, the presence of infection was correlated with higher scores on both the accepted CT severity index (p < 0.0001) and the modified CT severity index (p < 10-10). For both outcome parameters, however, a stronger correlation was seen with the modified index than with the accepted index.

Finally, a significant correlation (p = 0.0024) was seen between the modified severity index score and the development of organ failure, but the accepted severity index score did not correlate significantly with the development of organ failure (p = 0.0513).

Discussion
Previous sectionNext section

The introduction of the CT severity index in 1994 was a significant advance in the assessment of patients with acute pancreatitis [5, 8, 9]. This internationally accepted severity index, which is based on scoring the presence and degree of pancreatic inflammation and pancreatic necrosis, not only allows accurate differentiation of mild from severe pancreatitis but also numerically correlates with the patient's prognosis. However, despite the fact that the CT severity index has been successfully used to predict overall morbidity and mortality in patients with acute pancreatitis, recent literature has revealed the limitations of this currently accepted CT severity index [10-14]. First, the presence of organ failure [10], extrapancreatic parenchymal complications [11, 12], and peripancreatic vascular complications [14] does not significantly correlate with the score obtained with this index. Second, as reported in two independent studies [11, 14], the interobserver agreement for scoring CT scans using the current CT severity index is only moderate, with a reported percentage of agreement approximating 75%. Finally, as acknowledged by Balthazar et al. [8-9] and confirmed by others [14], no significant difference in morbidity and mortality is seen, when using the CT severity index, between patients who have 30-50% necrosis and patients who have more than 50% necrosis. Therefore, it is questionable whether these specific categorizations of the degree of pancreatic necrosis are necessary.

In light of these limitations, we hypothesized that a simplified CT severity index that incorporated features reflecting organ failure and extrapancreatic complications would be useful for predicting outcomes more accurately. The modified CT severity index differentiates only between presence or absence of acute fluid collections and, therefore, does not require a count of the collections. Furthermore, on the modified index, the presence of pancreatic necrosis is only scored as “no necrosis,” “minimal necrosis,” or “substantial necrosis,” thereby eliminating the unnecessary categorization between patients who have 30-50% necrosis and patients that have more than 50% necrosis. In addition, with moderate weighting (2 points or 20%), the presence of extrapancreatic findings, such as pleural fluid, ascites, extrapancreatic parenchymal abnormalities, vascular complications, or involvement of the gastrointestinal tract, can be incorporated into the analysis.

Our results confirm that the currently accepted CT severity index is indeed a powerful tool with which to predict morbidity in patients with acute pancreatitis. When comparing patients with mild pancreatitis and those with severe pancreatitis, we documented a statistically significant correlation between the numeric score obtained with the currently accepted index and the presence of infection, the need for surgery and percutaneous interventions, and the length of the hospital stay. However, we found no significant correlation between the score obtained with the accepted index and the length of the hospital stay when comparing patients with moderate pancreatitis and those with severe pancreatitis, nor did we see a significant correlation between the score obtained with the accepted index and the development of organ failure. The score obtained with the modified index, however, not only showed a stronger correlation for all parameters than the score obtained with the accepted index, but it also could be used to accurately predict the length of the hospital stay when comparing patients with moderate and those with severe pancreatitis and the development of organ failure. We speculate that the presence of ascites and pleural fluid may be responsible for the improved correlation, because they may be indicators of organ dysfunction.

Reflecting on the results from the interobserver agreement analysis, we were surprised to find a similar interobserver variability for both CT severity indexes because of the inherent simplifications of the modified index. Although the underlying reason for this similarity remains unclear, we postulate that the observers' agreement on the presence or absence of mild pancreatic necrosis remains challenging and that introducing new imaging features to be scored on the modified index may have increased the interobserver variability.

Our study had two important limitations. First, it was a nonrandomized retrospective study with a medium-sized sample. Second, from the consecutive patient cohort diagnosed with acute pancreatitis in our institution, we analyzed only the subgroup of patients who underwent contrast-enhanced CT within 1 week of the onset of symptoms. Therefore, to establish the true prognostic value of the modified CT severity index, the index probably has to be tested prospectively in all patients with acute pancreatitis at the time of presentation. However, because not all patients with acute pancreatitis need to have a CT examination and because patients with renal failure are unable to undergo a contrast-enhanced CT study, the methodology used in our study reflects, in some respects, current clinical practice.

In conclusion, compared with the currently used CT severity index, the modified CT severity index has a similar interobserver variability but correlates more closely with patient outcome in all the parameters studied, especially with the length of the hospital stay and the development of organ failure.

Address correspondence to K. J. Mortele ().

References
Previous sectionNext section
1. Bradley EL III. A clinically based classification system for acute pancreatitis: summary of the International Symposium on Acute Pancreatitis, Atlanta, GA, September 11 through 13, 1992. Arch Surg 1993; 128:586-590 [Google Scholar]
2. Steinberg W, Tenner S. Acute pancreatitis. N Engl J Med 1994; 330:1198-1210 [Google Scholar]
3. de Beaux AC, Palmer KR, Carter DC. Factors influencing morbidity and mortality in acute pancreatitis; an analysis of 279 cases. Gut 1995; 37:121-126 [Google Scholar]
4. Agarwal N. Assessment of severity in acute pancreatitis. Am J Gastroenterol 1991; 86:1385-1391 [Google Scholar]
5. Balthazar EJ. Acute pancreatitis: assessment of severity with clinical and CT evaluation. Radiology 2002; 223:603-613 [Google Scholar]
6. Ranson JHC, Rifkind KM, Roses DF, Fink SD, Eng K, Spencer FC. Prognostic signs and the role of operative management in acute pancreatitis. Surg Gynecol Obstet 1974; 139:69-81 [Google Scholar]
7. Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Crit Care Med 1985; 13:818-829 [Google Scholar]
8. Balthazar EJ, Robinson DL, Megibow AJ, Ranson JHC. Acute pancreatitis: value of CT in establishing prognosis. Radiology 1990; 174:331-336 [Google Scholar]
9. Balthazar EJ, Freeny PC, vanSonnenberg E. Imaging and intervention in acute pancreatitis. Radiology 1994; 193:297-306 [Google Scholar]
10. Lankish PG, Pflichthofer D, Lehnick D. No strict correlation between necrosis and organ failure in acute pancreatitis. Pancreas 2000; 20:319-322 [Google Scholar]
11. Mortele KJ, Mergo P, Taylor H, Ernst M, Ros PR. Renal and perirenal space involvement in acute pancreatitis: state-of-the-art spiral CT findings. Abdom Imaging 2000; 25:272-278 [Google Scholar]
12. Wiesner W, Studler U, Kocher T, Degen L, Buitrago-Tellez CH, Steinbrich W. Colonic involvement in non-necrotizing acute pancreatitis: correlation of CT findings with the clinical course of affected patients. Eur Radiol 2003; 13:897-902 [Google Scholar]
13. Inoue K, Hirota M, Beppu T, et al. Angiographic features in acute pancreatitis: the severity of abdominal vessel ischemic change reflects the severity of acute pancreatitis. J Pancreas 2003; 4:207-213 [Google Scholar]
14. Lecesne R, Tourel P, Bret PM, et al. Acute pancreatitis: interobserver agreement and correlation of CT and MR cholangiopancreatography with outcome. Radiology 1999; 211:727-735 [Google Scholar]
15. U.S. Department of Health and Human Services. International classification of diseases: ninth revision—clinical modification, 6th ed. Washington, DC: U.S. Department of Health and Human Services,2000 [Google Scholar]
16. Fagon JY, Chastre J, Novara A, Medioni P, Gibert C. Characterization of intensive care unit patients using a model based on the presence or absence of organ dysfunctions and/or infection: the ODIN model. Intensive Care Med 1993; 19:137-144 [Google Scholar]
17. Fleiss JL. Statistical methods for rates and proportions, 2nd ed. New York, NY: Wiley, 1981:212-255 [Google Scholar]

Recommended Articles

A Modified CT Severity Index for Evaluating Acute Pancreatitis: Improved Correlation with Patient Outcome

Full Access, , , ,
American Journal of Roentgenology. 2005;185:406-417. 10.2214/ajr.185.2.01850406
Abstract | Full Text | PDF (3458 KB) | PDF Plus (3627 KB) 
Full Access, , , , ,
American Journal of Roentgenology. 2014;202:479-492. 10.2214/AJR.13.11463
Abstract | Full Text | PDF (1418 KB) | PDF Plus (1472 KB) 
Full Access, , , ,
American Journal of Roentgenology. 2015;205:W32-W41. 10.2214/AJR.14.14056
Abstract | Full Text | PDF (1038 KB) | PDF Plus (1033 KB) 
Full Access, ,
American Journal of Roentgenology. 2014;203:1192-1204. 10.2214/AJR.13.12386
Abstract | Full Text | PDF (1122 KB) | PDF Plus (1179 KB) | Erratum 
Full Access,
American Journal of Roentgenology. 2011;197:1309-1321. 10.2214/AJR.11.7420
Abstract | Full Text | PDF (1195 KB) | PDF Plus (1215 KB) 
Full Access, ,
American Journal of Roentgenology. 2011;197:W375-W381. 10.2214/AJR.10.4339
Citation | Full Text | PDF (1075 KB) | PDF Plus (1086 KB)