Analgesic phenotypes in acute pancreatitis: Clustering analysis of two prospective global cohort studies

Pain management in acute pancreatitis (AP) is of critical therapeutic and prognostic significance. We aimed to identify patient phenotypes in AP based on pain management strategies and evaluate their prognostic impact.

Data from two major international prospective cohort studies of AP patients (PAINAP study, n = 2119 and APPRENTICE study, n = 1544) from 141 centers worldwide were collated. Demographic data and analgesic use (within 72 h of presentation) were used for latent class analysis of binary analgesic data. The number of clusters was determined by minimisation of Bayesian information criterion. Cluster assignment and AP outcomes were interrogated with multivariable mixed-effects logistic regression.

Overall, 3469 patients (median age 52; 47 % female) were analysed. There were 1015 (29.2 %) patients that had moderately severe to severe AP. Within the first 72 h, 494 (14.2 %) patients received non-steroidal anti-inflammatory drugs (NSAIDs), 1410 (40.6 %) weak opioids, 1347 (14.2 %) strong opioids, and 48 (1.4 %) epidural analgesia. There were significant variations in analgesic prescribing patterns across centers (p < 0.001, interclass correlation coefficient 43.8 %). Latent class analysis identified 5 unique patient clusters: early NSAIDs use, minimal analgesic use, strong opioid use, early multimodal analgesia use, and early weak opioid use. The cluster characterized by early NSAIDs was associated with non-severe AP (adjusted odds ratio 0.64, 95 % confidence interval 0.44-0.93, p = 0.02), which could suggest that the use of NSAIDs was perhaps driven by milder pain severity.

Unique clusters in the management of pain in AP were identified, with associations to severity of AP. Substantial centre-level variations exist in the patterns of analgesic prescribing globally, contributing to variations in outcomes.