The harmless acute pancreatitis score (HAPS) identifies non-severe patients: A systematic review and meta-analysis

We previously described a scoring system to identify patients with harmless acute pancreatitis as defined by absence of pancreatic necrosis, no need for artificial ventilation or dialysis, and non-fatal course. This scoring system, the Harmless Acute Pancreatitis Score (HAPS), can be quickly calculated from three parameters: absence of abdominal tenderness or rebound, normal hematocrit and normal creatinine level. We aim to assess the positive predictive value (PPV) of the HAPS by performing a meta-analysis of subsequently published studies.

We performed a literature search using Pubmed, Web of Science^TM and Google Scholar. We used random effects models, with maximum likelihood estimates, to estimate the PPV of HAPS. We produced forest plots and used the I² statistic to quantify heterogeneity.

Twenty reports covering 6374 patients were identified. The overall PPV based on 16 studies that closely followed the original description of the HAPS system was 97% (95%CI 95-99%) with significant heterogeneity (I² = 76%; P < 0.01). For 11 studies in which HAPS was used to identify patients with mild AP, the overall PPV dropped to 83% (74-91%). For 8 studies in which HAPS was used to predict non-fatal course the overall PPV was 98% (97-100%).

The HAPS, if used as originally defined, accurately identifies patients with non-severe AP who will not require ICU care and facilitate selection of patients who can be discharged after a short stay on a general ward or can even be cared for at home. This could free hospital beds for other purposes and decrease healthcare costs.