Severity assessment of acute pancreatitis using four prognostic factors – a nationwide multicenter study of 3682 cases in Japan

Dear Sir,

We read with great interest the article by Ikeura et al. ¹ assessing the efficacy of the prognostic factor score in the Japanese severity criteria for acute pancreatitis (AP). By analyzing 1159 patients with severe AP, they showed that the area under the curve (AUC) for the receiver-operating characteristic curve of the prognostic factor score for predicting AP-related mortality was 0.78, which was comparable to that for the well-known acute physiology and chronic health evaluation (APACHE) II score (AUC = 0.80). One weakness of their study is that mild AP cases were not included in the analysis due to the study design. In addition, severity assessment using the nine factors might be too complicated. In Japan, the Research Committee of Intractable Pancreatic Diseases, under the support of the Ministry of Health, Labor, and Welfare of Japan, has conducted nationwide epidemiological surveys of AP including both severe and mild cases every 4–5 years.^2–4 We recently reported that four (base excess or shock, renal failure, systemic inflammatory response syndrome criteria and age) out of the nine prognostic factors were associated with AP-related death by multivariate analysis in the 2011 survey. ⁵ Here, we validated the utility of the severity assessment system based on these four factors in 3682 AP cases collected for the 2003 (285 cases), 2007 (1146 cases), and 2011 (2251 cases) surveys.

We performed receiver-operating characteristic curve analysis to evaluate the predictive accuracy of the prognostic factor scores for mortality. The AUC was 0.80 for the nine prognostic factors. If the cut-off point was set at score = 3 as adopted currently, the sensitivity was 0.47 with a specificity of 0.93. The AUC for predicting mortality was 0.78 for the four prognostic factors, suggesting the comparable utility of these four factors in the severity assessment. The sensitivity was 0.53 with a specificity of 0.92 if the cut-off point was set at score = 2. If we define the patients fulfilling at least two out of the four prognostic factors as severe, 342/3682 (9.3%) patients had severe AP. The mortality of severe AP cases was 17.3% (59/342), whereas that of mild AP cases was 1.6% (52/3340) (Figure 1). If we choose only two factors (renal failure and age) that showed a stronger association with AP-related mortality among the four factors, ⁵ the AUC was decreased to 0.72. The sensitivity was 0.22 with a specificity of 0.98 if the cut-off point was set at score = 2.

Figure 1.

Mortality of the 3682 acute pancreatitis patients stratified by the four prognostic factor scores.

Our results suggested that severity assessment using the four prognostic factors was equivalent to the current Japanese severity criteria employing nine factors. Accurate assessment of severity in patients with AP is still a challenge. ⁶ Development of a more convenient and accurate severity assessment system would contribute to improving the outcomes of this intractable disease.